Impact of the Reconstruction Method on Delayed GastricEmptying After Pylorus-Preserving Pancreaticoduodenectomy:A Prospective Randomized Study

Dietmar Tamandl • Klaus Sahora • Johannes Prucker •

Rainer Schmid • Jens-Juul Holst • Johannes Miholic •

Peter Goetzinger • Michael Gnant

Published online: 12 October 2013

� Societe Internationale de Chirurgie 2013

Abstract

Background Delayed gastric emptying (DGE) is of con-

siderable concern in patients undergoing pylorus-preserv-

ing pancreaticoduodenectomy (PPPD). Prolonged hospital

stay, increased cost, and decreased quality of life add on to

interventions needed to treat DGE. This study was con-

ducted to determine if performing duodenojejunostomy via

the antecolic rather than the retrocolic route improved

incidence of DGE.

Methods Patients undergoing PPPD between April 2007

and November 2009 were randomized for either antecolic

or retrocolic reconstruction of the duodenojejunostomy.

DGE was then assessed by clinical criteria on postoperative

day (POD) 10. A paracetamol absorption test was also

administered with a liquid meal, and serial plasma levels of

intestinal peptides were measured.

Results Overall, 64 patients were amenable for analysis:

36 in the antecolic group and 28 in the retrocolic group.

The incidences of DGE on POD 10 were 17.6 and 23.1 %

(antecolic vs. retrocolic, respectively) (p = 0.628). The

two groups did not differ in regard to their median

(interquartile range) postoperative hospital length of stay

[13.0 (10.0–17.5) vs. 12.5 (11.0–17.0) days; p = 0.446],

time to regular diet [5 (5–7) vs. 5 (4–6) days; p = 0.353],

or morbidity (52.9 vs. 50.0 %; p = 0.777). The median

length of nasogastric tube decompression was similar in the

two groups [4 (3–7) vs. 3 (3–5) days; p = 0.600]. Levels of

paracetamol and glucagon-like peptide-1 were markedly

decreased in patients with DGE.

Conclusions Antecolic reconstruction after PPPD does

not improve the occurrence/the incidence of DGE and is

similar to retrocolic reconstruction with regard to second-

ary outcome parameters.

Introduction

Pancreaticoduodenectomy (PD) is the standard operation

for tumors of the pancreatic head, ampulla, and distal

common bile duct [1, 2]. Pylorus-preserving pancreatico-

duodenectomy (PPPD) is a procedure with equal short- and

long-term outcomes compared to the classic Whipple

pancreaticoduodenectomy, which has been demonstrated

by several randomized trials [3, 4]. However, PPPD seems

to be associated with a shorter operation time and less

intraoperative blood loss [4]. The idea of organ-preserving

surgery has led to the fact that whenever possible PPPD is

the standard operation instead of PD at many centers [5].

Delayed gastric emptying (DGE) is a common compli-

cation after PPPD, occurring in 5–70 % of patients after the

operation [5–9]. This high variation in incidence is due to

different use of definitions throughout the literature [6, 10,

11] and the fact that most studies dealing with this topic are

retrospective in nature. DGE is usually not a life-threat-

ening complication, but it prolongs hospital stay, decreases

quality of life, and increases treatment costs [6, 12].

D. Tamandl (&) � K. Sahora � J. Prucker � J. Miholic �P. Goetzinger � M. Gnant

Department of Surgery, Medical University of Vienna,

Waehringer Guertel 18-20, 10990 Vienna, Austria

e-mail: dietmar.tamandl@meduniwien.ac.at

M. Gnant

e-mail: michael.gnant@meduniwien.ac.at

R. Schmid

Clinical Institute of Medical and Chemical Laboratory

Diagnostics, Medical University of Vienna, Vienna, Austria

J.-J. Holst

Department of Biomedical Sciences, The Panum Institute,

University of Copenhagen, Copenhagen, Denmark

123

World J Surg (2014) 38:465–475

DOI 10.1007/s00268-013-2274-4

Several attempts to alleviate DGE have been reported,

most of which are applied postoperatively [13]. To our

knowledge, there are three randomized controlled trials

(RCTs) dealing with a modification of the operative pro-

cedure to improve the incidence of DGE [6, 14, 15]. The

first trial conducted in Japan was available at the time of

setting up the current study. It randomized patients to

undergo either antecolic or retrocolic reconstruction of the

duodenojejunostomy and found striking improvement of

DGE in the antecolic group (5 % antecolic vs. 50 % ret-

rocolic). Together with retrospective evidence [5, 7–9],

their results led to the belief in the surgical community that

antecolic reconstruction seems to be a safer method with

respect to decreased DGE. However, some shortcomings in

the RCT and some natural limitations in the retrospective

analyses raises the question of whether these findings can

be applied to all patients.

