PAPER

Critical Analysis
of a Large Series of Pancreaticogastrostomy
After Pancreaticoduodenectomy
Gerard V. Aranha, MD, FRCSC; Joshua M. Aaron, BS; Margo Shoup, MD

Hypothesis: Pancreaticogastrostomy is a safe opera-
tion for a variety of periampullary conditions.
Design: Retrospective review of a prospectively col-
lected database.
Setting: An academic tertiary care university hospital
and a Veterans Affairs hospital.
Patients: A total of 235 consecutive patients who un-
derwent pancreaticogastrostomy.
Main Outcome Measures: Indications for surgery,
preoperative risk factors, intraoperative and postopera-
tive variables, and factors that affect postoperative
complications.
Results: The most common initial symptoms were jaun-
dice (73.2%), weight loss (23.8%), and abdominal pain
(17.0%). The 4 most common indications for surgery were
pancreatic adenocarcinoma (41.3%), ampullary carci-
noma (17.0%), duodenal carcinoma (7.2%), and chronic
pancreatitis (7.2%). The median operating time was 6.5
hours. Median blood loss was 900 mL. The median in-
traoperative blood transfusion was 0 U. The median post-
operative length of stay was 9 days. Postoperative mor-
tality was 0.9%. The most common complications were
pancreatic fistulae (13.6%), 1 of which was thought to
cause 1 of 2 mortalities in this series. Pancreatic fistulae
developing after pancreaticogastrostomy were signifi-
cantly related to a low preoperative alkaline phospha-
tase level and surgery for nonpancreatic pathologic find-
ings. The presence of a fistula significantly increased the
postoperative length of hospital stay.
Conclusions: Pancreaticogastrostomy is a safe opera-
tion associated with low mortality and morbidity rates
and a pancreatic fistula rate of 13.6%. It should be con-
sidered as a suitable alternative for management of the
pancreatic remnant after pancreaticoduodenectomy.
Arch Surg. 2006;141:574-580

Author Affiliations: Division of
Surgical Oncology, Department
of Surgery, Loyola University,
Stritch School of Medicine,
Maywood, Ill (Drs Aranha and
Shoup and Mr Aaron), and
Surgical Service, Hines VA
Hospital, Hines, Ill
(Dr Aranha).
M
ANAGEMENT OF THE
pancreatic remnant af-
ter pancreaticoduode-
nectomy continues to
be a source of contro-
versy. This controversy is fueled by the fact
that leakage from the pancreaticoenteric
anastomosis is responsible for a large per-
centage of the morbidity and mortality that
follows pancreaticoduodenectomy. Trede
and Schwall1 from the Mannheim Clinic in
Mannheim, Germany, published data on
233 patients who had pancreaticojejunal
anastomosis and reported 25 pancreatic
leaks, for an incidence of 11%, and 20% of
the leaks led directly to postoperative deaths.
The Lahey Clinic2 and the Mayo Clinic3 re-
ported similar findings, with 34 pancreatic
leaks in 403 pancreaticoduodenectomies, for
an incidence of 8%, of which 26% was re-
lated directly to postoperative deaths.
Because leakage from the pancreatico-
enteric anastomosis has been the leading
cause of morbidity and mortality after pan-
creaticoduodenectomy, various tech-
niques for managing the pancreatic rem-
nant have been studied, including simple
ligation of the pancreatic duct,4,5 occlu-
sion of the pancreatic duct using a syn-
thetic rubber injection or fibrin glue,6,7 op-
timization of the blood supply of the edge
of the pancreatic remnant and meticu-
lous placement of sutures using magnifi-
cation, 8 the application of fibrin glue
sealant around the pancreaticojejunal anas-
tomosis,9 various modifications of pan-
creaticojejunostomy (either end-to-end or
end-to-side anastomosis),10-15 isolated
Roux-en-Y pancreaticojejunostomy,16,17
and pancreaticogastrostomy.18-32
The purpose of this study is to analyze
our experience with 235 pancreaticogas-
trostomies after pancreaticoduodenec-
tomy studied in a retrospective manner
from a prospectively collected database to
determine whether pancreaticogastros-
tomy is a safe and effective method for
managing the pancreatic remnant.

