Official Clinical Practice Journal of the

www.cghjournal.org March 2008 Volume 6 Number 3

Issue Highlights

Photodynamic Therapy for Cholangiocarcinoma

Chronic Abdominal Pain in Adolescents

Prognostic Factors in Acute Liver Failure

Supplementation Versus Dietary Counseling in Chronic Pancreatitis

I

M

K

*

Aodm2olntcRcr

ftuadmsFIsSssbletcwa

Mt

mage of the Month

etastatic Malignant Melanoma of the Gastrointestinal Tract

EELY R. PARISIAN,* JOEL E. MCFARLAND,‡ and ASHOK N. SHAH‡

Department of Internal Medicine, and ‡Department of Gastroenterology, University of Rochester Medical Center, Rochester, New York

a6midamppsmasn

no

1

2

3

af

75-year-old man with a history of malignant melanoma presentedwith fatigue and dyspnea for 4 months and melena in the setting

f increased nonsteroidal anti-inflammatory drug use over the past 7ays for knee pain. The patient first was found to have malignantelanoma of the right upper cheek and the right and left auricle in

003, with a second occurrence in 2005 on the posterior neck diagnosedn excision as Clark level IV with a Breslow depth of 6 mm. Sentinel

ymph node mapping and a positron emission tomography scan wereegative for evidence of metastatic disease. He presented to our insti-ution after an episode of presyncope. A review of systems was signifi-ant only for anorexia and a weight loss of 20 pounds over 3 months.elevant laboratory values on admission revealed a decreased hemato-rit from 42% to 23% over 1 month, and an international normalizedatio of 2.5 while on anticoagulation for atrial fibrillation.

Before esophagogastroduodenoscopy the patient received 4 units ofresh-frozen plasma and 5 units of packed red blood cells. Esophagogas-roduodenoscopy revealed numerous ulcerations including a pigmented,mbilicated, polypoid ulcer in the body of the stomach (Figure A), as wells a protuberant, ulcerated, and pigmented mass in the bulb of theuodenum (Figure B). Histology of the duodenal biopsy specimen showedelanocytes throughout and nuclei with differences in size, shape, and

taining intensity consistent with malignant melanoma (Supplementaryigure; supplementary material available online at www.cghjournal.org).mmunohistochemical stains showed that the neoplastic cells weretrongly reactive for Melan-A and HMB-45, but were not reactive for S-100.taging computerized tomography of the abdomen with contrast showedubcutaneous nodules and a nodule adjacent to the duodenal bulb con-istent with esophagogastroduodenoscopy findings in addition to multipleilateral pulmonary nodules, bilateral large adrenal masses, and multiple

iver lesions compatible with metastatic disease. There was no furthervidence of bleeding and the patient was discharged home with anticipa-ion of beginning treatment with chemotherapy or biologic agents afteronsultation with the oncology department. The risk of rebleeding out-eighed the benefit of maintaining anticoagulation for atrial fibrillationnd therefore Warfarin was discontinued at the time of discharge.

Malignant melanoma of the gastrointestinal tract is a rare neoplasm.alignant melanoma of the skin is one of the most common malignancies

o metastasize to the gastrointestinal tract.1 Studies have reported that

utopsies show gastrointestinal tract involvement in approximately 50% to0% of patients with malignant melanoma; however, clinical diagnosis isade in only 4.7% of all patients with malignant melanoma.2 Discourag-

ngly, many patients with resected metastatic melanoma will present withisease relapse years later. Several studies have reported relief of symptomsnd prolonged survival with surgical resection in patients with localized oroderate spread of disease. Patients with extensive metastasis and poor

erformance status are considered inoperable, as in the case that we haveresented here. Further, extraintestinal manifestations of disease have beenhown in 50% of patients who present with gastrointestinal tract involve-

ent.3 As a result, persistent vague abdominal pain, fatigue, melena, andnemia in patients with a history of malignant melanoma should promptuspicion of metastasis to the gastrointestinal tract and aggressive diag-ostic evaluation at the onset.

Supplementary DataNote: To access the supplementary material accompa-

ying this article, visit the online version of Clinical Gastroenter-logy and Hepatology at www.cghjournal.org.

References. Liang KV, Sanderson SO, Nowakowski GS, et al. Metastatic ma-

lignant melanoma of the gastrointestinal tract. Mayo Clin Proc2006;81:511–516.

. Panagiotou I, Brountzos EN, Bafaloukos D, et al. Malignant mela-noma metastatic to the gastrointestinal tract. Melanoma Res2002;12:169–173.

. Schuchter LM, Green R, Fraker D. Primary and metastatic diseasesin malignant melanoma of the gastrointestinal tract. Curr OpinOncol 2000;12:181–185.

The authors would like to give special thanks to Zhenhong Qu, MD,nd Ellen Giampoli, MD, of the University of Rochester Medical Centeror providing pathologic images.

© 2008 by the AGA Institute1542-3565/08/$34.00

doi:10.1016/j.cgh.2007.12.010

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2008;6:xxiv

Supplementary figure.

xxiv.e1 CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

I

T

I

D

Am

mpotptlosof

mage of the Month

ransmural Perforation of the Stomach by a Fishbone

GOR J. NASTASKIN, KANAT RANSIBRAHMANAKUL, and WALTER TRUDEAU

ivision of Gastroenterology, University of California Davis Medical Center, Sacramento, California

fawo

pttrqsiglofp

1

2

3

4

73-year-old Russian woman with a history of poorly con-trolled hypertension presented to the emergency depart-

ent with a 2-day history of epigastric abdominal pain.One day before her visit she ate a soup called uha, which isade from fish. Four hours later she developed sharp epigastric

ain radiating to the back. She denied fevers, nausea, vomiting,r diarrhea. The pain was not associated with food intake andhere were no factors that improved or worsened her pain. Herhysical examination was unremarkable except for mild epigas-ric tenderness to palpation without peritoneal signs. Basicaboratory test results were normal. Computerized tomographyf the abdomen showed a radiodense linear foreign body mea-uring 3 cm, extending transmurally through the anterior wallf the stomach, associated with mesenteric stranding, and no

ree air (Figure A). An esophagogastroduodenoscopy was per-

ormed and a fish bone penetrating the stomach wall (Figure B,rrow) with surrounding edema and ulcerations was noted. Thisas removed with a snare (Figure C). The patient was observedvernight and discharged home without complication.

More than 80% of foreign bodies that reach the stomach willass unimpeded through the gastrointestinal tract.1 However,hose thicker than 2 cm and longer than 5 cm tend to lodge inhe stomach.2 Objects that lodge in the gastric lumen canemain there for prolonged periods without adverse conse-uences.3 Complications of foreign body ingestion include ob-truction (occurring particularly at the pyloric sphincter andleocecal valve), perforation, and hemorrhage. Watchful waitingenerally is justified and may include administration of emetics,axatives, or spasmolytics, depending on the type and site ofbject. However, early endoscopic or surgical removal of largeoreign bodies from the stomach and of objects causing com-lications is recommended.4

References. Selivanov V, Sheldon GF, Cello JP, et al. Management of foreign

body ingestion. Ann Surg 1984;199:187–191.. Koch H. Operative endoscopy. Gastrointest Endosc 1977;24:65–

68.. Roark GD, Subramanyam K, Patterson M. Ingested foreign mate-

rial in mentally disturbed patients. South Med J 1983;76:1125–1127.

. Velitchkov NG, Grigorov GI, Losanoff JE, et al. Ingested foreignbodies of the gastrointestinal tract: retrospective analysis of 542cases. World J Surg 1996;20:1001–1005.

© 2008 by the AGA Institute1542-3565/08/$34.00

doi:10.1016/j.cgh.2007.12.024

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2008;6:xxvi

I

HK

*

Aastpdae(oew

itcbmwpotwp

mbpus(cuwn

mage of the Month

airball in the Stomach: A Case of Gastric TrichobezoarOLA HISAMUDDIN* and CHRISTOPHER P. BRANDT‡

Division of Gastroenterology and ‡Department of General Surgery, MetroHealth Medical Center, Case Western Reserve University, Cleveland, Ohio

agneagc

gubts

TiTape

no

1

2

3

43-year-old woman presented to the gastroenterology clinicwith symptoms of upper-abdominal discomfort, early satiety,

nd the sensation of a “ball rolling within the abdomen” foreveral months. She initially had consulted her obstetrician forhese symptoms with concerns that she could be pregnant. Theatient denied any history of gastrointestinal symptoms or ab-ominal surgery before the onset of her current symptoms. Thebdominal examination was remarkable for a palpable mass in thepigastrium. A computerized tomography scan of the abdomenFigure A) showed a foreign body (A) in the stomach. At laparot-my (Figure B), a large mass occupying the entire stomach andxtending into the pylorus was noted. A longitudinal gastrostomyas performed and a large foreign body was removed (Figure B).This was a trichobezoar measuring 840 g that was removed

ntact from the stomach (Supplementary figure; see supplemen-ary material online at www.cghjournal.org) without any compli-ations. An upper-gastrointestinal endoscopy was not performedefore surgery because the patient’s symptoms and CT of abdo-en was suggestive of a foreign body in the stomach in keepingith a bezoar. Furthermore, there was some suggestion of theatient eating her hair in the past. However, it became morebvious during the follow-up evaluation that she had a history ofrichotillomania and trichophagia that was most pronouncedhile she was pregnant 4 years before her initial presentation. Theatient then was referred for a psychiatric assessment.

The word bezoar is derived from the Persian language, whicheans “protection from the poison.” Historically, bezoars were

elieved to have the power of a universal antidote against anyoison. Gastric bezoars are the end result of the accumulation ofndigested, ingested foreign material that form masses in thetomach. They are classified as trichobezoar (hair), phytobezoarfruits and vegetables), or pharmacobezoar (drugs) on the basis ofomposition. Bezoars are encountered in less than 1% of patientsndergoing upper endoscopy.1 Trichobezoars develop in patientsith underlying psychiatric problems, trichophagia, trichotilloma-

ia, and gastric surgery. Delayed gastric emptying was considered

s a predisposing factor, however, a study revealed no evidence ofastroparesis in patients with bezoars.2 These patients present withonspecific symptoms such as abdominal pain, nausea, vomiting,arly satiety, anorexia, and weight loss. Nonetheless, they usuallyre diagnosed as incidental finding. Patients can present withastrointestinal bleeding related to the ulcers, which may beaused by peptic ulcer disease or pressure necrosis.

Diagnosis is made by radiologic means such as plain radio-raph, barium studies, and computerized tomography scan orpper-gastrointestinal endoscopy. Trichobezoars usually need toe removed either by endoscopy or surgery. Attempts to dissolvehem with enzymes such as cellulose and papain usually are un-uccessful.

Bezoars have a tendency to recur in up to 14% of patients.3

herefore, once the bezoar is removed by whichever means, it ismportant to treat the underlying or predisposing condition.hese include increased water intake, to alter the diet appropri-tely (eg, avoid persimmons and stringy vegetables) in case ofhytobezoars, to chew food carefully, and to seek psychiatricvaluation for trichobezoars.

Supplementary DataNote: To access the supplementary material accompa-

ying this article, visit the online version of Clinical Gastroenter-logy and Hepatology at www.cghjournal.org.

References. Kadian RS, Rose JF, Mann NS. Gastric bezoars: spontaneous

resolution. Am J Gastroenterol 1978;70:79–82.. Calabuig R, Navarro S, Carrio I, et al. Gastric emptying and bezo-

ars. Am J Surg 1989;157:287–290.. Robles R, Parrilla C, Escamilla J, et al. Gastrointestinal bezoars.

Br J Surg 1994;81:1000–1001.

© 2008 by the AGA Institute

1542-3565/08/$34.00

doi:10.1016/j.cgh.2007.12.013

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2008;6:xxviii

Supplementary figure.

xxviii.e1 IMAGE OF THE MONTH CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

I

G(K

D

AdspitcwufiCias

mage of the Month

astric Outlet Obstruction Caused by a Large Gallstone in the DuodenumBouveret’s Syndrome)ATSUNARI KISHI, KAZUKI YAMADA, and TOSHIRO SUGIYAMA

epartment of Gastroenterology, University of Toyama, Toyama, Japan

te

adac6Eta

1

2

58-year-old woman presented with a 7-day history ofsudden onset of severe vomiting followed by abdominal

istention and an inability to tolerate oral intake beyond smallips of water. Computerized tomography examination showedneumobilia, cholecystoduodenal fistula, and a 4.5 � 3.5 cm

mpacted gallstone in the duodenal bulb. Endoscopic examina-ion of the upper-gastrointestinal tract showed pyloric stenosisaused by the impacted stone (Figure A). The fistulous stomaas visualized in the posterior wall of the duodenal bulb (Fig-re B, white arrow). An upper-gastrointestinal series showed astula with the gallbladder and the common bile duct (Figure, thin arrow). These findings confirmed the diagnosis of an

mpacted gallstone (Figure C, wide arrow) in the duodenal bulbs the cause of gastric outlet obstruction. Because the endo-

copic stone retrieval and transendoscopic shock-wave litho-

ripsy was unsuccessful, surgical lithotomy consisting of simplenterotomy was performed.

Bouveret’s syndrome is a gastric outlet obstruction caused bylarge gallstone impacted in the duodenal bulb, and was first

escribed by Bonnet in 1841 and Bouveret in 1896. It almostlways presents with abdominal pain and vomiting and oc-urs most commonly in elderly women with a mean age of8.6 years.1 Endoscopic retrieval has a low success rate (10%).nterolithotomy or gastrotomy, with or without cholecystec-

omy and fistula repair, are the most common surgical ther-pies.2

References. Frattaroli FM, Reggio D, Guadalaxara A, et al. Bouveret’s syn-

drome: case report and review of the literature. Hepatogastroen-terology 1997;44:1019–1022.

. Cappell MS, Davis M. Characterization of Bouveret’s syndrome: acomprehensive review of 128 cases. Am J Gastroenterol 2006;101:2139–2146.

© 2008 by the AGA Institute1542-3565/08/$34.00

doi:10.1016/j.cgh.2007.12.027

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2008;6:e11

I

M

A

D

mage of the Month

irizzi With Pre-Bouveret’s Syndrome

LI SHAKOURI and SHOU–JIANG TANG

epartment of Digestive and Liver Diseases, UT Southwestern Medical Center, Dallas, Texas

Asm3mlAitTtagdtcfilttaobtaa

didtstdsssItsctmd

35-year-old woman was admitted with a 3-week history ofrecurrent biliary colic, jaundice, nausea, and nonbilious eme-

is after meals. Her serum bilirubin was 18 mg/dL (normal, 0.2–1.3g/dL), and her alkaline phosphatase was 1436 units/L (normal,

8–126 units/L). A right upper quadrant ultrasound showed com-on bile duct dilatation with evidence of cholelithiasis and mass-

ike structure in the gallbladder fossa. Murphy’s sign was negative.computed tomography scan of the abdomen/pelvis showed

ntrahepatic duct dilatation and common bile duct (CBD) dilata-ion up to 2.5 cm, just before the head of the pancreas (Figure A).here was a large hypodense lesion at this level with a hyperat-

enuating rim (Figure A, arrow), with abrupt duct narrowing there-fter. There was no pancreatic head mass. On endoscopic retro-rade cholangiopancreatography (ERCP), a 3- to 4-cm fillingefect was seen coming from the cystic duct and causing obstruc-ion within the CBD, ie, Mirizzi syndrome (Figure B). On endos-opy, there was a large protruding bulge at the lateral wall in therst portion of the duodenum (Figure C), corresponding to the

arge gallstone seen on cholangiogram. On fluoroscopy, the tip ofhe endoscope was “touching” the gallstone when we placed tip ofhe endoscope on the duodenal bulge. The large gallstone movedppropriately on gentle poking of the duodenal bulge with the tipf the endoscope. After biliary sphincterotomy, a 10F � 9-cmiliary stent was placed with good drainage. During cholecystec-omy, a 4-cm gallstone was removed from the CBD, and there waslmost complete obliteration of the cystic duct with formation offistulous tract from the gallbladder to the CBD.Mirizzi syndrome is a rare cause of obstructive jaundice pro-

uced by the impaction of a gallstone either in the cystic duct orn the gallbladder, resulting in stenosis of the extrahepatic bileuct.1,2 Although she had Mirrizi syndrome, the patient’s presen-ation was also concerning for Bouveret’s syndrome, which de-cribes gastric outlet obstruction caused by impacted gallstone inhe pylorus or duodenum.3 A single gallstone of at least 2.5 cm iniameter is the most common underlying cause of Bouveret’syndrome. The symptoms are epigastric pain, nausea, and eme-is. Jaundice is an uncommon physical finding with Bouveret’syndrome and is more commonly seen with Mirizzi syndrome.n patients with Mirizzi syndrome, a preoperative diagnosis andherapy by ERCP reduce the risk of biliary damage duringurgery. Although enterotomy or gastrotomy with or withoutholecystectomy and fistula repair is the most common surgicalherapy, there are reported cases of successful endoscopic treat-

ent of Bouveret’s syndrome with intracorporeal electrohy-raulic lithotripsy and mechanical lithotripsy.

© 2008 by the AGA Institute1542-3565/08/$34.00

doi:10.1016/j.cgh.2007.12.006

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2008;6:e12–e13

1

2

3

March 2008 CLINICAL GASTROENTEROLOGY AND HEPATOLOGY e13

References

. Lai EC, Lau WY. Mirizzi syndrome: history, present and futuredevelopment (review). ANZ J Surg 2006;76:251–257.

. Abou-Saif A, Al-Kawas FH. Complications of gallstone disease:

Mirizzi syndrome, cholecystocholedochal fistula, and gallstoneileus (review). Am J Gastroenterol 2002;97:249–254.

. Cappell MS, Davis M. Characterization of Bouveret’s syndrome: acomprehensive review of 128 cases (review). Am J Gastroenterol2006;101:2139–2146.

I

B

L

D

Avaelpdpuigsotslsnmstoa

o

mage of the Month

ouveret’s Syndrome: Diagnosis and Endoscopic Treatment

INCOLN E. V. V. C. FERREIRA, MARK D. TOPAZIAN, and TODD H. BARON

epartment of Medicine, Division of Gastroenterology & Hepatology, Mayo Clinic, Rochester, Minnesota

wc

mdm

no

1

2

3

n 82-year-old woman was admitted with a 1-day history ofepigastric abdominal pain accompanied by nausea and

omiting. She denied fever, chills, or melena. Serum amylasend lipase levels were increased markedly. Abdominal comput-rized tomography scan showed a 3.6-cm lamellated gallstoneodged in the second portion of the duodenum (Figure A). Theancreatic duct as well as the extrahepatic and intrahepatic bileucts were dilated, with air in the biliary tree. Free air wasresent in the right pericolic gutter. Abdominal exploration wasndertaken, but significant inflammation prevented definitive

ntervention. Esophagogastroduodenoscopy revealed a largereen pigmented stone impacted at the junction of the first andecond duodenum (Figure B). Endoscopic electrohydraulic lith-tripsy was performed, excavating a shaft through the center ofhe stone. The stone was fragmented further with an endo-copic dilating balloon, additional endoscopic electrohydraulicithotripsy, and mechanical lithotripsy, and the duodenal ob-truction was relieved completely. A large cholecystoduode-al fistula was seen (supplementary video; see supplementaryaterial online at www.cghjournal.org). A subsequent endo-

copic retrograde cholangiopancreatography revealed an addi-ional bile duct stone that was removed after biliary sphincter-tomy. The patient recovered fully and 5 months later remainssymptomatic.

Bouveret’s syndrome is a rare condition defined as gastric

utlet obstruction by a gallstone. Enterotomy or gastrotomy

ith or without cholecystectomy and fistula repair is the mostommon surgical therapy.

Endoscopy therapy is increasingly common and differentodalities of endoscopic treatment have been reported.1–3 En-

oscopic therapy for large stones such as this one facilitateinimally invasive therapy.

Supplementary DataNote: to access the supplementary material accompa-

ying this article, visit the online version of Clinical Gastroenter-logy and Hepatology at www.cghjournal.org.

References. Huebner ES, DuBois S, Lee SD, et al. Successful endoscopic

treatment of Bouveret’s syndrome with intracorporeal electrohy-draulic lithotripsy. Gastrointest Endosc 2007;66:183–184.

. Polistena A, Santi F, Tiberi R, et al. Endoscopic treatment ofBouveret’s syndrome. Gastrointest Endosc 2007;65:704–706.

. Gemmel C, Weickert U, Eickhoff A, et al. Successful treatment ofgallstone ileus (Bouveret’s syndrome) by using extracorporalshock wave lithotripsy and argon plasma coagulation. GastrointestEndosc 2007;65:173–175.

© 2008 by the AGA Institute1542-3565/08/$34.00

doi:10.1016/j.cgh.2007.12.055

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2008;6:e15

A

A

�

HIUcI�

vwl2

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2008;6:e17

BSTRACTS FROM AROUND THE WORLD

Click on citations to link to these articles and additional papers of interest.

�

p2�

oD�

ci2�

o

dditional Papers of Interest

Sagnelli E, Stroffolini T, Mele A, et al, and the Italianospitals Collaborating Group. Chronic hepatitis B in

taly: new features of an old disease—approaching theniversal prevalence of hepatitis B e antigen-negative

ases and the eradication of hepatitis D infection. Clinnfect Dis 2008;46:110 –113.

Abraides JG, Villanueva C, Bañares R, et al. Hepaticenous pressure gradient and prognosis in patientsith acute variceal bleeding treated with pharmaco-

ogic and endoscopic therapy. J Hepatol 2008;48:229 –

36. t

Del Piano M, Ballarè M, Carmagnola S, et al. DPEJlacement in cases of PEG insertion failure. Dig Liv Dis008:40:140 –143.

Biagi F, Campanella J, Bianchi PI, et al. The incidencef coeliac disease in adult first degree relatives. Dig Livis 2008;40:97–100.

Veijola L, Oksanen A, Linnala A, et al. Persistinghronic gastritis and elevated Helicobacter pylori antibod-es after successful eradication therapy. Helicobacter007;12:605– 608.

Allen KJ, Gurrin LC, Constantine CC, et al. Iron-verload-related disease in HFE hereditary hemochroma-

osis. N Engl J Med 2008;358:221–230.

C

M

CTETc

FIiht

IC.

OUc

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2008;6:261–262

ME Activities–Exams 1 and 2

iguel R. Arguedas, MD, Editor, CME Section

ME Credits:he American Gastroenterological Association (AGA) Institute is accredited by the Accreditation Council for Continuing Medicalducation to provide continuing medical education for physicians.he AGA Institute designates this educational activity for a maximum of 1.0 AMA PRA Category 1 Credit(s)™. Physicians should onlylaim credit commensurate with the extent of their participation in the activity.

aculty Disclosure:n accordance with the Accreditation Council for Continuing Medical Education’s Standards for Commercial Support of Continu-ng Medical Education, all faculty and planning partners must disclose any relevant financial relationship(s) or other relationship(s)eld within the past 12 months. The AGA Institute implements a mechanism to identify and resolve all conflicts of interest prioro delivering the educational activity to learners.

nstructions:ategory 1 credit can be earned by reading the relevant articles and taking these CME examinations online at http://www.cghjournal

org/content/cme. Answers can be obtained online after completing the exam(s).

bjectives:pon completion of these activities, participants should be able to demonstrate an increase in or affirmation of their knowledge of

linical medicine and evaluate the appropriateness of the clinical information as it applies to the provision of patient care.

EoH

T

Q

p

Q

p

Q

i

Q

p

262 CME ACTIVITIES CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

xam 1: Pancreatic Cancer: A Review of the Evidencen Causationart AR, et al, Authors

est ID No.: 0053 Contact hours: 1.0 Expiration Date: March 31, 2009

uestion 1:

b

Smoking accounts for approximately what percentage of

ancreatic cancer cases?

d

uestion 2:

b

d

uestion 3:

b

d

uestion 4:

b

d

a. 10%. 25%

c. 50%

. 75%

All of the following risk factors and exposures have beenositively associated with pancreatic cancer, except:

a. High folate consumption. History of cholecystectomy

c. Helicobacter pylori infection

. Asthma

All of the following genetic conditions are associated with anncreased risk of pancreatic cancer, except:

a. Familial adenomatous polyposis. Ataxia-telangiectasia

c. Peutz-Jeghers syndrome

. Tylosis

Which of the following statements regarding Helicobacterylori and pancreatic cancer is correct?

a. H pylori-induced gastritis reduces absorption of vitamin C.. H pylori infection results in decreased levels of gastrin.

c. H pylori infection results in decreased levels of secretin.

. H pylori-induced increase in IL-1 levels.

EDEH

T

Q

we

Q

Q

pf

Q

oa

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2008;6:263

xam 2: What Is the Role of Serological Testing in Celiacisease? A Prospective, Biopsy-Confirmed Study Withconomic Analysis

opper AD, et al, Authors

est ID No.: 0054 Contact hours: 1.0 Expiration Date: March 31, 2009

uestion 1:

b

According to the present study, which strategy is associated

ith the highest yield of finding a case of celiac disease onndoscopy?

d

uestion 2:

b

d

uestion 3:

b

d

uestion 4:

b

d

a. tTG-positive. EMA-positive

c. Either tTG or EMA-positive

. TTG-positive, followed by EMA-positive

A high TTG titer is seen in the following conditions, except:

a. SLE. Advanced heart failure

c. Diabetes mellitus

. Chronic liver disease

According to the results by Hopper et al, what percentage ofatients on a gluten-free diet and with villous atrophy onollow-up biopsies have tTG levels � 15U/mL?

a. �10%. �20%

c. �40%

. �80%

According to the authors, which factor may affect the resultsf studies that measure the sensitivity and specificity of tTGnd EMA, except?

a. Prevalence of celiac disease. Definition and ascertainment of the “gold standard”

c. Study design (prospective versus retrospective)

. The use of automated assays

A

�

aMPilC

rsowtwfvprhanndAvw

ofritaw1totcv

�

fIum

m

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2008;6:264–265

BSTRACTS FROM AROUND THE WORLD

Visit CGH online at www.cghjournal.org to link to these articles and additional papers of interest.

aLsJdtdtp

tdrefTids

�

AHLaB

bidhsb1dCcat

patbefi

Your Mom Was Right—Eat Your Fruitsnd Vegetablesillen AE, Subar AF, Graubard BI, et al, and Ziegler RG for the

rostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screen-ng Trial Project Team. Fruit and vegetable intake and preva-ence of colorectal adenoma in a cancer screening trial. Am Jlin Nutr 2007;86:1754 –1764.

Summary. There continues to be intense interest in theelationship between diet and colorectal neoplasia. Thistudy uses data from the prostate, lung, colorectal, andvarian cancer screening trial (PLCO) where 3057 patientsith prevalent adenoma were compared with 29,413 con-

rols. All patients underwent sigmoidoscopy and if a polypas found a full colonoscopy was recommended. A food

requency questionnaire was used to quantify fruit andegetable intake 12 months before study entry. When com-aring the lowest versus highest quintile of fruit intake, theisk of distal adenoma was less statistically significant in theighest quintile. An inverse association between adenomand total fruit intake was also detected, regardless of ade-oma number and pathology. Total vegetable intake wasot different based on adenoma although there were someifferences based on the presence of specific vegetable types.diet rich in fruit and deep yellow vegetables, dark green

egetables, and onions and garlic are modestly associatedith a reduced risk of colorectal adenoma.

Editor’s comment. As with all studies examining diet,ne has to question the amount of food source necessaryor consumption to achieve the end point to see if it isealistic. Additionally, how long has the food stuff beenngested prior to the determination of the endpoint? Forhe highest quintile, patients reported approximately 5 pyr-mid servings per day. This amount would likely far exceedhat most of us consume. It seems difficult to believe thatyear of such a diet prevents adenomas; it is more likely

hat such a diet, and perhaps other healthy life styles, werengoing for some time. Nevertheless, given the other po-ential health benefits of such a diet, even if you don’tonsume these food stuffs in the highest quintile and pre-ent an adenoma, maybe your mother is still correct.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Capsule Endoscopy Is Not the Answeror Surveillance in FAPaquinto G, Fornasarig M, Quaia M, et al. Capsule endoscopy isseful and safe for small-bowel surveillance in familial adeno-atous polyposis. Gastrointest Endosc 2008;67:61– 67.

Summary. Screening of the small bowel is recom-

ended in familial adenomatous polyposis (FAP) to detect i

nd remove adenomas which may progress to carcinoma.ittle is known about the yield of capsule endoscopy (CE) increening. Twenty-three patients with FAP underwent CE.ejunal / ileal polyps were found in 7 patients (30%); duo-enal polyps, however, were only seen in 4 of 11 (36%) andhe ampulla of vater was not found. Proximal jejunal polypsetected by CE were confirmed and removed by push en-eroscopy (PE) in 4 of 23 (23%) patients, and all removedolyps were tubular adenomas. No complications arose.

Editor’s comment. In FAP, duodenal adenomas, par-icularly those involving the ampulla are at highest risk forevelopment of carcinoma. Because these patients have al-eady undergone colectomy, the distal small bowel can beasily surveyed at the time of surveillance ileoscopy. There-ore, the role for capsule endoscopy is likely limited in FAP.his study, while showing that more distal polyps can be

dentified, supports the continued use of side viewing en-oscopy to examine the ampulla and push enteroscopy tourvey for proximal jejunal lesions.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Take Your Probiotic Before Yourntibioticickson M, D’Souza AL, Muthu N, et al. Use of probiotic

actobacillus preparation to prevent diarrhea associated withntibiotics: randomized double blind placebo controlled trial.MJ 2007;335:80 – 85.

Summary. Diarrhea is a common accompaniment ofroad spectrum antibiotic use leading to potential morbid-

ty and precipitation of Clostridium difficile colitis. This ran-omized double-blind placebo-controlled trial studied 135ospitalized patients receiving antibiotics. Patients con-umed a 100 g cocktail containing Lactobacillus casei, Lacto-acillus bulgaricus, and Streptococcus thermophilus twice daily for

week. Twelve percent of the probiotic group developediarrhea compared with 34% in the placebo group (P�.007).difficile diarrhea was identified in 17% of the placebo group

ompared with 0 in the probiotic group (P�.001) with anbsolute risk reduction of 17% and the number needed toreat of 6.

Editor’s comment. This well done trial suggests that arobiotic cocktail taken with broad-spectrum antibiotics inhospital setting not only may reduce diarrhea but poten-

ially decreases the likelihood of C difficile diarrhea which hasecome epidemic and more difficult to treat in the hospitalnvironment. Further study is warranted to confirm thesendings, given the widespread implications. Until such data

s available, it may be reasonable to adopt such a strategy.

�

HCGhI

LSc

Lav1

etpctpfspdeHardlHcotHbdtlqt

dtrmimrIHrt

�

TGMdc

rnpRg3a(SdlPI2

awoawmwdrtTcittct

A

�

te�

fc1�

c

March 2008 ABSTRACTS FROM AROUND THE WORLD 265

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Our Expanding Arsenal Againstepatitis B Virus Infection

han HLY, Heathcote J, Marcellin P, et al, and The 018 Studyroup. Treatment of hepatitis B e antigen-positive chronicepatitis with telbivudine or adefovir. A randomized trial. Ann

ntern Med 2007;147:745–754.

ai C–L, Gane E, Liaw Y–F, et al, and Brown NA for the Globetudy Group. Telbivudine versus lamivudine in patients withhronic hepatitis B. N Engl J Med 2007;357:2576 –2588.

ampertico P, Viganò M, Manenti E, et al. Low resistance todefovir combined with lamivudine: a 3-year study of 145 lami-udine-resistant hepatitis B patients. Gastroenterology 2007;33:1445–1451.

Summary. Hepatitis B virus (HBV) infection remains anxtraordinarily important public health issue worldwide. For-unately, our therapeutic armamentarium continues to ex-and, underscored by these recent publications. Chan et alompared 2 newer agents and showed that at 24 weeks ofreatment, telbivudine had more consistent HBV DNA sup-ression than adefovir. At 52 weeks, patients who switchedrom adefovir to telbivudine also had a greater HBV DNAuppression; at week 24, 3 times as many telbivudine treatedatients were HBV PCR negative. Lai et al compared telbivu-ine to lamivudine in chronic HBV infection with the primaryndpoint in this noninferiority trial of a reduction in serumBV DNA levels to fewer than 5 log 10 copies/mL. At 52 weeks,significantly higher proportion of HBeAg-positive patients

eceiving telbivudine, compared with those receiving lamivu-ine, had a therapeutic response (75% vs 67%) as well as histo-

ogic response (65% vs 56%). The mean time required for serumBV DNA to become undetectable by PCR was also signifi-

antly shorter in the telbivudine group, and overall, 60% vs 40%f the HBeAg-positive patients became PCR negative. Impor-antly, resistance developed in 5%, compared with 2.3% of theBeAg-positive and HBeAg-negative patients who received tel-

ivudine, compared with 11% and 10.7% who received lamivu-ine, respectively. Lastly, the report by Lampertico et al, suggesthat for those receiving long-term combination adefovir andamivudine, lamivudine resistant hepatitis B resistance is infre-uent (4%) and the prevention of virologic and clinical break-hrough was durable.

Editor’s comment. We continue to learn more aboutrug resistance with HBV treatment as we advance ourherapies for HBV now using well tolerated oral agents. Theesults using telbivudine are encouraging and suggest that

ore widespread use of these agents will make a significantmpact in the treatment of HBV infection. Unfortunately

any in the third world will not have the opportunity toeceive these drugs nor may recognize they are infected.ndeed there are many similarities between the HIV andBV epidemics. Further public health measures are war-

anted, and for diagnosis and prevention of HBV infection,

he use of these drugs should be more rapidly expanded. A

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Forget the Cookbook: Individualizereatment Duration for Hepatitis Cenotype 1 Patientsangia A, Minerva N, Bacca D, et al. Individualized treatment

uration for hepatitis C genotype 1 patients: a randomizedontrolled trial. Hepatology 2008;47:43–50.

Summary. Interferon and ribavirin are prescribed cur-ently as a standard 48-week treatment for HCV virus ge-otype 1 patients. This prospective randomized trial com-ared this standard regimen for 24, 48, or 72 weeks if HCVNA was negative at weeks 4, 8 or 12, respectively (variableroup). In the standard treatment group, 87.1%, 70.3%, and8.1% of patients who first achieved undetectable HCV RNAt 4, 8 or 12 weeks attained sustained virologic responseSVR), respectively. In the variable group, correspondingVR rates were 77.2%, 71.9%, and 63.5%. Independent pre-ictors of rapid virologic response (RVR) were HCV RNA

evels � 400,000 IU/mL and absence of advanced fibrosis.atients with this RVR and baseline viremia � 400,000U/mL achieved higher SVR when treated for 48 rather than4 weeks (86.8 vs 73.1, P �.14).

Editor’s comment. This pragmatic study assesses ther-py based on early response by persistent viremia. If thereas persistent viremia (albeit at low levels) at 12 weeks, andne persisted with treatment to 48 weeks, an SVR still waschievable. Conversely, if there was a � 2-log decline at 12eeks, patients were considered nonresponders and treat-ent was discontinued. These results are similar to thoseho were viremic at 24 weeks. Using a patient directeduration of therapy, not only could one achieve better cureates with longer duration of therapy, one could discon-inue patients’ treatment earlier given persistent viremia.hus, approximately a quarter of HCV genotype 1 patientsould be cured by therapy in only 24 weeks and an approx-mately comparable rate of patients may require extendedreatment for 72 weeks to enhance SVR. Such a tailoredherapy would provide clear benefit for patients given theost, complexity, and potential side effects of current HCVherapy.

dditional Papers of Interest

Jeong SH, Lee DJ, Kim YB, et al. Diagnostic value oferminal ileum intubation during colonoscopy. J Gastro-nterol Hepatol 2008;23:51–55.

Salem T, Molloy R, O’Dwyer P. Prospective, five-yearollow-up study of patients with symptomatic uncompli-ated diverticular disease. Dis Colon Rectum 2007;50:460 –1464.

Perry CD, Hutter MM, Smith DB, et al. Survival andhanges in comorbidities after bariatric surgery. Am J

nn Surg 2008;247:21–27.

E

PoC

Pirottampsf

palisnfiel

yPtoida

lBbwscpipismgpa

ie4wp

ditorial

cfficp

hutwpdsTfsagg

shtpp

onsrPvraltwTrpabuaPePsmstg

nlT

hotodynamic Therapy: Standardf Care for Palliation ofholangiocarcinoma?

alliation of malignant biliary obstruction in patients withunresectable hilar cholangiocarcinoma (CCA) is challeng-

ng for surgeons, interventional radiologists, endoscopists, andadiation oncologists alike. Compared with distal bile ductbstruction where palliation is relatively easily achieved opera-ively and non-operatively, hilar CCA poses particular difficul-ies. The tumors involve bifurcation and spread progressivelynd inexorably in a proximal direction such that eventually tooany branches of the intrahepatics are involved to permit stent

lacement to be effective. Indeed it, was Klatskin1 who de-cribed that these patients die of liver failure and cholangitisrom biliary obstruction.

Photodynamic therapy (PDT) is an ablative treatment forremalignant and malignant lesions. A photosensitizing drug isdministered followed by application of a specific wavelength ofight leading to intracellular activation of the drug and cellularnjury.2 In addition, thrombosis of vessels and immune re-ponse may lead to tumor destruction.3 The wavelength of lightecessary to cause activation is 630 nm and is provided by laserbers that emit light of that wavelength. When used in thesophagus, the fibers are easily placed into the esophagealumen because the endoscope is in a straightened position.

The first report of PDT for CCA was published more than 15ears ago.4 Since then there have been numerous studies usingDT for CCA — both basic science5–7 and clinical. It is impor-ant to note that the patient selection, primary outcomes, typesf fibers and fiber delivery vary between studies. It is also of

mportance to note that the passage of these fibers through auodenoscope poses special challenges, particularly in passageround the tight angle in the duodenum toward the bile duct.

The landmark clinical study using PDT for CCA was pub-ished by Ortner et al, nearly 10 years ago.8 Patients withismuth type III and IV CCA who did not have a fall inilirubin levels of at least 50% after bilateral stent placementere treated with PDT. Laser light was delivered cholangio-

copically using small caliber fibers. After treatment, a signifi-ant fall in bilirubin levels was achieved, and a sustained im-rovement in indices of quality of life was noted. The same lead

nvestigator then performed a randomized prospective trial oflastic biliary stents alone versus plastic biliary stents and PDT

n a similar patient population that failed to have a fall inerum bilirubin levels following bilateral biliary stent place-

ent.9 A significant increase in survival was seen in the PDTroup (median survival 493 days vs 98 days). PDT also im-roved biliary drainage and quality of life over stent placementlone.

Other groups have published similar results.10 –14 These stud-es from outside the US have used thin, flexible, 400-�m diam-ter fibers. In addition, in some series, patients were re-treated– 6 weeks later because of advanced Bismuth type IV tumorsith residual segmental duct occlusions/stenoses or tumor-

ositive biopsies at a 1-month follow-up examination.10 e

There are limited publications using PDT for CCA fromenters within the US. Indeed, nearly all reports have comerom 1 center.15–17 This group uses standard FDA-approvedbers designed for use in the esophagus without the use ofholangioscopy. Fiber breakage occurs in about one third ofatients.

In this issue of Clinical Gastroenterology and Hepatology, Ka-aleh et al,18 from Virginia publish their results using PDT fornresectable cholangiocarcinoma. Forty-eight patients werereated over a 5-year period. Of these, plastic biliary stents aloneere placed in 29 patients; PDT and plastic biliary stents werelaced in 19 patients. Laser delivery was achieved using stan-ard rigid 2.5-cm fibers that were preloaded into a 10F deliveryystem. One or 2 biliary segments were treated per session.herapy was repeated every 3 months until death or withdrawal

rom the study. The degree of decline in bilirubin levels wasimilar in the 2 groups. Kaplan–Meier survival analysis showed

statistically significant prolongation in survival in the PDTroup (mean 16.2 � 2.4 months) compared with the stent onlyroup (mean 7.4 � 1.6 months).

The study, while one of the first direct comparative survivaltudies from the US to show an improved survival using PDT,as a number of flaws. It is a small, retrospective study. Al-hough the majority of patients had Bismuth III and IV lesions,atients with Bismuth I and II lesions were included. Someatients in each group received chemoradiation therapy.

With so many options available for the palliation of biliarybstruction in patients with inoperable hilar cholangiocarci-oma — chemoradiation, brachytherapy, plastic and metaltents (endoscopically or percutaneously placed) — what can weecommend for these patients? Is there enough data to say thatDT should be given to all patients? Let’s start with the disad-antages of PDT. The treatment is not available at all centers. Itequires expertise in both endoscopy and photodynamic ther-py. The procedure is time consuming and can be quite pro-onged based on number of segments treated; suffice it to sayhat one should allot at least 90 minutes for the procedurehich entails stent removal, treatment, and stent replacement.he fibers available in the US are suboptimal for endoscopic

etrograde cholangiopancreatography use. They are stiff androne to breakage. Because of the stiffness, treatment is gener-lly limited to the main hepatic ducts since the fiber does notend around corners to reach intrahepatic branches. Finally, lets not forget the photosensitivity that occurs for 4 – 6 weeksfter therapy which may limit quality of life. The advantages ofDT are that it is reasonably well tolerated and seems to beffective. It can be repeated without a ceiling dosage effect.10

DT is the only treatment to date where there is evidence toupport an improvement in survival over plastic stent place-

ent alone for advanced CCA. In fact, one retrospective studyuggested that the survival with PDT and stents was similar tohose who underwent attempted curative, but incomplete sur-ical resection.19

Many questions about the role of PDT and CCA remain. Inearly all studies plastic biliary stents have been used. There is

ittle experience with self expandable metal stents and PDT.13

he downside to metal stent placement is that light may not

ffectively reach the tumor through the interstices of the stent

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2008;6:266–267

aqmpbbnPtiulWttss

sdltpramwriv

cqfacttatt

1

1

1

1

1

1

1

1

1

1

March 2008 267

nd/or be scattered during follow-up PDT, leading to inade-uate treatment. Some authors, however, routinely combineetal stent placement and PDT, though it requires 10 cm

arallel stents placed into the right and left systems followingaseline PDT to have access to both sides in the future ifilateral PDT is repeated (Richard Kozarek, personal commu-ication, November 2007). Most studies have used bilateralDT and stent placement. Is this necessary? What about pa-ients who have parenchymal atrophy on 1 side? Are we expos-ng those patients to a higher risk of cholangitis as compared tonilateral treatment? Do only patients with Bismuth III and IV

esions benefit? Should type I and II patients be offered PDT?hat are the effects of other treatments such as chemoradia-

ion — complementary or antagonistic? Does it make sense toreat patients with large masses when the treatment is relativelyuperficial? Finally, is cholangioscopic delivery of the laser lightuperior to fluoroscopic delivery?

In addition to time constraints and reimbursement, there areeveral reasons PDT availability in the US is limited. For en-oscopists in the US to embrace PDT, smaller and more flexible

aser fibers are needed. Furthermore, better delivery methods forhe fibers are also needed. For patients to embrace PDT, betterhotosensitizing agents are needed. Agents with a shorter du-ation of phototoxicity and more rapid onset (which wouldllow patients to receive the photosensitizing agent and treat-ent on the same day, rather than 48 hours apart) would beelcomed. Unfortunately, the number of patients with CCA is

elatively small such that the commercial benefit to companiess limited and thus there is not a financial incentive for ad-ancement in this area.

So, should photodynamic therapy be considered standard ofare for palliation of cholangiocarcinoma? The answer is aualified yes. The data suggest that PDT is an excellent optionor patients with unresectable CCA, especially for Bismuth IIInd IV lesions. However, there are no comparative trials withhemoradiation, and PDT availability is limited. Therefore,hese patients could be managed with standard palliative care atheir institution or referred to a specialized center with PDTvailability. Further comparative trials are needed to determinehe optimal regimen for palliation of obstructive jaundice inhese patients.

TODD H. BARON, MDDivision of Gastroenterology and Hepatology

Mayo ClinicRochester, Minnesota

References

1. Klatskin G. Adenocarcinoma of the hepatic duct at its bifurcationwithin the porta hepatitis. An unusual tumor with distinctive clin-ical and pathological features. Am J Med 1965;38:241–256.

2. Petersen BT, Chuttani R, Croffie J, et al. Photodynamic therapy forgastrointestinal disease. Gastrointest Endosc 2006;63:927–932.

3. Ortner MA, Dorta G. Technology insight: photodynamic therapy for

cholangiocarcinoma. Nat Clin Pract Gastroenterol Hepatol2006;3:459–467.

4. McCaughan JS Jr, Mertens BF, Cho C, et al. Photodynamic ther-apy to treat tumors of the extrahepatic biliary ducts: a casereport. Arch Surg 1991;126:111–113.

5. Wong Kee Song LM, Wang KK, Zinsmeister AR. Mono-L-aspartylchlorin e6 (NPe6) and hematoporphyrin derivative (HpD) in pho-todynamic therapy administered to a human cholangiocarcinomamodel. Cancer 1998;82:421–427.

6. Kiesslich T, Berlanda J, Plaetzer K, et al. Comparative character-ization of the efficiency and cellular pharmacokinetics of Foscan-and Foslip-based photodynamic treatment in human biliary tractcancer cell lines. Photochem Photobiol Sci 2007;6:619–627.

7. Oertel M, Schastak SI, Tannapfel A, et al. Novel bacteriochlorinefor high tissue-penetration: photodynamic properties in humanbiliary tract cancer cells in vitro and in a mouse tumour model. JPhotochem Photobiol B 2003 15;71:1–10.

8. Ortner MA, Liebetruth J, Schreiber S, et al. Photodynamic therapyof nonresectable cholangiocarcinoma. Gastroenterology 1998;114:536–542.

9. Ortner ME, Caca K, Berr F, et al. Successful photodynamic ther-apy for nonresectable cholangiocarcinoma: a randomized pro-spective study. Gastroenterology 2003;125:1355–1363.

0. Berr F, Wiedmann M, Tannapfel A, et al. Photodynamic therapyfor advanced bile duct cancer: evidence for improved palliationand extended survival. Hepatology 2000;31:291–298.

1. Wiedmann M, Berr F, Schiefke I, et al. Photodynamic therapy inpatients with non-resectable hilar cholangiocarcinoma: 5-yearfollow-up of a prospective phase II study. Gastrointest Endosc2004;60:68–75.

2. Zoepf T, Jakobs R, Arnold JC, et al. Palliation of nonresectablebile duct cancer: improved survival after photodynamic therapy.Am J Gastroenterol 2005;100:2426–2430.

3. Dumoulin FL, Gerhardt T, Fuchs S, et al. Phase II study ofphotodynamic therapy and metal stent as palliative treatment fornonresectable hilar cholangiocarcinoma. Gastrointest Endosc2003;57:860–867.

4. Zoepf T, Jakobs R, Arnold JC, et al. Photodynamic therapy forpalliation of nonresectable bile duct cancer—preliminary resultswith a new diode laser system. Am J Gastroenterol 2001;96:2093–2097.

5. Rumalla A, Baron TH, Wang KK, et al. Endoscopic application ofphotodynamic therapy for cholangiocarcinoma. Gastrointest En-dosc 2001;53:500–504.

6. Harewood GC, Baron TH, Rumalla A, et al. Pilot study to assesspatient outcomes following endoscopic application of photody-namic therapy for advanced cholangiocarcinoma. J GastroenterolHepatol 2005 Mar;20:415–420.

7. Prasad GA, Wang KK, Baron TH, et al. Factors associated withincreased survival after photodynamic therapy for cholangiocar-cinoma. Clin Gastroenterol Hepatol 2007;5:743–748.

8. Kahaleh M, Mishra R, Shami VM, et al. Unresectable cholangio-carcinoma: Comparison of Survival in Biliary Stenting alone vsStenting with Photodynamic Therapy. Clin Gatsroenterol Hepatol2008;6:290–297.

9. Witzigmann H, Berr F, Ringel U, et al. Surgical and palliativemanagement and outcome in 184 patients with hilar cholangio-carcinoma: palliative photodynamic therapy plus stenting is com-parable to r1/r2 resection. Ann Surg 2006;244:230–239.

doi:10.1016/j.cgh.2008.01.015

R

CV

E

*MM

LcdtahoHaccbwnpcstainlvlttaw

IBaphntiht

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2008;6:268–274

EVIEWS

hronic Hepatitis B: Preventing, Detecting, and Managingiral Resistance

MMET B. KEEFFE,* DOUGLAS T. DIETERICH,‡ JEAN–MICHEL PAWLOTSKY,§,� and YVES BENHAMOU¶

Division of Gastroenterology and Hepatology, Department of Medicine, Stanford University School of Medicine, Stanford, California; ‡Department of Medicine, The§

ount Sinai Medical Center, New York, New York; French National Reference Center for Viral Hepatitis B, C and Delta and Virology Laboratory, Hôpital Henri

ondor, Université Paris 12, Créteil, France; �INSERM U841, Créteil, France; and the ¶Service d’Hépatologie Hôpital Pitié-Salpêtrière, Université Paris 5, Paris, France

evcl

aahwtwiHmecurpcstr

3oscgpNrt

V

thfi

See Deng G et al on page 716 for companionarticle in the March 2008 issue ofGastroenterology.

icensed oral agents for antiviral therapy in patients withhronic hepatitis B virus (HBV) infection include lamivu-ine, adefovir, entecavir, and telbivudine. Emtricitabine,

enofovir, and the combination of tenofovir plus emtricit-bine in 1 tablet, which are licensed for the treatment ofuman immunodeficiency virus infection, are additionalff-label options for treating HBV infection. PreventingBV antiviral drug resistance to nucleoside/nucleotide an-

logues and appropriate management when resistance oc-urs has become a major focus in the management ofhronic hepatitis B. HBV antiviral drug resistance may beest prevented by using an agent or combination of agentsith a high genetic barrier to resistance, and 2 potentucleoside and nucleotide drugs with different resistancerofiles may prove to be the optimal first-line treatment forhronic hepatitis B. Frequent assessment of quantitativeerum HBV DNA remains the best approach to early detec-ion of resistance, and antiviral therapy should be modifieds soon as resistance is detected. Results from several clin-cal trials have shown that the addition or substitution ofewer antiviral agents can restore suppression of viral rep-

ication, normalize alanine aminotransferase levels, and re-erse histologic progression in patients with resistance toamivudine, but little information exists regarding the long-erm benefits of second-line treatment regimens. Despitehe substantial advances in treatment made to date, newgents with novel viral targets will be needed for patientsho ultimately may fail second- or third-line therapy.

t is estimated that 1.25 million people in the United Stateshave chronic hepatitis B, and there were 60,000 new hepatitis

virus (HBV) infections in 2004.1 Acute HBV infection may besymptomatic, result in self-limited or fulminant hepatitis, orrogress to chronic hepatitis, which can lead to cirrhosis orepatocellular carcinoma.2 Cirrhosis and hepatocellular carci-oma result in about 1 million deaths worldwide each year.3 Inhe United States, there has been a substantial decline in thencidence of acute hepatitis B over the past 15 to 20 years thatas coincided with the national strategy to eliminate HBV

ransmission by vaccination and public health measures.4 Nev-

rtheless, because of a high HBV prevalence rate in many de-eloping countries and population migrations to developedountries, chronic HBV infection is an important health prob-em in the United States and Europe.

Antiviral therapy in patients with chronic hepatitis B is associ-ted with improved outcomes.5,6 Over the past 10 years, oralntiviral agents available for patients with chronic HBV infectionave included lamivudine, adefovir, entecavir, and telbivudine,hich are licensed to treat chronic hepatitis B, and emtricitabine,

enofovir, and the combination of tenofovir plus emtricitabine,hich has been licensed for treatment of patients with human

mmunodeficiency virus (HIV) infection, but also is active againstBV infection.7 However, 20 years of experience with the treat-ent of HIV, as well as current HBV resistance data, point to the

mergence of HBV antiviral drug resistance,8 as illustrated by thease of lamivudine, the nucleoside that has the longest history ofse in patients with chronic HBV infection and the highest rate ofesistance. The development of resistance is associated with aoorer long-term prognosis.9,10 Clinicians need to be aware ofurrent approaches to resistance testing and of effective treatmenttrategies to minimize the emergence of resistant HBV strains, andhey should be cognizant of regimens that may be effective afteresistance has emerged.

Viral Replication and Selection ofResistant StrainsThe genome of HBV consists of partially double-stranded,

.2-kb, covalently closed, circular DNA (ccc DNA) comprising 4verlapping open reading frames.11 The viral genome is tran-cribed into 4 major subgenomic viral messenger RNAs, under theontrol of specific enhancers,12 and is the template for the pre-enomic RNA. The virus is encapsidated after binding of theolymerase and core to the pregenomic RNA in the cytoplasm.ucleocapsids are enveloped by budding into the endoplasmic

eticulum, after which they are secreted from the cell or return tohe nucleus to amplify the cccDNA reservoir.13

Viral mutations occur spontaneously during HBV replication.iral reverse transcriptases intrinsically are error prone and lack a

Abbreviations used in this paper: AASLD, American Association forhe Study of Liver Diseases; ALT, alanine aminotransferase; HBeAg,epatitis B e antigen; HBV, hepatitis B virus; HIV, human immunode-ciency virus.

© 2008 by the AGA Institute1542-3565/08/$34.00

doi:10.1016/j.cgh.2007.12.043

pTvdnipaaptw

tgoaermt

aHornc

t(aga

ouaa

aHMmrdv

ccd7lts

pieavthbavic2rbc

tlivcolooeca

Fsstt

March 2008 CHRONIC HEPATITIS B AND ANTIVIRAL RESISTANCE 269

roofreading function, allowing for replication errors to occur.hese replication errors result in the emergence of multiple HBVariant quasispecies that coexist and reach population densities inirect proportion to their relative replication fitness.11 This phe-omenon is responsible for the generation of significant diversity;

t has been shown that HBV genomes in one given patient dis-layed a rate of 1.4 to 3.2 � 10�5 nucleoside substitutions per year,value approximately 104 times greater than DNA genomes and

bout 10�2 less than that for HIV.14 A chronic HBV carrier canroduce up to 1013 virions per day, and, as a result, every nucleo-ide of the 3.2-kb HBV genome theoretically can be substitutedithin one patient every day.2

The dominant quasispecies is by definition the best adaptedo its host environment, and, as expected, random mutationsenerally will impair its fitness to a degree. Thus, the emergencef successful HBV variants resistant to an antiviral drug isffected by the mutation rate and viral load and is determinedventually by the replication fitness of the mutant virus inelation to the antiviral potency of the drug and the number of

utations required to confer resistance (ie, the genetic barriero resistance).15

Mutations Associated With Resistanceto Nucleoside/Nucleotide AnaloguesDefining the Consequences of ResistanceTreatment failure can be defined as primary or second-

ry. Primary treatment failure is the failure of a drug to reduceBV DNA levels by 1 � log10 IU/mL or greater within 3 months

f initiation of therapy, and secondary failure is defined as aebound of HBV replication by 1 � log10 IU/mL or greater fromadir in patients in whom treatment initially produced a de-rease in serum HBV DNA of 1 � log10 IU/mL or greater.16

Secondary treatment failure due to resistance to antiviralherapy can result in a decreased rate of hepatitis B e antigenHBeAg) seroconversion17; reversion of virologic, biochemical,nd histologic improvement18,19; increased rate of disease pro-ression5; severe exacerbations in the presence of cirrhosis20,21;nd risk for graft loss and death after liver transplantation.22

igure 1. Location of the major antiviral drug-resistant mutations as-ociated with LMV, LdT, ADV, TDF, and ETV11 (Reprinted with permis-ion). LMV, lamivudine; Ldt, telbuvidine; ADV, adefovir dipivoxil; TDF,enofovir DF; ETV, entecavir. POL/RT, polymerase reverse-transcrip-

aase.

Mutations in the Polymerase Gene andAntiviral ResistanceMutations resulting in resistance to nucleoside/nucle-

tide analogues mainly involve the viral polymerase gene (Fig-re 1).11,23–25 This gene contains 7 functional domains (A–G),nd mutations that give rise to nucleoside/nucleotide resistancere located essentially in domains A through E.23

Lamivudine ResistanceLamivudine inhibits viral reverse-transcriptase activity

s a competitive inhibitor of deoxycytidine triphosphate.23

igh-level lamivudine resistance results from M204V and204I mutations in the C domain.26 HBV variants with theseutations display reduced reverse-transcriptase activity and

eplication capacity, but compensatory mutations in the Bomain (V173L and L180M) restore the replication capacity ofirus with the M204V or M204I mutation.27,28

Results from 998 patients with HBeAg-positive compensatedhronic hepatitis B who took lamivudine for up to 6 years indi-ated that the proportion of patients with documented lamivu-ine-resistant mutations increased from 23% in year 1 to 43%, 55%,1%, and 65% in years 2 through 5, respectively. Patients with

amivudine-resistant mutations who were followed up for morehan 4 years had significantly higher risk for hepatic decompen-ation and liver disease–related severe adverse events.9

Adefovir ResistanceAdefovir inhibits priming of reverse transcription by

reventing the incorporation of deoxyadenosine triphosphatento the viral primer and inhibits viral-minus strand DNAlongation.29 Two mutations have been described that conferpproximately 5- to 10-fold reduced susceptibility to adefovir initro: N236T in the D domain of viral polymerase and A181V inhe B domain.23 Most recently, another rare mutation, I233V,as been suggested to result in primary resistance to adefovir,ut that mutation has not been confirmed to be associated withreduced susceptibility to adefovir in vitro.30 A study of adefo-

ir in 125 patients who were followed up for up to 240 weeksndicated that the cumulative probabilities of mutations asso-iated with a virologic breakthrough at 48, 96, 144, 192, and40 weeks were 0%, 3%, 11%, 18%, and 29%, respectively. Theespective values for mutations associated with virologic andiochemical breakthrough (alanine aminotransferase [ALT] in-reases) were 0%, 2%, 6%, 10%, and 11%.31,32

Entecavir ResistanceEntecavir is a deoxyguanosine analogue that is greater

han 100-fold more potent against HBV in culture than eitheramivudine or adefovir, and halts HBV DNA elongation afterncorporation of a few bases more than lamivudine or adefo-ir.33 Analysis of results from 673 patients treated with ente-avir indicated that 3% showed virologic rebound by 96 weeksf treatment. Three entecavir rebounds were attributable to

amivudine-resistant virus present at baseline, and none of thethers was associated with either entecavir genotypic resistancer loss of entecavir susceptibility.34 After administration ofntecavir to patients who were refractory to lamivudine, ente-avir resistance was detected in 1% of patients after 48 weeksnd an additional 9% after 96 weeks. Entecavir resistance was

ssociated with lamivudine-resistance– conferring mutations

prlcpw1

Hrtcp

sfaw4weo

s(Ntvlt

riociltt

gdrrWitc

w

ifiDdimtonh

t1mh6stiAmpbfp

fsar5tuHiidct

FcmmU

270 KEEFFE ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

lus additional changes at T184, S202, and/or M250.35 Recentesults have indicated that the cumulative probability of viro-ogic breakthrough associated with entecavir resistance in nu-leoside-naive patients was 0.8% over 4 years, and that foratients who received prior lamivudine therapy it was 39.5%,ith 1% having resistant virus after 1 year, 10% after 2 years,6% after 3 years, and 15% after 4 years of treatment.36

Telbivudine ResistanceTelbivudine targets the synthesis of positive-strand

BV DNA, which is thought to result in slower emergence ofesistance than lamivudine, which targets negative-strand syn-hesis.37 Results from the GLOBE trial indicated that the 2-yearumulative rate of virologic breakthrough for HBeAg-positiveatients was 21.6%, and for HBeAg-negative patients was 8.6%.38

Emtricitabine ResistanceEmtricitabine inhibits HBV polymerase in essentially the

ame manner as lamivudine, and the M204V mutation that con-ers resistance to lamivudine also results in resistance to emtricit-bine.39,40 Emtricitabine is effective as monotherapy in patientsith chronic HBV infection, reducing HBV DNA levels to less than00 copies/mL in 54% of patients over 48 weeks.39 However, at 48eeks, 9% to 16% of patients treated with 25 to 200 mg/day of

mtricitabine developed resistance-conferring mutations (M204Ir M204V with or without L180M or V173L).41

Tenofovir ResistanceTenofovir is an acyclic nucleotide analogue that retains

ignificant activity against HBV variants with mutationsM204V plus L180M), conferring resistance to lamivudine and

236T that results in resistance to adefovir.42,43 Tenofovir re-ains good activity in patients with lamivudine-resistant HBVariants, either as monotherapy or when added to continuingamivudine treatment.44,45 HBV mutations that confer resis-ance to tenofovir have not yet been identified.

Detecting, Preventing, and ManagingHepatitis B Virus ResistanceClinical evidence, as summarized earlier, indicates that

esistance to antiviral monotherapy: (1) is common, particularlyn patients treated with lamivudine; (2) increases with durationf treatment; and (3) is associated with poorer clinical out-omes. Many new antiviral agents, however, are available fornitial therapy and as second-line treatment for those failingamivudine. It is important that patients with lamivudine resis-ance be treated with agents that are not cross-resistant, such ashe nucleotide analogues adefovir or tenofovir.

The availability of genetic testing for identification of viralenotypes likely to show phenotypic resistance and the fact thatifferent antiviral agents do not have completely overlappingesistance profiles raises 3 questions: When should genotypicesistance testing be performed in patients with HBV infection?

hat approach(es) to initial treatment are least likely to resultn the rapid emergence of resistance? What treatment alterna-ives are most useful in patients who have developed resistance-onferring mutations?

Resistance TestingGenotypic resistance precedes phenotypic resistance

ith virologic breakthrough, which in turn precedes biochem- u

cal and/or clinical breakthrough. Phenotypic resistance is de-ned as virologic breakthrough (ie, an increase in serum HBVNA by �1 log10 above nadir after achieving virologic responseuring continued treatment), or virologic rebound (ie, increase

n serum HBV DNA to �20,000 IU/mL or to above pretreat-ent level after achieving virologic response during continued

reatment). Biochemical relapse is characterized as an increasef ALT to above the upper limit of normal after achievingormalization with treatment and may be associated with aepatitis flare (Figure 2).46,47

Development of resistance can be detected readily by moni-oring serum HBV DNA levels and is indicated by a greater than

log10 increase from the patient’s lowest level confirmed byeasurement on 2 assays. Patients taking lamivudine should

ave HBV DNA levels evaluated with a sensitive assay every 3 tomonths and those being treated with adefovir or entecavir

hould be monitored every 6 months after the first year ofreatment. Patients with advanced liver disease should be mon-tored every 3 months.47 Guidelines published by the Americanssociation for the Study of Liver Diseases (AASLD) recom-end serum HBV DNA monitoring every 3 to 6 months for

atients receiving therapy for chronic hepatitis B.46 It also haseen noted that monitoring the emergence of resistance-con-erring mutations may be the most sensitive way to assessatients who remain viremic on current treatments.48

Assays for Detection of Resistance-ConferringMutationsThe reference method for detection of resistance-con-

erring mutations is population-based sequencing (ie, a directequence analysis of the HBV polymerase gene). Line-probessays using probes for individual mutations are specific andeproducible and can detect a mutant representing as little as% of the viral population. However, this approach is limited inhat it can detect only known mutations. It requires periodicpdating with new probes specific to novel mutations becauseBV strains resistant to newly developed antiviral drugs are

solated and characterized.49,50 Other powerful technologies aren development, including high-throughput systems capable ofetecting polymorphisms in the entire HBV genome using genehip technology.51 This method is limited by the fact that it,oo, can detect only known polymorphisms and needs to be

igure 2. Serial changes in serum HBV DNA and ALT levels in asso-iation with emergence of antiviral-resistant HBV mutants. The firstanifestation of resistance is the detection of resistance-conferringutations (ie, genotypic resistance).46 (Reprinted with permission).LN, upper limit of normal.

pdated as resistance-conferring mutations are identified.50

fdbttm

losmwnirtua

estnrhsdilmit

tcofHldti

Kpemt

tdommpArha

ltpssoupvptHr

pcrccls

T

A udine

March 2008 CHRONIC HEPATITIS B AND ANTIVIRAL RESISTANCE 271

When to test. Clinicians might consider testing at theollowing times: (1) before selection of initial therapy, (2) routinelyuring the course of therapy, and (3) at the time of virologicreakthrough, before changing the therapeutic regimen. Evalua-ion of viral genetic mutations has been used extensively in pa-ients with HIV infection, and it is useful to review current recom-

endations for HIV genotypic resistance testing.Testing before initiation of therapy. Current guide-

ines recommend HIV drug resistance testing before initiationf therapy.52 This recommendation is justified by repeated ob-ervations of transmission of viral strains resistant to one or

ore antiretroviral drugs53; however, it is much less clearhether resistance testing is necessary before the initiation ofucleoside/nucleotide antiviral therapy in patients with HBV

nfection. At this time, it is therefore reasonable to suggest thatesistance testing before the initiation of nucleoside/nucleotidereatment is not warranted in patients with HBV infectionnless they acquired HBV from an infected patient undergoingntiviral treatment.

Testing during treatment. Although monitoring themergence of resistance-conferring mutations may be the mostensitive way to monitor patients taking nucleosides/nucleo-ides for the treatment of HBV infection,48 this approach doesot seem warranted on the basis of current information. Viralebound occurs well before a biochemical breakthrough orepatitis flare,46 and monitoring of serum HBV DNA with aensitive assay as recommended should permit clinicians toetect changes in viral load that signal loss of efficacy before

ncreases in ALT levels or clinical deterioration. Notably, guide-ines for management of HIV-infected patients make no recom-

endations for routine genotypic or phenotypic resistance test-ng in patients who maintain viral suppression duringherapy,52 and the same can be said for HBV infection.

Resistance testing at virologic breakthrough. HIVreatment guidelines recommend resistance testing beforehanging treatment regimens after virologic failure,52 and rec-mmendations for the management of patients with HBV in-ection also recognize the importance of resistance testing whenBV DNA testing indicates a 1 log10 or greater increase in viral

oad. Resistance testing should be performed, if possible, beforeiscontinuation of the failing regimen so that the results reflecthe latest detectable viral mutations. Results of resistance test-

able 1. Treatment Recommendations of the AASLD and US

Resistance to AASLD46

LMV Add ADV or TDF; stop LMV and switch to FTC/Tstop LMV and switch to ETV (pre-existing LMmutations predispose to ETV resistance)

ADV Add LMV; stop ADV and switch to FTC/TDF comto or add ETV

ETV Switch to or add ADV or TDFLdT Add ADV or TDF; stop LdT and switch to FTC/T

stop LdT and switch to ETV (pre-existing LdTmutations predispose to ETV resistance)

DV, adefovir dipivoxil; ETV, entecavir; FTC, emtricitabine; LdT, telbiv

ng then can be used to guide selection of new treatment.46 n

Management of Patients With ResistantVirusBoth the AASLD guidelines and those published by

eeffe et al47 provide recommendations for changing therapy inatients with HBV antiviral drug resistance to lamivudine, ad-fovir, or entecavir. The AASLD guidelines also provide recom-endations for new therapy in patients with resistance to

elbivudine (Table 1).46,47

Clinical trials performed to date have focused on second-lineherapy for patients who have developed resistance to lamivu-ine and have shown that adding on or switching to adefovir,r switching to entecavir, can resuppress viral replication, nor-alize ALT levels, and provide significant histologic improve-ent.54 –57 Tenofovir also has been shown to be effective in

atients with HBV resistance to both lamivudine and adefovir.44

ll data indicate that entecavir monotherapy should not beecommended in lamivudine-resistant patients because of theigh risk of entecavir resistance and adding a second drug islways preferable to a switch.

Rapti et al58 recently showed that the addition of adefovir toamivudine in HBeAg-negative patients with lamivudine resis-ance effectively suppressed serum HBV DNA in the majority ofatients without the development of adefovir resistance. In thistudy of patients with lamivudine resistance, 14 patients werewitched to adefovir, and 28 patients had adefovir added tongoing lamivudine therapy. Serum HBV DNA levels becamendetectable and ALT levels normalized in 71% and 90% ofatients, with no differences between the 2 groups. However,irologic and biochemical breakthrough occurred in 3 of 14atients (21%) who were switched to adefovir monotherapy 15o 18 months from the start of therapy, whereas suppression ofBV replication was persistent for up to 3 years in the group

eceiving combination adefovir plus lamivudine.Timing of the change in treatment also is important for

atients who experience viral breakthrough while receiving nu-leotide/nucleoside therapy. A prompt switch to new treatmentesults in more rapid viral suppression than does a delayedhange in therapy. A study of HBeAg-negative patients withhronic hepatitis B and genotypic and phenotypic resistance toamivudine indicated that adding adefovir when genotypic re-istance was detected (3– 6 log10 copies/mL of HBV DNA and

rithm for Patients With Antiviral-Resistant HBV46,47

US Algorithm47

ombination;istance

Add ADV (may be preferred over switch to ADV);switch to ETV (risk for subsequent ETVresistance); potential future management:add TDF or switch to FTC/TDF combination

tion; switch Add LMV (may be preferred over switch toLMV); switch to ETV (if no prior LMVresistance); potential future therapy: switchto FTC/TDF combination

Add or switch to ADV or TDFmbination;tance

; LMV, lamivudine; TDF, tenofovir DF.

Algo

DF cV res

bina

DF coresis

ormal ALT levels) was more effective than adding adefovir

wocpdnrmapi

accgraanHtt

ttTfcarfto

btHmblAedlr1r

ahwertdil

crvHwatto

pgatwawittplvsCattFt

idamrrcdilal

tonanHlYmvmaor

272 KEEFFE ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

hen phenotypic resistance was detected (�6 log10 copies/mLf HBV DNA and high ALT levels).59 Three months after thehange in treatment, serum HBV DNA was undetectable in allatients who had adefovir added when genotypic resistance wasetected, versus 46% of those who received adefovir when phe-otypic resistance was observed. The 2-year rates of virologicesponse were 100% and 78%, respectively, and ALT levels re-

ained normal in all patients who were given early addition ofdefovir. In contrast, ALT levels were normalized in 50% ofatients at 6 months, 72% at 12 months, and 93% at 24 months

n the group with late addition of adefovir.59

Avoiding Resistance Development:Selection of Initial TherapyCurrent treatment of HIV infection is based on creating

high genetic barrier to resistance. This refers to the number ofritical drug-resistant mutations required for the virus to over-ome the anti-HIV activity of the drug regimen.60 Currentuidelines for initial treatment of patients with HIV infectioneflect that view by recommending a combination of 3 or moregents from different classes having different mechanisms ofction.52 Single-drug class monotherapy or dual therapy withucleoside reverse-transcriptase inhibitors for the treatment ofIV has been shown to accelerate the development of resis-

ance, cross-resistance, and virologic failure, and also limit fu-ure treatment options.

Although data are not available for all antiviral agents usedo treat HBV infection, it is becoming clear that single-drugherapy may not provide a sufficiently high barrier to resistance.hat is certainly the case for lamivudine9 and it also may be true

or other agents. This has prompted consideration of 2-drugombinations as initial treatment in patients with chronic hep-titis B. Potential exceptions to the requirement of a 2-drugegimen to reduce or prevent resistance are entecavir and teno-ovir, with the cumulative experience suggesting that HBV an-iviral drug resistance occurs at very low rates of less than 1%ver 4 years of treatment, in the case of entecavir.36

Several small-scale studies have assessed the efficacy of com-ination therapy in treatment-naive patients with HBV infec-ion. Results from a 1-year comparison of 104 treatment-naiveBeAg-positive patients treated with telbivudine or lamivudineonotherapy or telbivudine plus lamivudine showed that tel-

ivudine alone and in combination therapy was superior toamivudine alone in reducing HBV DNA levels and achievingLT normalization, and that there were no significant differ-nces between results obtained with telbivudine versus telbivu-ine plus lamivudine.61 Moreover, addition of telbivudine to

amivudine, as might be expected, did not decrease the occur-ence of viral breakthrough (15.8% for lamivudine alone vs2.2% for the combination treatment) or the emergence ofesistance-associated mutations.61

The concept that combination therapy should use drugs thatre not cross-resistant is exemplified by a study of chronicepatitis B patients treated with lamivudine alone or combinedith adefovir for 1 year.62 Initially, both treatments resulted in

qually effective viral suppression, with 4 to 5 log10 HBV DNAeductions by week 16. However, by week 52, the combinationherapy group remained virally suppressed, with a median re-uction from baseline of 5.4 log10, whereas HBV DNA levels

ncreased in the lamivudine monotherapy group. That increase

argely was owing to the development of lamivudine resistance– d

onferring mutations, which occurred in 20 of the patients whoeceived lamivudine alone versus 2% of those treated with lami-udine plus adefovir.62 Another study of 30 treatment-naive,BeAg-positive patients showed that 48 weeks of treatmentith adefovir plus emtricitabine resulted in greater antiviralctivity than adefovir monotherapy; median HBV DNA reduc-ions at 48 weeks were �3.48 log10 versus �2.22 log10, respec-ively.63 No information was provided regarding the emergencef viral resistance.

The combination of lamivudine with an immune modulator,eginterferon alfa-2a, also has been shown to slow the emer-ence of lamivudine resistance in a 72-week study (48 weeks ofctive treatment and 24 weeks of additional follow-up evalua-ion) in which 537 treatment-naive HBeAg-negative patientsith chronic hepatitis B were treated with peginterferon alfa-2alone, lamivudine alone, or a combination of the two. At 48eeks, mutations conferring resistance to lamivudine developed

n 18% of patients treated with lamivudine alone versus 1% ofhose who received combination therapy.64 Results from longer-erm follow-up evaluation of 36 HBeAg-negative, anti-HBe–ositive, treatment-naive patients showed that treatment with

amivudine plus interferon alfa-2b for 1 year followed by lami-udine alone for 3 additional years is an approach that alsoubstantially may slow the emergence of lamivudine resistance.umulative rates of breakthrough viremia at the end of 1, 2, 3,nd 4 years of treatment were 0%, 14%, 32%, and 59%, respec-ively.65 These values compare with 23%, 46%, 55%, and 71% forhe first 4 years of lamivudine monotherapy in another study.9

urther study of an induction/maintenance approach to HBVherapy with other antiviral combinations is needed.

ConclusionsThe goal of therapy in patients with chronic hepatitis B

s rapid viral suppression and long-lasting maintenance of un-etectable levels of serum HBV DNA. Nucleoside/nucleotidenalogues rapidly are becoming the treatment of choice forost patients with chronic hepatitis B. These agents generally

equire indefinite administration but usually are well tolerated,esult in rapid viral suppression, improve Child–Pugh scores inirrhotic patients, and improve overall survival. The majorrawback of this treatment is the considerable risk of develop-

ng antiviral drug resistance, which occurs most frequently inamivudine-treated patients, but also has been shown for othergents (eg, adefovir, emtricitabine, and telbivudine) evaluated inong-term studies.