The paracetamol absorption test is a reliable method for

determining gastric emptying in healthy individuals [16,

17] and surgical patients [18, 19]. Because paracetamol is

absorbed only in the small intestine, gastric emptying is the

rate-limiting step in paracetamol uptake [16]. Therefore,

the profile of paracetamol concentration in plasma can be

used to assess how quickly a meal passes through the

stomach. Also, plasma levels of intestinal peptides, which

are secreted upon nutrient ingestion, have been used to

analyze the regulation of gastric emptying in previous

studies [19–21]. Glucagon-like peptide-1 (GLP-1), an in-

cretin that is secreted into the bloodstream within minutes

after nutrients reach the small intestine, has been shown to

correlate with gastric emptying [20]. GLP-1 levels rapidly

increased after a meal in patients who exhibited dumping

as a result of rapid gastric emptying after distal gastrec-

tomy, whereas patients without dumping showed an

attenuated peak of GLP-1 secretion [21]. Peptide YY (PYY

3-36) is secreted into the bloodstream upon contact of

nutrients with L-cells in the distal intestine [22]. It is

therefore a ‘‘late’’ indicator of gastric emptying.

We designed the current trial to determine if antecolic

reconstruction of the duodenojejunostomy was able to

decrease the incidence of DGE using clinical criteria. In

addition, patients had a test meal with a paracetamol

absorption test at postoperative day (POD) 10, which fur-

ther helped to elucidate the biology of gastric emptying in

these patients.

Materials and methods

This open, randomized trial was performed at a single

center in Austria. The study was conducted between April

2007 and November 2009 at the General Hospital of

Vienna, a tertiary referral center of the Medical University

of Vienna (MUV). The local ethics committee approved

this study (EK-No. 539/2006). All patients gave written

informed consent.

Eligible participants were adults between the ages of 18

and 90 years undergoing PPPD for cancer either of the

pancreatic head/uncinate process/ampulla or distal com-

mon bile duct or with a radiographically suspicious solid or

cystic tumor requiring pancreaticoduodenectomy. Exclu-

sion criteria were as follows: (1) distant metastases; (2)

locally unresectable tumors (arterial involvement or[1808involvement of the portal vein or superior mesenteric vein);

(3) invasion of the stomach; (4) prior surgical resection of

the stomach or duodenum. Exclusion criteria for the test

meal at POD 10 were (1) clinically significant anastomotic

dehiscence of any anastomosis or (2) postoperative pan-

creatitis that had been present for more than 10 days and

(3) hypersensitivity to paracetamol.

Patients were randomized preoperatively to undergo

either antecolic or retrocolic reconstruction of the duode-

nojejunostomy after the resection. The surgical procedure

and postoperative management were conducted according

to standard operating procedures of the MUV. On POD 10,

patients received a test meal consisting of 200 ml Fresubin

Protein Energy� (Fresenius Kabi, Graz, Austria), 50 ml

cream (36 % fat), and 1 g paracetamol (Ratiopharm Arz-

neimittel Vertriebs-GmbH, Vienna, Austria) according to a

protocol described by Strommer et al. [19]. This 250-ml

liquid meal was composed of 26.5 g carbohydrates, 21.1 g

proteins, and 31.4 g fat. It contained a total of 472 kcal. If

patients were discharged before POD 10, the meal was

administered at the day of discharge. Patients were fasted

overnight and received the test meal instead of breakfast on

POD 10. They were not allowed to take metoclopramide or

erythromycin before the meal for a period of 12 h. In case

of nausea, intravenous ondansetron was administered. All

patients were able to drink the meal within 5 min. Patients

remained sitting in bed and were not allowed to take

anything per mouth throughout the measurement period.

Blood samples were obtained before and 15, 30, 60, and

90 min after the meal. After that, patients resumed their

respective diet as planned for that day.

On POD 10, clinical parameters were observed and

recorded, including the type of diet, day of first full diet,

days of nasogastric tube (NGT) decompression, reinsertion

of an NGT, daily and total output of the NGT, and utili-

zation of prokinetic drugs (metoclopramide and erythro-

mycin). Data were recorded for up to 90 days after surgery

regarding the hospital length of stay (LOS), postoperative

complications, and mortality after discharge. Complica-

tions were graded using the Clavien-Dindo [23] staging

system. Other data acquired were demographic variables,

such as sex and date of birth, operating surgeon, histology

of removed specimen, and operating room LOS.

466 World J Surg (2014) 38:465–475

123

Surgery and postoperative care

After completion of the pancreaticoduodenectomy, the

reconstruction was performed as follows: The first jejunal

loop was brought up through a separate incision in the

transverse mesocolon usually to the right to the middle colic

vein. An end-to-side pancreaticojejunostomy was performed

in a single-layer fashion with 5-0 PDS-II� sutures (John-

son&Johnson Medical Products GmbH, Vienna, Austria).

The bile duct was then anastomosed end-to-side to the same

loop of jejunum with one layer of 5-0 PDS-II�. The stapled

duodenal remnant (usually 1–3 cm) was reopened, and the

pylorus was dilated with a blunt clamp from the luminal side.

Either antecolic or retrocolic end-to-side duodenojejunos-

tomy was then performed according to the randomization. A

single-layer anastomosis was performed, with the type of

suture and technique used (interrupted vs. running) at the

discretion of the operating surgeon. Silicone drains were

placed at the site of the pancreatic and bile duct anastomosis

and were managed as described below.

After surgery, patients typically recovered in the inten-

sive care unit overnight and were transferred to the floor on

POD 1. Patients received antibiotics only perioperatively.