(REPRINTED) ARCH SURG/ VOL 141, JUNE 2006 WWW.ARCHSURG.COM

574
Downloaded from www.archsurg.com on September 9, 2008
©2006 American Medical Association. All rights reserved.
 A B C

1 cm

Figure. Construction of the pancreaticogastrostomy. A, Sutures are placed from the posterosuperior gastric wall to the anterior pancreas body. B, A gastrotomy is
made, and sutures are placed from the posteroinferior gastric wall to the posterior pancreas body. C, Completed hepaticojejunostomy, gastrojejunostomy, and
pancreaticogastrostomy. Reprinted with permission from Excerpta Medica, Inc, from Aranha GV. “A Technique for Pancreaticogastrostomy.” Am J Surg.
1998;175:328-329. Copyright 1998.

the pancreas. Sutures entered the pancreas at least 2 cm from

METHODS
PATIENT DATA
We conducted a retrospective review of a prospective data-
base of 235 patients who underwent pancreaticogastrostomy
after pancreaticoduodenectomy at Loyola University Medical
Center and Hines VA Hospital between June 1, 1990, and July
31, 2005. All the patients who underwent pancreaticogastros-
tomy after pancreaticoduodenectomy were operated on by the
same surgeon (G.V.A.), with no patients having pancreatico-
duodenectomy reconstructed with pancreaticojejunostomy dur-
ing this period.
Clinical and pathologic data were obtained from the sur-
gery database, medical records, and interviews. Data obtained
for each patient included demographics; preoperative labora-
tory values; clinical symptoms and signs; procedures, includ-
ing computed tomography (CT), endoscopy, and the place-
ment of stents; and intraoperative data, such as the use of
prophylactic antibiotic agents, blood loss, operating time, and
units of blood transfused. Postoperative data, including patho-
logic findings, mortality, morbidity, and the use of octreotide,
were also analyzed. Analyses were performed using statistical
software (SPSS for Windows; SPSS Inc, Chicago, Ill). The ␹2
or Fisher exact test was used when appropriate. Statistical sig-
nificance was set at P=.05.
SURGICAL TECHNIQUE
All the patients underwent classic pancreaticoduodenectomy
with distal gastrectomy. The pancreatic remnant was mobi-
lized for 4 cm. Single layers of sutures of 3-0 silk were taken
from the posterosuperior gastric wall, at least 5 cm from the
cut edge of the stomach, to the anterior wall of the body of the
pancreas. A gastrotomy was made, and then sutures were placed
from the posteroinferior gastric wall to the posterior body of
the cut edge and exited 1 cm from the cut edge, and when the
sutures were tied, at least 1 to 2 cm of the pancreas was invagi-
nated into the stomach without a stent. The Figure demon-
strates the completed hepaticojejunostomy, gastrojejunos-
tomy, and pancreaticogastrostomy.33 Two drains were placed:
1 to drain the hepaticojejunostomy on the right side and 1 to
drain the pancreaticogastrostomy on the left side. All the pa-
tients received prophylactic antibiotics before surgery. Early in
the series, most patients received octreotide, but in the last 123
patients, octreotide was not used. Patients began receiving eryth-
romycin lactobionate, 250 mg intravenously every 6 hours, on
day 4. The nasogastric tube was removed on day 5, and meto-
clopramide hydrochloride was added intravenously every 8
hours. Proton pump inhibitors were used postoperatively to
prevent stress and marginal ulceration.
The patients are given a liquid diet on day 6 and, if tolerated,
advanced to a postgastrectomy diet on day 7. After 2 solid meals,
drainage from the left drain was measured for amylase content. If
the amylase level was less than 125 U/L, the drain was removed if
the volume was 200 mL/d or less. If the volume was greater than
200 mL/d, and the amylase level was normal, the patient was sent
home with the drain until the drainage decreased to less than 200
mL/d, and then the drain was removed. If the fluid was amylase
rich, the patient was fed, and if the volume did not change, the pa-
tient was sent home on an oral diet. If the fluid was amylase rich
and the volume increased after an oral diet, then a drain study was
performed by injecting dye into the drain. In 6 (19%) of 32 patients
with a pancreatic leak, the drain had eroded into the pancreatico-
gastrostomyanastomosisandhadtobepulledback.Thedrainplaced
along the biliary anastomosis was removed on day 5 if there was
no bile in the drain. This was done irrespective of the volume of
drainage.Apancreaticfistulawasdefinedasamylase-richfluidfrom
the pancreatic drain with 3 times the serum amylase level on the
first day after the patient eats a solid diet. Delayed gastric empty-
ing was defined as a need for nasogastric suction for more than 10
days after pancreaticoduodenectomy.

