March 2002 SELECTED SUMMARIES 831

1084, Lab Invest 1999;79:459-465, J p n J Cancer Res 1996;87:843- 848, Proc Natl Acad Sci U S A 1994;91:1858-1862, Jpn J Cancer Res 1996;87:1153-1159). Because CDH1 is a tumor suppressor gene, inactivation of this protein is believed to promote tumor formation. Although biallelic inactivation can occur through somatic mutations or allelic imbalance events (loss of heterozygosity), aberrant CpG dinucleotide methylation of the CDH1 promoter has also been shown to inactivate E-cadherin in both sporadic and inherited diffuse gastric cancer (Nat Genet 2000;26:16-17, Oncogene 2001;20:1525-1528).

However, despite our understanding of the molecular biology of E-cadherin and our recognition of CDH1 germline mutations as the underlying cause of HDGC, the optimal clinical management of HDGC family members is still unclear. The studies by Huntsman et al. and Chun et al. have shed light on this subject and have provided support for the role of prophylactic gastrectomy in the management of at least some of these families. Similar to the debate surrounding the role of prophylactic organ removal for other hereditary cancer predisposition syndromes such as hereditary breast ovary cancer syn- drome, which is caused by BRCA1 and BRCA2 germline mutations, the role of prophylactic gastrectomy in the management of HDGC families has been controversial (Cancer Genet Cytogenet 2000;122: 1-6, J Med Genet 1999;36:873-880). As in HDGC, prophylacti- cally removed breast and ovary specimens from BRCA mutation carriers have been shown to contain occult foci of malignant cells, even in the setting of negative mammographic and pelvic ultrasound examinations (J Clin Oncol 2000;18:2728-2732, Arch Pathol Lab Med 2000;124:378-381). Furthermore, prophylactic bilateral mas- tectomy reduces the risk of breast cancer by 90% in at-risk individ- uals but is also associated with negative effects on self-esteem and sexual relationships (Nat Med 2000;6:1024-1028). The effect of prophylactic gastrectomy on reducing the risk of gastric cancer in CDH1 mutation carriers is unknown at this time, but the results from Huntsman et al. and Chun et al. suggest it may provide a significant reduction in risk. Nonetheless, the morbidity of gastrectomy, which includes weight loss, lactose intolerance (50% of patients), fat mal- absorption and steatorrhea (66%-100% of patients), dumping syn- drome (20%-30% of patients), bacterial overgrowth, postprandial fullness, and iron and B12 vitamin deficiencies, presents a significant disincentive to embracing this as a universal management strategy (Acta Chir Scand 1988;154:37-41, Surgery 1990;108:488-494, Ann Surg 1975;182:415-429). Furthermore, the families who par- ticipated in the studies by Huntsman et al. and Chun et al. all had highly penetrant cancer-susceptibility phenotypes, which shifted the balance in favor of prophylactic gastrectomy. Indeed, because HDGC is believed to be only 70% penetrant in most families, a universal policy of prophylactic gastrectomy would result in 30% of HDGC mutation carriers receiving a nontherapeutic gastrectomy (Hum Mu- rat 1999;14:249-255). Yet, despite these concerns, the low success rates of curing even early stage diffuse gastric cancer combined with the unproven value of any current surveillance techniques to identify preclinical gastric cancers is a strong argument that prophylactic gastrectomy may be a viable option for some HDGC family members. In the future, identification of families that could potentially benefit from prophylactic gastrectomy will require a careful determination of the family history in addition to CDH1 mutation analysis of the at-risk individuals.

There are several issues raised by these studies that will only be resolved with more extended follow-up. Although it is easily accepted that prophylactic gastrectomies should prolong the life expectancy of individuals undergoing this procedure, that outcome is, at present, not proven. In addition, studies have identified breast and colon cancer in mutation carriers in at least some HDGC families, and it is

possible that individuals who undergo prophylactic gastrectomy may still be at increased risk for cancers at distant sites (Hum Mutat 1999;14:249-255, Hum Mol Genet 1999;8:607-610). Another cur- rently unresolved issue surrounding the use of prophylactic gastrec- tomy in CDHI mutation carriers is the quality of life after gastrec- tomy, which may improve because of the reduction of anxiety about developing stomach cancer or worsen because of the side effects associated with the procedure. An additional major issue is whether surveillance modalities or preventive strategies can be identified that will obviate the need for prophylactic gastrectomy in the future. Possible strategies that may serve such a purpose include endoscopic surveillance with optical coherent tomography or molecular marker analysis of body fluids, such as gastric lavage fluid. In regard to prevention, it has been shown in a small study of 2 HDGC families that half of the tumors demonstrated methylation of the CDH1

promoter as the "second hit," and it is tempting to consider the potential value of prophylactic treatment of these mutation carriers with demethylaring agents, such as 5-azadeoxycytidine (Nat Genet