They were not given somatostatin or prokinetics (meto-

clopramide and erythromycin) postoperatively on a routine

basis. Clear liquids were started on POD 1, and the diet was

advanced as tolerated by the patient. The NGT was

removed in the operating room and was reinserted only in

case of repeated vomiting or abdominal distension and the

inability to ingest food. Amylase and lipase were measured

in the drainage fluid on POD 5. If one of those levels

(amylase or lipase) was elevated, the drains remained in

place regardless of output. Serial measurements of pan-

creatic enzymes were performed at the discretion of the

treating surgeon, and the drains were removed if there was

no evidence of fistula.

Laboratory tests

Paracetamol plasma levels

For the paracetamol measurements, 5 ml of blood was

drawn into heparin-coated collection tubes (Vacutainer;

Becton–Dickinson Austria, Schwechat, Austria) at the

indicated time points. Paracetamol levels were measured as

described in an earlier study [24] (available upon request).

GLP-1 and PYY (3–36) plasma levels For the GLP-1 and

PYY (3–36) measurements, 4 ml of blood was drawn into

collection tubes containing 15 % K-EDTA and 250 kIU

aprotinin (Vacutainer) at the indicated time points, which

were placed on ice immediately. After centrifugation at

2,200 g for 10 min, plasma samples were aliquoted and

stored at -80 �C for later analysis. The methods to

measure plasma concentrations of GLP-1 and PYY (3–36)

have been described elsewhere in detail [25] (available

upon request).

Primary endpoint

The primary endpoint of this study was delayed gastric

emptying as defined by clinical criteria on POD 10. DGE is

defined as the NGT remaining in place beyond POD 10 and

one of the following criteria: (1) emesis after NGT

removal; (2) NGT reinsertion; (3) failure to progress with

the diet; (4) use of prokinetics after POD 10. If the NGT

was removed before day 10, two of the previous criteria

had to be fulfilled to qualify for DGE to be present.

Secondary endpoints

Secondary endpoints for this study included evaluation and

results of the following: (1) gastric emptying determined

by the paracetamol absorption test; (2) the kinetics of

intestinal peptides (GLP-1 and PYY) after ingestion of a

test meal; (3) postoperative LOS; (4) morbidity and

mortality.

Statistics and sample size calculation

The hypothesis was that antecolic reconstruction of the

duodenojejunostomy decreased the incidence of DGE after

PPPD. Based on the single prospective trial on this topic

that was available during our study designing [6], we

estimated the rate of DGE in the control arm (retrocolic) at

50 %, whereas the rate of DGE in the antecolic arm was

estimated at 10 %. We arrived at these figures because the

reported 5 % in the former reported trial was lower than in

all but one of the previous retrospective series [5]. With a

power of 80 % and a two-sided alpha of 0.05, we calcu-

lated that we required 20 patients per arm to demonstrate a

clinical significant difference. It was planned to accrue

patients for this protocol until 40 patients had undergone

the paracetamol test. However, because of the high post-

operative dropout rate we decided to analyze the entire

cohort on an intention-to-treat basis and reported the results

of the biochemical analyses as a subcohort. Randomization

was conducted prior to surgery, patients with an uneven

birth date (e.g., February 1) were allocated to the antecolic

arm. Patients who had an even birth date were treated with

retrocolic reconstruction.

SPSS 17.0 solftware (SPSS, Chicago, IL, USA) was

used for statistical analyses. Continuous data are repre-

sented as the median and interquartile range unless other-

wise indicated and were compared using Student’s t test.

For dichotomous variables, a v2 test was used. To compare

levels of paracetamol or intestinal peptides, the area under

World J Surg (2014) 38:465–475 467

123

the curve (AUC) was calculated and was used as a con-

tinuous variable to compare between groups. A value of

p \ 0.05 was considered to indicate significance in all

analyses.

Results

Patient recruitment

Between April 2007 and November 2009, a total of 82

patients who were scheduled to undergo PPPD at the

Department of Surgery at the Vienna General Hospital

were assessed for this study (Fig. 1). Among them, 71 were

randomized to undergo reconstruction using either ante-

colic (n = 38) or retrocolic (n = 33) duodenojejunostomy.

Two and three patients, respectively, in the antecolic and

retrocolic groups were not operated on according to pro-

tocol because the attending surgeon thought that the

randomized reconstruction method was technically not

feasible or was not possible for anatomic reasons (n = 2,

retrocolic group). Hence, 36 patients underwent the allo-

cated intervention in the antecolic group and 28 patients in

the retrocolic group. On POD 10, DGE was assessed

according to clinical criteria. However, two patients in each

group could not be evaluated owing to early discharge,

leaving 34 patients in the antecolic and 26 patients in the

retrocolic group for clinical analysis of DGE.

A test meal including a paracetamol absorption test was

conducted on or around POD 10. A significant number of

patients (eight in the antecolic group, six in the retrocolic

group) withdrew consent for this test meal at the day of

intervention. In six patients it was due to inability to ingest

the meal (n = 2), technical issues (n = 3), or other reasons

(n = 1). In all, 22 patients in the antecolic group and 18

patients in the retrocolic group underwent the test meal and

paracetamol absorption test. Two samples could not be

processed because of logistic reasons and one sample did

Fig. 1 Consolidated Standards of Reporting Trials (CONSORT) flow chart of all screened and randomized patients. DGE delayed gastric

emptying

468 World J Surg (2014) 38:465–475

123

not have enough blood for analysis, Thus, 21 (antecolic)

and 16 (retrocolic) patient samples were available for

measurement of plasma intestinal peptides. These numbers

are summarized in Fig. 1.