(REPRINTED) ARCH SURG/ VOL 141, JUNE 2006 WWW.ARCHSURG.COM

575
Downloaded from www.archsurg.com on September 9, 2008
©2006 American Medical Association. All rights reserved.
 Table 1. Characteristics and Preoperative Risk Factors Table 2. Indications for Pancreaticoduodenectomy
of 235 Patients*
Pathologic Finding Patients, No. (%)
Characteristic No.(%)
Adenocarcinoma 97 (41.3)
Age, median (range), y 68 (21-90) Ampullary carcinoma 40 (17.0)
Sex Duodenal carcinoma 17 (7.2)
Male 145 (61.7) Chronic pancreatitis 17 (7.2)
Female 90 (38.3) Common bile duct cancer 14 (6.0)
Jaundice 172 (73.2) Mucinous cystadenoma 10 (4.3)
Weight loss 56 (23.8) Neuroendocrine 9 (3.8)
Epigastric or back pain 40 (17.0) Intraductal papillary mucinous neoplasm 7 (3.0)
Itching 15 (6.4) Serous cystadenoma 6 (2.6)
Hypertension 98 (41.7) Cystic adenocarcinoma 2 (0.9)
Coronary artery disease 40 (17.0) Other 16 (6.8)
Diabetes mellitus 38 (16.2)
Previous cancer 28 (11.9)
COPD or asthma 21 (8.9)
Atrial fibrillation 19 (8.1) Table 3. Postoperative Complications
Hypothyroid 17 (7.2)
GERD 15 (6.4) Complication Patients, No. (%)
Peripheral vascular disease 12 (5.1)
Peptic ulcer disease 10 (4.3) Mortality 2 (0.9)
Pancreatic leak 32 (13.6)
Intra-abdominal abscess 14 (6)
Abbreviations: COPD, chronic obstructive pulmonary disease;
GERD, gastroesophageal reflux disease. Delayed gastric emptying 14 (6)
*Data are given as number (percentage) of patients except where Wound infection 13 (5.5)
indicated otherwise. Cardiac complications 8 (3.4)
Small intestinal obstruction 5 (2.1)
Pulmonary complications 5 (2.1)
Hemorrhage 4 (1.7)
RESULTS Gastric leak 3 (1.3)
Bile leak 2 (0.9)
Patient characteristics and preoperative risk factors are Reexploration 3 (1.3)
Miscellaneous 18 (7.7)
listed in Table 1. There were 145 men and 90 women
(median age, 68 years; range, 21-90 years. The most com-
mon initial symptoms included jaundice (73.2%), weight
loss (23.8%), and abdominal pain (17.0%). Median weight a massive upper gastrointestinal hemorrhage. Aside from
loss was 6.75 kg (range, 1.80-24.75 kg). Patients who were the 2 deaths, 191 patients underwent pancreaticoduo-
considered malnourished (ie, albumin level ⬍2.5 g/dL) denectomy without mortality.
were given parenteral or enteral nutrition for 10 to 14 A pancreatic fistula occurred in 32 patients (13.6%)
days before surgery. This occurred in fewer than 10 pa- and was the most common morbidity. Most of the 32 fis-
tients. Preoperative risk factors included hypertension, tulae closed with maintenance of drains and continued
diabetes mellitus, previous cancer, chronic obstructive oral intake. In 11 (34.4%) of these 32 patients, paren-
pulmonary disease or asthma, atrial fibrillation, coro- teral nutrition was necessary for fistula closure. Only 1
nary artery disease, peripheral vascular disease, peptic ul- patient had to return to surgery for closure of the fis-
cer, hypothyroidism, and pancreatitis. Indications for pan- tula. An intra-abdominal abscess occurred in 14 pa-
creaticoduodenectomy based on pathologic findings are tients (6.0%) and wound infections in 13 (5.5%). All the
given in Table 2. The 4 most common indications for patients with intra-abdominal abscess were treated with
pancreaticoduodenectomy were pancreatic adenocarci- either intraoperatively placed drains or new percutane-
noma (41.3%), ampullary carcinoma (17.0%), duode- ous drains by means of interventional radiology.
nal carcinoma (7.2%), and chronic pancreatitis (7.2%). Hemorrhage occurred in 4 patients (1.7%). Two of the
All the patients underwent preoperative CT, and 142 patients had hemorrhage from the gastrojejunal anasto-
(60.4%) had biliary stents placed via endoscopic retro- mosis that was controlled with endoscopic means, and the
grade cholangiopancreatography or percutaneous trans- other 2 had pseudoaneurysms. One pseudoaneurysm was
hepatic cholangiography. controlled at our institution (Loyola University Medical
The median operating time was 6.5 hours (range, 3.4- Center) with interventional radiology using coil emboli-
13.0 hours). Median blood loss was 900 mL (range, 200- zation. The other patient died at an outside institution of
7500 mL), and the median intraoperative transfusion was massive upper gastrointestinal bleeding and is presumed
0 U (range, 0-7 U). Octreotide was used after surgery in to have had a pseudoaneurysm. Bile leaks occurred in 2
112 patients and was not used in 123 patients. The me- patients (0.9%) and were treated with percutaneous trans-
dian postoperative hospital stay was 9 days (range, 5-83 hepatic stenting of the anastomosis, with resolution of the
days). Thirty-day and in-hospital mortality occurred in problem. Contained leaks from the gastrojejunostomy oc-
2 patients (0.9%) (Table 3). One patient died of acute curred in 3 patients (1.3%), and none required surgery. De-
respiratory distress syndrome on the 45th postoperative layed gastric emptying occurred in 14 patients (6.0%), and
day. The second patient died at an outside institution of all resolved with conservative measures.