Preventing emergence of resistance and virologic break-hrough may best be achieved by using an agent or combinationf agents with a high genetic barrier to resistance. This is clearlyot the case for lamivudine, which requires the selection of onlysingle mutation to result in high-level resistance and shouldo longer be used as monotherapy in patients with chronicBV infection, except possibly in patients with compensated

iver disease who become HBV DNA negative early in therapy.uen et al66 recently reported that serum HBV DNA measure-ents at week 4 and week 16 were useful in predicting lami-

udine response at year 5, with the week 4 measurement theost advantageous in allowing earlier prediction of response

nd/or failure to therapy owing to resistance. The combinationf lamivudine with other agents may increase the barrier toesistance, but large-scale, long-term studies still are needed to

ocument this effect. We also need more information about

lni

dlsdeaiwpauadtttl

Hopraedrvtns

1

1

1

1

1

1

1

1

1

1

2

2

2

2

2

2

2

2

2

2

3

3

March 2008 CHRONIC HEPATITIS B AND ANTIVIRAL RESISTANCE 273

ong-term viral suppression and emergence of resistance toewer agents (entecavir, telbivudine, tenofovir, and emtricitab-

ne) when they are used as monotherapy or in combination.A combination of 2 potent nucleosides/nucleotides with

ifferent resistance profiles may prove to be the optimal first-ine treatment for chronic hepatitis B, but it remains to behown in clinical trials. However, these studies are extremelyifficult to conduct and may not be feasible. Drugs such asntecavir and tenofovir have such low rates of resistance thatny study to determine improved outcomes, such as reductionn already very low rates of resistance and/or added efficacy,ould need huge numbers of patients treated for an extendederiod of time before any difference between combination ther-py and monotherapy might be shown. Thus, the decision tose combination therapy de novo might have to be made beforeppropriate information becomes available from formal ran-omized controlled trials. Further evaluation of the combina-ion of nucleosides/nucleotides with a course of interferonreatment also is warranted, given short-term results suggestinghat this approach may slow the emergence of resistance toamivudine.

Frequent, periodic assessment of viral load with a sensitiveBV DNA assay remains the best approach for early detection

f resistance, and testing frequency must be tailored to theatient’s disease severity and treatment history. The treatmentegimen should be modified as soon as resistance is detected;gents should be added or discontinued on the basis of knownffects of the detected mutations on the efficacy of specificrugs. Addition or substitution of newer antiviral agents canestore suppression of viral replication after emergence of lami-udine resistance. Nevertheless, new agents with viral targetshat differ from those of currently available drugs will beeeded for patients who have failed or ultimately may failecond or third treatment regimens.

References

1. Centers for Disease Control and Prevention (CDC). Viral hepatitisB. 2006. Available at: http://www.cdc.gov/ncidod/diseases/hepatitis/b/fact.htm. Accessed: July 17, 2007.

2. Glebe D. Recent advances in hepatitis B virus research: a Ger-man point of view. World J Gastroenterol 2007;13:8–13.

3. World Health Organization. Fact sheet no. 204. Hepatitis B. 2000.Available at: http://www.who.int/mediacentre/factsheets/fs204/en/. Accessed: July 17, 2007.

4. Wasley A, Miller JT, Finelli L, Centers for Disease Control andPrevention (CDC). Surveillance for acute viral hepatitis—UnitedStates, 2005. MMWR Surveill Summ 2007;56:1–24.

5. Liaw YF, Sung JJ, Chow WC, et al. Lamivudine for patients withchronic hepatitis B and advanced liver disease. N Engl J Med2004;351:1521–1531.

6. Arora G, Keeffe EB. Chronic hepatitis B with advanced hepaticfibrosis or cirrhosis: impact of antiviral therapy. Rev GastroenterolDisord 2007;7:63–73.

7. Zoulim F. In vitro models for studying hepatitis B virus drugresistance. Semin Liver Dis 2006;26:171–180.

8. Bartholomeusz A, Locarnini SA. Antiviral drug resistance: clinicalconsequences and molecular aspects. Semin Liver Dis 2006;26:162–170.

9. Lok AS, Lai CL, Leung N, et al. Long-term safety of lamivudinetreatment in patients with chronic hepatitis B. Gastroenterology2003;125:1714–1722.

0. Andreone P, Gramenzi A, Cursaro C, et al. High risk of hepato-

cellular carcinoma in anti-HBe positive liver cirrhosis patients

developing lamivudine resistance. J Viral Hepat 2004;11:439–442.

1. Locarnini S. Molecular virology of hepatitis B virus. Semin LiverDis 2004;24(Suppl 1):3–10.

2. Norder H, Courouce AM, Magnius LO. Complete genomes, phy-logenetic relatedness, and structural proteins of six strains of thehepatitis B virus, four of which represent two new genotypes.Virology 1994;198:489–503.

3. Beck J, Nassal M. Hepatitis B virus replication. World J Gastro-enterol 2007;13:48–64.

4. Okamoto H, Imai M, Kametani M, et al. Genomic heterogeneity ofhepatitis B virus in a 54-year-old woman who contracted theinfection through materno-fetal transmission. Jpn J Exp Med1987;57:231–236.

5. Bartholomeusz A, Locarnini S. Hepatitis B virus mutations asso-ciated with antiviral therapy. J Med Virol 2006;78(Suppl 1):S52–S55.

6. Locarnini S, Hatzakis A, Heathcote J, et al. Management ofantiviral resistance in patients with chronic hepatitis B. AntivirTher 2004;9:679–693.

7. Da Silva LC, Pinho JR, Sitnik R, et al. Efficacy and tolerability oflong-term therapy using high lamivudine doses for the treatmentof chronic hepatitis B. J Gastroenterol 2001;36:476–485.

8. Wright TL. Clinical trial results and treatment resistance withlamivudine in hepatitis B. Semin Liver Dis 2004;24(Suppl 1):31–36.

9. Dienstag JL, Goldin RD, Heathcote EJ, et al. Histological outcomeduring long-term lamivudine therapy. Gastroenterology 2003;124:105–117.

0. Yuen MF, Kato T, Mizokami M, et al. Clinical outcome and viro-logic profiles of severe hepatitis B exacerbation due to YMDDmutations. J Hepatol 2003;39:850–855.

1. Si Ahmed SN, Tavan D, Pichoud C, et al. Early detection of viralresistance by determination of hepatitis B virus polymerase mu-tations in patients treated by lamivudine for chronic hepatitis B.Hepatology 2000;32:1078–1088.

2. Mutimer D, Pillay D, Shields P, et al. Outcome of lamivudineresistant hepatitis B virus infection in the liver transplant recipi-ent. Gut 2000;46:107–113.

3. Zoulim F. Mechanism and viral persistence and resistance tonucleoside and nucleotide analogs in chronic hepatitis B virusinfection. Antiviral Res 2004;64:1–15.

4. Hussain M, Lok AS. Mutations in the hepatitis B virus polymerasegene associated with antiviral treatment for hepatitis B. J ViralHepat 1999;6:183–194.

5. Poch O, Sauvaget I, Delarue M, et al. Identification of four con-served motifs among the RNA-dependent polymerase encodingelements. EMBO J 1989;8:3867–3874.

6. Allen MI, Deslauriers M, Andrews CW, et al. Identification andcharacterization of mutations in hepatitis B virus resistant tolamivudine. Hepatology 1998;27:1670–1677.

7. Delaney WE 4th, Yang H, Westland CE, et al. The hepatitis B viruspolymerase mutation rtV173L is selected during lamivudine ther-apy and enhances viral replication in vitro. J Virol 2003;77:11833–11841.

8. Fu L, Cheng YC. Role of additional mutations outside the YMDDmotif of hepatitis B virus polymerase in L(�)SddC (3TC) resis-tance. Biochem Pharmacol 1998;55:1567–1572.

9. Seigneres B, Martin P, Werle B, et al. Effects of pyrimidine andpurine analog combinations in the duck hepatitis B virus infectionmodel. Antimicrob Agents Chemother 2003;47:1842–1852.

0. Curtis M, Zhu Y, Borroto-Esoda K. HBV rtI233V polymerase vari-ant remains sensitive to adefovir (abstr). J Hepatol 2007;46(Suppl 1):S26–S27.

1. Hadziyannis SJ, Tassopoulos NC, Heathcote EJ, et al. Long-

term therapy with adefovir dipivoxil for HBeAg-negative chronic

3

3

3

3

3

3

3

3

4

4

4

4

4

4

4

4

4

4

5

5

5

5

5

5

5

5

5

5

6

6

6

6

6

6

6

GUC5

s

274 KEEFFE ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

hepatitis B for up to 5 years. Gastroenterology 2006;131:1743–1751.

2. Borroto-Esoda K, Arterburn S, Snow A, et al. Final analysis ofvirological outcomes and resistance during 5 years of adefovirdipivoxil monotherapy in HBeAg-negative patients (abstr). J Hepa-tol 2006;44(Suppl 2):S179–S180.

3. Langley DR, Walsh AW, Baldick CJ, et al. Inhibition of hepatitis Bvirus polymerase by entecavir. J Virol 2007;81:3992–4001.

4. Colonno RJ, Rose R, Baldick CJ, et al. Entecavir resistance is rarein nucleoside naïve patients with hepatitis B. Hepatology 2006;44:1656–1665.

5. Tenney DJ, Rose RE, Baldick CJ, et al. Two-year assessment ofentecavir resistance in lamivudine-refractory hepatitis B viruspatients reveals different clinical outcomes depending on theresistance substitutions present. Antimicrob Agents Chemother2007;51:902–911.

6. Colonno RJ, Rose RE, Pokornowski K, et al. Four year assess-ment of ETV resistance in nucleoside-naïve and lamivudine re-fractory patients (abstr). J Hepatol 2007;46(Suppl 1):S294.

7. Seifer M, Patty A, Dukhan D, et al. Telbivudine (LdT) preferentiallyinhibits second (�) strand HBV DNA synthesis (abstr). Gastroen-terology 2005;128(Suppl 2):A742–A743.

8. Lai CL, Gane E, Hsu CW, et al. Two-year results from the GLOBEtrial in patients with hepatitis B: greater clinical and antiviralefficacy for telbivudine (LdT) vs lamivudine (abstr). Hepatology2006;44(Suppl 1):222A.

9. Lim SG, Ng TM, Kung N, et al. A double-blind placebo-controlledstudy of emtricitabine in chronic hepatitis B. Arch Intern Med2006;166:49–56.

0. Yang H, Qi X, Sabogal A, et al. Cross-resistance testing of next-generation nucleoside and nucleotide analogues against lamivu-dine-resistant HBV. Antivir Ther 2005;10:625–633.

1. Gish RG, Trinh H, Leung N, et al. Safety and antiviral activity ofemtricitabine (FTC) for the treatment of chronic hepatitis B infec-tion: a two-year study. J Hepatol 2005;43:60–66.

2. Delaney WE 4th, Ray AS, Yang H, et al. Intracellular metabolismand in vitro activity of tenofovir against hepatitis B virus. Antimi-crob Agents Chemother 2006;50:2471–2477.

3. Lada O, Benhamou Y, Cahour A, et al. In vitro susceptibility oflamivudine-resistant hepatitis B virus to adefovir and tenofovir.Antivir Ther 2004;9:353–363.

4. van Bömmel F, Zollner B, Sarrazin C, et al. Tenofovir for patientswith lamivudine-resistant hepatitis B virus (HBV) infection andhigh HBV DNA level during adefovir therapy. Hepatology 2006;44:318–325.

5. van der Eijk AA, Hansen BE, Niesters HG, et al. Viral dynamicsduring tenofovir therapy in patients infected with lamivudine-resis-tant hepatitis B virus mutants. J Viral Hepat 2005;12:364–372.

6. Lok AS, McMahon BJ. Chronic hepatitis B. Hepatology 2007;45:507–539.

7. Keeffe EB, Dieterich DT, Han SH, et al. A treatment algorithm forthe management of chronic hepatitis B virus infection in theUnited States: an update. Clin Gastroenterol Hepatol 2006;4:936–962.

8. Tillmann HL. Antiviral therapy and resistance with hepatitis Bvirus infection. World J Gastroenterol 2007;13:125–140.

9. Lok AS, Zoulim F, Locarnini S, et al. Monitoring drug resistance inchronic hepatitis B virus (HBV)-infected patients during lamivu-dine therapy: evaluation of performance of INNO-LiPA HBV DRassay. J Clin Microbiol 2002;40:3729–3734.

0. Zoulim F. New nucleic acid diagnostic tests in viral hepatitis.Semin Liver Dis 2006;26:309–317.

1. Tran N, Berne R, Chann R, et al. European multicenter evaluationof high-density DNA probe arrays for detection of hepatitis B virusresistance mutations and identification of genotypes. J Clin Mi-crobiol 2006;44:2792–2800.

2. Department of Health and Human Services (DHHS). Guidelines t

for the use of antiretroviral agents in HIV-1-infected adults andadolescents. October 10, 2006. Available at: http://AIDSinfo.nih.gov/contentfiles/Adultandadolescent6L.pdf. Accessed: July17, 2007.

3. Tang JW, Pillay D. Transmission of HIV-1 drug resistance. J ClinVirol 2004;30:1–10.

4. Sherman M, Yurdaydin C, Sollano J, et al. Entecavir for treatmentof lamivudine-refractory, HBeAg-positive chronic hepatitis B. Gas-troenterology 2006;130:2039–2049.

5. Peters MG, Hann HWH, Martin P, et al. Adefovir dipivoxil alone orin combination with lamivudine in patients with lamivudine-resis-tant chronic hepatitis B. Gastroenterology 2004;126:91–101.

6. Perrillo R, Hann HW, Mutimer D, et al. Adefovir dipivoxil added toongoing lamivudine in chronic hepatitis B with YMDD mutanthepatitis B virus. Gastroenterology 2004;126:81–90.

7. Chang TT, Gish RG, Hadziyannis SJ, et al. A dose-ranging study ofthe efficacy and tolerability of entecavir in lamivudine-refractorychronic hepatitis B patients. Gastroenterology 2005;129:1198–1209.

8. Rapti I, Dimou E, Mitsoula P, et al. Adding-on versus switching-toadefovir therapy in lamivudine-resistant HBeAg-negative chronichepatitis B. Hepatology 2007;45:307–313.

9. Lampertico P, Vigano M, Manenti E, et al. Adefovir rapidly sup-presses hepatitis B in HBeAg-negative patients developing geno-typic resistance to lamivudine. Hepatology 2005;42:1414–1419.

0. Luber AD. Genetic barriers to resistance and impact on clinicalresponse. EJIAS: eJournal of the International AIDS Society2005;7:69:1–11. Available at: http://www.medscape.com/viewarticle/504524_print. Accessed: March 20, 2007.

1. Lai CL, Leung N, Teo EK, et al. A 1-year trial of telbivudine,lamivudine, and the combination in patients with hepatitis B eantigen-positive chronic hepatitis B. Gastroenterology 2005;129:528–536.

2. Sung JJY, Lai JY, Zeuzem S, et al. A randomised double-blindphase II study of lamivudine (LAM) compared with lamivudineplus adefovir dipovoxil (ADV) for treatment naïve patients withchronic hepatitis B (CHB): week 52 analysis (abstr). J Hepatol2003;38(Suppl 2):25–26.

3. Lau G, Cooksley H, Ribeiro RM, et al. Randomised, double-blindstudy comparing adefovir dipivoxil plus emtricitabine combinationtherapy versus adefovir alone in HBeAg(�) chronic hepatitis B:efficacy and mechanisms of treatment response (abstr). Hepa-tology 2004;40(Suppl 1):272A.

4. Marcellin P, Lau GK, Bonino F, et al. Peginterferon alfa-2a alone,lamivudine alone, and the two in combination in patients withHBeAg-negative chronic hepatitis B. N Engl J Med 2004;351:1206–1217.

5. Nikolaidis NL, Giouleme OI, Tziomalos KA, et al. Interferon/long-term lamivudine combination therapy in anti-HBe positive chronichepatitis B patients. J Gastroenterol Hepatol 2005;20:1721–1725.

6. Yuen MF, Wong DKH, Fung J, et al. Predictive value of HBV DNAlevels at frequent time points during early and maintenancephase of 5-year lamivudine and mutational profiles of reversetranscriptase (RT) and surface (S) genes (abstr). Hepatology2006;44:557A.

Address requests for reprints to: Emmet B. Keeffe, MD, Division ofastroenterology and Hepatology, Department of Medicine, Stanfordniversity Medical Center, 750 Welch Road, Suite 210, Palo Alto,alifornia 94304-1509. e-mail: ekeeffe@stanford.edu; fax: (650) 498-692.The development of this manuscript was supported by an unre-

tricted educational grant from Gilead Sciences (Foster City, CA).The authors wish to thank Filippo Cavalieri, PhD, for editorial con-

ributions and assistance in the preparation of the manuscript.

P

A

*N

PwislracaacowHidcmordfi

Wmdttiecm

dctficpiac

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2008;6:275–282

ancreatic Cancer: A Review of the Evidence on Causation

NDREW R. HART,*,‡ HUGH KENNEDY,‡ and IAN HARVEY*

Population Health, School of Medicine, Health Policy and Practice, University of East Anglia, Norwich, Norfolk, United Kingdom; ‡Department of Gastroenterology,

orfolk & Norwich University Hospital National Health Service Trust, Colney, Norfolk, Norwich, United Kingdom

nt

iatupprtototobmrildcitr5c

ror[rtgdcto2p

d

See CME exam on page 262.

ancreatic cancer kills more than 250,000 people each yearorldwide and has a poor prognosis. The aim of this article

s to critically review the epidemiologic evidence for expo-ures that may either increase or decrease the risk. A Med-ine search was performed for epidemiologic studies andeviews published up to April 2007. Consistent evidence of

positive association was found for family history andigarette smoking. Many studies documented a positivessociation with diabetes mellitus and chronic pancreatitis,lthough the etiologic mechanisms are unclear. Other asso-iations were detected, but the results were either inconsistentr from few studies. These included positive associationsith red meat, sugar, fat, body mass index, gallstones, andelicobacter pylori, and protective effects of increasing par-

ty, dietary folate, aspirin, and statins. There was no evi-ence linking alcohol or coffee consumption with an in-reased risk of pancreatic cancer. The associations withany exposures need to be clarified from further epidemi-

logic work in which there is both precise measurement ofisk factors, adjustment for potential confounders, and, forietary studies, information recorded on the method ofood preparation and pattern of consumption. Such work ismportant to reduce the incidence of this fatal disease.

orldwide, adenocarcinoma of the pancreas causes morethan a quarter of a million deaths annually, is the 13th

ost common cancer and the eighth most frequent cause ofeath from cancer.1 Survival rates are among the worst for anyumor, with a mortality to incidence ratio of 98%.1 Most pa-ients present with advanced disease and management optionsnclude chemotherapy, pain control, and relief of jaundice byither surgical or endoscopic means. The cause of pancreaticancer is largely unknown and there are currently no acceptedodels of the disease.This review aims to critically review the epidemiologic evi-

ence for exposures and, in areas where there is sparse oronflicting evidence, suggest studies that may provide clarifica-ion. In writing this article, the Medline database was searchedor published articles between 1950 and April 2007. The principalnitial search terms used, which were combined with pancreaticancer, were as follows: family history, occupation, diabetes,ancreatitis, cholecystectomy, Helicobacter pylori, smoking, par-

ty, meat, fat, sugar, fruits, vegetables, coffee, alcohol, physicalctivity, body mass index, aspirin, and statins. For several terms

omprehensive reviews had been published on specific areas,

amely occupations, smoking, diabetes, coffee, and alcohol, andhese are referenced.

Two types of epidemiologic study design can be used tonvestigate exposures for pancreatic cancer, namely case-controlnd cohort studies, with both having advantages and disadvan-ages. Case-control studies recruit a wide selection of cases andse accurate measures of risk factors because the numbers ofarticipants included are relatively small. However, a majorroblem can be recall bias, where it is difficult to ensure accu-ate information on risk factors before the onset of symptomshat potentially were involved in the cause of the disease. An-ther limitation is selection bias if control groups are chosenhat are not representative of the general population. The sec-nd method of study is cohort studies, and their strength ishat the exposure data collected truly represent that before thenset of symptoms. Also, there is less selection bias becauseoth participants who develop the disease and those who re-ain well are drawn from the same population. In writing this

eview we decided that recall bias for many exposures such asllnesses, occupation, smoking, and medications is likely to beow and therefore included both study designs. However, forietary exposures, recall bias for food items exists in case-ontrol studies, which is removed by a cohort design. Therefore,n this review only dietary data from the latter type is includedo ensure the recorded diet truly represents that associated withisk. However, we acknowledge that there are at least a further5 case-control studies that have examined diet and pancreaticancer.

Family History and Genetic DisordersA family history of pancreatic cancer in a first-degree

elative is associated with an increased risk of pancreatic cancerf between 2.5 to 5.3 times.2– 6 The risk increases the moreelatives are affected, with a risk of 6.4 (95% confidence intervalCI], 1.8 –16.4) in those with 2 affected relatives, increasing to aisk of 32.0 (95% CI, 10.2–74.7) in those with 3 affected rela-ives.7 The increased risk in family members could be owing toenetic factors and/or lifestyle habits that may be similar inifferent generations, such as smoking. A case-control studyonducted in Michigan addressed this possibility6 and reportedhat pancreatic cancer in a first-degree relative (parent, sibling,ffspring) was associated with an increased risk of the disease of.5 (95% CI, 1.3– 4.7) times, after adjusting for smoking in theatient’s relative. Interaction existed between the 2 factors in

Abbreviations used in this paper: BMI, body mass index; CI, confi-ence interval; RR, relative risk.

© 2008 by the AGA Institute1542-3565/08/$34.00

doi:10.1016/j.cgh.2007.12.041

tw

caissbwg(tlitoecac1(wdtrc

cpnmcacgtctmstregnec

dIcsfs

ctptiivoteslscnsdsc

e1tcfihitectarm

tmatbc7hascew1ACppttww

276 HART ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

hat the risk was 6.0 times greater (95% CI, 2.0 –18.3) in thoseho had an affected relative who also smoked.

A number of genetic disorders are associated with an in-reased risk of pancreatic cancer, although no common geneticbnormality has been identified in all these conditions. Thesenclude hereditary pancreatitis, familial adenomatous polypo-is, familial atypical multiple mole melanoma, Peutz-Jeghersyndrome, hereditary nonpolyposis colorectal cancer, familialreast cancer, ataxia-telangiectasia, and cystic fibrosis. Patientsith hereditary pancreatitis have a 53 (95% CI, 23–105) timesreater risk of disease with a cumulative lifetime risk of 40%95% CI, 9%–71%),8 and in those with Peutz-Jeghers syndromehe relative risk is 132 (95% CI, 44 –261) with a cumulativeifetime risk of 36%.9 In this condition, germline mutationsn the STK 11 gene have been identified,10 which acts as aumor-suppressor gene in the earliest steps of the pathogenesisf hamartomas into adenocarcinomas. Finally, Maisonneuvet al11 collected reports of patients presenting with pancreaticancer who also suffered from cystic fibrosis in North Americand Europe. By using population data, the relative risk ofancer in individuals with cystic fibrosis was 5.3 (95% CI, 2.4 –0.1), with an early age of onset at a median age of 35 yearsrange, 18 –58 y). Furthermore, there is a case-report of a patientith cystic fibrosis and bilateral lung transplantation who un-erwent a Whipple procedure for pancreatic adenocarcinoma athe age of 12 years.12 Because all these genetic conditions areare, the contributions they make to all cases of pancreaticancer are small.

Occupational ExposuresThe effect of occupational exposures was assessed in a

omprehensive meta-analysis of data from 92 studies that re-orted effects for 23 agents.13 The results showed that chlori-ated hydrocarbon solvents and related compounds had aeta-risk ratio of 1.4 (95% CI, 1.0 –1.8), and for nickel and nickel

ompounds the meta-risk ratio was 1.9 (95% CI, 1.2–3.2). For othergents, small but nonsignificant risks were reported, includinghromium compounds, polycyclic aromatic hydrocarbons, or-anochlorine insecticides, silica dust, and asbestos (meta-risk ra-io, 1.1; 95% CI, 0.9 –1.5). Although most studies have notonfirmed an association with occupational asbestos exposure,here are ecologic data to support this hypothesis from com-

unities with high levels of asbestos in the drinking waterupplies.14 A second meta-analysis of 14 epidemiologic studieshat assessed risk in occupations exposed to formaldehydeeported a pooled relative risk of 1.1 (95% CI, 1.0 –1.3).15 How-ver, this was limited to embalmers, anatomists, and patholo-ists (relative risk [RR], 1.3; 95% CI, 1.0 –1.7). Currently, there iso plausible biological mechanism because most biologicalffects of formaldehyde are at the site of contact with thehemical.

Hormonal FactorsIncreasing parity may prevent pancreatic cancer by re-

ucing insulin-like growth factor levels and body iron stores.nsulin-like growth factors are involved in the development ofancer and promote cellular proliferation and inhibit apopto-is.16,17 Pregnancy can induce changes in the insulin-like growthactor axis and an analysis of plasma insulin-like growth factors

howed a lower concentration in women with 4 or more births q

ompared with nulliparous women (180 vs 212 ng/mL, P forrend � .003).18 Secondly, during pregnancy, there is a markedhysiologic reduction in total body iron stores. Two case-con-rol studies have reported that increased iron consumption andncreased serum iron concentrations are associated with anncreased risk of pancreatic cancer.19,20 Free iron may be in-olved in carcinogenesis by inducing DNA damage by causingxidative stress.21 Several epidemiologic studies have reportedhat parity reduces the risk of pancreatic cancer,22–26 which mayxplain why women have a lower incidence of the disease. Thesetudies recorded a reduction in risk of pancreatic cancer of ateast 20% in women who had 4 or 5 children. Two Scandinaviantudies did not report an association, although these did notontrol for the possible confounding effect of smoking.27,28 Fi-ally, a cohort investigation from Japan, which did adjust formoking, again did not show an effect.29 The epidemiologic evi-ence is inconsistent, despite the biological mechanisms, andhould be clarified by future studies considering all potentialonfounders.

Concomitant IllnessIn patients with diabetes mellitus, a meta-analysis of 36

pidemiologic studies reported an odds ratio of 1.82 (95% CI,.66 –1.89).30 The mechanism of the diabetes association needso be studied, investigating possibilities that include a commonausative agent, the metabolic consequences of diabetes, and,nally, residual confounding. A possible mechanism is thatyperglycemia promotes hyperinsulinemia and activation of

nsulin-like growth factor 1, which stimulates cell prolifera-ion.31 A potential difficulty with this association between dis-ases is the issue of reverse causality, in that destruction of isletells by malignant tissue may precipitate diabetes, rather thanhe latter actually being a causative factor.32 However, thisssociation is likely to be real because epidemiologic studieseport an increased risk in patients diagnosed with diabetes

ore than 10 years before cancer onset.33

Acute and chronic pancreatitis have been shown consistentlyo be risk factors for pancreatic cancer.34 –38 The biological

echanisms are again unknown, but possibly related to dam-ge induced by the inflammatory process. The data suggest thathis is not a spurious association owing to pancreatic cancereing misdiagnosed initially as chronic pancreatitis. The asso-iation still persists when cases of cancer diagnosed more thanyears after an initial diagnosis of pancreatitis are analyzed.37 Aistorical cohort study of 2015 subjects with chronic pancre-titis, who were followed up for a mean of 7.4 years, reported atandardized incidence ratio of 26.3 (95% CI, 19.9 –34.2), with aumulative 4% risk of pancreatic cancer at 20 years follow-upvaluation.38 Two Swedish cohort studies monitoring patientsith pancreatitis34,36 reported relative risks of 2.2 (95% CI,.6 –2.9) and 5.0 (95% CI, 4.1– 6.1) for any form of pancreatitis.study of American Veterans found an odds ratio of 3.42 (95%

I, 1.98 –5.91) for pancreatic cancer in patients with all types ofancreatitis.37 The risks are lower after an episode of acuteancreatitis compared with chronic disease.34,36 Furthermore,he effect of chronic pancreatitis does not seem to be mediatedhrough alcohol because the risk was similar36,37 in patients inhom the cause of pancreatitis was alcohol-related comparedith non–alcohol-induced pancreatitis. Finally, data are re-

uired in investigating the risk in patients with chronic pan-

cp

ftaccpWkajP(fa

cmdacnammlccmmiwd0

hcimcqelhabHctptsbegc

rmppecitrWcwadsss

tOHfscpcittbi

epcmrrproTosasmfiip

March 2008 CAUSE OF PANCREATIC CANCER 277

reatitis who also have developed diabetes as a complication ofancreatitis.

Gallstones and cholecystectomy have been reported as riskactors for pancreatic cancer with risks in the range 1.3 to 2.8imes,5,39 – 43 although other studies have not found any associ-tions.44 – 49 A possible mechanism is that cholecystectomy in-reases the circulating levels of the gastrointestinal hormone,holecystokinin-pancreozymin, which has trophic effects on theancreas including pancreatic hyperplasia and hypertrophy.50,51

hy discrepancies exist in the epidemiologic literature is un-nown, although possibilities include other causative factorsssociated with stone development, which have not been ad-usted for in the analyses. In the US Nurse Health and Healthrofessionals Follow-up Studies, the risk decreased from 1.31

95% CI, 0.93–1.83) to 1.11 (95% CI, 0.78 –1.56) after correctingor other factors including body mass index (BMI) and physicalctivity.52

Other illnesses are associated with an increased risk of pan-reatic cancer including asthma,53 familial atypical multipleole melanoma,54 and other primary malignancies.55 Asthma

oubled the risk of pancreatic cancer, which was not altered bydjustment for smoking, and, furthermore, other respiratoryonditions, namely emphysema and chronic bronchitis, wereot associated.53 Familial atypical multiple mole melanoma isn autosomal-dominant condition characterized by malignantelanoma of the skin and atypical precursor lesions. Germlineutations in the p16 (CDKN2A) gene have been reported in at

east a quarter of patients and the cumulative risk of pancreaticancer by age 75 years is 17%.54 Finally, the incidence of pan-reatic cancer is increased secondary to other smoking-relatedalignancies including lung cancer (RR, 1.3; 95% CI, 1.0 –1.6 inen; RR, 2.5; 95% CI, 1.9 –3.2 in women), head and neck cancer

n women (RR, 1.8; 95% CI, 1.2–2.5), and bladder cancer inomen (RR, 1.5; 95% CI, 1.1–2.0).55 Interestingly, a markedecrease in risk after lymphoma in men has been reported (RR,.2; 95% CI, 0.0 – 0.8).55

H pylori InfectionThere are several plausible biological mechanisms for

ow H pylori infection could increase the risk of pancreaticancer. These are secondary to the H pylori–induced gastritisncluding a reduced absorption of anti-oxidants such as vita-

in C,56 and hypergastrinemia, which stimulates pancreaticancer cell growth.57 Also, increased secretin levels, as a conse-uence of decreased somatostatin production, have trophicffects on the pancreatic ductal epithelium.58 Three epidemio-ogic studies, 1 cohort59 and 2 case-control investigations,60,61

ave investigated this hypothesis. These reported a positivessociation, with the 2 largest recording an approximate dou-ling of the risk of pancreatic cancer in those infected with

pylori.59,60 To confirm consistency of the association, furtherohort studies are required because such methodology ensureshat the H pylori status truly represents whether infection isresent or not before the development of cancer. The associa-ion has clinical implications in that individuals at high risk,uch as those with a family history of diabetes mellitus, coulde screened for infection. The H pylori hypothesis may helpxplain other clinical associations such as patients with a partialastrectomy for ulcer disease having a greater risk of pancreatic

ancer.62– 64 w

Cigarette SmokingSmoking is reported consistently as an environmental

isk factor for pancreatic cancer and accounts for approxi-ately 25% of all pancreatic cancers.65 Carcinogens reach the

ancreas via the blood steam or refluxed bile, and nitrosamines,resent in cigarettes, induce pancreatic tumors in animal mod-ls.66 Cigarette smoking approximately doubles the risk of pan-reatic cancer and the effect is related to its duration andntensity.67–75 The importance of these data prompted the In-ernational Agency for Research in Cancer to state that “ciga-ette smoking is an important cause of pancreatic cancer.”75

hether passive smoking may be involved is unknown. A case-ontrol study conducted in Canada reported that nonsmokersho had been exposed to such smoking, both as a child and asn adult, had a small but nonsignificant increased risk of theisease (odds ratio, 1.21; 95% CI, 0.60 –2.44).76 Smoking alwayshould be measured and adjusted for in etiologic epidemiologictudies of pancreatic cancer and encouraging nonsmokinghould reduce the incidence of the disease.

Total Energy IntakeThe effect of total energy intake has been reported, to

he best of our knowledge, in only 2 prospective cohort studies.ne from Finland, in men, found an inverse trend (P � .05).77

owever, a prospective investigation from Iowa, in women,ound no association with total caloric intake.78 The relation-hip may be confounded by physical activity because there is aorrelation between physical activity and energy intake, andhysical activity itself may be protective against pancreaticancer.79 – 81 Neither study adjusted for the effects of total phys-cal activity. In most of the investigations into diet discussed inhe following sections, risks associated with individual foodypes were adjusted for total energy intake. This is importantecause energy adjustment helps to control for several factors

ncluding body size, metabolic rate, and physical activity.

MeatThe hypothesis that meat increases risk is based on

xperimental data that cooking meat at high temperatures,articularly red meat, produces heterocyclic amines and poly-yclic aromatic hydrocarbons, which are carcinogenic in ani-als.82 To date, 7 cohort studies specifically have investigated

ed meat consumption and pancreatic cancer, with 3 studieseporting a significant positive association,69,83,84 3 studies re-orting no association,67,77,85 and 1 study reporting a reducedisk.86 In studies reporting a positive association, a high intakef red meat approximately doubled the risk of disease.69,83,84

he inconsistency in the evidence may be related to the methodf meat preparation in that cooking at higher temperatures,uch as frying, produces more carcinogens than preparing meatt lower temperatures. Further clarification from epidemiologictudies therefore is required, particularly data on the cooking

ethod. For chicken and poultry, most cohort studies did notnd any association with the disease.69,77,84,85 If red meat were

nvolved, then it potentially could be switched in public healthrograms.

Sugar IntakeA high dietary sugar intake leads to hyperinsulinemia,

hich may be carcinogenic by altering the cell cycle, inhibiting

airbtqcnlrasitafwUsgmUiatiif

oibcew1cahtlSdh

aafaictstl

bg

firahi9fPhwf0rqf

oCchSCiHwt1ico

ficritESis11ppcMbil

278 HART ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

poptosis, and down-regulating insulin-like growth factor bind-ng protein 1. Prospective studies involving biomarkers haveeported double the risk of pancreatic cancer in subjects withoth higher fasting serum glucoses and fasting insulin concen-rations.87,88 However, cohort investigations using food-fre-uency data, have either found no association with total su-rose89 or total sugar intake.90 One study reported a positiveonsignificant association with the highest quintile of glycemic

oad (1.53; 95% CI, 0.96 –2.45),89 and another study found noelationship to this measure.90 Glycemic load is a measure thatllows the carbohydrate content of individual foods to be clas-ified according to their postprandial effects and their effects onnsulin levels. The source of sugar and the pattern of consump-ion may be important. An investigation of 77,797 Swedish mennd women reported a relative risk of 1.95 (95% CI, 1.10 –3.46)or the highest versus the lowest consumption of added sugar,ith a significant trend across quartiles (P � .03).91 In thenited States, women consuming the highest intake of sugar-

weetened soft drinks had a 1.57 (95% CI, 1.02–2.41) timesreater risk of disease,92 although no association was found inen. Soft drinks are the leading source of added sugar in theS diet contributing to a high glycemic load. In summary, there

s supportive evidence for an effect of sugar from experimentalnd biomarker data and for added sugar consumption, al-hough questionnaire data on total sugar intake is unsupport-ve. Clarification is required from further cohort investigationsnvestigating the intake and pattern of sugar consumptionrom different sources.

Dietary FatDietary fat entering the duodenum initiates the release

f cholecystokinin, which stimulates pancreatic hyperplasia,ncreasing its susceptibility to carcinogens.93 Fat intake haseen investigated in 3 cohort studies with no consistent asso-iations detected.77,84,85 In the Finnish male smokers study,nergy-adjusted saturated fat intake was associated positivelyith pancreatic cancer (highest vs lowest quintile: hazard ratio,.60; 95% CI, 0.96 –2.64; P trend � .02).77 No significant asso-iations were detected with total fat, polyunsaturated fattycids, or monounsaturated fatty acids. In the Multiethnic Co-ort Study84 in Hawaii and Los Angeles no associations withhe percentage of total energy from fat, saturated fat, or cho-esterol were detected. Similarly, in the US Nurses’ Healthtudy, no associations were reported with the intake of total fat,ifferent types of fat, and cholesterol.85 Other prospective co-ort studies investigating dietary fat are needed.

Fruit and Vegetable ConsumptionFruits and vegetables contain anti-oxidants that have

nticarcinogenic properties. Cohort studies have not reportedny protective associations between pancreatic cancer and totalruit or vegetable intake,39,67,69,77,94 –96 although, again, smallssociations could be undetected because of measurement errorn the dietary assessment methods. However, in a Swedishohort study a significant inverse association was observed withhe cruciferous vegetable cabbage (�1 serving/wk vs never con-umption: hazard ratio, 0.62; 95% CI, 0.39 – 0.99).96 This pro-ective effect may be related to the high content of glucosino-

ates, which are degraded into isothiocyanates, which inhibit f

oth pancreatic carcinogenesis in animal models and therowth of human pancreatic cancer cell lines.

Consistent evidence exists for a protective effect of dietaryolate. Folate is important in DNA synthesis and repair throughts ability to donate methyl groups. The Swedish cohort studyeported a highly protective effect, reducing the risk of pancre-tic cancer to 0.25 (95% CI, 0.11– 0.59) in those consuming theighest quintile of folate.97 Similar findings also were reported

n the male Finnish smokers’ investigation98 (hazard ratio, 0.52;5% CI, 0.31– 0.87 for highest vs lowest quintile of dietaryolate) and in combined data from the Nurses’ and Healthrofessionals follow-up studies (RR, 0.66; 95% CI, 0.42–1.03 forighest vs lowest quartile).99 In a nested case-control analysisithin the Finnish study, men in the highest tertile of serum

olate intake had a statistically significant (55%; 95% CI, 0.18 –.76) reduction in risk.100 Before conclusions can be reachedegarding fruit and vegetable intake, more data again are re-uired from other cohort studies, particularly on individualoods.

BeveragesReviewing the studies on the potential carcinogenicity

f alcohol and coffee, The International Agency for Research onancer concluded that there was little evidence to support

asual associations.101,102 Since these statements, similar dataave been reported in the US Health Professionals Follow Uptudy,103 The Nurses Health Study,103 The Japan Collaborativeohort Study for Evaluation of Cancer Risk,104 and The Swed-

sh Twin Registry Cohort Study.10 However, The Iowa Women’sealth Study, reported a nonsignificant positive associationith increased alcohol intake (P trend � .11) and a doubling of

he risk with an increased coffee consumption (RR, 2.2; 95% CI,.1– 4.3).78 The data that the risk of pancreatic cancer is similar

n patients with alcohol-related and non–alcohol-relatedhronic pancreatitis argues against a role for the direct effectsf alcohol in causes.

Anthropometric Measurements andPhysical ActivityAnthropometric measures including BMI, waist circum-

erence, and waist-hip ratio are dependent on several factorsncluding diet and physical activity. A meta-analysis of 6 case-ontrol and 8 cohort studies of 6391 cases of pancreatic cancereported an increased risk of 2% (95% CI, 1%–3%) for each unitncrease in BMI.105 Since this meta-analysis in 2003, 8 prospec-ive studies have been published with conflicting results. Theuropean Prospective Investigation Into Cancer and Nutritiontudy found a nonsignificant increased cancer risk with increas-

ng BMI (RR, 1.09; 95% CI, 0.95–1.24 per 5 kg/m2, and aignificant effect of both a larger waist-to-hip ratio (RR/0.1,.24; 95% CI, 1.04 –1.48) and waist circumference (RR/10 cm,.13; 95% CI, 1.01–1.26).80 Positive associations also were re-orted for increasing BMI and waist circumference in 2 Swedishopulation-based cohorts106 and in The American Cancer So-iety Cancer Prevention Study II Nutrition Cohort.107 In the US

ultiethnic Cohort Study, an increased risk was found in men,ut a protective association was found in women.108 However,

n 2 Japanese cohorts, one reported no effect of BMI at base-ine109 and in the second reported a protective effect that was

ound in men and no association in women.110 In the Iowa

Wopmvd

tNn2imiwiuapo2aao

psatictoIt1fepcttapddmotsuquaAcrmp

wudf

pccstcudpicstC

tmcFdwswc

March 2008 CAUSE OF PANCREATIC CANCER 279

omen’s Health Study, no effect of increasing BMI was foundn the risk for pancreatic cancer. Although most data suggest aositive association, inconsistencies need to be clarified thatay be related to the prevalence and definition of obesity and

ariations in the cut-off points used for analysis of BMI in theifferent studies.

Physical activity may affect the risk of pancreatic cancerhrough its effects on glucose metabolism and insulin levels.ine cohort studies in this area were identified, with 6 showingo effect,107–112 1 showing a nonsignificant protective effect,80 andshowing a protective effect of increased physical activity.53,81 An

nvestigation from the United States reported a protective effect ofoderate physical activity, with those in the highest quintile hav-

ng a significant 55% reduction in the risk of pancreatic cancerith a significant trend across quintiles (P � .001).81 This

nvestigation in male health professionals and female nursessed frequency of exercise questionnaires validated against di-ries of physical activity and physiologic measurements such asulse rate. Of the investigations that showed no effect, severalnly had limited questions on physical activity,107,109,110,112 andlacked generalizability.111,112 Before a conclusion on physical

ctivity and disease risk can be reached, further investigationsre required using validated questionnaires measuring all formsf physical activity in representative populations.

Potential Effects of DrugsAspirin inhibits cyclooxygenase enzymes, which may

revent carcinogenesis by enhancing cellular immune re-ponses, inhibiting prostaglandin synthesis, and influencingpoptosis, tumorigenesis, and angiogenesis.113 Aspirin also ei-her could increase or decrease the risk of pancreatic cancer byts ability to influence different lipoxygenase enzymes, whichan have either anticarcinogenic or procarcinogenic proper-ies.114 The only trial that has assessed aspirin in the incidencef pancreatic cancer was conducted in the Women’s Healthnitiative in the United States.115 In the Women’s Health Ini-iative, nearly 40,000 women were randomized to receive either00 mg aspirin every other day or placebo, with an averageollow-up time of 10.1 years. No statistically significant differ-nce was noted between test and placebo groups, with a re-orted relative risk of 1.42 (95% CI, 0.81–2.49) for pancreaticancer. However, the dose of aspirin was relatively small, wasaken intermittently, and only 73% of aspirin users took at leastwo thirds of the study medication. Whether higher doses ofspirin, taken more often, would be chemopreventive in otheropulations is unknown. A meta-analysis of aspirin use andisease risk, using data from 7 epidemiologic studies, plus theata from the earlier-reported trial, reported a summary esti-ate of 0.98 (95% CI, 0.86 –1.13).116 However, this conclusion

n aspirin may be difficult to interpret because not all assessedhe daily use of aspirin115,117,118 and several did not adjust formoking.117–119 The latter probably is important because aspirinse and smoking may be associated negatively and conse-uently the latter could mask any protective effects of aspirinse itself. However, 4 observational epidemiologic studies didssess the daily use of aspirin and adjusted for smoking.120 –123

cohort study in postmenopausal women reported a signifi-ant reduction in the incidence of pancreatic cancer with aelative risk of 0.40 (95% CI, 0.20 – 0.82) in those taking aspirin

ore than 6 times per week.120 In the other 3 studies either a

ositive,122 negative,121 or no effect123 was found, all of which

ere statistically not significant. The role of aspirin therefore isnclear and needs to be clarified in epidemiologic studies withetailed measures of the drug dose and consideration of con-ounders.

A second group of medications that could reduce the risk ofancreatic cancer are the statins (3-hydroxy-3-methylglutaryl–oenzyme A reductase inhibitors). These drugs reduce serumholesterol by inhibiting the rate-limiting step in cholesterolynthesis. Statins also have effects on many cellular processeshat may protect against the development and progression ofancer. In addition to cholesterol there are several other prod-cts derived from 3-hydroxy-3-methylglutaryl– coenzyme A re-uctase including farnesyl pyrophosphate and geranylgeranylyrophosphate. These affect guanosine triphosphatase signal-

ng proteins whose functions can influence cell proliferation. Aohort study of nearly half a million US veterans reported thattatin use of more than 6 months was associated with a reduc-ion in pancreatic cancer risk, with an odds ratio of 0.33 (95%I, 0.26 – 0.41; P � .01).124

ConclusionsSmoking and family history are unequivocal risk fac-

ors for pancreatic cancer. The data are consistent for diabetesellitus and chronic pancreatitis, but importantly the biologi-

al mechanisms are unknown and need further investigation.or most other exposures the evidence is either inconsistent orerived from relatively few studies. Clarification is required inhich there is, first, accurate measurement of exposures, and,

econd, consideration adjustment for other risk factors. Suchork is important to reduce the incidence of this highly fatal

ancer.

References

1. Parkin D, Bray F, Ferlay J, et al. Global cancer statistics. CACancer J Clin 2005;55:74–108.

2. Falk R, Pickle L, Fontham E, et al. Life-style risk factors forpancreatic cancer in Louisiana: a case-control study. Am J Epi-demiol 1988;128:324–336.

3. Ghadirian P, Liu G, Gallinger S, et al. Risk of pancreatic canceramong individuals with a family history of cancer of the pan-creas. Int J Cancer 2002;97:807–810.

4. Fernandez E, La Vecchia C, D’Avanzo B, et al. Family history andthe risk of liver, gallbladder, and pancreatic cancer. CancerEpidemiol Biomarkers Prev 1994;3:209–212.

5. Silverman D, Schiffman M, Everhart J, et al. Diabetes mellitus,other medical conditions and familial history of cancer as riskfactors for pancreatic cancer. Br J Cancer 1999;80:1830–1837.

6. Schenk M, Schwartz A, O’ Neal E, et al. Familial risk of pancre-atic cancer. J Natl Cancer Inst 2001;93:640–644.

7. Klein A, Brune K, Petersen G, et al. Prospective risk of pancre-atic cancer in familial pancreatic cancer kindreds. Cancer Res2004;64:2634–2683.

8. Lowenfels A, Maisonneuve P, DiMagno E, et al, and The Inter-national Heriditary Pancreatitis Study Group. Hereditary pancre-atitis and the risk of pancreatic cancer. J Natl Cancer Inst1997;89:442–446.

9. Giardiello F, Brensinger J, Tersmette A, et al. Very high risk ofcancer in familial Peutz-Jeghers syndrome. Gastroenterology2000;119:1447–1453.

10. Jenne D, Reimann H, Nezu J, et al. Peutz-Jegher’s syndrome iscaused by mutations in a novel serine threonine kinase. Nat

Genet 1998;18:38–43.

280 HART ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

11. Maisonneuve P, Marshall B, Lowenfels A. Risk of pancreaticcancer in patients with cystic fibrosis. Gut 2007;56:1327–1328.

12. Petrowsky H, Schuster H, Irani S, et al. Pancreatic cancer incystic fibrosis after bilateral lung transplantation. Pancreas2006;33:430–432.

13. Ojajarvi I, Partanen T, Ahlbom A, et al. Occupational exposuresand pancreatic cancer: a meta-analysis. Occup Environ Med2000;57:316–324.

14. Marsh GM. Critical review of epidemiological studies related toingested asbestos. Environ Health Perspect 1983;53:49–56.

15. Collins J, Esmen N, Hall T. A review and meta-analysis of form-aldehyde exposure and pancreatic cancer. Am J Ind Med 2001;39:336–345.

16. Yu H, Rohan T. Role of the insulin-like growth factor family incancer development and progression. J Natl Cancer Inst 2000;92:1472–1489.

17. Jones J, Clemmons D. Insulin-like growth factors and their bind-ing proteins: biological actions. Endocr Rev 1995;16:3–34.

18. Holmes M, Pollak M, Hankinson SE. Life-style correlates ofplasma insulin-like growth factor 1 and insulin-like growth factorbinding protein 3 concentrations. Cancer Epidemiol BiomarkersPrev 2002;11:862–867.

19. Silverman D, Swanson C, Gridely G, et al. Dietary and nutritionalfactors and pancreatic cancer: a case-control study based ondirect interviews. J Natl Cancer Inst 1998;90:1710–1719.

20. Friedman G, Van Den Eeden S. Risk factors for pancreaticcancer: an explanatory study. Int J Epidemiol 1993;22:30–37.

21. Toyokuni S. Iron-induced carcinogenesis: the role of redox reg-ulation. Free Radic Biol Med 1996;20:533–566.

22. Bueno de Mesquita H, Maisonneuve P, Moerman C, et al.Anthropometric and reproductive variables and exocrine carci-noma of the pancreas: a population-based case-control study inThe Netherlands. Int J Cancer 1992;52:24–29.

23. Teras L, Patel A, Rodriguez C, et al. Parity, other reproductivefactors, and risk of pancreatic cancer mortality in a large cohortof U.S. women (United States). Cancer Causes Control 2005;16:1035–1040.

24. Fernandez E, La Vecchia C, D’Avanzo B, et al. Menstrual andreproductive factors and pancreatic cancer risk in women. Int JCancer 1995;62:11–14.

25. Kreiger N, LaCroix J, Sloan M. Hormonal factors and pancreaticcancer in women. Ann Epidemiol 2001;11:563–567.

26. Skinner HG, Michaud D, Colditz G, et al. Parity, reproductivefactors, and the risk of pancreatic cancer in women. CancerEpidemiol Biomarkers Prev 2003;12:433–438.

27. Karlson B, Wuu J, Hsieh C, et al. Parity and the risk of pancreaticcancer: a nested case-control study. Int J Cancer 1988;77:224–227.

28. Kvale G, Heuch I, Nilssen S. Parity in relation to mortality andcancer incidence: a prospective study of Norwegian women. IntJ Epidemiol 1994;23:691–699.

29. Lin Y, Kikuchi S, Tamakoshi A, et al. Association of menstrualand reproductive factors with pancreatic cancer risk in women:findings of the Japan Collaborative Cohort Study for evaluationof cancer risk. J Gastroenterol 2006;41:878–883.

30. Huxley R, Ansary-Moghaddam A, Berrington de Gonzalez A, et al.Type-II diabetes and pancreatic cancer: a meta-analysis of 36studies. Br J Cancer 2005;92:2076–2083.

31. Ohmura E, Okada M, Onoda N, et al. Insulin-like growth factor Iand transforming growth factor alpha as autocrine growth fac-tors in human pancreatic cell growth. Cancer Res 1990;50:103–107.

32. Permert J, Ihse I, Jorfeldt L, et al. Improved glucose metabolismafter subtotal pancreatectomy for pancreatic cancer. Br J Surg1993;80:1047–1050.

33. Zendehdel K, Nyren O, Ostenson C, et al. Cancer incidence in

patients with type 1 diabetes mellitus: a population-basedcohort study in Sweden. J Natl Cancer Inst 2003;95:1797–1800.

34. Ekbom A, McLaughlin J, Karlsson B-M, et al. Pancreatitis andpancreatic cancer: a population-based study. J Natl Cancer Inst1994;86:625–627.

35. Malka D, Hammel P, Maire F, et al. Risk of pancreatic adeno-carcinoma in chronic pancreatitis. Gut 2002;51:849–852.

36. Karlson B-M, Ekbom A, Josefsson S, et al. The risk of pancreaticcancer following pancreatitis: an association due to confound-ing? Gastroenterology 1997;113:587–592.

37. Bansal P, Sonnenberg A. Pancreatitis is a risk factor for pancre-atic cancer. Gastroenterology 1995;109:247–251.

38. Lowenfels A, Maisonneuve P, Calallini G, et al. Pancreatitis andthe risk of pancreatic cancer. N Engl J Med 1993;328:1433–1437.

39. Shibata A, Mack TM, Paganini-Hill A, et al. A prospective study ofpancreatic cancer in the elderly. Int J Cancer 1994;58:46–49.

40. Cuzick J, Babiker A. Pancreatic cancer, alcohol, diabetes melli-tus and gall-bladder disease. Int J Cancer 1989;43:415–421.

41. Ekbom A, Yuen J, Karlsson BM, et al. Risk of pancreatic andperiampullar cancer following cholecystectomy—a population-based cohort study. Dig Dis Sci 1996;41:387–391.

42. Kalapothaki V, Tzonou A, Hsieh C, et al. Tobacco, ethanol,coffee, pancreatitis, diabetes mellitus, and cholelithiasis asrisk factors for pancreatic carcinoma. Cancer Causes Control1993;4:375–382.

43. Chow W, Johansen C, Gridley G, et al. Gallstones, cholecystec-tomy and risk of cancers of the liver, biliary tract and pancreas.Br J Cancer 1999;79:640–644.

44. Gullo L. Risk of pancreatic cancer and periampullary cancerfollowing cholecystectomy. Ann Oncol 1999;10(Suppl 4):127–128.

45. Ye W, Lagergren J, Nyren O, et al. Risk of pancreatic cancer aftercholecystectomy: a cohort study in Sweden. Gut 2001;49:678–681.

46. Haines AP, Moss AR, Whittemore A, et al. A case-control studyof pancreatic carcinoma. J Cancer Res Clin Oncol 1982;103:93–97.

47. Bueno de Mesquita H, Maisonneuve P, Moerman C, et al.Aspects of medical history and exocrine carcinoma of the pan-creas: a population-based case-control study in the Nether-lands. Int J Cancer 1992;52:17–23.

48. Maringhini A, Moreau J, Melton L, et al. Gallstones, gallbladdercancer, and other gastrointestinal malignancies. Ann Intern Med1987;107:30–35.

49. Ichimiya H, Kono S, Ikeda M, et al. Cancer mortality amongpatients undergoing cholecystectomy for benign biliary dis-eases. Jap J Cancer Res 1986;77:579–583.

50. Rosenberg L, Duguid W, Brown RA. Cholecystectomy stimulateshypertrophy and hyperplasia in the hamster pancreas. 1984;37:108–111.

51. Watanapa P, Williamson R. Experimental pancreatic hyperplasiaand neoplasia: effects of dietary and surgical manipulation. Br JCancer 1993;67:877–884.

52. Schernhammer E, Michaud D, Leitzmann M, et al. Gallstones,cholecystectomy, and the risk for developing pancreatic cancer.Br J Cancer 2002;86:1081–1084.

53. Stolzenberg-Solomon R, Pietinen P, Taylor P, et al. A prospectivestudy of medical conditions, anthropometry, physical activity,and pancreatic cancer in male smokers (Finland). CancerCauses Control 2002;13:417–426.

54. Vasen H, Gruis N, Frants R, et al. Risk of developing pancreaticcancer in families with familial atypical multiple mole melanomaassociated with a specific 19 deletion of p16 (p16-Leiden). IntJ Cancer 2000;87:809–811.

55. Neugut A, Ahsan H, Robinson E. Pancreas cancer as a second

March 2008 CAUSE OF PANCREATIC CANCER 281

primary malignancy. A population-based study. Cancer 1995;76:589–592.

56. Annibale B, Capurso G, Delle Fave G. Consequences of Helico-bacter pylori infection on the absorption of micronutrients. DigLiver Dis 2002;34(Suppl 2):S72–S77.

57. Jensen RT. Involvement of cholecystokinin/gastrin-related pep-tides and their receptors in clinical gastrointestinal disorders.Pharmacol Toxicol 2002;91:333–350.

58. Risch HA. Etiology of pancreatic cancer with a hypothesis con-cerning the role of N-nitroso compounds and excess gastricacidity. J Natl Cancer Inst 2003;95:948–960.

59. Stolzenberg-Solomon R, Blaser MLP, Perez-Perez G, et al. Heli-cobacter pylori seropositivity as a risk factor for pancreaticcancer. J Natl Cancer Inst 2001;93:937–941.

60. Raderer M, Wrba F, Kornek G, et al. Association between Heli-cobacter pylori infection and pancreatic cancer. Oncology 1998;55:16–19.

61. Nilsson H, Stenram U, Ihse I, et al. Re: Helicobacter pyloriseropositivity as a risk factor for pancreatic cancer. J NatlCancer Inst 2002;94:632–633.

62. Tascilar M, van Rees BP, Sturm PDJ, et al. Pancreatic cancerafter remote peptic ulcer surgery. J Clin Pathol 2002;55:340–345.

63. Mack TM, Yu MC, Hanisch R, et al. Pancreas cancer and smok-ing, beverage consumption and past medical history. J NatlCancer Inst 1986;76:49–60.

64. Ross AHM, Smith MA, Anderson JR, et al. Late mortality aftersurgery for peptic ulcer. N Engl J Med 1982;307:519–522.

65. Lowenfels A, Maisonneuve P. Epidemiology and risk factors forpancreatic cancer. Best Pract Res Clin Gastroenterol 2006;20:197–209.

66. Rivenson A, Hoffmann D, Prokopczyk B, et al. Induction of lungand exocrine pancreas tumours in F344 rats by tobacco-specificand areca-derived N-nitrosamines. Cancer Res 1988;48:6912–6917.

67. Coughlin S, Calle E, Patel A, et al. Predictors of pancreaticcancer mortality among a large cohort of United States adults.Cancer Causes Control 2000;11:915–923.

68. Doll R, Peto R, Wheatley K, et al. Mortality in relation to smok-ing: 40 years’ observation on male British doctors. BMJ 1994;309:901–911.

69. Zheng W, McLaughlin J, Gridley G, et al. A cohort study ofsmoking, alcohol consumption, and dietary factors for pancre-atic cancer (United States). Cancer Causes Control 1993;4:477–482.

70. MacMahon B, Yen S, Trichopoulos D, et al. Coffee and cancer ofthe pancreas. N Engl J Med 1981;304:630–633.

71. Wynder EL, Hall NEL, Polansky M. Epidemiology of coffee andpancreatic cancer. Cancer Res 1983;43:3900–3906.

72. Whittemore AS, Paffenbarger RS, Anderson K, et al. Early pre-cursors of pancreatic cancer in college men. J Chron Dis 1983;36:251–256.

73. Lin Y, Tamakoshi A, Kawamura T, et al. A prospective cohortstudy of cigarette smoking and pancreatic cancer in Japan.Cancer Causes Control 2002;13:249–254.

74. Silverman D, Dunn J, Hoover R, et al. Cigarette smoking andpancreas cancer: a case-control study based on direct inter-views. J Natl Cancer Inst 1994;86:1510–1516.

75. International Agency for Research on Cancer. Tobacco smoking.IARC Monogr Eval Carcinog Risks Hum 1986;38:83–126.

76. Villeneuve P, Johnson K, Mao Y, et al. Environmental tobaccosmoke and the risk of pancreatic cancer: findings from a Cana-dian population-based case-control study. Can J Public Health2004;95:32–37.

77. Stolzenberg-Solomon R, Pietinen P, Taylor P, et al. Prospectivestudy of diet and pancreatic cancer in male smokers. Am J

Epidemiol 2002;155:783–792.

78. Harnack L, Anderson K, Zheng W, et al. Smoking, alcohol,coffee, and tea intake and incidence of cancer of the exocrinepancreas: The Iowa Women’s Health Study. Cancer EpidemiolBiomarkers Prev 1997;6:1081–1086.

79. Stolzenberg-Solomon R, Pietinen P, Taylor P, et al. A prospectivestudy of medical conditions, anthropometry, physical activity,and pancreatic cancer in male smokers (Finland). CancerCauses Control 2002;13:417–426.

80. Berrinton de Gonzalez A, Spencer E, Bueno-de-Mesquita HB,et al. Anthropometry, physical activity, and the risk of pancreaticcancer in the European Prospective Investigation into Cancerand Nutrition. Cancer Epidemiol Biomarkers Prev 2006;15:879–885.

81. Michaud D, Giovannucci E, Willett W, et al. Physical activity,obesity, height, and the risk of pancreatic cancer. JAMA 2001;286:921–929.

82. Yoshimoto M, Tsutsumi M, Iki K, et al. Carcinogenicity of het-erocyclic amines for the pancreatic duct epithelium in hamsters.Cancer Lett 1999;143:235–239.

83. Larsson S, Hakanson N, Permert J, et al. Meat, fish, poultry andegg consumption in relation to risk of pancreatic cancer; aprospective study. Int J Cancer 2006;118:2866–2870.

84. Nothlings U, Wilkens L, Murphy SP, et al. Meat and fat intake asrisk factors for pancreatic cancer: the multiethnic cohort study.J Natl Cancer Inst 2005;97:1458–1465.

85. Michaud DGE, Willett W, Colditz G, et al. Dietary meat, dairyproducts, fat, and cholesterol and pancreatic cancer risk in aprospective study. Am J Epidemiol 2003;157:1115–1125.

86. Isaksson B, Jonsson F, Pedersen N, et al. Lifestyle factors andpancreatic cancer risk: a cohort study from the Swedish TwinRegistry. Int J Cancer 2002;98:480–482.

87. Stolzenberg-Solomon R, Graubard B, Chari S, et al. Insulin,glucose, insulin resistance, and pancreatic cancer in malesmokers. JAMA 2005;294:2872–2878.

88. Jee S, Ohrr H, Sull J, et al. Fasting serum glucose level andcancer risk in Korean men and women. JAMA 2005;293:194–202.

89. Michaud D, Liu S, Giovannucci E, et al. Dietary sugar, glycaemicload, and pancreatic cancer risk in a prospective study. J NatlCancer Inst 2002;94:1293–1300.

90. Silvera S, Rohan T, Jain M, et al. Glycaemic index, glycaemicload, and pancreatic cancer risk (Canada). Cancer Causes Con-trol 2005;16:431–436.

91. Larsson S, Bergkvist L, Wolk A. Consumption of sugar andsugar-sweetened foods and the risk of pancreatic cancer in aprospective study. Am J Clin Nutr 2006;84:1171–1176.

92. Schernhammer E, Hu F, Giovannucci E, et al. Sugar-sweetenedsoft drink consumption and risk of pancreatic cancer in twoprospective cohorts. Cancer Epidemiol Biomarkers Prev 2005;14:2098–2105.

93. Roebuck B, Kaplita P, Edwards B, et al. Effects of dietary fatsand soybean protein on azaserine-induced pancreatic carcino-genesis and plasma cholecystokinin in the rat. Cancer Res1987;47:1333–1338.

94. Mills PK, Beeson WL, Abbey DE, et al. Dietary habits and pastmedical history as related to fatal pancreas cancer amongAdventists. Cancer 1988;61:2578–2585.

95. Sauvaget C, Nagano J, Hayashi M, et al. Vegetables and fruitintake and cancer mortality in the Hiroshima/Nagasaki LifeSpan Study. Br J Cancer 2003;88:689–694.

96. Larsson S, Hakansson N, Naslund I, et al. Fruit and vegetableconsumption in relation to pancreatic cancer risk: a prospectivestudy. Cancer Epidemiol Biomarkers Prev 2006;15:301–305.

97. Larsson S, Hakansson N, Giovannucci E, et al. Folate intake andpancreatic cancer incidence: a prospective study of Swedishwomen and men. J Natl Cancer Inst 2006;98:407–413.

98. Stolzenberg-Solomon R, Pietinen P, Barrett M, et al. Dietary and

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

1

PU

282 HART ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

other methyl-group availability factors and pancreatic cancerrisk in a cohort of male smokers. Am J Epidemiol 2001;153:680–687.

99. Skinner HG, Michaud D, Giovannucci E, et al. A prospectivestudy of folate intake and the risk of pancreatic cancer in menand women. Am J Epidemiol 2004;160:248–258.

00. Stolzenberg-Solomon R, Albanes D, Nieto F, et al. Pancreaticcancer risk and nutrition-related methyl-group availability indica-tors in male smokers. J Natl Cancer Inst 1999;91:535–541.

01. International Agency for Research on Cancer. Coffee, tea, mate,methylxanthines and methylglyoxal. IARC Monogr Eval CarcinogRisks Hum 1991;51:1–513.

02. International Agency for Research on Cancer. Alcohol drinking.IARC Monogr Eval Carcinog Risks Hum 1988;44:1–378.

03. Michaud D, Giovannucci E, Willett W, et al. Coffee and alcoholconsumption and the risk of pancreatic cancer in two prospec-tive United States Cohorts. Cancer Epidemiol Biomarkers Prev2001;10:429–437.

04. Lin Y, Tamakoshi A, Kawamura T, et al. Risk of pancreaticcancer in relation to alcohol drinking, coffee consumption andmedical history: findings from the Japan Collaborative cohortstudy for evaluation of cancer risk. Int J Cancer 2002;99:742–746.

05. Berrington de Gonzalez A, Sweetland S, Spencer E. A meta-analysis of obesity and the risk of pancreatic cancer. Br JCancer 2003;89:519–523.

06. Larsson S, Permert J, Hakansson N, et al. Overall obesity,abdominal obesity, diabetes and cigarette smoking in relationto the risk of pancreatic cancer in two Swedish population-based cohorts. Br J Cancer 2005;93:1310–1315.

07. Patel A, Rodriguez C, Bernstein L, et al. Obesity, recreationalphysical activity, and risk of pancreatic cancer in a large USCohort. Cancer Epidemiol Biomarkers Prev 2005;14:459–466.

08. Nothlings U, Wilkens L, Murphy S, et al. Body mass index andphysical activity as risk factors for pancreatic cancer: the Multi-ethnic Cohort Study. Cancer Causes Control 2007;18:165–175.

09. Lin Y, Kikuchi S, Tamakoshi A, et al. Obesity, physical activityand the risk of pancreatic cancer in a large Japanese cohort. IntJ Cancer 2007;120:2665–2671.

10. Luo J, Iwasaki M, Inoue M, et al. Body mass index, physicalactivity and the risk of pancreatic cancer in relation to smokingstatus and history of diabetes: a large-scale population-basedcohort study in Japan—the JPHC study. Cancer Causes Control2007;18:603–612.

11. Lee I, Sesso H, Oguma Y, et al. Physical activity, body weight,and pancreatic cancer mortality. Br J Cancer 2003;88:679–

683. e

12. Sinner P, Schmitz K, Anderson K, et al. Lack of association ofphysical activity and obesity with incident pancreatic cancer inelderly women. Cancer Epidemiol Biomarkers Prev 2005;14:1571–1573.

13. Dannenberg A, Altorki N, Boyle J, et al. Cyclo-oxygenase 2: apharmacological target for the prevention of cancer. LancetOncol 2001;2:544–551.

14. Ding X, Kuszynski C, El-Metwally T, et al. Lipoxygenase inhibitioninduced apoptosis, morphological changes, and carbonic anhy-drase expression in human pancreatic cancer cells. BiochemBiophys Res Commun 1999;266:392–399.

15. Cook N, Lee I, Gaziano JM, et al. Low-dose aspirin in the primaryprevention of cancer: the women’s health study: a randomisedcontrolled trial. JAMA 2005;294:47–55.

16. Larsson S, Giovannucci E, Bergkvist L, et al. Aspirin and non-steroidal anti-inflammatory drug use and risk of pancreatic can-cer: a meta-analysis. Cancer Epidemiol Biomarkers Prev 2006;1:2561–2564.

17. Schreinemachers D, Everson RB. Aspirin use and lung, colonand breast cancer incidence in a prospective study. Epidemiol-ogy 1994;5:138–146.

18. Ratnasinghe L, Graubard B, Kahle L, et al. Aspirin use andmortality from cancer in a prospective cohort study. AnticancerRes 2004;24:3177–3184.

19. Friis S, Sorensen H, McLaughlin J, et al. A population-basedcohort study of the risk of colorectal and other cancers amongusers of low-dose aspirin. Br J Cancer 2003;88:684–688.

20. Anderson K, Johnson T, Lazovich D, et al. Association betweennonsteroidal anti-inflammatory drugs and the incidence of pan-creatic cancer. J Natl Cancer Inst 2002;94:1168–1171.

21. Menezes R, Huber K, Mahoney M, et al. Regular use of aspirinand pancreatic cancer risk. BMC Public Health 2002;2:18.

22. Schernhammer E, Kang J, Chan A, et al. A prospective study ofaspirin use and the risk of pancreatic cancer in women. J NatlCancer Inst 2004;96:22–28.

23. Jacobs E, Connell C, Rodriquez C, et al. Aspirin use and pan-creatic cancer mortality in a large United States cohort. J NatlCancer Inst 2004;96:524–528.

24. Khurana V, Sheth A, Caldito G, et al. Statins reduce the risk ofpancreatic cancer in humans: a case-control study of half amillion veterans. Pancreas 2007;34:260–265.

Address requests for reprints to: Dr Andrew R. Hart, MB, CHB.opulation Health, School of Medicine, Health Policy and Practice,niversity of East Anglia, Norwich, Norfolk, NR4 7TJ, United Kingdom.

-mail: a.hart@uea.ac.uk; fax: (0044) 1603-593752.

C

CS

J

*

Crtsic

CwdcaiccWesl2mPpodotgauifi

Caitefl

oon7

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2008;6:283–289

LINICAL IMAGING

omputerized Tomography Enterography and Its Role inmall-Bowel Imaging

OEL G. FLETCHER,* JAMES HUPRICH,* EDWARD V. LOFTUS, JR,‡ DAVID H. BRUINING,‡ and JEFF L. FIDLER*

Department of Radiology, and ‡Division of Gastroenterology & Hepatology, Mayo Clinic Rochester, Rochester, Minnesota

pCfg

atatsdpmgwccdpomancelioc

ppr

G

omputerized tomography enterography is being adoptedapidly by many institutions as the primary technique usedo image the small bowel. The purpose of this article is toummarize how the examination is performed, and to spec-fy appropriate indications and alternatives, performanceharacteristics, and anticipated future developments.

omputerized tomography (CT) enterography is a CT ofthe abdomen and pelvis designed to image the lumen and

all of the small bowel. Such optimization requires luminalistention (with a larger than normal volume of ingested oralontrast material), high-resolution imaging, and the appropri-te phase of intravenous contrast enhancement. Although pos-tive (high attenuation) oral contrast can be used, the oralontrast used for CT enterography is generally a neutral entericontrast, meaning that its CT attenuation is similar to water.

hen used with iodinated intravenous contrast agents at CTnterography, neutral oral contrast agents maximize the con-picuity of hypervascular pathologies such as inflamed boweloops or masses. For a CT enterography examination, 1500 to000 mL of oral contrast is ingested over approximately 60inutes before the examination to distend the small bowel.

olyethylene glycol solution and low-concentration barium sus-ension (which contains sorbitol) are the most commonly usedral contrast agents because they result in superior small-bowelistention relative to water, but may cause self-limited diarrhear gas shortly after the examination.1 Although the examina-ion is highly robust in an outpatient setting, gastroenterolo-ists should be aware that hospitalized patients or patients whore unable to drink large volumes of contrast usually cannotndergo CT enterography successfully. Contraindications also

nclude severe contrast dye allergies and significant renal insuf-ciency.

To adequately visualize the small-bowel lumen and wall atT enterography, slice thicknesses of 3 mm or less are used inn overlapping fashion. Because of the increased number ofmages generated during a CT enterography examination, gas-roenterologists will find it most helpful to review images in anlectronic fashion, with routine coronal images generated, toacilitate visualization of tortuous small-bowel loops and fistu-ae.

Maximal mural enhancement of the normal small bowelccurs 50 seconds after intravenous contrast material injectionr 14 seconds after peak aortic enhancement2; CT imagingormally is initiated at this time. However, CT scanning up to

0 seconds after contrast injection permits routine hepatic

hase imaging, but does not compromise the ability to identifyrohn’s disease.3,4 Imaging of the small bowel may be per-

ormed in multiple vascular phases in the setting of occultastrointestinal (GI) bleeding (see later).

Normal Findings at ComputerizedTomography EnterographyGastroenterologists should be familiar with the normal

ppearance of the small bowel at CT enterography. Althoughhe jejunum possesses the valvulae conniventes, which are seens thin, closely spaced folds resulting in a feathery appearance,he ileum possesses few, if any, folds. The small-bowel wallhould be less than 3 mm in thickness when the lumen isistended by fluid. When imaging is performed in the enterichase of maximal small-bowel enhancement, the jejunum nor-ally will have greater attenuation than the ileum, with a

radual decrease in attenuation of the enhancing small-bowelall as one moves toward the terminal ileum. Intramural fat

an be a normal finding in the terminal ileum, but is a sign ofhronic inflammation elsewhere in the small bowel. In Crohn’sisease, it often is accompanied by mucosal hyperenhancement,ointing to both active and chronic inflammation. The jejunumccasionally is collapsed at enterography, which can be a nor-al finding in the minority of cases, but the ileum almost

lways is distended adequately.5 Notwithstanding this weak-ess, CT enterography is acceptable for most small-bowel indi-ations, particularly given the fact that many small-bowel dis-ases, such as Crohn’s, or masses will mildly obstruct theumen as a result of transmural inflammation, edema, ornvolvement. Moreover, mural and mesenteric pathologiesften are occult at luminal imaging such as fluoroscopy orapsule endoscopy.

Imaging Findings at ComputerizedTomography EnterographyAlthough multiple indications exist for CT enterogra-

hy (Table 1), the overwhelming indication is known or sus-ected Crohn’s disease. In this setting, the goal of CT enterog-aphy is to identify active Crohn’s disease, CT findings

Abbreviations used in this paper: CT, computerized tomography;I, gastrointestinal; MR, magnetic resonance.

© 2008 by the AGA Institute1542-3565/08/$34.00

doi:10.1016/j.cgh.2007.12.049

ifi(oephsasisitMbmhti

twafpmt

bsi2p

tPttgwa(pteradl

fwnMmppt

Ccistoltpe

fs

FN

T

S

M

284 FLETCHER ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

ndicating bowel wall and adjacent mesenteric inflammation,stulizing disease (including abscess), small-bowel obstruction

if present), and any extraintestinal manifestations. Findingsf active inflammatory Crohn’s disease include mural hyper-nhancement, thickening, and stratification, in addition toeri-enteric fat stranding and the comb sign. Mural hyperen-ancement is segmental attenuation greater than adjacentmall-bowel loops and is the most sensitive CT finding forctive inflammatory Crohn’s disease.6 Mural thickening is con-idered present when the small-bowel wall is greater than 3 mmn thickness and the lumen is filled with oral contrast. Althoughegmental mural hyperenhancement alone is a nonspecific find-ng for Crohn’s disease, asymmetric mural enhancement andhickening are pathognomonic for Crohn’s disease (Figure 1).

ural stratification refers to a laminated appearance to theowel wall and is seen with more advanced small-bowel inflam-ation (Figure 2).7 With mural stratification, the mucosa is

yperenhancing, with the submucosa being of lower attenua-ion representing edema (water attenuation) or inflammatorynfiltrate (soft-tissue attenuation).