Patient characteristics

Demographic and tumor-specific details according to the

reconstruction method are presented in Table 1. Age, sex,

and operation duration were similarly distributed between

groups. A nonsignificant trend of more patients (20 vs. 13)

with ductal adenocarcinoma was observed in the antecolic

group.

Primary endpoint

At POD 10, six patients in the antecolic group (17.6 %) and

six in the retrocolic group (23.1 %) had clinically evident

DGE (p = 0.628). There was no difference in the absolute

number of patients requiring NGT reinsertion (seven in each

group). The duration of decompression and the median daily

and total NGT output were similar in the two groups. Patients

in the antecolic arm resumed a regular diet at POD 5 (med-

ian), similar to patients in the retrocolic reconstruction arm.

There was no difference in administration of prokinetic

drugs between the two groups (Table 2).

Secondary endpoints

Eighteen (52.9 %) patients in the antecolic group devel-

oped some kind of postoperative complication compared to

13 (50.0 %) of patients in the retrocolic group (p = 0.777).

The nature and grading of complications is summarized in

Table 2. One patient in each group died within 90 days

after the operation. The median postoperative LOS was

also not different between the treatment groups (Table 2).

The characteristics of patients with DGE (n = 12)

compared to patients without DGE (n = 48) are depicted

In Table 3. Patients with DGE had more complications

than patients without DGE (12 vs. 19, p \ 0.001), although

there was no specific pattern of these complications. The

median hospital LOS was 17.5 days in patients with DGE

compared to 12.0 days in patients without DGE. As

expected, the DGE patients resumed their regular diet later,

received prokinetics longer, and had more days of NGT

decompression than patients without DGE.

Paracetamol absorption test and plasma levels

of intestinal peptides as surrogate parameters for DGE

A test meal also containing 1 g paracetamol was admin-

istered on POD 10 (see Methods). Paracetamol levels were

measured at baseline and 15, 30, 60, and 90 min after

ingestion of the meal. Patients with DGE had less uptake of

paracetamol, which indicates that the drug remained longer

in the stomach since paracetamol is solely absorbed in the

jejunum. At any time point after ingestion of the test meal,

the paracetamol levels were lower in patients with DGE

(Fig. 2a). The AUC in the group of patients with DGE was

lower than that in patients without DGE. The maximum

concentration of paracetamol was reached later when DGE

was present.

With respect to the reconstruction method, there was no

difference in the uptake or plasma levels of paracetamol.

The plasma levels were similar if compared at individual

time points, and the AUCs in the antecolic and retrocolic

groups were not different. A similar trend was observed for

GLP-1, which is secreted by intestinal L-cells when

nutrients reach the upper jejunum. Patients with DGE had a

slower rise of plasma levels and a lower AUC than patients

with no clinical evidence of DGE (Fig. 2b). There was no

difference in GLP-1 levels with regard to treatment groups.

Table 1 Patient characteristics

according to groups

IQR interquartile range, IPMN

intraductal papillary mucinous

neoplasm, SCN/MCN serous/

mucinous cystic neoplasm

* The p values are derived from

the v2 test for categoric

variables and from Student’s

t-test for continuous variables

Characteristic Antecolic group (n = 36) Retrocolic group (n = 28) p*

Age (years), median and IQR 67.1 (55.7–75.3) 65.4 (55.6–70.6) 0.688

Sex (M/F) 17/19 12/16 0.803

Histology: malignant/benign 28/8 20/8 0.561

Pathology

Ductal adenocarcinoma 20 13 0.267

Distal cholangiocarcinoma 1 1

Ampullary cancer 4 5

IPMN 2 0

SCN/MCN 1 1

Chronic pancreatitis 5 7

Other 3 1

Surgery duration (min), median

and IQR

273 (245–300) 280 (243–340) 0.970

World J Surg (2014) 38:465–475 469

123

For PYY, the AUC in patients with DGE was slightly

lower than in patients without DGE, but the difference did

not reach statistical significance (p = 0.060) (Fig. 2c). The

AUC for PYY was similar in the antecolic and retrocolic

groups. These findings are summarized in Table 4.

Discussion

We report the outcomes of a prospective study investigat-

ing the effect of antecolic versus retrocolic reconstruction

on the incidence of DGE in patients who underwent PPPD.

In contrast to the existing literature [5–9], we did not find a

difference between these groups (Table 5). Furthermore,

we performed a paracetamol absorption test and measured

plasma levels of intestinal peptides, which confirmed these

findings.