(REPRINTED) ARCH SURG/ VOL 141, JUNE 2006 WWW.ARCHSURG.COM

576
Downloaded from www.archsurg.com on September 9, 2008
©2006 American Medical Association. All rights reserved.
 A second operation was required in 3 patients (1.3%).
In the first patient, the surgery was for peritonitis. At ex- Table 4. Preoperative, Intraoperative, and Postoperative
ploration, it was found that the pancreatic remnant had Variables in Patients With vs Without a Pancreatic Fistula*
separated completely from the stomach. In this case, the
No Fistula Fistula P
pancreatic duct was simply oversewn and the gas- Characteristic (n = 203) (n = 32) Value
trotomy closed. This patient survived. The second pa-
Age, y
tient had a drain included in the closure of the fascia, and
⬍70 113 (84) 21 (16)
when the pancreatic drain was removed, it was severed. ⱖ70 90 (89) 11 (11)
.29
This patient was taken back to the operating room, and Duration of surgery, h
the drain remnant was found in the wound just after the ⱕ6.5 110 (89) 13 (11)
.15
anterior fascia was opened; therefore, reentry into the ab- ⬎6.5 93 (83) 19 (17)
domen was not necessary. The third patient who had a Octreotide therapy
second surgery was the one who had surgical closure of Yes 94 (84) 18 (16)
.30
No 109 (89) 14 (11)
a persistent pancreatic fistula 3 months after surgery. Of Intraoperative blood
the 112 patients receiving octreotide, 12 (10.7%) devel- transfusion
oped a fistula, and this was not significant compared with Yes 65 (81) 15 (19)
.10
the 11 (8.9%) of 123 patients who developed a fistula with- No 138 (89) 17 (11)
out octreotide administration. Estimated blood loss, mL
Table 4 lists certain preoperative, intraoperative, and ⱕ1000 128 (90) 15 (10)
.08
⬎1000 75 (82) 17 (18)
postoperative variables in relation to fistula formation. Age, Preoperative stent
duration of surgery, use of octreotide, intraoperative blood Yes 127 (89) 15 (11)
transfusion, intraoperative blood loss, preoperative bili- .09
No 76 (82) 17 (18)
ary stenting, preoperative endoscopic retrograde cholan- Preoperative ERCP or PTC
giopancreatography or percutaneous transhepatic cholan- Yes 49 (86) 8 (14)
.92
giography, and a preoperative albumin level of 3 g/dL had No 154 (87) 24 (13)
Alkaline phosphatase, U/L
no effect on fistula formation. However, patients with el-
ⱕ110 46 (74) 16 (26)
evated alkaline phosphatase levels had a significantly lower ⬎110 157 (91) 16 (9)
.001
incidence of postoperative pancreatic fistula than those Albumin, g/dL
whose alkaline phosphatase level was not elevated. Also, ⱕ3 44 (90) 5 (10)
.43
patients who had pancreatic adenocarcinoma and pancre- ⬎3 159 (85) 27 (15)
atic pathologic abnormalities had a much lower incidence Pancreatic pathologic findings
Yes 135 (91) 13 (9)
of pancreatic fistula than those who had other types of .005
No 68 (78) 19 (22)