Because the transmural inflammation of Crohn’s disease ex-ends into the mesentery, there is stranding in the peri-enteric fat,hich is associated with increased serum concentrations of C-re-ctive protein.8 Fibro-fatty proliferation is distinct from increasedat density and refers to the surgical and imaging findings of fattyroliferation along the mesenteric side of the bowel, whichay participate in the Crohn’s inflammatory response, and

ypically reflects chronic Crohn’s inflammation.9

The comb sign heralds significant inflammation in the smallowel or colon and refers to engorged vasa recta (or the vesselsupplying the bowel).10 When these vessels become enlarged dur-ng acute inflammation, they resemble the teeth of a comb (Figure). The comb sign is associated with increased serum C-reactiverotein levels and length of hospital stay during Crohn’s flares.11

Penetrating disease is seen frequently at CT enterography inhe Crohn’s disease population (up to 20% of the time12).enetrating complications, particularly entero-enteric and en-erocolic fistulas, frequently are unsuspected clinically yet iden-ified incidentally during CT enterography (Figure 2). Fistulasenerally appear as hyperenhancing extra-enteric tracks with orithout air or fluid.13 Although routine CT can detect fistulaend abscess,14 CT enterography can describe the type of fistulaie, its connections), its size, whether or not it is associated withroximal obstruction, whether it arises from active inflamma-ory disease, or whether it is associated with abscess.13 In ourxperience, smaller fistulas associated with active disease willesolve with medical treatment. Perianal fistulas or fistulasrising in the postoperative setting unassociated with Crohn’sisease may not be hyperenhancing, potentially because of their

able 1. Indications for CT Enterography

Type of examination Indication

ingle-phase CT enterography Rule out Crohn’s diseaseStage Crohn’s disease activityRule out small-bowel massAssess for complications of celiac

sprueUnexplained diarrhea

ultiphase CT enterography Occult GI bleeding

ack of acute inflammatory response. b

In a large retrospective review of patients, Bruining et al12

ound that extraintestinal manifestations of Crohn’s diseaseere seen in about 12% to 15% of patients, including sacroiliitis,ephrolithiasis, and portal or mesenteric vein thrombosis.oreover, non–inflammatory bowel disease–related abnor-alities including pulmonary nodules, adrenal nodules, lym-

homa, and other malignancies were detected. Certainly,atients with Crohn’s disease also have an increased suscep-ibility to enteric and hepatic malignancies and lymphoma.15

CT enterography also can be used for a variety of non-rohn’s indications. Besides active Crohn’s disease, the most

ommon finding at CT enterography is a mass. Masses cannclude carcinoid tumors, adenocarcinomas, lymphomas, GItromal tumors, and Meckel’s diverticula (and their complica-ions). Carcinoid tumors are the most frequently seen tumors inur practice, and are seen as hyperenhancing polypoid or carpet

esions that can invade the mesentery, with characteristic soft-issue attenuation mesenteric metastases (Figure 3). The ap-earance of other small-bowel tumors and celiac disease at CTnterography has been described elsewhere.6

Performance of ComputerizedTomography Enterography for ActiveCrohn’s DiseaseBefore 2000, sensitivity estimates of CT enterography

or active inflammatory Crohn’s disease were based largely onurgical or fluoroscopic series without a reliable endoscopic

igure 1. Asymmetric mural hyperenhancement in Crohn’s disease.ote additional presence of intramural fat, indicating there also has

een chronic inflammation.

rstasagirr

fdrs.taoe

(

March 2008 CT ENTEROGRAPHY 285

eference standard. Since that time, most assessments havehown a sensitivity for active inflammatory Crohn’s disease onhe order of 80% to 90%,3,4,16,17 with some showing sensitivityround 60%.18,19 When quantitative and visual measures ofegmental hyperenhancement are compared with endoscopynd small-bowel biopsy, the odds ratio for active disease isreater than 10, and the sensitivity for histologic inflammations about 80%, similar to endoscopic assessment alone.3 Weecently conducted a head-to-head trial comparing CT enterog-

aphy, capsule endoscopy, ileocolonoscopy, and small-bowel 2

ollow-through in 42 patients with suspected or known Crohn’sisease. The sensitivity of capsule endoscopy and CT enterog-aphy was almost identical (83% and 82%, respectively), with thepecificity of CT being significantly higher (89% vs 53%; P �02).16 In addition, although we excluded patients with symp-oms of obstructive disease, 17% of the patients had asymptom-tic strictures, which precluded capsule assessment. In a sec-ndary analysis of 27 patients who received both capsulendoscopy and CT enterography, capsule endoscopy identified

igure 2. (A) Enterocolic fistula (arrows) arising from the top of annflamed jejunal loop (large arrowhead) and connecting to the hepaticexure. (B) The proximal jejunum shows mural stratification (large ar-ow), with the distal jejunum being decompressed (arrowhead). (B) En-orged vasa recta (or the comb sign; small arrows) supply the (C) mildly

nflamed proximal jejunal loop (arrows), which contains an enterolitharrowhead).

Fiflrgi

additional patients with active Crohn’s disease who had no

estctpbi

eppidrsC

CttecfcfCsirrCtfimmea

frConnapeippoiateecaCa

Fi(t(mo

286 FLETCHER ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

vidence of Crohn’s disease at CT imaging, but again showedignificantly lower specificity (53% vs 100%). Although a recentrial by Voderholzer et al18 showed superior performance ofapsule endoscopy versus CT enteroclysis (ie, with a nasojejunalube) for patients undergoing both tests, 27% of their studyopulation was excluded from undergoing capsule endoscopyecause of small-bowel strictures at CT enteroclysis. Moreover,

igure 3. Ileal carcinoid with typical mesenteric metastasis. (A) Axialmage shows an ileal loop with focal mural thickening and enhancementarrowhead) associated with a soft-tissue mesenteric mass with excen-ric calcifications and stranding to adjacent small-bowel loops (arrow).B) Coronal image shows similar findings (bowel loops, arrowheads;

esenteric metastasis, arrow). The surgical specimen showed a clusterf mural carcinoids with metastases to local nodes.

n the 41 patients who underwent CT enterography and capsule e

ndoscopy, the capsule failed to reach the colon in 25% of theatients. In the Voderholzer et al18 experience, only 1 of 41atients had a retained capsule. In summary, CT enterography

s moderately sensitive and highly accurate for active Crohn’sisease and can exclude significant strictures that may result inetained capsules. Nevertheless, capsule endoscopy may identifyome patients with mild active inflammatory disease occult atT imaging.

Clinical Contribution of ComputerizedTomography EnterographyCT enterography provides objective evidence of active

rohn’s disease or small-bowel mass proximal to the reach ofhe endoscope, as well as providing visualization of the mesen-ery and liver. Studies examining the clinical benefit of CTnterography are beginning to emerge.12,20 Higgins et al20 re-ently had gastroenterologists review clinical information be-ore the revelation of CT enterography findings, to make alinical assessment on whether or not patients would benefitrom corticosteroid treatment. Subsequent to these predictions,T enterography information was provided. Their study

howed that CT enterography did not replicate original clin-cal impressions, and changed the impression of corticoste-oid benefit in 61% of these patients. Similarly, in our expe-ience, approximately half of the patients with penetratingrohn’s disease had no symptoms or signs of a fistula at

heir initial clinical assessment.21 Detection of unsuspectedstulizing Crohn’s disease resulted in either escalation ofedical therapy or surgical/procedural intervention in theajority of these patients. Prospective trials detailing the

ffect of CT enterography results on clinical decision makingre needed.

Multiphase Computerized TomographyEnterography for OccultGastrointestinal BleedingAlthough earlier radiologic literature was disappointing

or the ability of CT to detect the causes of occult GI bleeding,22

ecent work with multiphase CT enterography has shown thatT has an important role to play in this clinical setting. Becausef the requirement for rapid multiphase imaging, this tech-ique is best performed on a 64-channel (or higher) CT scan-er.23 Neutral oral contrast and iodinated intravenous contrastre the same as for a single-phase examination. Scanning iserformed from the diaphragm to the symphysis pubis duringach of 3 phases, with scanning initiated when a region-of-nterest attenuation threshold in the aorta is reached. The firsthase is a bolus-triggered arterial phase, followed by an enterichase (20 seconds after trigger), and a delayed phase (70 sec-nds after trigger). Once the region-of-interest trigger threshold

s reached by iodinated contrast arriving in the aorta, therterial phase is initiated after a short 6-second delay. Duringhe arterial phase, vascular ectasias such as AVMs and theirarly draining veins are seen. The enteric phase often highlightsnhancing tumors, with delayed images showing that iodinatedontrast is accumulating in the small-bowel lumen, indicatingctive bleeding (Figure 4). Huprich et al23 showed the benefit ofT enterography in identifying multiple small-bowel tumorsnd vascular lesions detected and undetected at capsule

ndoscopy, with a larger prospective study underway. In the

sepm

orpfc

eCefoarispaffl

c

March 2008 CT ENTEROGRAPHY 287

etting of occult GI bleeding, CT enterography and capsulendoscopy are not competitive imaging modalities but com-lementary ones, because findings from each imaging studyay be nonspecific.

Appropriate Use of Cross-SectionalAlternatives to ComputerizedTomography EnterographyTable 2 lists alternative tests preferred over CT enter-

graphy in certain clinical settings. Because CT and magneticesonance (MR) enteroclysis (after nasojejunal tube placement)ermit the rapid infusion of large volumes of enteric contrast ataster rates, these examinations better show small-bowel caliber

hanges reflecting low-grade obstruction. Similarly, because CT (

nterography can result in collapsed jejunal loops, we performT or MR enteroclysis to investigate further nonspecific capsule

ndoscopy findings in the jejunum. CT enteroclysis is not per-ormed for patients with known or suspected Crohn’s disease orccult GI bleeding because of the acceptable performance char-cteristics of CT enterography, patient objections to, and timeequired for nasojejunal tube placement. Because of the markedncrease in signal differences between pathology and anatomictructures near the anus, we perform MR imaging to stageerianal disease or investigate potential J-pouch complicationsfter ileal pouch–anal anastomosis. MR enterography is pre-erred to CT enterography when imaging to determine if in-ammed bowel loops or fistulas have responded to therapy

igure 4. Active bleeding at multiphase CT enterography. (A) Arterial,B) enteric, and (C) delayed phase images show progressive focal con-rast accumulation (arrows in A, B, and C) in an ileal angiodysplasiaonfirmed at intraoperative endoscopy. (C) Additional focus (arrows) ofontrast on delayed image is also presumed to be active bleeding.

F(tc

owing to lack of radiation), and in patients with contraindica-

tittlol4i

fs

ipawi

T

M

P

M

T

���������

N

288 FLETCHER ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

ions to CT. In addition, there are several advantages to MRmaging not seen at CT enterography, which include the abilityo perform MR fluoroscopy to observe bowel peristalsis in realime and dynamically characterize stenotic lesions.24 Table 3ists the relative strengths of CT versus MR enterography. Inur experience, CT enterography is used frequently as a first-

ine test given its speed (scanning takes seconds rather than5– 60 min), robustness (it rarely fails to generate high-quality

mages), spatial and temporal resolution (which are important

able 2. Cross-Sectional Alternatives to CT Enterographyand Suggested Indications

Type of examination Indication

R or CT enteroclysis Suspected low-grade partial small-bowelobstruction

Questionable or nonspecific capsuleendoscopy findings, particularly ifjejunal

elvic MR Staging of perianal fistulasSuspected complications after ileal

pouch–anal anastomosisR enterography Monitoring response to therapy

Fluoroscopic, dynamic characterization ofstenotic or adhesive lesions

Contraindications to CT

t

or visualizing small masses and fistulizing disease), and CTcanner access.

The FutureWe believe that CT enterography will continue to be

ncorporated into wider clinical measures of Crohn’s disease,articularly given the promise that objective CT findings suchs mural hyperenhancement can be quantitated.3,25 Moreover,e anticipate that continuing technical developments in CT

mage reconstruction will substantially reduce the radiation

igure 5. Potential for dramatic radiation dose reduction at CT enter-graphy shown using dual-energy CT enterography. (A) Normal doseT enterography is achieved by combining the output from 2 radio-raph tubes (each with half the normal dose), and shows mild wallhickening and mural hyperenhancement in the ascending colon (arrow)ompared with the normal enhancement in the descending colon (ar-owhead). (B) The 140-kV tube and (C) the 80-kV tube also are diag-ostic, but were performed at half the radiation dose. Note the in-reased conspicuity of the iodine signal showing mural enhancement at

able 3. Relative Strengths and Weaknesses of MR vs CTEnterography

MR Strengths CT

�� Visualize bowel wall ����� Bowel wall signal ��� Mesenteric signal to noise ratio ���� Spatial resolution ���

Temporal resolution ����� Ionizing radiation ��� Dynamic characterization of stenotic lesions �

Amount of data ��Length of examination ��

OTE. Relative strength: ��� � �� � �; weekness: “–.”

FoCgtcrnc

he lower tube energy.

drptttFeecee

1

1

1

1

1

1

1

1

1

1

2

2

2

2

2

2

RM

March 2008 CT ENTEROGRAPHY 289

ose at CT enterography, which is already the same or less thanoutine abdominal CT. The use of dual-source CT systems willermit wider use of low-energy CT scanning, which will increasehe conspicuity of hypervascular inflammation and permit fur-her radiation dose reduction (Figure 5). Developments in MRechnology will facilitate rapid growth of MR enterography.inally, we believe that there will be routine correlation of CTnterography and capsule endoscopy findings, and that thesexaminations will be used in combination in more patients,ombining the exquisite mucosal visualization of capsulendoscopy with the mural and mesenteric images of CTnterography.

References

1. Young B, Fletcher J, Booya F, et al. Head-to-head comparison oforal contrast agents for cross-sectional enterography: smallbowel distention, timing and side effects. J Comput Assist To-mogr 2008;32:32–38.

2. Schindera ST, Nelson RC, DeLong DM, et al. Multi-detector rowCT of the small bowel: peak enhancement temporal window—initial experience. Radiology 2007;243:438–444.

3. Bodily KD, Fletcher JG, Solem CA, et al. Crohn disease: muralattenuation and thickness at contrast-enhanced CT enterogra-phy—correlation with endoscopic and histologic findings of in-flammation. Radiology 2006;238:505–516.

4. Booya F, Fletcher JG, Huprich JE, et al. Active Crohn disease: CTfindings and interobserver agreement for enteric phase CT enter-ography. Radiology 2006;241:787–795.

5. Wold P, Fletcher J, Johnson C, et al. Assessment of small bowelCrohn disease: noninvasive perioral enterography compared withother imaging methods and endoscopy-feasibility study. Radiol-ogy 2003;229:275–281.

6. Paulsen SR, Huprich JE, Fletcher JG, et al. CT enterography as adiagnostic tool in evaluating small bowel disorders: review ofclinical experience with over 700 cases. Radiographics 2006;26:641–662.

7. Macari M, Megibow AJ, Balthazar EJ. A pattern approach to theabnormal small bowel: observations at MDCT and CT enterogra-phy. AJR Am J Roentgenol 2007;188:1344–1355.

8. Colombel JF, Solem CA, Sandborn WJ, et al. Quantitative mea-surement and visual assessment of ileal Crohn’s disease activityby computed tomography enterography: correlation with endo-scopic severity and C reactive protein. Gut 2006;55:1561–1567.

9. Desreumaux P, Ernst O, Geboes K, et al. Inflammatory alterationsin mesenteric adipose tissue in Crohn’s disease. Gastroenterol-ogy 1999;117:73–81.

0. Meyers MA, McGuire PV. Spiral CT demonstration of hypervascu-larity in Crohn’s disease: “vascular jejunization of the ileum” orthe “comb sign”. Abdom Imaging 1995;20:327–332.

1. Lee S, Ha H, Yang S, et al. CT of prominent pericolic or perienteric

vasculature in patients with Crohn’s disease: correlation with 4

clinical disease activity and findings on barium studies. AJR Am JRoentgenol 2002;179:1029–1036.

2. Bruining DH, Fletcher JG, Tremaine WJ, et al. Prevalence ofpenetrating disease and extraintestinal manifestations ofCrohn’s disease with CT enterography. Gastroenterology 2007;132:650.

3. Vogel J, da Luz Moreira A, Baker M, et al. CT Enterography forCrohn’s disease: accurate preoperative diagnostic imaging. DisColon Rectum 2007;50:1761–1769.

4. Maconi G, Sampietro GM, Parente F, et al. Contrast radiology,computed tomography and ultrasonography in detecting internalfistulas and intra-abdominal abscesses in Crohn’s disease: aprospective comparative study. Am J Gastroenterol 2003;98:1545–1555.

5. Bernstein CN, Blanchard JF, Kliewer E, et al. Cancer risk inpatients with inflammatory bowel disease: a population-basedstudy. Cancer 2001;91:854–862.

6. Solem CA, Loftus LV, Fletcher JG, et al. Small bowel (SB) imagingin Crohn’s disease (CD): a prospective, blinded, 4-way compari-son trial. Gastroenterology 2005;128(Suppl 2):A-74.

7. Hassan C, Cerro P, Zullo A, et al. Computed tomography entero-clysis in comparison with ileoscopy in patients with Crohn’sdisease. Int J Colorectal Dis 2003;18:121–125.

8. Voderholzer WA, Beinhoelzl J, Rogalla P, et al. Small bowelinvolvement in Crohn’s disease: a prospective comparison ofwireless capsule endoscopy and computed tomography entero-clysis. Gut 2005;54:369–373.

9. Hara A, Leighton J, Sharma V, et al. Small bowel: preliminarycomparison of capsule endoscopy with barium study and CT.Radiology 2004;230:260–265.

0. Higgins PD, Caoili E, Zimmermann M, et al. Computed tomographicenterography adds information to clinical management in smallbowel Crohn’s disease. Inflamm Bowel Dis 2007;13:262–268.

1. Booya F. CT enterography findings in fistulizing Crohn’s disease(abstract 1982 GI-e). In: Radiological Society of North AmericaScientific Assembly and Annual Meeting, Chicago, IL, 2004:768.

2. Hara AK, Leighton JA, Sharma VK, et al. Small bowel: preliminarycomparison of capsule endoscopy with barium study and CT.Radiology 2004;230:260–265.

3. Huprich JE, Fletcher JG, Alexander J, et al. Obscure GI bleeding:64-channel multiphase, multiplanar CT enterography has a role.Radiology 2008;246:562–571.

4. Fidler J. MR imaging of the small bowel. Radiol Clin North Am2007;45:317–331.

5. Loftus E. Objective measures of disease activity: alternatives tosymptom indices. Rev Gastroenterol Dis 2007;7:S8–S16.

Address requests for reprints to: J. G. Fletcher, MD, Department ofadiology, 200 First St. SW, Mayo Building, West 2-B, Rochester,innesota 55905. e-mail: fletcher.joel@mayo.edu; fax: (507) 266-

609.

E

US

MPB

*

BupsagsplppgftPKt7acawmbatoswwEs

Cnfaw

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2008;6:290–297

NDOSCOPY CORNER

nresectable Cholangiocarcinoma: Comparison of Survival in Biliarytenting Alone Versus Stenting With Photodynamic Therapy

ICHEL KAHALEH,* RAJNISH MISHRA,* VANESSA M. SHAMI,* PATRICK G. NORTHUP,* CARL L. BERG,*ENNY BASHLOR,* PETRA JONES,* KRISTI ELLEN,* GEOFFREY R. WEISS,‡ CHRISTIANA M. BRENIN,‡

ARBARA E. KURTH,‡ TYVIN A. RICH,§ REID B. ADAMS,� and PAUL YEATON*

Digestive Health Center, ‡Cancer Center, §Departments of Radiation Oncology and �Surgery, University of Virginia Health System, Charlottesville, Virginia

om

tqbmjtapts

iast

auPcfi

nqtc(ws

srd

cP

See Editorial on page 266.

ackground & Aims: Photodynamic therapy (PDT) fornresectable cholangiocarcinoma is associated with im-rovement in cholestasis, quality of life, and potentiallyurvival. We compared survival in patients with unresect-ble cholangiocarcinoma undergoing endoscopic retro-rade cholangiopancreatography (ERCP) with PDT andtent placement with a group undergoing ERCP with stentlacement alone. Methods: Forty-eight patients were pal-

iated for unresectable cholangiocarcinoma during a 5-yeareriod. Nineteen were treated with PDT and stents; 29atients treated with biliary stents alone served as a controlroup. Multivariate analysis was performed by using Modelor End-Stage Liver Disease score, age, treatment by chemo-herapy or radiation, and number of ERCP procedures andDT sessions to detect predictors of survival. Results:aplan–Meier analysis demonstrated improved survival in

he PDT group compared with the stent only group (16.2 vs.4 months, P<.004). Mortality in the PDT group at 3, 6,nd 12 months was 0%, 16%, and 56%, respectively. Theorresponding mortality in the stent group was 28%, 52%,nd 82%, respectively. The difference between the 2 groupsas significant at 3 months and 6 months but not at 12onths. Only the number of ERCP procedures and num-

er of PDT sessions were significant on multivariatenalysis. Adverse events specific to PDT included 3 pa-ients with skin phototoxicity requiring topical therapynly. Conclusions: ERCP with PDT seems to increaseurvival in patients with unresectable cholangiocarcinomahen compared with ERCP alone. It remains to be provedhether this effect is attributable to PDT or the number ofRCP sessions. A prospective randomized multicentertudy is required to confirm these data.

holangiocarcinoma is the second most common malig-nancy arising within the liver and is associated with sig-

ificant morbidity and mortality.1 The majority of patients areound to be unresectable on presentation2– 4; their survival ispproximately 3 months without intervention and 4 – 6 months

ith biliary decompression.5–10 Successful palliation of biliary

bstruction remains the main goal for reducing morbidity andortality in these patients with unresectable disease.6,11

Hepaticojejunostomy is associated with a 30-day postopera-ive mortality rate between 7% and 24%,12–16 and significantuality of life after surgery is only improved in a minorityecause of time needed to recover.9,17,18 Endoscopy with place-ent of biliary stents has become the standard to palliate

aundice in this population, with less morbidity and mortalityhan that associated with surgery.13,18 –23 Limitations of thispproach are related to (1) the ability to decompress affectedroximal segments and (2) recurrent stent occlusion, becausehese stents offer no ability to remodel malignant tis-ues.6,7,12,24 –26

Photodynamic therapy (PDT) is an evolving therapy thatnvolves the intravenous administration of a photosensitizinggent followed by its activation by using light illumination of apecific wavelength, resulting in ischemic necrosis27–29 propor-ional to tissue oxygenation.30,31

PDT was demonstrated to reduce xenografted human chol-ngiocarcinoma tumor volume by 60% in a mouse model.32 Inncontrolled human studies in which porfimer sodium– basedDT was combined with stenting, there was improvement inholestasis and survival, and few complications related to por-mer sodium were observed.33–36

A prospective, randomized, controlled trial confirmed a sig-ificant advantage attributable to PDT in relief of jaundice,uality of life, and survival.37 The improvement of survival inhe PDT group was such that it was considered unethical toontinue the study after the first 39 patients were randomized20 received PDT). This study was criticized because enrollmentas limited to those patients in whom technically successful

tenting did not result in successful drainage.38

We reported our experience with PDT in treatment of unre-ectable cholangiocarcinoma and compared its efficacy andesults with patients palliated with only endoscopic biliaryrainage.

Abbreviations used in this paper: ERCP, endoscopic retrogradeholangiopancreatography; MELD, Model for End-Stage Liver Disease;DT, photodynamic therapy.

© 2008 by the AGA Institute1542-3565/08/$34.00

doi:10.1016/j.cgh.2007.12.004

iatbawtT

adfilo

tuocTtar

iwre(lwit

Odsgpbepa5

s(PwaAdwtd

pwi

dlwoltd

dPaocp

j

T

AGM

MMT

CRM

CN

P

M

March 2008 PDT IN CHOLANGIOCARCINOMA 291

Materials and MethodsPatientsSixty-four consecutive patients were referred to our

nstitution with cholangiocarcinoma between December 2001nd November 2006. Sixteen patients (25%) underwent resec-ion. The remaining 48 patients were palliated with endoscopiciliary stents, and of these, 19 received PDT after its availabilityt the University of Virginia in December 2004. All patientsere captured in a dedicated database and followed prospec-

ively. Patient pretreatment characteristics are summarized inable 1.

Therapy with PDT was offered to all patients with unresect-ble cholangiocarcinoma and those with resectable lesionseemed inoperable. No patients had contraindications to por-mer sodium, such as compromised kidney or hepatic function,

eukopenia or thrombocytopenia, or evidence of cancer of an-ther organ.

Staging and Tissue DiagnosisAll patients had clinical and radiologic features charac-

eristic for cholangiocarcinoma. Bismuth classification was doc-mented for all patients (Type I, tumors below the confluencef the left and right hepatic ducts; Type II, tumors reaching theonfluence but not involving the left or right hepatic ducts;ype III, tumors occluding the common hepatic duct and either

he right (IIIa) or left (IIIb) hepatic duct; Type IV, tumors thatre multicentric or that involve the confluence and both the

able 1. Pretreatment Clinical Characteristics of All thePatients in the Study, Showing ComparisonBetween the Groups With PDT and Stent VersusStent Only

PDT group(n � 19)

Stent group(n � 29)

Pvalue

ge (y) (mean) 65.3 67.4 .61ender (male, female) 11, 8 13, 16ean bilirubin at

presentation8.3 mg/dL 12.8 (6.6) mg/dL .08

ean bilirubin after 90 days 3.5 mg/dL 6.3 mg/dL .09ean MELD on presentation 14.6 18.3 .03

umor extensionBismuth IV 9 10Bismuth III 7 10Bismuth II 1 8Bismuth I 2 1

hemotherapy (n � 22) 11 11adiotherapy (n � 19) 9 10ortality3 mo 0% 27.60% .0116 mo 16% 51.70% .0112 mo 56% 82.10% .636

holangitis 7 10 .25o. of interventions, median

(range)3.0 (1–8) 2.0 (1–13) .053

atient alive at the end ofstudy

8 2

ean & median survival(mo)

16.2 (8) 7.4 (5) .003

ight and left hepatic ducts). 7

Pathologic diagnosis was established in 69% of cases, whichs in accordance with published studies.39,40 Tissue diagnosisas established by endoscopic retrograde cholangiopancreatog-

aphy (ERCP) with triple sampling41 in 20 of 26 (77%) cases, byndoscopic ultrasound with fine-needle aspiration in 11 of 1478%) cases, percutaneous liver biopsy in 8 of 8 cases, and byaparotomy/laparoscopy in 4 of 4 cases. Staging was performedith computed tomography and/or magnetic resonance imag-

ng. Resectability was defined according to the criteria of Vau-hey and Blumgart.42

Materials and TechniquesDuodenoscopes (TJF-140, TJF-160, and TJF-160VF;

lympus America, Center Valley, PA) were used for all proce-ures. Biliary cannulation was performed with triple lumenphincterotomes. After biliary sphincterotomy, a cholangio-ram defined the extent of ductal involvement. Selective decom-ression of all opacified, dilated segments was attempted, withougie and balloon dilation to assist in the placement of poly-thylene stents (7F, 8.5F, and 10F diameter [Figures 1– 4]). Allrocedures were performed by 1 of 2 dedicated pancreaticobili-ry endoscopists (M.K. or P.Y.), each performing in excess of00 ERCPs annually.

Photodynamic TherapyEach patient to whom PDT was offered underwent a

pecific, detailed educational process by a dedicated team memberP.B., P.J., or K.E.), after which informed consent was obtained.orfimer sodium (Photofrin; Axcan Pharma Inc, Quebec, Canada)as used as a photo sensitizing agent, administered intravenouslyt a dose of 2 mg/kg body weight 48 hours before illumination.

diode laser system (InGaAIP Laser Diode; Diomed Inc, An-over, MA) with a maximum power output of 2000 mW and aavelength of 633 � 3 nm was used as a light source, delivered

hrough a 3.0-m length fiber having a 2.5-cm-long cylindricaliffuser at its distal end (Pioneer Optics, Windsor Locks, CT).

During ERCP, the biliary anatomy was defined, and appro-riate locations for therapy were identified, after which guide-ire access was obtained and dilation was used as necessary to

ntroduce the diffuser within the malignant stricture.The diffuser was inserted into a 10F sheath of a plastic stent

elivery system (MAJ-1419; Olympus America), placed at theevel of the stricture to be treated (Figures 5 and 6). The sheetas positioned at the appropriate level, after being advancedver a wire. Photoactivation was performed at 620 nm with a

ight dose of 180 J/cm2, fluence of 0.250 W/cm2, and irradia-ion time of 750 seconds. One or 2 segments were treated at theiscretion of the endoscopist.

After photoactivation, only plastic stents were inserted toecompress opacified radicals proximal to the treated lesion.DT was repeated at 3-month intervals (Figure 7) at which timell stents were replaced; stents were exchanged earlier in the casef premature occlusion or migration to maintain optimal de-ompression. All patients received periprocedure antibiotic pro-hylaxis.

Definition of EventsSuccessful therapy was defined by relief of cholangitis,

aundice, and pruritus with decrease of bilirubin to less than

5% of the pretreatment value within 30 days.43 Complications

rd

ddl3pwa

(tTtumpsbamclLcrdm

tiaB

(rwtmwM3

(p

pbdnumfces

Fs

292 KAHALEH ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

elated to ERCP were defined following consensus criteria44 andifferentiated from complications related to PDT.

Follow-UpThose patients undergoing therapy including PDT un-

erwent ERCP with PDT every 3 months until death or with-rawal from the study. All patients were followed in clinic with

aboratory values at least 1 month after intervention and everymonths thereafter or earlier in case of complications. In

atients who received plastic stents without PDT, repeat ERCPas performed if indicated (recurrent cholestasis, cholangitis,nd/or pruritus) until patient refusal or death.

Statistical AnalysisStatistical analysis was performed with SAS, version 9.1

SAS Institute, Cary, NC). Statistical significance was assumedo be at an alpha of .05, and all statistical tests were two-sided.he Fisher exact test or �2 tests were used to compare propor-

ions, and the Student t test or Wilcoxon signed rank test wassed for continuous data. Multivariable logistic regressionodels were constructed with 6-month and 1-year survival end

oints. Univariate time-to-event survival models were con-tructed by using the Kaplan-Meier method and comparedetween groups with the log-rank test. Multivariate survival wasnalyzed with a Cox proportional hazards model and theethod of maximum likelihood estimates. Variables were in-

luded in the multivariate model if they achieved a P value ofess than .20 in the univariate comparison (Model for End-Stageiver Disease [MELD], number of sessions), or if they werelinically expected to be important in the disease process (age,adiation, or chemotherapy). MELD and bilirubin are not in-ependent variables, so both were not included in the finalodel. It was believed that MELD was a better representation of

1Figure 1. Cholangiogram showing a Bismuth III lesion (patient A).

he subjects’ severity of disease and was therefore chosen to ben the final model instead of total bilirubin. This study waspproved by the University of Virginia Institutional Reviewoard for Health Sciences Research.

ResultsForty-eight patients (24 male), mean age, 66.6 years

range, 26 –94 years) were palliated. A total of 19 (40%) patientseceived ERCP and PDT, whereas 29 (60%) underwent ERCPith stenting alone. The pretreatment clinical characteristics of

he study population are summarized in Table 1. Mean andedian pretreatment carbohydrate antigen 19-9 (n � 41/48)as 3203 and 227 U/mL (range, 3–59,100 U/mL), respectively.ean and median pretreatment carcinoembryonic antigen (n �

6/48) was 42.9 and 3.5 (range, 0.7– 650), respectively.Nineteen patients were found to have Bismuth IV cancers

40%), 17 had Bismuth III (35%), 9 had Bismuth II (19%), and 3atients had Bismuth I (6%) (Table 1).

All patients had clinical and cross-sectional imaging com-atible with cholangiocarcinoma. No patients were excludedecause of a large mass. The median number of ERCP proce-ures was 3.0 (range, 1– 8) in the PDT group, including a meanumber of 1.6 (range, 1–3) PDT sessions. The stent only groupnderwent a median of 2.0 (range, 1–13) ERCP procedures toaintain biliary decompression. There was no significant dif-

erence between age, gender, preprocedure total bilirubin, car-inoembryonic antigen, and carbohydrate antigen 19-9 levels inach group (Table 1). The stent only group had a statisticallyignificant (P � .03) higher preprocedure MELD score (18.3 vs

igure 2. After placement of 2 stents on each side, selective access ofegments VI and VII is achieved with a guidewire (patient A).

4.6) compared with the PDT group.

cap

cuob(

v(owwg

pwwOsc

t

adln

rg

bg((ms.mm

2msac(pEshbttet

dea

f

T

PARCME

T

OBRDOOHWP

T

March 2008 PDT IN CHOLANGIOCARCINOMA 293

Biliary DecompressionIn both groups, serum bilirubin was successfully de-

reased (�75% of the original value) (Figure 8). After 3 months,greater than 50% reduction of bilirubin was noted in 26

atients.In both groups the decrease in bilirubin was significant when

omparing pretreatment versus post-treatment bilirubin (Fig-re 8) (P � .008 in the PDT group and P � .0001 in the stentnly group). However, when comparing the degree of decreaseetween the 2 groups, no significant difference was observed

P � .78).

Chemotherapy and RadiationA total of 22 (46%) patients received chemotherapy with

arious combinations of gemcitabine and capecetabine. Eleven58%) were in the PDT group, and 11 (38%) were in the stentnly group (P � .17). A total of 19 patients also were treatedith concomitant extracorporeal radiation. Of these, 9 (47%)ere included in the PDT group and 10 (34%) in the stent onlyroup (P � .37).

Complications and Adverse EventsComplications in the stent only group included 10

atients developing cholangitis after endoprosthetic therapy,ith 2 patients dying as a consequence. Post-ERCP pancreatitisas observed in 4 patients and duodenal perforation in one.ther adverse events in the stent group included hepatic ab-

cess (1), Billroth II perforation (1), and non-ST elevation myo-ardial infarction (2).

In the PDT group, 7 patients (37%) developed cholangitisreated with antibiotics alone. One patient developed a hepatic

able 2. Multivariate Analysis of Factors Predictive ofIncreased Survival

Variable P value Hazard ratio 95% Confidence interval

DT .0062 0.27 0.105–0.689ge .9001 0.998 0.969–1.028adiation .185 0.308 0.054–1.757hemotherapy .9285 1.074 0.228–5.059ELD .6947 1.016 0.938–1.102RCP sessions .0002 0.65 0.517–0.818

able 3. Table Comparing Studies Performed by Using ERCPCholangiocarcinoma

Study Year N (M/F) Study type

rtner et al35 1998 9 Single armerr et al36 2000 23 Single armumalla et al34 2001 6 Single armumoulin et al50 2003 24 Single armrtner et al37 2003 20 Randomizedrtner et al37 2003 31 Nonrandomizedarewood et al57 2005 8 Single armitzigmann et al39 2006 68 Single armrasad et al40 2007 25 Single arm

B, total bilirubin; N/A, not available.

bscess with prolonged cholangitis, requiring percutaneousrainage. Other complications included hemobilia (2) and cho-

ecystitis (2). Both patients with cholecystitis were managedonsurgically.

Among the adverse events specific to PDT, 3 patients expe-ienced skin phototoxicity, one World Health Organizationrade 3; all were managed with topical therapy.

SurvivalAt the time of analysis, 10 patients were living, with 8

eing in the PDT group. Cause of death included tumor pro-ression (n � 30), cholangitis (n � 2), pulmonary embolusn � 1), multiorgan failure (n � 4), and myocardial infarctionn � 1). Overall mean and median survival was 8.5 � 7.2 and 7

onths (range, 0.25–36 months), respectively. Kaplan-Meierurvival analysis (Figure 9) showed statistically significant (P �003) prolonged survival in the PDT group (mean, 16.2 � 2.4

onths, compared with the stent only group (mean, 7.4 � 1.6onths).Overall survival at 3, 6, and 12 months was 83%, 33%, and

7%, respectively. In the PDT group the 3-, 6-, and 12-monthortality rates were 0%, 16%, and 56%, respectively. The corre-

ponding mortality rates in the stent only group were 28%, 52%,nd 82%, respectively. This difference was statistically signifi-ant at 3 (P � .01) and 6 months (P � .01) but not at 12 monthsP � .08). On analysis of maximum likelihood estimates, factorsredictive of survival were treatment with PDT and number ofRCP sessions. There was no significance related to age, MELDcore, chemotherapy, or radiation (Table 2). The proportionalazards survival model was re-run with an interaction termetween exposure to PDT and number of ERCP sessions, andhere was a trend toward statistical significance of the interac-ion term (P � .08). In this new model, the influence of PDTxposure was persistently statistically significant (P � .004), andhe remaining conclusions of the study were unchanged.

DiscussionAlmost 80% of patients with cholangiocarcinoma are

iagnosed at an unresectable stage.3,8,45 Effective palliation isssential, because biliary drainage and prevention of cholestasisre crucial to prevent pruritus, cholangitis, and death.

The approach to palliative biliary decompression has evolvedrom surgery with its attendant morbidity and mortality16,22

PDT With Photofrin Sodium for Palliation of

dian TB beforeand afterDT (mg/dL)

Mean PDTsessions(range)

Mediansurvival

(mo)

Adverse events:phototoxicity,cholangitis

6 � 6 1.5 (1–2) 14.6 1 (11%), 0 (0%)2 � 1.1 3 (1–5) 11.1 3 (13%), 8 (35%)7 � 1.3 2.3 (1–2) �6 2 (33%), 2 (33%)3 � 2.6 (mean) 1–2 9.9 2 (8%), 5 (21%)A (decreased) 2.4 (1–5) 16.4 2 (10%), 5 (25%)8 � 3.1 1.5 (1–4) 14.2 3 (10%), 6 (19%)7 � 1.1 2 (1–5) 9.2 2 (25%), 2 (25%)A (decreased) 2 (1–6) 12 8 (12%), 38 (56%)1 � 3.5 (mean) 1.6 (1–4) 13.4 1 (4%), 2 (8%)

With

Me

P

18.11.2.

13.N/

11.7.N/6.

atpnaie

trdtatf

“tids

wiTwdmP

Fu

F(

F

294 KAHALEH ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

nd more recently from percutaneous to endoscopic management;he latter has been shown superior to a percutaneous approach inrevention of cholestasis and improved mortality.46 Unfortu-ately, the benefit of ERCP with stent placement is often obvi-ted by proximal tumor obstruction.6,7,10,12 To address thisssue, multiple studies have been conducted to investigate theffect of combining bile duct stent insertion and PDT.33–36,47,48

PDT is thought to destroy cancer and neovascular cells ando induce tumor thrombosis through formation of cytotoxiceaction products, including singlet oxygen radicals.27,28,30 Un-erstanding that survival in cholangiocarcinoma is related tohe efficacy of biliary decompression, PDT offers a logical mech-nism to use. The efficacy of stenting is limited by stent pa-ency; unlike benign conditions, no tissue remodeling is af-ected by stenting tumor. PDT offers the possibility of

igure 3. Selective access of segments V and VIII is then achieved by

sing a sphincterotome preloaded with a guidewire (patient A). (

remodeling” tumoral mass,49 which might enhance or prolonghe decompressive effect. Accepting this hypothesis, PDT couldmprove cholestasis and survival in the setting of incompleteecompression with stents by opening previously inaccessibleegments, an observation noted by ourselves and others.

In most prospective trials (Table 3) and in our series, PDTas associated with a significant reduction in bilirubin and

ncrease in survival compared with historical data.34 –36,39,40,43,50

his study compared the efficacy of PDT with biliary stentingith a group receiving only biliary plastic stenting. Notableifferences with most previous published reports were thatultiple sessions of ERCP were offered to both groups, and

DT was not restricted to a single session. Our results seem

igure 4. Fluoroscopic image after drainage of all dilated segmentspatient A).

igure 5. Malignant stricture (Bismuth I) before PDT treatment

patient B).

sicmusdpsda

r

UsgmAmswtP

rii

F(

F

FttmddM

F(sdc

March 2008 PDT IN CHOLANGIOCARCINOMA 295

imilar to those of Ortner et al,37 in which a dramatic increasen median survival after PDT (16.4 months) was observed,ompared with 3.3 months in the patients receiving stent place-ent alone. Unlike the study by Ortner et al, patients who

nderwent successful stenting were not excluded from ourtudy, and bilirubin values in both groups were significantlyecreased (Figure 8). Our aggressive approach to biliary decom-ression, as advocated by Prat et al,51 appeared to influenceurvival independently of PDT administration (Table 2). Theesign of this study does not resolve the possibility that andditive effect would be observed combining PDT and stenting.

In a recent study by Zoepf et al,47 in which a hematoporphy-in derivative (Photosan-3; Scotia Pharmaceuticals, Guildford,

igure 6. PDT application at the level of the malignant stricturepatient B).

s

K) was used as a photosensitizer, patients were randomized totents only or PDT with stents. Stents were changed in bothroups every 3 months. Survival in the PDT group was 21onths, compared with 7 months in the stent only group.lthough this suggests an additive effect of PDT, there was noultivariate analysis of the number of ERCP sessions as a

urrogate of decompression, and by design, those living longerould have had more ERCP sessions. An alternative explana-

ion of the observed results would be improved efficacy withhotosan-3 compared with Photofrin.

A recent retrospective study40 reported 25 patients with un-esectable cholangiocarcinoma treated with PDT; on multivar-ate analysis, the presence of a visible mass on imaging andncreasing time between diagnosis and PDT predicted a poorer

igure 8. Mean change in bilirubin at entry and after 3 months ofreatment. When comparing the PDT group with the stent only group,here is no statistical difference in either the pretreatment or post-treat-ent values (P � .09), and there is not any statistical difference in theegree of decrease between the 2 groups (P � .78). In both groups,ecompression was successful by criteria defined in Materials andethods.

igure 9. Cumulative survival of patients treated with PDT and stentsolid line) versus stent only group (broken line). Kaplan–Meier analysishowed estimated mean survival of 16.2 � 2.4 months (95% confi-ence interval, 7–27 months) after PDT vs 7.4 � 1.6 months (95%onfidence interval, 3–7 months), respectively, which was statistically

ignificant (P � .003). igure 7. Cholangiogram at 3-month follow-up (patient B).

sms4pov

ufsecyl

atipoaP

ccodrdibrfi

1

1

1

1

1

1

1

1

1

1

2

2

2

2

2

2

2

2

2

T

4

9

296 KAHALEH ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

urvival rate after PDT. Therefore, early treatment with PDTight preserve hepatic function. We used the MELD score to

tratify disease severity in our study population. Thirty-four of8 patients (71%) presented with a MELD �14. Of these 34atients, 23 (68%) were in the stent only group; despite thisbservation, MELD was not an independent predictor of sur-ival by multivariate analysis (Table 2).

Two groups have attributed improved survival in this pop-lation to the use of chemotherapy and radiation.52,53 We of-

ered chemoradiation to all patients; 22 patients (11 PDT, 11tent only) received chemotherapy, of whom 19 also receivedxternal beam radiation (9 PDT, 10 stent only). This therapyonferred no additional survival advantage by multivariate anal-sis, confirming the results of other studies that show onlyimited response in cholangiocarcinoma.54,55

Finally, as seen in several other studies (Table 3), directdverse effects of PDT were minor and largely related to pho-otoxicity.34,36,56 The incidence of bacterial cholangitis was sim-lar in both the groups (35%) and is compatible with theublished results of 25%–35%.34,50,57 Cholangitis accounted fornly 5% of all deaths in our series, again suggesting the value ofggressive biliary decompression. Costs related to ERCP andDT are provided in Table 4.

In summary, ERCP with PDT appears to improve survivalompared with ERCP with biliary stenting alone in unresectableholangiocarcinoma. Although our results are comparable withther published results, the effect we observed is not indepen-ent of the number of ERCP sessions, raising a complex issueelated to adequacy of stenting. In addition, the uncontrolledesign of this study prevented definitive conclusions from be-

ng drawn. To definitely prove that PDT confers an additionalenefit compared with ERCP with stenting alone, a prospective,andomized, controlled trial is required with strict criteria de-ning successful stenting (Figures 3–9 and Table 4).

References

1. Ahrendt SA, Nakeeb A, Pitt HA. Cholangiocarcinoma. Clin LiverDis 2001;5:191–218.

2. de Groen PC, Gores GJ, LaRusso NF, et al. Biliary tract cancers.N Engl J Med 1999;341:1368–1378.

3. Jarnagin WR, Fong Y, DeMatteo RP, et al. Staging, resectability,and outcome in 225 patients with hilar cholangiocarcinoma. AnnSurg 2001;234:507–519.

4. Reed DN Jr, Vitale GC, Martin R, et al. Bile duct carcinoma: trendsin treatment in the nineties. Am Surg 2000;66:711–715.

5. Liu CL, Lo CM, Lai EC, et al. Endoscopic retrograde cholangio-pancreatography and endoscopic endoprosthesis insertion in pa-tients with Klatskin tumors. Arch Surg 1998;133:293–296.

able 4. Comparison of Costs Between ERCP With BiliaryStent Placement and PDT

3268 ERCP with biliary stentPhysician charge: $1440Medicare payment: $375.74Highest commercial payment: $631.18

6570 PDTPhysician charge: $125.00Medicare payment: $53.46Highest commercial payment: $91.65

6. Polydorou AA, Cairns SR, Dowsett JF, et al. Palliation of proximal

malignant biliary obstruction by endoscopic endoprosthesisinsertion. Gut 1991;32:685–689.

7. Ducreux M, Liguory C, Lefebvre JF, et al. Management of malig-nant hilar biliary obstruction by endoscopy: results and prognosticfactors. Dig Dis Sci 1992;37:778–783.

8. Farley DR, Weaver AL, Nagorney DM. “Natural history” of unre-sected cholangiocarcinoma: patient outcome after noncurativeintervention. Mayo Clin Proc 1995;70:425–429.

9. Baer HU, Stain SC, Dennison AR, et al. Improvements in survivalby aggressive resections of hilar cholangiocarcinoma. Ann Surg1993;217:20–27.

0. Deviere J, Baize M, de Toeuf J, et al. Long-term follow-up ofpatients with hilar malignant stricture treated by endoscopicinternal biliary drainage. Gastrointest Endosc 1988;34:95–101.

1. Gibson RN, Yeung E, Hadjis N, et al. Percutaneous transhepaticendoprostheses for hilar cholangiocarcinoma. Am J Surg 1988;156:363–367.

2. Lai EC, Tompkins RK, Mann LL, et al. Proximal bile duct cancer:quality of survival. Ann Surg 1987;205:111–118.

3. Havlik R, Sbisa E, Tullo A, et al. Results of resection for hilarcholangiocarcinoma with analysis of prognostic factors. Hepato-gastroenterology 2000;47:927–931.

4. Kapoor VK, Pradeep R, Haribhakti SP, et al. Intrahepatic segmentIII cholangiojejunostomy in advanced carcinoma of the gallblad-der. Br J Surg 1996;83:1709–1711.

5. Chaudhary A, Dhar P, Tomey S, et al. Segment III cholangiojeju-nostomy for carcinoma of the gallbladder. World J Surg 1997;21:866–871.

6. Guthrie CM, Banting SW, Garden OJ, et al. Segment III cholan-giojejunostomy for palliation of malignant hilar obstruction. Br JSurg 1994;81:1639–1641.

7. Figueras J, Llado L, Valls C, et al. Changing strategies in diagno-sis and management of hilar cholangiocarcinoma. Liver Transpl2000;6:786–794.

8. Blom D, Schwartz SI. Surgical treatment and outcomes in carci-noma of the extrahepatic bile ducts: the University of Rochesterexperience. Arch Surg 2001;136:209–215.

9. Bismuth H, Castaing D, Traynor O. Resection or palliation: priorityof surgery in the treatment of hilar cancer. World J Surg 1988;12:39–47.

0. Cotton PB. Endoscopic methods for relief of malignant obstruc-tive jaundice. World J Surg 1984;8:854–861.

1. Classen M, Hagenmuller F. Biliary drainage. Endoscopy 1983;15(Suppl 1):221–229.

2. Smith AC, Dowsett JF, Russell RC, et al. Randomised trial ofendoscopic stenting versus surgical bypass in malignant lowbileduct obstruction. Lancet 1994;344:1655–1660.

3. Soehendra N, Reynders-Frederix V. Palliative bile duct drainage: anew endoscopic method of introducing a transpapillary drain.Endoscopy 1980;12:8–11.

4. Schmassmann A, von Gunten E, Knuchel J, et al. Wallstentsversus plastic stents in malignant biliary obstruction: effects ofstent patency of the first and second stent on patient complianceand survival. Am J Gastroenterol 1996;91:654–659.

5. Born P, Rosch T, Bruhl K, et al. Long-term outcome in patientswith advanced hilar bile duct tumors undergoing palliative endo-scopic or percutaneous drainage. Z Gastroenterol 2000;38:483–489.

6. Davids PH, Groen AK, Rauws EA, et al. Randomised trial ofself-expanding metal stents versus polyethylene stents for distalmalignant biliary obstruction. Lancet 1992;340:1488–1492.

7. Hsi RA, Rosenthal DI, Glatstein E. Photodynamic therapy in thetreatment of cancer: current state of the art. Drugs 1999;57:725–734.

8. Webber J, Herman M, Kessel D, et al. Current concepts ingastrointestinal photodynamic therapy. Ann Surg 1999;230:

12–23.

2

3

3

3

3

3

3

3

3

3

3

4

4

4

4

4

4

4

4

4

4

5

5

5

5

5

5

5

5

Hl0

DEr

March 2008 PDT IN CHOLANGIOCARCINOMA 297

9. Oleinick NL, Evans HH. The photobiology of photodynamic ther-apy: cellular targets and mechanisms. Radiat Res 1998;150:S146–S156.

0. Nelson JS, Liaw LH, Orenstein A, et al. Mechanism of tumordestruction following photodynamic therapy with hematoporphy-rin derivative, chlorin, and phthalocyanine. J Natl Cancer Inst1988;80:1599–1605.

1. Pass HI. Photodynamic therapy in oncology: mechanisms andclinical use. J Natl Cancer Inst 1993;85:443–456.

2. Wong Kee Song LM, Wang KK, Zinsmeister AR. Mono-L-aspartylchlorin e6 (NPe6) and hematoporphyrin derivative (HpD) in pho-todynamic therapy administered to a human cholangiocarcinomamodel. Cancer 1998;82:421–427.

3. McCaughan JS Jr, Mertens BF, Cho C, et al. Photodynamic ther-apy to treat tumors of the extrahepatic biliary ducts: a casereport. Arch Surg 1991;126:111–113.

4. Rumalla A, Baron TH, Wang KK, et al. Endoscopic application ofphotodynamic therapy for cholangiocarcinoma. Gastrointest En-dosc 2001;53:500–504.

5. Ortner MA, Liebetruth J, Schreiber S, et al. Photodynamic therapyof nonresectable cholangiocarcinoma. Gastroenterology 1998;114:536–542.

6. Berr F, Wiedmann M, Tannapfel A, et al. Photodynamic therapyfor advanced bile duct cancer: evidence for improved palliationand extended survival. Hepatology 2000;31:291–298.

7. Ortner ME, Caca K, Berr F, et al. Successful photodynamic ther-apy for nonresectable cholangiocarcinoma: a randomized pro-spective study. Gastroenterology 2003;125:1355–1363.

8. Gores GJ. A spotlight on cholangiocarcinoma. Gastroenterology2003;125:1536–1538.

9. Witzigmann H, Berr F, Ringel U, et al. Surgical and palliativemanagement and outcome in 184 patients with hilar cholangio-carcinoma: palliative photodynamic therapy plus stenting is com-parable to r1/r2 resection. Ann Surg 2006;244:230–239.

0. Prasad GA, Wang KK, Baron TH, et al. Factors predicting survivalin patients with cholangiocarcinoma treated with photodynamictherapy. Clin Gastroenterol Hepatol 2007;5:743–748.

1. Jailwala J, Fogel EL, Sherman S, et al. Triple-tissue sampling atERCP in malignant biliary obstruction. Gastrointest Endosc 2000;51:383–390.

2. Vauthey JN, Blumgart LH. Recent advances in the managementof cholangiocarcinomas. Semin Liver Dis 1994;14:109–114.

3. De Palma GD, Pezzullo A, Rega M, et al. Unilateral placement ofmetallic stents for malignant hilar obstruction: a prospectivestudy. Gastrointest Endosc 2003;58:50–53.

4. Cotton PB, Lehman G, Vennes J, et al. Endoscopic sphincterot-omy complications and their management: an attempt at consen-sus. Gastrointest Endosc 1991;37:383–393.

5. Bismuth H, Nakache R, Diamond T. Management strategies inresection for hilar cholangiocarcinoma. Ann Surg 1992;215:

31–38. o

6. Speer AG, Cotton PB, Russell RC, et al. Randomised trial ofendoscopic versus percutaneous stent insertion in malignantobstructive jaundice. Lancet 1987;2:57–62.

7. Zoepf T, Jakobs R, Arnold JC, et al. Palliation of nonresectablebile duct cancer: improved survival after photodynamic therapy.Am J Gastroenterol 2005;100:2426–2430.

8. Wiedmann M, Berr F, Schiefke I, et al. Photodynamic therapy inpatients with non-resectable hilar cholangiocarcinoma: 5-yearfollow-up of a prospective phase II study. Gastrointest Endosc2004;60:68–75.

9. Berr F, Tannapfel A, Lamesch P, et al. Neoadjuvant photodynamictherapy before curative resection of proximal bile duct carcinoma.J Hepatol 2000;32:352–357.

0. Dumoulin FL, Gerhardt T, Fuchs S, et al. Phase II study ofphotodynamic therapy and metal stent as palliative treatment fornonresectable hilar cholangiocarcinoma. Gastrointest Endosc2003;57:860–867.

1. Prat F, Chapat O, Ducot B, et al. Predictive factors for survival ofpatients with inoperable malignant distal biliary strictures: apractical management guideline. Gut 1998;42:76–80.

2. Foo ML, Gunderson LL, Bender CE, et al. External radiationtherapy and transcatheter iridium in the treatment of extrahe-patic bile duct carcinoma. Int J Radiat Oncol Biol Phys 1997;39:929–935.

3. Morganti AG, Trodella L, Valentini V, et al. Combined modalitytreatment in unresectable extrahepatic biliary carcinoma. Int JRadiat Oncol Biol Phys 2000;46:913–919.

4. Kelley ST, Bloomston M, Serafini F, et al. Cholangiocarcinoma:advocate an aggressive operative approach with adjuvant chemo-therapy. Am Surg 2004;70:743–749.

5. Goldstein RM, Stone M, Tillery GW, et al. Is liver transplantationindicated for cholangiocarcinoma? Am J Surg 1993;166:768–772.

6. Ortner M. Photodynamic therapy in the biliary tract. Curr Gastro-enterol Rep 2001;3:154–159.

7. Harewood GC, Baron TH, Rumalla A, et al. Pilot study to assesspatient outcomes following endoscopic application of photody-namic therapy for advanced cholangiocarcinoma. J GastroenterolHepatol 2005;20:415–420.

Address requests for reprints to: Michel Kahaleh, MD, Digestiveealth Center Box 800708, University of Virginia Health System, Char-

ottesville, VA 22908-0708. e-mail: mk5ke@virginia.edu; fax: 434-924-491.Part of this work was presented as an oral presentation during

igestive Diseases Week 2007, American Society for Gastrointestinalndoscopy Topic Forum #140, Washington, DC, May 21, 2007, witheference: Gastrointest Endosc 2007;65:AB96.

The authors acknowledge the strong support by the technical staff

f the Digestive Health Center’s Clinic and Endoscope Staff.

PC

JE

D

Bfftcsckc5CaitTa2t5(dspaciiSptmm

Wcgjdo

bpttsc

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2008;6:298–301

erformance Characteristics of the Suspected Blood Indicator Feature inapsule Endoscopy According to Indication for Study

ONATHAN M. BUSCAGLIA, SAMUEL A. GIDAY, SERGEY V. KANTSEVOY, JOHN O. CLARKE, PRISCILLA MAGNO,LAINE YONG, and GERARD E. MULLIN

ivision of Gastroenterology and Hepatology, Johns Hopkins University School of Medicine, Johns Hopkins Hospital, Baltimore, Maryland

aistgoat

ttt

lewsfiJtfrcltr

rfpenddippst

pi

ackground & Aims: The suspected blood indicator (SBI)eature of wireless capsule endoscopy (WCE) was developedor rapid screening of intestinal lesions with bleeding poten-ial. Our aim was to assess the accuracy and performanceharacteristics of the SBI according to the indications fortudy in a large cohort of patients. Methods: We reviewedollected data on all WCE studies performed at Johns Hop-ins Hospital from January 2006 to June 2007. Study indi-ations were as follows: anemia of unknown origin (n �3), obscure gastrointestinal bleeding (n � 112), suspectedrohn’s disease (n � 122), and other (n � 4). Concordantnd discordant findings between gastroenterologists’ read-ngs and SBI were recorded for each patient. Results: Aotal of 221 lesions with bleeding potential was detected.he overall sensitivity, specificity, positive predictive value,nd negative predictive value for the SBI were 56.4%, 33.5%,4.0%, and 67.3%, respectively. For actively bleeding lesions,he SBI sensitivity and positive predictive value were only8.3% and 70%, respectively. The sensitivity was highest64%) in patients undergoing WCE for suspected Crohn’sisease, with a negative predictive value of 80.4%. The sen-itivity was only 58.3% and 41.3%, respectively, in studieserformed for obscure gastrointestinal bleeding andnemia. Conclusions: Performance characteristics of theurrently available SBI feature in WCE are suboptimal andnsufficient to screen for lesions with bleeding potential. Evenn patients with active intestinal bleeding, the sensitivity ofBI was less than 60%, which is lower than previously re-orted. However, in patients with suspected Crohn’s disease,

he high sensitivity and negative predictive value of SBI mayake it a useful tool for the detection of large areas of abnor-al mucosa.

ireless capsule endoscopy (WCE) was first developed foradvanced imaging of the small intestine.1–3 Although its

linical efficacy is most proven in the diagnosis of obscureastrointestinal bleeding,4 – 6 it also has been helpful as an ad-

unct to radiologic studies for patients with suspected Crohn’sisease, celiac disease, small-bowel tumors, anemia of unknownrigin, chronic abdominal pain, and other indications.7–11

The WCE system (Given Imaging, Yoqneam, Israel) in com-ination with RAPID software (Duluth, Georgia) has a sus-ected blood indicator (SBI) feature that initially was developedo facilitate the screening of possible sites of active bleeding,hereby significantly reducing the reading time spent by a phy-ician-reviewer.12 The SBI function works by identifying red-

olored pixels on the viewing screen, and then automatically

nnotating the study scroll bar with a red hash mark; thusndicating to the reader that a potential bleeding lesion or otherite of pathology may be present. Previously reported studies onhe accuracy of the SBI either have been dedicated solely toastrointestinal bleeding,13 or were limited by a small numberf enrolled patients.12,14 The aim of our study was to assess theccuracy and performance characteristics of the SBI accordingo the indications for the study in a large cohort of patients.

MethodsConsecutive patients undergoing WCE at our institu-

ion from January 2006 through June 2007 were reviewed forhe study. Permission to review patient records was granted byhe Johns Hopkins University Institutional Review Board.

All patients were asked to refrain from eating or drinking ateast 8 hours before swallowing the Given M2A video capsulendoscope. Laxative bowel preparation was not used. Patientsere allowed to eat and drink 4 hours after the start of their

tudy. Each WCE study was interpreted by 1 of 5 board-certi-ed/board-eligible gastroenterologists (J.M.B., S.A.G., P.M.,

.O.C., and E.Y.). All 5 readers had experience reviewing greaterhan 50 cases each. Images were reviewed at a speed of 8 to 15rames per second. Captured thumbnail images and summaryeports were re-examined and verified by a separate, board-ertified gastroenterologist (G.E.M.) who had experience with ateast 250 cases. There was greater than 95% concordance be-ween the verifying reader (G.E.M.) and each of the 5 initialeviewers.

After each study, the interpreting physician was asked toecord all endoscopic findings within our WCE database. Theollowing pathologic lesions were considered significant as itertains to the SBI function of the Given software: ulcers,rosions, arteriovenous malformations, red spots, varices, ve-ous ectasias, blood, and blood clots. Physicians were asked toocument whether the SBI accurately predicted the lesionsiscovered within each patient’s study. Performance character-

stics of the SBI were defined by concordance between what wasositively sensed by the Given software system, and what wasositively detected by the physician-reviewer. Cases in which aignificant lesion was found by the reviewer, but not detected byhe SBI, also were documented. Those cases in which SBI

Abbreviations used in this paper: GI, gastrointestinal; NPV, negativeredictive value; PPV, positive predictive value; SBI, suspected blood

ndicator; WCE, wireless capsule endoscopy.© 2008 by the AGA Institute

1542-3565/08/$34.00

doi:10.1016/j.cgh.2007.12.029

cfrd

bosC(tbssnmy

ss3mv01i3ta

ofi1bp

ttoapsn

Fo

T

MFM

T

UEA

RVVB

T

T

T

S

S

S

S

March 2008 PERFORMANCE OF THE SUSPECTED BLOOD INDICATOR 299

orrectly predicted the findings in one part of the study, butailed to detect the lesions in another part of the study, wereecorded as well. The indications for each WCE study also wereocumented for all patients.

ResultsA total of 291 patient WCE recordings were studied

etween January 2006 and June 2007. When the total numberf studies was divided according to clinical indications, theingle most common indication was suspicion of small-bowelrohn’s disease in patients with diarrhea and abdominal pain

n � 122; 41.9%). Obscure gastrointestinal (GI) bleeding washe second most common indication (n � 112; 38.5%), followedy anemia of unknown origin (n � 53; 18.2%). Four additionaltudies were performed for other indications (1.4%). These 4tudies were not included in the analysis, making the finalumber of WCE studies equal to 287 (Table 1). There were 125ales (43.6%) and 162 females (56.4%), with a mean age of 53

ears.A total of 221 pathologic lesions were identified in 287

tudies by 5 interpreting physicians (Table 2). Mucosal redpots were the most common lesions recognized (69 of 221;1.2%), followed by erosions (55 of 221; 24.9%), arteriovenousalformations (41 of 221; 18.6%), ulcers (41 of 221; 18.6%),

enous ectasias (11 of 221; 5%), and intestinal varices (1 of 221;.5%). There were 15 cases in which blood was recognized, and2 separate lesions were documented to account for the bleed-

ng: 4 ulcers, 3 erosions, and 5 arteriovenous malformations. Incases (1.4%), blood clots or active bleeding was documented in

he small bowel by the interpreting physician, yet no discern-ble lesion was recognized (Table 2).

able 1. Patient Demographics and Study Indications

Indicationfor WCE

Anemia ofunknown

origin(n � 53)

Obscure GIbleeding

(n � 112)

SuspectedCrohn’sdisease

(n � 122)Total (%)

(n � 287)

ales 25 64 36 125 (43.6)emales 28 48 86 162 (56.4)ean age, y 53.7 58.9 46.3 53.0

able 2. Pathologic Lesions With Bleeding Potential Foundin 287 WCE Studies

Lesion

Withoutactive

bleeding (%)Actively

bleeding (%) Total (%)

lcers 37 (16.7) 4 (1.8) 41 (18.6)rosions 52 (23.5) 3 (1.4) 55 (24.9)rteriovenousmalformations

36 (16.3) 5 (2.3) 41 (18.6)

ed spots 69 (31.2) 0 69 (31.2)arices 1 (0.5) 0 1 (0.5)enous ectasias 11 (5.0) 0 11 (5.0)lood clots and/or activebleeding

— 3 (1.4) 3 (1.4)

total 206 (93.2) 15 (6.8) 221 (100)

There was a similar proportion of overall patients in thebscure GI bleeding group and the anemia group with positivendings detected by the interpreting physician: 32.1% (36 of12) and 32.1% (17 of 53), respectively. In the suspected small-owel Crohn’s disease group, 20.5% (25 of 122) of patients hadositive findings.

Table 3 highlights the concordance between the reader andhe SBI function of the software. Performance characteristics ofhe SBI did not differ among the 5 different readers. Of a totalf 287 studies, 44 patients had a significant lesion with anssociated SBI (true positive). In 139 patients, an SBI wasresent without any discernable disease (false positive). Thetudy was negative in 70 patients without an SBI present (trueegative), and 34 patients had a lesion recognized by the phy-

igure 1. Example of a false-negative SBI in a patient with obscure-vert GI bleeding found to have lymphonodular hyperplasia on resec-

able 3. Concordance Between Reader and theSBI Function

Anemia ofunknown

originObscure GIbleeding

SuspectedCrohn’sdisease Total

otal number ofanalyzed recordings

53 112 122 287

BI present, significantlesion present (truepositive), n

7 21 16 44

BI present, significantlesion absent (falsepositive), n

24 55 60 139

BI absent, significantlesion absent (truenegative), n

12 21 37 70

BI absent, significantlesion present (falsenegative), n

10 15 9 34

ion of the terminal ileum.17 PillCam (Given Imaging, Yogneam, Israel)

s1t

2ip(t

t3ui5bwot

tcw3www

uohid

Iswifc

lihisf

Sp6dpiimT6ipstf

cmsfbrsS6iceit

taCcpsdsg

s

T

SSPN

N

T

SP

N

300 BUSCAGLIA ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

ician-reviewer without an associated SBI (false negative, Figure). Concordance values based on each of the 3 major indica-ions for the study also are listed in Table 3.

Table 4 lists the performance characteristics of the SBI in all87 WCE studies, as well as performance characteristics accord-

ng to each clinical indication. The overall sensitivity, specificity,ositive predictive value (PPV), and negative predictive value

NPV) of the SBI were 56.4%, 33.5%, 24.0%, and 67.3%, respec-ively.

In patients undergoing WCE for anemia of unknown origin,he sensitivity, specificity, PPV, and NPV of the SBI were 41.3%,3.3%, 22.6%, and 54.5%, respectively (Table 4). In patientsndergoing WCE for obscure GI bleeding, the sensitivity, spec-

ficity, PPV, and NPV of the SBI were 58.3%, 27.6%, 27.6%, and8.3%, respectively. When performed for the suspicion of small-owel Crohn’s disease, the sensitivity, specificity, PPV, and NPVere 64.0%, 38.1%, 21.1%, and 80.4%, respectively. Factoring innly those cases with active bleeding, the overall sensitivity ofhe SBI was 58.3%, with a PPV of 70.0% (Table 5).

When accounting for all patients studied, the sensitivity ofhe SBI was highest (64%) in those undergoing WCE for suspi-ion of small-bowel Crohn’s disease. The specificity of the SBIas poor for all 3 indications, with the highest value being8.1% in the suspected Crohn’s group (Table 4). The NPV alsoas highest in this same group, with a value of 80.4%. The PPVas poor in all groups, with a peak value of 27.6% in patientsith obscure GI bleeding.

DiscussionSince the invention of capsule endoscopy, it has been

sed most often and most effectively for the diagnosis ofbscure GI bleeding.15,16 More recently, however, its indicationsave broadened with its usefulness shown in the diagnosis of

nflammatory bowel disease and the evaluation of chronic ab-ominal pain.7–11

Early in its development, the WCE software system (Givenmaging) offered an SBI feature with each of its recordings. Thismall red line appearing at the top of the time tracing initiallyas designed as a rapid screening tool, allowing physicians to

dentify areas of active bleeding quickly.12 Although the SBIeature has been available for more than 5 years, reports of itslinical usefulness are relatively sparse, usually completed on a

able 4. Performance Characteristics of the SBI for the Pred

Indicationfor WCE

Anemia ofunknown origin

Obscurbleed

ensitivity 41.3 (19.4–66.6) 58.3 (40.1pecificity 33.3 (19.1–51.1) 27.6 (18.3PV 22.6 (10.3–41.5) 27.6 (18.3PV 54.5 (32.7–74.9) 58.3 (40.1

OTE. Percentages (95% confidence intervals) shown.

able 5. Sensitivity and PPV of the SBI in 15 WCE StudiesWith Blood Clots or Active Bleeding

ensitivity 58.3 (28.6–83.6)PV 70.0 (35.4–91.9)

aOTE. Percentages (95% confidence intervals) shown.

imited number of patients, and showing significant variationsn performance characteristics.12–14 Furthermore, other studiesave not reported the accuracy of SBI when capsule endoscopy

s used for indications other than active bleeding (ie, suspectedmall-bowel Crohn’s disease, anemia of unknown origin, and soorth).

In 2003, Liangpunsakul et al12 first reported the accuracy ofBI in 24 patients undergoing WCE. Eighteen studies wereerformed for occult GI bleeding or iron-deficiency anemia, andfor abdominal pain. Overall, the sensitivity of the SBI in the

etection of small-bowel lesions with bleeding potential wasoor at 25.7%. However, if only actively bleeding lesions in the

ntestine were considered, the sensitivity of the SBI was signif-cantly better (81.2%). Signorelli et al13 also studied the perfor-

ance characteristics of the SBI in 100 consecutive patients.he overall sensitivity was poor (40.9%), and it only increased to0.9% in the group of patients with red blood or active bleeding

n the small bowel. D’Halluin et al14 looked at a larger cohort ofatients (n � 156) with obscure GI bleeding. The overall sen-itivity of the SBI for detecting low-risk or high-risk lesions inhe small bowel still was low (37%), and it increased to only 57%or the detection of active bleeding.

The aim of our study was to verify the SBI accuracy in a largeohort of patients (n � 287), and to examine the SBI perfor-ance characteristics according to the indications for the

tudy. Altogether, our findings suggest that when accountingor all possible lesions with bleeding potential in the smallowel, the sensitivity of the SBI may be higher than originallyeported (56.4% vs 25%– 41%). More importantly, however, theensitivity in our study was not improved significantly when theBI was used only in patients with active bleeding (58.3% vs0%– 81% in previous studies). This finding has significant clin-

cal relevance, showing that the current version of the SBIannot be a substitute for a complete and detailed review of thentire WCE study; therefore, it should not serve as the primarynterpretation modality by those physicians who do not haveime to read the entire report from mouth to cecum.

The sensitivity of the SBI in our study was highest in pa-ients undergoing WCE for the evaluation of diarrhea andbdominal pain with a clinical suspicion of small-bowelrohn’s disease (64%). Previous publications showed signifi-

antly lower overall sensitivity values of 25% to 41% in studieserformed for a similar indication.12–14 It appears from ourtudy that with an NPV of 80.4%, the use of an SBI to aid in theetection of larger areas of abnormal mucosa in patients withuspected inflammatory bowel disease warrants further investi-ation.

The main limitation of our study was its retrospective de-ign. A large prospective study will be necessary to verify the SBI

of Pathologic Lesions With Bleeding Potential

SuspectedCrohn’s disease Total

0) 64.0 (42.6–81.3) 56.4 (44.7–67.4)3) 38.1 (28.6–48.6) 33.5 (27.2–40.4)3) 21.1 (12.9–32.2) 24.0 (18.2–31.0)0) 80.4 (65.6–90.1) 67.3 (57.3–76.0)

iction

e GIing

–74.–39.–39.–74.

ccuracy and performance characteristics shown by our retro-

sswWc

ptdws(Cac

1

1

1

1

1

1

1

1

Hm2

dSsYir

March 2008 PERFORMANCE OF THE SUSPECTED BLOOD INDICATOR 301

pective review. Improvements in the existing Given softwareystem could increase the diagnostic yield of the SBI, and thusill make it a reliable screening tool for patients undergoingCE. Until then, there is no substitution for an expert physi-

ian reviewing the study in full.In conclusion, our large retrospective study showed that the

erformance characteristics of the currently available SBI fea-ure in WCE are suboptimal and insufficient for adequateetection of lesions with bleeding potential. Even in patientsith active small intestinal bleeding, the sensitivity of the SBI

till is less than 60%, which is lower than previously reported60%– 81%). However, in patients with suspected small-bowelrohn’s disease, our findings indicate that the high sensitivitynd NPVs of SBI warrant further investigation of this tool as alinical aid in the detection of large areas of abnormal mucosa.

References

1. Iddan G, Meron G, Glukhovsky A, et al. Wireless capsule endos-copy. Nature 2000;405:417.

2. Meron GD. The development of the swallowable video capsuleM2A. Gastrointest Endosc 2000;52:817–819.

3. ASGE Technology Assessment Committee. Wireless capsule en-doscopy. Gastrointest Endosc 2002;56:621–624.

4. Saurin JC, Delvaux M, Vahedi K, et al. Clinical impact of capsuleendoscopy compared to push enteroscopy: 1-year follow-upstudy. Endoscopy 2005;37:318–323.

5. Hartman D, Schmidt H, Bolz G, et al. A prospective two-centerstudy comparing wireless capsule endoscopy with intraoperativeenteroscopy in patients with obscure GI bleeding. GastrointestEndosc 2005;61:826–832.

6. Ell C, Remke S, May A, et al. The first prospective controlled trialcomparing wireless capsule endoscopy with push enteroscopy inchronic gastrointestinal bleeding. Endoscopy 2002;34:685–689.

7. Eliakim R, Fischer D, Suissa A, et al. Wireless capsule videoendoscopy is a superior diagnostic tool in comparison to bariumfollow-through and computerized tomography in patients with sus-pected Crohn’s disease. Eur J Gastroenterol Hepatol 2003;15:363–367.

8. Costamagna G, Shah SK, Riccioni ME, et al. A prospective trialcomparing small bowel radiographs and video capsule endoscopyfor suspected small bowel disease. Gastroenterology 2002;

123:999–1005. a

9. Herreiras JM, Caunedo A, Rodriguez-Tellez M, et al. Capsuleendoscopy in patients with suspected Crohn’s disease in nega-tive endoscopy? Endoscopy 2003;35:1–5.

0. Bardan E, Nadler M, Chowers Y, et al. Capsule endoscopy for theevaluation of patients with chronic abdominal pain. Endoscopy2003;35:688–689.

1. Vasquez-Iglesias J, Gonzalez-Conde B, Estevez-Prieto E, et al. Aprospective study of COX-2 inhibitors versus nonspecific NSAIDsinduced small bowel lesions using video capsule endoscopy.Endoscopy 2003;35:A183.

2. Liangpunsakul S, Mays L, Rex DK. Performance of Given sus-pected blood indicator. Am J Gastroenterol 2003;98:2676–2768.

3. D’Halluin PN, Delvaux M, Lapalus MG, et al. Does the suspectedblood indicator improve the detection of bleeding lesions bycapsule endoscopy? Gastrointest Endosc 2005;61:243–249.

4. Signorelli C, Villa F, Rondonotti E, et al. Sensitivity and specificityof the suspected blood identification system in video capsuleenteroscopy. Endoscopy 2005;37:1170–1173.

5. Soussan BE, Antonietti M, Herve S, et al. Diagnostic yield andtherapeutic implications of capsule endoscopy in obscure gas-trointestinal bleeding. Gastroenterol Clin Biol 2004;28:1068–1073.

6. Gupta R, Lakhtakia S, Tandan M, et al. Capsule endoscopy inobscure gastrointestinal bleeding—an Indian experience. IndianJ Gastroenterol 2006;25:188–190.

7. Buscaglia J, Carroll C, Daniels J, et al. Benign lymphoid hyper-plasia: a rare cause of obscure overt bleeding in an adult. Gas-trointest Endosc 2007;66:1248–1250.

Address requests for reprints to: Jonathan M. Buscaglia, MD, Johnsopkins Hospital, 1830 E. Monument Street, Room 7100-A, Balti-ore, Maryland 21205. e-mail: jbuscaglia@jhmi.edu; fax: (410) 955-

108.Jonathan Buscaglia, MD, initiated the study design, conducted the

ata analysis, and prepared the manuscript. Samuel Giday, MD, andergey Kantsevoy, MD, PhD, aided in the data analysis and the manu-cript preparation. John Clarke, MD, Priscilla Magno, MD, and Elaineong, MD, collectively interpreted greater than 60% of the WCE stud-

es, and each edited the manuscript draft. Gerard Mullin, MD, MHS,eviewed all WCE studies, contributed to the study design and data

nalysis, and edited the manuscript draft.