2000;26:16-17). Although outstanding issues remain regarding the management of

CDH1 mutation carriers in HDGC families, several recommendations can be made at this time. First, familial gastric cancer patients should be referred to a health care provider who can provide genetic coun- seling and testing to best identify individuals who are at enhanced risk for gastric cancer in these families. Second, the ability of surveil- lance regimens including upper endoscopy with biopsy, endoscopic ultrasonography, and/or chromoendoscopy is not proven and possibly of no value in identifying early and potentially curable gastric cancer. Third and final, prophylactic gastrectomy is a viable management strategy but should be reserved for those individuals who have undergone extensive genetic assessment and preoperative counseling. The rapidly unfolding story about HDGC that has transpired in the last 2 years has shown that the power of genetics can alter our understanding and management of this disease, however, additional questions have now been raised that need to be answered in turn.

WILLIAM M. GRADY, M.D.

RICHARD M. PEEK, JR., M.D.

Reply. Germline CDH1 mutations occur only in families with dif- fuse gastric cancer and not in families where index cases have the more common intestinal type gastric cancer (Cancer Res 1998;58:4086- 4089). The importance of including careful histopathologic review when ascertaining families for genetic testing (J Med Genet 1999; 36:873-880) is highlighted by our accumulated experience. A total of 57 families have been analyzed to date: CDH1 mutations were found in 7 of 21 (33%) HDGC kindred and in none of the others. Recent publications may be of assistance in the management of these families. First, through the International Gastric Cancer Linkage Consortium (IGCLC), data from several pedigrees with HDGC caused by germline CDH1 mutation has been combined to estimate the penetrance of gastric and breast cancer in gene carriers (Gastroenter- ology 2001;121;1348-1353). The lifetime accumulated risk for women with a germline truncating CDH1 mutation was calculated to be 83% (95% confidence interval [CI], 58%-100%) and 67% for men (95% CI, 39%-99%). Women have a lifetime risk of breast

cancer of 39% (95% CI, 12%-84%). Although all 10 prophylactic gastrectomy specimens reported had

microscopic foci of diffuse gastric cancer, and no proven screening technology exists, a decision to undergo prophylactic surgery does not need to be rushed because the biological behavior of these early lesions is not known (N Engl J Med 2001;344:1904-1909, Cancer 2001; 92:181-187). This is particularly true for female CDH1 mutation

832 SELECTED SUMMARIES GASTROENTEROLOGY Vol. 122, No. 3

carriers of reproductive age because it is uncertain whether a preg- nancy could be maintained after gastrectomy. Weight loss is an inevitable consequence of gastrectomy, however, the procedure may have surprisingly little effect on eventual lean body mass (Surgery 2001;130:612-619, World J Surg 1997;21:416-421). All 6 pa- tients in our families who underwent gastrectomy have had to sig- nificantly modify their diet and eating habits to avoid dumping and diarrhea. However, this complication is not insurmountable, and all patients returned to work or full-time school within 4 months of the procedure (Surgery 2001;130:612-619).

In families with HDGC without E-cadherin mutations and in families with missense mutations in CDH1, we would not recom- mend prophylactic surgery at this moment, until penetrance in these families has been characterized.

The development of an animal model for HDGC would help in development of novel screening techniques and evaluating potential pharmaco-prevention strategies. Finally, as Drs. Grady and Peek suggest, "more questions" are inevitable consequences of research on any new clinical entity. We hope, through the work of the IGCLC, to accumulate the required data for evidence-based management of affected families.

CARLOS CALDAS, M.D. DAVID HUNTSMAN, M.D.

PREVENTION OF VARICES REBLEEDING: ARE DRUGS BETTER AFTER ALL?

Villanueva C, MinanaJ, Ortiz F Gallego A, Soriano G, Torras X, Sainz S, Boadas J, Cusso X, Guarner C, Balanzo J (Department

of Gastroenterology, Hospital dela Santa Creui sant pau, Bar-

celona, Spain). Endoscopic ligation compared with combined

treatment wi th nadolol and isosorbide mononitrate to prevent

recurrent variceal bleeding. N Engl J Med 2001 ;345 :647-

655.

A randomized controlled trial was performed on 144 pa-

tients with cirrhosis who had been hospitalized for bleeding

esophageal varices. The patients were randomized to receive

either a combinat ion of medical therapy (72 patients) or esoph-

ageal variceal band ligation (72 patients). Ligation was re-

peated every 2 -3 weeks unti l the varices were eradicated.