So far, one well-designed RCT from Japan has investi-

gated this topic and found a relevant improvement of DGE

in patients with antecolic reconstruction (5 vs. 50 % in the

retrocolic group). Some of the findings of their study,

however, limit applicability to our patients [6]. First, their

study had to be stopped early because of the extensive

benefit in the intervention group that ethically would not

allow proceeding with treatment in the control group. This

observed difference could also be due to temporary fluc-

tuations, which would have leveled out more during the

course of the trial. Second, the median postoperative stay in

Table 2 Parameters of delayed gastric emptying and postoperative morbidity according to treatment arms

Parameter Antecolic group (n = 34) Retrocolic group (n = 26) p

Section A: generala

No. of patients requiring

NGT

7 (20.6 %) 7 (26.9 %) 0.788

NGT (days) 4 (3–7) 3 (3–5) 0.600

Daily retention (ml) 474 (42–1,366) 800 (0–1,936) 0.482

Median total retention

(ml)

2,370 (300–4,100) 1,050 (0–5,810) 0.650

Section B: with vs.

without NGTaAll patients Patients with

NGT

Patients without

NGT

All patients Patients with

NGT

Patients without

NGT

p*

Regular diet resumed

(POD)

5 (5–7) 8 (3–8) 5 (4–6) 5 (4–6) 5 (5–13) 5 (4–6) 0.353**

Metoclopramide (days) 0 (0–3) 6 (3–10) 0 (0–1) 2 (0–3) 5 (3–7) 0 (0–3) 0.269**

Erythromycin (days) 0 0 (0–6) 0 0 0 (0–2) 0 0.576**

Hospital LOS (days) 13 (10.5–17.5) 17 (11–29) 12 (9–15) 12.5 (11–17) 17 (12–20) 12 (10.5–15.0) 0.446**

Section C: postoperative complications

Any complications 18 (52.9 %) 13 (50.0 %) 0.777

Grade I 2 (5.9 %) 1 (3.8 %)

Grade II 11 (32.3 %) 8 (30.8 %)

Grade III 4 (11.8 %) 3 (11.5 %)

Grade IV 0 0

Grade V 1 (2.9 %) 1 (3.8 %)

Delayed gastric

emptying

6 (17.6 %) 6 (23.1 %) 0.628

Bile leak 2 (5.9 %) 2 (7.7 %) 0.593

Pancreatic fistula 5 (14.7 %) 4 (15.4 %) 0.629

Intraabdominal abscess 2 (5.9 %) 2 (7.7 %) 0.491

Infection 7 (20.6 %) 1 (3.8 %) 0.157

Bleeding 2 (5.9 %) 1 (3.8 %) 0.552

Ileus 2 (5.9 %) 0 0.138

Other 1 (2.9 %) 1 (3.8 %) 0.420

Two patients in each group were not amenable for analysis of postoperative outcome (see Fig. 1)

NGT nasogastric tube, IQR interquartile range, POD postoperative day, LOS length of staya Unless otherwise stated, the results are expressed as the median and the interquartile range (IQR)

* The p values were derived from v2 test for categoric variables and from Student’s t-test for continuous variables

** These p values apply to the comparison of all patients

470 World J Surg (2014) 38:465–475

123

the antecolic group was 28 days compared to 48 days in

the retrocolic group, a number that is more than double the

reported average hospital stay after PPPD in Europe [5]

and even more if compared to that in the United States

[26].

Other studies dealing with this topic include the series of

Hartel et al. [5], who analyzed 100 patients between 1996

and 2001 while practicing in Bern, when retrocolic

reconstruction was always performed. They compared their

findings to a cohort of 100 patients with antecolic recon-

struction treated by the same surgeons in Heidelberg

between 2002 and 2003. They found an improvement in

their DGE rates from 24 to 5 % and concluded that this

difference was due to the antecolic reconstruction method.

It is, however, difficult to dissect all those differences in

perioperative management between two centers in two

Table 3 Postoperative

parameters in relation to

delayed gastric emptying

Unless otherwise stated, the

results are expressed as the

median and the interquartile

range (IQR)

Some patients had more than

one complication

DGE delayed gastric emptying

* The p values are derived from

the v2 test for categoric

variables and from Student’s

t-test for continuous variables

Parameter DGE according to clinical criteria p*

Yes (n = 12) No (n = 48)

NGT (days) 4 (3–7) 0 \0.001

Daily retention (ml) 887 (42–1,936) 0 \0.001

Total retention (ml) 2,875 (300–4,250) 0 \0.001

Regular diet resumed (POD) 8 (5–10) 5 (4–6) 0.003

Metoclopramide (days) 6 (3–10) 0 (0–3) \0.001

Erythromycin (days) 0 (0–6) 0 \0.001

Hospital LOS (days) 17.5 (13.5–25.0) 12 (10.5–15.5) 0.003

Postoperative complications (no.)

Any 12 19 \0.001

Bile leak 0 4 0.321

Pancreatic fistula 3 6 0.227

Intraabdominal abscess 2 2 0.085

Infection 2 6 0.108

Bleeding 1 2 0.090

Ileus 1 1 0.100

Other 1 1 0.465

Fig. 2 Plasma levels of paracetamol (a), glucagon-like peptide-1 (b),

and peptide YY (3–36) (c) according to the reconstruction group (top

panels) and to patients with or without delayed gastric emptying after

pylorus-preserving pancreaticoduodenectomy (bottom panels). Val-

ues are the means. Error bars represent the standard error of the mean

World J Surg (2014) 38:465–475 471

123

countries, let alone the effect of the learning curve, the case

load (100 PPPDs in 6 vs. 2 years), the difference in med-

ical complications (17 vs. 8 %, p = 0.054), and so on. The

rest of the relevant literature comprises retrospective

studies with sometimes exceedingly high rates of DGE (up

to 70 and 80 %) [7–9].