O

PP

P

G

BvfiaewtBswufcScTostt5PmntatsPofifi

Ttrc

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2008;6:302–308

RIGINAL ARTICLES—ALIMENTARY TRACT

atient Predictors of Esophageal Stricture Development Afterhotodynamic Therapy

ATRICK YACHIMSKI, WILLIAM P. PURICELLI, and NORMAN S. NISHIOKA

astrointestinal Unit and Wellman Center for Photomedicine, Massachusetts General Hospital, Boston, Massachusetts

ea

ng(pspcti

tHafscomao

twaantt

ifdpnrm

bro

See Curvers W et al on page 670 for companionarticle in the March 2008 issue ofGastroenterology.

ackground & Aims: The most common significant ad-erse event after photodynamic therapy (PDT) with por-mer sodium is esophageal stricture formation. This studyssessed whether pretreatment variables, including priorndoscopic therapy for Barrett’s esophagus, are associatedith post-PDT stricturing. Methods: Data from all pa-

ients who had undergone PDT with porfimer sodium forarrett’s esophagus with high-grade dysplasia, intramuco-

al carcinoma, or T1 cancer at our institution since 1997ere reviewed. Results: One hundred sixteen patientsnderwent 160 courses of PDT. The incidence of strictureormation after index PDT was 16% (19/116). For all PDTourses, the overall incidence of stricture was 23% (37/160).tricture rate was significantly higher after a second PDTourse compared with index PDT (43% vs 16%, P � .0007).here was no association between post-PDT stricture devel-pment and age, gender, body mass index, or prior endo-copic mucosal resection. Patients who developed a stric-ure had a longer length of Barrett’s esophagus beforereatment than those who did not develop a stricture (7.7 vs.7 cm for index PDT only, P � .025; 7.4 vs 5.7 cm for allDT courses, P � .007). Length of Barrett’s esophagus,ultiple PDT courses, and presence of intramucosal carci-

oma on pretreatment pathology were independent predic-ors of post-PDT stricture in a stepwise logistic regressionnalysis controlling for treatment variables, includingreatment length. Conclusions: An increased risk oftricture development was seen after multiple courses ofDT. An association between post-PDT stricture and lengthf Barrett’s esophagus but not treatment length was alsoound. Endoscopic mucosal resection did not appear tonfluence the likelihood of stricture development after por-mer sodium– based PDT.

he term Barrett’s esophagus is used to describe metaplasia ofthe distal esophageal mucosa where columnar intestinal-

ype mucosa replaces squamous mucosa. This metaplasia is aesult of chronic distal esophageal acid exposure1 and is typi-

ally encountered in adults with gastroesophageal reflux dis-

ase. Barrett’s esophagus is a precursor lesion to esophagealdenocarcinoma.2

The sequence from Barrett’s esophagus to esophageal ade-ocarcinoma is believed to occur by stepwise neoplastic pro-ression. Barrett’s esophagus progresses to low-grade dysplasiaLGD); LGD progresses to high-grade dysplasia (HGD); HGDrogresses to adenocarcinoma. Among individuals with long-egment Barrett’s esophagus (�3 cm), 1 in 200 per year (or 0.5%er year) will develop adenocarcinoma.3 Esophageal adenocar-inoma now has the fourth highest incidence of all gastroin-estinal tract cancers diagnosed in the United States4 and isncreasing at a rapid rate.

Distal esophagectomy is considered by most physicians to behe standard of care for patients with Barrett’s esophagus andGD. For patients undergoing esophagectomy for HGD, oper-

tive mortality rates of approximately 2% have been reportedrom high-volume centers.5,6 However, morbidity rates are con-iderably higher.7 Elderly patients and patients with medicalomorbidities are often rejected as surgical candidates, whereasther patients decline surgical management. These patientsight instead benefit from nonoperative treatment of HGD,

nd this has prompted the development of endoscopic meansf ablating Barrett’s mucosa.

Photodynamic therapy (PDT) is a Food and Drug Adminis-ration–approved endoscopic treatment for Barrett’s esophagusith HGD.8 During PDT, patients receive a photosensitizinggent followed by illumination with red laser light to achieveblation of Barrett’s mucosa. The most common clinically sig-ificant adverse effect of PDT is esophageal stricture forma-ion.9 In some published series, more than 30% of patientsreated with PDT developed esophageal stricture.10,11

The reason strictures form after PDT with porfimer sodiums not known. It has been hypothesized that the deep, circum-erential tissue injury and resulting inflammatory reaction pro-uced by porfimer sodium PDT incites a fibrotic response thatroduces stricturing. However, post-treatment oral steroids doot appear to reduce the likelihood of stricture formation orecurrence.12 Potential endoscopic treatment parameters that

ight be associated with the formation of strictures after PDT

Abbreviations used in this paper: ALA, aminolevulinic acid; BMI,ody mass index; CI, confidence interval; EMR, endoscopic mucosalesection; HGD, high-grade dysplasia; LGD, low-grade dysplasia; OR,dds ratio; PDT, photodynamic therapy.

© 2008 by the AGA Institute1542-3565/08/$34.00

doi:10.1016/j.cgh.2007.12.001

itm

pad

P2I

tcAltggp

pfl(s

tgildpcps

HpiTctiwdspr

iSwpcdcc1l

1lJe

babttepoe

asdat

uiatt

9Sat

wincc(bcE

tnf

uwPpmEw

March 2008 PATIENT PREDICTORS OF ESOPHAGEAL STRICTURE DEVELOPMENT AFTER PDT 303

nclude light dose,13 centering balloon length,11,12, multiplereatments of the same esophageal segment,10 and pre-treat-

ent of nodular mucosa.10

The purpose of this study was to determine whether anyretreatment variables including prior endoscopic therapy aressociated with an increased likelihood of esophageal strictureevelopment after PDT.

Patients and MethodsRecords were reviewed for all patients who underwent

DT in a single large urban teaching hospital between 1997 and006. This retrospective review was approved by the Partnersnstitutional Review Board.

Patients who underwent porfimer sodium– based PDT forreatment of Barrett’s esophagus with HGD, intramucosal car-inoma, or T1 adenocarcinoma were included in this analysis.ll patients in the analysis were required to have undergone at

east 1 follow-up endoscopy at our institution after PDT. Pa-ients who underwent PDT for palliative treatment of esopha-eal adenocarcinoma stage T2 or greater (6 patients), esopha-eal squamous cell carcinoma (3 patients), or gastric cancer (3atients) were excluded.

Sources of data included endoscopy reports, pathology re-orts, clinic notes, and, when available, endoscopy unit nursingow sheets. All records were reviewed by a single investigator

P.Y.), who was unaware of clinical outcome (stricture versus notricture) at the time of data extraction.

Length of Barrett’s esophagus was determined by the dis-ance between the squamocolumnar junction and the top of theastric folds. All patients underwent a biopsy protocol consist-ng of 4-quadrant jumbo forceps biopsies every 2 cm along theength of Barrett’s esophagus to stage and grade the extent ofysplasia. Histopathologic diagnosis was determined by an ex-erienced gastrointestinal pathologist at our institution. Forases in which diagnostic endoscopy and biopsy had beenerformed at a referring institution, pathology specimens wereent to our institution for diagnostic confirmation before PDT.

All patients underwent endoscopic ultrasound before PDT.igh-frequency probe endoscopic ultrasound was not routinelyerformed. If there was evidence of a focal or nodular lesion on

nitial endoscopy and if endoscopic ultrasound excluded stage2 or greater disease, endoscopic mucosal resection (EMR)ould be performed. No additional or alternative endoscopicreatment was offered before PDT. High-dose proton pumpnhibitor therapy (omeprazole 40 mg twice daily or equivalent)as prescribed for all patients before PDT and continued in-efinitely after PDT. Some patients underwent esophageal pHtudies if they experienced heartburn or reflux symptoms onroton pump inhibitor therapy; however, pH studies were notoutinely performed.

A standardized PDT protocol was used that consisted ofntravenous porfimer sodium (Photofrin; Axcan Pharma, Mont-aint-Hilaire, Quebec, Canada) administration on day 0. Aeight-based dose of 2 mg/kg was administered, except for aeriod of time before 1998 when the maximum dose wasapped at 150 mg. Patients returned to the endoscopy unit onay 2 for initial light exposure. The majority of cases used aylindrical diffusing fiber that was passed through the workinghannel of a standard diagnostic gastroscope. Fiber lengths of, 2.5, or 5 cm were used on the basis of the desired treatment

ength. Centering balloons were used in some patients between B

998 and 2005. The target esophageal mucosa was exposed toaser light at a wavelength of 630 nm and a total energy of 150/cm. Any deviation from this protocol was documented in thendoscopy report.

A single course of PDT can have 1 or 2 light exposures separatedy 48 hours. Patients returned to the endoscopy unit on day 4 forpossible second light exposure. This second light exposure coulde used to either extend the length of treatment and/or re-treathe previously treated area if the mucosal PDT effect appearedo be uneven. Day 4 light exposure was withheld only if thextent of mucosal injury was judged to be severe, if theatient was experiencing significant symptoms of dysphagia,dynophagia, or chest discomfort, or if the patient otherwisexhibited poor tolerance to PDT after initial light exposure.

All patients returned for surveillance endoscopy 3 monthsfter PDT. Endoscopy was performed sooner when post-PDTtricture was suspected on the basis of dysphagia. Stricture wasefined as esophageal luminal narrowing preventing passage ofstandard gastroscope. All dilations were performed with a

hrough-the-scope esophageal balloon dilator.The majority of patients receiving PDT at our institution

ndergo all endoscopic follow-up at our institution. However,n certain cases such as extensive travel, patients underwentdditional endoscopic evaluation at an outside institution. Inhese cases endoscopic reports were retrieved and reviewed forhis analysis.

Statistical AnalysisAll statistical analyses were performed with SAS version

.1 software (SAS Institute, Cary, NC). For univariate analyses,tudent t test or Wilcoxon rank sum testing was performed fornalysis of continuous variables. Chi-square test or Fisher exactest was used for analyses of binary or ordinal variables.

After univariate analysis, stepwise logistic regression analysisas performed. Multiple treatment variables were incorporated

nto the model, including porfimer sodium dose, light dose,umber of light exposures, treatment length, and use of aentering balloon versus cylindrical diffusing fiber. Patientharacteristics analyzed included age, gender, body mass indexBMI), length of Barrett’s esophagus, pathology (treated as ainary variable [HGD versus intramucosal and/or submucosalarcinoma]), history of prior stricture, and history of priorMR.

The threshold for statistical significance was defined as awo-sided P value less than .05. Because of the exploratoryature of the analysis, Bonferroni correction was not per-

ormed.

ResultsPretreatment Patient CharacteristicsOne hundred sixteen patients with Barrett’s esophagus

nderwent 160 courses of PDT. Seventy-nine patients under-ent a single course of PDT, whereas 31 patients underwent 2DT courses, 5 patients underwent 3 PDT courses, and 1atient underwent 4 PDT courses. Seven patients were part of aulticenter study and were included in a previous report.10

ighty-one percent of patients were male, and 99% were white,ith a mean age of 70.4 years at the time of index PDT. Mean

MI was 28.6 kg/m2 at the time of index PDT.

pbhatept

frw

wHsd

bell

wraro

pwpwEEe

(ts

1Tif

atsm

isTPrSa

Ppc

stT

Psgo(1(slt

304 YACHIMSKI ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

History of esophageal stricture was present in 6 of 116atients (5%). Five patients had undergone esophageal surgeryefore PDT. Three had undergone Nissen fundoplication, and 2ad undergone partial esophagectomy for treatment of esoph-geal cancer and were referred for PDT for recurrent cancer. Ofhese latter 2 patients, one had also been previously treated forsophageal cancer with chemotherapy and radiation. No otheratients reported a history of esophageal or mediastinal radia-ion.

The location of the squamocolumnar junction and gastricolds was documented in 96% of endoscopy reports. In theemainder of reports, only total length of Barrett’s esophagusas described.

Before index PDT, the mean length of Barrett’s esophagusas 6.0 cm. Fifty-nine patients (51%) had been diagnosed withGD as the most advanced pathology on pre-PDT biopsy

pecimens, whereas the remaining 57 patients (49%) had beeniagnosed with intramucosal carcinoma.

Endoscopic Mucosal ResectionThirty patients underwent a total of 31 EMR sessions

efore index PDT (Figure 1). EMR was performed by usingither a saline-assisted cap and snare technique or a bandigator device (Duette; Cook Medical, Bloomington, IN) fol-owed by snare resection.

The mean number of EMR specimens resected per sessionas 1.7 (range, 1– 6). Nine patients underwent more than 1

esection in a given session, whereas the remainder underwentsingle resection. Two patients experienced post-EMR bleeding

equiring immediate further endoscopic therapy. There were nother observed complications of EMR.

Pathologic examination of EMR specimens revealed nondys-lastic Barrett’s esophagus in 2 patients, Barrett’s esophagusith LGD in 1 patient, Barrett’s esophagus with HGD in 4atients, intramucosal carcinoma in 18 patients, and carcinomaith submucosal involvement in 3 patients. In 1 patient, theMR specimen was not available for review, and in 1 patient theMR specimen contained no epithelial tissue. No patient wasxcluded from PDT on the basis of the EMR findings.

Median time between EMR and index PDT was 61 daysrange, 16 –762 days). There was no association between meanime from EMR to index PDT and development of post-PDT

Figure 1. Patients undergoing EMR before PDT.

tricture (119 days in the group with post-PDT stricture versus

05 days in the group without post-PDT stricture, P � .25).here was no difference in the mean number of EMR resections

n patients who did and did not develop post-PDT stricture (1.7or both groups).

Photodynamic Therapy VariablesThe majority of patients received 2 light exposures, with

maximum light dose of 150 J/cm each session and withouthe use of a centering balloon. Documentation of porfimerodium dose was unavailable for 15 PDT courses, and docu-

entation of light dose was missing for 13 courses.

Stricture RateThe overall incidence of stricture was 23% (37/160). The

ncidence of stricture after index PDT was 16% (19/116). Thetricture rate after a second PDT course was 43% (16/37).he increase in stricture rate after a second course versus indexDT was statistically significant (43% vs 16%, P � .0007; riskatio, 2.64; 95% confidence interval [CI], 1.52– 4.59) (Figure 2).tricture rates of 33% (2/6) after 3 PDT courses and 0% (0/1)fter 4 PDT courses were observed.

Of the 19 patients who developed a stricture after indexDT, 6 underwent a second PDT course. Three of these 6atients (50%) developed recurrent stricture after the secondourse of PDT.

Median number of dilations required to treat post-PDTtricture was 3 (range, 0 to 20). In 1 patient, a stricture wasraversed with the gastroscope without use of a balloon dilator.here were no observed complications of stricture dilation.

Univariate and Logistic Regression AnalysisIn univariate analyses of patients undergoing index

DT only, factors associated with the development of post-PDTtricture were length of Barrett’s esophagus (7.7 � 4.2 cm in theroup with post-PDT stricture vs 5.7 � 3.1 cm in group with-ut post-PDT stricture, P � .025) and PDT treatment length6.5 � 1.4 cm in the group with post-PDT stricture and 5.6 �.6 cm in the group without post-PDT stricture, P � .03)Table 1). There was no apparent association between post-PDTtricture and any other treatment variable including number ofight exposures, light dose, and use of a diffusing fiber ratherhan centering balloon.

Figure 2. Stricture rate: index PDT vs second PDT course.

soPst.p

27hrewwlur

buP[ca1t9c1

itcmsWplpimd

T

NAMBPPP

L

L

L

L

TP

a

T

NAMBPPP

L

L

L

L

TP

a

T

I

A

March 2008 PATIENT PREDICTORS OF ESOPHAGEAL STRICTURE DEVELOPMENT AFTER PDT 305

In analyses of all 160 PDT courses, an association per-isted between development of post-PDT stricture and lengthf Barrett’s esophagus (7.4 � 3.7 cm in the group with post-DT stricture vs 5.7 � 3.2 cm in the group without post-PDTtricture, P � .007). However, no association was seen withreatment length (Table 2). A trend toward significance (P �08) was observed for the association between light dose andost-PDT stricture.

Among patients with prior stricture who underwent PDT,2% (8/37) developed a post-PDT stricture, as compared with a% (9/123) post-PDT stricture rate among patients without aistory of esophageal stricture (P � .03). The unadjusted riskatio for post-PDT stricture in a patient with history of priorsophageal stricture was 2.32 (95% CI, 1.27– 4.23), comparedith a patient without a prior stricture. In addition, patientsith intramucosal or submucosal carcinoma appeared more

ikely than patients with HGD to develop post-PDT stricture onnivariate analysis (31% [23/75] vs 16% [14/85]; unadjusted riskatio, 1.86; 95% CI, 1.03–3.35).

Stepwise logistic regression analyses were performed foroth index PDT alone and for all PDT courses. For patientsndergoing index PDT, the only variable associated with post-DT stricture was length of Barrett’s esophagus (odds ratio

OR], 1.23; 95% CI, 1.04 –1.44) (Table 3). In analyses of all PDTourses, length of Barrett’s esophagus was confirmed to bessociated with post-PDT stricture (OR, 1.27; 95% CI, 1.10 –.47). Additional variables associated with post-PDT stricture inhis inclusive analysis were multiple PDT courses (OR, 3.15;5% CI, 1.43– 6.94) and presence of intramucosal or submucosalarcinoma on pretreatment pathology (OR, 4.62; 95% CI, 1.66 –2.89) (Table 3).

able 1. Univariate Predictors of Stricture After Index PDT

Stricture No stricture P value

19 97ge 67.9 � 12.4 y 70.9 � 9.7 y .24ale gender 15/19 (79%) 79/97 (81%) .76MI (kg/m2) 29.6 � 7.5 28.4 � 6.4 .58rior EMR 7/19 (37%) 23/97 (24%) .26rior stricture 2/19 (11%) 4/97 (4%) .25orfimer sodium

dose173.1 � 41.6 mg 170.7 � 41.6 mg .86

ight dose300 J/cm 0/19 (0%) 4/92 (4%) .65150 J/cm 18/19 (95%) 83/92 (90%)130 J/cm 1/19 (5%) 5/92 (5%)

ight exposures2 15/19 (79%) 87/97 (90%) .241 4/19 (21%) 10/97 (10%)

ight deliveryCylindrical fiber 18/19 (95%) 83/97 (86%) .46Centering balloon 1/19 (5%) 14/97 (14%)

ength Barrett’sesophagus

7.7 � 4.2 cm 5.7 � 3.1 cm .025

reatment length 6.5 � 1.4 cm 5.6 � 1.6 cm .03athologic

diagnosisHGD 8/19 (42%) 51/97 (53%) .40Intramucosal

carcinomaa11/19 (58%) 46/97 (47%)

Includes 3 patients with submucosal (T1sm) disease. a

DiscussionPDT is an established endoscopic technique for ablat-

ng Barrett’s esophagus with HGD and/or early-stage intraepi-helial neoplasia. Patients with medical comorbidities that pre-lude surgical management and patients who decline surgicalanagement might be appropriate candidates for PDT. In

ome patients, multiple courses of PDT might be required.ith close biopsy surveillance, many patients might not exhibit

rogression to invasive cancer during extended post-PDT fol-ow-up. Other patients might experience downgrading of dys-lasia or complete ablation of Barrett’s epithelium. In the latter

nstance, glandular Barrett’s mucosa buried beneath neosqua-ous epithelium is a concern and has been associated with

isease recurrence.14,15

able 2. Univariate Predictors of Stricture After PDT (AllTreatment Courses)

Stricture No stricture P value

37 123ge 69.2 � 11.3 y 70.4 � 10.1 y .54ale gender 28/37 (76%) 99/123 (80%) .53MI (kg/m2) 28.9 � 6.8 28.3 � 6.0 .77rior EMR 11/37 (30%) 26/123 (21%) .28rior stricture 8/37 (22%) 9/123 (7%) .03orfimer sodium

dose169.5 � 39.9 mg 169.2 � 39.5 mg .94

ight dose300 J/cm 5/35 (14%) 5/112 (4%) .08260 J/cm 1/35 (3%) 0/112 (0%)225 J/cm 0/35 (0%) 1/112 (1%)150 J/cm 28/35 (80%) 98/112 (88%)130 J/cm 1/35 (3%) 8/112 (7%)

ight exposures2 30/37 (81%) 111/123 (90%) .151 7/37 (19%) 12/123 (10%)

ight deliveryDiffusing fiber 33/37 (89%) 102/122 (84%) .41Centering balloon 4/37 (11%) 20/122 (16%)

ength Barrett’sesophagus

7.4 � 3.7 cm 5.7 � 3.2 cm .007

reatment length 6.2 � 1.5 cm 5.7 � 1.6 cm .23athologic

diagnosisHGD 14/37 (38%) 71/123 (58%) .03Intramucosal

carcinomaa23/37 (62%) 52/123 (42%)

Includes 3 patients with submucosal (T1sm) disease.

able 3. Predictors of Stricture After PDT: LogisticRegression Analysis

P value OR 95% CI

ndex PDTLength Barrett’s esophagus .01 1.23 1.04–1.44

ll PDT coursesLength Barrett’s esophagus .001 1.27 1.10–1.47Multiple PDT courses .004 3.15 1.43–6.94Pathology (Intramucosalcarcinomaa vs HGD)

.003 4.62 1.66–12.89

Includes 3 patients with submucosal (T1sm) disease.

Porstrs

rac(CsdtItmbcnpp

ilsP2s

fldm1rsIawtesvbd

wa

mmfdbpw

ovspEstctnlp

oaostdtsmtlea

siacnvpfi

e

T

OOPKP

a

306 YACHIMSKI ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

Esophageal stricture occurred at a rate of 16% after indexDT and 23% after all PDT courses in this study. A minimumf 3 months of endoscopic follow-up at our institution wasequired for inclusion in this study. The majority of post-PDTtrictures develop within 3 weeks after treatment,13,16 so it isherefore likely that this analysis, although retrospective, accu-ately captures and reports the true incidence of post-PDTtricture in this cohort.

Post-PDT stricture should be suspected in any patient whoeports dysphagia after PDT. Strictures are typically superficialnd might be effectively dilated with standard endoscopic ac-essories. Either through the scope balloon dilators or SavaryCook Medical Inc, Bloomington, IN) dilators16 might be used.omplications including esophageal perforation, while not ob-

erved in our study, are a possibility in patients undergoingilation of post-PDT stricture. As evident in our cohort, mul-iple dilations might be required to achieve stricture resolution.ntralesional steroid injection, stent placement, or home dila-ion techniques have been used by other investigators in a

inority of patients with refractory strictures. Patients shoulde instructed to adhere to a liquid or soft-solid diet, with highalorie oral supplements as necessary to maintain adequateutrition, until dysphagia has resolved. Overall, relatively fewatients experience significant weight loss as a consequence ofost-PDT stricture.10

Among patients who develop post-PDT stricture, the major-ty will achieve complete stricture resolution. In recently pub-ished follow-up data from the original PDT with porfimerodium (PORPDT) study, 94% of subjects who developed post-DT stricture were stricture-free at the completion of the initial-year study phase; and only 3 subjects developed recurrenttricture between years 2 and 5 of follow-up.17

Previous studies have identified treatment variables that in-uence the likelihood of stricture development, including lightose, centering balloon length, and pre-treatment of nodularucosa. Panjehpour et al13 demonstrated that a light dose of

15 J/cm led to a 15.3% incidence of severe stricture (defined asequiring 6 or more dilations), as compared with a 5%– 6%evere stricture rate in patients treated with 85 J/cm or 95 J/cm.n a longitudinal series of 100 patients, stricture rate was 52%fter use of a 3-cm centering balloon, as compared with 28%hen 5-cm or 7-cm centering balloons were used.11 However,

he reverse trend was observed in a smaller series evaluating theffect of oral steroid therapy on post-PDT stricture, with a 31%tricture rate in patients treated with 7-cm centering balloonsersus a 7% stricture rate in patients treated with 5-cm centeringalloons.12 Preliminary analysis of the PORPDT phase III trialata identified associations between nodule pre-treatment, as

able 4. Stricture Rate After Porfimer Sodium PDT for Barrett

Author, year Maximum light dose Light ex

verholt et al, 20038 100–250 J/cm 1–verholt et al, 200510 130 J/cmrasad et al, 200719 130–200 J/cm 1–eeley et al, 200729 300–400 J/cmresent study 130–300 J/cm 1–

Includes 26 patients who received hematoporphyrin derivative photosens

ell as repeated treatments on the same esophageal segmentnd stricture development.10

Our data confirm the findings of prior studies in identifyingultiple treatment courses as a factor associated with develop-ent of post-PDT stricture (Table 4). Less attention has been

ocused on treatment-independent patient variables that pre-ict stricture formation. This study identifies an associationetween length of Barrett’s esophagus and development ofost-PDT stricture after multivariable analysis. This associationas independent of treatment length.

Our data also suggest an association between intramucosalr submucosal carcinoma on pretreatment pathology and de-elopment of post-PDT stricture. In this study, dysplasia/tumortage was assessed by forceps biopsy and endosonography in allatients, whereas only one fourth of patients had undergoneMR. EMR is superior to endoscopic ultrasound in detectingubmucosal tumor invasion.18 It is conceivable that some pa-ients in our cohort might have had undetected submucosalarcinoma, which might have been more accurately staged ifhey had undergone EMR. However, although this study wasot designed to assess the efficacy of PDT in our cohort,

ong-term follow-up of these patients has failed to detect theresence of such carcinomas in the vast majority of patients.

A pathophysiologic basis for the associations between lengthf Barrett’s esophagus and depth of involvement, respectively,nd post-PDT stricture might be explained by the distributionf porfimer sodium. If there is preferential uptake of porfimerodium by neoplastic tissue, including dysplastic Barrett’s epi-helium, then this tissue might be subject to a greater extent orepth of tissue injury after light exposure. If this is the case,hen stricture rate would be expected to be lower after porfimerodium PDT of nondysplastic Barrett’s or esophageal squa-

ous mucosa. However, this hypothesis cannot be assessed onhe basis of available evidence, because there are limited pub-ished experience with PDT ablation of nondysplastic Barrett’spithelium and no data on the in vivo effect of PDT on esoph-geal squamous epithelium.

Additional limitations of this study include the relativelymall sample size, which limits statistical power. This is a factorn virtually all single-center studies of PDT for Barrett’s esoph-gus, even when considering data from relatively high-volumeenters. A second limitation of this study is its retrospectiveature. The associations reported in this study should undergoalidation in a prospective analysis. Furthermore, because PDTrotocols vary among institutions, the external validity of ourndings is uncertain.

It is also possible that post-PDT stricture might be influ-nced by additional variables not included in this analysis. A

sophagus

res Patients (N)

Stricture rate

Index PDT All PDT courses

82 18% 30%132 12% 36%131a 23% 27%50 30% 42%

116 16% 23%

’s E

posu

22222

itizer.

draapte

otittcBpipps

rPdeh(bt(l

tfdtweovtdpd

phEcsspsP

emwIla

yspbs

atqSr4iwlp

etrIbptntt

gawsfHPtd

March 2008 PATIENT PREDICTORS OF ESOPHAGEAL STRICTURE DEVELOPMENT AFTER PDT 307

etailed medication, tobacco, and alcohol history had not beeneliably documented in a significant portion of our patients,nd these variables were therefore not incorporated into thenalysis. Because esophageal pH studies were not routinelyerformed to assess degree of acid suppression, it is conceivablehat differential degrees of acid suppression might have influ-nced stricture development in individual patients.

During the period of this study, porfimer sodium PDT wasffered to patients with Barrett’s esophagus and HGD or in-ramucosal carcinoma, as well as some patients with T1 diseasenvolving the submucosa (T1sm). The risk of post-PDT stric-ure was discussed with all patients before PDT. On the basis ofhe findings of this analysis, we offer the following additionalounseling: (1) patients with very long segments (�6 –7 cm) ofarrett’s esophagus might be at increased risk for stricture; (2)atients with intramucosal carcinoma or submucosal tumor

nvolvement might be at increased risk for stricture; and (3)atients can safely undergo EMR for either staging or thera-eutic intent before PDT, without increased risk of post-PDTtricture.

This last finding is in direct contrast with recently reportedesults in which an association between prior EMR and post-DT stricture was found.19 This discrepancy might be due toifferences in the patient cohort and/or PDT protocol. Forxample, the incidence of prior esophageal stricture was muchigher in the Mayo group of patients (24%) than in our cohort

5% before index PDT). Treated patients might therefore haveeen more prone to recurrent esophageal stricture. In addition,he typical light dose used in the Mayo protocol was higher200 J/cm vs 150 J/cm), and an association between intensity ofight dose and stricture has been demonstrated previously.

In addition, little detail is available regarding the compara-ive performance of EMR in the 2 studies. EMR might be usedor focal resection of a single esophageal nodule or for widerebulking of dysplastic Barrett’s epithelium. EMR of greaterhan 75% of total esophageal circumference has been associatedith post-EMR stricture.20 It is therefore conceivable that the

xtent of EMR might impact the likelihood of stricture devel-pment after EMR and PDT combination therapy. Time inter-al and esophageal healing between EMR and PDT mightheoretically impact the likelihood of post-PDT stricture. Ourata do not support this hypothesis, although the number ofatients who underwent both EMR and PDT limits power toetect differences.

The presence or absence of an association between EMR andost-PDT stricture is clinically important because EMR volumeas increased with the assistance of dedicated and easy-useMR devices.21,22 Appropriate, safe, and continued use of thisombination therapy for patients with Barrett’s esophagushould be contingent on its long-term efficacy, as well as itshort- and long-term side effect profile. As an increasing pro-ortion of patients undergo EMR (for either therapeutic ortaging purposes) before PDT (Figure 1), the impact of EMR onDT outcomes might become further elucidated.

For patients at high risk for post-PDT stricture, alternativendoscopic ablation techniques might provide alternative treat-ent options. Esophageal stricture is generally not observedhen aminolevulinic acid (ALA) is used as the photosensitizer.

n addition, ALA is administered orally, and photosensitivity isimited to 24 – 48 hours. High rates of both pathologic response

nd disease-free survival have been reported during multiple-

ear follow-up in patients undergoing ALA-PDT.23,24 However, aignificant proportion of patients undergoing ALA-PDT reportrocedural discomfort including chest pain.24,25 ALA-PDT cane associated with profound nausea and vomiting, hypoten-ion, and arrhythmia, and a fatality has been reported.25

Emerging endosopic ablation techniques for Barrett’s esoph-gus include radiofrequency ablation and cryoablation. A mul-icenter study of a circumferential balloon-based radiofre-uency ablation device (HALO360 System; BÂRRX Medical, Inc,unnyvale, CA) in 100 subjects with nondysplastic Barrett’seported high remission rates, no buried glands in more than300 biopsies, and no post-treatment strictures.26 Similar find-

ngs have recently been reported in abstract form for patientsith HGD.27 Preliminary data suggest that cryoablation with

iquid nitrogen spray can achieve a histopathologic response inatients with HGD or intramucosal carcinoma.28

In this respect, the endoscopic management of Barrett’ssophagus serves as a paradigm for the future of endoscopicherapy. Endoscopic therapeutic options include PDT, EMR,adiofrequency ablation, or a combination of these modalities.deally, choice among these techniques should be determinedy data regarding patient outcomes, as opposed to providerreferences or available local expertise. Determining which pa-ients are likely to experience which outcomes (positive andegative) of which therapeutic modality merits further atten-ion in endoscopic research. Prospective, multicenter, con-rolled trials will be required to answer these questions.

In summary, our experience with PDT for Barrett’s esopha-us identifies an association between length of Barrett’s esoph-gus and development of post-PDT stricture. This associationas independent of treatment length. Patients with intramuco-

al or submucosal carcinoma also appear to be at increased riskor post-PDT stricture when compared with patients withGD. The risk of stricture increases with multiple PDT courses.rior EMR did not influence the likelihood of post-PDT stric-ure. No complications were observed in patients undergoingilation of post-PDT stricture.

References

1. Singh P, Taylor RH, Colin-Jones DG. Esophageal motor dysfunc-tion and acid exposure in reflux esophagitis are more severe ifBarrett’s metaplasia is present. Am J Gastroenterol 1994;89:349–356.

2. Falk GW. Barrett’s esophagus. Gastroenterology 2002;122:1569–1591.

3. Shaheen NJ, Crosby NA, Bozymski EM, et al. Is there publicationbias in the reporting of cancer risk in Barrett’s esophagus?Gastroenterology 2000;119:587–589.

4. US Cancer Statistics Working Group. United States cancer sta-tistics: 1999–2002—incidence and mortality web-based report.Atlanta: US Department of Health and Human Services, Centersfor Disease Control and Prevention and National Cancer Institute,2005. Available at: www.cdc.gov/cancer/npcr/uscs.

5. Ferguson MK, Mannheim KS. Resection for Barrett’s mucosa withhigh grade dysplasia: implications for prophylactic photodynamictherapy. J Thorac Cardiovasc Surg 1997;114:824–829.

6. Reed MF, Tolis G Jr, Edil BH, et al. Surgical treatment of esoph-ageal high-grade dysplasia. Ann Thorac Surg 2005;79:1110–1115.

7. Chang LC, Oelschleger BK, Quiroga E, et al. Long-term outcome ofesophagectomy for high-grade dysplasia or cancer found duringsurveillance for Barrett’s esophagus. J Gastrointest Surg 2006;

10:341–346.

1

1

1

1

1

1

1

1

1

1

2

2

2

2

2

2

2

2

2

2

GB6

308 YACHIMSKI ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

8. Overholt BF, Panjehpour M, Halberg DL. Photodynamic therapy forBarrett’s esophagus with dysplasia and/or early stage carci-noma: long-term results. Gastrointest Endosc 2003;58:183–188.

9. Overholt BF, Panjehpour M. Photodynamic therapy for Barrett’sesophagus. Gastrointest Endosc Clin N Am 1997;7:207–220.

0. Overholt BF, Lightdale CJ, Wang KK, et al. Photodynamic therapywith porfimer sodium for ablation of high-grade dysplasia in Bar-rett’s esophagus: international, partially blinded, randomizedphase III trial. Gastrointest Endosc 2005;62:488–498.

1. Overholt BF, Panjehpour M, Haydek JM. Photodynamic therapy forBarrett’s esophagus: follow-up in 100 patients. Gastrointest En-dosc 1999;49:1–7.

2. Panjehpour M, Overholt BF, Haydek JM, et al. Dysplasia and earlycancer in Barrett’s esophagus and effect of oral steroids onstricture formation. Am J Gastroenterol 2000;95:2177–2184.

3. Panjehpour M, Overholt BF, Phan MN, et al. Optimization of lightdosimetry for photodynamic therapy of Barrett’s esophagus: ef-ficacy vs incidence of stricture after treatment. Gastrointest En-dosc 2005;61:13–18.

4. Ban S, Mino M, Nishioka N, et al. Histologic aspects of photody-namic therapy for dysplasia and early adenocarcinoma arising inBarrett’s esophagus. Am J Surg Pathol 2004;28:1466–1473.

5. Van Laethem JL, Peny MO, Salmon I, et al. Intramucosal carci-noma arising under squamous re-epithelialisation of Barrett’soesophagus. Gut 2000;46:574–577.

6. Overholt BF. Photodynamic therapy strictures: who is at risk?Gastrointest Endosc 2007;65:67–69.

7. Overholt BF, Wang KK, Burdick JS, et al. Five-year efficacy andsafety of photodynamic therapy with Photofrin in Barrett’s high-grade dysplasia. Gastrointest Endosc 2007;66:460–468.

8. Larghi A, Lightdale CJ, Memeo L, et al. EUS followed by EMR forstaging of high-grade dysplasia and early cancer in Barrett’sesophagus. Gastrointest Endosc 2005;62:16–23.

9. Prasad GA, Wang KK, Buttar NS, et al. Predictors of strictureformation after photodynamic therapy for high-grade dysplasia inBarrett’s esophagus. Gastrointest Endosc 2007;65:60–66.

0. Katada C, Muto M, Manabe T, et al. Esophageal stenosis afterendoscopic mucosal resection of superficial esophageal lesions.Gastrointest Endosc 2003;57:165–169.

1. Pacifico R, Wang KK, Wongkersong LM, et al. Combined endo-

scopic mucosal resection and photodynamic therapy versus D

esophagectomy for management of early adenocarcinoma inBarrett’s esophagus. Clin Gastroenterol Hepatol 2003;1:252–257.

2. Wolfson HC, Hemminger LL, Raimondo M, et al. Photodynamictherapy and endoscopic mucosal resection for Barrett’s dyspla-sia and early esophageal adenocarcinoma. South Med J 2004;97:827–830.

3. Mackenzie G, Selvasekar C, Jamieson N, et al. Low incidence ofesophageal adenocarcinoma following optimal regimen of ALAPDT for high grade dysplasia in Barrett’s esophagus. Gastroin-test Endosc 2007;65:AB132.

4. Pech O, Gossner L, May A, et al. Long-term results of photody-namic therapy with 5-aminolevulinic acid for superficial Barrett’scancer and high-grade intraepithelial neoplasia. Gastrointest En-dosc 2005;62:24–30.

5. Hage M, Siersema PD, van Dekken H, et al. 5-aminolevulinic acidphotodynamic therapy versus argon plasma coagulation for ab-lation of Barrett’s oesophagus: a randomized trial. Gut 2004;53:785–790.

6. Sharma VK, Wang KK, Overholt BF, et al. Balloon-based circum-ferential, endoscopic radiofrequency ablation of Barrett’s esoph-agus: 1-year follow-up. Gastrointest Endosc 2007;65:185–195.

7. Gondrie JJ, Peters F, Curvers WL, et al. Radiofrequency ablationof Barrett’s esophagus containing high-grade dysplasia. Gastroi-ntest Endosc 2007;65:AB135.

8. Dumot JA, Vargo JJ, Zuccaro G, et al. Preliminary results ofcryotherapy ablation for esophageal high grade dysplasia (HGD)or intra-mucosal cancer (IMC) in high risk non-surgical patients.Gastrointest Endosc 2007;52:AB110.

9. Keeley SB, Pennathur A, Gooding W, et al. Photodynamic therapywith curative intent for Barrett’s esophagus with high-grade dys-plasia and superficial esophageal cancer. Ann Surg Oncol 2007;14:2406–2410.

Address requests for reprints to: Patrick Yachimski, MD, Blake 4astrointestinal Unit, Massachusetts General Hospital, 55 Fruit St,oston, MA 02115. e-mail: pyachimski@partners.org; fax: 617-724-832.A version of these data was presented in abstract form at Digestive

iseases Week, May 19–24, 2007, Washington, DC.

PL

MJ

*M

Bcdtplcafscderucofpeweaipvaba

Tllea1itp

ednvbl

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2008;6:309–313

eptic Ulcerations Are Related to Systemic Rather Than Local Effects ofow-Dose Aspirin

ARTIJN G. H. VAN OIJEN,*,‡ JEANNE P. DIELEMAN,‡ ROBERT J. F. LAHEIJ,*,‡ MIRIAM C. J. M. STURKENBOOM,‡,§

AN B. M. J. JANSEN,* and FREEK W. A. VERHEUGT�

Department of Gastroenterology, and �Department of Cardiology, Radboud University Nijmegen Medical Center, Nijmegen, The Netherlands; and the ‡Department of

edical Informatics, and the §Department of Epidemiology & Biostatistics, Erasmus MC, University Medical Center, Rotterdam, The Netherlands

asdlc

uwtvEm

lculcdb

Ccatt5afPsdTpprcdm

ackground & Aims: Effervescent calcium carbasalate is aalcium-salt of acetylsalicylic acid causing less local gastricamage than acetylsalicylic acid at high doses in healthy con-

rols. The aim of the study was to investigate the incidence ofeptic ulcers in a population-based cohort using bioequiva-

ent low-dose acetylsalicylic acid (80 mg) or effervescent cal-ium carbasalate (100 mg). Methods: Incident acetylsalicyliccid or effervescent calcium carbasalate users were identifiedrom the Integrated Primary Care Information database. Thetudy cohort comprised 19,819 subjects: 11,891 on acetylsali-ylic acid and 7928 on effervescent calcium carbasalate. Inci-ence rates for documented peptic ulcer disease confirmed byndoscopy were calculated and time-dependent adjusted Coxegression analysis was used to compare the risk of pepticlcers for patients using acetylsalicylic acid or effervescentalcium carbasalate. Results: During an average 1.85 yearsf follow-up evaluation, 115 ulcers were found. The riskor developing a peptic ulcer during drug use was: 3.07er 1000 person-years for acetylsalicylic acid and 4.31 forffervescent calcium carbasalate. The risk of peptic ulcersas not statistically significantly higher in patients using

ffervescent calcium carbasalate than in acetylsalicyliccid users (adjusted hazard ratio, 1.39; 95% confidencenterval, 0.92–2.12). Conclusions: The incidence rate ofeptic ulcer disease is similar in patients using low-dose effer-escent calcium carbasalate compared with regular low-dosecetylsalicylic acid. This implicates that peptic ulcers seem toe related to systemic rather than to local effects of low-dosecetylsalicylic acid.

he beneficial effect of platelet aggregation inhibitors forsecondary prevention of cardiovascular diseases is well estab-

ished.1–3 The most frequently used drugs for this purpose areow-dose salicylates. In the Netherlands acetylsalicylic acid andffervescent calcium carbasalate (a calcium salt of salicylic acids)ccount for about 4.5 million prescriptions (2,707,800 and,767,100, respectively) per year.4 Both drugs work through

rreversible inhibition of cyclooxygenase-1, thereby preventinghe conversion of arachidonic acid into thromboxane A2, alatelet agonist.

Salicylates also are known for their local and systemic sideffects in the upper-gastrointestinal tract, even at low dailyosages of 80 mg. The formation of prostaglandins required forormal gastric mucosal integrity is blocked as a result of irre-ersible inhibition of cyclooxygenase-1.5,6 The main differenceetween acetylsalicylic acid and effervescent calcium carbasalate

ies in the local gastrotoxic effects. Acetylsalicylic acid is a weak

cid, whereas effervescent calcium carbasalate is a soluble calcium-alt. Poor disintegration of the acetylsalicylic acid tablet may causerug concentrations to be higher in certain parts of the stomach,

eading to a direct irritating effect on the gastric mucosa, whereasalcium carbasalate should not have this problem.7

The effect of both medications on the occurrence of pepticlcers has been studied in a randomized clinical cross-over trialherein 20 healthy volunteers were administered high equipo-

ent doses of acetylsalicylic acid (650 mg 3 times/day) or effer-escent calcium carbasalate (826.8 mg 3 times/day) for 5 days.8

ffervescent calcium carbasalate caused fewer gastroduodenalucosal erosions than acetylsalicylic acid.Currently, no data are available comparing the effect of

ong-term use of acetylsalicylic acid and effervescent calciumarbasalate on the occurrence of clinically symptomatic pepticlcers in standard low doses for the prevention of cardiovascu-

ar disease in the general population. We therefore assessed andompared the incidence of peptic ulcers among users of low-ose acetylsalicylic acid and users of effervescent calcium car-asalate in a population-based primary care setting.

MethodsSettingAll data were retrieved from the Integrated Primary

are Information project, a general practice research databaseontaining data from electronic patient records of a group ofbout 150 general practitioners in the Netherlands. Details ofhe database have been described elsewhere.9,10 Briefly, the da-abase contains the complete medical records of more than00,000 patients. The electronic records contain coded andnonymous data on patient demographics, reasons for visit (inree text), diagnoses (using the International Classification forrimary Care11 and free text) from general practitioners andpecialists, referrals, laboratory findings, hospitalizations, andrug prescriptions, including indications and dosage regimen.o maximize completeness of the data, general practitionersarticipating in the Integrated Primary Care Informationroject are not allowed to maintain a system of paper-basedecords aside from the electronic medical records. The systemomplies with European Union guidelines on the use of medicalata for medical research and has been proven valid for phar-acoepidemiologic research.12 The Scientific and Ethical Advi-

Abbreviation used in this paper: CI, confidence interval.© 2008 by the AGA Institute

1542-3565/08/$34.00

doi:10.1016/j.cgh.2007.12.018

sa

apiwpCnoctetep

mboepltatwa

fnrppific

simecsttab

c

cutddnopPcpcatfijscdcdAsw

hsab1

so6accmfaUpt

ducwp

pr3vp

a(

310 VAN OIJEN ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

ory Board of the Integrated Primary Care Information projectpproved the study.

Source Population and Exposure CohortsThe study cohort started from January 1, 1995, and was

nalyzed to September 30, 2003. The source population com-rised all subjects with at least 1 year of valid history available

n the database. This meant that patients had to be registeredith the general practitioner for at least 1 year and the generalractitioner had to be participating in the Integrated Primaryare Information project for at least 1 year. This year waseeded to be able to identify incident use of acetylsalicylic acidr effervescent calcium carbasalate and to assess the presence ofomorbidity at baseline. From the source population we iden-ified all new users of either low-dose acetylsalicylic acid orffervescent carbasalate calcium (80 –100 mg/day). These pa-ients were followed up from start of prescription until thearliest of transferal out of the practice, death, end of the studyeriod, or occurrence of a peptic ulcer.

In this exposure cohort, the duration of use of individualedication was calculated by dividing the prescribed quantity

y the prescribed daily dosage. From this we determined peri-ds of exposure, which were classified as acetylsalicylic acid,ffervescent calcium carbasalate, or nonuse. Person-time of ex-osure was accumulated during the follow-up period for calcu-

ation of incidence rates. Subjects were allowed to switch be-ween acetylsalicylic acid and effervescent calcium carbasalatend have treatment interruptions. Hence, subjects could con-ribute person-time to multiple exposure categories. Switchingas considered in the analysis by including previous use ofnother type of salicylic acid as a covariate.

Identification and Ascertainment ofPeptic UlcersThe first occurrence of documented peptic ulcer disease

or each individual was identified through searches on diag-oses and free text indicators for peptic ulcers. The medicalecords of all potential cases were reviewed manually by ahysician unaware of the research question and classified aseptic ulcers, if confirmed by upper-gastrointestinal endoscopy

n secondary care. In addition, the date of referral for endoscopyor first occurrence of (symptoms for) a peptic ulcer was verifiedn the patient record. Cases were attributed to the exposureategory in which they occurred.

CovariatesAs covariates in this cohort analysis, we considered age,

ex, patients’ history of peptic ulcer disease, other comorbidityncluding the indication for use and the use of interfering

edication. As important concomitant medications we consid-red nonsteroidal anti-inflammatory drugs, oral steroids, cal-ium antagonists, anticoagulates other than acetylsalicylic acid,elective serotonin reuptake inhibitors, and diuretics because ofheir gastrointestinal side effects. The use of gastric acid secre-ion–inhibiting medication, such as proton pump inhibitorsnd H2-receptor antagonists, was considered as a covariateecause of their beneficial and preventive effect on peptic ulcers.

Data AnalysisStandard descriptive statistics were used to describe the

haracteristics of the acetylsalicylic acid and effervescent cal- t

ium carbasalate users. Pearson’s chi-squared statistics weresed to compare distributions of categoric variables betweenhe cohorts and the Student t test was used to compare the ageistributions. Crude incidence rates of documented peptic ulcerisease confirmed by endoscopy were calculated by dividing theumber of peptic ulcer cases by the corresponding person-yearsf the exposure category in which they occurred. Ninety-fiveercent confidence intervals (CIs) were calculated based on aoisson distribution. Cox regression analysis was used to cal-ulate hazard ratios and 95% CIs for the risk of documentedeptic ulcer confirmed by endoscopy during use of effervescentalcium carbasalate compared with acetylsalicylic acid (used asreference � 1). Adjusted Cox regression then was performed

o calculate the risk (hazard ratios) of documented and con-rmed peptic ulcer occurrence in our study population, ad-

usted for important (univariate P � .10) covariates. Becausewitching between exposure categories might occur during theourse of follow-up evaluation we analyzed exposure as a time-ependent factor. Additional stratified adjusted analyses wereonducted to explore the effect of a history of peptic ulcerisease and acid-suppressive medication use as effect modifiers.ll analyses were performed using SAS statistical software (ver-

ion 8.2, SAS Institute Inc, Cary, NC). Statistical significanceas accepted as a 2-sided P value of less than .05.

ResultsThe source population comprised 364,683 patients who

ad on average 2.7 years of follow-up evaluation. Within thistudy population, 11,891 patients used low-dose acetylsalicyliccid and 7928 used low-dose effervescent calcium carbasalate ataseline. The average follow-up period of these patients was.85 years.

Baseline characteristics of the study population are pre-ented in Table 1. The mean age (�SD) of the total populationf platelet aggregation inhibitor users in this population was7.5 years (�13.5 y) and was comparable between acetylsalicyliccid and effervescent calcium carbasalate users (P � .12). Con-omitant use of nonsteroidal anti-inflammatory drugs, calciumhannel blockers, and acid-suppressive medication was com-on in our population (19%, 21%, and 14%, respectively). Ef-

ervescent calcium carbasalate users used significantly morecid-suppressive medication and coumarin derivates (P � .05).sers of effervescent calcium carbasalate also had a higherrevalence of heart failure, rheumatoid arthritis, and osteoar-hritis (Table 1).

The total duration of acetylsalicylic acid use was 3,068,033ays (8399.82 y) and the mean duration of acetylsalicylic acidse per patient was 691.19 days (SD, 582.31). For effervescentalcium carbasalate the total duration of use in our populationas 2,426,427 days (6643.20 y), with a mean duration of use peratient of 653.73 days (SD, 590.21).

During the follow-up period we identified 115 documentedeptic ulcers confirmed by endoscopy. The overall incidenceate of peptic ulcer disease during acetylsalicylic acid use was.07 per 1000 person-years, which was comparable with effer-escent calcium carbasalate users (incidence rate, 4.31 per 1000erson-years; P � .68) (Table 2).

A total of 871 patients (7%) switched from acetylsalicyliccid to effervescent calcium carbasalate and 1124 patients14%) switched from effervescent calcium carbasalate to ace-

ylsalicylic acid (P � .01). These switchers were more likely to

ecwop9rru2c1adf(

mtia0

g2rd2ata

fspaed

pbdgdbs

T

D

C

C

Na

cb

T

AE

T

AE

AFHU

UU

UM

a

ab

ec

o

March 2008 PEPTIC ULCER IN LOW-DOSE ASPIRIN USERS 311

xperience a peptic ulcer (hazard ratio, 9.4; 95% CI, 6.2–14.4)ompared with all nonswitchers. Other factors associatedith the risk of peptic ulcers in our population were agelder than 65 years (hazard ratio, 2.06; 95% CI, 1.28 –3.33),atients’ history of peptic ulcer disease (hazard ratio, 15.4;5% CI, 8.9 –26.6), use of acid-inhibiting medication (hazardatio, 2.39; 95% CI, 1.60 –3.57), use of selective serotonineuptake inhibitors (hazard ratio, 3.78; 95% CI, 2.15– 6.64),se of coumarin derivatives (hazard ratio, 4.26; 95% CI,.38 –7.62), and the use of effervescent calcium carbasalateompared with the use of acetylsalicylic acid (hazard ratio,.70; 95% CI, 1.13–2.57). From Table 3 we can conclude that,fter adjustment for all known risk factors for peptic ulcerisease, effervescent calcium carbasalate shows a similar riskor peptic ulcer disease as compared with acetylsalicylic acidhazard ratio, 1.39; 95% CI, 0.92–2.12).

Stratified analyses for patients with and without a docu-ented history of peptic ulcer disease showed that the rela-

ive risk was higher in the group without peptic ulcer diseasen the past 1.75 (95% CI, 1.13–2.72) than in the group with

history of peptic ulcer disease in the past 0.90 (95% CI,.26 –3.09). The unadjusted relative risk in the group using

able 1. Baseline Characteristics

Acetylsalicylicacid

(n � 11,891)

Effervescent calciumcarbasalate(n � 7928)

emographicsMales, n (%) 6,575 (55) 4,347 (55)Mean age, y (SD) 67.6 (13.1) 67.3 (14.1)�65 y, n (%) 7270 (61) 4864 (61)

oncurrent drug use, n (%)Nonsteroidal anti-

inflammatory drugsa2216 (19) 1561 (20)

Oral steroids 533 (4) 344 (4)Calcium channel blocker 2532 (21) 1600 (20)Acid-suppressive

medicationb1574 (13) 1218 (15)

Selective serotoninreuptake inhibitors

347 (3) 227 (3)

Coumarin derivates 526 (4) 439 (6)omorbidity, n (%)History of validated

peptic ulcer disease104 (1) 90 (1)

Ischemic heart disease 10,565 (89) 7128 (90)Heart failure 572 (5) 516 (7)Diabetes mellitus 1461 (12) 987 (12)Rheumatoid arthritis 281 (2) 329 (4)Osteoarthritis 548 (5) 501 (6)

OTE. Bolded entries: P � .05.Nonselective nonsteroidal anti-inflammatory drugs and selective cy-looxygenase-2 inhibitors.H2-receptor antagonists and proton pump inhibitors.

able 2. Incidence Rates per 1000 Person-Years for Both Ac

Number of patientsPers

ex

cetylsalicylic acid 11,891 2

ffervescent calcium carbasalate 7928 11,594

astric acid inhibitors (proton pump inhibitor or histamine-–receptor antagonists) was not significantly higher (hazardatio, 1.95; 95% CI, 0.90 – 4.22) compared with patients whoid not use this cotherapy (hazard ratio, 1.57; 95% CI, 0.96 –.56). After adjustment for age, sex, history of peptic ulcer,nd concurrent medication use these relative risks changedo 1.40 (95% CI, 0.62–3.18) for gastric acid inhibitor usersnd remained at 1.57 (95% CI, 0.96 –2.57) for the nonusers.

DiscussionThis study showed that patients using low-dose ef-

ervescent calcium carbasalate have a similar risk of endo-copically documented peptic ulcer disease compared withersons using a similar low dose of acetylsalicylic acid, afterdjusting for other known risk factors for peptic ulcer dis-ase such as age, history of peptic ulcer, and concomitantrug use.

A possible explanation for the higher than expected risk ofeptic ulcers in effervescent calcium carbasalate users coulde the result of differences in chemical properties of therugs. As a weak acid, acetylsalicylic acid is less soluble in theastric, low-pH environment and therefore may cause localamage and be less absorbed than effervescent calcium car-asalate. As a consequence equivalent molar oral doses ofalicylic acid that are formulated differently could result in

alicylic Acid and Effervescent Calcium Carbasalate

arsd

No. of events(% of patients)

Incidence rate(per 1000 patient-years) P value

65 (0.55) 3.07 .68

able 3. Unadjusted and Adjusted Hazard Ratios forDocumented Peptic Ulcers Confirmedby Endoscopy

Hazard ratios for documented pepticulcers (95% CI)

Unadjusted Adjusteda

cetylsalicylic acidb Reference Referenceffervescent calciumcarbasalate

1.70 (1.13–2.57) 1.39 (0.92–2.12)

ge �65 y 2.06 (1.28–3.33) 1.97 (1.18–3.30)emale sex 0.70 (0.47–1.05) 0.74 (0.48–1.15)istory of peptic ulcers 15.4 (8.90–26.6) 10.1 (5.23–19.6)se of nonsteroidal anti-inflammatory drugs

1.06 (0.68–1.64) 1.27 (0.79–2.02)

se of acid suppressors 2.39 (1.60–3.57) 1.33 (0.82–2.14)se of selective serotoninreuptake inhibitors

3.78 (2.15–6.64) 3.76 (1.97–7.16)

se of coumarin derivatives 4.26 (2.38–7.62) 3.44 (1.85–6.39)edication switchc 9.43 (6.17–14.4) 6.82 (4.34–10.7)

Adjustment for age, sex, history of peptic ulcer, medication switch,nd concurrent medication use.Acetylsalicylic acid is used only as a reference for comparison withffervescent calcium carbasalate.Switch from acetylsalicylic acid to effervescent calcium carbasalater vice versa.

etyls

on-yepose

1,167

50 (0.63) 4.31

ds

atcdcoceded

csrompcssfwmdsaeatpmapetra

cuWpmptoboef

dueppo

lta

stlHtmocictgat

nviaci

1

1

1

1

312 VAN OIJEN ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

ifferent plasma levels of salicylic acid and therefore result inystemic toxic effects.

Previous studies on differently formulated acetylsalicyliccids have been conducted. Several studies have been under-aken to study the toxicity of high-dose buffered and enteric-oated aspirin.13,14 Enteric-coated aspirin significantly re-uces (local) gastric toxicity, but buffered aspirins wereomparable with plain aspirin.13–15 To our knowledge, thenly study that compared both effervescent carbasalate cal-ium and plain acetylsalicylic acid was the study by Murrayt al8 in 20 healthy volunteers. They concluded that the highoses of the drugs are equivalent therapeutically, but thatffervescent calcium carbasalate caused less gastric mucosalamage than acetylsalicylic acid.

Being an observational study in an existing database ofomplete patient records, misclassification and confoundinghould be taken into consideration while interpreting theesults. Confounding by (contra-)indication could be a pointf concern. Because of its gastroprotective image, physiciansay prescribe effervescent calcium carbasalate preferably to

atients with known upper-gastrointestinal symptoms (ie,hanneling). In our population about 10% of the patientswitched between both forms of medication studied. Patientswitched equally often from plain acetylsalicylic acid to ef-ervescent calcium carbasalate and vice versa. However, thereas no evidence of strong channeling because concurrentedications used for or known to cause gastrointestinal

isease were not different between both groups. If any, themall differences in the prevalence of concomitant diseasend drugs cannot explain the similarity of peptic ulcer dis-ase occurrence during effervescent calcium carbasalate orcetylsalicylic acid use. However, patients who switched be-ween both formulations were at higher risk for developingeptic ulcer compared with patients who did not switch. Thisay be contributed to the underlying reason for switching. Ifpatient will be admitted to a hospital, or presents to the

rescribing physician with intolerance (eg, dyspepsia) forither effervescent calcium carbasalate or acetylsalicylic acid,he treating facility may choose to switch the patient’s aspi-in formulation. Unfortunately, in our database no data werevailable regarding the reason for switching.

Misclassification of the outcome may have occurred be-ause we restricted the analysis to endoscopically confirmedlcers only to avoid inclusion of false-positive peptic ulcers.e thus may have underestimated the incidence rate of

eptic ulcer disease. This is only a point of concern if theisclassification is differential, which may be the case if

hysicians are more likely to request an endoscopy for pa-ients using one salicylic acid rather than the other. Becausef the gastroprotective image of effervescent calcium car-asalate we believe that, if any, differential misclassificationf the outcome would only reduce the risk estimate becausendoscopies probably are less likely to be performed in ef-ervescent calcium carbasalate users.

The primary outcome of this study was the occurrence ofocumented peptic ulcer disease, confirmed by endoscopy. Wesed free text to find the combination of peptic ulcer andndoscopy, but unfortunately the exact anatomic location orathologic reports were not documented in the studied generalractitioner database. For additional insights into the etiology

f gastroduodenal damage by different low-dose aspirin formu-

ations, an endoscopy study should be performed comparinghese formulations, including the studied acetylsalicylic acidnd effervescent calcium carbasalate.

The combined effects of aspirin and clopidogrel have beentudied widely in large cardiovascular randomized clinicalrials. In our adjusted analysis, we adjusted for anticoagu-ants (eg, coumarin derivates), as covariates in our analysis.

owever, during the period of inclusion in the study cohort,he Dutch Cardiology guidelines and medicine reimburse-

ent schemes did not advise or reimburse the long-term usef clopidogrel as cotherapy of acetylsalicylic acid or efferves-ent calcium carbasalate. Only recently, patients will be re-mbursed for a half-year of double therapy with aspirin andlopidogrel after an acute coronary event or elective percu-aneous coronary intervention or coronary artery bypassraft surgery intervention. For this reason we did not havedequate data about clopidogrel use and could not includehat in our analysis.

In conclusion, the incidence rate of endoscopically diag-osed peptic ulcers is similar in patients using low-dose effer-escent calcium carbasalate compared with low-dose acetylsal-cylic acid. This suggests that systemic effects of low-dosespirin may be more important than local effects, which impli-ates that any formulation of aspirin should be used cautiouslyn high-risk patients.

References

1. Pearson TA, Blair SN, Daniels SR, et al. AHA guidelines forprimary prevention of cardiovascular disease and stroke: 2002update. Circulation 2002;106:388–391.

2. Verheugt FWA. Aspirin, the poor man’s statin. Lancet 1998;351:227–228.

3. Verheugt FWA. In search of a super-aspirin for the heart. Lancet1997;349:1409–1410.

4. Facts and figures 2005. The Netherlands: Foundation for Phar-maceutical Statistics, 2006.

5. Garcia Rodriguez LA, Hernandez-Diaz S. Risk of uncomplicatedpeptic ulcer among users of aspirin and nonaspirin nonsteroi-dal antiinflammatory drugs. Am J Epidemiol 2004;159:23–31.

6. Derry S, Loke YK. Risk of gastrointestinal haemorrhage withlong term use of aspirin: meta-analysis. BMJ 2000;321:1183–1187.

7. Cryer B, Feldman M. Effects of very low dose daily, long-termaspirin therapy on gastric, duodenal, and rectal prostaglandinlevels and on mucosal injury in healthy humans. Gastroenterology1999;117:17–25.

8. Murray FE, Hudson N, Atherton JC, et al. Comparison of effectsof effervescent calcium carbasalate and aspirin on gastrodu-odenal mucosal damage in human volunteers. Gut 1996;38:11–14.

9. Vlug AE, van der Lei J, Mosseveld BM, et al. Post-marketingsurveillance based on electronic patient records: the IPCI project.Methods Inf Med 1999;38:339–344.

0. van der Lei J, Duisterhout JS, Westerhof HP, et al. The introduc-tion of computer-based patient records in The Netherlands. AnnIntern Med 1993;119:1036–1041.

1. Lamberts H, Wood M, Hofmans-Okkes IM. International primarycare classifications: the effect of fifteen years of evolution. FamPract 1992;9:330–339.

2. WHO Collaborating Centre for Drug Statistics Methodology. ATCindex with DDDs. Oslo, Norway: WHO Collaborating Centre forDrug Statistics Methodology, 2002.

3. Lanza FL, Royer GL Jr, Nelson RS. Endoscopic evaluation of

the effects of aspirin, buffered aspirin and enteric-coated

1

1

oMe

M

March 2008 PEPTIC ULCER IN LOW-DOSE ASPIRIN USERS 313

aspirin on gastric and duodenal mucosa. N Engl J Med1980;303:136–138.

4. Kelly JP, Kaufman DW, Jurgelon JM, et al. Risk of aspirin-associated major upper-gastrointestinal bleeding with enteric-coated or buffered product. Lancet 1996;348:1413–1416.

5. Garcia Rodriguez LA, Hernandez-Diaz S, de Abajo FL. Association be-tween aspirin and upper gastrointestinal complications: systematic

review of epidemiological studies. Br J Clin Pharm 2001;52:563–571. p

Address requests for reprints to: Martijn van Oijen, MSc, Departmentf Gastroenterology and Hepatology, Radboud University Nijmegenedical Center, PO Box 9101, 6500 HB Nijmegen, The Netherlands.-mail: M.vanOijen@MDL.umcn.nl; fax: (31) 243540103.The authors want to express special thanks to Fabian van der Sluis,D, Erasmus MC, Rotterdam, for his contribution to the validation

rocess.

WB

AW

*G

BdcldthfcwfisaItsTaaavagwsdslfra

T1atUb

tr

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2008;6:314–320

hat Is the Role of Serologic Testing in Celiac Disease? A Prospective,iopsy-Confirmed Study With Economic Analysis

NDREW D. HOPPER,* MARIOS HADJIVASSILIOU,‡ DAVID P. HURLSTONE,* ALAN J. LOBO,* MARK E. McALINDON,*ILLIAM EGNER,§ GRAEME WILD,§ and DAVID S. SANDERS*

Department of Gastroenterology, ‡Department of Neurology, Royal Hallamshire Hospital, Sheffield, United Kingdom; and the §Department of Immunology Northern

eneral Hospital Sheffield, United Kingdom

tettssR(crAotHttbbb

c6orcictbrfuac

statll

it

See CME exam on page 263.

ackground & Aims: The optimal serologic tests for theetection of celiac disease and follow-up assessment remainsontroversial. Our aim was to evaluate all current immuno-ogic assays for diagnosing celiac disease using the gold stan-ard of duodenal biopsy. We also assessed whether tissue

ransglutaminase (tTG) antibody is a quantitative marker foristologic severity. Methods: Consecutive adult patients re-

erred for gastroscopy without a previous known diagnosis ofeliac disease were recruited (group 1). Concurrently, patientsith a known diagnosis of celiac disease on a gluten-free diet

or more than 1 year undergoing repeat duodenal biopsy weredentified (group 2). All patients had duodenal biopsies anderologic analysis performed for immunoglobulin(Ig) A andntibodies to human immunoglobulin (Ig)A-tTG, IgA-gliadin,gG-gliadin, and IgA-endomysial antibody. Results: Twohousand patients were recruited in the first group. Seventy-even (3.9%) patients were diagnosed with new celiac disease.he sensitivity, specificity, positive predictive value, and neg-tive predictive value for IgA tTG were 90.9%, 90.9%, 28.6%,nd 99.6%. When adopting a 2-step approach using tTG firstnd then EMA the sensitivity, specificity, positive predictivealue, and negative predictive value was 85.7%, 98.6%, 71.7%,nd 99.7%, respectively. The use of nondeamidated IgA/IgGliadin antibodies conferred no additional diagnostic benefithen considering the detection of adult celiac disease. In the

econd group 48 patients with celiac disease on a gluten-freeiet were identified. Sixteen of 48 of these patients had per-isting villous atrophy, but 7 of 16 (44%) had a normal tTGevel. Conclusions: IgA tTG alone is a sensitive markeror celiac disease. A normal tTG level does not predictecovery of villous atrophy in patients with celiac disease ongluten-free diet.

he prevalence of celiac disease in the United States, Eu-rope, and the United Kingdom is between 0.75% and

%.1– 4 In addition, we now accept that there may be manyssociated conditions and symptoms that warrant serologicesting for celiac disease (adopting a case-finding approach).ltimately, the diagnosis of celiac disease still requires a small-owel biopsy showing villous atrophy.5– 8

There are a number of serologic tests that have been reportedo be accurate in identifying patients who then should be

eferred for a duodenal biopsy. However, the optimal serologic

est or test strategy remains controversial. Previously, anti–ndomysial antibody (EMA) had been reported as an accurateest with a sensitivity greater than 90% and a specificity greaterhan 98%.9 –11 However, it is recognized that EMA requires aubjective immunofluorescence method and has limited sub-trate resources (either monkey esophagus or umbilical cord).12

ecently, the introduction of anti–tissue transglutaminasetTG) antibody testing using either guinea pig or human re-ombinant tTG has led investigators to suggest that humanecombinant tTG may have higher sensitivities than EMA.12,13

n advantage of automated tTG testing is higher throughputf samples and also the opportunity to obtain a quantitativeiter using the enzyme-linked immunosorbent assay method.owever, there are more false-positive results associated with

TG testing.14,15 This lower specificity has led some investigatorso describe a 2-step method (tTG first and, if positive, followedy EMA—if there is a positive EMA result then proceed toiopsy) to avoid patients undergoing unnecessary duodenaliopsies.12,13

Despite the high accuracy of EMA and tTG, seronegativeeliac disease still occurs. This has been reported to account for.4% (8 of 126) of all cases of celiac disease,16 and appears toccur more often with lesser degrees of atrophy.17,18 For thiseason duodenal biopsy in patients with a high suspicion ofeliac disease still is recommended even if the serologic testings negative.19 Currently published multicenter trials with largeohorts that have tried to compare EMA and tTG accuracy haveaken 1 of 2 approaches. Either they have performed a duodenaliopsy in patients who were antibody positive or they haveecruited patients known to have a high risk of celiac disease,or example, iron-deficiency anemia.16,20,21 However, we werenable to identify any study that evaluated all of the currentntibody tests in a large series of low-risk adult patients withoncomitant duodenal biopsy performed in all patients.

Once a diagnosis of celiac disease has been made, there is noingle method that allows for assessment of compliance. Ques-ioning patients for symptom response to a gluten-free diet anddietary assessment can be used as a noninvasive marker, but

his may not correlate with intestinal damage.22 However, vil-ous recovery on a gluten-free diet may take up to 24 months oronger in adults.23,24 There may be a role for serology in the

Abbreviations used in this paper: EMA, endomysial antibody; IEL,ntraepithelial lymphocyte; IG, immunoglobulin; NPV, negative predic-ive value; PPV, positive predictive value; tTG, tissue transglutaminase.

© 2008 by the AGA Institute1542-3565/08/$34.00

doi:10.1016/j.cgh.2007.12.008

ab

sne

d(tpAuao4npgcdosf

ohaaadvpaaDh

wgdh

ddpSp

f

soalmMv3

(Ie(lpph

rfcuat

[gdappwradugf

es

pdidard5vi

vv

March 2008 EVALUATION OF CURRENT SEROLOGIC TESTS IN CELIAC DISEASE 315

ssessment of compliance but, to date, EMA has been shown toe a poor predictor of villous recovery.25

For these reasons the aim of our study was to assess whicherologic tests or test strategy is optimal for both the recog-ition and follow-up evaluation of patients with celiac dis-ase.

MethodsPatient RecruitmentPatient recruitment took place in a single endoscopy

epartment at the Royal Hallamshire Hospital in SheffieldUnited Kingdom). Consecutive adult patients referred for gas-roscopy were recruited by a single endoscopist over a 26-montheriod (patient group 1 was recruited from January 2004 topril 2006). The Department of Gastroenterology currentlyses a policy of taking 4 duodenal biopsy specimens routinelys part of the endoscopic examination. Patient consent wasbtained and the gastroscopy examination was performed withbiopsy specimens taken from the second part of the duode-

um. At the same time a blood sample was obtained from eachatient and analyzed for total immunoglobulin A (IgA), IgA-liadin, IgG-gliadin, IgA-tTG, and IgA-EMA. Patients were ex-luded from this cohort if they had a known diagnosis of celiacisease, a coagulopathy (international normalized ratio � 1.3r platelet count of �80), active gastrointestinal bleed or auspected carcinoma observed during the examination, or re-used to participate in the study.

Patients then found to have villous atrophy (on duodenal bi-psy) with supportive serology and symptoms were classified asaving celiac disease. Patients with villous atrophy (confirmed onsecond review of the sample to ensure a well-oriented sample)

nd a negative antibody profile were classified as having seroneg-tive celiac disease after further assessment. To further support theiagnosis of antibody-negative celiac disease alternative causes ofillous atrophy were excluded (such as Giardia and Helicobacterylori infection, selective IgA deficiency) and a human lymphocytentigen profile was checked. To confirm the diagnosis of seroneg-tive celiac disease these patients were required to have the DQ2 orQ8 pattern consistent with celiac disease and a clinical andistologic response to a gluten-free diet.26,27

Correlation of Tissue TransglutaminaseAntibody Titer With Degree of VillousAtrophyTo assess the relationship between tTG and histology

e studied both patients diagnosed with celiac disease fromroup 1 and any patients not in that study but who were newlyiagnosed with celiac disease during the same time frame andad the same antibody assessment (in our center).

A second cohort of patients (group 2) formed a known celiacisease group. These were patients undergoing gastroscopy anduodenal biopsy for assessment of histologic remission. Theseatients had been on a gluten-free diet for greater than 1 year.erology was obtained in an identical manner to that describedreviously.

Serology and HistologyAll duodenal biopsy specimens were fixed in buffered

ormalin and embedded in paraffin wax. Standard, 3-�m–thick s

ections at 3 levels were stained with H&E. If changes suggestivef celiac disease were present in a biopsy specimen it was gradedccording to the modified Marsh criteria using the most severeesion present28: Marsh 0, normal appearance; Marsh 1, normal

orphology with raised intraepithelial lymphocytes (IELs);arsh 2, raised IEL with crypt hyperplasia; Marsh 3a, partial

illous atrophy; Marsh 3b, subtotal villous atrophy; and Marshc, total villous atrophy.29

Total IgA was measured on a Behring BN2 nephelometerSiemens Healthcare, Frankfurt, Germany). IgA gliadin,gG gliadin, and human tTG antibodies were assayed onnzyme-linked immunosorbent assay kits from AeskulisaAESKU. DIAGNOSTICS, Wendelsheim, Germany). Sero-ogic samples with a titer greater than 15 U/mL were taken asositive. IgA EMA was detected by immunofluorescence onrimate esophagus sections from The Binding Site (Birming-am, UK).

Ethical approval was obtained from the South Sheffieldesearch and ethics committee. Statistical analysis was per-ormed using SPSS version 10.0 (SPSS Inc, Chicago, IL). Allomparisons between sensitivities of tTG and EMA were madesing the Fisher exact test, and the mean values of tTG levelsnd age groups were compared with an independent sample test.

ResultsAssessment of the Clinical Performanceof Available Serology TestsIn group 1 there were 2000 patients recruited (1167

58.3%] females; mean age, 55.8 y; range, 16 –94 y). From thisroup a total of 77 patients were diagnosed with new celiacisease, giving a prevalence for celiac disease in all patientsttending for gastroscopy of 3.9% (prevalence previously re-orted by our group19). The histologic grading of villous atro-hy for these 77 patients was as follows: 29 with Marsh 3a, 30ith Marsh 3b, and 18 with Marsh 3c lesions. During the

ecruitment period a total of 2220 patients had a gastroscopy,nd 220 patients were excluded (previous diagnosis of celiacisease, 48; gastrointestinal bleed, 101; probable carcinoma, 12;nable to tolerate gastroscopy, 8; coagulopathy, 12; follow-upastroscopy during study period, ie, already recruited, 36; re-used to be included in the trial, 3).

The indications for performing the gastroscopy for all thexaminations and in patients found to have celiac disease arehown in Figure 1.

In group 1, we compared the prevalence of symptoms inatients found to have celiac disease (n � 77) with those whoid not (n � 1923). Patients with celiac disease had a signif-

cantly higher prevalence of weight loss (15.6% vs 5.3%) andiarrhea (42.9% vs 5.2%). Patients without celiac disease hadsignificantly higher prevalence of dyspepsia (17.3% vs 1%),

eflux (13.8% vs 1%), and dysphagia (7.2% vs 0%). The celiacisease group was significantly younger (mean age, 48.0 vs6.1 y) and contained a higher percentage of females (70.1%s 57.9%) (significance is defined as P � .05 for all compar-sons).

Table 1 shows the sensitivity, specificity, positive predictivealue (PPV), and negative predictive value (NPV) for the indi-idual antibodies and their combinations using tTG and EMA

imultaneously or in a 2-step method. Also shown in Table 1 is

tdswsr

ednann

0

ptwopyfip3

316 HOPPER ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

he number of patients who would require a duodenal biopsyepending on which serologic strategy is used and the corre-ponding number of new celiac disease cases that potentiallyould be missed by taking the different approaches. Table 2

hows a corresponding economic analysis for the different se-ologic and biopsy strategies.

When specifically considering serology-negative celiac dis-ase there were 6 patients in the cohort of 2000 who had celiacisease but were tTG and EMA negative. All 6 were IgA-gliadinegative; 1 of the 6 patients (who presented with anemia) hadn IgA deficiency but, despite this, this patient also had aegative IgG gliadin. Seven celiac disease patients were tTGegative, but 1 of these 7 patients had a positive EMA.

The overall prevalence of IgA deficiency in our series was

.7% (14 of 2000), of which only 1 patient had celiac disease. 3

Correlating Serology With Histologic Severity(Newly Diagnosed Patients With CeliacDisease Not on a Gluten-Free Diet)During the study period we identified a total of 114

atients with newly diagnosed celiac disease. Seventy-seven pa-ients were obtained from group 1 (77 of 2000) and a further 37ho were found to have celiac disease in the department bututside of group 1 recuitment due to being on other endosco-ist’s lists. The mean age of the whole group (n � 114) was 47.3ears (range, 17– 85 y), and 69.2% (79 of 114) were female. Thendings on histology showed that 37.7% (43 of 114) of theseatients had a Marsh 3a lesion, 38.6% (44 of 114) had a Marshb lesion, and 23.7% (27 of 114) had a Marsh 3c lesion.