Alternatively, the medical group patients received a combina-

tion of Nadolol (average dose, 96 mg) and isosorbide (average

dose, 66 mg). The primary end points were recurrent bleeding,

complications, and death. Dur ing a median follow-up period

of 21 months, the probabili ty of recurrence of rebleeding in the

medical group, in which 24 patients (33%) rebled, was sig-

nificantly less (P = 0.04) than in the l igation group, in which

35 patients (48.2%) bled. In the medical group, the rebleeding

rate was significantly less in those with a significant portohy-

potensive response to the medication (18% vs. 54%, P =

0.001). There were 9 major complications in the l igation

group, 7 bleeding esophageal ulcers and 2 aspiration pneumo-

nias, significantly more (P = 0.05) than the 2 patients in the

medication group, who both suffered from bradycardia and

dyspnea. O f interest, ascites developed in only 18.7% of pa-

tients wi th a portohypotensive response, compared with

66.6% (P < 0.001) in nonresponsive patients. Thir ty patients

(41.6%) in the l igation group died compared with 23 patients

(32%) in the medication group (P = 0.52), but significantly

(P = 0.01) only 3 patients with a hemodynamic response died.

Comment . Cirrhotic patients with large, grade 3 -4 esophageal varices are at great risk from life threatening bleeding. Therefore, the prevention of the initial bleed is of major importance because pro- phylactic medical therapy with beta blockers may actually reduce mortality (N Engl J Med 1991;324:1532-1538, Ann Intern Med 1992;77:59-70). Failing that, if the patient has already bled, then the prevention of rebleeding, and death, is a principal aim of hepa- tologists today. Until recently, the major forms of therapy were either the use of beta blockers (Lancet 1990;336:153-156) or repeated courses of endoscopic sclerotherapy to ablate the varices and improve survival (Gastroenterology 1989;96:1087-1092), or both (Gastroen- terology 1992; 102:1760-1763). Endoscopic band ligation was intro- duced principally to reduce the complication rate of sclerotherapy (Am J Surg 1990;159:21-26). Subsequently, with randomized con- trolled trials, it was also shown to be more effective than sclerotherapy to prevent rebleeding (N Engl J Med 1992;326:1527-1532, Lancet 1993;342:391-394, Ann Intern Med 1993;119:1-7), and this was confirmed by meta analysis (Ann Intern Med 1995;123:280-287). Now, meta analysis of further trials indicates that prophylactic liga- tion reduces initial bleeding (Hepatology 2001;33:802-807). This is especially important in centers where optimal supervision and fol- low-up of medical therapy is difficult (N Engl J Med 1999;340:988- 993). Thus, ligation has become the treatment of choice for the overall management of esophageal varices (Hepatology 1995;22:332- 354). At the same time, venodilatation by long-acting nitrates was shown to lower portal pressure in cirrhotic patients (Gastroenterology 1989;96:1110-1118) and to be as effective as beta blockers in preventing the first variceal hemorrhage (Gastroenterology 1993;104: 1460-1465). Longer term studies over several years showed that although this efficacy in preventing bleeding is maintained, nitrate therapy was associated with a worrying increase in deaths caused by hepatic failure (Gastroenterology 1997;113:1632-1639). After the demonstration that the portohypotensive effect of a combination of beta blocker and long-acting nitrate was greater than the effect of either drug alone, several studies showed that this combination was also superior in the prevention of variceal bleeding (Ann Intern Med 1991;114:869-873, Lancet 1996;348:1677-1681). At the same time, the present group showed that this drug combination was also more effective than sclerotherapy (N Engl J Med 1996;334:1624- 1629). The present trial has advanced further to show that this combination of beta blockers and long-acting nitrates are more effec- tive than "the state of art" therapy, band tigation, in the management and prevention of rebleeding in cirrhotic patients who have bled from varices. Because this observation, if valid, could result in a major change in our management, the study deserves close scrutiny.

In a comparison of 2 therapies, such as the present study, the first question to be asked is whether both therapies were carried out optimally. Looking at the investigators' first study, the efficacy of the same drug combination, at a median follow-up of 18 months, achieved a rebleeding rate of 25.6% (N Engl J Med 1996;334:1624- 1629), which is similar to the 33% that rebled at a median of 21 months in the present study. However, in the first study the rebleed- ing rate in the sclerotherapy group was 53.5%, at the same follow-up period. This is surprisingly similar to the present rate in the ligation group, in which the overall rebleeding rate was 49%, the vast majority from esophageal varices, but with a longer mean follow-up period of 2 years. How does this compare to the literature? In the first major study by Steigmann et al. (N Engl J Med 1992;326:1527-