Table 4 Plasma levels of paracetamol, GLP-1, and PYY after the test meal on POD 10

Parameter Reconstruction method p* Delayed gastric emptying p*

Antecolic (n = 22) Retrocolic (n = 18) Yes (n = 8) No (n = 32)

Paracetamol

Level (lmol/l)

Baseline 0.14 (0–0.60) 0.20 (0–0.88) 0.267 0.13 (0–0.38) 0.14 (0–0.61) 0.325

15 min 2.33 (0.69–5.72) 1.98 (1.37–2.80) 0.757 1.10 (0.53–2.28) 2.33 (1.37–6.50) 0.002

30 min 2.20 (1.73–6.55) 3.85 (2.19–5.30) 0.779 2.38 (0.51–3.00) 4.26 (1.75–6.60) 0.006

60 min 3.20 (2.56–6.00) 3.92 (2.30–5.40) 0.943 2.83 (1.30–3.75) 3.77 (2.70–6.53) 0.025

90 min 3.57 (2.25–6.72) 4.15 (3.00–5.70) 0.508 3.08 (1.94–4.39) 4.02 (2.94–6.80) 0.043

AUCa 13.24 (7.01–27.78) 16.13 (11.28–23.0) 0.827 10.53 (6.50–16.13) 18.0 (10.04–29.37) 0.004

Cmax

At 15 min 3/22 5/18 0/8 8/32

At 30 min 4/22 3/18 1/8 6/32

At or after 60 min 15/22 10/18 7/8 18/32

GLP-1

Level (pmol/l)

Baseline 16 (11–25) 11.5 (7.5–27.5) 0.795 11 (5–25) 15 (10–26) 0.352

15 min 36 (18–91) 31.5 (13–61.5) 0.275 21 (13–42) 44 (17–91) 0.008

30 min 62 (20–187) 49.5 (22–82) 0.296 21 (11–55) 60 (24–187) 0.001

60 min 37 (22–68) 30 (17.5–59.0) 0.401 30 (10–49) 35.5 (24–68) 0.034

90 min 30 (22–54) 42 (19.5–70) 0.955 27 (13–54) 37.5 (23–70) 0.131

AUCb 5.69 (1.29–21.12) 3.62 (1.34–11.41) 0.240 1.26 (0.68–9.07) 6.62 (2.54–23.97) 0.006

Cmax

At 15 min 9/20 5/16 1/6 13/30

At 30 min 7/20 6/16 1/6 12/30

At or after 60 min 4/20 5/16 4/6 5/30

PYY

Level (pg/ml)

Baseline 40 (30–64) 30 (30–43) 0.288 30 (30–43) 36 (30–61) 0.155

15 min 57 (30–82) 37 (30–52) 0.209 45 (30–71) 50 (30–75) 0.333

30 min 56 (37–112) 49 (35–67) 0.268 49 (38–58) 52 (35–110) 0.014

60 min 66 (48–124) 51 (40–85) 0.109 63 (48–64) 65 (40–118) 0.073

90 min 72 (36.0–119.5) 51.5 (43–84) 0.216 53.5 (45–57) 64 (34–111) 0.063

AUCb 15.58 (6.65–39.08) 6.96 (3.60–14.147) 0.226 6.54 (6.52–10.29) 14.17 (3.63–7.74) 0.060

Cmax

At or before 30 min 5/19 8/15 3/7 10/27

At 60 min 9/19 5/15 3/7 11/27

At 90 min 5/19 2/15 1/7 6/27

The results are expressed as the median and the interquartile range (IQR) except for Cmas, which are absolute numbers

GLP-1 glucagon-like peptide-1, PYY peptide YY (3–36), AUC area under the curve, Cmax maximum serum concentration reached at the

indicated time pointa Arbitrary unitsb Arbitrary units 910-3

* The p values are derived from Student’s t-test for comparison of continuous variables

472 World J Surg (2014) 38:465–475

123

During the final analyses for our study, a group from

India published the results of an RCT with similar design

[15]. It did not show any influence of the gastroenteric

anastomosis with regard to DGE. Their trial, however, also

included patients who underwent a classic Whipple pro-

cedure, so the impact on patients with an intact pylorus

after PD was still unclear.

The current study was conducted at a single center and

87 % of the procedures were performed or assisted by a

single surgeon (P.G.). Although the randomization method

was simple and did not balance for any potential differ-

ences, the two treatment groups were homogeneously dis-

tributed and did not show any evident differences in

preoperative or intraoperative properties. The difference in

the final number of patients in the two arms is explained by

patients dropping out of the trial after being screened for

eligibility. One way to overcome this issue would have

been to blind patients to the intervention, which we did not

plan to do when we assembled this study.

The addition of the functional test of paracetamol

absorption and measurement of intestinal peptide plasma

levels further strengthened our observations, although they

were not standardized and ubiquitously available tests [16,

19]. We decided not to perform an invasive test such as

scintigraphy [13] because we thought that it was not feasible

in our patients. We anticipated that such a test could decrease

compliance and that we would not be able to accrue enough

patients for this protocol. In fact, even with this noninvasive

intervention we used for this study, eight and six patients,

respectively, withdrew consent for the test meal around POD

10. The plasma paracetamol concentration after the test meal

closely mirrored our clinical impression of DGE in this patient

population. However, we found that the interpretation of the

paracetamol uptake test did not have a better predictive value

in the individual patient. Prolonged LOS was better estimated

with clinical criteria for DGE (data not shown).