Patients with partial or subtotal villous atrophy (Marsh 3a or

Figure 1. Referral symptoms for gastroscopy.(A) Patient’s symptoms that prompted referral forgastroscopy. The total was 2114 as a result ofsome patients (n � 2000) having more than 1symptom. (B) Patients’ symptoms that promptedreferral for gastroscopy and were found to haveceliac disease. The total was 96 as a result ofsome patients (n � 77) having more than 1symptom.

b lesion) had a significantly lower average (mean) tTG titer

(tcgIMsU(

p.t1

d(

fi0h1a

tps13

a(tdwta

T

OOI

B

E

IIB

Nrpb

T

OOI

E

N

March 2008 EVALUATION OF CURRENT SEROLOGIC TESTS IN CELIAC DISEASE 317

168.1 U/mL and 165.0 U/mL, respectively) than patients withotal villous atrophy (255 U/mL) (P � .05). From group 1, whenonsidering patients with changes in the duodenal biopsy sug-estive of potential celiac disease, 39 were found to have raisedELs (Marsh 1) and 2 patients had a lesion consistent with a

arsh grade 2. Patients with Marsh 1 or 2 lesions also had aignificantly lower average tTG titer (27.7 U/mL and 23.0/mL, respectively) than patients with villous atrophy (P � .05)

Figure 2).The sensitivity of EMA increased from 79% in patients with

artial villous atrophy to 100% in total villous atrophy (P �01). A similar observation was made for tTG sensitivity (usinghe standard cut-off level of �15), with 86.0% in Marsh 3a to00% in Marsh 3c (P � .05) (Figure 3).

Correlating Serology With Histologic Severity(Repeat Duodenal Biopsy in Patients WithCeliac Disease on a Gluten-Free Diet forMore Than 1 Year)In group 2 there were 48 patients with known celiac

isease (gluten-free diet for �1 y). The mean age was 52.7 yearsrange, 21–78 y), with 68.8% (33 of 48) female patients. The

able 1. Evaluation of Different Serologic Strategies That Co

Serologic tests used torefer for biopsy Sensitivity, (%) Specificity, (%)

nly tTG positive 90.9 (82.4–94.5) 90.9 (89.5–92.1) 28.6nly EMA positive 87.0 (77.7–92.8) 98.0 (97.4–98.6) 64.4

f tTG positive and thenEMA positive (2-step)

85.7 (76.2–91.8) 98.6 (98.0–99.0) 71.7

oth tTG positive andEMA positive

85.7 (76.2–91.8) 98.6 (98.0–99.0) 71.7

ither tTG positive orEMA positive

92.2 (84.0–96.4) 90.3 (88.9–91.6) 27.6

gG gliadin positive 48.1 (37.3–59.0) 95.8 (94.9–99.6) 31.6gA gliadin positive 49.4 (38.5–60.2) 89.6 (88.2–90.1) 16.0oth IgG gliadin andIgA gliadin positive

36.4 (26.5–47.5) 98.8 (98.2–99.2) 54.9

OTE. Individual serology is listed and the practice of using both tTGelied on (as the indication for biopsy) the last 2 columns show howotentially would be undetected. The 95% confidence intervals are shy our group.19

able 2. Proposed Cost of Different Strategies Using tTG and

Serologic test used torefer for biopsy

Resultingnumber of patients

undergoingduodenal biopsy

per 2000

Celiac diseascases identifi

per 2000

nly tTG positive 245 70nly EMA positive 104 67

f tTG positive andthen EMA positive

92 66

ither tTG positive orEMA positive

257 71

OTE. The cost of a duodenal biopsy in our hospital is converted to US doll

ndings on histology showed that 43.8% (21 of 48) had a Marshlesion, 12.5% (6 of 48) had a Marsh 1 lesion, 10.4% (5 of 48)

ad a Marsh 2 lesion, 18.8% (9 of 48) had a Marsh 3a lesion,0.4% (5 of 48) had a Marsh 3b lesion, and 4.2% (2 of 48) hadMarsh 3c lesion.Figure 4 shows the percentage of patients with a positive

TG or EMA result for each grade of villous atrophy. Theercentage of patients with a positive EMA or tTG both wereignificantly higher in patients with villous atrophy (EMA, 7 of6; tTG, 7 of 16) than in those without villous atrophy (EMA,of 32; tTG, 4 of 32) (EMA, P � .02; tTG, P � .05).

DiscussionThis study performed concurrent serologic testing and

duodenal biopsy in all adult patients referred for endoscopyn � 2000). Previous investigators have used either serologicesting as a means of determining which patients should un-ergo a biopsy or have performed a routine duodenal biopsyithout serologic testing in tandem. Our results suggest that

TG is the appropriate first serologic test but thereafter whether2-step approach then is taken (with EMA) is likely to remain

e Used for Testing for Celiac Disease

(%) NPV, (%)

Resultingnumber of patients

undergoingduodenal biopsy

per 2000

Missed cases ofceliac disease

out of 77

.3–34.5) 99.6 (99.2–99.8) 245 7

.9–73.0) 99.4 (99.0–99.7) 104 10

.8–79.9) 99.4 (99.4–99.0) 92 11

.8–79.9) 99.4 (99.4–99.0) 92 11

.5–33.4) 99.7 (99.3–99.8) 257 6

.9–40.5) 97.9 (97.1–98.4) 114 40

.9–21.2) 97.8 (97.0–98.4) 238 39

.4–67.7) 97.4 (96.7–98.1) 51 49

MA together or sequentially is shown. If the serology results only arey biopsies would be performed in our cohort and how many cases

n parentheses. The “Only tTG positive” data were reported previously

A as Listed in Table 1

Celiac diseasecases missed

out of apossible 77

Proposed totalcost for 2000,

$/£

Cost per celiacdisease diagnosis,

$/£

7 (1 in 11) 53,880/28,210 780/40310 (1 in 8) 38,260/20,032 570/29911 (1 in 7) 40,200/21,051 610/319

6 (1 in 13) 81,950/42,906 1150/604

uld B

PPV,

(23(54(61

(61

(22

(23(11(41

and Eman

own i

EM

eed

ars ($110/£58), and a tTG and EMA antibody test are $13/£7 each.

si(aphmmsaaHdcococ

mnabnmtRgpt

othps

Fcarwfft

Fdpt(

Frgpa

318 HOPPER ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

ubject to individual clinical judgment. If a single antibody tests used in isolation tTG will detect most cases of celiac diseasetTG sensitivity, 90.9%), however, this is offset by the expense ofgreater number of endoscopies with duodenal biopsy and theotential discomfort to those patients. This is reflected by tTGaving a PPV of only 28.6%. For this reason a 2-step approachay be cost effective (Table 2), but this will result in cliniciansissing some cases of celiac disease as a result of the lower

ensitivity of EMA. Because of this discordance between EMAnd tTG positive results, one could test for both concurrentlynd opt to biopsy all patients who have any positive antibody.owever, in our data set, 1 in 13 cases of undetected celiacisease still would be missed as a result of serology-negativeeliac disease. On this basis we cannot make a recommendationn which test is the most cost effective or accurate for differentlinical presentations; however, we still would recommend du-denal biopsy in any cases in which a clinician is suspicious ofeliac disease irrespective of a negative antibody result.

What is apparent from our evaluation is the poor perfor-ance of the gliadin antibodies. In this study, the gliadins gave

o extra diagnostic benefit when used specifically to detectdult celiac disease. However, better results for gliadins haveeen shown in children.20 Variation may occur as a result of theumerous antigliadin assay kits. In addition, they may be aarker of extraintestinal manifestations of gluten sensitivity in

he absence of enteropathy (eg, neurologic manifestations).30

ecently, with the emergence of newly developed IgA syntheticliadin– derived deamidated peptide, investigators have shownromising accuracy when used on its own or combined withTG.31,32

igure 2. Graph showing patients with newly diagnosed potential andonfirmed celiac disease. The tTG titer at presentation was groupedccording to the Marsh grade of the duodenal biopsy. The black barsepresent the average (mean) tTG values. The average titer for Marsh 3cas significantly higher than that of 3a or 3b (P � .01), and the average

or Marsh grade 1 and 2 were significantly lower than the average titersor Marsh grades 3a–c. One dot can represent more than 1 patient if the

TG level was identical. tTG, IgA tissue transglutaminase antibody. (

We have reported a sensitivity for tTG of 90.9% and for EMAf 87.0%. These sensitivities are at the lower end of the spec-rum when compared with previously reported studies.10 Per-aps the reasons for this are as follows: first, our study was arospective evaluation and other prospective studies also con-istently have shown lower sensitivities for tTG and EMA.21,26

igure 3. Sensitivity of tTG (cut off �15 U/mL) and EMA antibody forifferent degrees of villous atrophy in newly diagnosed patients withotential and confirmed celiac disease. Sensitivity for both EMA andTG was significantly higher in Marsh grade 3c than in Marsh grade 3aP � .01 for EMA and P � .05 for tTG).

igure 4. Percentage of positive IgA tTG (cut off �15 U/mL) and EMAesults in patients with celiac disease on a gluten-free diet (�1 y)rouped according to Marsh grade. The percentage of patients with aositive EMA or tTG were both significantly higher in patients with villoustrophy (EMA, 7 of 16; tTG, 7 of 16) than in those without villous atrophy

EMA, 3 of 32; tTG, 4 of 32) (EMA, P � .02; tTG, P � .05).

Sttptstp

Astlewpbc

badfawds(chdPs

giaMcmftwa(ttcrabsprcdtr9

t

qArvrpsb

rwbseawds

tpuadtdts

1

March 2008 EVALUATION OF CURRENT SEROLOGIC TESTS IN CELIAC DISEASE 319

econd, our study was performed in a low (or lower)-risk grouphan other biopsy-confirmed studies, which reduced our ascer-ainment bias. Finally, unlike most previous studies all of ouratients underwent a biopsy (irrespective of antibody status),herefore we believe our data truly assesses the prevalence oferonegative celiac disease. It is not surprising that by adoptinghis strategy this has resulted in a reduced sensitivity by com-arison with previously published data.

The PPV of IgA tTG in our study was 28.6% (Table 1).lthough disappointing by comparison with other published

tudies (range, 21.8%– 67%),33–35 there may be several reasons forhis. A high tTG titer has been shown to occur in patients withiver disease, diabetes,15,36 and in up to 40% of patients withnd-stage heart failure.37 This might be reflected in our cohort,hich aimed to be relatively low risk for celiac disease (overallrevalence of celiac disease, 3.9%). However, our cohort mighte considered an unwell group (with other comorbidities) be-ause they were coming to the hospital for gastroscopy.

There have been a few reports of a reduced sensitivity foroth EMA and tTG in patients with lesser degrees of villoustrophy.17,18,38 We found similar results in our cohort of celiacisease patients. The sensitivity of tTG and EMA decreasedrom 100% and 100%, respectively, in patients with total villoustrophy (Marsh 3c) to 85% and 79%, respectively, in patientsith partial villous atrophy (Marsh 3a) (Figure 3). However,espite our observation that the tTG titer in Marsh 3c lesions isignificantly higher than in lesser degrees of villous atrophyFigure 2), conversely others have reported that a high tTG titeran occur in patients without villous atrophy.39 If we used aigher cut-off value than 15 U/mL (manufacturer’s recommen-ation), for example, higher than 100 U/mL, this increased thePV in our series to 75% (with a corresponding sensitivity,pecificity, and NPV of 20%, 97%, and 72.7%, respectively).

We opted for the presence of a Marsh grade 3 lesion to be theold standard for validating the serologic tests. This approachs concurrent with the majority of previous reports,16,20,26,35,38

lthough there have been a few studies that also incorporatedarsh 2 lesions.13,40 The concept of celiac disease or potential

eliac disease without villous atrophy is well described.28 Minorucosal changes (Marsh 1 and 2) may normalize with a gluten-

ree diet and improvement of symptoms. However, in our studyhere were only 2 patients with Marsh 2 lesions and thus thisould have had a negligible effect on our tTG validation. Therere other causes of raised IELs (Marsh 1) in the duodenumapart from celiac disease) for which we did not assess.28 Whenrying to detect adult celiac disease using either tTG or EMA,he inclusion of Marsh 1 lesions (by expanding our diagnosticriteria for celiac disease; Figures 2 and 3) in our data set wouldeduce the sensitivity further. It is unclear which patients withraised IEL require follow-up evaluation but a repeat biopsy oriopsy after a gluten-free challenge and HLA testing has beenuggested.41 It would perhaps be an option to exclude theseatients with a Marsh 1 (n � 39) or Marsh 2 (n � 2) biopsyesult completely from both our celiac and control groups andlassify them as having potential or unknown celiac disease: inoing so the resulting sensitivity, specificity, PPV, and NPV forTG from our series would be 90.9%, 91.3%, 30.0%, 99.6%,espectively, and for EMA it would be 87.0%, 98.2%, 66.3%, and9.5%, respectively.

We consider that tTG offers clear advantages owing to au-

omation (allowing higher throughput of samples) and it is a 1

uantitative test (using enzyme-linked immunosorbent assay).ll patients in our study had a duodenal biopsy; thus, the

educed PPV is perhaps a reflection of real clinical practice. Iniew of this observation we believe that EMA should not beegarded as an obsolete test. We did not assess IgG tTG in ouratients because it was not available in our laboratory at thetart of the study. Initial reports of the new IgG-tTG assays haveeen encouraging.21,42

Our data for the role of tTG in the assessment of histologicemission of celiac patients on a gluten-free diet are concurrentith previous reports.43– 45 We would not recommend that tTGe used in isolation as a marker of histologic severity or remis-ion. Forty-four percent (7 of 16) of celiac patients at follow-upvaluation (on a gluten-free diet for �1 y) who had villoustrophy (Marsh 3a– c) also had a tTG of less than 15 U/mL. Weould recommend that remission should be based on repeatuodenal biopsy, symptom response, dietary questioning, anderologic status as a composite assessment.46

In conclusion, IgA tTG alone is a sensitive marker for de-ecting celiac disease, but because of its poor PPV in clinicalractice it has not superseded the use of IgA-EMA testing. These of nondeamidated IgA/IgG gliadin antibodies confers nodditional diagnostic benefit when specifically considering theetection of adult celiac disease. Finally, a normal tTG is unableo predict recovery of villous atrophy in patients with celiacisease (on a gluten-free diet for �1 y) and we would suggesthat a duodenal biopsy always should be considered if patientstill are symptomatic.

References

1. Fasano A, Berti I, Gerarduzzi T, et al. Prevalence of celiac diseasein at-risk and not-at-risk groups in the United States: a largemulticenter study. Arch Intern Med 2003;163:286–292.

2. Bingley PJ, Williams AJ, Norcross AJ, et al. Undiagnosed coeliacdisease at age seven: population based prospective birth cohortstudy. BMJ 2004;328:322–323.

3. West J, Logan RF, Hill PG, et al. Seroprevalence, correlates, andcharacteristics of undetected coeliac disease in England. Gut2003;52:960–965.

4. Sanders DS, Patel D, Stephenson TJ, et al. A primary care cross-sectional study of undiagnosed adult coeliac disease. Eur J Gas-troenterol Hepatol 2003;15:407–413.

5. Revised criteria for diagnosis of coeliac disease. Report of Work-ing Group of European Society of Paediatric Gastroenterology andNutrition. Arch Dis Child 1990;65:909–911.

6. United European Gastroenterology Working Group. When is acoeliac a coeliac? Report of a working group of the United Euro-pean Gastroenterology Week in Amsterdam, 2001. Eur J Gastro-enterol Hepatol 2001;13:1123–1128.

7. National Institutes of Health. Consensus Development Confer-ence Statement on Celiac Disease, June 28-30, 2004. Gastro-enterology 2005;128(Suppl 1):S1–S9.

8. AGA Institute Medical Position Statement on. the Diagnosis andManagement of Celiac Disease. Gastroenterology 2006;131:1977–1980.

9. Lewis NR, Scott BB. Systematic review: the use of serology toexclude or diagnose coeliac disease (a comparison of the en-domysial and tissue transglutaminase antibody tests). AlimentPharmacol Ther 2006;24:47–54.

0. Hill ID. What are the sensitivity and specificity of serologic testsfor celiac disease? Do sensitivity and specificity vary in differentpopulations? Gastroenterology 2005;128(Suppl 1):S25–S32.

1. Rostom A, Dube C, Cranney A, et al. The diagnostic accuracy of

1

1

1

1

1

1

1

1

2

2

2

2

2

2

2

2

2

2

3

3

3

3

3

3

3

3

3

3

4

4

4

4

4

4

4

Sa

re

320 HOPPER ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

serologic tests for celiac disease: a systematic review.Gastroenterology 2005;128(Suppl 1):S38–S46.

2. Hill PG, McMillan SA. Anti-tissue transglutaminase antibodiesand their role in the investigation of coeliac disease. Ann ClinBiochem 2006;43:105–117.

3. Hill PG, Forsyth JM, Semeraro D, et al. IgA antibodies to humantissue transglutaminase: audit of routine practice confirms highdiagnostic accuracy. Scand J Gastroenterol 2004;39:1078–1082.

4. Volta U, De Franceschi L, Molinaro N, et al. Frequency andsignificance of anti-gliadin and anti-endomysial antibodies in au-toimmune hepatitis. Dig Dis Sci 1998;43:2190–2195.

5. Clemente MG, Musu MP, Frau F, et al. Antitissue transglutami-nase antibodies outside celiac disease. J Pediatr GastroenterolNutr 2002;34:31–34.

6. Collin P, Kaukinen K, Vogelsang H, et al. Antiendomysial andantihuman recombinant tissue transglutaminase antibodies inthe diagnosis of coeliac disease: a biopsy-proven European mul-ticentre study. Eur J Gastroenterol Hepatol 2005;17:85–91.

7. Abrams JA, Diamond B, Rotterdam H, et al. Seronegative celiacdisease: increased prevalence with lesser degrees of villousatrophy. Dig Dis Sci 2004;49:546–550.

8. Tursi A, Brandimarte G, Giorgetti GM. Prevalence of antitissuetransglutaminase antibodies in different degrees of intestinaldamage in celiac disease. J Clin Gastroenterol 2003;36:219–221.

9. Hopper AD, Cross SS, Hurlstone DP, et al. Pre-endoscopy sero-logical testing for coeliac disease: evaluation of a clinical deci-sion tool. BMJ 2007;334:729.

0. Tonutti E, Visentini D, Bizzaro N, et al. The role of antitissuetransglutaminase assay for the diagnosis and monitoring of coe-liac disease: a French-Italian multicentre study. J Clin Pathol2003;56:389–393.

1. Reeves GE, Squance ML, Duggan AE, et al. Diagnostic accuracyof coeliac serological tests: a prospective study. Eur J Gastroen-terol Hepatol 2006;18:493–501.

2. Ciacci C, Cirillo M, Cavallaro R, et al. Long-term follow-up of celiacadults on gluten-free diet: prevalence and correlates of intestinaldamage. Digestion 2002;66:178–185.

3. Tursi A, Brandimarte G, Giorgetti GM, et al. Endoscopic andhistological findings in the duodenum of adults with celiac dis-ease before and after changing to a gluten-free diet: a 2-yearprospective study. Endoscopy 2006;38:702–707.

4. Grefte JM, Bouman JG, Grond J, et al. Slow and incompletehistological and functional recovery in adult gluten sensitive en-teropathy. J Clin Pathol 1988;41:886–891.

5. Dickey W, Hughes DF, McMillan SA. Disappearance of endomy-sial antibodies in treated celiac disease does not indicate histo-logical recovery. Am J Gastroenterol 2000;95:712–714.

6. Feighery C, Weir DG, Whelan A, et al. Diagnosis of gluten-sensi-tive enteropathy: is exclusive reliance on histology appropriate?Eur J Gastroenterol Hepatol 1998;10:919–925.

7. Shidrawi RG, Przemioslo R, Davies DR, et al. Pitfalls in diagnos-ing coeliac disease. J Clin Pathol 1994;47:693–694.

8. Dickson BC, Streutker CJ, Chetty R. Coeliac disease: an updatefor pathologists. J Clin Pathol 2006;59:1008–1016.

9. Oberhuber G, Granditsch G, Vogelsang H. The histopathology ofcoeliac disease: time for a standardized report scheme for pa-thologists. Eur J Gastroenterol Hepatol 1999;11:1185–1194.

0. Hadjivassiliou M, Grunewald RA, Davies-Jones GA. Gluten sensi-tivity as a neurological illness. J Neurol Neurosurg Psychiatry2002;72:560–563.

1. Agardh D. Antibodies against synthetic deamidated gliadin pep- a

tides and tissue transglutaminase for the identification ofchildhood celiac disease. Clin Gastroenterol Hepatol2007;5:1276–1281.

2. Sugai E, Vazquez H, Nachman F, et al. Accuracy of testing forantibodies to synthetic gliadin-related peptides in celiac disease.Clin Gastroenterol Hepatol 2006;4:1112–1117.

3. Bardella MT, Trovato C, Cesana BM, et al. Serological markersfor coeliac disease: is it time to change? Dig Liver Dis 2001;33:426–431.

4. Chan AW, Butzner JD, McKenna R, et al. Tissue transglutaminaseenzyme-linked immunosorbent assay as a screening test forceliac disease in pediatric patients. Pediatrics 2001;107:E8.

5. Lock RJ, Stevens S, Pitcher MC, et al. Is immunoglobulin Aanti-tissue transglutaminase antibody a reliable serologicalmarker of coeliac disease? Eur J Gastroenterol Hepatol 2004;16:467–470.

6. Vecchi M, Folli C, Donato MF, et al. High rate of positive anti-tissue transglutaminase antibodies in chronic liver disease. Roleof liver decompensation and of the antigen source. Scand JGastroenterol 2003;38:50–54.

7. Peracchi M, Trovato C, Longhi M, et al. Tissue transglutaminaseantibodies in patients with end-stage heart failure. Am J Gastro-enterol 2002;97:2850–2854.

8. Rostami K, Kerckhaert JP, Tiemessen R, et al. The relationshipbetween anti-endomysium antibodies and villous atrophy in coe-liac disease using both monkey and human substrate. Eur JGastroenterol Hepatol 1999;11:439–442.

9. Freeman HJ. Strongly positive tissue transglutaminase anti-body assays without celiac disease. Can J Gastroenterol2004;18:25–28.

0. Bazzigaluppi E, Roggero P, Parma B, et al. Antibodies to recom-binant human tissue-transglutaminase in coeliac disease: diag-nostic effectiveness and decline pattern after gluten-free diet.Dig Liver Dis 2006;38:98–102.

1. Mahadeva S, Wyatt JI, Howdle PD. Is a raised intraepitheliallymphocyte count with normal duodenal villous architecture clin-ically relevant? J Clin Pathol 2002;55:424–428.

2. Korponay-Szabo IR, Dahlbom I, Laurila K, et al. Elevation of IgGantibodies against tissue transglutaminase as a diagnostic toolfor coeliac disease in selective IgA deficiency. Gut 2003;52:1567–1571.

3. Gillett HR, Freeman HJ. Comparison of IgA endomysium antibodyand IgA tissue transglutaminase antibody in celiac disease. CanJ Gastroenterol 2000;14:668–671.

4. Vahedi K, Mascart F, Mary JY, et al. Reliability of antitransglu-taminase antibodies as predictors of gluten-free diet compliancein adult celiac disease. Am J Gastroenterol 2003;98:1079–1087.

5. Kaukinen K, Sulkanen S, Maki M, et al. IgA-class transglutami-nase antibodies in evaluating the efficacy of gluten-free diet incoeliac disease. Eur J Gastroenterol Hepatol 2002;14:311–315.

6. Kaukinen K, Peraaho M, Lindfors K, et al. Persistent small bowelmucosal villous atrophy without symptoms in coeliac disease.Aliment Pharmacol Ther 2007;25:1237–1245.

Address requests for reprints to: Dr Andrew D. Hopper, 15 Nairntreet, Sheffield, S10 1UL, United Kingdom. e-mail: andydhopper@ol.com; fax: (44) 114-2712692.Some data were described previously as part of an alternative study

eported in the British Medical Journal 2007;334:729–733. Any ref-rence to this subset of results has been denoted throughout the

rticle.

SH

P

D

Bmsnvbaopvbhutccs(Hnswwp(cbpd

ISstapm

dtdabsrta

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2008;6:321–328

ymptom Severity but Not Psychopathology Predicts Visceralypersensitivity in Irritable Bowel Syndrome

ATRICK P. J. VAN DER VEEK,* YANDA R. VAN ROOD,‡ and AD A. M. MASCLEE*

epartments of *Gastroenterology and Hepatology and ‡Psychiatry, Leiden University Medical Center, Leiden, The Netherlands

tpwbSuihsr

cl(a

tradm

p(tguwbcotats

tsMAd

ackground & Aims: Visceral hypersensitivity is a hall-ark of irritable bowel syndrome (IBS), but the relation-

hip with clinical symptoms and psychological factors hasot been fully established. We aimed to (1) evaluate theseariables in a large cohort of IBS patients, recruited fromoth hospital and general practice, and in healthy controlsnd (2) assess which of these factors predicts the occurrencef visceral hypersensitivity in IBS. Methods: Rectal com-liance and perception (intensity, perception thresholds;isual analogue scale, 0 –100 mm) were assessed by a rectalarostat study (ramp distention) in 101 IBS patients and 40ealthy volunteers. IBS symptom severity was scored bysing a 14-day 5-item diary. Anxiety, depression, somatiza-

ion, vigilance, pain coping, dysfunctional cognitions, psy-honeuroticism, and quality of life were assessed with psy-hometric questionnaires. Results: Rectal compliance wasignificantly reduced in IBS patients compared with controlsP < .01), as were thresholds for pain (27 � 15 vs 35 � 8 mm

g; P < .01) and urge (P < .05). Levels of anxiety, depression,euroticism, somatization, and dysfunctional cognitions wereignificantly increased in IBS patients versus controls,hereas pain coping and quality of life were significantlyorse. Hypersensitivity to rectal distention occurred in 33% ofatients and was associated with increased symptom severityP � .016), but not with demographic characteristics or psy-hological disturbances. Conclusions: Hypersensitivity toalloon distention occurs in 33% of IBS patients and isredicted by symptom severity but not by psychological oremographic characteristics.

rritable bowel syndrome (IBS) is characterized by recurrentabdominal discomfort or pain and disturbed bowel habits.1

everal pathophysiologic mechanisms have been suggested inymptom generation, including altered intestinal motility,2 au-onomic dysfunction,3,4 inflammation,5,6 and immune systemlterations.6 – 8 Particularly, visceral hypersensitivity appears tolay an important role9,10 and has been proposed as a biologicarker of IBS.11

Visceral hypersensitivity might result from disturbances atifferent levels of the brain-gut axis, in which peripheral sensi-ization of intestinal nerve endings,12 hyperexcitability of spinalorsal horn neurons,13 and altered central processing of visceralfferent information14 are implicated. Abnormalities in regionalrain activation, especially in areas involved in pain processinguch as the anterior cingulate cortex and thalamus, have beeneported in IBS patients in response to rectal balloon disten-ion.15 These regions belong to the emotional limbic system and

re involved in psychological and cognitive events.16,17

IBS symptomatology is associated with psychological fac-ors, and these might affect clinical outcome.18 For instance,sychological distress is more prevalent among IBS patientsho seek health care.19 Little is known about the relationshipetween psychological variables and visceral hypersensitivity.uch information is relevant because it might provide a betternderstanding of the pathogenesis of IBS and thereby improve

ts treatment. The few studies that explored this relationshipave been criticized because of methodologic shortcomingsuch as sample size and patient selection (tertiary refer-als).9,11,19

The aims of the present study were to (1) explore in a largeohort of IBS patients the prevalence of rectal hypersensitivity,evels of psychological distress, and IBS symptom severity and2) assess which demographic, clinical, and psychological vari-bles predict the occurrence of visceral hypersensitivity in IBS.

MethodsParticipantsThis study was part of a large randomized controlled

rial of psychological treatment in IBS, the results of which wereecently published.20 IBS patients between 18 and 65 years ofge were invited to participate. Baseline evaluation includedetailed psychological assessment, rectal barostat measure-ents, and IBS symptom severity scores.To obtain a representative sample from the IBS population,

atients were recruited from both the hospital IBS populationpatients referred to the outpatient Department of Gastroen-erology of the Leiden University Medical Center) and theeneral population through local advertisement. Healthy vol-nteers were recruited through advertisement for comparisonith the patient sample. All eligible participants were screenedy one of the investigators (P.vd.V). Each patient met Rome IIriteria for IBS.1 Exclusion criteria were organic disease, previ-us abdominal surgery (except cholecystectomy and appendec-omy), and pregnancy. Use of antispasmodics, laxatives, bulkinggents, and occasional use of analgesics was permitted. We usedhe Mini International Neuropsychiatric Interview (Dutch ver-ion 5.0.0)21 to exclude patients with severe psychopathology

Abbreviations used in this paper: CSFBD, Cognitive Scale for Func-ional Bowel Disorders; IBS, irritable bowel syndrome; MMPI, Minne-ota Multiphasic Personality Inventory; NVM, Netherlands version ofMPI; PCCL, Pain Coping and Cognition List; SAS, Somatosensorymplification Scale; SCL-90, Symptom Checklist 90; SD, standardeviation; SF-36, Short Form 36; VAS, visual analogue scale.

© 2008 by the AGA Institute1542-3565/08/$34.00

doi:10.1016/j.cgh.2007.12.005

(fe

ScaprpaiW

r(

a

ibstcs

Ip

Cacl

w

((t

hfc

mtp

S

mt

sifwoc

rcbh

tvilp

taqn8

watcwaotwlbdmm

tcaaeepate

upi

322 VAN DER VEEK ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

psychosis or risk of suicide). Informed consent was obtainedrom each participant. The Leiden University Medical Centerthics committee had approved the study protocol.

BarostatAn electronic barostat (Synectics Visceral Stimulator;

ynectics Medical, Stockholm, Sweden) was used to assess rectalompliance and perception. This device measures rectal motorctivity as volume changes in a rectal balloon, in which constantressure is maintained by injecting air when the rectal wallelaxes and aspirating air during rectal contraction. Intrabagressure is directly measured via a separate lumen. Maximalirflow is 38 mL/s. Pressure and volume are continuously mon-tored and recorded on a personal computer (Polygram for

indows SVS module; Synectics Medical).

VisceroperceptionPerception of urge to defecate and abdominal pain during

ectal distention was quantified on a 100-mm visual analogue scaleVAS). End points ranged from none to intolerable.

Demographic CharacteristicsThe demographic group characteristics of interest were

ge, sex, and level of health care (general practice or referral).

Symptom SeverityPatients and controls rated the severity of any abdom-

nal discomfort, abdominal pain, constipation, diarrhea, andloating daily for 14 days on a 5-point Likert scale (0, noymptoms; 1, mild; 2, moderate; 3, severe; 4, very severe symp-oms) by using a symptom diary card. A composite score wasomputed by summing up the 14-day mean scores for eachymptom (range, 0 –20).

Psychological AssessmentA battery of questionnaires was administered to both

BS patients and control subjects to determine the followingsychological characteristics of each group.

Anxiety and depression. We used the Symptomhecklist 90 (SCL-90) to measure levels of anxiety (10 items)nd depression (16 items). The SCL-90 is a validated survey andonsists of 90 items addressing a range of physical and psycho-ogical problems.22

Psychoneuroticism. The level of psychoneuroticismas determined by summing up all 90 items of the SCL-90.

Somatization. We used the abridged Dutch versionNVM) of the Minnesota Multiphasic Personality InventoryMMPI) to measure somatization, which is 1 of 5 subscales onhis questionnaire.23

The role of the abovementioned psychological factors in IBSas been studied previously.9,10,19 In addition, we considered the

ollowing psychological variables relevant, because they mightonfound the abovementioned determinants.

Vigilance. We used the previously validated 10-item So-atosensory Amplification Scale (SAS)24 to determine the extent

o which an individual is likely to report enhanced perception ofhysical symptoms (ie, lower cognitive perception thresholds).

Cognitions. The recently developed 31-item Cognitive

cale for Functional Bowel Disorders (CSFBD) was used to f

easure patients’ levels of dysfunctional cognitions concerningheir IBS.25

Pain coping. Pain coping was measured by 1 of 4ubscales of the Pain Coping and Cognition List (PCCL). Thisnventory has been widely used in The Netherlands and awaitsuture validation. Patients were asked to rate the extent tohich they agreed with 11 statements concerning pain copingn a 7-point scale, ranging from “I completely disagree” to “Iompletely agree.”

Somatic symptoms. The SCL-90 was also used toecord non–IBS-related somatic symptoms. There are 12 itemsoncerning general complaints, including headache, vertigo,ackache, myalgia, difficulties with breathing, intolerance forigh or low temperatures, dysphagia, etc.

Quality of life. Quality of life was assessed by usinghe validated Short Form 36 (SF-36) questionnaire.26 This sur-ey measures quality of life in 8 domains, ie, physical function-ng, social functioning, role limitations due to physical prob-ems and emotional problems, mental health, vitality, bodilyain, and general health.

Experimental DesignA small standardized, low caloric breakfast was permit-

ed at 8:00 AM on the day of the barostat recordings. Afterrrival at our department at 10:00 AM, subjects filled out alluestionnaires consecutively. Each participant was allowed theecessary time to complete the questionnaires, which took0 –90 minutes on average.

After completion, the rectum was evacuated by using a tapater enema. Participants were then placed in a hospital bed,nd with the subject in the left lateral position, a lubricated,ightly folded, highly compliant polyethylene bag (maximumapacity, 1000 mL) tied to the end of a multilumen tube (19F)as inserted through the anus and positioned in the rectal

mpulla. Bag position was checked by manual inflation of 150 mLf air and subsequent retraction of the catheter until prevented byhe external anal sphincter. After balloon deflation, the catheteras introduced an additional 2 cm, secured to the subject’s upper

eg by a piece of tape, and connected to the barostat. The hospitaled was placed in a 15-degree recumbent supine position (Tren-elenburg) to avoid interference of abdominal mass with barostateasurements. Barostat measurements commenced approxi-ately 4 hours after the light breakfast.The experimental protocol consisted of a slow ramp disten-

ion to assess rectal compliance. Intrabag pressure was in-reased at a rate of 1 mm Hg/min, starting at 5 mm Hg, untilmaximum of 30 mm Hg. Patients rated the urge to defecate

nd level of abdominal pain on the 100-mm VAS scale at allven pressures (6, 8, . . . 30 mm Hg). After the experiment hadnded, the rectal balloon was deflated and removed, and eacharticipant was provided with a 14-day symptom diary card andstamped envelope to return the diary. Subjects were instructed

o start filling out their symptom diary on the day after thexperiment.

Barostat AnalysisDynamic compliance was assessed by calculating vol-

me increments for each individual pressure step in each studyarticipant. Compliance was defined by the largest volume

ncrement (ie, the steepest slope of the pressure-volume curve)

or each participant and averaged over groups. Perception

sPw

dc

aw8

dcMrtblsstm

octaaubc

y0(tlrsittS

R

T

D

S

P

Q

Na

b

March 2008 PREDICTORS OF VISCERAL HYPERSENSITIVITY IN IBS 323

cores were expressed as the mean score at each pressure step.erception thresholds were defined as the first pressure level athich perception scores exceeded 10 mm.

Visceral HypersensitivityPatients with a pain perception threshold �2 standard

eviations (SDs) below the mean threshold in controls wereonsidered to be hypersensitive to balloon distention.

Statistical AnalysisWe aimed to enroll at least 40 subjects in each group to be

ble to detect a 5-mm Hg difference in mean pain threshold, whiche considered clinically relevant, with a power of 0.80 and SD ofmm Hg on the basis of previous studies by our group.All statistical analyses were carried out with SPSS for Win-

ows, version 11.0.1 (SPSS Inc, Chicago, IL). Demographicharacteristics were compared between groups by Student t test,

ann-Whitney, or �2 analysis as appropriate. Differences inectal compliance and visceroperception were analyzed for sta-istical significance by using mixed models, with patient num-ers as indicator for repeated measurements. One model ana-

yzed pressure, volume, and pressure by volume interaction aseparate contributors to the model; a second model did theame for pressure, visceral perception, and pressure by percep-ion interaction. Compliance, perception of urge and pain at

aximum rectal pressure (30 mm Hg), and perception thresh-

able 1. Baseline Demographic, Clinical, and Psychological C

Characteristic IBS

emographicsAge (y)Female sex (%)Bowel habit (%)

DiarrheaConstipationAlternatingNot specifiedNormal (controls)

ymptomsIBS symptom score (0–20)

sychological profileAnxiety (10–50)Depression (16–80)Somatic symptoms (12–60)Psychoneuroticism (90–450)Dysfunctional cognitions (31–217)Vigilance (0–40)Pain coping (6–1)Somatization (0–2)

uality of life (0–100)Physical functioningRole limitations–physicalBodily painMental healthRole limitations–emotionalSocial functioningVitalityGeneral health

OTE. Score ranges from best to worst are indicated after each paraP � .01 vs healthy controls.

P � .05 vs healthy controls.

lds for urge were compared by Mann–Whitney (patients versusontrols) or Kruskal–Wallis analysis (IBS subgroups). Becausehe pain threshold during ramp distention was not reached inll participants, the best estimates for the mean pain thresholdnd SD were obtained by maximum likelihood estimation bysing software for parametric survival models. Normal distri-ution for the pain scores was assumed. These estimates wereompared by log-rank analysis.

Finally, binary logistic regression and backward stepwise anal-sis (method, likelihood ratio; entry at 0.05 probability, removal at.10 probability) was performed to identify demographic, clinicalsymptom severity), and psychological characteristics that predicthe occurrence of visceral hypersensitivity. Age, gender, health careevel, predominant bowel habit, postinfectious symptom onset,ectal compliance, symptom severity, anxiety, depression, somaticymptoms, psychoneuroticism, dysfunctional cognitions regard-ng functional bowel disorders, vigilance, pain coping, somatiza-ion, and quality of life (general health subscale) were entered inhe analysis as separate predictors. Data were expressed as mean �D. The level of significance was set at P � .05.

ResultsSubject CharacteristicsWe screened 130 patients, 26 of whom did not meet

ome II criteria, and 40 healthy volunteers. Two patients de-

cteristics of IBS Patients and Healthy Controls

nts (N � 101) Healthy controls (N � 40)

0 � 13.9 39.7 � 15.073 63

34 035 024 08 00 100

4 � 2.5a 0.43 � 0.57

4 � 4.6b 12.2 � 3.75 � 6.9a 20.7 � 8.33 � 5.6a 15.0 � 3.78 � 31.9a 113.3 � 30.73 � 35.8a 85.7 � 37.37 � 5.8 7.7 � 4.74 � 1.0b 3.7 � 0.86 � 0.4a 0.3 � 0.3

0 � 20.4a 94.1 � 10.50 � 42.0a 87.2 � 28.61 � 19.6a 90.3 � 16.12 � 16.3 78.5 � 13.48 � 35.3 91.0 � 26.82 � 23.7a 90.9 � 14.35 � 16.9a 70.8 � 15.82 � 18.8a 75.1 � 14.6

r. Data are expressed as mean � SD.

hara

patie

42.

4.

13.22.18.

123.110.

9.3.0.

82.60.62.75.80.73.58.61.

mete

cd1tt

rgua

ptssntshlt(

Ivad((I

Ftdm(Cam

Fa

T

ADSPDVPS

N mete

324 VAN DER VEEK ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

lined to participate in the barostat study, and 1 patient wasiagnosed with conversion disorder. All healthy volunteers and01 patients provided informed consent and were included inhe final analysis. Thirty-one patients (31%) were recruitedhrough the outpatient department, and 70 patients (69%) were

igure 1. (A) Dynamic rectal compliance (mL/mm Hg) in 101 IBS pa-ients (squares) and 40 controls (triangles). Compliance was significantlyecreased in patients compared with controls. Data are expressed asean � standard error of the mean. (B) Dynamic rectal compliance

mL/mm Hg) in IBS-D, IBS-C, and IBS-A patients and 40 controls.ompliance was significantly increased in IBS-C compared with IBS-Dnd IBS-A, but similar compared with controls. Data are expressed as

able 2. Psychological Profile of Patients Recruited From the

Psychological profile Referra

nxiety (10–50) 1epression (16–80) 2omatic symptoms (12–60) 1sychoneuroticism (90–450) 12ysfunctional cognitions (31–217) 10igilance (0–40)ain coping (6–1)omatization (0–2)

OTE. Score ranges from best to worst are indicated after each para

ean � standard error of the mean. c

ecruited through advertisement. All patients in the latterroup had previously consulted a physician and had been eval-ated for their abdominal symptoms. Healthy controls werelso recruited through advertisement.

Demographic, clinical, and psychological characteristics ofatients and controls are listed in Table 1. Mean age and maleo female ratio were not different between groups. Symptomeverity and levels of anxiety, depression, psychoneuroticism,omatization, other somatic symptoms, and dysfunctional cog-itions were all slightly but significantly increased in IBS pa-ients compared with healthy controls. Pain coping scores wereignificantly reduced in IBS. Compared with controls, patientsad impaired quality of life on 6 of 8 SF-36 subscales. Psycho-

ogical measures were not different between patients from theertiary referral center and those from the general populationTable 2).

Rectal Compliance and PerceptionRectal compliance was significantly reduced in the

BS group compared with healthy control subjects (29.7 � 13s 41.8 � 18 mL/mm Hg, P � .0001) (Figure 1A). Subgroupnalysis showed that rectal compliance was particularly re-uced in patients with a diarrhea-predominant bowel habit

IBS-D; P � .04) and those with alternating bowel habitIBS-A; P � .05) compared with constipation-predominantBS (IBS-C) (Figure 1B).

igure 2. Intensity of urge perception in 101 IBS patients (squares)nd 40 controls (triangles). Urge did not differ between patients and

iary Referral Center and From the General Population

ter (N � 31) General population (N � 70)

3.4 13.7 � 5.05.1 22.7 � 7.64.0 18.2 � 6.322.5 124.6 � 35.734.2 110.6 � 36.74.2 10.4 � 6.31.0 3.4 � 1.00.3 0.6 � 0.4

r. Data are expressed as mean � SD.

Tert

l cen

2.9 �2.2 �8.6 �2.1 �9.6 �8.2 �3.5 �0.7 �

ontrols. Data are expressed as mean � standard error of the mean.

H2caIi2dag

wceopllg

wcn

wHw1tp

mr1pt

F4p.

T

CUPTT

Nma

b

c

Ft(d

March 2008 PREDICTORS OF VISCERAL HYPERSENSITIVITY IN IBS 325

Urge perception at high rectal pressure distention (30 mmg) was not significantly different between IBS patients (6.6 �

.7 cm) and controls (6.1 � 2.6 cm) (P � .30). The pressure-urgeurves were also not significantly different between patientsnd controls (pressure by group interaction, P � .82; Figure 2).n contrast, pain perception at high rectal pressure was signif-cantly increased in IBS patients compared with controls (2.5 �.7 vs 1.0 � 1.4 cm, P � .003), and the pressure-pain curvesiffered significantly between groups (pressure by group inter-ction, P � .0001; Figure 3). No differences between IBS sub-roups were found (Table 3).

Perception ThresholdsUrge thresholds were reached in all participants, but they

ere somewhat reduced in IBS patients (15.6 � 6.1 mm Hg)ompared with controls (18 � 6.0 mm Hg) (P � .042). No differ-nces were found between IBS subgroups (Table 3). In contrast,nly 10 of 40 control subjects (25%), compared with 55 of 101 IBSatients (54%), reached the threshold for rectal pain during bal-

oon distention (�2 � 10.01, P � .002) (Figure 4). Maximumikelihood estimation of the mean pain threshold and SD in eachroup and subsequent log-rank analysis showed that the threshold

igure 3. Intensity of pain perception in 101 IBS patients (squares) and0 controls (triangles). Pain perception was significantly increased inatients compared with controls (pressure by group interaction, P �

0001). Data are expressed as mean � standard error of the mean.

able 3. Rectal Compliance and Perception in IBS Patients,

MeasureIBS-D

(N � 34)IBS-C

(N � 35

ompliance (mL/mm Hg) 27.2 � 11 35.2 � 1rge at 30 mm Hg (cm) 6.5 � 2.9 6.7 � 2ain at 30 mm Hg (cm) 2.2 � 2.6 2.9 � 2hreshold urge (mm Hg) 16.9 � 6.6 14.2 � 5hreshold pain (mm Hg) 31.3 � 18 23.6 � 1

OTE. Data for the group with unknown bowel habit are not shown bean � SD.

P � .01 compared with controls.P � .05 compared with IBS-D and IBS-A.

P � .05 compared with controls.

as significantly reduced in IBS patients (27.5 � 15 mm Hg)ompared with controls (35.3 � 8.2 mm Hg) (P � .0009) but didot differ between IBS subgroups (Table 3).

Visceral HypersensitivityThe threshold for hypersensitivity to balloon distention

as set at 18.9 mm Hg (35.3 minus 16.4 [ie, 2 � 8.2 SD] mmg). Thirty-three IBS patients (33%), compared with 0 controls,ere identified as hypersensitive to balloon distention (�2 �7.06, P � .0001) (Table 4). Thus, pain thresholds fell outsidehe range of control subjects in approximately 1 in 3 IBSatients.

Predictors of Visceral HypersensitivityOf all tested variables, only IBS symptom severity re-

ained as a predictor of visceral hypersensitivity in the logisticegression analysis (odds ratio, 1.25; 95% confidence interval,.04 –1.50; P � .016). Table 5 lists demographic, clinical, andsychological characteristics in hypersensitive and normosensi-ive patients. IBS symptom scores were significantly higher in

ubgroups, and Healthy Controls

S patients

Controls(N � 40)

IBS-A(N � 24)

All patients(N � 101)

26.6 � 11 29.7 � 13a 41.8 � 186.7 � 2.9 6.6 � 2.7 6.1 � 2.62.6 � 3.0 2.5 � 2.7a 1.0 � 1.4

15.6 � 6.1 15.6 � 6.1c 18.0 � 6.029.6 � 15 27.5 � 15a 35.3 � 8.2

se of the small number of patients (N � 8). Data are expressed as

igure 4. Individual pain thresholds in IBS patients and healthy con-rols. Significantly more patients (N � 55, 54%) compared with controlsN � 10, 25%) reached the pain threshold before the end of the rampistention (dotted line, 30 mm Hg).

IBS S

IB

)

4b

.6

.7

.63

ecau

h2

tppp

aphiurdn

bprsrahsbtrmptwp

dsaadsbhpoonpda

csphIttp

pbFlirstimroill

T

IC

a

T

AFRB

PDI

GADSPDVPSAL

a

[b

326 VAN DER VEEK ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

ypersensitive compared with normosensitive patients (5.4 �.5 vs 4.0 � 2.4, P � .007). No other differences were found.

DiscussionThe present study showed that (1) visceral hypersensi-

ivity is an important feature of IBS but is not present in allatients, and (2) hypersensitivity to rectal balloon distention isredicted by IBS symptom severity but not by demographic orsychological characteristics.

Our data confirmed previous findings that rectal compliancend pain thresholds are reduced and that the intensity of painerception is increased in IBS patients when compared withealthy controls. Urge intensity at any given pressure was sim-

lar in patients and controls, with slightly lower thresholds forrge in IBS patients. Our observation that pain perceptionather than urge is increased is consistent with other reportsemonstrating decreased perception thresholds in IBS only foroxious stimuli and not for stool.11

It is presumed that a phasic distention protocol (ie, rapidalloon inflation to predefined pressure levels) is the preferredrocedure to test visceral hypersensitivity, because this would elicitectal sensations at lower volumes or pressures compared withlow ramp distention.27,28 However, we chose to perform onlyamp distentions because we considered rectal compliance to ben important factor in the model on predictors of visceralypersensitivity, and compliance is best measured by means oflow ramp distention.28 Phasic distentions were not performedecause assessment of sensory thresholds during phasic disten-ions after preceding ramp distention might introduce perceptualesponse bias, and phasic distentions before ramp distention

ight affect subsequent rectal compliance measurements. Theain thresholds we observed during ramp distention are similaro those reported by others using phasic distentions,10,29,30

hich supports previous findings that the type of distentionrocedure (phasic, ramp, etc) does not affect perception.31

One of our main findings is that hypersensitivity to balloonistention was less likely to occur in patients with milderymptoms. This challenges the view that visceral hyperalgesia is

biologic marker of IBS,11 because hypersensitivity might bebsent in Rome II–positive patients with mild symptoms. Theifference in the proportion of hypersensitive patients in thetudy by Mertz et al11 (95%) and our study (33%) might in parte due to the use of different parameters to define visceralypersensitivity. Mertz et al11 used 3 parameters to score rectalerception simultaneously (ie, perception thresholds, intensityf sensations, and altered viscerosomatic referral), whereas wenly identified patients having decreased pain thresholds andot those having decreased discomfort thresholds or alteredain referral patterns. It is, of course, essential to use equalefinitions of visceral hypersensitivity when comparing its prev-

able 4. Visceral Hypersensitivity in IBS Patients andHealthy Controls

Hypersensitive Normosensitive

BS (N � 101) 33 (33%)a 68 (67%)ontrols (N � 40) 0 (0%) 40 (100%)

P � .001 compared with controls.

lence between studies. Because no accepted definition of vis- c

eral hypersensitivity is currently available, we decided to use atatistical point of view and consider patients with a painerception threshold �2 SDs below the mean threshold inealthy controls as hypersensitive to rectal balloon distention.

n general, this method is accepted to define outliers. Althoughhis cutoff is arbitrary, our data suggested that hypersensitivityo rectal distention is not a suitable biologic marker to identifyatients with IBS.

The pathophysiology of visceral hyperalgesia in IBS remainsoorly understood. Recent evidence suggested that distur-ances might occur at different levels of the brain-gut axis.irst, sensitization of peripheral nerve endings at the intestinal

evel might occur during or after acute inflammation,12,13 lead-ng to higher excitability and/or increased firing of these neu-ons. Second, some studies suggested that alterations in thepinal dorsal horn neurons might provide an explanation forhe extended viscerosomatic referral pattern that is often seenn IBS.11,12 Third, altered processing of afferent visceral infor-

ation in the brain, particularly in the prefrontal cortex, ante-ior cingulated cortex, and thalamus, has repeatedly been dem-nstrated in IBS patients.15,32 These regions are not only

nvolved in pain processing but are also part of the emotionalimbic system and are therefore involved in numerous psycho-ogical and cognitive events.16,17 Because nociception (becoming

able 5. Demographic, Clinical, and PsychologicalCharacteristics of Hypersensitive andNormosensitive IBS Patients

CharacteristicHypersensitive

(N � 33)Normosensitive

(N � 68)

ge (y) 40.7 � 12.4 42.6 � 14.5emale sex (%) 73 74ecruitment (%) advertisement 68 71owel habit (%)Diarrhea 33 34Constipation 46 29Alternating 18 27Not specified/normal 3 10

ostinfectious (%) 11 13ynamic compliance 31.8 � 14.9 28.6 � 11.3

BS composite score 5.4 � 2.5a 4.0 � 2.4Discomfort 1.38 � 0.8b 1.17 � 0.62Pain 1.34 � 0.95c 0.98 � 0.72Constipation 0.73 � 0.64c 0.37 � 0.56Diarrhea 0.45 � 0.86 0.48 � 0.69Bloating 1.37 � 0.79c 1.01 � 0.75

eneral health 62.7 � 16.4 60.5 � 19.9nxiety 13.9 � 5.0 13.2 � 4.4epression 23.1 � 6.5 22.3 � 7.1omatic symptoms 19.0 � 4.5 18.0 � 6.1sychoneuroticism 126.5 � 32.2 122.5 � 32.0ysfunctional cognitions 106.8 � 35.3 111.9 � 36.1igilance 9.2 � 5.3 9.9 � 6.1ain coping 3.5 � 1.1 3.3 � 0.9omatization 0.6 � 0.4 0.6 � 0.4ntispasmodics (%) 15 12axatives or bulking agents (%) 30 31

P � .007 vs normosensitive patients (range, 0 [no symptoms] to 20worst imaginable]).P � .072 vs normosensitive patients.

P � .02 vs normosensitive patients.

abpictficdwdccspcl

lsradmddtpftta

pcwmwvvtdodInc

oisunth

rdwcp

sbfI

1

1

1

1

1

1

1

1

1

1

March 2008 PREDICTORS OF VISCERAL HYPERSENSITIVITY IN IBS 327

ware of a painful stimulus) and emotional pain managementoth occur in similar regions of the brain, we hypothesized thatsychological disturbances are related to visceral hypersensitiv-

ty. However, our results did not support this hypothesis, be-ause none of the psychological variables we studied predictedhe occurrence of hypersensitivity to balloon distention. Thesendings substantiated previous observations that psychologicalharacteristics such as anxiety, somatization, and neuroticismo not correlate with sensory thresholds.9,11,19 Similar resultsere obtained in a recent study in which multivariate analysisemonstrated that abdominal pain and bloating were signifi-antly associated with altered rectal perception, whereas psy-hological symptoms such as anxiety were not.33 Our data alsohowed that rectal hyperalgesia is not associated with othersychological factors (vigilance, dysfunctional cognitions, painoping), demographic characteristics (age, gender), quality ofife, or predominant bowel habit.

Previously Whitehead et al34 proposed a model for psycho-ogical factors that influence pain perception in IBS. It wasuggested that low pain thresholds in IBS are influenced by 2elated cognitive traits, ie, selective attention to gut sensationsnd a tendency to interpret these sensations as symptoms ofisease. Our data showed that neither vigilance (selective so-atic attention) nor cognitions regarding functional bowel

isorders (interpretation of normal sensations as symptoms ofisease) were different between hypersensitive and normosensi-ive IBS patients. These findings suggested that hypersensitiveatients do not perceive or manage their symptoms differentlyrom normosensitive patients. Although vigilance and cogni-ions on functional bowel disorders differed significantly be-ween patients and controls, these parameters were not associ-ted with increased rectal sensitivity.

We aimed to obtain a representative sample from the IBSopulation by recruiting patients both from the outpatientlinic and by advertisement. Levels of psychological distressere low and did not differ significantly between groups. Oneight argue that low levels of psychopathology explained whye found no correlation between psychological variables and

isceral hypersensitivity, because a certain degree of parameterariability was required for correlations to be detected. Al-hough some studies found significantly more psychologicalisturbances in IBS patients recruited from tertiary care,18,19,35

ne of these studies found no relation between psychologicalistress and visceral hypersensitivity in clinic patients withBS,19 supporting our finding that visceral hypersensitivity wasot affected by psychopathology, regardless of level of healthare.

Allowing patients to take antispasmodics, laxatives, and,ccasionally, analgesics during barostat measurements is a lim-

tation of this study because it might interfere with visceralensitivity and affect sensory thresholds in general. Althoughse of these medications was similar in hypersensitive andormosensitive patients, prohibiting the use of these medica-ions might have further increased the number of patients withypersensitivity to balloon distention in both groups.

In conclusion, we found that patients with IBS had impairedectal compliance and reduced sensory thresholds to rectalistention compared with controls. Visceral hypersensitivityas present in one third of our IBS population and was asso-

iated with increased symptom severity. Although psychological

arameters did not predict the occurrence of visceral hypersen-

itivity, this does not exclude a common neuropsychologicalasis in the pathophysiology of IBS. Future studies shouldocus on the role of the brain-gut axis in the development ofBS.

References

1. Thompson WG, Longstreth GF, Drossman DA, et al. Functionalbowel disorders and functional abdominal pain. Gut 1999;45(Suppl 2):II43–II47.

2. Chey WY, Jin HO, Lee MH, et al. Colonic motility abnormality inpatients with irritable bowel syndrome exhibiting abdominal painand diarrhea. Am J Gastroenterol 2001;96:1499–1506.

3. Aggarwal A, Cutts TF, Abell TL, et al. Predominant symptoms inirritable bowel syndrome correlate with specific autonomic ner-vous system abnormalities. Gastroenterology 1994;106:945–950.

4. Van der Veek PP, Swenne CA, Van de Vooren CA, et al. Viscer-osensory-cardiovascular reflexes: altered baroflex sensitity in irri-table bowel syndrome. Am J Physiol 2005;289:R970–976.

5. Rodriguez LA, Ruigomez A. Increased risk of irritable bowel syn-drome after bacterial gastroenteritis: cohort study. BMJ 1999;318:565–566.

6. Gwee KA, Collins SM, Read NW, et al. Increased rectal mucosalexpression of interleukin 1beta in recently acquired post-infec-tious irritable bowel syndrome. Gut 2003;52:523–526.

7. Gonsalkorale WM, Perrey C, Pravica V, et al. Interleukin 10 ge-notypes in irritable bowel syndrome: evidence for an inflammatorycomponent? Gut 2003;52:91–93.

8. Van der Veek PP, van den Berg M, Kroon YE, et al. Role of tumornecrosis factor-alpha and interleukin-10 gene polymorphisms inirritable bowel syndome. Am J Gastroenterol 2005;40:2510–2516.

9. Whitehead WE, Holtkotter B, Enck P, et al. Tolerance for rectosig-moid distention in irritable bowel syndrome. Gastroenterology1990;98:1187–1192.

0. Bouin M, Plourde V, Boivin M, et al. Rectal distention testing inpatients with irritable bowel syndrome: sensitivity, specificity,and predictive values of pain sensory thresholds. Gastroenterol-ogy 2002;122:1771–1777.

1. Mertz H, Naliboff B, Munakata J, et al. Altered rectal perceptionis a biological marker of patients with irritable bowel syndrome.Gastroenterology 1995;109:40–52.

2. Mayer EA, Gebhart GF. Basic and clinical aspects of visceralhyperalgesia. Gastroenterology 1994;107:271–293.

3. Olivar T, Cervero F, Laird JM. Responses of rat spinal neurones tonatural and electrical stimulation of colonic afferents: effect ofinflammation. Brain Res 2000;866:168–177.

4. Verne GN, Himes NC, Robinson ME, et al. Central representationof visceral and cutaneous hypersensitivity in the irritable bowelsyndrome. Pain 2003;103:99–110.

5. Silverman DH, Munakata JA, Ennes H, et al. Regional cerebralactivity in normal and pathological perception of visceral pain.Gastroenterology 1997;112:64–72.

6. Bishop S, Duncan J, Brett M, et al. Prefrontal cortical functionand anxiety: controlling attention to threat-related stimuli. NatNeurosci 2004;7:184–188.

7. Bush G, Luu P, Posner MI. Cognitive and emotional influences inanterior cingulate cortex. Trends Cogn Sci 2000;4:215–222.

8. Drossman DA, McKee DC, Sandler RS, et al. Psychosocial factorsin the irritable bowel syndrome: a multivariate study of patientsand nonpatients with irritable bowel syndrome. Gastroenterology1988;95:701–708.

9. Guthrie E, Creed F, Fernandez L, et al. Cluster analysis of symp-toms and health seeking behaviour differentiates subgroups ofpatients with severe irritable bowel syndrome. Gut 2003;52:

1616–1622.

2

2

2

2

2

2

2

2

2

2

3

3

3

3

3

3

mCl

D

fe

328 VAN DER VEEK ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

0. van der Veek PP, van Rood YR, Masclee AA. Clinical trial: short-and long-term benefit of relaxation training for irritable bowelsyndrome. Aliment Pharmacol Ther 2007;26(6):943–952.

1. Sheehan DV, Lecrubier Y, Sheehan KH, et al. The Mini-Interna-tional Neuropsychiatric Interview (M.I.N.I.): the development andvalidation of a structured diagnostic psychiatric interview forDSM-IV and ICD-10. J Clin Psychiatry 1998;59(Suppl 20):22–33.

2. Derogatis LR, Rickels K, Rock AF. The SCL-90 and the MMPI: astep in the validation of a new self-report scale. Br J Psychiatry1976;128:280–289.

3. Dahlstrom GW, Welsh GS, Dahlstrom LE. An MMPI handbook:volume I—clinical interpretation. Minneapolis: University of Min-nesota Free Press, 1972.

4. Speckens AE, Spinhoven P, Sloekers PP, et al. A validation studyof the Whitely Index, the Illness Attitude Scales, and the Somato-sensory Amplification Scale in general medical and general prac-tice patients. J Psychosom Res 1996;40:95–104.

5. Toner BB, Stuckless N, Ali A, et al. The development of a cogni-tive scale for functional bowel disorders. Psychosom Med 1998;60:492–497.

6. Brazier JE, Harper R, Jones NM, et al. Validating the SF-36 healthsurvey questionnaire: new outcome measure for primary care.BMJ 1992;305:160–164.

7. Sun WM, Read NW, Prior A, et al. Sensory and motor responsesto rectal distention vary according to rate and pattern of ballooninflation. Gastroenterology 1990;99:1008–1015.

8. Whitehead WE, Delvaux M. Standardization of barostat proce-dures for testing smooth muscle tone and sensory thresholds inthe gastrointestinal tract: the Working Team of Glaxo-WellcomeResearch, UK. Dig Dis Sci 1997;42:223–241.

9. Chang L, Munakata J, Mayer EA, et al. Perceptual responses inpatients with inflammatory bowel disease. Gut 2000;47:497–

505. m

0. Naliboff BD, Munakata J, Fullerton S, et al. Evidence for twodistinct perceptual alterations in irritable bowel syndrome. Gut1997;41:505–512.

1. Hammer HF, Phillips SF, Camilleri M, et al. Rectal tone, disten-sibility, and perception: reproducibility and response to differentdistensions. Am J Physiol 1998;274:G584–G590.

2. Ringel Y, Drossman DA, Turkington TG, et al. Regional brainactivation in response to rectal distension in patients with irrita-ble bowel syndrome and the effect of a history of abuse. Dig DisSci 2003;48:1774–1781.

3. Posserud I, Syrous A, Lindström L, et al. Altered rectal perceptionin irritable bowel syndrome is associated with symptom severity.Gastroenterology 2007, doi:10.1053/j.gastro.2007.07.024.

4. Whitehead WE, Palsson OS. Is rectal pain sensitivity a biologicalmarker for irritable bowel syndrome: psychological influences onpain perception. Gastroenterology 1998;115:1263–1271.

5. Longstreth GF, Hawkey CJ, Mayer EA, et al. Characteristics ofpatients with irritable bowel syndrome recruited from threesources: implications for clinical trials. Aliment PharmacolTher 2001;15:959–964.

Address requests for reprints to: Dr P. P. J. van der Veek, Depart-ent of Gastroenterology and Hepatology, Leiden University Medical

enter, Building 1, C4-P, PO Box 9600, 2300 RC, Leiden, The Nether-ands. e-mail: P.P.J.van_der_Veek@lumc.nl; fax: �31-71-5248115.

This study was supported by a grant from the Dutch Digestiveiseases Foundation (NVGE).We thank Saskia le Cessie of the Department of Medical Statistics

or statistical advice and our colleagues at the Department of Gastro-nterology and Hepatology for assistance in the barostat measure-

ents.

CN

N

*D

BappsomqowDwHo(Ddtd(co1(l(fs

Fsiaaiwpcwii

cRiap

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2008;6:329–332

hronic Abdominal Pain and Depressive Symptoms: Analysis of theational Longitudinal Study of Adolescent Health

ADER N. YOUSSEF,* KATHERINE ATIENZA,* ANNETTE L. LANGSEDER,* and RICHARD S. STRAUSS‡

Center for Pediatric Irritable Bowel & Motility Disorders, Goryeb Childrens Hospital at Atlantic Health, Morristown, New Jersey; and ‡Johnson and Johnson, Product

evelopment Unit, Titusville, New Jersey

Siif

2iobs

s(igIC1mDwbrtrt

tsatm

owdfcq

SD

ackground & Aims: Abdominal pain is common indolescence. The aim of this study was to determine therevalence of depressive symptoms in a large cohort ofatients with frequent abdominal pain. Methods: A pro-pective, cross-sectional, nationally representative samplef children aged 13 to 18 years (mean age, 16.2 � 1.7 y; 49%ale) completed in-home interviews and separate in-school

uestionnaires for the National Longitudinal Study in Ad-lescent Health (the Add Health Study). Depressed moodas assessed with the Center for Epidemiologic Studiesepression Scale. Subjective measures of abdominal painere reported by 20,745 adolescents from wave 1 of the Addealth Study. Frequency of abdominal pain over the previ-

us 1 year was rated as rare (0 –1 episode/wk), moderate2–3 episodes/wk), or daily (>4 episodes/wk). Results:aily pain is reported in 3.2% of adolescents, with an ad-itional 14% reporting pain as moderate in frequency. Six-

een percent of all adolescents are at risk for developingepression. The risk for depression goes from 16% to 45%P < .001) when the pain is daily. Compared with rare pain,hildren with daily pain were more likely to miss school 10r more times per year (46% vs 19%, P < .001), cry (12.1% vs%, P < .001), feel sad (25.2% vs 5.3%, P < .001), and lonely25.2% vs 6.4%, P < .001). Children with daily pain wereikely to consider life a failure versus those with no pain10.2% vs 3.3%, P < .001). Conclusions: Adolescents withrequent abdominal pain are at increased risk for depres-ive symptoms, social isolation, and missing school.

requent abdominal pain is a common complaint amongadolescents in the United States, with 17% of high school

tudents in small community, school-based samples experienc-ng abdominal pain once a week.1 The exact prevalence ofbdominal pain is unknown but does increase with age throughdolescence and affects more females. Frequent abdominal painn children and adolescents has been associated with socialithdrawal, school absenteeism, and a poor quality of life foratients and their parents.2 Other associated restrictions in-lude decreased appetite, not playing with friends, and less timeith hobbies.3 Suffering from frequent abdominal pain earlier

n life often leads to increased health care use as adults and alsos more likely to result in an anxiety disorder later in life.4

Psychiatric morbidities such as anxiety and depression areommon in adults who suffer from frequent abdominal pain.ecent literature has examined the relationship of suicidal

deation and depression in adults who suffer from both organicnd nonorganic abdominal pain. Adult patients with chronic

ain have higher rates of suicidal ideation.5 A recent study by

mith et al5 suggested that abdominal pain in adults may be anndependent risk factor for suicidal ideation. Moreover, suicidaldeation increases with symptom severity and perceived inter-erence with life in adults with irritable bowel syndrome.6

The lifetime prevalence of depression through adolescence is0%.7 One study noted the rate of weekly depressive symptoms

n 15-year-olds to be 49% in females and 34% in males.7 The aimf this study was to examine on a national level the associationetween frequent abdominal pain in adolescents and depressiveymptoms.

MethodsA prospective, cross-sectional, nationally representative

ample of adolescents in the United States aged 13 to 18 yearsmean age, 16.2 � 1.7 y; 49% male) completed both in-housenterviews and at-school questionnaires for the National Lon-itudinal Study in Adolescent Health (the Add Health Study).nitiated in 1994 under a grant from the National Institute ofhild Health and Human Development with co-funding from7 other federal agencies, the Add Health Study is the largest,ost comprehensive survey of adolescents ever undertaken.8

ata at the individual, family, school, and community levelsere collected in 2 waves between 1994 and 1996. In addition toaseline demographic data, measures of health status, health-elated behaviors, and social function including school activi-ies, friends, and home activities were collected. A parentalespondent gave information on household income and paren-al education.

The sample population of adolescents was stratified accordingo region, school type, ethnicity, and urban areas. The currenttudy used data from the in-home interview sample of 20,745dolescents (wave 1 in April–December 1995). Additional informa-ion on the Add Health Study design, sampling strategy, and

ethodology are available at www.cpc.unc.edu/addhealth.Depressed mood was assessed with the Center for Epidemi-

logic Studies Depression Scale (CES-D). The CES-D initiallyas designed to screen adults for symptoms of depressionuring the previous week. It is not a psychiatric diagnostic toolor clinical depression but identifies individuals at risk for alinical diagnosis of depression. The Add Health Study containsuestions about depressive symptoms that correspond to those

Abbreviations used in this paper: Add Health, National Longitudinaltudy in Adolescent Health; CES-D, Center for Epidemiologic Studiesepression Scale.

© 2008 by the AGA Institute1542-3565/08/$34.00

doi:10.1016/j.cgh.2007.12.019

fseC1emffiasi

dttctwdw

famma

sdWlpassrtt

wicCpsiw

rda

rt.hw.

twav(

tof

crsiofa

Fwdt

330 YOUSSEF ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

ound in the 20-question CES-D. Depressive symptoms mea-ured included feeling sad, feel like crying, feeling lonely, lownergy, and life as failure. The modified the Add Health StudyES-D scale is valid for use by adolescents.9 A cut-off score of6 has been used to define cases of depression in adults, butstimates the prevalence of depression in adolescents to beore than 50%. Roberts et al10 determined that scores of 24 in

emales and 22 in males maximized the sensitivity and speci-city of the CES-D for predicting major depressive disordermong adolescents. We thus defined risk of depression in thistudy as a CES-D score of 23 or greater in adolescents, as othernvestigators have.10

The Add Health Study in-home questionnaire asked stu-ents how often they had a stomachache or upset stomach inhe past 12 months. Respondents did not have the opportunityo characterize or quantify the duration of abdominal painomplaints during the day. Frequency of abdominal pain overhe past year was rated as rare if there were 0 to 1 episodes pereek, occasional if there were 2 to 3 episodes per week, andaily if there were daily or almost-daily episodes of pain pereek.

Data AnalysisVariables in our analysis were selected on the basis of

actors previously described as relevant to depressive disordersnd adolescent mental health. These items included sociode-ographics, health (subjective self-rating of abdominal pain),ental health issues (sadness, life is a failure), and school

bsenteeism.Data were analyzed with SPSS version 12.0 (Chicago, IL)

tatistical software. Bivariate analyses of outcomes and socio-emographic factors were conducted with chi-squared analysis.e analyzed potential predictive variables with multivariate

ogistic regression for the outcome of abdominal pain andersistent moderate/severe depressive symptoms (CES-D, �23)mong adolescents. Logistic regression models were con-tructed in a forward fashion, adding variables bivariate analy-es and stepwise modeling. Significant variables (P � .05) wereetained in the model, and additional demographic variableshat were significant such as school absenteeism and participa-ion in social activities were retained in the final model.

ResultsDaily pain was reported in 3.2% of the study population

ith an additional 14% reporting abdominal pain as moderaten frequency. In this representative population, 16% of adoles-ents had significant depressive symptoms as defined by aES-D score greater than 23. Adolescents with daily abdominalain were significantly more likely to have increased CES-Dcores of 23 or greater compared with those with rare abdom-nal pain (45% vs 3.2%, P � .001). Of those at risk for depressionith daily abdominal pain, females outnumbered males 2:1.

Effect of Abdominal Pain on SchoolAttendanceAdolescents who suffered from daily abdominal pain

eported an increased prevalence of missing school 10 or moreays per year as compared with those with minimal or no

bdominal pain (46% vs 19%, P � .001) (Figure 1). A

Effect of Abdominal Pain on Social ActivitiesAdolescents who suffered from daily abdominal pain

eported less participation in social activities as compared withhose with minimal or no abdominal pain (0.6% vs 1.9%, P �04) (Figure 2). Social activities included hobbies, sports, andanging out with friends. Children with daily abdominal painere less likely to participate in active sports (19% vs 30%, P �

001) or to participate in school activities (P � .002).

Associated Depressive SymptomsAdolescents with daily abdominal pain were more likely

o feel tired, cry, feel sad, and feel lonely as compared with thoseith no abdominal pain (P � .001). Adolescents with dailybdominal pain were more likely to consider their lives a failureersus those with no abdominal pain (10.7% vs 3.3%, P � .001)Table 1).

Children with daily pain were more likely to have changedheir primary residence in the prior 5 years (P � .05). Frequencyf abdominal pain or risk of depression was not related toamily socioeconomic status.

DiscussionUtilizing the National Longitudinal Study of Adoles-

ent Health, a prospective US population based registry, theelationship between frequent abdominal pain and depressiveymptoms was examined. Adolescents who experience abdom-nal pain more than 4 times a week have an increased prevalencef school absenteeism (�10 days per year), fatigue, crying, andeelings of sadness and loneliness as compared with healthydolescents who had abdominal pain less than once a week.

igure 1. Effect of abdominal pain on school attendance. Adolescentsho suffered from daily abdominal pain reported an increased inci-ence of missing school more than 10 days per year as compared withhose with minimal or no abdominal pain (46% vs 19%, *P � .001).

lso, adolescents with frequent abdominal pain were more

lfpcwrf

aayeledr

avipceimscdqfid

fp

tdodp

stsaCpaaimfa

ttsmsssa

ddmycatta2sdspsbsls

ac

Fwsp

T

FFFF

March 2008 ADOLESCENT ABDOMINAL PAIN AND DEPRESSION 331

ikely to restrict their daily activities and consider their lives aailure compared with children with rare episodes of abdominalain. From the Add Health Study population, 16% of adoles-ents are at risk for depression as defined by the CES-D score,hich dramatically increases to 45% when those patients expe-

ience frequent abdominal pain. About 40% of children withrequent abdominal pain meet criteria for depression.11

More than 15% of the Add Health Study responders hadbdominal pain more than 2 times per week. This was a largend diverse sample of children followed up prospectively for 1ear to assess abdominal pain complaints. Previous studiesstimating abdominal pain prevalence in children have beenimited by their smaller single-community samples. Ghandourt al12 reported that 20% of adolescent females experience ab-ominal pain more than once a week, which is similar to 17%eported by Hyams et al1 in high school students.

Numerous articles have shown that children with chronicbdominal pain were more likely to have increased health ser-ice use, restrictions in daily living, and suffer from functionalmpairments in adulthood.3,4,11,13 In this cohort of adolescentatients identified as having daily abdominal pain, we haveonfirmed the significant negative impact on quality of life asvidenced by the increased days of school missed and dimin-shed participation in social activities. School absenteeism, not

eeting friends, loss of appetite, and sleep disturbances areome of the restrictions of daily living in children and adoles-ents with chronic abdominal pain that may need to be ad-ressed in future intervention studies. These restriction in dailyuality of life may indeed contribute to the most worrisomending in our study that more than 10% of adolescents withaily abdominal pain believed their life was a failure.

In adults, chronic abdominal pain is an independent riskactor for suicidal ideation.5 Adults who experience abdominal

igure 2. Effect of abdominal pain on social activities. Adolescentsho suffered from daily abdominal pain reported less participation inocial activities as compared with those with minimal or no abdominalain (0.6 vs 1.9, *P � .04).

ain from IBS are thought to have suicidal ideations because ofF

he belief that their disorder is hopeless. Also, their perceivedisease severity independently increases suicidal ideations.6 Rec-gnition of frequent abdominal pain as a chronic pain syn-rome allows for identification of adolescents at risk for de-ression and self-injurious behavior.

Little et al14 were able to document positive depressioncreens in children with chronic abdominal pain and nongas-rointestinal somatic complaints. Specific somatic complaintsuch as musculoskeletal pain 3 times a week in boys and girlsnd weekly headaches in girls may be clues for depression.15

hildren with chronic abdominal pain have more somatic com-laints compared with healthy children, which includes head-che, stomachache, backache, and morning fatigue, possibly asresponse to stress.4,13 Although it was not an aim of this study,

t would have been interesting to measure the frequency ofusculoskeletal pain and headaches among adolescents with

requent abdominal pain and possibly to determine an associ-tion with depressive symptoms as well.

Although the Add Health Study relies on self-reported symp-oms, the overall quality of the data is considered quite high. Inhis study, measures that can be validated externally such aself-reported height and weight were correlated highly with

easured height and weight.16 In addition, self-reported friend-hip nominations also have been validated externally.17 Thistudy shows an association with abdominal pain and depressiveymptoms in adolescents, which is more significant when thebdominal pain occurs daily.

The limitations of this study were that the Add Health studyid not provide extensive medical information on the respon-ents, particularly regarding whether any interventions wereade or treatment was started. Also, it has been more than 10

ears since the data gathered in the Add Health Study wereollected and published. One may question whether the datare reflective of current adolescents. The recent publication ofhe Youth Risk Behavior Surveillance System has confirmedhat the rates of depressive symptoms and suicide attemptsmong adolescents have not changed significantly from 1991 to005.18 In addition, with the recent insights of Smith el al5

uggesting that abdominal pain in adults may be an indepen-ent risk factor for suicidal ideation and may be related to bothymptom severity and duration, the data reported here haveotential important public health concerns for adolescents whouffer similarly from chronic abdominal pain. It would haveeen ideal to have a control group for this large cohort totrengthen the association of frequent abdominal pain in ado-escents and increased risk for depressive symptoms, but thistudy analyzed previously gathered data.