pathologic abnormality. Finally, a pancreatic fistula was as- Hospital stay, 9 (5-34) 18 (7-83) ⬍.001
sociated with a doubling in the median length of hospital median (range), d
stay.
Abbreviations: ERCP, endoscopic retrograde cholangiopancreatography;
PTC, percutaneous transhepatic cholangiography.
COMMENT *Data are given as number (percentage) of patients except where
otherwise indicated.

Pancreaticogastrostomy became a reality when Tripodi

and Sherwin34 first reported successful transplantation
of the pancreas into the stomach in 1934. This finding
was confirmed by Person and Glenn.35 Waugh and
Clagett36 at the Mayo Clinic were the first to use pancre-
aticogastrostomy in the clinical setting in 1946. Further
impetus to the success of the operation was provided by
Mackie et al37 in 1975. Used infrequently initially, pan-
creaticogastrostomy has been used much more fre-
quently in the past 20 years.24-32 It is well-known that leak-
age from the pancreaticojejunal anastomosis and its
consequences are the leading causes of mortality after pan-
creaticoduodenectomy. Data from the late 1970s and early
1980s confirm this. More recently, Yeo et al38 reported
on 650 consecutive pancreaticoduodenectomies per-
formed at The Johns Hopkins Hospital. They noted that
26 patients (4.0%) required repeated surgery. Of the 26
patients, 9 (34.6%) died after the second operation. In
all of these patients, death could be directly related to
leakage from the pancreaticojejunal anastomosis. Sev-
eral theoretical physiologic and technical advantages to
performing pancreaticogastrostomy have been de-
scribed.28 A physiologic advantage is believed to be that
pancreatic enzymes are inactivated by the acidic gastric
fluid. In addition, the stomach does not contain entero-
kinase, which is required for the conversion of trypsino-
gen to trypsin and the subsequent activation of other pro-
teolytic enzymes. A lack of enzyme activation may prevent
autodigestion of the anastomosis. In addition, the alka-
line and pancreatic secretions may aid in preventing mar-
ginal ulceration. The proximity of the pancreas to the pos-
terior wall of the stomach allows for potentially less
tension on the anastomosis. The excellent blood supply
to the stomach wall is favorable to anastomotic healing,
and the thickness of the stomach wall holds sutures well.
Nasogastric decompression provides for continuous emp-
tying of the stomach and, therefore, less tension on the
pancreaticogastrostomy anastomosis, a benefit not pos-
sible with a pancreaticojejunal anastomosis. A review of
the world literature on pancreaticogastrostomy from 1946
to 1997 by Mason39 seems to confirm the safety of pan-
creaticogastrostomy. Of 813 patients who underwent pan-
creaticogastrostomy, 32 (4.0%) developed pancreatico-
cutaneous fistulae. Of these fistulae, 3 (0.4%) were thought
to be the primary cause of death. In a meta-analysis of