The biology of intestinal peptides such as GLP-1 and

PYY (3-36) is complex, and many physiologic effects and

interactions have been described [22, 27]. Therefore, we

decided not to interpret any of our baseline findings based

on their plasma levels because there is little evidence about

their meaning after PPPD. We chose to use only the time

course of GLP-1 and PYY concentration as a surrogate for

passage of nutrients to the proximal and distal intestine,

respectively [19]. A sampling period longer than 90 min

would probably have been more appropriate for better

understanding the PYY levels. Similar to other studies [20,

21, 28], the early GLP-1 peak correlated well with the

clinical impression of gastric emptying. The authors are

aware, however, that this is just an indirect indicator and

cannot replace an imaging study.

A limitation of this trial could be that it was finally

underpowered to assess the observed difference of only

5 % in DGE between groups. Although we chose the

numbers for our sample size calculation based on the

available literature at the time, the estimated improvement

in DGE by the operative intervention was too high. The

incidence of DGE was six patients in each group, or 17.6

and 23.1 %, respectively. For this absolute difference of

5.5 % we would now calculate a sample size of 840

patients for each arm, a number that would not enable us to

perform this trial reasonably. There is a risk that the results

presented here are related to a type II error as the number

of patients who were eventually evaluable for the primary

endpoint was too small to draw definitive conclusions.

Even on an intention-to-treat basis analysis (38 and 33

patients, respectively, in the two groups), there would have

to be at least 12 events in the retrocolic and 6 events in the

antecolic group in order to show a statistically significant

difference.

Another drawback of this study limiting future inter-

pretation is that the definition of DGE has since become

outdated because of the consensus definition proposed by

the International Study Group of Pancreatic Surgery

(ISGPS) published in November of 2007 [11]. We decided

in 2006 to use the definition of the Johns Hopkins group

Table 5 Series dealing with reconstruction of the duodenojejunostomy after PPPD

Study Region Study type Total patients

(antecolic/retrocolic)

DGE antecolic

(%)

DGE retrocolic

(%)

Tani [6] Japan RCT 20/20 5 50

Gangavatiker [15]a India RCT 32/36 34 28

Horstmann [9] Germany Retrospective 51/0 12 NA

Sugiyama [7] Japan Retrospective 12/18 8 72

Hartel [5] Germany Retrospective 100/100 5 24

Murakami [8] Japan Retrospective 78/54 10 81

Current series Austria Prospective 36/28 17.6 23.1

PPPD pylorus-preserving pancreaticoduodenectomy; RCT randomized controlled trial; NA not applicablea Includes patients with PPPD and a Whipple operation

World J Surg (2014) 38:465–475 473

123

because we found it to be a well-published and very useful

way of evaluating DGE [10]. We identified 13 individuals

who qualified for the diagnosis of DGE using the ISGPS

definition, 13 patients with the NGT in place longer than

3 days, and 11 with failure to eat a regular diet after POD

7. In summary, we had 11 patients with grade A DGE, 1

patient with grade B DGE, and 1 with grade C DGE.

Because this definition of DGE was not designated in the

methods that were prospectively established, we did not

report these findings in the ‘‘Results’’ section. However,

this new definition seems to be applicable and is used

internationally [29, 30], so we wanted to report these data

at this point.

Conclusions

We have performed the first prospective trial in a Western

center regarding the effect of the duodenojejunal recon-

struction method on the incidence of postoperative DGE

after PPPD. We found no differences in the results regardless

of whether the duodenojejunostomy was performed via the

antecolic route or the retrocolic route. We confirmed the

clinical findings with a test meal performed on POD 10.

However, to truly and finally answer this question, a trial

with a noninferiority design is probably necessary, which is

of questionable feasibility in this setting.

Conflict of interest The authors do not disclose any potential

conflict of interest.

References

1. Whipple AO, Parsons WB, Mullins CR (1935) Treatment of

carcinoma of the ampulla of Vater. Ann Surg 102:763–779

2. Traverso LW, Longmire WP Jr (1978) Preservation of the

pylorus in pancreaticoduodenectomy. Surg Gynecol Obstet

146:959–962

3. Tran KT, Smeenk HG, van Eijck CH et al (2004) Pylorus pre-

serving pancreaticoduodenectomy versus standard Whipple pro-

cedure: a prospective, randomized, multicenter analysis of 170

patients with pancreatic and periampullary tumors. Ann Surg

240:738–745

4. Diener MK, Knaebel HP, Heukaufer C et al (2007) A systematic

review and meta-analysis of pylorus-preserving versus classical

pancreaticoduodenectomy for surgical treatment of periampullary

and pancreatic carcinoma. Ann Surg 245:187–200

5. Hartel M, Wente MN, Hinz U et al (2005) Effect of antecolic

reconstruction on delayed gastric emptying after the pylorus-

preserving Whipple procedure. Arch Surg 140:1094–1099

6. Tani M, Terasawa H, Kawai M et al (2006) Improvement of

delayed gastric emptying in pylorus-preserving pancreaticoduo-

denectomy: results of a prospective, randomized, controlled trial.