The prevalence of frequent abdominal pain increases withge into adolescence, which is similar to depression. The asso-iation of frequent abdominal pain and anxiety in children is

able 1. Abdominal Pain and Associated DepressiveSymptoms

Rare pain Daily pain P value

eel sad 5.3% 25.3% �.001eel like crying 1.0% 12.1% �.001eel lonely 6.4% 25.2% �.001eel low energy 4.8% 19.1% �.001

eel life is a failure 3.3% 10.7% �.001

waitodsttaregdm

gooodqbs

1

1

1

1

1

1

1

1

1

1

atJ

Pf

dHtifC(

332 YOUSSEF ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

ell established.1,2,13,15 More than 80% of children with frequentbdominal pain have a comorbid anxiety disorder and approx-mately 40% have depressive symptoms.13 In this current studyhere was a dramatic increase from those patients who sufferednly rarely from abdominal pain versus those with daily ab-ominal pain at risk for depression as defined by the CES-Dcore. Adolescent females with frequent abdominal pain appearo be the target population at highest risk for depression.13 Inhe current study we found a similar prevalence of frequentbdominal pain among females compared with males, with aatio of 2:1. In a recent systematic review of the literaturexploring suicidal behavior and irritable bowel syndrome, Spie-el et al confirmed that chronic abdominal pain is an indepen-ent predictor of suicidal behavior in adults, particularly fe-ales.19

Early recognition of hopelessness and other symptoms sug-estive of depression in adolescents who present for evaluationf frequent abdominal pain is a key finding and offers anpportunity to provide more effective coping strategies to beffered earlier to these patients. An effective strategy for ab-ominal pain early in adolescence may have a positive effect onuality of life and decrease the associated psychosocial comor-idities associated with the chronic abdominal pain syndromeseen in adults.

References

1. Hyams JS, Burke G, Davis PM, et al. Abdominal pain and irritablebowel syndrome in adolescence: a community-based study. J Pe-diatr 1996;129:220–226.

2. Youssef NN, et al. Quality of life for children with functionalabdominal pain: a comparison study of patients’ and parents’perceptions. Pediatrics 2006;117:54–59.

3. Roth-Isigkeit, et al. Pain among children and adolescents: restric-tions in daily living and triggering factors. Pediatrics 2005;115:e152–e162.

4. Campo JV, Comer DM, Jansen-McWilliams L, et al. Recurrentpain, emotional distress, and health service use in childhood.J Pediatr 2002;141:76–83.

5. Smith MT, Edwards RR, Robinson RC, et al. Suicidal ideation,plans, and attempts in chronic pain patients: factors associatedwith increased risk. Pain 2004;111:201–208.

6. Miller V, Hopkins L, Whorwell PJ. Suicidal ideation in patients withirritable bowel syndrome. Clin Gastroenterol Hepatol 2004;3:1064–1068.

7. Rushton JL, Forcier M, Schectman RM. Epidemiology of depres-sive symptoms in the National Longitudinal Study of AdolescentHealth. J Am Acad Child Adolesc Psychiatry 2002;41:199–205.

8. Bearman PS, Jones J, Udry JR. The National Longitudinal Study ofAdolescent Health: research design 1997. Chapel Hill, NC: Caro-lina Population Center, 1997. Available at: http://www.

cpc.unc.edu/addhealth. b

9. Radloff LS. The use of the Center for Epidemiologic StudiesDepression Scale in adolescent and young adults. J Youth Ado-lesc 1991;20:149–166.

0. Roberts RE, Lewinsohn PM, Seeley JR. Screening for adolescentdepression: a comparison of depression scales. J Am Acad ChildAdolesc Psychiatry 1990;30:58–66.

1. Campo JV, et al. Recurrent abdominal pain, anxiety, and depres-sion in primary care. Pediatrics 2004;113:817–824.

2. Ghandour RM, et al. Headache, stomachache, backache, andmorning fatigue among adolescent girls in the United States.Arch Pediatr Adolesc Med 2004;158:797–803.

3. Burke P, Elliott M, Fleissner R. Irritable bowel syndrome andrecurrent abdominal pain: a comparative review. Psychosomatics1999;40:277–281.

4. Little CA, Williams SE, Puzanovova M, et al. Multiple somaticsymptoms linked to positive screen for depression in pediatricpatients with chronic abdominal pain. J Pediatr GastroenterolNutr 2007;44:58–62.

5. Egger HL, Costello EJ, Erkanli A, et al. Somatic complaints andpyschopathology in children and adolescents: stomach aches,musculoskeletal pains and headaches. J Am Acad Child AdolescPsychiatry 1999;38:852–860.

6. Goodman E, Hinden B, Khandelwal S. Accuracy of teen andparental reports of obesity and body mass index. Pediatrics2000;106:52–58.

7. Strauss RS, Pollack HA. Social marginalization of overweightchildren. Arch Pediatr Adolesc Med 2003;157:746–752.

8. Eaton DK, et al. Youth risk behavior surveillance—United States2005. MMWR Surveillance Summaries 2006;55:1–108.

9. Spiegel B, Schoenfeld P, Nabiloff B. Systematic review: the prev-alence of suicidal behaviour in patients with chronic abdominalpain and irritable bowel syndrome. Aliment Pharmacol Ther2007;26:183–193.

Address correspondence to: Nader N. Youssef, MD, Center for Pedi-tric Irritable & Motility Disorders, Goryeb Children’s Hospital at Atlan-ic Health, 100 Madison Avenue, Internal Box 82, Morristown, Newersey.

The Marguerite and Joseph P. Goryeb Endowment to the Center forediatric Irritable Bowel & Motility Disorders provided partial funding

or this study.This research used data from Add Health, a program project

esigned by J. Richard Udry, Peter S. Bearman, and Kathleen Mullanarris, and funded by a grant P01-HD31921 from the National Insti-

ute of Child Health and Human Development, with cooperative fund-ng from 17 other agencies. Persons interested in obtaining data filesrom Add Health should contact Add Health, Carolina Populationenter, 123 West Franklin Street, Chapel Hill, NC 27516-2524addhealth@unc.edu).

Special acknowledgment is owed to Ronald R. Rindfuss and Bar-

ara Entwisle for assistance in the original design.

SC

EPJ

*CoA

Bdfitsigtiirbtdr1svlnccrwwrnwrvto

GspcacMoa

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2008;6:333–338

haring Genetic Test Results in Lynch Syndrome: Communication Withlose and Distant Relatives

LENA M. STOFFEL,*,‡,§ BETH FORD,‡ ROWENA C. MERCADO,‡ DARASHANA PUNGLIA,‡ WENDY KOHLMANN,�

EGGY CONRAD,¶ AMIE BLANCO,¶ KRISTEN M. SHANNON,# MARK POWELL,** STEPHEN B. GRUBER,�

ONATHAN TERDIMAN,¶ DANIEL C. CHUNG,§,#,‡‡ and SAPNA SYNGAL*,‡,§

Division of Gastroenterology, Brigham and Women’s Hospital, Boston, Massachusetts; ‡Division of Population Sciences, **Department of Biostatistics andomputational Biology, Dana-Farber Cancer Institute, Boston, Massachusetts; §Harvard Medical School, Boston, Massachusetts; �Cancer Genetics Clinic, Universityf Michigan Medical Center, Ann Arbor, Michigan; ¶University of California San Francisco Cancer Center, San Francisco, California; and the #Center for Cancer Risk

nalysis, ‡‡Gastroenterology Unit, Massachusetts General Hospital, Boston, Massachusetts

d2a

tpmscierpgeagwttu

tLs

actFHMCfdr

s

ackground & Aims: Clinical genetic testing can helpirect cancer screening for members of Lynch syndromeamilies; however, there is limited information about fam-ly communication of genetic test results. Methods: Aotal of 174 probands who had genetic testing for Lynchyndrome were enrolled through 4 US cancer genetics clin-cs. Subjects were asked whether they had disclosed theirenetic test results to first-, second-, and third-degree rela-ives. Univariate and multivariate analyses were used todentify clinical and demographic factors associated withnforming immediate and extended family of genetic testesults. Results: One hundred seventy-one of 174 pro-ands (98%; 95% confidence interval, 95%–100%) reported

hat they had disclosed their genetic test result to a first-egree relative. Communication of test results to otherelatives occurred significantly less often, with only 109 of62 (67%; 95% confidence interval, 59%–74%) subjects withecond- or third-degree relatives sharing their results. Indi-iduals with a pathogenic mutation were significantly moreikely to inform distant relatives than were subjects with aegative or indeterminate test result (odds ratio, 2.49; 95%onfidence interval, 1.14 –5.40). Probands’ age, sex, and can-er status did not influence communication of genetic testesults. Lack of closeness and concerns that relatives wouldorry or not understand the implications of test resultsere the primary reasons for not sharing genetic test

esults. Conclusions: Most individuals who undergo ge-etic testing for Lynch syndrome share their test resultsith first-degree family members; however, these results

each more distant relatives significantly less often. Inter-entions to improve communication of genetic test resultso members of the extended family are necessary to provideptimal cancer prevention care to at-risk families.

enetic testing plays an increasing role in the care ofpatients at risk for cancer as a result of hereditary cancer

yndromes. Lynch syndrome, also known as hereditary nonpoly-osis colorectal cancer, is the most common hereditary colorectalancer syndrome, and accounts for approximately 3% to 5% ofll diagnosed colorectal cancer cases.1 Genetic testing is clini-ally available for mutations in the DNA mismatch repair genes

LH1, MSH2, and MSH6, which are the most common causesf Lynch syndrome. Because of the increased risk for colorectal

nd extracolonic cancers, individuals at risk for Lynch syn-

rome require a colonoscopy every 1 to 2 years starting at age0 to 25,2 and women should have screening for endometrialnd ovarian cancers or consider prophylactic hysterectomy.2

Identification of a pathogenic mutation through geneticesting confirms the clinical diagnosis of Lynch syndrome androvides an opportunity to stratify cancer risk for other familyembers. Individuals who undergo genetic testing for Lynch

yndrome appear to be more likely to adhere to recommendedancer screening guidelines.3 A recent survey of cancer centersn the United States indicated that the demand for geneticvaluation services for familial cancer syndromes has increasedapidly over the past decade.4 Although genetic testing is ex-ensive, economic analyses have supported the clinical utility ofenetic testing for Lynch syndrome and have shown that costffectiveness increases substantially when the benefits of testingre extended to probands’ family members.5 Studies have sug-ested that most relatives of patients with colorectal cancerould be interested in genetic testing for cancer predisposi-

ion6; however, there are limited data regarding how informa-ion about genetic testing actually is communicated in familiesndergoing molecular evaluation for Lynch syndrome.

The objectives of our study were to examine how geneticesting information is communicated in families at risk forynch syndrome, and to identify factors associated with disclo-ure of genetic test results to close and distant family members.

MethodsWe conducted a cross-sectional questionnaire study

mong individuals with a personal or family history fulfillinglinical criteria for Lynch syndrome. Subjects were recruitedhrough 4 cancer genetics clinics in the United States: Dana-arber Cancer Institute (Boston, MA), Massachusetts Generalospital (Boston, MA), University of Michigan (Ann Arbor,I), and University of California San Francisco (San Francisco,

A). Eligible subjects included individuals whose personal oramily history fulfilled Bethesda Guidelines for Lynch syn-rome.7 All participants were age 18 years or older, and wereequired to read and write English.

Abbreviations used in this paper: FDR, first-degree relative; SDR,econd-degree relative; TDR, third-degree relative.

© 2008 by the AGA Institute1542-3565/08/$34.00

doi:10.1016/j.cgh.2007.12.014

cPcatwcwmtct

rnntdaO(La

ghaaasvlttfw

rbctto

tvrrtstwrFa

tutawiwwmSa

gwmMng

1rltrotwts

tt(scRawifiwLrAcascw

334 STOFFEL ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

Eligible individuals were identified through visits at 1 of 4ancer genetics clinics or through referral by a family member.otential subjects were invited to enroll in the study either at alinical visit or by mail. Individuals approached by mail receivedn initial study packet with an introduction letter and ques-ionnaire, as well as a decline to participate form. Individualsho did not return study materials after 2 follow-up telephone

alls and 2 mailings were considered nonresponders. Subjectsho had undergone genetic testing were enrolled at least 3onths after their genetic test result had been disclosed to

hem. Questionnaire data were scanned and entered into aomputerized database. The study was approved by the insti-utional review board of each participating study site.

A total of 466 eligible individuals were approached for en-ollment. Of these, 270 (58%) completed the study question-aire, 34 (7%) declined to participate, and 158 (34%) wereonresponders. Women and college graduates were more likelyo complete the study questionnaires. There were no significantifferences between other demographic characteristics (such asge and cancer status) of study responders and nonresponders.f the 270 subjects who completed study questionnaires, 174

64%) reported that they had undergone genetic testing forynch syndrome; only these individuals were included in thisnalysis.

MeasuresThe study questionnaires collected standard demo-

raphic data including age, sex, race, ethnicity, marital status,ousehold income, level of education, and type of health insur-nce. Subjects provided details about personal cancer historynd were asked to estimate their own risk for developing cancernd whether they had ever undergone genetic testing for Lynchyndrome. In a detailed family history section, subjects pro-ided information about number of siblings; children; preva-ence of cancers among first-, second-, and third-degree rela-ives; as well as history of genetic testing and specific geneticest results for individuals in their immediate and extendedamily who had undergone genetic testing. A family pedigreeas constructed for each subject.Subjects were asked, “Have you shared your genetic test

esult with any of the following people: mother, father, sisters,rothers, spouse/partner, daughters, sons, aunts/uncles, orousins?” Subjects were asked to choose among reasons whyhey had or had not shared genetic test results with each ofhose family members and were permitted to select more thanne response.

Statistical AnalysisFor subjects who reported having undergone genetic

esting, each participant’s family history and pedigree was re-iewed to determine that they had at least one living first-degreeelative (FDR) and at least 1 living second- or third-degreeelative (SDR/TDR). Subjects who indicated they had sharedheir genetic test result with their mother, father, brothers,isters, or children were classified as having disclosed the resulto an FDR. Subjects who indicated they had shared their resultith uncles, aunts, or cousins were classified as having disclosed

esults to an SDR/TDR. Subjects without at least one livingDR or SDR/TDR were not included in the corresponding

nalysis. t

The potential effects of clinical and demographic factors onhe decision to disclose genetic test results were explored usingnivariate tests of association (Fisher Exact test and Studenttest). Factors that were found to be significant on univariatenalysis or that were believed to have empiric clinical relevanceere included in multivariable logistic regression models to

dentify variables associated with sharing genetic test resultsith FDRs and SDRs/TDRs. Generalized estimating equationsere used to account for potential clustering of results amongembers of the same family. Analyses were performed using

AS software (SAS Institute, Cary, NC). All P values are 2-sidednd a P value of less than .05 was considered significant.

ResultsSubject CharacteristicsMost of the 174 subjects who reported having under-

one genetic testing for Lynch syndrome were women (70%),hite (91%), college graduates (69%), and married (76%). Theean age of the participants was 46.7 years (range, 18 –79 y).ore than half of study participants (61%) had a cancer diag-

osis, and 104 (60%) individuals had a confirmed positiveenetic mutation associated with Lynch syndrome (Table 1).

Disclosure of Genetic Test Results toFirst-Degree Family MembersOverall, 171 of 174 (98%; 95% confidence interval, 95–

00) subjects reported that they had shared their genetic testesult with at least one FDR (Figure 1). Only 5 subjects with aiving parent reported that they had not disclosed their geneticest results to their mother or father, 4 had not disclosed theiresults to a sister, and 2 had not informed a brother. Nearly 90%f subjects with children informed their sons or daughters ofheir test result, and most of the others indicated that theyould wait until their children were older before discussing the

esting. There were no observed differences in rates of disclo-ure to FDRs by probands’ genetic test results or sex.

Disclosure of Genetic Test Results to MoreDistant RelativesOverall, 109 of 162 (67%) subjects with living second- or

hird-degree family members reported that they had sharedheir genetic test result with one or more of these SDRs/TDRsFigure 1). Of the 97 individuals whose genetic test resulthowed a pathogenic mutation (positive test), 73 (75%) dis-losed their test result to a relative beyond their nuclear family.ates of disclosure to SDRs/TDRs were significantly lowermong subjects with indeterminate or true negative results,ith only 24 of 42 (57%) and 12 of 23 (52%), respectively,

ndicating they had shared test results with relatives beyondrst degree (P � .03) (Table 1). In univariate analysis, subjectsho had more relatives diagnosed with cancers associated withynch syndrome appeared more likely to share their genetic testesult with family members beyond FDRs (P � .05) (Table 1).side from the genetic test result, there were no other signifi-ant associations between the disclosure of genetic test result ton SDR/TDR and probands’ sociodemographic characteristicsuch as sex, age, level of education, race, marital status, havinghildren, and personal history of cancer. Similarly, individualsith a higher level of cancer worry, previous history of genetic

esting in the family, or prior evaluation at a high-risk/genetics

T

MS

R

E

M

C

T

C

P

M

H

M

S

E

Na

b

u

March 2008 SHARING GENETIC TEST RESULTS IN LYNCH SYNDROME 335

able 1. Characteristics of the Study Population (N � 174) and Univariate Analysis of Factors Predicting Disclosure ofGenetic Test Results to Any Family Members Beyond First Degree

Characteristics

All subjects who hadgenetic testing

(N � 174)

Subject disclosure to family members

Told beyondFDR

(N � 109)a

Did not tell beyondFDR

(N � 53)a

P valuebFrequency % Frequency % Frequency %

ean age (range) 46.7 18–79 45.96 45.26 .73ex .20Male 52 29.9 28 59.6 19 40.4Female 122 70.1 81 70.4 34 29.6

ace .15White 157 90.8 96 65.3 51 34.7Non-white 16 9.3 13 86.7 2 13.3Unknown/missing 1 —

ducation .72Less than college graduate 53 31.0 34 69.4 15 30.6At least college graduate 118 69.0 73 65.8 38 34.2Unknown/missing 3 — 2 — — —arital status 1.00Married 132 75.9 81 66.9 40 (33.1)Not married 38 21.8 26 68.4 12 (31.6)Unknown/missing 4 2.3 2 — 1 —

ancer diagnosis .50Yes 106 60.9 66 69.5 29 30.5No 68 39.1 43 64.2 24 35.8

est results .03Positive 104 59.8 73 75.3 24 24.7Indeterminate 47 27.0 24 57.1 18 42.9True negative 23 13.2 12 52.2 11 47.8

ancer worry .55Low 58 33.5 35 66.0 18 34.0Moderate 65 37.6 44 72.1 17 27.9High 50 28.9 30 62.5 18 37.5Unknown/missing 1 —

rior genetic testing in family 1.00Yes 128 73.6 82 67.2 40 32.8No 39 22.4 24 68.6 11 31.4Do not know 7 4.0 3 — 2 —utation previously identified in family .48Yes 113 64.9 75 69.4 33 30.6No 61 35.1 34 63.0 20 37.0

istory of Lynch syndrome cancer in 1 ormore relatives

1.00

Yes 155 89.1 98 67.1 48 32.9No 19 10.9 11 68.8 5 31.2ean number of relatives with Lynch

syndrome cancers3.71 (�2.0) 4.04 (�2.1) 3.38 (�1.8) .05

ubjects with children .72Yes 122 70.1 76 68.5 35 31.5No 52 29.9 33 64.7 18 35.3

ver evaluated in a genetics/high-risk clinic .78Yes 154 89.0 98 67.6 47 32.4No 19 11.0 10 62.5 6 37.5Unknown/missing 1 — 1 — — —

OTE. Bolded entries have a significance of P � .05.There were 162 subjects who indicated they have at least 1 SDR or TDR to tell.Comparison between subjects who did and did not tell beyond FDRs, the Fisher exact test was used for all categoric variables; the t test was

sed for all continuous variables.

cTo4

priS5

tffftrod

tfuwagaurbiTieb

u

ntm(mtgn

cgdtgrBgtltrfdfltrc

samffddfmt

Fm

T

P

S

P

M

Nfe

336 STOFFEL ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

linic were not more likely to disclose their result to SDRs/DRs (Table 1). There were no differences in rates of disclosuref genetic test results among subjects enrolled from any of thestudy sites.In multivariate analysis controlling for subjects’ sex and

ersonal and family history of cancer, having a genetic testesult that revealed a pathogenic mutation was the only signif-cant predictor of disclosing the test result to one or moreDRs/TDRs (odds ratio, 2.49; 95% confidence interval, 1.14 –.40) (Table 2).

Reasons for Disclosing or Not DisclosingGenetic Test ResultsSubjects were asked to choose from a list of reasons why

hey had or had not disclosed genetic test results to one or moreamily members (Figure 2). The most frequently cited reasonsor sharing a genetic test result with family members were asollows: (1) to inform them of their risk, (2) to encourageesting, and (3) to obtain emotional support. Only 1 in 10espondents reported that they informed their family membersf their genetic test result because their physician told them too so.

The most frequently cited reasons for not sharing geneticest results were as follows: (1) they were not close to theiramily members, (2) concern that family members would notnderstand the test result, and (3) they did not want relatives toorry. Only 3 individuals listed concerns about confidentialitys a reason why they did not share their test result. None listeduilt about their result as a reason for not sharing informationbout their testing. Among other reasons were wanting to waitntil young children were older before informing them aboutesults of genetic tests, inability to contact specific family mem-ers, and concern that the information would be too distress-

ng. Although 53 of 162 (32.7%) individuals with living SDRs/DRs said they did not share genetic test results beyond their

mmediate family, only 26 (15%) reported that they had delib-rately withheld their genetic test results from any family mem-er.

DiscussionIn a multicenter study of 174 individuals who had

igure 1. Percentage of participants disclosing test results to familyembers, by test result. Error bars represent 95% confidence intervals.

ndergone genetic testing for Lynch syndrome, we found that a

early all (98%) had disclosed their test results to FDRs; andwo thirds had shared these results with more distant family

embers such as cousins, aunts, or uncles. Seventy-three of 9775%) subjects whose testing identified a gene mutation com-

unicated this result to one or more SDRs/TDRs. Having arue positive genetic test result was the only clinical or demo-raphic factor significantly associated with disclosure of a ge-etic test result to relatives beyond the immediate family.

To date, most of the research examining family communi-ation about genetic testing has focused on individuals under-oing evaluation for hereditary breast and ovarian cancer syn-rome. These studies have shown that most patients that areested for BRCA 1 and 2 mutations inform their FDRs of theirenetic test result; however, communication to more distantelatives occurs less frequently.8 In the predominantly femaleRCA 1 and 2 cohorts, patients were more likely to discloseenetic test results to other female family members rather thano male relatives also at risk,8,9 and positive results were moreikely to be shared, as compared with negative or uninformativeest results.8,9 Previous studies on disclosure of genetic testesults in Lynch syndrome families, each of which reported onewer than 40 subjects, also found that genetic test results areisclosed less frequently to at-risk relatives outside the nuclearamily.10 –12 One study suggested that male probands were lessikely than female probands to communicate genetic test resultso family members11 and proposed that male patients mightequire additional counseling resources to ensure appropriateommunication of results.

Our findings, from a much larger Lynch syndrome cohort,how that probands’ sex, race, cancer history, and concernsbout privacy and confidentiality do not appear to influenceost patients’ decisions about sharing genetic test results with

amily; however, the genetic test result is still a significantactor. Most subjects who did not disclose test results said theyid not intentionally withhold this information. The fact thatisclosure was nearly universal among FDRs, but occurred lessrequently among more distant relatives, suggests that subjects

ay not be fully aware of the potential impact of their geneticest result on health care of SDRs/TDRs and/or may encounter

able 2. Multivariate Analysis of Factors PredictingDisclosure to any Family Members BeyondFirst Degree

Characteristic Odds ratio (95% CI) P value

ersonal history of cancerYes 1.23 (0.61–2.51) .57No —

exFemale 1.68 (0.82–3.44) .17Male —

ositive mutation carrierYes 2.49 (1.14–5.40) .02No —ean number of relatives with Lynch

syndrome cancers0.13a .21

OTE. Analysis used generalized estimating equation to control foramily effects, with exchangeable working correlation � �0.02. Boldntry has a significance of P �.05.

Parameter estimate.

ot

apam(garanchaaode

rdpspa

rtfstrr

aamobntfnTiaftbtm

Frsit

March 2008 SHARING GENETIC TEST RESULTS IN LYNCH SYNDROME 337

ther barriers to communicating results to more distant rela-ives.

Why is sharing genetic test results important for other rel-tives in a family with Lynch syndrome? In the case of a trueositive test, identifying a clearly pathogenic gene mutation inproband confirms the diagnosis and allows unaffected familyembers to ascertain their cancer risk through informative

and less expensive) mutation-specific testing. A true positiveenetic test result affects clinical management for FDRs, SDRs,nd TDRs, who, through testing, would learn whether theyequire high-risk cancer screening. By comparison, a true neg-tive genetic test result (which occurs when an individual testsegative for a mutation previously identified in the family)hanges clinical management for only the tested individual andis/her progeny (children/grandchildren), providing reassur-nce that they did not inherit the increased cancer risk. Because

true negative test result does not change management forther relatives who are not direct descendants, lower rates ofisclosure of true negative test results to SDRs/TDRs might bexpected.

In the case of an “indeterminate/uninformative” genetic testesult (which occurs when an individual’s genetic test resultoes not reveal a mutation and there has not been a mutationreviously identified in the family), the diagnosis of Lynchyndrome can be neither confirmed nor rejected and the im-ortance of disclosing this test result is less obvious. However,

igure 2. (A) Reasons for disclosingesults to family members. (B) Rea-ons for not disclosing results to fam-

ly members. *Subjects were permit-ed to select more than 1 option.

n indeterminate/uninformative result still may have clinical n

elevance. Discussion of the genetic testing provides an oppor-unity to share information about Lynch syndrome with otheramily members who may be at risk and may benefit frompecialized cancer surveillance. In our cohort more than half ofhe subjects with an indeterminate/uninformative genetic testesult met Amsterdam Criteria and still would be considered atisk for Lynch syndrome.

Overall, our results show that in most cases informationbout a genetic test result does reach members of the immedi-te and extended family. In 75% of cases in which a pathogenicutation was identified, this information was shared with one

r more SDRs/TDRs. If this information led other family mem-ers to get tested, this would support models of cost effective-ess for genetic testing, suggesting benefits are likely to extendo family members beyond the proband and his/her immediateamily. With regard to individuals with an indeterminate ge-etic test result, our finding that rates of disclosure to SDRs/DRs are significantly lower (57%) may reflect the difficulty in

nterpreting the clinical significance of an uninformative resultnd conveying this information in a way patients and theiramilies can understand. Some prior studies have suggestedhat individuals with an indeterminate genetic test result maye falsely reassured that their cancer risk is now lower becausehere was no mutation identified, and may not feel that infor-

ation about the genetic test result is important to dissemi-

ate.8

odpsbaiorpttbtrwPctgt

jqlmtrnnictqtiFofm

aLcwcesipidoi

rf

1

1

1

1

oSf

uCET

338 STOFFEL ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

One of the major challenges in improving the effectivenessf genetic testing for cancer prevention is to develop ways toistribute this information to family members and health careroviders. At present, most genetic testing is conducted inpecialized centers where patients meet with genetic counselorsefore and after disclosure of their genetic test result. It gener-lly is accepted that health care providers have an obligation tonform probands of implications of the genetic diagnosis forther family members and encourage them to share their testesult with relatives.13 In the United States, privacy laws preventhysicians from disseminating this information without a pa-ient’s consent. However, simply telling patients that they needo inform family members of their genetic test result may note enough. Our findings suggest that even under ideal condi-ions, some patients do not inform relatives of their genetic testesult, in many cases because of concerns their family membersill worry, or will not understand, or because they are not close.roviding patients with a detailed letter describing the impli-ations of the test result and giving them an annotated copy ofhe family tree indicating which family members should receiveenetic testing information may help ensure that this informa-ion is shared with others who may benefit.

There are several limitations to consider in this study. Sub-ects who were willing to spend 30 minutes completing studyuestionnaires may have been more motivated and thus more

ikely to communicate their test result to family members, orore likely to report that they had. We asked subjects whether

hey had shared their genetic test result with specific types ofelatives (sisters, aunts, cousins, and so forth); however, we didot have total counts of each proband’s living relatives, so it wasot possible to calculate the proportion of at-risk relatives

nformed of genetic test results. We did not have any way toonfirm that subjects actually shared genetic testing informa-ion with relatives who were at risk, nor could we assess theuality of the communication or the accuracy of the informa-ion transmitted to family members. Consequently, our find-ngs may overestimate the true rates of genetic test disclosure.inally, our study did not collect information about outcomesf disclosure of genetic testing information to relatives; there-ore we could not ascertain whether clinical care of family

embers changed as a result of genetic testing.Despite these limitations, our report provides useful data

bout patterns of communication about genetic testing inynch syndrome families. Our study of patients from 4 USancer centers shows that genetic test results often are sharedith members of the immediate family, but less often are

ommunicated to more distant at-risk relatives. Because it isxpected that a greater share of genetic testing will move frompecialized cancer centers to physician’s private offices, it ismportant to develop and implement strategies that will helpatients communicate with family members about genetic test-

ng. Specific interventions, such as providing patients withocumentation of their genetic test result and screening rec-mmendations and providing them with strategies for dissem-

nating their test result to at-risk family members, may help to L

emove barriers to family communication and to improve ef-ectiveness of genetic testing for cancer prevention.

References

1. Lynch HT, Watson P, Shaw TG, et al. Clinical impact of moleculargenetic diagnosis, genetic counseling, and management of he-reditary cancer. Part II: hereditary nonpolyposis colorectal carci-noma as a model. Cancer 1999;86(Suppl):2457–2463.

2. Giardiello FM, Brensinger JD, Petersen GM. AGA technical reviewon hereditary colorectal cancer and genetic testing. Gastroenter-ology 2001;121:198–213.

3. Wagner A, van Kessel I, Kriege MG, et al. Long term follow-up ofHNPCC gene mutation carriers: compliance with screening andsatisfaction with counseling and screening procedures. Fam Can-cer 2005;4:295–300.

4. Epplein M, Koon KP, Ramsey SD, et al. Genetic services forfamilial cancer patients: a follow-up survey of National CancerInstitute Cancer Centers. J Clin Oncol 2005;23:4713–4718.

5. Ramsey SD, Clarke L, Etzioni R, et al. Cost-effectiveness ofmicrosatellite instability screening as a method for detectinghereditary nonpolyposis colorectal cancer. Ann Intern Med 2001;135:577–588.

6. Kinney AY, Choi YA, DeVellis B, et al. Interest in genetic testingamong first-degree relatives of colorectal cancer patients. Am JPrev Med 2000;18:249–252.

7. Rodriguez-Bigas MA, Boland CR, Hamilton SR, et al. A NationalCancer Institute Workshop on Hereditary Nonpolyposis ColorectalCancer Syndrome: meeting highlights and Bethesda guidelines.J Natl Cancer Inst 1997;89:1758–1762.

8. Claes E, Evers-Kiebooms G, Boogaerts A, et al. Communicationwith close and distant relatives in the context of genetic testingfor hereditary breast and ovarian cancer in cancer patients. Am JMed Genet A 2003;116:11–19.

9. Patenaude AF, Dorval M, DiGianni LS, et al. Sharing BRCA1/2test results with first-degree relatives: factors predicting whowomen tell. J Clin Oncol 2006;24:700–706.

0. Peterson SK, Watts BG, Koehly LM, et al. How families commu-nicate about HNPCC genetic testing: findings from a qualitativestudy. Am J Med Genet C Semin Med Genet 2003;119:78–86.

1. Gaff CL, Collins V, Symes T, et al. Facilitating family communica-tion about predictive genetic testing: probands’ perceptions.J Genet Couns 2005;14:133–140.

2. Mesters I, Ausems M, Eichhorn S, et al. Informing one’s familyabout genetic testing for hereditary non-polyposis colorectal can-cer (HNPCC): a retrospective exploratory study. Fam Cancer2005;4:163–167.

3. Offit K, Groeger E, Turner S, et al. The “duty to warn” a patient’sfamily members about hereditary disease risks. JAMA 2004;292:1469–1473.

Address requests for reprints to: Elena M. Stoffel, MD, MPH, Divisionf Gastroenterology, Brigham and Women’s Hospital, 75 Francistreet, Boston, Massachusetts 02115. e-mail: estoffel@partners.org;ax: (617) 632-4088.

Supported by an American College of Gastroenterology Junior Fac-lty Award (2004), a GlaxoSmithKline Institute for Digestive Healthlinical Research Award (2004), and K07 NCI CA 120448-01 (all to.M.S.), and K24 NCI CA 113433 (S.S.). Drs Syngal, Chung, anderdiman have consultant/advisory relationships with Myriad Genetic

aboratories, Salt Lake City, Utah.

OA

PP

JJ

*‡

BieCpaabaSbbqliw(sowBtRwS0Caa

Idsitfeieq

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2008;6:339–345

RIGINAL ARTICLES—LIVER, PANCREAS,ND BILIARY TRACT

rognostic Implications of Lactate, Bilirubin, and Etiology in Germanatients With Acute Liver Failure

OHANNES HADEM,* PENELOPE STIEFEL,* MATTHIAS J. BAHR,* HANS L. TILLMANN,‡ KINAN RIFAI,*URGEN KLEMPNAUER,§ HEINER WEDEMEYER,* MICHAEL P. MANNS,* and ANDREA S. SCHNEIDER*

§

Department of Gastroenterology, Hepatology and Endocrinology, and Department of Visceral and Transplant Surgery, Hannover Medical School, Hannover; andMedical Clinic and Policlinic II, University of Leipzig, Leipzig, Germany

snvddatFopadriBwtaCBm(w

tttaetEDp

uuMpa

ackground & Aims: Among the potentially helpfulndicators of poor prognosis in acute liver failure (ALF) aretiology, encephalopathy grade, blood lactate, and King’sollege Criteria (KCC). The accuracy of these parameters inredicting transplantation or death shows significant vari-tion in different countries. Methods: We retrospectivelynalyzed 102 patients with ALF treated at our institutionetween 1996 and 2005. Baseline parameters, simplifiedcute physiology score III (SAPS-III), KCC, Model for End-tage Liver Disease (MELD) score, and a novel score ofilirubin, lactate, and etiology (BiLE score) were comparedetween transplant-free survivors and patients who re-uired liver transplantation or died, by using multivariate

inear regression analysis and receiver operating character-stics (ROC). Results: The most common causes of ALFere indeterminate liver failure (21%), acute hepatitis B

18%), acetaminophen ingestion (16%), and Budd-Chiariyndrome (9%). Transplantation-free survival was 38%, 44%f patients underwent liver transplantation, and 18% diedithout transplantation. Eight-week survival was 77%. TheiLE score was the best predictor of death or need of

ransplantation, with 79% sensitivity and 84% specificity.OC analysis revealed a better performance of BiLE scorehen compared with bilirubin, lactate, MELD score, andAPS-III (area under the curve: 0.87 � 0.04, 0.73 � 0.51,.73 � 0.52, 0.71 � 0.05, and 0.68 � 0.59, respectively).onclusions: The simple, combined BiLE score emerged

s the best predictor of poor outcome in our patient cohortnd should be prospectively evaluated in other populations.

n 1970, Trey and Davidson1 defined acute liver failure (ALF)as an acute deterioration of liver function resulting in the

evelopment of encephalopathy within 8 weeks of the onset ofymptoms in a patient with a previously healthy liver. Accord-ng to the recent position paper of the American Association forhe Study of Liver Diseases, ALF is defined as decline in liverunction of less than 26 weeks’ duration, absence of cirrhosis,vidence of coagulation abnormality (usually indicated by annternational normalized ratio [INR] �1.5), and any degree ofncephalopathy. Patients with Wilson’s disease, vertically ac-

uired HBV, or autoimmune hepatitis might be included in

pite of possible cirrhosis if the disease has only been recog-ized for less than 26 weeks.2 The composition of etiologiesaries among geographic regions, an issue that has been ad-ressed by a number of studies.3 King’s College Hospital Lon-on reported 57% acetaminophen toxicity and 9% viral hepatitismong 1014 patients between 1973 and 1991.4 A large prospec-ive study involving 17 sites in the United States (Acute Liverailure Group) collected data of 308 patients regarding etiol-gy, clinical and laboratory features, and outcome of patientsresenting with ALF. The most common etiologies were acet-minophen overdose (39%), indeterminate (17%), idiosyncraticrug reactions (13%), and viral hepatitis A and B (12%).5 Earlyeports from France indicated viral hepatitis to be the mostmportant cause of ALF in central Europe,6 with acute hepatitis

accounting for up to 45% of cases.7 The diversity of etiologiesas just recently assessed by an international survey of 6 liver

ransplantation centers summarizing 284 ALF cases. Althoughcetaminophen was the dominant etiology in London andopenhagen with 55% and 74% of cases, respectively, hepatitis(including acute exacerbations of chronic hepatitis B) was theost prominent etiology in Hong Kong (74%) and Karachi

38%). In addition, a high proportion of cases in Karachi (35%)ere due to hepatitis E.8

A major problem in the management of ALF is the predic-ion of spontaneous recovery. Of numerous proposed prognos-ic criteria for ALF, the King’s College criteria (KCC) have beenhe most commonly used and most frequently evaluated.9 KCCre probably the best validated tool currently available.10 How-ver, the predictive value of KCC in the U.S. has been ques-ioned11 and has not been sufficiently evaluated in centralurope, where acetaminophen is not the leading cause of ALF.12

epending on etiology, KCC were shown to have a positiveredictive value of 79%– 88% and negative predictive value of

Abbreviations used in this paper: ALF, acute liver failure; AUC, areander the curve; BiLE, bilirubin-lactate etiology; ICU, intensive carenit; INR, international normalized ratio; KCC, King’s College Criteria;ELD, Model for End-Stage Liver Disease; OLT, orthotopic liver trans-lantation; ROC, receiver operating characteristics; SAPS-III, simplifiedcute physiology score III.

© 2008 by the AGA Institute1542-3565/08/$34.00

doi:10.1016/j.cgh.2007.12.039

5ABpasdtrccpgDjp

hi(pdDIc

tplaaf

wO

smdpaaSdpcltobw[itacp

iiAcAmwut(

(mdtptMac

Fa

T

B

Nd

340 HADEM ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

0%– 65%.13 In France, Clichy criteria are widely used to selectLF patients in need of orthotopic liver transplantation (OLT).esides the presence of hepatic encephalopathy, they includeatients’ age and factor V levels.14 Only a few studies havettempted to directly compare the Clichy criteria and KCC,uggesting superiority of the KCC.15,16 In acetaminophen-in-uced ALF, blood lactate levels above 3.0 mmol/L had a sensi-ivity of 76% and specificity of 97% in predicting death,17 butesults diverged in non–acetaminophen-related ALF.18 –22 Ac-ording to the large prospective U.S. multi-center study, en-ephalopathy grade at admission and etiology were the onlyredictors of outcome.5 Recently, Yantorno et al23 have sug-ested a cut-off value above 30 of the Model for End-Stage Liverisease (MELD) score as predictor of poor outcome. The ob-

ectives of our retrospective analysis were to evaluate theserognostic parameters in a central European cohort.

MethodsPatients and Clinical AssessmentWe retrospectively identified 210 patients with acute

epatic dysfunction treated at the intensive care unit of ournstitution between 1996 and 2005. Of those, 102 patientsFigure 1) fulfilled the diagnostic criteria of ALF as initiallyublished by Trey and Davidson1 as well as those recentlyefined by the American Association for the Study of Liveriseases (ie, hepatic encephalopathy, acute-onset increase of

NR �1.5, and absence of signs of chronic liver disease inlinical and ultrasound examination).2

The study was approved by the human research committee ofhe Medical School Hannover, thereby confirming that therotocol conformed to the ethical guidelines of the 1975 Dec-

aration of Helsinki. Informed consent for retrospective datacquisition was obtained from all patients whose current postalddress was available or from patients’ next of kin in case of aatal outcome.

All parameters except the maximum encephalopathy gradeere documented at admission to intensive care unit (ICU).utcome end points included liver transplantation, death, and

igure 1. Outcome of 102 patients included in the retrospective

wnalysis.

urvival without transplantation for at least 8 weeks after ad-ission at ICU. Hepatic encephalopathy was graded on a stan-

ard scale of I–IV as described previously.24 Simplified acutehysiology score (SAPS) III, which is currently being evaluated aspredictive model in critical care patients, was calculated by using

n EXCEL (Microsoft Corp, Redmond, WA) file provided by theAPS-III Outcomes Research Group (www.saps3.org). KCC wereetermined as described.9 The calculation of MELD score waserformed according to the equation: MELD score � [9.57 � loge

reatinine (mg/dL) � 3.78 � loge bilirubin (mg/dL) � 11.20 �oge INR � 6.43].25 In addition to the established predictiveools, we introduced a simple combined bilirubin-lactate-etiol-gy score (or BiLE score) that can be calculated right at theeginning of the ICU stay. The calculation of BiLE score, whichas developed empirically, is BiLE score � (baseline bilirubin

�mol/L]/100) � baseline lactate [mmol/L] � 4 [in case ofndeterminate ALF, Budd-Chiari syndrome, or phenprocoumonoxicity] � 2 [in case of acetaminophen toxicity] � 0 [in case ofny other ALF etiology] (Table 1). These parameters were thenompared between the cohort of transplant-free survivors andatients who needed liver transplantation or died.

Diagnoses were based on accepted diagnostic criteria thatnvolved history, laboratory values, ultrasound imaging, and inndividual cases histopathologic examination. IndeterminateLF was assumed when the above mentioned diagnostic pro-edures including toxicology screens and serology for hepatitis, B, and C, herpes simplex virus, varicella zoster virus, cyto-egalovirus, and Epstein-Barr virus as well as autoantibodiesere inconclusive. Decision to enroll a patient in the highrgency liver transplantation program was made according tohe guidelines of the Eurotransplant International Foundationsee www.eurotransplant.nl for details).

Statistical AnalysisAll calculations were performed with SPSS software

version 13.0; SPSS Inc, Chicago, IL). Results are expressed asedians and ranges unless otherwise stated. Differences in

iscrete variables were tested with the �2 test or the Fisher exactest, wherever appropriate. Continuous variables were com-ared with the t test for normally distributed variables and withhe Mann-Whitney test for non-normally distributed variables.

ultivariate linear regression analysis was performed to evalu-te prognostic value. Receiver operating characteristic (ROC)urves were calculated for the most relevant parameters.

ResultsDemographic Characteristics andClinical DataOf the 102 patients with acute liver failure, 72 (71%)

able 1. BiLE Score

iLE scoreBilirubin (�mol/L)/100 � lactate (mmol/L)�4 (in case of indeterminate ALF, Budd-Chiari syndrome, or

phenprocoumon toxicity)�2 (in case of acetaminophen toxicity)�0 (in case of any other ALF etiology)

OTE. Sensitivity and specificity in predicting liver transplantation oreath were calculated by using a BiLE score cut-off value �6.9.

ere women. The median age of the group was 38 years (range,

1omw(tudeIimpwT

nmicsdtAWa(o

lstF

F

T

AWM

PIWHFSCAAABALMKSBB

NA

March 2008 PROGNOSTIC IMPLICATIONS OF LACTATE, BILIRUBIN, AND ETIOLOGY IN ALF 341

6 –74 years). Exact data on the duration of jaundice beforenset of encephalopathy were available for 54 patients. Theedian interval between jaundice and onset of encephalopathyas 6 days (range, 0 – 40 days) in the overall cohort, 1 day

range, 0 –39 days) in patients surviving without OLT (referredo as OLT-free survival), and 9 days (0 – 40 days) in those whonderwent transplantation or died (referred to as OLT oreath) (P � .01). Median value for individual maximum hepaticncephalopathy grade was III (range, I–IV) for the whole group,I (range, I–IV) in case of OLT-free survival, and III (range, I–IV)n case of OLT or death (P � .001). A serum creatinine of 2

g/dL (177 �mol/L) or greater was present in 6 of 39 (15%)atients with OLT-free survival versus 17 of 63 (27%) patientsith OLT or death (not significant). Details are presented inable 2.

Etiology of Acute Liver FailureThe distribution of etiologies of ALF was heteroge-

eous without predominance of one particular cause. Indeter-inate ALF accounted for 21 patients (21%). Acute hepatitis B

nfection led to 18 cases (18%, 1 patient with hepatitis B/Doinfection), and 16 acetaminophen ingestions (16%) were ob-erved. Another 9 patients (9%) had acute Budd-Chiari syn-rome. Phenprocoumon toxicity and idiosyncratic drug reac-ions were deemed responsible for 7 (7%) and 5 (5%) cases ofLF, respectively. Diagnoses of Amanita phalloides ingestion,ilson’s disease, hepatitis A, ischemic hepatitis (“shock liver”),

nd halothane reaction accounted for 5 (5%), 5 (5%), 4 (4%), 44%), and 3 (3%), respectively. Other causes were seen in 5 (5%)

able 2. Characteristics of Patients at Admission According t

VariableSurvival without

transplantation (n � 39)

ge (y) 37 (16–74)omen, n (%) 23 (59)aximum encephalopathy grade II (I–IV)I (n) 15II (n) 11III (n) 10IV (n) 3

rothrombin time (% of normal) 24 (8–75)NR 3.5 (1.2–7.4)BC (per �L) 9.4 (4.6–23.5)emoglobin (g/dL) 13.6 (8.0–19.9)actor V (% of normal) 29 (4–165)odium (mmol/L) 136 (121–147)reatinine (�mol/L) 69 (33–416)LT (U/L) 2633 (55–16,726)ST (U/L) 2154 (40–60,620)lkaline phosphatase (U/L) 151 (19–333)ilirubin (�mol/L) 103 (10–624)mmonia (�mol/L) 41 (6–185)actate (mmol/L) 2.9 (0.6–13.1)ELD score 28 (13–49)CC fulfilled, n (%) 7 (18)APS-III 54 (39–98)iLE score 4.3 (0–14.7)iLE score �6.9, n (%) 6 (16)

OTE. All variables are expressed as medians (and ranges), unlessbbreviation: WBC, white blood count.

f ALF patients. Among them were one patient with acute w

ymphoblastic leukemia and another one with myelodysplasticyndrome who developed veno-occlusive disease after stem cellransplantation. An illustration of ALF etiologies is given inigure 2.

igure 2. Causes and outcomes of ALF. Percentage of each etiology

tcome

iver transplantationor death (n � 63) P value

95% Confidence intervalof differences

39 (17–74) NS —49 (78) NS —III (I–IV) �.001 —

8 — —11 — —25 — —19 — —

20 (4–48) NS —4.0 (1.5–10.0) .05 —

11.2 (0.2–42.1) .01 0.7–6.412.6 (5.2–17.2) .005 0.5–2.6

20 (4–63) NS —137 (124–153) NS —82 (28–1746) NS —

1062 (12–10,011) .01 249–2681769 (23–8122) .05 —197 (8–597) .04 1–95263 (20–848) �.001 75–220

72 (16–634) .005 16–904.7 (1.2–26.7) .001 1.29–5.1734 (14–62) �.001 3–1035 (58) �.001 —61 (42–103) .004 —9.7 (1.4–30.0) �.001 4.4–8.450 (79) �.001 —

wise stated.

o Ou

L

other

ith regard to the whole study cohort is given on top of each bar.

iopPebtahwdg

aprtb

dr

A(i8ops1aAt

pw(�

6

T

LBMKB

T

AWMPWHCSCAAABALCMKBBRD

NA

342 HADEM ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

Comparison of Clinical and LaboratoryFeatures by EtiologyA comparison of the 3 most common etiologies of ALF

n our cohort, ie, indeterminate ALF, hepatitis B, and acetamin-phen ingestion, is given in Table 3. There is a notably highroportion of women (18 of 21, 86%) in indeterminate ALF.atients with indeterminate ALF also had a higher maximumncephalopathy grade (median III) and higher bilirubin level,ut ALT and C-reactive protein were markedly lower than in theotal cohort, respectively. In contrast, acetaminophen toxicityffected men and women equally and was associated withigher ALT values. Although information on jaundice durationas available for 54 patients only, time interval between jaun-ice and onset of encephalopathy appeared to vary betweenroups.

Outcome of Acute Liver FailureAs shown in Figure 1, overall survival 8 weeks after

dmission at ICU was 77% (79 of 102 patients). Thirty-nineatients (38%) survived without transplantation, and 45 (44%)eceived a liver graft. Five patients (5%) died after transplanta-ion, 5 (5%) died awaiting a graft, and 13 (13%) died withouteing listed for transplantation. In the latter group, causes of

able 4. Predicting Death or Need of Transplantation

Parameter Cut-off value Sensitivity (%) Specificity

actate 3.5 mmol/L 59 66ilirubin 140 �mol/L 67 64ELD score 32 65 69CC 58 82

able 3. Characteristics of Patients at Admission According t

Variable Indeterminate (n � 21)

ge (y) 43 (28–73)omen, n (%) 18 (86)aximum encephalopathy grade III (I–IV)rothrombin time (% of normal) 22 (8–43)BC (per �L) 11.0 (7.2–29.5)emoglobin (g/dL) 12.9 (7.3–17.0)oagulation factor V (% of normal) 21 (7–54)odium (mmol/L) 138 (128–153)reatinine (�mol/L) 72 (28–485)LT (U/L) 1062 (75–7766)ST (U/L) 803 (23–5389)lkaline phosphatase (U/L) 212 (114–597)ilirubin (�mol/L) 310 (73–660)mmonia (�mol/L) 106 (24–634)actate (mmol/L) 4.1 (1.3–24.2)-reactive protein (mg/L) 9 (4–40)ELD score 34 (23–50)CC fulfilled, n (%) 18 (86)iLE score 11.3 (7.1–29.8)iLE score �6.9, n (%) 21 (100)eceived transplant, n (%) 13 (62)ied, n (%) 6 (29)

OTE. All variables are expressed as medians (and ranges), unless obbreviation: WBC, white blood count.

iLE score 6.9 79 84

eath were multiorgan failure, septicemia, malignancies, andespiratory failure, among others.

Eight-week outcome was strongly influenced by the cause ofLF. Survival without transplantation was high in hepatitis A

4 of 4 patients, 100%), A phalloides toxicity (4 of 5, 80%),diosyncratic drug reaction other than phenprocoumon (4 of 5,0%), and acetaminophen ingestion (12 of 16, 75%). Low ratesf survival without OLT were seen in Wilson’s disease (0 of 5atients, 0%), indeterminate ALF (1 of 21, 5%), Budd-Chiariyndrome (1 of 9, 11%), and phenprocoumon toxicity (1 of 7,4%). Transplantation was performed in only 12% of the acet-minophen group as compared with 62% of the indeterminateLF group. Women were more likely to require liver transplan-

ation or die than men (Figure 1).

Predictive Factors for a Fatal Outcome

Several clinical and laboratory parameters as well asrognostic models emerged as discriminators between survivalithout OLT on the one hand and OLT or death on the other

Table 2). Lactate (cutoff �3.5 mmol/L), MELD score (cutoff32), and KCC achieved a sensitivity and specificity of 59% and

6%, 65% and 69%, and 58% and 82%, respectively (Table 4). No

Positive predictive value (%) Negative predictive value (%)

74 4975 5477 5583 55

e 3 Most Common Etiologies

Fulminant hepatitis B (n � 18) Acetaminophen (n � 16)

34 (20–57) 37 (20–52)11 (61) 9 (56)

II (I–IV) II (I–IV)19 (4–39) 24 (8–48)

8.6 (5.4–17.2) 11.6 (5.6–35.3)13.3 (11.6–15.5) 13.0 (6.3–19.9)

28 (4–165) 19 (7–42)137 (123–142) 138 (128–147)65 (3–482) 87 (33–1746)

3345 (666–9792) 3497 (89–8471)1435 (220–8108) 2474 (222–9826)

193 (19–333) 179 (45–331)214 (67–624) 58 (120–193)94 (6–218) 83 (17–496)

4.3 (0.6–21.1) 3.2 (0.9–26.7)22 (4–148) 38 (2–208)33 (20–44) 27 (13–62)2 (11) 4 (25)

6.9 (3.9–23.7) 1.9 (0–25.2)8 (44) 3 (19)9 (50) 2 (13)4 (22) 3 (19)

ise stated.

(%)

o th

therw

89 71

st

aesIbvpasBh(s7wApAmK

mrpsrnps7Tta

copmvctmn

o

FsSps

Fp(a

T

LBMBS

March 2008 PROGNOSTIC IMPLICATIONS OF LACTATE, BILIRUBIN, AND ETIOLOGY IN ALF 343

tatistical difference was observed for sodium, creatinine, pro-hrombin time, and factor V.

Multivariate linear regression analysis revealed that bilirubinnd lactate were the most predictive individual laboratory param-ters in our patients. We attempted to create a prognostic scoreimple enough for bedside use that could be calculated right afterCU admission. Therefore, baseline bilirubin (�mol/L) was dividedy 100, thereby transferring the values of this parameter to thealue range of baseline lactate (mmol/L). To take into account theredictive importance of ALF etiology, the sum of bilirubin/100nd lactate was then modified by adding positive or negativeummands for certain etiologic entities. The novel combinediLE score (calculation formula given in Table 1) achievedighest significance in predicting liver transplantation or death

Figure 3; Table 4). With a cut-off value of 6.9, its sensitivity,pecificity, and positive and negative predicitive values were9%, 84%, 89%, and 71%, respectively. Sensitivity of BiLE scoreas particularly high (100%) in patients with indeterminateLF. In addition, the performance of the BiLE score in com-arison to other parameters was measured by ROC analysis.rea under the curve (AUC) values revealed a better perfor-ance of the BiLE score in comparison with bilirubin, lactate,CC, MELD score, and the SAPS III score (Figure 4, Table 5).

DiscussionALF is estimated to cause approximately 3.5 deaths per

illion people per year in industrialized countries.26 The cur-ently largest prospective study on ALF, which enrolled 308atients in the United States, demonstrated a 43% chance ofurvival without OLT and an overall survival of 67%.5 Our studyepresented a large cohort of patients with acetaminophen andon–acetaminophen-induced ALF in central Europe. Of 102atients, 38% survived without transplantation. The overallurvival 8 weeks after ICU admission in our patient group was7%, slightly higher than in the recently published U.S. study.his might be due to differences between the European (Euro-

ransplant) and U.S. (United Network for Organ Sharing) liver

igure 3. Lactate, MELD score, and BiLE score discriminate betweenurvivors without liver transplantation and liver transplantation or death.urvivors without liver transplantation (white boxes) are compared withatients who required transplantation or died (grey boxes). Box plotshow medians as well as interquartile ranges. Error bars show ranges.

llocation systems (mortality on the waiting list was 5% in our A

ohort [data not shown] versus 10% in the U.S. study).5 Obvi-usly, our data might be subject to referral bias. Analyzingatients from a single center might on the other hand imply aore homogeneous patient cohort as a result of low interindi-

idual variability regarding diagnostic criteria and standard ofare. Finally, we did not include patients with acute exacerba-ions of chronic hepatitis B, although arguably this scenario

ight be difficult to distinguish from and overlap with fulmi-ant hepatitis B–induced ALF.8

Recent studies have shown a remarkable geographic diversityf ALF etiologies.3–5,7–9,14,17 Our data found indeterminate ALF

igure 4. ROC curves for prediction of death or need of liver trans-lantation. (A) Comparison of lactate and bilirubin with BiLE score.

B) Comparison of MELD score and SAPS-III with BiLE score. Abbrevi-tions are shown in Table 2.

able 5. Predicting Death or Need of Transplantation: ROC

Parameter n AUC � SE (95% CI) P value

actate 96 0.73 � 0.52 (0.63–0.83) �.001ilirubin 102 0.73 � 0.51 (0.64–0.83) �.001ELD score 101 0.71 � 0.05 (0.61–0.82) �.001iLE score 101 0.87 � 0.04 (0.80–0.95) �.001APS-III 87 0.68 � 0.59 (0.57–0.79) .005

bbreviations: SE, standard error; 95% CI, 95% confidence interval.

(ACIm(Antwwskslltsswka

mptftaptAp

spGfagp

pwtppsoolsbi

tecpvc

teasitbsatp

trwuleaosplBblcast(oaa

spIsravimwow

344 HADEM ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

21%) and acute hepatitis B (18%) as the most common etiologicLF entities, whereas acetaminophen ingestion and Budd-hiari syndrome accounted for only 16% and 9%, respectively.

n that sense, overall survival rates of greater than 75% are evenore surprising, considering the low survival rate without OLT

17% in the U.S. study, only 5% in our study) in patients withLF of indeterminate origin.5 The high incidence of indetermi-ate ALF in Germany is one of the most interesting results ofhis study. As in the U.S. ALF study, 71% of patients wereomen. But interestingly, women accounted for 86% of patientsith indeterminate ALF. Whether women are innately more

usceptible to ALF or are more commonly exposed to differentinds of drugs remains to be determined.27 Other statisticallyignificant features of indeterminate ALF were the low levels ofiver enzymes, C-reactive protein, the relatively high bilirubinevels, and a higher encephalopathy grade. One can speculatehat a high proportion of cases of indeterminate ALF take aubacute course with late ICU admission, which has beenhown to be associated with poor prognosis in other reports7 asell as in this study. Furthermore, it would be interesting tonow whether infectious complications contribute to enceph-lopathy in these patients.28

The survival rate without OLT of 38% in acute hepatitis Bight be an underestimation of today’s prognosis of these

atients. Although several case reports have suggested that earlyreatment with lamivudine in patients with subfulminant orulminant hepatitis B might reduce the need for life-saving liverransplantation,29 there is no agreement whether antiviral ther-py should be instituted in this setting. In fact, a randomizedlacebo-controlled trial from India did not support lamivudinereatment in acute hepatitis B.30 In our center, patients withLF have been treated with lamivudine since 2000 (ie, 12 of 18atients with acute hepatitis B in the present cohort).

Patients with ALF caused by phenprocoumon toxicity repre-ented a unique feature of this German patient cohort. Phen-rocoumon, the most widely used coumarin anticoagulant inermany, has been associated with hepatitis and subacute liver

ailure (2% and 0.2%, respectively, of phenprocoumon-associ-ted adverse events in Germany).31 The poor outcome in thisroup was probably influenced by contraindications to trans-lantation imposed by the multimorbidity of these patients.

The major aim of our study was to evaluate commonly usedrognostic models in ALF, comparing transplant-free survivorsith patients who underwent transplantation or died. Limita-

ions come from the retrospective data analysis. A number ofatients had to be excluded from the study because of incom-lete documentation of hepatic encephalopathy at ICU admis-ion. Our study was also limited by the fact that treatment andutcome might have changed during the 9-year study period;ne example is the treatment of fulminant hepatitis B with

amivudine as discussed above.29 Finally, defining poor progno-is as death or liver transplantation might have implicated aias related to the decision whether to perform transplantation

n an individual patient.We found that in German ALF patients with a high propor-

ion of indeterminate ALF and acute viral hepatitis maximumncephalopathy grade, bilirubin, lactate, KCC, and MELD scoreould all discriminate between transplant-free survivors andatients who needed liver transplantation or died. Our obser-ation is in line with a recent study on 144 patients with ALF

aused by acute viral hepatitis from India. The authors found

hat the presence of 3 of 6 clinical parameters (among them,ncephalopathy grade and jaundice-encephalopathy interval)nd to a lesser extent also MELD and KCC discriminatedurvivors from nonsurvivors.32 The prognostic value of MELDn acetaminophen-induced liver injury has recently been inves-igated prospectively in 460 patients. MELD was not found toe superior over KCC or INR in predicting death.33 Two othertudies in patients with non-acetaminophen ALF and seroneg-tive hepatitis suggested that KCC might be of similar predic-ive strength as MELD, and that its application results in highatient survival rates.34,35

Multivariate linear regression analysis of our data revealedhat bilirubin and lactate were the most predictive of all pa-ameters evaluated. Although the prognostic value of lactate isell-recognized in acetaminophen-induced ALF,17,36 its clinicaltility in non–acetaminophen-related ALF is less well-estab-

ished because of the small patient number examined andtiologic heterogeneity of this group of ALF patients. Dabos etl20 studied 59 patients with non–acetaminophen-related ALF,f whom 34 underwent liver transplantation or died and hadignificantly higher lactate levels. MacQuillan et al22 presentedrobability curves of nonsurvival on the basis of blood lactate

evels in 27 patients with non–acetaminophen-related ALF.aseline and 8-hour lactate levels did not differ significantlyetween 7 spontaneous survivors and 20 patients who needed

iver transplantation or died. Finally, hyperlactatemia was notonsidered to be of prognostic value in 63 patients with non–cetaminophen-related ALF. However, the interpretation of thistudy by Taura et al21 is complicated by the fact that only liverransplant candidates were included. Because 86 ALF cases84%) of our patient cohort were non–acetaminophen-related,ur study added evidence to the assumption that lactate ismong the best predictors of a fatal outcome even in non–cetaminophen-related ALF.

Bilirubin and lactate in combination with etiology wereubsequently integrated into a novel, empirically developedrognostic score that can be calculated at bedside right afterCU admission. This combined BiLE score achieved the highestensitivity and specificity in predicting poor outcome and wasevealed to be superior over other parameters in the ROCnalysis. The particular advantage of BiLE score might be itsery high sensitivity in predicting death or transplantation inndeterminate ALF. It would therefore be interesting to learn

ore about the implications of this parameter in areas of theorld in which non-acetaminophen ALF is predominant. In ourpinion, BiLE score deserves to be validated in other centersith other patient cohorts.

References

1. Trey C, Davidson CS. The management of fulminant hepaticfailure. In: Popper H, Schaffner F, eds. Progress in liver diseases.New York: Grune and Stratton, 1970:282–298.

2. Polson J, Lee WM. AASLD position paper: the management ofacute liver failure. Hepatology 2005;41:1179–1197.

3. Goldberg E, Chopra S. Fulminant hepatic failure: definition, etiol-ogy, and prognostic indicators. In: Uptodate 2004, version 12.2.Available at: http://www.uptodateonline.com. Accessed on May15, 2004.

4. O’Grady JG, Portmann B, Williams R. Fulminant hepatic failure. In:Schiff L, Schiff R, eds. Diseases of the liver. Philadelphia: Lippin-cott, 1993:852–876.

5. Ostapowicz G, Fontana RJ, Schiødt FV, et al. Results of a pro-

1

1

1

1

1

1

1

1

1

1

2

2

2

2

22

2

2

2

2

3

3

3

3

3

3

3

oH

March 2008 PROGNOSTIC IMPLICATIONS OF LACTATE, BILIRUBIN, AND ETIOLOGY IN ALF 345

spective study of acute liver failure at 17 tertiary care centers inthe United States. Ann Intern Med 2002;137:947–954.

6. Bernuau J, Rueff B, Benhamou JP. Fulminant and subfulminantliver failure: definitions and causes. Semin Liver Dis 1986;6:97–106.

7. O’Grady JG. Acute liver failure. In: O‘Grady JG, Lake JR, HowdlePD, eds. Comprehensive clinical hepatology. London: Mosby,2000:30.1–30.20.

8. Bernal W, Schiødt FV, Hamid S, et al. Etiologies and outcomes foracute liver failure at 6 sites around the world. Dig Dis Sci 2005;24:A318.

9. O’Grady JG, Alexander GJM, Hayllar KM, et al. Early indicators ofprognosis in fulminant hepatic failure. Gastroenterology 1989;97:439–455.

0. Renner EL. How to decide when to list a patient with acute liverfailure for liver transplantation? Clichy or King’s College criteria,or something else? J Hepatol 2007;46:554–557.

1. Lee WM. Acute liver failure in the United States. Semin Liver Dis2003;23:217–226.

2. Escorsell A, Mas A, de la Mata M. Acute liver failure in Spain:analysis of 267 cases. Liver Transplant 2007;13:1389–1395.

3. Anand AC, Nightingale P, Neuberger JM. Early indicators of prog-nosis in fulminant hepatic failure: an assessment of the King’scriteria. J Hepatol 1997;26:62–68.

4. Bismuth H, Samuel D, Castaing D, et al. Orthotopic liver trans-plantation in fulminant and subfulminant hepatitis: the PaulBrousse experience. Ann Surg 1995;222:109–119.

5. Izumi S, Langley PG, Wendon J, et al. Coagulation factor V levelsas a prognostic indicator of fulminant hepatic failure. Hepatology1996;23:1507–1511.

6. Pauwels A, Mostefa-Kara N, Florent C, et al. Emergency livertransplantation for acute liver failure. J Hepatol 1993;17:124–127.

7. Bernal W, Donaldson N, Wyncoll D, et al. Blood lactate as anearly predictor of outcome in paracetamol-induced acute liverfailure: a cohort study. Lancet 2002;359:558–563.

8. Bihari D, Gimson AE, Lindridge J, et al. Lactic acidosis in fulmi-nant hepatic failure. Some aspects of pathogenesis and progno-sis. J Hepatol 1985;1:405–416.

9. Murphy ND, Kodakat SK, Wendon JA, et al. Liver and intestinallactate metabolism in patients with acute hepatic failure under-going liver transplantation. Crit Care Med 2001;29:2111–2118.

0. Dabos KJ, Newsome PN, Parkinson JA, et al. Biochemical prog-nostic markers of outcome in non-paracetamol-induced fulminanthepatic failure. Transplantation 2004;27:200–205.

1. Taura P, Martinez-Palli G, Martinez-Ocon J, et al. Hyperlactatemiain patients with non-acetaminophen-related acute liver failure.World J Gastroenterol 2006;12:1949–1953.

2. MacQuillan GC, Seyam MS, Nightingale P, et al. Blood lactate,but not serum phosphate levels can predict patient outcome in

fulminant hepatic failure. Liver Transpl 2005;9:1073–1079. h

3. Yantorno SET, Trentadue JJ, Ruf AE. The model for end stage liverdisease (MELD): a useful tool to assess prognosis in fulminanthepatic failure. Liver Transl 2004;10:C36.

4. Lee WM. Acute liver failure. N Engl J Med 1993;329:1862–1872.5. Wiesner R, Edwards E, Freeman R, et al. Model for end-stage liver

disease (MELD) and allocation of donor livers. Gastroenterology2003;124:91–96.

6. Hoofnagle JH, Carithers RL, Shapiro C, et al. Fulminant hepaticfailure: summary of a workshop. Hepatology 1995;21:240.

7. Third National Health and Nutrition Examination Survey 1988-1994, NHANES III Second Laboratory Data File (CD-ROM, series11, no 2A), US Department of Health and Human Services,National Center for Health Statistics. Hyattsville, MD. Centers forDisease Control and Prevention, 1998.

8. Vaquero J, Polson J, Chung C, et al. Infection and the progressionof encephalopathy in ALF. Gastroenterology 2003;125:755–764.

9. Tillmann HL, Hadem J, Leifeld L, et al. Safety and efficacy oflamivudine in patients with severe acute or fulminant hepatitis B,a multicenter experience. J Viral Hepat 2006;13:256–263.

0. Kumar M, Satapathy S, Monga R, et al. A randomized controlledtrial of lamivudine to treat acute hepatitis B. Hepatology 2007;45:97–101.

1. Schimanski CC, Burg J, Möhler M, et al. Phenprocoumon-inducedliver disease ranges from mild acute hepatitis to (sub-) acute liverfailure. J Hepatol 2004;41:67–74.

2. Dhiman RK, Jain S, Maheshwari U, et al. Early indicators ofprognosis in fulminant hepatic failure: an assessment of theModel for End-Stage Liver Disease (MELD) and King’s CollegeHospital Criteria. Liver Transpl 2007; Mar 16 [epub ahead ofprint].

3. Schmidt LE, Larsen FS. MELD score as a predictor of liver failureand death in patients with acetaminophen-induced liver injury.Hepatology 2007;45:789–796.

4. Katoonizadeh A, Decaestecker J, Wilmer A, et al. MELD score topredict outcome in adult patients with non-acetaminophen-in-duced acute liver failure. Liver Int 2007;27:329–334.

5. Wigg AJ, Gunson BK, Mutimer DJ. Outcomes following liver trans-plantation for seronegative acute liver failure: experience duringa 12-year period with more than 100 patients. Liver Transpl2005;1:27–34.

6. Schmidt LE, Larsen FS. Prognostic implications of hyperlac-tatemia, muliple organ failure, and systemic inflammatory re-sponse syndrome in patients with acetaminophen-induced acuteliver failure. Crit Care Med 2006;34:337.

Address requests for reprints to: Johannes Hadem, MD, Departmentf Gastroenterology, Hepatology, and Endocrinology, Medical Schoolannover, Carl-Neuberg-Strasse 1, 30625 Hannover, Germany. e-mail:

adem.johannes@mh.hannover.de; fax: �49-511-532-4896.

Bo

BL

*o

BcpaiWmFfvpswowiaetitlpibcfimap

Abnrmmoapa

t

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2008;6:346–352

leeding Stomal Varices: Case Series and Systematic Reviewf the Literature

RET J. SPIER,* ABDULLAH A. FAYYAD,* MICHAEL R. LUCEY,* ERIC A. JOHNSON,* MYRON WOJTOWYCZ,‡

AYTON RIKKERS,§ BRUCE A. HARMS,§ and MARK REICHELDERFER*

Section of Gastroenterology and Hepatology, Department of Medicine, ‡Department of Radiology, and the §Department of Surgery, University of Wisconsin School

f Medicine and Public Health, Madison, Wisconsin

lfwifd

pfpdhcvtd

loHteptcarmaari

aw1

ct

ackground & Aims: Bleeding stomal varices are aommon problem in patients with surgical stomas andortal hypertension, and remain difficult to diagnosend manage. Methods: We identified all patients at ournstitution with bleeding stomal varices from 1989 to 2004.

e surveyed all patients undergoing ileal pouch–anal anasto-osis from 1997 to 2007 for bleeding anastomotic varices.

inally, we performed a systematic review of the literatureocusing on diagnosis and treatment of bleeding stomalarices that included 74 English language studies of 234atients. Results: We identified 8 patients with bleedingtomal varices. Recognition of stomal varices typicallyas delayed, particularly when failing to examine thestomy without the appliance. Stomal variceal bleedingas confirmed by Doppler ultrasound or angiographic

maging. Simple local therapy usually stopped bleeding,lbeit temporarily. Sclerotherapy was effective, but at thexpense of unacceptable stomal damage. Decompressiveherapy was required for secondary prophylaxis, includ-ng transjugular intravascular transhepatic shunts (2 pa-ients), surgical portosystemic shunts (2 patients), andiver transplantation (1 patient). No patient with an ilealouch–anal anastomosis developed anastomotic bleed-

ng from varices. Conclusions: Primary prevention ofleeding stomal varices requires avoidance of creating entero-utaneous stomas in patients with portal hypertension. Care-ul inspection of the uncovered ostomy is essential for bleed-ng stomal varices diagnosis. Once identified, conservative

easures will stop bleeding temporarily with definitive ther-py required, including transjugular intravascular transhe-atic shunts, surgical shunts, or liver transplantation.

lthough varices arising at sites other than the gastroesoph-ageal junction account for only 5% of all variceal bleeding,

leeding stomal varices (BSV) at surgically created mucocuta-eous anastomoses, are a common cause of recurrent hemor-hage in patients with portal hypertension and surgical sto-

as.1 Previous reports of the incidence, natural history, andanagement of BSV have been anecdotal or small series with

nly short-term follow-up periods.2– 68 In this article, we combinereport of our case series of BSV patients and a survey of our

atients who received an ileal pouch–anal anastomosis (IPAA) withsystematic review of the literature on stomal varices.

MethodsWe retrieved the records of all patients with BSV seen at

he University of Wisconsin Hospital and Clinics and the Wil-

iam S. Middleton Memorial Veterans Hospital (Madison, WI)rom 1989 to 2004. Eight patients were identified (Table 1),ith records retrieved from the respective electronic charts after

nstitutional review board approval. If patients were lost toollow-up evaluation, we used the social security database toetermine health status (alive vs deceased) as of June 18, 2007.69

To examine the possibility that IPAA reconstruction (as op-osed to permanent ileostomy) reduces the risk of bleedingrom portal hypertension, we reviewed all IPAA procedureserformed from 1997 to 2007. Charts were reviewed for evi-ence of gastrointestinal hemorrhage in the presence of portalypertension. Surrogate markers for portal hypertension in-luded radiographic imaging revealing cirrhotic-appearing liver,arices, splenomegaly, recanalized umbilical vein, portal venoushrombosis, or clinical history of chronic liver disease becauseirect pressure measurements were not obtained.

Search Strategy for Systematic ReviewRelevant studies were identified by searching the fol-

owing databases on July 3, 2007: Medline, Cochrane Databasef Clinical Trails, and Cumulative Index to Nursing and Alliedealth Literature (Figure 1). The databases were searched using

he following key words or their combinations: stoma, stomal,ctopic, ileostomy, colostomy, bleed, bleeding, hemorrhage,ortal hypertension, cirrhosis, end stage liver disease. In addi-ion, we reviewed reference lists from all included studies andonsulted experts. The review included all English languagebstracts and manuscripts, and was performed by a singleesearcher, with previous less formal searches performed on

ultiple occasions. We included all study designs. Two of theuthors (BJS and MR) independently reviewed the titles andbstracts of available literature for possible inclusion in theeview. A data collection form was used to extract study time,ntervention, treatment, and various outcome measures.

ResultsCase SeriesDemographics. There were 4 men and 4 women with

mean age at onset of bleeding of 54 years (range, 21–74 y); 7ere from the University of Wisconsin Hospital and Clinics andwas from the William S. Middleton Memorial Veterans Hos-

Abbreviations used in this paper: BSV, bleeding stomal varices; CT,omputerized tomography; IPAA, ileal pouch–anal anastomosis; TIPS,ransjugular intrahepatic portosystemic shunting.

© 2008 by the AGA Institute1542-3565/08/$34.00

doi:10.1016/j.cgh.2007.12.047

pwpmTaspcci

Aov81ktstB

wcppfBt

awesasph

t

T

Ahp rative

T

March 2008 STOMAL VARICES 347

ital. Cirrhosis caused portal hypertension in 6 of 8 patients,hereas the other 2 had portal vein thrombosis. The first 3atients in the series had ileostomal varices arising from per-anent ileostomies placed after colectomy for ulcerative colitis.he indications for ileostomy at our institutions then changedfter 1984 when IPAA replaced Brooke ileostomy for recon-truction after total colectomy (Table 1). Overall, 9 stomas werelaced in our 8 patients: 4 patients had ileostomies, 3 hadolostomies, and 1 patient had both a colostomy and an ilealonduit. The mean interval from stomal surgery to first bleed-ng was 8.5 years (range, 1 mo to 23 y).

Clinical presentation and physical examination.ll patients presented with gastrointestinal bleeding, whichften was recurrent and massive. There was a history of stomalariceal bleeding (made retrospectively) before diagnosis in 6 ofpatients with a mean delay in diagnosis of 24 months (range,mo to 13 y; Table 2). The diagnosis of portal hypertension wasnown in 5 patients but not in the other 3 patients. Removal ofhe appliance, a maneuver often omitted during previous epi-odes of bleeding, was necessary to see the full extent of theypical surface changes of hypertensive stomopathy (see later).efore diagnosis, all patients had significant bleeding that often

able 1. Patient Characteristics

Patient (sex) Cause of PVH Type of ostomyAge

w

1 (F) AIH Ileostomy2 (F) PSC Ileostomy3 (F) HCV Ileostomy4 (M) HBC, HCV End colostomy

5 (M) NASH End sigmoid colostomy6 (M) HCC with PVT Loop ileostomy and colostomy7 (F) PVT caused by

pylephlebitisTransverse loop colostomy

8 (M) Cardiac cirrhosis Ileostomy

IH, Autoimmune hepatitis; APR, abdominoperitoneal resection; F, fepatitis C virus; HIV, human immunodeficiency virus; M, male; NASHortal venous hypertension; PVT, portal venous thrombosis; UC, ulce

Figure 1. Selection of the articles for the systematic review. F

as rapid and in one case lead to shock requiring rapid resus-itation: all patients required a blood transfusion (2–17 unitser patient). Spurting or gushing of blood was reported in 6atients. Indeed, the clinical record reported that blood spurtedor a distance of 2 feet in 1 patient after appliance removal.efore the diagnosis of BSV, the treating physicians often at-

ributed bleeding to stomal irritation or granulation tissue.The appearance of the stomas and surrounding skin was

bnormal at the time of diagnosis in all patients; changes thatere obvious once the appliance was removed. Typically, the

nteric side was congested with a raspberry hue with the skinide showing a purple discoloration and dilated veins giving theppearance of protruding blue ridges. A large varix was nevereen on the mucosal surface of the stoma. Bleeding could berovoked by stomal manipulation and was observed to be ofigh pressure in all instances.

Imaging. Doppler ultrasound findings included mul-iple, enlarged (�2 mm) stomal veins with abnormal flow (Ta-

ostomylaced Reason for ostomy

Time with ostomy beforestomal variceal bleed

Total colectomy for UC 23 yTotal colectomy for UC 10 yTotal colectomy for UC 6 yHIV lymphoma with visceral

perforation4 y

APR for rectal cancer 1 moBladder cancer invading rectum 11 yStreptococcus viridans

septicemia with multipleabdominal abscesses

9 y

Total colectomy for diverticularbleeding

5 y

e; HBC, hepatitis B cirrhosis; HCC, hepatocellular carcinoma; HCV,alcoholic steatohepatitis; PSC, primary sclerosing cholangitis; PVH,colitis.

able 2. Duration of Prior Bleeding and Techniques Used toDiagnose BSV

Patient (sex)Bleeding before

definitive diagnosisModality used to diagnose

stomal varices

1 (F) Yes, 1 mo Doppler ultrasound2 (F) No Spurting 6–12 in; raspberry

stoma; Doppler ultrasound3 (F) Yes, 13 y Magnetic resonance angiogram;

Doppler ultrasound4 (M) Yes, 4 mo Doppler ultrasound; spurting

with needle puncture5 (M) Yes, 7 mo Tagged red blood cell scan;

traditional angiogram6 (M) Yes, 1 mo CT angiogram showed PVT;

spurting at home with ostomychange

7 (F) Yes, 2 y CT angiogram and traditionalangiogram

8 (M) No Vigorous venous oozing;Doppler ultrasound

whenas p

32383217

734754

64

emal, non

, female; M, male; PVT, portal venous thrombosis.

bibbonttpoti

lhss

aFtihs

aeabSp

fppsaT

aple

hvssm

aAmotur

(f5vhauspa

spp

Fotaditm(gl

348 SPIER ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

le 2). Computerized tomography (CT) scanning was diagnosticn 2 patients, showing portal hypertension and stomal varices inoth (Figure 2). A tagged red blood cell scan localized activeleeding to the stoma in 1 patient with subsequent percutane-us angiography confirming stomal varices. Magnetic reso-ance angiography confirmed the diagnosis as well as definedhe vascular anatomy in 1 patient being considered for liverransplantation. Upper- and lower-gastrointestinal endoscopicrocedures were performed in all patients (often on multipleccasions), with internal ileal varices seen in none of our pa-ients; gastroesophageal varices were found in 2 patients, lead-ng to the diagnosis of portal hypertension.

Treatment. At least temporary hemostasis was estab-ished by pressure with sandbags, cautery with silver nitrate,aving the patient assume a recumbent position, and localuturing. One patient repeatedly cauterized his stoma with

igure 2. (A) CT angiogram performed 2 years after the first episodesf stomal bleeding in patient 7. The study was performed during selec-ive injection of iodinated contrast medium into her superior mesentericrtery with scans obtained during the portal venous phase. Tortuousilated veins are present at the left-sided ileostomy. (B) Reprojected CT

mage from the same study during the portal venous phase (anteropos-erior view of the abdomen). The patient has chronic occlusion of theain portal vein, immediately superior to the superior mesenteric vein

open arrow). The veins of the stoma (solid arrow) are enlarged, con-ested, and in communication with multiple prominent body wall col-

ateral vessels (curved arrows).

ilver nitrate at home. o

Pharmacotherapy including octreotide for acute bleedingnd propranolol for prevention was tried in 2 patients (Table 3).reehand sclerotherapy of the stoma was performed in 3 pa-ients, using 5% sodium morrhuate in 0.5- to 2-mL amounts pernjection (up to 5 mL total).66 Needle injection often provokedigh-pressure venous bleeding. In 1 patient, injection sites wereelected and marked by ultrasound.66

Sclerotherapy invariably produced significant stomal dam-ge, including stricturing of the stoma, requiring dilation andventually causing retraction of the stoma, poor sealing of theppliance, and fecal leakage. In 1 patient, pain after injectioneyond the squamocolumnar junction required hospitalization.ubsequent injections on the mucosal side caused no pain, butroduced significant stomal damage.

Percutaneous embolization of a bleeding varix was per-ormed in 2 patients in whom local therapies failed. Decom-ressive therapy was required for secondary prophylaxis in 4atients, including transjugular intrahepatic portosystemichunting (TIPS) (2 patients), surgical shunts (2 patients),nd liver transplantation (1 patient previously undergoingIPS).

Survival and outcomes. The mean follow-up periodfter diagnosis was 5 years (range, 1 mo to 25 y). None of ouratients died from bleeding, although 7 died from underlying

iver disease (Table 3).70 Overall survival after the first bleedingpisode ranged from 5 months to 25 years.

Two patients received local therapy only. One patient, whoad inoperable metastatic hepatocellular carcinoma and portalein thrombosis, lived for 5 months with repeated surgicaluturing and self-administration of silver nitrate cautery. Theecond patient refused further therapy and died of undeter-

ined causes after 2 years.Among those treated with sclerotherapy, 1 patient survived

lmost 2 years while undergoing sclerotherapy every 3 months.nother patient survived 4.5 years with the combination ofaintenance propranolol plus sclerotherapy and subcutaneous

ctreotide for acute bleeding. Scheduled sclerotherapy every 1o 2 months controlled bleeding for 3 years in the third patientntil recurrent BSV and sclerotherapy-induced stomal damageequired TIPS.

Definitive therapy was used in 4 patients. The first patientpatient 3) underwent TIPS without further bleeding until liverailure necessitated liver transplantation. This patient survived

years after transplant and 25 years after the initial stomalariceal bleed. The second patient (patient 5) underwent trans-epatic coil embolization of the stomal varices along with TIPS,nd survived 1.5 years after the initial bleed. Two patientsnderwent shunt surgery for portal hypertension, including auccessful end-to-side portocaval shunt (patient 1), which wasatent at 4 years, and a mesocaval shunt (patient 7) that faileds a result of shunt thrombosis.

Anastomotic Bleeding in Ileal Pouch–AnalAnastomosis RecipientsWe identified 489 patients with IPAAs who were ob-

erved for a mean of 6.5 years (range, 1–9.5 y), including 10atients with surrogate markers of portal hypertension (4 hadrimary sclerosing cholangitis). Pouch variceal bleeding did not

ccur in any patient.

2

otpsga

fihpavr

lumt

tmai

ifmtpp

v

dpcD

llde3nbpavfsb

T

PC

E

SL

T

O

March 2008 STOMAL VARICES 349

Systematic ReviewThere were 74 case reports and case series consisting of

34 patients (Table 4).Imaging. Doppler ultrasound, CT, and magnetic res-

nance angiography may identify large variceal collections inhe region of the stoma, aid the diagnosis of cirrhosis and/orortal hypertension, and show patency of the portal venousystem.11,26,27,30,37,58 There were 14 reports of percutaneous an-iography showing stomal varices as well as localization ofctive bleeding.2,8,12,13,15,18,21,22,25,31,46,50,53,56

Treatment and outcome. The systematic review con-rmed that local therapy is effective as initial treatment of acuteemorrhage. Local pressure and recumbent body positioningrovide rapid control of BSV.43,46 Surgical suturing, ligation,nd cautery are useful approaches if the bleeding site can beisualized and have been documented in 27 patients in 16eports.2,4,8,11,14,17,18,26,29,43,45,54,57,62,64,66

Data regarding the use of �-blockers for secondary prophy-axis were limited to 19 patients in 8 reports; �-blockers weresed as monotherapy with subsequent bleeding occurring inost instances.16,30,32,42,43,58,65,67 Octreotide was not reported in

he systematic review.There were 9 reports of sclerotherapy used in 14 pa-

ients.4,17,28,38,40,57,58,64,66 One report described visualization andarking of varices using Doppler ultrasound before sclerother-

py. There were no reports of injury to the stoma by repeatednjection of sclerosant.

There were 20 reports of local surgical approaches, includ-ng stomal revision or relocation, often without long-termollow-up evaluation.2,4,6,14 –16,19,24,27,29,38,43,45,47,52,53,55,57,66,67 The

ore extensive local procedure of mucocutaneous disconnec-ion was documented in 3 reports comprising at least 13atients.6,15,24 Recurrent bleeding was common after theserocedures.

Placement of surgical shunts (including mesocaval, portoca-

able 3. Acute and Chronic Therapy With Complication and S

Patient(sex)

Treatment at firstpresentation Therapy (in chronologic order)

1 (F) Pressure with cautery End-to-side portacaval shunt2 (F) Octreotide drip 50

mcg/hPropranolol 60 mg orally twice a

day; sclerotherapy withultrasound guidance; pressure;self-administered octreotide50 mg as necessary

3 (F) Sclerotherapy withcautery � 2

Surgical revision of ostomy � 2;propranolol 80 mg orally twicea day; TIPS; OLTx

4 (M) Sclerotherapy Sclerotherapy (monthly)

5 (M) Embolization withcoiling

TIPS

6 (M) Pressure and cautery;surgical suturing

Pressure and self-applied cauterywith silver nitrate

7 (F) Embolization withcoiling

Mesocaval shunt

8 (M) Pressure Pressure

LTx, orthotopic liver transplant; PCA, patient-controlled analgesia.

al, and splenorenal shunts) for control of bleeding and portalD

ecompression was documented in 13 reports comprising 21atients.2,8,12,13,21,22,46,49,50,71 In patients with well-compensatedirrhosis (Child–Pugh A or low Model for End-Stage Liverisease), shunting reduced the need for repeated intervention.72

Therapy with coil embolization after direct angiography al-owed for acute control of bleeding, although vessel recana-ization or new stomal variceal formation eventually man-ated further intervention.3,31,34,35,49,56,61 The use of coilmbolization at the time of TIPS placement was reported in

patients, but there was no evidence to suggest this tech-ique is better than TIPS alone.3,26,61 TIPS has controlledleeding successfully from stomal varices in 19 reports com-rising 40 patients.3,4,7,9,10,26,27,30,33,39,41,48,51,59 – 61,64,67,68 For ex-mple, Vidal et al60 reported successful therapy in stomalarices in 7 of 8 patients, 1 of whom required shunt dilationor recurrent bleeding and 1 of whom eventually underwenturgical portacaval shunt; the overall cumulative rate of re-leeding was 23% and 31% at 1 and 2 years, respectively. Inter-

able 4. Our Study Data Totaled With the Data CompiledFrom the Systematic Review

Our study Systematic review Total

atients, n 8 226 234onservative therapySclerotherapy 3 14 17Surgical suturing, ligation,

or cautery1 27 28

�-blocker 2 19 21Recumbent positioning 1 2 3

ndovascular therapyTIPS 2 40 42Coil embolization 2 18 20

urgical shunt 2 21 23iver transplant 1 13 14

al Data

sttreatment complications Alive vs deadSurvival after first

stomal variceal bleed

e Dead 4 yrotherapy caused significantarring/stricturing requiringlation, leading to retractionthe stoma and the bag not

ting

Dead 4 y 7 mo

iance did not fit because oflerotherapy

Dead 25 y 7 mo

with sclerotherapy requiringA

Dead 1 y 10 mo

e Dead 1 y 7 mo

e Dead 5 mo

mbosis of mesocaval shuntth subsequent rebleedingm stomal varices

Alive 5 y 3 mo

e Dead 2 y 4 mo

urviv

Po

NonScle

scdioffit

Applsc

PainPC

Non

Non

Throwifro

Non

eaths caused by bleeding 0 6 6

evt

ts

mdettcsBig3a

4rpvpuppsatr

rDaenoa

pBtFdbdbstmwscd

pi

mttpe

psmf

1

Fv

350 SPIER ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

stingly, there were several reports of rebleeding by ectopicarices after TIPS despite lowering the portosystemic gradiento less than 12 mm Hg.59,60,64

Finally, our experience, along with 8 reports of 13 pa-ients, confirmed that liver transplantation controls BSVuccessfully.4,6,20,35,64 – 66

DiscussionStomal varices, which caused significant morbidity,

ultiple hospitalizations, and blood transfusions, were firstescribed in 1968 in a study of the use of surgical colonxclusion to treat hepatic encephalopathy.45 Subsequently,here were case reports and small series leading to a mispercep-ion that BSV is a rare phenomenon.21,56,73 This orthodoxy washallenged by several reports from the Mayo Clinic; with 1tudy reporting a 27% incidence (17 of 62 patients) ofSV.20,43,65 The Mayo Clinic investigators described the follow-

ng risk factors for stomal varices in primary sclerosing cholan-itis patients with ostomies who developed BSV65: at least stagefibrosis on liver biopsy and splenomegaly, esophageal varices,

nd ascites.Internal anastomoses did not lead to bleeding varices in thepatients from the first Mayo series who underwent IPAA

econstruction after colectomy in the presence of significantortal hypertension (2 had splenomegaly and 1 had esophagealarices). Additional reports from the same institution of 40atients with primary sclerosing cholangitis and IPAA followedp for 21 months revealed no ileal varices or bleeding from theouch.74,75 Our study provides further evidence that primaryrevention of stomal variceal formation is achievable whentomas are avoided in patients with known portal hypertension,nd suggests that the unique venous anatomy of stomas (owingo the mucocutaneous connection) is what leads to the majorisk of variceal bleeding.

BSV may be missed. Visual inspection of the stoma afteremoving the appliance is critical and often is overlooked.oppler ultrasound should be the first imaging study ordered

nd usually will confirm the diagnosis of stomal varices, withvaluation of the liver and portal venous system allowing diag-osis of underlying cirrhosis or portal venous thrombosis. CTr magnetic resonance also will show stomal variceal collectionsnd provide important additional anatomic information.

Local therapies (including local pressure, positioning of theatient, suturing, or cautery) are effective at acute control ofSV and are always the first step. Our study shows that certain

ypes of secondary prophylaxis have unacceptable morbidity.irst, sclerotherapy should be avoided on account of stomalamage and/or recurrent BSV in nearly all patients. Second,ased on our systematic review, we suggest that mucocutaneousisconnection and surgical relocation of the stoma should note used on account of recurrent bleeding and the perioperativeurgical risk. Finally, data from our series, as well as the sys-ematic review, suggest that pharmacotherapy is ineffective as

onotherapy. Although the use of subcutaneous octreotideas not described in the systematic review, our experience

uggests that it may reduce acute hemorrhage and should beonsidered for intermittent use in patients who are not candi-ates for definitive therapy.

TIPS has become the preferred treatment modality in mostatients in need of portal decompression, with surgical shunt-

ng reserved for TIPS failures.72 TIPS is also an effective treat-

ent for BSV. However, there are reports of bleeding even whenhe trans-sinusoidal pressure gradient has been reduced to lesshan 12 mm Hg.59,60,64 These data suggest that the thresholdortal pressure gradient leading to BSV may be lower than forsophageal or gastric varices.

Liver transplantation is appropriate for patients with decom-ensated liver failure and BSV. Based on our data, and ourystematic review of the literature, we have proposed a treat-

ent algorithm for management of patients with bleedingrom stomal varices (Figure 3).

References

1. Kinkhabwala M, Mousavi A, Iyer S, et al. Bleeding ileal varicositydemonstrated by transhepatic portography. AJR Am J Roentgenol1977;129:514–516.

2. Adson MA, Fulton RE. Ileal stoma and portal-hypertension—uncommon site of variceal bleeding. Arch Surg 1977;112:501–504.

3. Ahari HK, Feldman L, Kaufman JA, et al. Vascular and interven-tional case of the day—case 1: peristomal varices. AJR Am JRoentgenol 1999;173:829–832.

4. Alkari B, Shaath NM, El-Dhuwaib Y, et al. Transjugular intrahe-patic porto-systemic shunt and variceal embolisation in the man-agement of bleeding stomal varices. Int J Colorectal Dis 2005;20:457–462.

5. Arcidiaco M, Dicorato G, Pasqualin G, et al. Portal hypertensionand enterostomy: a dangerous combination. Colostomy-inducedvarices as a rare cause of gastrointestinal hemorrhage. Report ofa case and review of the literature. Ann Ital Chir 1999;70:61–65.

6. Beck DE, Fazio VW, Grundfestbroniatowski S. Surgical-manage-ment of bleeding stomal varices. Dis Colon Rectum 1988;31:343–346.

7. Bernstein D, Yrizarry J, Reddy KR, et al. Transjugular intrahepaticportosystemic shunt in the treatment of intermittently bleedingstomal varices. Am J Gastroenterol 1996;91:2237–2238.

8. Cameron AD, Fone DJ. Portal hypertension and bleeding ilealvarices after colectomy and ileostomy for chronic ulcerative coli-tis. Gut 1970;11:755–759.

9. Carrafiello G, Lagana D, Giorgianni A, et al. Bleeding from peris-tomal varices in a cirrhotic patient with ileal conduit: treatmentwith transjugular intrahepatic portocaval shunt (TIPS). Emerg Ra-diol 2007;13:341–343.

0. Chavez DR, Snyder PM, Juravsky LI, et al. Recurrent ileal conduithemorrhage in an elderly cirrhotic man. J Urol 1994;152:951–

igure 3. Suggested algorithm for management of bleeding stomalarices. Epi, epinephrine; MELD, Model for End-Stage Liver Disease.

953.

1

1

1

1

1

1

1

1

1

2

2

2

2

2

2

2

2

2

2

3

3

3

3

3

3

3

3

3

3

4

4

4

4

4

4

4

4

4

4

5

5

5

5

5

5

March 2008 STOMAL VARICES 351

1. Choi JW, Lee CH, Kim KA, et al. Ectopic varices in colonic stoma:MDCT findings. Korean J Radiol 2006;7:297–299.

2. Cooper MJ, Mackie CR, Dhorajiwala J, et al. Hemorrhage fromileal varices after total proctocolectomy. Am J Surg 1981;141:178–179.

3. Crooks KK, Hensle TW, Heney NM, et al. Ileal conduit hemor-rhage secondary to portal hypertension. Urology 1978;12:689–693.

4. Eade MN, Williams JA, Cooke WT. Bleeding from an ileostomycaput medusae. Lancet 1969;2:1166–118.

5. Eckhauser FE, Sonda LP, Strodel WE, et al. Parastomal ilealconduit hemorrhage and portal-hypertension. Ann Surg 1980;192:620–624.

6. Farquharson AL, Bannister JJ, Yates SP. Peristomal varices—lifethreatening or luminal? Ann R Coll Surg Engl 2006;88:W6–W8.

7. Finemore RG. Repeated haemorrhage from a terminal colostomydue to mucocutaneous varices with coexisting hepatic metastaticrectal adenocarcinoma: a case report. Br J Surg 1979;66:806.

8. Firlit RS, Firlit CF, Canning J. Exsanguinating hemorrhage fromurinary ileal conduit in patient with portal hypertension. Urology1978;12:710–711.

9. Foulkes J, Wallace DM. Hemorrhage from stomal varices in anileal conduit. Br J Urol 1975;47:630.

0. Fucini C, Wolff BG, Dozois RR. Bleeding from peristomal varices:perspectives on prevention and treatment. Dis Colon Rectum1991;34:1073–1078.

1. Goldstein MB, Brandt LJ, Bernstein LH, et al. Hemorrhage fromileal varices: a delayed complication after total proctocolectomyin a patient with ulcerative colitis and cirrhosis. Am J Gastroen-terol 1983;78:351–354.

2. Goldstein WZ, Edoga J, Crystal R. Management of colostomalhemorrhage resulting from portal-hypertension. Dis Colon Rec-tum 1980;23:86–90.

3. Graeber GM, Ratner MH, Ackerman NB. Massive hemorrhagefrom ileostomy and colostomy stomas due to mucocutaneousvarices in patients with coexisting cirrhosis. Surgery 1976;79:107–110.

4. Grundfestbroniatowski S, Fazio V. Conservative treatment ofbleeding stomal varices. Arch Surg 1983;118:981–985.

5. Hamlyn AN, Morris JS, Lunzer MR, et al. Portal-hypertension withvarices in unusual sites. Lancet 1974;2:1531–1534.

6. Han SG, Han KJ, Cho HG, et al. A case of successful treatmentof stomal variceal bleeding with transjugular intrahepatic porto-systemic shunt and coil embolization. J Korean Med Sci 2007;22:583–587.

7. Handschin AE, Weber M, Weishaupt D, et al. Contrast-enhancedthree-dimensional magnetic resonance angiography for visualiza-tion of ectopic varices. Dis Colon Rectum 2002;45:1541–1544.

8. Hesterberg R, Stahlknecht CD, Roher HD. Sclerotherapy for mas-sive enterostomy bleeding resulting from portal hypertension. DisColon Rectum 1986;29:275–277.

9. Hollands MJ. Parastomal hemorrhage from an ileal conduit sec-ondary to portal-hypertension. Br J Surg 1982;69:675.

0. Johnson PA, Laurin J. Transjugular portosystemic shunt for treat-ment of bleeding stomal varices. Dig Dis Sci 1997;42:440–442.

1. Kishimoto K, Hara A, Arita T, et al. Stomal varices: treatment bypercutaneous transhepatic coil embolization. Cardiovasc Inter-vent Radiol 1999;22:523–525.

2. Konate A, Oberti F, Aube C, et al. Stomal varices treated by glueembolization. Gastroenterol Clin Biol 2007;31:300–302.

3. Lagier E, Rousseau H, Maquin P, et al. Treatment of bleedingstomal varices using transjugular intrahepatic portosystemicshunt. J Pediatr Gastroenterol Nutr 1994;18:501–503.

4. Lashley DB, Saxon RR, Fuchs E, et al. Bleeding ileal conduitstomal varices: diagnosis and management using transjugulartranshepatic angiography and embolization. Urology 1997;50:

612–614.

5. Macedo TA, Andrews JC, Kamath PS. Ectopic varices in thegastrointestinal tract: short- and long-term outcomes of percuta-neous therapy. Cardiovasc Intervent Radiol 2005;28:178–184.

6. Mallick IH, Pearson HJ. Quiz HQ 36. An exceptional cause ofbleeding from “stoma”. Varices around mucus fistula as a resultof portal hypertension complicating primary biliary cirrhosis.J Gastrointest Liver Dis 2007;16:83, 115.

7. Minami S, Okada K, Matsuo M, et al. Treatment of bleedingstomal varices by balloon-occluded retrograde transvenous oblit-eration. J Gastroenterol 2007;42:91–95.

8. Morgan TR, Feldshon SD, Tripp MR. Recurrent stomal varicealbleeding. Successful treatment using injection sclerotherapy. DisColon Rectum 1986;29:269–270.

9. Morris CS, Najarian KE. Transjugular intrahepatic portosystemicshunt for bleeding stomal varices associated with chronic portalvein occlusion: long-term angiographic, hemodynamic, and clini-cal follow-up. Am J Gastroenterol 2000;95:2966–2968.

0. Mosquera DA, Walker SJ, McFarland JB. Bleeding stomal varicestreated by sclerotherapy. J R Coll Surg Edinb 1988;33:337.

1. Nayar M, Saravanan R, Rowlands PC, et al. TIPSS in the treat-ment of ectopic variceal bleeding. Hepatogastroenterology 2006;53:584–587.

2. Noubibou M, Douala HC, Druez PM, et al. Chronic stomal varicealbleeding after colonic surgery in patients with portal hyperten-sion: efficacy of beta-blocking agents? Eur J Gastroenterol Hepa-tol 2006;18:807–808.

3. Peck JJ, Boyden AM. Exigent ileostomy hemorrhage. A complica-tion of proctocolectomy in patients with chronic ulcerative colitisand primary sclerosing cholangitis. Am J Surg 1985;150:153–158.

4. Poilleux J, Damamme B. Colostomy-induced varices in portalhypertension. Chirurgie 1992;118:252–258.

5. Resnick RH, Ishihara A, Chalmers TC, et al. A controlled trial ofcolon bypass in chronic hepatic encephalopathy. Gastroenterol-ogy 1968;54:1057–1069.

6. Ricci RL, Lee KR, Greenberger NJ. Chronic gastrointestinal-bleed-ing from ileal varices after total proctocolectomy for ulcerative-colitis—correction by mesocaval shunt. Gastroenterology 1980;78:1053–1058.

7. Roberts PL, Martin FM, Schoetz DJ, et al. Bleeding stomal vari-ces—the role of local treatment. Dis Colon Rectum 1990;33:547–549.

8. Ryu RK, Nemcek AA, Chrisman HB, et al. Treatment of stomalvariceal hemorrhage with TIPS: case report and review of theliterature. Cardiovasc Intervent Radiol 2000;23:301–303.

9. Samaraweera RN, Feldman L, Widrich WC, et al. Stomal vari-ces—percutaneous trans-hepatic embolization. Radiology 1989;170:779–782.

0. Scaletscky R, Wright JK, Shaw J, et al. Ileal conduit venousvarices from portal-hypertension as a cause of recurrent, mas-sive hemorrhage—case-report and review of the literature. J Urol1994;151:417–419.

1. Shibata D, Brophy DP, Gordon FD, et al. Transjugular intrahepaticportosystemic shunt for treatment of bleeding ectopic varices withportal hypertension. Dis Colon Rectum 1999;42:1581–1585.

2. Smith S, Wiener ES, Starzl TE, et al. Stoma-related varicealbleeding: an under-recognized complication of biliary atresia.J Pediatr Surg 1988;23:243–245.

3. Stansby G, Meyrick-Thomas J, Lewis AA. Pericolostomy varices.J R Coll Surg Edinb 1990;35:109–110.

4. Thomas DJ, Abercrombie GF. Peri-stomal varices—an unusualcause of bleeding from an ileal conduit. Br J Urol 1993;71:355–356.

5. Thompson J. Caput medusae: peristomal varices. J ET Nurs

1993;20:216–219.

5

5

5

5

6

6

6

6

6

6

6

6

6

6

7

7

7

7

7

7

Ga5

352 SPIER ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

6. Toumeh KK, Girardot JD, Choo IW, et al. Percutaneous transhe-patic embolization as treatment for bleeding ileostomy varices.Cardiovasc Intervent Radiol 1995;18:179–182.

7. Tricas JML, Celaya GZ, Lizoain JLH. Colostomal variceal hemor-rhage in a patient with portal-hypertension. Rev Esp Enferm Dig1991;79:437–438.

8. Van der Wouden EJ, D Westerveld B. Photo quiz—a patient withdiabetes mellitus and recurrent peristomal bleeding—what isyour diagnosis? Portal hypertension. Neth J Med 2006;64:314.

9. Vangeli M, Patch D, Terreni N, et al. Bleeding ectopic varices—treatment with transjugular intrahepatic porto-systemic shunt(TIPS) and embolisation. J Hepatol 2004;41:560–566.

0. Vidal V, Joly L, Perreault P, et al. Usefulness of transjugularintrahepatic portosystemic shunt in the management of bleedingectopic varices in cirrhotic patients. Cardiovasc Intervent Radiol2006;29:216–219.

1. Vuillemin E, Aube C, Rifflet H, et al. Recurrent bleeding stomalvarices treated with transjugular intrahepatic portosystemicshunt and stomal varices embolization. J Radiol 1998;79:690–692.

2. Wang MM, McGrew W, Dunn GD. Variceal bleeding from anileostomy stoma. South Med J 1985;78:733–737.

3. Watkins RM. Variceal hemorrhage from a colostomy due to portal-hypertension secondary to intra-hepatic metastases from rectal-carcinoma. BMJ 1981;282:189–190.

4. Weinberg GD, Matalon TAS, Brunner MC, et al. Bleeding stomalvarices—treatment with a transjugular intrahepatic portosys-temic shunt in 2 pediatric-patients. J Vasc Interv Radiol 1995;6:233–236.

5. Wiesner RH, Larusso NF, Dozois RR, et al. Peristomal varicesafter proctocolectomy in patients with primary sclerosing cholan-gitis. Gastroenterology 1986;90:316–322.

6. Wolfsen HC, Kozarek RA, Bredfeldt JE, et al. The role of endoscopicinjection sclerotherapy in the management of bleeding peristomalvarices. Gastrointest Endosc 1990;36:472–474.

7. Wong RCK, Berg CL. Portal hypertensive stomapathy: a newly

described entity and its successful treatment by placement t

of a transjugular intrahepatic portosystemic shunt. Am JGastroenterol 1997;92:1056–1057.

8. Zimmerman G, Smith DC, Taylor FC, et al. Recurrent urinaryconduit bleeding in a patient with portal-hypertension—manage-ment with a transjugular intrahepatic portosystemic shunt. Urol-ogy 1994;43:748–751.

9. Social Security Death Index Interactive Search. Available athttp://ssdi.rootsweb.com/cgi-bin/ssdi.cgi. Accessed: June 18,2007.

0. Graham DY, Smith JL. The course of patients after varicealhemorrhage. Gastroenterology 1981;80:800–809.

1. Warren WD, Millikan WJ, Henderson JM, et al. 10 years portalhypertensive surgery at Emory—results and new perspectives.Ann Surg 1982;195:530–542.

2. Henderson JM, Boyer TD, Kutner MH, et al. Distal splenorenalshunt versus transjugular intrahepatic portal systematic shuntfor variceal bleeding: a randomized trial. Gastroenterology 2006;130:1643–1651.

3. Conte JV, Arcomano TA, Naficy MA, et al. Treatment of bleedingstomal varices—report of a case and review of the literature. DisColon Rectum 1990;33:308–314.

4. Kartheuser AH, Dozois RR, LaRusso NF, et al. Comparison ofsurgical treatment of ulcerative colitis associated with primarysclerosing cholangitis: ileal pouch-anal anastomosis versusBrooke ileostomy. Mayo Clin Proc 1996;71:748–756.

5. Kartheuser AH, Dozois RR, Wiesner RH, et al. Complications andrisk-factors after ileal pouch-anal anastomosis for ulcerative-co-litis associated with primary sclerosing cholangitis. Ann Surg1993;217:314–320.

Address requests for reprints to: Mark Reichelderfer, MD, Section ofastroenterology and Hepatology, University of Wisconsin Hospitalnd Clinics, 600 Highland Avenue, H6/516-5124, Madison, Wisconsin3792. e-mail: mxr@medicine.wisc.edu; fax: (608) 265-5677.The authors would like to thank Dr David Feldstein for his contribu-

ions to this article.

DC

S

D

BpotecfptmwrefmmBrymt1ktswstffw

Ctc4Mtnfmocqaiv

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2008;6:353–359

ietary Counseling Versus Dietary Supplements for Malnutrition inhronic Pancreatitis: A Randomized Controlled Trial

IDDHARTH SINGH, SHALLU MIDHA, NAMRATA SINGH, YOGENDRA KUMAR JOSHI, and PRAMOD KUMAR GARG

epartment of Gastroenterology and Human Nutrition, All India Institute of Medical Sciences, New Delhi, India

eoafwilhiocHiCoao

ieapatartvtmppshn

psf

Ciccct

ackground & Aims: Up to 50% of patients with chronicancreatitis (CP) are malnourished. There are limited datan the role of dietary intervention in improving the nutri-ional status of such patients. The aim was to compare thefficacy of medium chain triglyceride (MCT)– enrichedommercial dietary supplements with dietary counselingor homemade food in the management of malnutrition inatients with CP. Methods: In a randomized controlled

rial, consecutive undernourished patients with CP (bodyass index [BMI] <18.5 kg/m2) at a tertiary care hospitalere randomized to receive either dietary counseling for

egular homemade food or commercial MCT-enriched di-tary supplements for a period of 3 months to compensateor the dietary calorie deficit. All patients received standard

anagement for CP including pancreatic enzyme supple-ents. Primary outcome measure was improvement inMI. Results: Sixty malnourished patients with CP were

andomized to counseling group (n � 29; mean age, 32 � 10ears; male, 83%) and supplementation group (n � 31;ean age, 28 � 10 years; male, 84%). BMI increased in both

he counseling group and supplementation group (17.2 �.7 vs 18.1 � 1.8 kg/m2, P � .001; 16.7 � 1.6 vs 18.2 � 1.6g/m2, P � .001). There were similar improvements inriceps skinfold thickness, dietary intake, fecal fat, and paincore during a period of 3 months in both groups. Thereas, however, no significant difference between the coun-

eling and supplementation groups with regard to any ofhe outcome measures. Conclusions: Dietary counselingor a balanced homemade diet is as good as commercialood supplements in improving malnutrition in patientsith CP.

hronic pancreatitis (CP) is characterized by pancreaticinflammation and fibrosis, eventually leading to destruc-

ion of pancreatic parenchyma and loss of exocrine and endo-rine function. The most common cause of CP is alcohol in0%– 80% of cases, but it is idiopathic in 20%– 60% of cases.1

alnutrition is a common feature of CP, particularly in pa-ients with alcoholic and idiopathic tropical CP. Although mal-utrition has been thought of as a cause of or contributory

actor in the pathogenesis of CP, others and we have shown thatalnutrition is an effect and not a cause of CP.2,3 The etiology

f malnutrition in these patients is multifactorial. Maldigestionaused by decreased pancreatic exocrine secretion and inade-uate bicarbonate delivery to the duodenum leading to second-ry inactivation of enzymes and bile acids by gastric acid is anmportant cause for malnutrition. Abdominal pain, nausea,

omiting, and postprandial satiety contribute by limiting di-

tary intake.4 Self-imposed dietary restriction caused by the fearf inducing pain also contributes to poor nutrition.5 Physicianslso, as a general habit, advise patients with CP to reduce theirat intake to a minimum.6 In a study, we found that patientsith CP had a selective dietary fat restriction due to food fads

n spite of being adequately supplemented with pancreaticipase and other enzymes and analgesics.2 Patients with alco-olic CP are malnourished as a result of continued alcohol

ntake as an important cause of undernutrition. Developmentf CP-related complications like diabetes, pancreatic pseudo-yst, and pancreatic cancer also lead to nutritional decline.4

ypermetabolic state with increased resting energy expenditures another cause of malnutrition in 30%–50% of patients withP.7 The degree of undernutrition has a negative impact on theutcome of these patients.8,9 Micronutrient deficiency mightlso contribute to decreased antioxidant capacity and increasedxidative stress in these patients.10

Although a lot of emphasis has been laid on treating abdom-nal pain by way of analgesics, pancreatic enzyme supplements,ndoscopic therapy, and surgery, not many studies have lookedt strategies for improving nutrition in these patients beyondancreatic enzyme supplementation. Recently, commerciallyvailable dietary supplements containing hydrolyzed oligopep-ides and medium chain triglycerides (MCT) have been regardeds useful for improving nutrition in patients with CP. These areeadily digestible, well-tolerated preparations. Moreover, pa-ients might be more compliant, thinking it is of medicinalalue. Recent studies have shown that the presence of MCT inhe commercially available food supplements results in only a

inimal stimulation of postprandial CCK release and exocrineancreatic secretion, which might decrease abdominal pain inatients with CP.11 No study has explored the efficacy of theseupplements as compared with simple dietary counseling withomemade food in improving the nutritional status of mal-ourished patients with CP.

We conducted a randomized controlled trial (RCT) to com-are the efficacy of MCT-enriched commercially available foodupplements with dietary counseling for regular homemadeood for the management of malnutrition in patients with CP.

Abbreviations used in this paper: BMI, body mass index; CCK-RF,CK releasing factor; CP, chronic pancreatitis; CHI, creatinine height

ndex; CT, computed tomography; ERCP, endoscopic retrogradeholangiopancreatography; MRCP, magnetic resonance cholangiopan-reatography; MCT, medium chain triglycerides; MUAC, mid upper armircumference; RCT, randomized controlled trial; TSF, triceps skinfoldhickness.

© 2008 by the AGA Institute1542-3565/08/$34.00

doi:10.1016/j.cgh.2007.12.040

ccsposoiiprnsfio�ai(opctfdsb(cac

ohu

e

pcTbucasaa

taqftrr

tat

cbdteKi

aw1ic

smn

twdsas

pqeept4

cgdppatt

gwhtPmdsaP

i

354 SINGH ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

MethodsWe conducted an RCT in our tertiary care academic

enter. Consecutive patients with CP attending the pancreaslinic at our center were included in the study during the periodtarting August 2000 –July 2003. The diagnosis of CP was sus-ected on the basis of suggestive clinical features, ie, recurrentr chronic abdominal pain and/or presence of diabetes and/orteatorrhea. The diagnosis was confirmed if there was evidencef pancreatic calcification and/or ductal changes in the form of

rregularity, dilation, and/or stricture of pancreatic duct onmaging studies that included ultrasonography and/or com-uted tomography (CT) of the abdomen and/or endoscopicetrograde cholangiopancreatography (ERCP)/magnetic reso-ance cholangiopancreatography (MRCP).12 Patients were as-essed for their nutritional status. Patients with malnutritionormed the study group. Patients were considered malnour-shed if their body mass index (BMI) was less than 18.5 kg/m2,r if they had lost significant weight (defined as recent loss of10% of their usual body weight within the last 6 months) asresult of the primary disease.13 The patients with the follow-

ng associated conditions were excluded from the study:1) clinically apparent steatorrhea in the form of large, bulky,ily stools because any improvement in the nutritional status ofatients with steatorrhea would have been attributed to pan-reatic enzyme supplementation and not to dietary interven-ion; (2) cancer of the pancreas; (3) biliary obstruction in theorm of deranged liver function test results and dilated bileuct on ultrasound; (4) patients currently undergoing endo-copic or surgical therapy; (5) patients with uncontrolled dia-etes; (6) patients with acute exacerbation of pancreatitis;7) patients with large pseudocyst (�6 cm in size); (8) patientsurrently consuming alcohol �40 g/day; (9) opioid analgesicddicts; and (10) patients with comorbid conditions likehronic liver disease.

All the patients underwent a battery of tests for the diagnosisf CP and its complications. These tests included the following:ematology and serum biochemistry and imaging, includingltrasound abdomen and CT. If required, MRCP was done.

Nutritional and Dietary AssessmentThe patients underwent a detailed nutritional and di-

tary assessment.Nutritional assessment was assessed by the anthropometric

rofile of patients. Anthropometry included BMI, which wasalculated by using the formula, Weight (in kg)/Height2 (in m).he triceps skinfold thickness (TSF) was measured midwayetween acromion process of scapula and olecranon process bysing skinfold calipers (Harpenden). The mid upper arm cir-umference (MUAC) was measured in the left upper arm, with

nonstretch tape, at the mid-point between the tip of thehoulder and the tip of the elbow (olecranon process and thecromion). Three readings were recorded for each parameter,nd the mean was calculated.

For dietary assessment, a detailed dietary history was ob-ained from each subject at the time of entry into the study by

trained dietitian through an interview by using a food fre-uency questionnaire. The frequency of consumption of rawoodstuff before the onset of disease was elicited. A record of allhe food consumed during the past 24 hours was made with theecall method. Nutrient intake was calculated with the 24-hour

ecall method.14 The result was estimated in accordance with t

he standard Table of Food Composition in India.15 Themounts of proteins, fats, carbohydrates, and the calories forhese were computed.

Protein metabolism was assessed by nitrogen balance andreatinine height index (CHI). Nitrogen balance was calculatedy using the formula: [Nitrogen intake/day � Nitrogen output/ay]. Nitrogen intake was calculated from 24-hour dietary pro-ein intake (total protein intake/6.25), and nitrogen output wasstimated from 24-hour urinary nitrogen excretion by usingjeldahl’s method and an additional 5 mg/kg of nitrogen for

ntegumental and other losses.16,17

CHI is a ratio of the patient’s 24-hour creatinine excretionnd the expected normal creatinine excretion. CHI is calculatedith the following formula: [(Measured urinary creatinine �00)/Ideal urinary creatinine for a given height]. Urinary creat-

nine is an estimate of body muscle mass calculated as urinaryreatinine in grams per 24 hours.

Exocrine and endocrine pancreatic functions were also as-essed. Endocrine function was done by using blood sugar

easurement (fasting and postprandial). Diabetes was diag-osed on the basis of World Health Organization criteria.18

Exocrine function was assessed by measuring fecal chymo-rypsin concentration by spectrophotometric method.19 Stoolsere collected for 24 hours from patients receiving a normaliet; stools were subsequently homogenized, weighed, andtored at �20°C. Fecal fat was measured to quantify fat lossccording to van de Kamer et al.20 Patients were given 1 g/kg fatupplement per day for 3 days before stool collection.

A pain score was devised to assess the severity of abdominalain in patients with CP. It was calculated on the basis of fre-uency of pain (No episode of pain in last 12 months � 0/onepisode per 3–12 months � 1/one episode per 3 months � 2/onepisode per month � 3/one episode per week � 4/two episodeser week or continuous � 6) of CP, and treatment/severity (noreatment � 0/oral analgesics � 2/parenteral analgesics �/hospitalization � 6) of the pain.

RandomizationAll the study patients were randomly assigned, by using

omputer-generated random number list, to either of the 2roups, dietary counseling or dietary supplementation. Ran-om allocation sequence, enrollment, and assigning partici-ants to the 2 groups were done by separate individuals. Thearticipants knew what intervention they were getting, thosedministering the intervention knew what was being adminis-ered, but the person assessing the outcome was blinded to thereatment the patient was receiving.

The daily nutrient requirement of the patients in bothroups was calculated on the basis of Harris Benedict equation,hich takes into account the present weight, age, sex, andeight of the patient, and this value was then multiplied by 1.9o compensate for hypermetabolic state in chronic disease.21

atients in both the groups received pancreatic enzyme supple-entation (4 capsules 3 times a day to be taken at the start of,

uring, and at the end of meals). Each enteric-coated micro-phere capsule contained lipase 8000 USP, amylase 30,000 USP,nd protease 30,000 USP (Digestomen-P; Menarini Raunaqharma Limited, India).

In the dietary counseling group, the patient’s usual dietaryntake was assessed by 24-hour recall and food frequency ques-

ionnaire, and the calorie deficit in the diet was calculated as the

decncpcace

ddmtewd5wcnfbfi

mm

dnc

Ridfg

o

T

AMDE

DPP

a

March 2008 NUTRITIONAL INTERVENTION IN CP 355

ifference between recommended calories and actual intake. Anxpert dietitian counseled and encouraged the patient to in-rease the dietary intake by eating small frequent servings oformal homemade diet including all food groups, that is,ereals, pulses, milk, vegetables, fruit, sugar, and oil. No em-hasis was placed on using any particular type of oil. A diethart was prescribed for the required amount of calories. Thepproximate distribution of calories into the nutrients wasarbohydrates 60%, proteins 10%–15%, and fat 25%–30% of totalnergy intake.

In the dietary supplementation group, the average currentaily intake of calories by the patients was calculated, and theeficit in the calorie intake (required minus actual) was supple-ented by commercially available polymeric formula feeds (Nu-

ren 1.0; Nestle India Ltd, India). One 250-mL serving of suchnteral formula feed provided 250 kcal and 9.5 g proteins alongith vitamins, minerals, choline, taurine, and carnitine. Theistribution of calories into the nutrients was carbohydrates1%, proteins 16%, and fat 33% of total energy intake. The feedas MCT-enriched with 25% of fat as MCT. It contained 50%

asein and 50% whey as the protein source, had a high level ofatural antioxidants, and was lactose-, cholesterol-, and gluten-

ree. To check for compliance of the patients, they were asked toring back empty tins of the supplement consumed at everyollow-up visit. The calculated dietary intake target was similarn both groups.

Outcome MeasuresTo evaluate improvement in anthropometry, improve-

ent in BMI at 3 months was taken as the primary outcome

Figure 1. CONSORT chart.

easure. Secondary outcome measures included TSF, MUAC, P

ietary intake (the total calorie intake of patient in each group),itrogen balance, pancreatic exocrine function (assessed byhange in fecal fat), and pain score.

The study was approved by the Indian Council of Medicalesearch. All the patients were included in the study after an

nformed written consent. The study was conducted in accor-ance with the humane and ethical principles of research setorth in the Helsinki guidelines. The study followed CONSORTuidelines.22

Sample SizeIn the absence of any previous data on this subject and

n the basis of an estimate of around 30 –50 new patients of CP

able 1. Clinical and Imaging Features at Baseline:Counseling Versus Supplementation Groups

VariablesCounseling(n � 29)

Supplementation(n � 31)

ge (y) 32 (10)a 28 (10)a

ale sex (n) 24 26uration of disease (y) 3.3 (2.7)a 4.4 (3.3)a

tiologyAlcoholic (n) 14 11Idiopathic (n) 15 20

iabetes (n) 7 6seudocyst 6 (21%) 5 (16%)ancreatic calcification 16 (55%) 20 (64%)

Mean (standard deviation).

� NS for all variables.

pbC8cpsg(l((ppc

ww

itapaaS

dd(ap

dmi

e

pvpg(tgt(

pespshm

T

HF

S

S

S

SSS

S

S

356 SINGH ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

er year in our hospital, we took a sample size of 60 patients toe recruited during a 3-year period. A total of 201 patients withP were seen in our hospital during the 3-year period. Of these,4 were not malnourished and hence did not meet the inclusionriteria. Of the remaining, 35 patients were excluded because ofresence of biliary obstruction (n � 9), clinically apparentteatorrhea (n � 7), comorbidities like chronic liver disease,astric outlet obstruction, intestinal or pulmonary tuberculosisn � 4), continued heavy alcohol intake (n � 4), presence ofarge pseudocyst (n � 3), uncontrolled diabetes mellitusn � 3), opioid addiction (n � 2), carcinoma of the pancreasn � 1), acute exacerbation of pancreatitis (n � 1), andatients undergoing endoscopic therapy (n � 1). Twenty-twoatients refused to participate. Hence, 60 patients were in-luded in the study (Figure 1, CONSORT chart).

Statistical AnalysisThe distributions of fecal fat and fecal chymotrypsin

ere normalized by log transformation. Intergroup comparisonas done with independent sample t test. Intragroup compar-

able 2. Hematology and Biochemical Profile at Baseline inCounseling Versus Supplementation Groups

Variable Counseling Supplementation P value

emoglobin (g/dL) 13.21 � 0.42 13.52 � 1.23 NSasting sugar(mg/dL)

112.00 � 36.00 114.77 � 72 NS

erum bilirubin(mg/dL)

0.59 � 0.10 0.88 � 0.65 NS

erum totalprotein (g/dL)

7.55 � 0.64 8.14 � 0.41 NS

erum albumin(g/dL)

4.21 � 0.44 4.57 � 0.47 NS

erum AST (IU) 43.30 � 24.93 33.22 � 18.97 NSerum ALT (IU) 42.95 � 29.87 29.28 � 17.21 NSerum alkalinephosphate (IU)

253.21 � 220.54 140.07 � 79.29 NS

erum calcium(mg/dl)

10.14 � 0.42 9.50 � 2.10 NS

tool chymotrypsin(unit/g stool)

3.41 � 2.75 4.42 � 3.04 NS

son at periods of follow-up was studied by using paired sampletest. All parameters were compared with intention-to-treat

nalysis. No ancillary analysis was performed. The data areresented as mean (standard deviation) or median (range) asppropriate. A P value of �.05 was taken as significant. Allnalyses were performed with the SPSS software (SPSS 12.0;PSS Inc, Chicago, IL).

ResultsBaseline Clinical ProfileThe clinical characteristics of patients randomized to

ietary counseling and to dietary supplementation have beenescribed in Table 1. The mean age of the patients was 30 years�10); 83% were men. The etiology of CP was alcoholic in 40%nd idiopathic in 59%. The baseline hematology and biochemicalarameters of patients in both the groups are given in Table 2.

Of the 60 patients studied, 29 were randomized to receiveietary counseling alone, whereas 31 received dietary supple-ents. There was no evidence of any adverse events in either

ntervention group. The flow of patients is given in Figure 1.Effect of intervention on nutritional status. The

ffect of intervention on nutritional status is shown in Figure 2.The 2 groups were not different in their anthropometric

rofile at baseline (Table 3). At the end of 3 months of inter-ention, there was significant improvement in all the anthro-ometric parameters including BMI, MUAC, and TSF in bothroups. BMI increased from 17.2 � 1.7 to 18.1 � 1.8 kg/m2

P � .001) in the dietary counseling group and from 16.7 � 1.6o 18.2 � 1.6 kg/m2 (P � .001) in the dietary supplementationroup. There was, however, no significant difference betweenhe counseling and supplementation groups at 3 months18.1 � 1.8 vs 18.2 � 1.6 kg/m2; P � NS).

There was no significant difference in the parameters ofrotein metabolism at baseline in the 2 groups (Table 4). At thend of 3 months, the dietary supplementation group showedignificant improvement in protein metabolism as assessed by aositive nitrogen balance and a high CHI. This trend was noteen in those receiving dietary counseling. The 2 groups were,owever, still not different in terms of protein metabolism at 3onths.

Figure 2. Effect of intervention on nu-tritional status.

admgcaTd

feg((e2

slcsdm

mcdpan

gpugmww

ptttjashbTawgiChnh

setmas

T

BBMT

Na

b .

T

NCU

Na

b

March 2008 NUTRITIONAL INTERVENTION IN CP 357

Effect of intervention on dietary intake. There wassignificant improvement in the dietary intake of fat, carbohy-rates, and proteins as well as total energy in both groups at 3onths (Figure 3, Table 5). In the dietary supplementation

roup, there were 41.25% and 66.66% increases in the meanalorie and fat intakes, respectively, of which 42.14% of caloriesnd 37.5% of fat were provided by the dietary supplements.here was no difference between the 2 groups in terms ofietary intake at baseline and during subsequent follow up.

Effect of intervention on pancreatic exocrineunction. There was a significant decrease in the fecal fatxcretion, suggesting an improved fat absorption in bothroups at 3 months. Fecal fat decreased from 14.8 to 8.0 g/dP �.007) in the dietary counseling group and from 12.8 to 6.9 g/dP �.001) in the dietary supplementation group. There was, how-ver, no significant difference in the fecal fat excretion between thegroups at baseline or during subsequent follow-up.

Effect of intervention on pain score. The 2 groupshowed significant improvement in the pain score during fol-ow-up. It decreased from 5.6 to 3.3 (P � .001) in the dietaryounseling group and from 5.0 to 3.4 (P � .001) in the dietaryupplementation group. There was, however, no significantifference between the 2 groups at baseline and at the end of 3onths.

DiscussionPatients with CP might experience maldigestion and

alnutrition.4 Chronic inflammation and fibrosis in the glandan destroy exocrine tissue, leading to inadequate delivery ofigestive enzymes to the duodenum in the prandial and post-randial periods and subsequent maldigestion. Maldigestion isugmented by inadequate bicarbonate delivery to the duode-um, with secondary inactivation of enzymes and bile acids by

able 3. Anthropometric Variables at Baseline and 3 Months

Parameter

Counseling

Baseline (n � 29) 3 Mo (n � 25) P v

ody weight (kg) 47.1 � 6.3 50.1 � 7.0 .0MI (kg/m2) 17.2 � 1.7 18.1 � 1.8 .0UAC (cm) 22.2 � 2.4 23.4 � 2.5 .0

SF (mm) 7.7 � 3.6 8.8 � 5.5 .0

OTE. Values shown as mean � standard deviation.Paired t test within the group.t test between supplementation and counseling groups at 3 months

able 4. Parameters of Protein Metabolism at Baseline and

Parameter

Counseling

Baseline (n � 28) 3 Mo (n � 24)

itrogen balance (g/day) 6.5 � 4.8 7.1 � 4.0HI 48.7 � 15.4 54.1 � 24.1rinary creatinine (g/day) 0.7 � 0.2 0.8 � 0.3

OTE. Values shown as mean � standard deviation.Paired t test within the group.

t test between supplementation and counseling groups at 3 months.

astric acid.23 Abdominal pain, sitophobia, nausea, vomiting,ostprandial satiety, and ongoing alcohol abuse might contrib-te to poor oral intake. Gastric dysmotility and mechanicalastric outlet obstruction from fibrosis in the pancreatic headight contribute to malnutrition and clinical decline. Patientsith CP might at times experience profound steatorrhea andeight loss.4

The standard therapy for CP is centered on management ofain. Little emphasis has been paid on adequate nutrition ofhese patients. Aggressive nutrition therapy in CP might havehe ability to change the course of the disease and amelioratehe clinical outcome.24 Nutrition in CP patients has twin ob-ectives; first and foremost is improvement of the malnutrition,nd second is to decrease pain by decreasing pancreatic glandtimulation. Recently, commercially available food supplementsave been actively propagated for their easy digestibility andetter tolerance for patients with pancreatic insufficiency.he homogenized form of these foods might cause pancre-tic stimulation for a shorter period of time as comparedith complex solids.25 They also contain hydrolyzed oli-opeptides and MCT, which have been shown to cause min-mal pancreatic stimulation by way of decreased secretion ofCK.11 The efficacy of these food supplements vis-à-visomemade diet, with regard to their ability to improve theutritional status and symptomatology of patients with CP,as not been studied.

We found in the present RCT that although both dietaryupplementation with commercial food preparations and di-tary counseling improved the nutritional status of the pa-ients, commercially available food supplements were no

ore efficacious than routine dietary counseling for a bal-nced homemade diet in improving the overall nutritionaltatus of patients. The significant improvement in anthro-

unseling Versus Supplementation Groups

Supplementation

Baseline (n � 31) 3 Mo (n � 24) P valuea P valueb

46.0 � 6.4 50.9 � 6.9 .001 .71816.7 � 1.6 18.2 � 1.6 .001 .79321.6 � 2.3 23.4 � 2.0 .001 .9797.1 � 3.6 9.7 � 3.9 .001 .524

nths: Counseling Versus Supplementation Groups

Supplementation

luea Baseline (n � 31) 3 Mo (n � 24) P valuea P valueb

5 5.4 � 3.9 8.8 � 5.5 .003 .2050 43.3 � 13.9 56.3 � 18.9 .007 .7274 0.6 � 0.3 0.8 � 0.3 .001 .925

: Co

alue

01010101

3 Mo

P va

.76

.21

.12

paidf

msfwifFqaegeh

sbseasftdft

shecdtwt8a1

tpscn(pvlm

teciauetnsnT

(pCftag

hi

T

ECPF

Na

b

358 SINGH ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

ometric parameters, dietary intake, pancreatic exocrine test,nd pain scores in both groups was perhaps related to overallmprovement in their dietary intake coupled with the stan-ard management for CP in a systematic manner and regularollow-up.

One of the important reasons for malnutrition in CP is theisconstrued belief of patients, sometimes reinforced by phy-

icians, that fat intake should be restricted to a minimum forear of inducing pain. With simple dietary counseling for aholesome food intake, we were able to alleviate undernutrition

n these patients. Restriction of fat leads to impalatability ofood besides decreasing the overall caloric value of the food.at intake need not be altered in quantity or modified inuality (use of MCTs) because these patients are receivingdequate pancreatic lipase supplementation by way of exog-nous pancreatic enzymes. Moreover, it has been shown thatastric lipase secretion is also increased in CP, and thisnzyme can achieve about 30% of the lipolysis observed inealthy volunteers.26,27

The use of MCT in place of long chain triglycerides has beenhown to reduce postprandial CCK release.28 CCK, in turn, haseen implicated as one of the factors responsible for pancreatictimulation and subsequent pain in patients with CP.11 How-ver, only a single small study of 10 patients with CP has shown

reduction in pain in patients receiving MCT-enriched foodupplements.29 CCK release is dependent on CCK releasingactor (CCK-RF). In the presence of trypsin and other proteases,his releasing factor is easily degraded. However, CP leads toecreased secretion of proteases, leading to loss of the negativeeedback on CCK-RF and elevated CCK levels.30 Supplementa-ion with pancreatic enzymes, including proteases, could pos-

able 5. Dietary Intake at Baseline and 3 Months: Counselin

Parameter

Counseling

Baseline (n � 29) 3 Mo (n � 25) P

nergy (kcal/day) 2188 � 672 2575 � 141arbohydrates (g/day) 346 � 140 377 � 147roteins (g/day) 73 � 24 85 � 26ats (g/day) 52 � 24 67 � 29

OTE. Values are shown as mean � standard deviation.Paired t test within the group.

Figure 3. Effect of intervention on dietary intake.

t test between supplementation and counseling groups at 3 months.

ibly result in destruction of CCK-RF, so that CCK levels are notigh enough to result in pain. Moreover, MCTs also have a lownergy density, are not very palatable, and hence have a poorompliance and might induce side effects such as nausea, ab-ominal pain, and diarrhea.24 Hence, it is usually recommendedhat MCTs be advised only to those patients who do not gaineight adequately in spite of standard management for CP and

hose with persistent steatorrhea. It is generally believed that0% of the patients can be managed by dietary counseling,nalgesics, and pancreatic enzyme supplements, whereas only0%–15% might require oral nutritional supplements.31

The improvement in protein anabolism, as assessed by ni-rogen balance, was better in patients on supplements as com-ared with dietary counseling. However, the difference was notignificant at the end of 3 months between the 2 groups, whichould be due to a small sample size. The reasons for betteritrogen balance in the intervention group could be related to

1) consumption of adequate proteins through nutritional sup-lements, (2) low dietary protein intake due to predominantlyegetarian diet in the counseling group, and (3) better assimi-ation or absorption of oligopeptides present in the supple-

ents.The weight gain observed in both groups could be attributed

o (1) increased calorie, fat, and protein intake; (2) use ofxogenous pancreatic enzyme supplements resulting in de-reased fecal fat; and (3) decreased pain experienced by patientsn both groups, with resultant increase in calorie consumption,ttributable to use of pancreatic enzyme supplements and reg-lar patient follow-up. Although the possibility of Hawthorneffect cannot be excluded in our study, it is less likely becausehere was definite objective improvement in weight gain andutritional status after the intervention in the groups undertudy.32 The Hawthorne/protocol effect is a component of theonspecific benefit from improved routine care within a trial.33

he effect, however, is minimal at best.34

There are several clinical implications of this study.1) Homemade diet, which is cheaper, more palatable, and morehysiologic, improves the nutritional status of patients withP. (2) Patients with CP should be advised to take small,

requent, homemade balanced meals. Dietary counseling by arained dietitian should be made an integral part of the man-gement of these patients. (3) Nutritional supplements are notenerally required; they might be used only in special situations.

Thus, we concluded that dietary counseling for a balancedomemade diet is as good as commercial food supplements in

mproving malnutrition in patients with CP.

rsus Supplementation Groups

Supplementation

ea Baseline (n � 31) 3 Mo (n � 24) P valuea P valueb

2053 � 689 2900 � 195 .004 .182324 � 126 407 � 147 .004 .66469 � 22 97 � 30 .001 .14848 � 18 80 � 39 .829 .195

g Ve

valu

.069

.031

.044

.102

1

1

1

1

1

1

1

1

1

1

2

2

2

2

2

2

2

2

2

2

3

3

3

3

3

Ata

March 2008 NUTRITIONAL INTERVENTION IN CP 359

References

1. Garg PK, Tandon RK. Survey on chronic pancreatitis in the Asia-Pacific region. J Gastroenterol Hepatol 2004;19:998–1004.

2. Joshi YK, Midha S, Singh N, et al. Muscle and visceral proteinsare preserved with a positive nitrogen balance in patients withchronic pancreatitis. Gastroenterology 2005;128(Suppl 2):A471.

3. Abu-Bakare A, Taylor R, Gill GV, et al. Tropical or malnutrition-related diabetes: a real syndrome? Lancet 1986;1:1135–1138.

4. Petersen JM, Forsmark CE. Chronic pancreatitis and maldiges-tion. Semin Gastrointest Dis 2002;13:191–199.

5. Goebell H, Hotz J, Hoffmeister H. Hypercaloric nutrition as aetio-logical factor in chronic pancreatitis. Z Gastroenterol 1980;18:94–97.

6. Vaona B, Armellini F, Bovo P, et al. Food intake of patients withchronic pancreatitis after onset of the disease. Am J Clin Nutr1997;65:851–854.

7. Hebuterne X, Hastier P, Peroux JL, et al. Resting energy expen-diture in patients with alcoholic chronic pancreatitis. Dig Dis Sci1996;41:533–539.

8. Thorsgaard Pedersen N, Nyboe Andersen B, Pedersen G, et al.Chronic pancreatitis in Copenhagen: a retrospective study of 64consecutive patients. Scand J Gastroenterol 1982;17:925–931.

9. Worning H. Chronic pancreatitis: pathogenesis, natural historyand conservative treatment. Clin Gastroenterol 1984;13:871–894.

0. Braganza JM, Schofield D, Snehalatha C, et al. Micronutrientantioxidant status in tropical compared with temperate-zonechronic pancreatitis. Scand J Gastroenterol 1993;28:1098–1104.

1. Shea JC, Bishop MD, Parker EM, et al. An enteral therapy con-taining medium-chain triglycerides and hydrolyzed peptides re-duces postprandial pain associated with chronic pancreatitis.Pancreatology 2003;3:36–40.

2. Tandon RK, Sato N, Garg PK, et al. Chronic pancreatitis: Asia-Pacific consensus report. J Gastroenterol Hepatol 2002;17:508–518.

3. WHO physical status: the use and interpretation of anthropome-try—report of a WHO Expert Consultation. WHO Technical Reportseries number 854, 1995.

4. Thimmayamma BVS. A hand book of schedules and guidelines insocio – economic and diet survey. Hyderabad, India: NationalInstitute of Nutrition, ICMR, 1987.

5. Deosthale YG. Nutritive value of Indian foods: some recent stud-ies. Indian J Med Res 1978;68(Suppl):1–16.

6. Hiller A, Plazin J, Van Slyke DD. A study of conditions for Kjeldahldetermination of nitrogen in proteins: description of methodswith mercury as catalyst, and titrimetric and gasometric measure-ments of the ammonia formed. J Biol Chem 1948;176:1401–1420.

7. Food and Agriculture Organization/World Health Organization.Report of a joint FAO/WHO ad hoc expert committee. WHO Tech-

nical Report series 1974 no. 52, WHO, Geneva. o

8. Jarrett RJ. Impaired glucose tolerance and diabetes: WHO crite-ria. Br Med J (Clin Res Ed) 1981;282:990.

9. Kasper P, Moller G, Wahlefeld A. New spectrophotometric esti-mation of chymotrypsin in stool. Clin Chem 1984;30:1753–1757.

0. Van de Kamer JH, Bokkel-Huinik HB, Weyers HA. Rapid methodfor determination of fat in feces. J Biol Chem 1949;177:347–355.

1. Dickerson RN, Vehe KL, Mullen JL, et al. Resting energy expen-diture in patients with pancreatitis. Crit Care Med 1991;19:484–490.

2. Altman DG, Schulz KF, Moher D, et al. The revised CONSORTstatement for reporting randomized trials: explanation and elab-oration. Ann Intern Med 2001;134:663–694.

3. Thiruvengadam R, DiMagno EP. Inactivation of human lipase byproteases. Am J Physiol 1988;255:G476–G481.

4. Giger U, Stanga Z, DeLegge MH. Management of chronic pancre-atitis. Nutr Clin Pract 2004;19:37–49.

5. Miller LJ, Clain JE, Malagelada JR, et al. Control of human post-prandial pancreatic exocrine secretion: a function of the gas-troduodenal region. Dig Dis Sci 1979;24:150–154.

6. Moreau J, Bouisson M, Balas D, et al. Gastric lipase in alcoholicpancreatitis: comparison of secretive profiles following pentagas-trin stimulation in normal adults and patients with pancreaticinsufficiency. Gastroenterology 1990;99:175–180.

7. Carriere F, Grandval P, Renou C, et al. Quantitative study ofdigestive enzyme secretion and gastrointestinal lipolysis inchronic pancreatitis. Clin Gastroenterol Hepatol 2005;3:28–38.

8. Symersky T, Vu MK, Frolich M, et al. The effect of equicaloricmedium- chain and long-chain triglycerides on pancreas enzymesecretion. Clin Physiol Funct Imaging 2002;22:307–311.

9. Freedman SD. Peptamen: a novel therapy in patients with chronicpancreatitis (abstract). Gastroenterology 1997;112:A267.

0. Spannagel AW, Green GM, Guan D, et al. Purification and char-acterization of a luminal cholecystokinin-releasing factor from ratintestinal secretion. Proc Natl Acad Sci U S A 1996;93:4415–4420.

1. Meier R, Ockenga J, Pertkiewicz M, et al. ESPEN guidelines onenteral nutrition: pancreas. Clin Nutr 2006;25:275–284.

2. Franke RH, Kaul JD. The Hawthorne experiments: first statisticalinterpretation. Am Sociol Rev 1978;43:623–643.

3. Braunholtz DA, Edwards SJ, Lilford RJ. Are randomized clinicaltrials good for us (in the short term)? evidence for a “trial effect”.J Clin Epidemiol 2001;54:217–224.

4. Jones S. Was there a Hawthorne effect? Am J Sociol 1992;98:451–468.

Address requests for reprints to: Pramod Kumar Garg, MD, DM,ssociate Professor, Department of Gastroenterology, All India Insti-

ute of Medical Sciences, New Delhi 110029, India. e-mail: pkgarg@iims.ac.in; fax: �91-11-26588663.The study was supported by a research grant from the Indian Council

f Medical Research, New Delhi.

B

AM

Ga

*L

BgThrdlfNqhncwiIplriCtvdpa

TappaobatiHhtchhgc

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2008;6:360–363

RIEF COMMUNICATIONS

nti-Aging Therapy With Human Growth Hormone Associated Withetastatic Colon Cancer in a Patient With Crohn’s Colitis

IL Y. MELMED,* SHANE M. DEVLIN,* GEORGE VLOTIDES,* DEEPTI DHALL,‡ SORAYA ROSS,* RUN YU,*nd SHLOMO MELMED*

‡

Department of Medicine and Department of Anatomic Pathology, Cedars-Sinai Medical Center, David Geffen School of Medicine, University of California,os Angeles, California

hpgh

irspdnphlsse(c

dHhycin

rdwtndbh

hgsf

ackground & Aims: The nonapproved use of humanrowth hormone (HGH) for anti-aging has been increasing.heoretical concerns for neoplastic potentiation by HGHave been raised, but not proven clinically. Methods: Weeport the case of a 68-year-old man with colonic Crohn’sisease who was found to have aggressive metastatic co-

on cancer. The patient had been receiving HGH therapyor anti-aging purposes for 7 years before presentation.ormal and malignant colonic tissue was examined for

ualitative and quantitative molecular profiles of growthormone (GH) and its signaling molecules, using immu-ohistochemistry and RNA extraction with polymerasehain reaction amplification. Results: Immunoreactivityas more robust in tumor tissue than in normal colon for

nsulin-like growth factor-1 receptor (IGF-1R) but not forGF, GH, or GH receptor. RNA extraction with quantitativeolymerase chain reaction showed that IGF-1R and vascu-

ar endothelial growth factor expression, but not IGF-1, GHeceptor, or suppressor of cytokine signaling-2, were highern tumor than in normal colonic tissue. Conclusions:olorectal cancer development concurrent with administra-

ion of HGH for anti-aging purposes occurred in an indi-idual already at increased risk for colon cancer. This un-erscores the need for further investigation of theroneoplastic potential of GH supplementation fornti-aging.

he use of growth hormone (GH) in adults is approved forpatients with proven pituitary deficiency, or those with

cquired immune deficiency syndrome–associated wasting.1 Ap-roximately 100,000 individuals obtain GH for anti-aging pur-oses annually in the United States without a prescription, andbout 30% of GH prescriptions in the United States are forff-label use such as anti-aging.2,3 Because of the potentialenefits of GH administration on bone density, muscle mass,nd cardiovascular health in GH-deficient individuals, the prac-ice of administering GH to healthy adults has become increas-ngly common as a result of claims of anti-aging properties.

owever, a recent meta-analysis of GH administration inealthy adults showed no important benefit to body composi-ion and that side effects were common.4 Furthermore, con-erns have been raised regarding the carcinogenic potential ofuman growth hormone (HGH).5 Several epidemiologic studiesave reported an association between increased insulin-likerowth factor-1 (IGF-1) serum levels and the risk of colorectal

ancer.6 – 8 Acromegaly, a condition of endogenous GH excess,

as been reported to be associated with increased risk of neo-lasia including colorectal adenomas and carcinoma,9,10 anduidelines for colorectal cancer screening in these individualsave been published.11

Despite the growing off-label use of HGH for anti-aging andts potential carcinogenic risk, there is a paucity of publishedeports attributing its use to the occurrence of cancer in thisetting. We describe a case of metastatic colon cancer in aatient with long-standing Crohn’s disease who was receivingaily anti-aging treatment with HGH for years before the diag-osis of cancer. We performed a molecular assessment for theresence of IGF-1, IGF-1 receptor (IGF-1R), GH, and growthormone receptor (GHR) in malignant and nonmalignant co-

onic tissue. We also performed molecular analysis for suppres-or of cytokine signaling (SOCS)-2, a GH-dependent negativeignaling molecule implicated in the regulation of intestinalpithelial cell growth,12 and vascular endothelial growth factorVEGF), an IGF-1– dependent angiogenic gene that promotesolon cancer growth.13,14

Case ReportA 68-year-old man with long-standing colonic Crohn’s

isease was admitted with hematochezia and abdominal pain.e was receiving prednisone 6 mg/day and mesalamine, andad never taken immunomodulator or biologic therapy. For 7ears before admission, the patient sought therapy at anti-aginglinics, and was maintained with testosterone 200 mg weeklynjection, dehydroepiandrosterone 100 mg/day, and recombi-ant HGH (0.6 mg subcutaneously twice daily).

Colonoscopy identified a 7-cm cecal mass, and laparotomyevealed widely metastatic colon cancer. The patient had un-ergone 2 colonoscopies within 2 years of presentation, both ofhich showed changes consistent with chronic, active colitis in

he right colon, but no definite dysplasia or malignancy; it isot known whether the patient had ever had a colonoscopy forysplasia screening. At the time of admission, his carcinoem-ryonic antigen level was 38 ng/mL (normal, �3 ng/mL) andis IGF-1 level was 245 ng/mL (normal, 36 –237 ng/mL). The

Abbreviations used in this paper: GH, growth hormone; GHR, growthormone receptor; HGH, human growth hormone; IGF-1, insulin-likerowth factor-1; IGF-1R, insulin-like growth factor-1 receptor; SOCS,uppressor of cytokine signaling; VEGF, vascular endothelial growthactor.

© 2008 by the AGA Institute1542-3565/08/$34.00

doi:10.1016/j.cgh.2007.12.017

pPaGmc

nfiubt

nwbbfia3aimMn

FinccfGcfpGtMnr

March 2008 HGH AND COLON CANCER 361

atient died from his illness within 5 months of diagnosis.osthumously, normal and malignant colonic tissue were ex-mined for qualitative and quantitative molecular profiles ofH, GHR, IGF-1, IGF-1R, SOCS-2, and VEGF-A by using im-unohistochemistry and RNA extraction with polymerase

hain reaction amplification.

MethodsHuman Subject’s ApprovalPermission to review medical records and to obtain

ormal and malignant colonic tissue from archived, formalin-xed, paraffin-embedded sections from laparotomy for molec-lar investigation was obtained from the institutional reviewoard at Cedars-Sinai Medical Center (Los Angeles, CA) afterhe patient’s death.

igure 1. Immunofluorescent stain-ng of GH, GHR, IGF-1, and IGF-1R inormal colon mucosa and colon car-inoma. Paraffin sections of normalolon mucosa and colon carcinomarom the patient were stained for GH,HR, IGF-1, or IGF-1R (red), andounterstained with Hoechst 33342or nuclei (blue). (A) Increased cyto-lasmic signal for IGF-1R, but not forH, GHR, or IGF-1, in tumor relative

o normal colon tissue. (B and C)erged images for IGF-1R staining in

ormal and malignant colonic tissue,

espectively.

ImmunohistochemistryParaffin slices (5 �m thick) of resected malignant and

ormal tissue were deparaffinated and rehydrated, and antigensere retrieved by microwave heating. Processed slices werelocked with 5% fetal bovine serum and 0.6% Tween 20 inuffered saline, and incubated with primary antibody at 1:100or 2 hours followed by rhodamine-labeled goat anti-mousemmunoglobulin G or donkey anti-rabbit immunoglobulin Gt 1:200 for 1 hour. Slices were counterstained with Hoechst3342 at 1:5000. Stained slices were preserved in Prolong Goldntifade reagent (Invitrogen, Carlsbad, CA). Antibodies usedncluded rabbit polyclonal anti-GH (Dako, Carpinteria, CA),

ouse monoclonal anti-GHR (MAB263; Abcam, Cambridge,A), rabbit polyclonal anti–IGF-1 (H-70; Santa Cruz Biotech-

ology, Santa Cruz, CA), and rabbit polyclonal anti–IGF-1R

(TJ(

teptvMsaOadA3nGGNsIGGNsVGGB5

ww

a

ncWsti

tewmmshpsw1kvifinIecfG

cwuamelmibttn

362 MELMED ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

C-20; Santa Cruz). Sections were observed with a Nikon EclipseE200 fluorescence microscope (Nikon Corporation, Tokyo,

apan), and pictures were acquired with a SPOT digital cameraDiagnostic Instruments, Inc Sterling Heights, MI).

RNA Extraction and Quantitative-PolymeraseChain ReactionRNA extraction was performed from 10-�m–thick sec-

ions of freshly cut normal and tumor formalin-fixed, paraffin-mbedded tissue, respectively, using the RNeasy formalin-fixed,araffin-embedded kit (Qiagen). Reverse transcription of 1.0 �gotal RNA was performed with M-MLV-Reverse Transcriptase (In-itrogen). Q-PCR was performed with SYBR Green PCR Master

ix (Applied Biosystems, Foster City, CA) in a 20-�L reactionystem, containing 10 �L Master Mix, 500 nmol/L primer (each)nd appropriate DNA aliquots, and analyzed with the Bio-Rad iQ5ptical System Software (Bio-Rad Laboratories, Hercules, CA). To

llow the use of fragmented RNA, the following primers wereesigned specifically to amplify small amplicons: GH (Genebankccession NM_000515): sense 5’-CAGGAGTGTCTTCGCCAACA-’, anti-sense 5’-TCCCCATCAGCGTTTGGAT-3’ (amplicon 101t); GHR (Genebank Accession NM_000163): sense 5’-CAACCA-ATCCACCCATTGC-3’, anti-sense 5’-GAATCCCAGTTAAACT-ACGTTCAG-3’ (amplicon 61 nt); IGF-I (Genebank AccessionM_000618): sense 5’-GCTGGTGGATGCTCTTCAGTT-3’, anti-

ense 5’-CCCTGTGGGCTTGTTGAAAT-3’ (amplicon 64 nt);GF-IR (Genebank Accession NM_000875): sense 5’-ACGTGAA-ATCCGCCATTCT-3’, anti-sense 5’-CCTAGGATGAGGCGAAG-TTT-3’ (amplicon 70 nt); SOCS-2 (Genebank AccessionM_003877): sense 5’-GCAAGGATAAGCGGACAGGTC-3’, anti-

ense 5’-AGAGCGGTTTGGTCAGATAAAGG-3’ (amplicon 72 nt);EGF-A (Genebank Accession NM_001025366): sense 5’-GCTT-CCATTCCCCACTTG-3’, anti-sense 5’-TTTCCTCTTTCTGCTG-TTTCC-3’ (amplicon 97 nt); and �-actin (Genebank AccessionC002409): sense 5’-TTGAATGATGAGCCTTCGTG-3’, anti-sense’-GCCTTCATACATCTCAAGTTGG-3’ (amplicon 67 nt).