Ann Surg 243:316–320

7. Sugiyama M, Abe N, Ueki H et al (2004) A new reconstruction

method for preventing delayed gastric emptying after pylorus-

preserving pancreatoduodenectomy. Am J Surg 187:743–746

8. Murakami Y, Uemura K, Sudo T et al (2008) An antecolic Roux-

en-Y type reconstruction decreased delayed gastric emptying

after pylorus-preserving pancreatoduodenectomy. J Gastrointest

Surg 12:1081–1086

9. Horstmann O, Becker H, Post S et al (1999) Is delayed gastric

emptying following pancreaticoduodenectomy related to pylorus

preservation? Langenbecks Arch Surg 384:354–359

10. DeOliveira ML, Winter JM, Schafer M et al (2006) Assessment

of complications after pancreatic surgery: a novel grading system

applied to 633 patients undergoing pancreaticoduodenectomy.

Ann Surg 244:931–937 discussion 937–939

11. Wente MN, Bassi C, Dervenis C et al (2007) Delayed gastric

emptying (DGE) after pancreatic surgery: a suggested definition

by the International Study Group of Pancreatic Surgery (ISGPS).

Surgery 142:761–768

12. Nguyen TC, Sohn TA, Cameron JL et al (2003) Standard vs.

radical pancreaticoduodenectomy for periampullary adenocarci-

noma: a prospective, randomized trial evaluating quality of life in

pancreaticoduodenectomy survivors. J Gastrointest Surg 7:1–9

discussion 9–11

13. Yeo CJ, Barry MK, Sauter PK et al (1993) Erythromycin accel-

erates gastric emptying after pancreaticoduodenectomy: a pro-

spective, randomized, placebo-controlled trial. Ann Surg 218:

229–237 discussion 237–228

14. Bassi C, Falconi M, Molinari E et al (2005) Reconstruction by

pancreaticojejunostomy versus pancreaticogastrostomy following

pancreatectomy: results of a comparative study. Ann Surg

242:767–771 discussion 771–763

15. Gangavatiker R, Pal S, Javed A et al (2011) Effect of antecolic or

retrocolic reconstruction of the gastro/duodenojejunostomy on

delayed gastric emptying after pancreaticoduodenectomy: a ran-

domized controlled trial. J Gastrointest Surg 15:843–852

16. Heading RC, Nimmo J, Prescott LF et al (1973) The dependence

of paracetamol absorption on the rate of gastric emptying. Br J

Pharmacol 47:415–421

17. Medhus AW, Sandstad O, Bredesen J et al (1999) Delay of

gastric emptying by duodenal intubation: sensitive measurement

of gastric emptying by the paracetamol absorption test. Aliment

Pharmacol Ther 13:609–620

18. Morioka J, Miyachi M, Niwa M et al (2008) Gastric emptying for

liquids and solids after distal gastrectomy with Billroth-I recon-

struction. Hepatogastroenterology 55:1136–1139

19. Strommer L, Raty S, Hennig R et al (2005) Delayed gastric

emptying and intestinal hormones following pancreatoduode-

nectomy. Pancreatology 5:537–544

20. Schirra J, Katschinski M, Weidmann C et al (1996) Gastric

emptying and release of incretin hormones after glucose ingestion

in humans. J Clin Invest 97:92–103

21. Yamamoto H, Mori T, Tsuchihashi H et al (2005) A possible role

of GLP-1 in the pathophysiology of early dumping syndrome.

Dig Dis Sci 50:2263–2267

22. Ballantyne GH (2006) Peptide YY(1–36) and peptide YY(3–36).

Part I. Distribution, release and actions. Obes Surg 16:651–658

23. Dindo D, Demartines N, Clavien PA (2004) Classification of

surgical complications: a new proposal with evaluation in a

cohort of 6,336 patients and results of a survey. Ann Surg

240:205–213

24. Rumack BH, Matthew H (1975) Acetaminophen poisoning and

toxicity. Pediatrics 55:871–876

25. Miholic J, Hoffmann M, Holst JJ et al (2007) Gastric emptying of

glucose solution and associated plasma concentrations of GLP-1,

GIP, and PYY before and after fundoplication. Surg Endosc

21:309–314

26. Lavu H, Kennedy EP, Mazo R et al (2010) Preoperative

mechanical bowel preparation does not offer a benefit for patients

who undergo pancreaticoduodenectomy. Surgery 148:278–284

474 World J Surg (2014) 38:465–475

123

27. Baggio LL, Drucker DJ (2007) Biology of incretins: GLP-1 and

GIP. Gastroenterology 132:2131–2157

28. Blom WA, Lluch A, Vinoy S et al (2006) Effects of gastric

emptying on the postprandial ghrelin response. Am J Physiol

Endocrinol Metab 290:E389–E395

29. Malleo G, Crippa S, Butturini G et al (2010) Delayed gastric emp-

tying after pylorus-preserving pancreaticoduodenectomy: validation

of International Study Group of Pancreatic Surgery classification and

analysis of risk factors. HPB (Oxford) 12:610–618

30. Welsch T, Borm M, Degrate L et al (2010) Evaluation of the

International Study Group of Pancreatic Surgery definition of

delayed gastric emptying after pancreatoduodenectomy in a high-

volume centre. Br J Surg 97:1043–1050

World J Surg (2014) 38:465–475 475

123