Statistical AnalysisResults are expressed as the mean � SD and differences

ere evaluated using the Student t test (statistical significanceas set at P � .05).

ResultsLight histology showed distinct areas of normal colon

nd poorly differentiated adenocarcinoma (not shown). Immu- c

oreactivity was more robust in tumor tissue than in normalolon for IGF-1R, but not for IGF, GH, or GHR (Figure 1).

hen RNA was subjected to Q-PCR, IGF-1R and VEGF expres-ion were shown to be higher in the tumor than in normalissue. However, IGF-1, GHR, and SOCS-2 expression were notncreased (Figure 2).

DiscussionThe attributable risk of cancer in patients receiving GH

herapy has not been established definitively. In a meta-analysisvaluating the reported incidence of neoplasia in individualsith GH deficiency treated with exogenous GH, there was aodest but significant increased incidence of overall cancerortality, colorectal cancer, and Hodgkin’s disease.15 However,

urveillance studies of GH replacement for GH-deficient adultsave not shown enhanced cancer incidence or mortality. Theotential risk of adenomatous polyps and cancer death in theetting of excess GH may be owing to increased serum IGF-1,hich can stimulate cell growth and inhibit apoptosis.16 IGF-Rs are expressed on human colorectal cancer cells and IGF-1 isnown to be a stimulator of colorectal cancer cell growth initro.17 IGF-1R also has been shown to play an important rolen angiogenesis and growth of human colon cancer cells.18 Ourndings of increased IGF-1R and VEGF expression supports theotion that VEGF acts to promote tumor angiogenesis via

GF-1 signaling.13 Despite our inability to detect SOCS-2 inither normal or malignant tissue, SOCS-2 may yet be impli-ated in this patient’s cancer because haplotype insufficiencyor SOCS-2 promotes colonic polyposis in the setting of excessH in mice.19

In this case report, the patient had long-standing Crohn’solitis, a known risk factor for colorectal cancer. The degree tohich exogenous HGH anti-aging treatment may have contrib-ted to the malignant transformation of colonic epithelium orccelerated growth of a pre-existing neoplasm cannot be deter-ined on the basis of our findings. However, 2 colonoscopic

valuations within 2 years of presentation did not detect ma-ignancy or dysplasia, suggestive of an accelerated process of

alignant transformation and/or growth. IGF-1, which wasncreased in this patient’s serum at the time of admission, haseen shown both to inhibit apoptosis and to allow progressionhrough the cell cycle, which could potentiate both malignantransformation as well as accelerated growth of an existingeoplasm.20 Conversely, inflammatory bowel disease has been

Figure 2. Increased expression of IGF-1R andVEGF, but not GHR or IGF-1, in tumor relative tonormal colon tissue as determined by quantita-tive reverse-transcription polymerase chain re-action. **P � .01.

haracterized by GH resistance and relative IGF-1 deficiency,21 a

fiHCcacrf

cfcfieIrittic

ivmitp

1

1

1

1

1

1

1

1

1

1

2

2

2

2

2

2

2

Mf

Dt0

March 2008 HGH AND COLON CANCER 363

nding that has prompted clinical investigations for the use ofGH and other growth factors for the treatment of bothrohn’s disease and ulcerative colitis. A randomized, placebo-

ontrolled study showed significant short-term improvement indults with Crohn’s disease treated with HGH.22 However, con-erns about malignant transformation of chronically inflamed,apidly dividing colonic epithelial tissue in response to growthactors have been raised.23

Our findings of increased tumor tissue IGF-1R expression asompared with normal colon lends support to an etiologic roleor the GH/IGF-1 axis in the development and progression ofolon cancer, as has been described previously.24,25 However, ourndings do not discriminate between accelerated growth influ-nced by increased local GH signaling or increased circulatingGF-1. Furthermore, the molecular basis of colitis-related colo-ectal cancer is thought to involve multiple genetic alterations,ncluding mutations in p53, APC, and k-ras.26 The assessment ofhese markers and description of the complex interplay betweenhese molecular alterations and endocrine and paracrine signalss important to understanding the molecular pathogenesis ofancer in this patient, but beyond the scope of this report.

Colorectal cancer development concurrently with the admin-stration of HGH for anti-aging purposes occurred in an indi-idual already at increased risk for colon cancer with inflam-atory bowel disease. This underscores the need for further

nvestigation of the proneoplastic potential of GH supplemen-ation in individuals for anti-aging as its use for this nonap-roved indication becomes more widespread.

References

1. Gharib H, Cook DM, Saenger PH, et al. American Association ofClinical Endocrinologists medical guidelines for clinical practicefor growth hormone use in adults and children—2003 update.Endocr Pract 2003;9:64–76.

2. Perls TT, Reisman NR, Olshansky SJ. Provision or distribution ofgrowth hormone for “antiaging”: clinical and legal issues. JAMA2005;294:2086–2090.

3. Vance ML. Can growth hormone prevent aging? N Engl J Med2003;348:779–780.

4. Liu H, Bravata DM, Olkin I, et al. Systematic review: the safetyand efficacy of growth hormone in the healthy elderly. Ann InternMed 2007;146:104–115.

5. Melmed S. Supplemental growth hormone in healthy adults: theendocrinologist’s responsibility. Nat Clin Pract Endocrinol Metab2006;2:119.

6. Ma J, Pollak MN, Giovannucci E, et al. Prospective study ofcolorectal cancer risk in men and plasma levels of insulin-likegrowth factor (IGF)-I and IGF-binding protein-3. J Natl Cancer Inst1999;91:620–625.

7. Giovannucci E, Pollak MN, Platz EA, et al. A prospective study ofplasma insulin-like growth factor-1 and binding protein-3 and riskof colorectal neoplasia in women. Cancer Epidemiol BiomarkersPrev 2000;9:345–349.

8. Kaaks R, Toniolo P, Akhmedkhanov A, et al. Serum C-peptide,insulin-like growth factor (IGF)-I, IGF-binding proteins, and colorec-tal cancer risk in women. J Natl Cancer Inst 2000;92:1592–1600.

9. Jenkins PJ. Acromegaly and cancer. Horm Res 2004;62(Suppl 1):

108–115.

0. Domenech-Santasusana M, Carles J, Goday A, et al. Associationof acromegaly and two malignancies: colorectal cancer and non-Hodgkin’s lymphoma. Ann Oncol 1994;5:659.

1. Jenkins PJ, Fairclough PD. Screening guidelines for colorectalcancer and polyps in patients with acromegaly. Gut 2002;51(Suppl 5):V13–V14.

2. Miller ME, Michaylira CZ, Simmons JG, et al. Suppressor ofcytokine signaling-2: a growth hormone-inducible inhibitor of in-testinal epithelial cell proliferation. Gastroenterology 2004;127:570–581.

3. Wu Y, Yakar S, Zhao L, et al. Circulating insulin-like growth factor-Ilevels regulate colon cancer growth and metastasis. Cancer Res2002;62:1030–1035.

4. Warren RS, Yuan H, Matli MR, et al. Induction of vascular endo-thelial growth factor by insulin-like growth factor 1 in colorectalcarcinoma. J Biol Chem 1996;271:29483–29488.

5. Swerdlow AJ, Higgins CD, Adlard P, et al. Risk of cancer inpatients treated with human pituitary growth hormone in the UK,1959-85: a cohort study. Lancet 2002;360:273–277.

6. Sperling MA, Saenger PH, Hintz R, et al. Special editorial: growthhormone treatment and neoplasia-coincidence or consequence?J Clin Endocrinol Metab 2002;87:5351–5352.

7. Lahm H, Amstad P, Wyniger J, et al. Blockade of the insulin-likegrowth-factor-I receptor inhibits growth of human colorectal can-cer cells: evidence of a functional IGF-II-mediated autocrine loop.Int J Cancer 1994;58:452–459.

8. Reinmuth N, Fan F, Liu W, et al. Impact of insulin-like growthfactor receptor-I function on angiogenesis, growth, and metasta-sis of colon cancer. Lab Invest 2002;82:1377–1389.

9. Michaylira CZ, Ramocki NM, Simmons JG, et al. Haplotype insuf-ficiency for suppressor of cytokine signaling-2 enhances intesti-nal growth and promotes polyp formation in growth hormone-transgenic mice. Endocrinology 2006;147:1632–1641.

0. Giovannucci E. Insulin, insulin-like growth factors and colon can-cer: a review of the evidence. J Nutr 2001;131:3109S–3120S.

1. Theiss AL, Fruchtman S, Lund PK. Growth factors in inflammatorybowel disease: the actions and interactions of growth hormoneand insulin-like growth factor-I. Inflamm Bowel Dis 2004;10:871–880.

2. Slonim AE, Bulone L, Damore MB, et al. A preliminary study ofgrowth hormone therapy for Crohn’s disease. N Engl J Med2000;342:1633–1637.

3. Sinha A, Nightingale J, West KP, et al. Epidermal growth factorenemas with oral mesalamine for mild-to-moderate left-sidedulcerative colitis or proctitis. N Engl J Med 2003;349:350–357.

4. Hakam A, Yeatman TJ, Lu L, et al. Expression of insulin-likegrowth factor-1 receptor in human colorectal cancer. Hum Pathol1999;30:1128–1133.

5. Weber MM, Fottner C, Liu SB, et al. Overexpression of theinsulin-like growth factor I receptor in human colon carcinomas.Cancer 2002;95:2086–2095.

6. Itzkowitz SH. Molecular biology of dysplasia and cancer in inflam-matory bowel disease. Gastroenterol Clin North Am 2006;35:553–571.

Address requests for reprints to: Gil Y. Melmed, MD, Cedars-Sinaiedical Center, 8635 West Third Street, #960-W, Los Angeles, Cali-

ornia 90048. e-mail: melmedg@cshs.org; fax: (310) 967-0131.Supported by the National Institutes of Health (CA 75979 to S.M.), the

oris Factor Molecular Endocrinology Laboratory, and by a National Insti-utes of Health–sponsored Gastroenterology training grant (T32 DK7180-31 to G.Y.M.). S.M. is an ad hoc scientific advisor to Eli Lilly.

G.Y.M. and S.M.D. contributed equally to this article.

RG

MP

*D

BotcM(tapaiCpa

SGIwrwC

2plirw(pppb

pts

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2008;6:364–366

ituximab Therapy for Refractory Biliary Strictures in Immunoglobulin4–Associated Cholangitis

ARK TOPAZIAN,* THOMAS E. WITZIG,‡ THOMAS C. SMYRK,§ JOSE S. PULIDO,� MICHAEL J. LEVY,*ATRICK S. KAMATH,* and SURESH T. CHARI*

Miles and Shirley Fiterman Center for Digestive Diseases, Department of Medicine, Division of Gastroenterology and Hepatology, ‡Department of Internal Medicine,

ivision of Hematology and Oncology, §Department of Pathology, and �Department of Ophthalmology, Mayo Clinic, Rochester, Minnesota

mwam

plaEhiwmmBtrtbnaEspcta

tdicsuhiiaCsc

Ea

See Ghazale AH et al on page 706 forcompanion article in the March 2008 issue ofGastroenterology.

ackground & Aims: Biliary strictures occur in a thirdf patients with autoimmune pancreatitis and have beenermed immunoglobulin G subclass 4 (IgG4) associatedholangitis (IAC). IAC often responds to steroid therapy.ethods: A patient with autoimmune pancreatitis and

IAC) refractory to steroids and 6-mercaptopurine wasreated with rituximab, a monoclonal antibody directedgainst the CD20 antigen on B lymphocytes. Results: Theatient’s biliary strictures improved after rituximab ther-py, permitting removal of his biliary stents. Systemic man-festations of IgG4-associated disease also improved.onclusions: Rituximab may be a treatment option foratients with refractory or recurrent autoimmune pancre-titis or IAC.

teroid-responsive biliary strictures occur in autoimmunepancreatitis (AIP), and have been termed immunoglobulin

subclass 4 (IgG4) associated cholangitis (IAC).1 Patients withAC who relapse after withdrawal of steroids often improveith azathioprine.2 This report describes a patient with IAC and

efractory hilar biliary strictures who responded to treatmentith rituximab, a monoclonal antibody directed against theD20 antigen on B cells.

Case ReportA 64-year-old man developed pancreatitis in the fall of

004. Serum liver tests were normal except for an alkalinehosphatase level of 130 U/L (normal, 80 –120 U/L). History,

aboratory evaluation, abdominal ultrasound, and computer-zed tomography scan did not reveal an etiology. Magneticesonance imaging showed an irregular main pancreatic ductith a stenosis in the neck. His serum IgG4 level was 568 mg/dL

normal, 8 –140 mg/dL). Endoscopic ultrasound showed a hy-oechoic, enlarged pancreas. Endoscopic retrograde cholangio-ancreatography (ERCP) showed multifocal strictures of theancreatic duct. Cholangiography was normal. Pancreatic ductrushings were negative for malignancy.

Based on the increased serum IgG4 level and compatibleancreatography, a diagnosis of AIP was made and he wasreated with an 8-week tapering course of oral prednisone

tarting with 50 mg/day. His postprandial pain recurred a

onth after completing steroid therapy, and he was re-treatedith a 12-week course of prednisone with immediate symptom-tic improvement. Steroid therapy was complicated by ophthal-ic herpes zoster.In the fall of 2005 he became jaundiced. His serum alkaline

hosphatase level was 378 U/L, his aspartate aminotransferaseevel was 215 U/L, his total bilirubin level was 12.7 mg/dL, hislbumin level was 3.1 g/dL, and his IgG4 level was 993 mg/dL.RCP showed a stenosis of the biliary confluence and commonepatic duct (Figure 1A). Biliary brushings for cytology, digital

mage analysis, and fluorescent in situ hybridization (FISH)ere negative for malignancy. Intraductal biliary biopsy speci-ens showed a lymphoplasmacytic inflammatory infiltrate withany IgG4-positive cells, compatible with IAC (Figure 2A and

). Biliary stents were placed and his jaundice resolved. He wasreated with a 12-week tapering course of oral prednisone, butepeat ERCP showed a persistent hilar stenosis extending fur-her into the intrahepatic ducts than previously. Intraductalrushings and biopsy specimens again were negative for malig-ancy. He received another 12-week course of oral prednisones well as oral 6-mercaptopurine at a dose of 1.5 mg/kg/day.RCP was repeated 3 months later; the hilar strictures wereomewhat improved, and the biliary stents were removed. Theatient was hospitalized 2 weeks later with jaundice andholangitis. Cholangiography again showed a tight stenosis ofhe biliary confluence, and 3 plastic biliary stents were placednd subsequently exchanged every 3 months.

In the fall of 2006, after 8 months of 6-mercaptopurineherapy, the patient lost weight and developed steatorrhea withecreased vision in the right eye. His serum bilirubin level

ncreased to 4.4 mg/dL and remained increased despite ex-hange of his biliary stents. A positron emission tomographycan showed mild to moderate [18F]-2-fluoro-deoxy-D-glucoseptake consistent with inflammatory disease in the hepaticilum, the pancreas, and the right eye. Ophthalmologic exam-

nation revealed deep choroidal infiltrates. Magnetic resonancemaging showed thickening of the pancreatic body and tail withn irregular pancreatic duct, and hilar bile duct thickening.holangiography showed worsening of his intrahepatic biliary

trictures, and it was difficult to fill his intrahepatic ducts withontrast despite occlusion cholangiography (Figure 1B). Addi-

Abbreviations used in this paper: AIP, autoimmune pancreatitis;RCP, endoscopic retrograde cholangiopancreatography; IAC, IgG4-ssociated cholangitis; IgG4, immunoglobulin G subclass 4.

© 2008 by the AGA Institute1542-3565/08/$34.00

doi:10.1016/j.cgh.2007.12.020

ta

gpficclcpBk

mdmmwnjggspta3issHNw(5Ih

ccp

oootepSibss

aonitp

aletndbid

dapctrst

Ftpgt

Fipst

March 2008 RITUXIMAB FOR IgG4–ASSOCIATED CHOLANGITIS 365

ional biliary biopsy specimens again showed findings of IAC,nd biliary brushings were negative for malignancy.

EUS was repeated and showed a hypoechoic pancreas. EUS-uided Tru-cut biopsy of the pancreatic body showed chronicancreatitis with a lymphoplasmacytic infiltrate and storiformbrosis (Figure 2C). Immunostains showed many IgG4-positiveells, supporting the diagnosis of AIP (Figure 2D). The lympho-ytic infiltrate was composed of both CD3- and CD20-positiveymphocytes. Stains for � and � light chains did not show lighthain restriction. Peripheral blood flow cytometric immuno-henotyping showed no monotypic B-cell population. T- and-cell quantitative markers showed normal CD4 and naturaliller cell numbers. The serum IgG4 level was 1480 mg/dL.

A diagnosis was made of AIP and IAC with ocular involve-ent, refractory to steroids and 6-mercaptopurine, which were

iscontinued. Rituximab immunotherapy was initiated at 375g/m2 weekly for 4 doses in November of 2006. Within aonth of initiating rituximab therapy the patient’s sense ofell being improved steadily; he regained weight and resumedormal activities. Vision in his right eye returned to normal. His

aundice and symptoms of steatorrhea resolved. An upper-astrointestinal hemorrhage occurred and was attributed toastric varices; a transjugular liver biopsy in January of 2007howed cirrhosis with rare, scattered IgG4-positive cells. Oralropranolol was begun. Repeat ophthalmologic examination ofhe right eye showed the choroidal lesions were now atrophicnd inactive. The patient was placed on maintenance rituximab75 mg/m2 every 3 months. ERCP performed 4 months after

nitiation of rituximab showed improvement in his hilar biliarytrictures (Figure 1C) and his biliary stents were removed. Theerum IgG4 level was 1020 mg/dL at the time of stent removal.e remained anicteric and without evidence of cholangitis.ine months after stent removal his total serum bilirubin levelas 1.1 mg/dL, his alkaline phosphatase level was 267 IU/L

normal, 45–115 IU/L), his aspartate aminotransferase level was8 IU/L, his alanine aminotransferase level was 40 IU/L, and hisgG4 level was 299 mg/dL. The patient consented to review ofis medical record for research purposes.

DiscussionAIP is an inflammatory disorder characterized by pan-

reatic enlargement, multifocal pancreatic duct strictures, in-reased serum IgG4 levels, and a pancreatic IgG4-positive lym-

igure 1. (A) Cholangiogram showing hilar biliary strictures beforereatment. (B) Progression of strictures after steroid and 6-mercapto-urine therapy. Intrahepatic ducts fill poorly despite occlusion cholan-iography. (C) Improvement 4 months after the initiation of rituximabherapy.

hoplasmacytic infiltrate with narrowing of small ducts and s

bliterative phlebitis. Bile duct strictures are reported in a thirdf patients with AIP and have been termed IAC.3 IAC also mayccur in patients without pancreatitis. Stenoses may occur inhe intrapancreatic bile duct, biliary confluence, and/or periph-ral intrahepatic ducts. Histology of resected IAC shows IgG4-ositive lymphoplasmacytic infiltrates in the bile duct wall.4

imilar infiltrates have been reported in some cases of thyroid-tis, Mikulicz’s disease, chronic sialadenitis, retroperitoneal fi-rosis, tubulointerstitial nephritis, and autoimmune hypophy-itis,5–7 and these all may be manifestations of an IgG4-relatedystemic disease.1,8

IAC typically responds to oral steroid therapy.1,3,4,9 Relapsefter withdrawal of steroids is common, particularly with hilarr intrahepatic duct strictures, and additional steroid or immu-omodulator therapy often is successful.4 The patient described

n this report did not respond to steroid or 6-mercaptopurineherapy. Rituximab therapy resulted in cholangiographic im-rovement and allowed his biliary stents to be removed.

Rituximab is a chimeric IgG1 monoclonal antibody directedgainst CD20, a phosphoprotein expressed on the surface of Bymphocytes. Initially used to treat lymphoma, rituximab hasfficacy in some immune-mediated diseases, including rheuma-oid arthritis,10 thrombotic thrombocytopenic purpura,11 be-ign orbital pseudolymphoma,12 and pemphigus vulgaris13 (aisease characterized by the presence of pathogenic IgG4 anti-odies). The mechanism of action of rituximab in these diseases

s presumed to be B-cell depletion, resulting in decreased pro-uction of pathogenic autoantibodies.

Because the patient was treated solely with rituximab, whichepletes only B cells, this case suggests that B lymphocytes playcentral role in the pathogenesis of IAC and AIP. The lym-

hoplasmacytic infiltrates that characterize AIP include poly-lonal B cells, plasma cells, and T cells.14 In addition to infil-rating affected tissues, B lymphocytes affect the inflammatoryesponse by interaction with regulatory T cells.15 The patient’serum IgG4 level initially remained increased, sharply decliningoward normal 7 months after rituximab therapy was begun.

igure 2. (A) Biliary biopsy specimen showing a lymphoplasmacyticnfiltrate. (B) IgG4 stain of the biliary biopsy specimen shows abundantositive plasma cells, consistent with IAC. (C) Pancreas Tru-cut biopsyhows features of chronic pancreatitis with a lymphoplasmacytic infil-ration and storiform fibrosis, consistent with AIP. (D) Pancreatic IgG4

tain shows abundant positive plasma cells.

TbtbdphI

bmsrs

IabppaHum

sita

rtaaoaa

1

1

1

1

1

1

1

1

1

1

FDSm

a

366 TOPAZIAN ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 3

he initially persistent increase of IgG4 levels is not surprisingecause immunoglobulins are secreted by mature plasma cellshat are not depleted by rituximab. This phenomenon also haseen reported in patients with rheumatoid arthritis10 and Wal-enstrom’s macroglobulinemia16 who receive rituximab. Theatient’s clinical and cholangiographic response occurred whileis serum IgG4 level was increased persistently, suggesting that

gG4 was not directly pathogenic.Cirrhosis in this case most likely was caused by chronic

iliary obstruction caused by IAC. A liver biopsy performed 2onths after the initiation of rituximab therapy did show rare,

cattered IgG4-positive cells. By improving biliary strictures,ituximab therapy might lead to some reversal of hepatic fibro-is.17,18

Could our patient have a biliary malignancy rather thanAC? Pseudotumors of the biliary tree have been described inssociation with AIP.19 The diagnosis of IAC in this case wasased on the marked increase of the serum IgG4 levels, theresence of co-existent AIP, and compatible biliary histology. Aositron emission tomography scan showed mild to moderatectivity suggestive of inflammation rather than malignancy.istologic and hematologic evaluations for lymphoma werenrevealing. It is unlikely that this patient’s biliary stricture wasalignant.The required duration of rituximab therapy is unclear. The

hort-term efficacy of rituximab therapy has been documentedn rheumatoid arthritis10 and other inflammatory diseases, buthe duration of response and the benefit of maintenance ther-py is uncertain. Further data are needed in this regard.

In conclusion, rituximab therapy was effective in this case ofefractory IgG4-associated cholangitis. Rituximab may be areatment option for patients with refractory or recurrent IgG4-ssociated disease. Rituximab has few side effects, and may ben attractive therapeutic option in patients who are intolerantf steroids or suffer relapses of IgG4-associated disease, as wells in patients such as this one with AIP and refractory IgG4-ssociated cholangitis.

References

1. Björnsson E, Chari S, Smyrk T, et al. IgG4 associated cholangitis:description of an emerging clinical entity based on review of theliterature. Hepatology 2007;45:1547–1554.

2. Ghazale A, Chari S, Smyrk T, et al. Value of serum IgG4 in thediagnosis of autoimmune pancreatitis and in distinguishing itfrom pancreatic cancer. Am J Gastroenterol 2007;102:1646–1653.

3. Nishino T, Toki F, Oyama H, et al. Biliary tract involvement inautoimmune pancreatitis. Pancreas 2005;30:76–82.

4. Ghazale A, Chari S, Zhang L, et al. IgG4-associated cholangitis(IAC): clinical profile and response to therapy. Gastroenterology2008;134:706–715.

5. Ohara H, Nakazawa T, Sano H, et al. Systemic extrapancreaticlesions associated with autoimmune pancreatitis. Pancreas

2005;31:232–237. T

6. Yamamoto M, Takahashi H, Ohara M, et al. A new conceptualiza-tion for Mikulicz’s disease as an IgG4-related plasmacytic dis-ease. Mod Rheumatol 2006;16:335–340.

7. Chari S, Smyrk T, Levy M, et al. Diagnosis of autoimmune pan-creatitis: the Mayo Clinic experience. Clin Gastroenterol Hepatol2006;4:1010–1016.

8. Hamed G, Tsushima K, Yasuo M, et al. Inflammatory lesions ofthe lung, submandibular gland, bile duct and prostate in a patientwith IgG4-associated multifocal systemic fibrosclerosis. Respirol-ogy 2007;12:455–457.

9. Matsushita M, Yamashina M, Ikeura T, et al. Effective steroidpulse therapy for the biliary stenosis caused by autoimmunepancreatitis. Am J Gastroenterol 2007;102:220–221.

0. Cohen S, Emery P, Greenwald M, et al. Rituximab for rheumatoidarthritis refractory to anti-tumor necrosis factor therapy: resultsof a multicenter, randomized, double-blind, placebo-controlled,phase III trial evaluating primary efficacy and safety at twenty-fourweeks. Arthritis Rheum 2006;54:2793–2806.

1. Scully M, Cohen H, Cavenagh J, et al. Remission in acute refrac-tory and relapsing thrombotic thrombocytopenic purpura follow-ing rituximab is associated with a reduction in IgG antibodies toADAMTS-13. Br J Haematol 2007;136:451–461.

2. Witzig T, Inwards D, Habermann T, et al. Treatment of benignorbital pseudolymphomas with the monoclonal anti-CD20 anti-body rituximab. Mayo Clin Proc 2007;82:692–699.

3. Ahmed A, Spigelman Z, Cavacini L, et al. Treatment of pemphigusvulgaris with rituximab and intravenous immune globulin. N EnglJ Med 2006;355:1772–1779.

4. Kojima M, Sipos B, Klapper W, et al. Autoimmune pancreatitis:frequency, IgG4 expression, and clonality of T and B cells. Am JSurg Pathol 2007;31:521–528.

5. Yang Z, Novak A, Ziesmer S, et al. Attenuation of CD8(�) T-cellfunction by CD4(�)CD25(�) regulatory T cells in B-cell non-Hodgkin’s lymphoma. Cancer Res 2006;15:10145–10152.

6. Ghobrial I, Fonseca R, Greipp P, et al. Initial immunoglobulin M’flare’ after rituximab therapy in patients diagnosed with Walden-strom macroglobulinemia: an Eastern Cooperative OncologyGroup Study. Cancer 2004;101:2593–2598.

7. Bonis P, Friedman S, Kaplan M. Is liver fibrosis reversible? N EnglJ Med 2001;344:452–454.

8. Hammel P, Couvelard A, O’Toole D, et al. Regression of liverfibrosis after biliary drainage in patients with chronic pancreatitisand stenosis of the common bile duct. N Engl J Med 2001;344:418–423.

9. Kurihara T, Itoi T, Sofuni A, et al. Pseudotumor of the bile ductassociated with autoimmune pancreatitis. Endoscopy 2007;epub ahead of print.

Address requests for reprints to: Dr Mark Topazian, Miles and Shirleyiterman Center for Digestive Diseases, Department of Medicine,ivision of Gastroenterology and Hepatology, Mayo Clinic, 200 Firsttreet SW, Rochester, Minnesota 55906. e-mail: topazian.mark@ayo.edu; fax: (507) 266-3939.Dr Witzig serves on the Genentech and BiogenIDEC advisory boards

nd receives research support from these companies for clinical trials.

his information was disclosed to participants.

LRHpb

RSD

caoleDprTltspp

Rtb

24c5wv

dihttdhwpt

i

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2008;6:367

ETTER TO THE EDITOReaders are encouraged to write letters to the editor concerning articles that have been published in CLINICAL GASTROENTEROLOGY AND

EPATOLOGY. Short, general comments are also considered, but use of the Letters to the Editor section for publication of original data inreliminary form is not encouraged. Letters should be typewritten and submitted electronically to http://www.editorialmanager.com/cgh. Please

e sure to send 2 hard copies of any figures to the editorial office.

e46r

acI

1

2

3

4

5

6

ibavirin Posology, Observance, andustained Virologic Responseear Editor:The article by Reddy et al1 seems very encouraging for clini-

ians and patients; minor ribavirin dose reductions do notppear to significantly affect sustained virologic response (SVR)f genotype 1 hepatitis C with 48 weeks of treatment if cumu-

ative exposure is more than 60% of the prescribed dose. How-ver, it is also confusing. The seminal article used for Food andrug Administration and European Agency enregistrationointed out that 1000 –1200 mg ribavirin is better than 800 mgibavirin for 48 weeks of treatment of genotype 1 hepatitis C.2

his dose difference is between 60% and 100%! It will be prob-ematic if clinicians conclude that the efficacy difference be-ween ribavirin posology of 100% and, for example, 70% is verymall, in opposition to the rule 80-80-80.2 Importance of com-liance and observance will be weakened and also the interest ofrescribing erythropoietin in case of ribavirin-induced anemia.When looking more carefully at pooled data in the article by

eddy et al,1 this apparent discordance could be explained; inhat article there is no analysis taking into account the resultsy low or high pretreatment viremia.

In the article by Hadziyannis et al,3 SVR difference between4-week treatment with 1000 –1200 mg ribavirin (42%) and8-week treatment with 800 mg ribavirin (41%) is not signifi-ant, especially in case of low pretreatment viral load (52% and5%, respectively). Recently the possibility to treat only for 24eeks patients with low pretreatment viral load and rapid

irologic response (RVR) at week 4 has been shown.4,5

So it is not surprising to illustrate that 48 weeks with stan-ard dose of ribavirin or only greater than 60% dose of ribavirin

s equivalent if the data analysis does not differentiate low andigh pretreatment viral load. In genotype 1 patients, low pre-reatment viral load patients especially with RVR are easier toreat than high viral load patients, and a decrease of ribavirinose could be effective with less consequences in SVR than forigh viral load patients if treatment duration is 48 weeks; 24eeks with full dose 1000 –1200 mg ribavirin in case of lowretreatment viral load is the standard of care if RVR is ob-ained.

The real clinical questions from the article by Reddy et al1 are

n case of high pretreatment viral load, is 60% of ribavirin dose

quivalent to standard dose (1000 –1200 mg) ribavirin during8 weeks? In case of low pretreatment viral load and RVR, is0% of ribavirin dose equivalent to standard dose (800 mg)ibavirin during 24 weeks?

Therefore, we ask Reddy et al for data reanalysis, taking intoccount the pretreatment viral load, if possible with differentutoff (the most discriminant cutoff could be around 400,000U/mL6) and also the RVR.

P. COUZIGOUJ. FOUCHERL. CASTERAV. DE LEDINGHENHepatoGastroEnterology DepartmentUniversity Victor Segalen BordeauxHaut Levêque HospitalPessac, France

. Reddy KR, Shiffman ML, Morgan TR, et al. Impact of ribavirin dosereductions in hepatitis C virus genotype 1 patients completingpeginterferon alfa-2a/ribavirin treatment. Clin Gastroenterol Hepa-tol 2007;5:124–129.

. Mc Hutchison JG, Manns M, Patel K, et al. Adherence to combi-naison therapy enhances sustained response in genotype 1-in-fected patients with chronic hepatitis C. Gastroenterology 2002;123:1061–1069.

. Hadziyannis SJ, Sette H Jr, Morgan TR, et al. Peginterferon alpha2a (40 KD) and ribavirin combination therapy in chronic hepatitisC: randomized study of the effect of duration and ribavirin dose.Ann Intern Med 2004;140:346–355.

. Zeuzem S, Buti M, Ferenci P, et al. Efficacy of 24 weeks treatmentwith peginterferon alfa-2b plus ribavirin in patients with chronichepatitis C infected with genotype 1 and low pretreatment viremia.J Hepatol 2006;44:97–103.

. Jensen D, Morgan TR, Marcellin P, et al. Early identification of HCVgenotype 1 patients responding to 24 weeks peginterferon al-pha-2a (40 kd)/ribavirin therapy. Hepatology 2006;43:954–956.

. Zeuzem S, Fried M, Reddy KR, et al. Improving the clinical rele-vance of pre-treatment viral load as a predictor of sustained viro-logical response SVR in patients infected with hepatitis C genotype1 treated with peginterferon alpha2a (40KD) (pegasys) plus riba-virin (copegus). Hepatology 2006;44:267A.

doi:10.1016/j.cgh.2007.09.008

SUBSCRIPTIONS

Correspondence regarding subscriptions or change ofaddress should be directed to CLINICAL GASTROENTEROLOGY

AND HEPATOLOGY, Elsevier, Periodicals Department, 6277 SeaHarbor Dr, Orlando, FL 32887-4800. Change of address no-tices, including both the old and new addresses of the sub-scriber and the mailing label, should be sent at least one monthin advance. Customer Service: (800) 654-2452; outside theUnited States and Canada, (407) 563-6020. Yearly subscrip-tion rates: United States and possessions: individuals, $224;institutions, $396; students and residents, $114; single issues,$38. All other countries: individuals, $279; institutions, $455;students and residents, $153; single issues, $38. For all areasoutside the United States and possessions, there is no addi-tional charge for surface delivery. To receive student/residentrate, orders must be accompanied by name of affiliated insti-tution, date of term, and the signature of program/residencycoordinator on institution letterhead. Orders will be billed atindividual rate until proof of status is received. Checks shouldbe made payable to Elsevier and sent to Clinical Gastroenterologyand Hepatology, Elsevier, Periodicals Department, PO Box628239, Orlando, FL 32862-8239. Prices are subject tochange without notice. Current prices are in effect for backvolumes and back issues. Single issues, both current and back,exist in limited qualities and are offered for sale subject toavailability. Back issues sold in conjunction with a subscrip-tion are on a prorated basis.

CLAIMS FOR MISSING ISSUES

Claims for missing issues can be honored only up to 3months for domestic addresses and 6 months for foreign ad-dresses. Duplicate copies will not be sent to replace onesundelivered through failure to notify Elsevier of change ofaddress.

REPRINTS

For 100 or more copies of an article in this publication,please contact the Commercial Reprints Department, ElsevierScience Inc., 360 Park Avenue South, New York, New York10010-1710. Tel: (212) 633-3813; fax: (212) 633-3820; e-mail: reprints@elsevier.com.

ADVERTISING REPRESENTATIVE

For display advertising, contact: Cunningham Associ-ates, 180 Old Tappan Rd, Old Tappan, NJ 07675; telephone(201) 767-4170; fax (201) 767-8065. For classified advertis-ing, contact: Jay Hong, 360 Park Avenue South, New York,NY 10010-1710; telephone (212) 633-3713; fax (212) 633-3820.

STATEMENT OF RESPONSIBILITY

The ideas and opinions expressed in Clinical Gastroenter-ology and Hepatology do not necessarily reflect those of the Amer-ican Gastroenterological Association (AGA) Institute, the Editor,or the Publisher. Publication of an advertisement or other productmention in Clinical Gastroenterology and Hepatology should not beconstrued as an endorsement of the product or the manufacturer’sclaims. Readers are encouraged to contact the manufacturer withany questions about the features or limitations of the productsmentioned. Neither the American Gastroenterological Associa-tion (AGA) Institute nor the Publisher assumes any responsibilityfor any injury and/or damage to persons or property arising out ofor related to any use of the material contained in this periodical.The reader is advised to check the appropriate medical literatureand the product information currently provided by the manufac-turer of each drug to be administered to verify the dosage, themethod and duration of administration, or contraindications. It isthe responsibility of the treating physician or other health careprofessional, relying on independent experience and knowledge ofthe patient, to determine drug dosages and the best treatment forthe patient.

PERMISSIONS

Copyright © 2008 by The AGA Institute. All rightsreserved. No part of this publication may be reproduced ortransmitted in any form or by any means now or hereafter known,electronic or mechanical, including photocopy, recording, or anyinformation storage and retrieval system, without permission inwriting from the Publisher. Printed in the United States ofAmerica.

Correspondence regarding permission to reprint all or part ofany article published in this journal may be sought directly fromElsevier’s Global Rights Department in Oxford, UK: phone215-239-3804 or �44 (0)1865 843830, fax �44 (0)1865853333, e-mail healthpermissions@elsevier.com. Requests mayalso be completed on-line via the Elsevier homepage (www.elsevier.com/permissions).

The appearance of the code at the bottom of the first page of anarticle in this journal indicates the copyright owner’s consent thatcopies of the article may be made for personal or internal use, orfor the personal or internal use of specific clients, for thoseregistered with the Copyright Clearance Center, Inc. (222 Rose-wood Drive, Danvers, MA 01923; (978) 750-8400; www.copyright.com). This consent is given on the condition that thecopier pay the stated per-copy fee for that article through theCopyright Clearance Center, Inc. for copying beyond that per-mitted by Sections 107 or 108 of the US Copyright Law. Thisconsent does not extend to other kinds of copying, such ascopying for general distribution, for advertising or promotionalpurposes, for creating new collective works, or for resale. Absenceof the code indicates that the material may not be processedthrough the Copyright Clearance Center, Inc.

ThThe e Official fficial Clinical Practicelinical Practice Journal urnal of t the AGA Institutee AGA Institute

Information for Readers

EDITORC. Mel Wilcox

SENIOR ASSOCIATE EDITORMichael B. Fallon

ASSOCIATE EDITORSMiguel R. ArguedasMohamad A. EloubeidiCharles O. Elson

EDITORIAL STAFFThoba Khumalo, Managing EditorBrook Simpson, Assistant Managing EditorSarah Williamson, Medical IllustratorLindsey Guerin, Editorial AssistantShannon Dean, Editorial AssistantErin Dubnansky, Senior Director of Scholarly PublishingChristine Charlip, Division Director of Publications

EDITORIAL BOARD

Neena Abraham, Houston, TXJulio Bai, Buenos Aires, ArgentinaAlan Barkun, Montreal, CanadaDeepak Bhasin, Chandigarh, IndiaLaurence Blendis, Tel Aviv, IsraelSteven R. Brandt, Baltimore, MDAlan Buchman, Chicago, ILPeter Bytzer, Copenhagen, DenmarkMichael Camilleri, Rochester, MNMarcia Canto, Baltimore, MDBrooks Cash, Bethesda, MDNaga Chalasani, Indianapolis, INHenry L. Chan, Hong Kong, ChinaWilliam D. Chey, Ann Arbor, MIMarcia Cruz-Correa, San Juan, PRByron Cryer, Dallas, TXRobert Fraser, Daw Park, South AustraliaMichael Goggins, Baltimore, MD

Benjamin Gold, Atlanta, GATakuji Gotoda, Tokyo, JapanIan Gralnek, Haifa, IsraelSteven-Huy B. Han, Los Angeles, CAStephen A. Harrison, San Antonio, TXJeremy Jass, Montreal, QuebecSunanda Kane, Chicago, ILMeredith Kilgore, Birmingham, ALRichard Kozarek, Seattle, WAUri Ladabaum, San Francisco, CAAngel Lanas, Zaragoza, SpainMichael Levy, Rochester, MNStephen McClave, Louisville, KYKlaus Monkemuller, Magdeburg, GermanyKoenraad Mortele, Boston, MAJohn Pandolfino, Chicago, ILEamonn Quigley, Cork, IrelandJose Remes-Troche, Mexico City, Mexico

David Rubin, Chicago, ILBruce Sands, Boston, MAMark Schattner, New York, NYRobert Schoen, Pittsburgh, PANick Shaheen, Chapel Hill, NCPrateek Sharma, Kansas City, MOKirti Shetty, Washington, DCRhonda F. Souza, Dallas, TXBrennan Spiegel, Los Angeles, CARichard Sterling, Richmond, VAJonathan P. Terdiman, San Francisco, CAPier A. Testoni, Milan, ItalyRadu Tutuian, Zurich, SwitzerlandShyam Varadarajulu, Birmingham, ALWilliam Whitehead, Chapel Hill, NCKhay-Guan Yeoh, Singapore, Republic of

Singapore

EDITOR EMERITUSMichael Camilleri

OFFICERS OF THE AGA INSTITUTE

Nicholas F. LaRusso, PresidentRochester, Minnesota

Robert S. Sandler, President-ElectChapel Hill, North Carolina

Gail Hecht, Vice PresidentChicago, Illinois

Damian H. Augustyn, Secretary/TreasurerSan Francisco, CA

AGA Institute: AGA Education, Practice and Research

Published by Elsevier, 1600 John F. Kennedy Boulevard, Philadelphia, Pennsylvania 19103-2899. Clinical Gastroenterology and Hepatology (ISSN1542-3565), © 2008 by the AGA Institute, is published monthly by Elsevier, 360 Park Avenue South, New York, NY 10010. Business and EditorialOffices: 1600 John F. Kennedy Boulevard, Suite 1800, Philadelphia, PA 19103-2899. Accounting and Circulation Offices: 6277 Sea Harbor Dr,Orlando, FL 32887-4800. Periodicals postage paid at New York, NY and additional mailing offices. Postmaster: Send change of address toClinical Gastroenterology and Hepatology, Elsevier, Periodicals Department, 6277 Sea Harbor Dr, Orlando, FL 32887-4800.

ThThe e Official fficial Clinical Practicelinical Practice Journal urnal of t the AGA Institutee AGA Institute

IMAGES OF THE MONTHxxiv Metastatic Malignant Melanoma of the Gastrointestinal Tract

K. R. Parisian, J. E. McFarland, and A. N. Shah

xxvi Transmural Perforation of the Stomach by a FishboneI. J. Nastaskin, K. Ransibrahmanakul, and W. Trudeau

xxviii Hairball in the Stomach: A Case of Gastric TrichobezoarK. Hisamuddin and C. P. Brandt

ELECTRONIC IMAGES OF THE MONTHAvailable only online at www.cghjournal.org

e11 Gastric Outlet Obstruction Caused by a Large Gallstone in the Duodenum (Bouveret’sSyndrome)K. Kishi, K. Yamada, and T. Sugiyama

e12 Mirizzi With Pre-Bouveret’s SyndromeA. Shakouri and S.-J. Tang

e15 Bouveret’s Syndrome: Diagnosis and Endoscopic TreatmentL. E. V. V. C. Ferreira, M. D. Topazian, and T. H. Baron

CME ACTIVITIESTake the CME exams online at http://www.cghjournal.org/content/cme

261 Exam 1: Pancreatic Cancer: A Review of the Evidence on CausationM. R. Arguedas

263 Exam 2: What Is the Role of Serological Testing in Celiac Disease? A Prospective, Biopsy-Confirmed Study With Economic AnalysisM. R. Arguedas

ON THE COVER(Top Figure) Reprojected CT image during the portal venous phase of study (anteroposterior view of the abdomen). The patient has chronicocclusion of the main portal vein, immediately superior to the superior mesenteric vein (open arrow). The veins of the stoma (solid arrow) areenlarged, congested, and in communication with multiple prominent body wall collateral vessel (curved arrows) See Speir et al on page 346.(Bottom Figure) Esophagogastroduodenoscopy displaying a fish bone penetrating the stomach wall (arrow). See Nastaskin et al on page xxvi.

ThThe e Official fficial Clinical Practicelinical Practice Journal urnal of t the AGA Institutee AGA Institute

Contents Vol. 6, No. 3, March 2008

ABSTRACTS FROM AROUND THE WORLD

264 Visit CGH online at www.cghjournal.org to link to these and additional papers of interestC. M. Wilcox

ELECTRONIC ABSTRACTS FROM AROUND THE WORLD

e17 Available only at www.cghjournal.orgC. M. Wilcox

EDITORIAL

266 Photodynamic Therapy: Standard of Care for Palliation of Cholangiocarcinoma?T. H. Baron

REVIEWS

link 268 Chronic Hepatitis B: Preventing, Detecting, and Managing Viral ResistanceE. B. Keeffe, D. T. Dieterich, J.–M. Pawlotsky, and Y. Benhamou

See companion article by Deng G et al on page 716 in the March issue of Gastroenterology.

ECME 275 Pancreatic Cancer: A Review of the Evidence on CausationA. R. Hart, H. Kennedy, and I. Harvey

CLINICAL IMAGING

283 Computerized Tomography Enterography and Its Role in Small-Bowel ImagingJ. G. Fletcher, J. Huprich, E. V. Loftus, Jr, D. H. Bruining, and J. L. Fidler

ENDOSCOPY CORNER

290 Unresectable Cholangiocarcinoma: Comparison of Survival in Biliary Stenting AloneVersus Stenting With Photodynamic TherapyM. Kahaleh, R. Mishra, V. M. Shami, P. G. Northup, C. L. Berg, P. Bashlor, P. Jones, K. Ellen, G. R. Weiss,C. M. Brenin, B. E. Kurth, T. A. Rich, R. B. Adams, and P. Yeaton

298 Performance Characteristics of the Suspected Blood Indicator Feature in CapsuleEndoscopy According to Indication for StudyJ. M. Buscaglia, S. A. Giday, S. V. Kantsevoy, J. O. Clarke, P. Magno, E. Yong, and G. E. Mullin

ORIGINAL ARTICLES—ALIMENTARY TRACT

link 302 Patient Predictors of Esophageal Stricture Development After Photodynamic TherapyP. Yachimski, W. P. Puricelli, and N. S. Nishioka

See companion article by Curvers W et al on page 670 in the March issue of Gastroenterology.

309 Peptic Ulcerations Are Related to Systemic Rather Than Local Effects of Low-DoseAspirinM. G. H. Van Oijen, J. P. Dieleman, R. J. F. Laheij, M. C. J. M. Sturkenboom, J. B. M. J. Jansen, andF. W. A. Verheugt

ECME 314 What Is the Role of Serologic Testing in Celiac Disease? A Prospective, Biopsy-ConfirmedStudy With Economic AnalysisA. D. Hopper, M. Hadjivassiliou, D. P. Hurlstone, A. J. Lobo, M. E. McAlindon, W. Egner, G. Wild, andD. S. Sanders

321 Symptom Severity but Not Psychopathology Predicts Visceral Hypersensitivity inIrritable Bowel SyndromeP. P. J. Van der Veek, Y. R. Van Rood, and A. A. M. Masclee

329 Chronic Abdominal Pain and Depressive Symptoms: Analysis of the NationalLongitudinal Study of Adolescent HealthN. N. Youssef, K. Atienza, A. L. Langseder, and R. S. Strauss

333 Sharing Genetic Test Results in Lynch Syndrome: Communication With Close andDistant RelativesE. M. Stoffel, B. Ford, R. C. Mercado, D. Punglia, W. Kohlmann, P. Conrad, A. Blanco, K. M. Shannon, M. Powell,S. B. Gruber, J. Terdiman, D. C. Chung, and S. Syngal

ORIGINAL ARTICLES—LIVER, PANCREAS, AND BILIARY TRACT

339 Prognostic Implications of Lactate, Bilirubin, and Etiology in German Patients WithAcute Liver FailureJ. Hadem, P. Stiefel, M. J. Bahr, H. L. Tillmann, K. Rifai, J. Klempnauer, H. Wedemeyer, M. P. Manns, andA. S. Schneider

346 Bleeding Stomal Varices: Case Series and Systematic Review of the LiteratureB. J. Spier, A. A. Fayyad, M. R. Lucey, E. A. Johnson, M. Wojtowycz, L. Rikkers, B. A. Harms, and M. Reichelderfer

353 Dietary Counseling Versus Dietary Supplements for Malnutrition in ChronicPancreatitis: A Randomized Controlled TrialS. Singh, S. Midha, N. Singh, Y. K. Joshi, and P. K. Garg

BRIEF COMMUNICATIONS

360 Anti-Aging Therapy With Human Growth Hormone Associated With Metastatic ColonCancer in a Patient With Crohn’s ColitisG. Y. Melmed, S. M. Devlin, G. Vlotides, D. Dhall, S. Ross, R. Yu, and S. Melmed

link 364 Rituximab Therapy for Refractory Biliary Strictures in Immunoglobulin G4–AssociatedCholangitisM. Topazian, T. E. Witzig, T. C. Smyrk, J. S. Pulido, M. J. Levy, P. S. Kamath, and S. T. Chari

See companion article by Ghazale AH et al on page 706 in the March issue of Gastroenterology.

LETTER TO THE EDITOR

367 Ribavirin Posology, Observance, and Sustained Virologic ResponseP. Couzigou, J. Foucher, L. Castera, and V. de Ledinghen

Access to the full content of Clinical Gastroenterology and Hepatology Online is available to all subscribers!

New! AGA members have seamless access to full Clinical Gastroenterology and Hepatology content from theAGA Web site. Simply sign in to http://www.cghjournal.org, visit the Publications section of the Website, andclick on Clinical Gastroenterology and Hepatology. You will be directed to the Journal Website and will haveaccess to all content without having to supply a different username and password. Members can also visit http://www.cghjournal.org directly and click on “AGA Memebers Log-in.”

Non-member subscribers must create an online user account and activate their subscription to access the full textof articles on Clinical Gastroenterology and Hepatology Online. To activate your individual online subscription,please visit http://www.cghjournal.org and click on “Activate Online Access.” To activate your account, you willyour subscriber acccount or member number, which you can find on your mailing label (AGA members must usetheir 10-digit member number, including leading zeros). If you have any questions regarding your onlinesubscription, please call Elsevier Customer Service at (800) 654-2452.

Sample mailing label

1GAST V91-4 1234567-8J.H. DOE, MD531 MAIN STCENTER CITY, NY 10001

Š This is yoursubscriptionaccount number

Institutional access to Clinical Gastroenterology and Hepatology Online will be allowed only by limited sitelicensing. Further instructions will be available online.

Personal subscriptions to Clinical Gastroenterology and Hepatology Online are for individual use only and maynot be transferred. Use of Clinical Gastroenterology and Hepatology Online is subject to agreement to the termsand conditions of use as indicated online.

This is yourAmericanGastroenterologicalAssociation membernumber

‹ 0000101864JANE DOE531 MAIN STCENTER CITY, NY 10001

INFORMATION FOR AUTHORS

ContentCLINICAL GASTROENTEROLOGY AND HEPATOLOGY publishesoriginal papers, review articles, brief communications, andspecial category articles on all aspects of the digestive tract andthe liver. Manuscripts must be prepared in accordance with the“Uniform Requirements for Manuscripts Submitted to Bio-medical Journals” developed by the International Committeeof Medical Journal Editors (http://www.icmje.org).

EthicsCLINICAL GASTROENTEROLOGY AND HEPATOLOGY strongly dis-courages the submission of more than one article dealing withrelated aspects of the same study. In almost all cases, a singlestudy is best reported in a single paper.

The Journal editors consider research/publication miscon-duct to be a serious breach of ethics and will take action asnecessary to address such misconduct, which includes submis-sion or publication of information that:

● Is intentionally erroneous,

● Has been published elsewhere by a different author with-out acknowledgment (plagiarism),

● Has been published elsewhere by the same author with-out acknowledgment (duplicate publication), or

● Is subsequently published elsewhere by the same authorwithout acknowledgment, attribution, or permissionfrom the AGA Institute, as holder of the copyright, toreprint or adapt the material.

Breaches in these standards may result in proscribed submis-sion for all authors of the concerned manuscript and, whenappropriate, notification of the authors’ institutions. All au-thors are fully responsible for the content of the manuscript.

The publication of abstracts is not considered duplicatepublication but should be disclosed in the cover letter accom-panying the manuscript submission.

AuthorshipEach author must have participated sufficiently in the work totake public responsibility for the content of the paper and mustapprove of the final version of the manuscript. Authorship shouldbe based on substantive contributions to each of the following:

● Conception and design of the study;

● Generation, collection, assembly, analysis and/or inter-pretation of data;

● Drafting or revision of the manuscript;

● Approval of the final version of the manuscript.

National Institutes of Health (NIH) FundingIf a manuscript is accepted for publication and was supported inpart, or in whole, by the NIH, the author may request that the

manuscript be automatically posted to PubMed Central (PMC). Ifthe author wishes to make this request, he or she should check theappropriate box located on the Copyright Assignment Formunder the NIH Funding portion of the form. In addition, theauthor must provide the NIH grant number. If an author makesthis request, the manuscript will automatically be entered in thePMC system and will be available to the public, free of charge,twelve months from the manuscript’s publication date.

Disclosure Policy

I. Authors (Original Articles and Review Articles)The following information must be included on the titlepage of submitted manuscripts:

• A list for all authors that discloses any financial arrange-ment (e.g., consultancies, stock ownership, equity inter-ests, patent-licensing arrangements, research support,major honoraria, etc.) they may have with a companywhose product figures prominently in the submittedmanuscript or with a company making a competingproduct. Interactions that occur from the start of theresearch activity in the specific program until the timewhen the paper is anticipated to be published or one yearfrom submission date, whichever is longer, are pertinent.In the absence of any conflict of interest, authors mustmake the statement that there is no conflict to disclose.

• A statement of all funding sources supporting the workand all institutional or corporate affiliations.

• Individuals who provided writing assistance for the manuscriptand the funding source for this assistance must be disclosed.

• Investigators must disclose potential conflicts to studyparticipants and must state whether they have done so.

• All phase II or III trials pertaining to a commercialproduct (pharmaceutical or device) require:

a. A statement that the statistical analysis of the entire datasets pertaining to efficacy (specifically primary and majorsecondary efficacy endpoints) and safety (specifically, se-rious adverse events as defined in federal guidelines) havebeen independently confirmed by a biostatistician who isnot employed by the corporate entity (the name of thebiostatistician must be provided); and

b. A statement from the corresponding author that he orshe had full access to all of the data and takes fullresponsibility for the veracity of the data and analysis.All authors for each manuscript will also need to com-plete and submit to the journal’s editorial office theAuthorship Responsibility portion of the CopyrightAssignment, Authorship Responsibility, NIH Fund-ing, Financial Disclosure, Institutional Review Board/Animal Care Committee Approval, and Sponsorshipform upon submission of a manuscript.

ThThe e Official fficial Clinical Practicelinical Practice Journal urnal of t the AGA Institutee AGA Institute

• Authors must describe the role of the study sponsor(s), ifany, in the study design; in the collection, analysis, andinterpretation of data; in the writing of the report; and inthe decision to submit the report for publication.

II. Reviewers and Editorialists

• Reviewers and editorialists will be disqualified fromreviewing or from writing an editorial if they:

a. have had an ongoing collaboration, original publica-tions or grants with the authors within the previoustwo years, except in the case of being a part of amulti-center group from a different site; or

b. are from the same institution as the authors.

• Reviewers and editorialists should disclose at the time of sub-mission of their review or editorial any financial arrangement(e.g., consultancies, stock ownership, equity interests, patent-licensing arrangements, research support, major honoraria, etc.)they may have with a company whose product figures prom-inently in the submitted manuscript or with a company mak-ing a competing product. Interactions that occur from the startof the research activity in the specific program until the timewhen the paper is anticipated to be published or one year fromsubmission date, whichever is longer, are pertinent.

Continuing Medical Education (CME)Each issue of CLINICAL GASTROENTEROLOGY AND HEPATOLOGY

will contain two to three continuing medical education examsassociated with articles that appear in the issue. AGA memberscan take the exams online free of charge. Non-AGA members arerequired to pay a $15 processing fee. For CME exams thataccompany original articles, readers can claim 1.0 AMA PRACategory 1 Credits.™

Reviewers of manuscripts can also claim CME credit. Afterreviewing a manuscript on the Journal’s manuscript trackingsystem, Editorial Manager, the reviewer will be prompted toclaim up to 3.0 AMA PRA Category 1 Credits.™

Cover LetterCLINICAL GASTROENTEROLOGY AND HEPATOLOGY strongly en-courages authors to suggest 3 to 4 referees (include the mailingaddress, electronic address, phone, and fax numbers) and theAssociate Editor they believe best qualified to review theirpaper. Authors may also list a non-preferred Associate Editorand non-preferred referees, but the ultimate selection of anAssociate Editor and referees is at the sole discretion of theEditor and Associate Editor, respectively.

State reasons for deviations, if any, from standard format and clarifyany potential conflict related to the exclusive nature of the publica-tion. The cover letter must also categorize the manuscript into one oftwo groups: Alimentary Tract or Liver/Pancreas/Biliary.

Manuscript Preparation

Submission. Submit 1 complete manuscript typed in 12-pointfont size and double-spaced with 1-inch margins. Limit of 15pages (approximately 4000 words) for original articles or about 20pages (5000–6000) words for review articles, including tables,figures, and references (2 figures � 1 typed page). All manuscriptssubmitted to CLINICAL GASTROENTEROLOGY AND HEPATOLOGY

are made available for online review. Authors may submit theirmanuscripts, with figures and tables, electronically via our web-

site, http://www.editorialmanager.com/cgh. Complete instruc-tions for online submission are located on the website.

Note: If you have submitted your manuscript electronically,you do not need to submit a hard copy. If your manuscript isaccepted you will be asked to provide the editorial office withhard copies of the missing materials.

Other Enquiries. Visit http://authors.elsevier.com/trackpaper.html for the facility to track accepted articles and set up e-mailalerts to inform you of when an article’s status has changed.The Author’s Homepage (http://www.elsevier.com/authors)also provides detailed artwork guidelines, copyright informa-tion, frequently asked questions, and more.

Contact details for questions arising after acceptance of anarticle, especially those relating to proofs, are provided afterregistration of an article for publication.

Manual Submission. Manual submissions may be mailed to:AGA Institute, 4930 Del Ray Avenue, Bethesda, MD 20814,Attention: CGH Editorial Office. Authors of manual submissionswill incur a nonrefundable charge of $75. Manual submissionsmust be accompanied by a check (payable to the AGA Institute),purchase order, or credit card information (card type, card num-ber, cardholder’s name, and expiration date).

Title PageTitle—Use no abbreviations. Limit: 120 characters with spaces.Short Title—Limit: 45 characters with spaces.Authors—Include first names of all authors and name and full

location of department and institution where work was performed.Grant Support—List grant support and other assistance.Abbreviations—List abbreviations alphabetically. (Note: In

general, the use of abbreviations is discouraged.)Correspondence—Provide name, complete address, e-mail ad-

dress, telephone number, and fax number of corresponding author.Financial Disclosures—All authors must disclose any finan-

cial arrangement(s) they may have with a company whoseproduct figures prominently in the submitted manuscript orwith a company making a competing product.

Writing Assistance—The names and funding source forindividuals who provided writing assistance must be listed.

AbstractLimit: 250 words. Do not use abbreviations, footnotes, or

references.Authors should submit a structured abstract of no more than

250 words organized into the following categories as applicable:Background & Aims: Describe the importance of the study

and the precise research objective(s) or study question(s).Methods: Methods should include information on the fol-

lowing aspects of study design when applicable. The methodssection may employ subheadings at the discretion of the author.

● Design—describe the basic study design, e.g., random-ized controlled trial, cross-sectional study, cohort study,case series, survey, etc.

● Setting—specify whether the study was conducted in aprimary or tertiary care setting, in an ambulatory careclinic or hospital, in the general community, etc.

● Participants—indicate the number of study subjects and howthey were selected, recruited, and assigned to the intervention.

INFORMATION FOR AUTHORS (Continued)

● Intervention—report the method of administration andduration of the intervention.

Results: Provide the main outcomes of the study, includingconfidence intervals or P values. Report the absolute values andrisk differences so that readers can determine the absolute, aswell as the relative, impact of the results.

Conclusions: State only conclusions that are directly sup-ported by the evidence and the implications of the findings.

Body of PaperDescribe ethical guidelines followed; cite approval of institu-tional human research review committee or animal welfarecommittee; describe in detail hazardous procedures or chemi-cals involved, including precautions observed.

Outline statistical methods used.Identify drugs and chemicals used by generic name (if trade-

marks are mentioned, give manufacturer name and city).

ReferencesCite references in order of appearance in text using super-

scripted Arabic numerals.Cite personal communications and unpublished data di-

rectly in text without being numbered.Conform abbreviations to those used in Index Medicus.Conform style and punctuation to CLINICAL GASTROENTER-

OLOGY AND HEPATOLOGY requirements:

Article (list 3 authors followed by et al):13. Meltzer SJ, Ahnen DJ, Battifour H, et al. Protooncogene

abnormalities in colon cancers and adenomatous polyps.Gastroenterology 1987;92:1174–1180.

Book:18. Day RA. How to write and publish a scientific paper.

Philadelphia: Institute for Scientific Information, 1979.

Article in Book:22. Costa M, Furness JB, Llewellyn-Smith IF. Histochemistry

of the enteric nervous system. In: Johnson LR, ed. Phys-iology of the gastrointestinal tract. Volume 1. 2nd ed.New York: Raven, 1987:1–40.

TablesTables should be prepared without the use of tabs; most

table editor programs can be uploaded successfully. If yourtable contains decimal fractions, please round your numbers totwo places after the decimal point.

FiguresImages: Images can be clinical, pathologic (gross or micro-

scopic), endoscopic, or radiographic. They should be of highquality (300 dpi or greater, clear, and in good focus) andillustrate the diagnosis well.

Photographs: Photographs of identifiable patients must beaccompanied by written permission to publish from the patient.

Line Art and Graphs: Graphs, charts, and other line art may bereformatted and/or redrawn by our Medical Illustration Department(if needed) for consistency with the overall style of the AGA Institutejournals. Please be sure that any graphs or line art you submit are ata resolution of at least 150 dpi so that they are readable to reviewers.

Cost: Authors will be required to pay for the printing of colorfigures ($650 for the first color figure and $100 each for additionalfigures). If the manuscript is reviewed with color figures, it must bepublished with color figures with printing fees paid for by the author.If the author does not wish to pay for printing color figures, then the

authors upload grayscale or black-and-white files only to allow reviewof the data as they will ultimately be published.

Figure Legends: Please do not embed or flatten the textinto the image files. Figure legends should be typed andsubmitted in .rtf (rich text format). This text will be refor-matted in the style of the AGA Institute journals.

Accepted Figure File Formats: We support the followingamong dozens of file formats: .bmp, .gif, .jpg, .pbm, .pcx, .png, .tif,.eps, .xbm, .psd, and .tga files. When sending image files, please donot embed them in Word. You may submit mixed file formats(image1.jpg, image2.tif, image3.eps, etc.).

Preferred Figure File Formats: .tiff, .psd, and .jpg. If youhave created Photoshop image files containing separate layers witharrows or text, please send us the layered files (unflattened).

Image File Formats not Supported at this Time: Chem-Draw, CorelDraw, Canvas, FreeHand, Excel, SigmaPlot, QuarkXpress, and Equation Editor. You may export image files fromthese programs as PDF, JPEG, or other acceptable file formats.

File Naming Convention: Figures should be named con-secutively such as “figure 1.tif,” “figure 2.jpg,” etc., with thefile extension appended (.tif, .jpg, .eps, etc). Each figure shouldbe saved as a separate electronic file.

Color Figures: Files should be submitted in the CMYK colorspace. Authors are encouraged to present color figures in a mannerthat will allow the data to be interpreted by colorblind readers.CLINICAL GASTROENTEROLOGY AND HEPATOLOGY suggests thatauthors present dual labeled images in green and magenta ratherthan in green and red. See the website if the Jfly data depositoryfor Drosophila researchers (http://jfly.iam.u-tokyo.ac.jp/color/) formore information on how to make figures and presentationsintelligible for a colorblind audience.

Font: If your figures include text, an 8 to 10 point font shouldbe used. Acceptable fonts are “sans serif” fonts such as Arial,Helvetica, and Myriad. Examples of unacceptable fonts (“serif”fonts) are Times, Palatino, and Georgia. Lettering should beginwith an upper case letter, followed by lower case lettering.

Multiple Panel Figures: Do not mount multiple part figures.Please submit each panel (image) separately. However, you maysubmit a multiple panel version to suggest the order in which youwould like the panels arranged. You may also include a written,suggested layout. Each individual panel should be of the highestpossible quality (300 dpi or higher) at actual print size.

Reporting Clinical TrialsAs of January 1, 2008, it is mandatory for authors to provide fullregistration of their clinical trial(s). A clinical trial is defined as anyresearch project that prospectively assigns human subjects to inter-vention and comparison groups of study the cause-and-effect relation-ship between a medical intervention and a health outcome. The trialmust have at least one prospectively assigned concurrent control orcomparison group in order to trigger the requirement for registration.

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY has adoptedthe recommendations put forth by the International Committeeof Medical Journal Editors (ICMJE). For more information on theICMJE recommendations, please go to http://www.icmje.org/.Appropriate online trial registries include: www.clinicaltrials.gov, www.isrctn.org, www.umin.ac.jp/ctr/index.htm, www.actr.org.au, www.trialregister.nl, or any primary registers that participatein the World Health Organization’s International Clinical TrialRegistry Platform. The clinical trial registry URL and the clinicaltrial number must be included in the body of the manuscript and

INFORMATION FOR AUTHORS (Continued)

must be provided in the manuscript management system uponsubmission.

Reporting Meta-Analyses of Genetic StudiesAs of January 1, 2008 CLINICAL GASTROENTEROLOGY AND HEPATOLOGY

will require all metaanalyses of genetic studies to be registered and tofollow the Human Genome Epidemiology Network™ (HuGENet)guidelines. To review the guidelines, go to http://www.cdc.gov/genomics/hugenet/reviews/guidelines.htm/.

Image of the MonthImage of the Month presents a striking clinical image that ismeant to challenge and inform readers. The Image of the Monthis presented as a short case report including up to three images.The section is intended to illustrate and teach important medicalpoints. If you would like to submit an image for publication inthe Journal, please follow the instructions below.

● Images can be either clinical, pathologic (gross or microscopic),endoscopic, or radiographic. They should be of high quality andillustrate the diagnosis well. Images for hardcopy submissionshould be original color or black and white photographic or digitalprints 5 � 7 inches in size and should be submitted in duplicate.

● Arrows or other symbols, or label identifiers should bemarked only on the second print. The back of each printand slide must be labeled with the last names of thecontributors and an arrow indicating the top of image.

● The case should be described in one typed, double-spaced pageor less. Format should be as follows: Short pertinent history,physical examination and laboratory findings, and initial clinicalcourse. The image(s) (1 to 2, no more than 2) should then bedescribed with all labeled structures explained in the text.

● The answer should discuss the image findings and the diagnosisin no more than one double-spaced, typed page. The diagnosisand discussion should make an important medical teachingpoint and include from 1 to 3 pertinent references. Informationregarding the specific patient in terms of follow-up and responseto therapy should be given as appropriate.

● No more than three authors are allowed on each submission.Contributors must provide their names, addresses, phone, ande-mail addresses. Contributors must sign and return the copyrightform which assigns copyright to the AGA Institute and attest thatthe figure has not been submitted or published elsewhere.

● Image of the Month submissions must be submittedonline at http://www.editorialmanager.com/cgh.

Electronic Image of the MonthDue to a high number of Image of the Month submissions and theincreasing use of video in reports on clinical cases, CLINICAL GASTROEN-TEROLOGY AND HEPATOLOGY publishes Electronic Image of the Month,where some accepted images or those that include video clips can bepublished online only. When an image is accepted for the ElectronicImage of the Month section, authors will receive a decision letter request-ing approval to publish their article online only. If an author does notagree to these terms the article will not be considered further for publi-cation. If authors choose to have their accepted image published in theElectronic Image of the Month section, their article will be posted on ourwebsite, www.cghjournal.org, within 2 to 3 months.

To submit your Image of the Month to CLINICAL GASTRO-ENTEROLOGY AND HEPATOLOGY, log-on to http://www.

editorialmanager.com/cgh. Authors are required to follow theImage of the Month guidelines provided above.

Letters to the EditorLetters to the Editor offer opinions on manuscripts submitted to CLINICAL

GASTROENTEROLOGY AND HEPATOLOGY. Text should not exceed 250words with a limit of 3 references; no more than 3 authors are allowed oneach submission. All letters become the property of CLINICAL GASTRO-ENTEROLOGY AND HEPATOLOGY and are subject to editing by the editors.Letters commenting on manuscripts are sent to the authors of thosemanuscripts for a response. Letters are selected for their importance andnot all letters submitted can be published. Letters should be submittedwithin a month from the publication of the manuscripts.

Brief CommunicationsBrief Communications are short reports of preliminary or limitedresults of original research, observations, or case series on the causes,mechanisms, diagnosis, course, treatment, and prevention of digestivediseases. The formatting guidelines for brief communication are asfollows: Abstract must not exceed 175 words, manuscript body mustnot exceed 1500 words, 1–2 figures or tables, and 15–20 references.

Case ReportsThe journal will publish only a select number of high quality andcritical case reports. Authors have the option of reformatting theircase report as a Brief Communication or Image of the Month.

PublicationAccepted manuscripts are sent to the publisher, Elsevier, soonafter acceptance. Once authors have had the opportunity to reviewgalleys of their manuscripts, these author-corrected proofs will beuploaded to Articles in Press on www.cghjournal.org and will beindexed on PubMed, several weeks ahead of print.

Manuscripts are copyedited to make them consistent withJournal style; if a particular section in the manuscript is not clearor requires additional information, the copy editor will directquestions to the author. These questions, or author queries, willappear in the margins of the proofs that are sent to the author.

The time between acceptance and publication is currentlyabout 3 months. The corresponding author can expect proofs ofthe article approximately 2 months after acceptance. Authorsreceive proofs for the primary purpose of checking the accuracyof the typesetting; authors are not to revise or rewrite theirarticles at this stage. If after acceptance of their paper, authorsbecome aware of important information they believe should beadded to their manuscript, they should contact the editor ofCLINICAL GASTROENTEROLOGY AND HEPATOLOGY.

Authors are required to return proofs to the publisher within 48hours.

Reprints. Forms for ordering article reprints will be sent with proofsto authors and should be returned with the corrected proofs. Authors donot receive free reprints, and therefore are responsible for ordering theirown reprints (minimum order, 100) from the publisher.

Editorial OfficeThe address for the submission of manuscripts or correspon-dence is: C. Mel Wilcox, MD, Editor, CLINICAL GASTROEN-TEROLOGY AND HEPATOLOGY, AGA Institute, 4930 Del RayAvenue, Bethesda, Maryland 20814. E-mail: cgh@gastro.org;tel: (301) 654-2055 ext. 683.© 2008 by the AGA Institute

INFORMATION FOR AUTHORS (Continued)

Copyright Assignment, Authorship Responsibility, NIH Funding, Financial Disclosure,Institutional Review Board/Animal Care Committee Approval, and Sponsorship

Copyright Assignment. In consideration of the American Gastroenterological Association (AGA) Institute (the “AGA Institute”)taking action to review and credit the below-identified submission (the “Manuscript”), and for other valuable consideration, the receipt andsufficiency of which is hereby acknowledged, the undersigned authors and/or creators (the “Authors”), jointly and severally, hereby transfer,convey, and assign to the AGA Institute, free and clear of any liens, licenses or encumbrances, the entire right, title, and interest in and to theManuscript throughout the world, including without limitation in and to any and all copyrights for the Manuscript (including but not limitedto rights to copy, publish, excerpt, collect royalties and make derivative works) in print, electronic, Internet, broadcast, and all other forms andmedia now or hereafter known, and for any and all causes of action heretofore accrued in Authors’ favor for infringement of the aforesaidcopyrights, to have and to hold the same unto the AGA Institute, its successors and assigns, for and during the existence of the aforesaidcopyrights, and all renewals and extensions thereof. At any time and from time to time hereafter, the Authors shall upon the AGA Institute’swritten request take any and all steps and execute, acknowledge and deliver to the AGA Institute any and all further instruments and assurancesnecessary or expedient in order to vest the aforesaid copyrights and causes of action more effectively in the AGA Institute. The Authors retainthe nonexclusive permission to use all or part of the Manuscript in future works of their own in a noncompeting way, provided proper copyrightcredit is given to the AGA Institute. Should the AGA Institute finally determine that it will not publish the Manuscript, the AGA Instituteagrees to assign its rights therein back to the Authors. (Note: material prepared by employees of the federal government in the course of theirofficial duties may not be copyrightable.)

Authorship Responsibility. I, the undersigned Author, certify that I have participated sufficiently in the intellectual content, theanalysis of data, if applicable, and the writing of the Manuscript, to take public responsibility for it. I have reviewed the final version of theManuscript, believe it represents valid work, and approve it for publication. As an Author of this Manuscript, I certify that, except to the extentexpressly credited to others in the text of the Manuscript: the entire Manuscript is an original creation of the Authors; and none of the materialin the Manuscript has been published previously, is included in another manuscript, or is currently under consideration for publication elsewhere.I also certify that this Manuscript has not been accepted for publication elsewhere, and that I have not assigned, licensed, or otherwise transferredany right or interest in the Manuscript to anyone. Moreover, should the AGA Institute or the editors of CLINICAL GASTROENTEROLOGYAND HEPATOLOGY request the data upon which the Manuscript is based, I shall produce it. Authors are responsible for applying forpermission for both print and electronic rights for all borrowed materials and are responsible for paying any fees related to the applications ofthese permissions.

Institutional Review Board/Animal Care Committee Approval. I, the undersigned Author, certify that my institution has approvedthe protocol for any investigation involving humans or animals and that all experimentation was conducted in conformity with ethical andhumane principles of research.

Sponsorship. I, the undersigned author, certify that I had full access to all of the data in this study and I take responsibility for theintegrity of the data and the accuracy of the data analysis.

Manuscript title:

Signatures: Each Author must sign and date this statement and assignment. In the case of a work made for hire, the employer(s) must also sign.For example, for any Manuscript including any portion created in the course of employment for another (whether as a regular employee or asan independent contractor) requires the signature of the relevant employer(s).

Print Name: Date: Print Name: Date:

Print Name: Date: Print Name: Date:

Employer signature(s) as Author (required for works made for hire): Employer signature(s) as Author (required for works made for hire):

(Employer) (Employer)

By: By:

Print Name: Date: Print Name: Date:

▫ If this Manuscript exists in the public domain because it was written as part of the official duties of the Authors as employees of the U.S. government, check this box.

National Institutes of Health (NIH) Funding. My manuscript was supported in part, or in whole, by the NIH. In accordance with the NIHPublic Access Policy, I would like to request that my manuscript, should it be accepted for publication, be submitted to PubMed Central (PMC).I understand that my manuscript will therefore be freely accessible by the public via PMC twelve months from the date of its publication.▫ I would like my manuscript automatically submitted to PMC, should it ultimately be accepted. My NIH grant number is

Financial Disclosure. Check the appropriate box and sign where indicated. All Authors must sign one of the statements below.▫ I, the undersigned Author, certify that I have no financial arrangements (e.g., consultancies, stock ownership, equity interests, patent-licensingarrangements, research support, major honoraria, etc.) with a company whose product figures prominently in the submitted manuscript or with a companymaking a competing product except as disclosed on a separate attachment. All funding sources supporting the work are acknowledged on the title page.

▫ I, the undersigned Author, certify that I have included on the title page of the manuscript any financial arrangements (e.g., consultancies,stock ownership, equity interests, patent-licensing arrangements, research support, major honoraria, etc.) that I have with a company whoseproduct figures prominently in the submitted manuscript or with a company making a competing product. All funding sources supporting thework are acknowledged on the title page.