(REPRINTED) ARCH SURG/ VOL 141, JUNE 2006 WWW.ARCHSURG.COM

577
Downloaded from www.archsurg.com on September 9, 2008
©2006 American Medical Association. All rights reserved.
pancreatic fistula after pancreaticoduodenectomy, Bar-
toli et al40 also suggested that pancreaticogastrostomy was
safer than pancreaticojejunostomy. Yeo et al41 pub-
lished the only prospective randomized trial of pancre-
aticogastrostomy vs pancreaticojejunostomy after pan-
creaticoduodenectomy. In this study, there was no
difference in the mortality and morbidity whether the
stomach or jejunum was anastomosed to the pancreatic
remnant.
In the present study, only 2 deaths (0.9%) occurred.
One death was due to acute respiratory distress syn-
drome and occurred 45 days after the operation. The
second death can be related to leakage from the pancre-
aticogastrostomy because this patient, who had a pancre-
aticoduodenectomy for an ampullary carcinoma, was dis-
charged from the hospital with a drain on the ninth
postoperative day. Five days after discharge, he was found
unconscious by a family member and was taken by am-
bulance to another institution. Here, after undergoing CT,
he returned to the emergency department where, accord-
ing to his family members, blood rushed out of his mouth
and nose, resulting in immediate death. We can only con-
jecture that this death was due to a pseudoaneurysm as
a result of a leak from the pancreaticogastrostomy. No
autopsy was performed. Late bleeding in a patient who
has had pancreaticoduodenectomy, especially one who
has had a leak, is often due to pseudoaneurysms. These
patients should undergo immediate CT angiography, fol-
lowed by conventional angiography, with embolization
if a pseudoaneurysm is demonstrated.42
Recent published large series38,43 with pancreaticojeju-
nal anastomosis revealed a reexploration rate of 3% to 4%,
of which 34% of the mortality was related directly to leak-
age from the pancreaticojejunal anastomosis. In a previ-
ous study,44 5 (5%) of 97 patients undergoing pancreati-
cojejunostomy needed reexploration, and in 3 of these 5
patients, death was due to leakage at the pancreaticojeju-
nal anastomosis. In the same study,44 reexploration was not
required in 117 patients undergoing pancreaticogastros-
tomy. This has also been reported by other research-
ers.45,46 In the present series, reexploration was required in
3 patients (1.3%), and all 3 survived.
Previous studies have attempted to define the cause of
fistula formation after pancreaticoduodenectomy. In the ar-
ticle by Yeo et al,41 the surgeon’s experience and pancreati-
coduodenectomy performed for nonpancreatic pathologic
findings were strongly associated with pancreatic fistula for-
mation. The soft texture of the pancreas, as seen in nonpan-
creatic pathologic abnormalities, was found by Lin et al47 to
be significantly associated with pancreatic fistula. In the pre-
sent study, although we did not have sufficient information
on the texture of the pancreatic remnant, fistula formation
was significantly related to pancreaticoduodenectomy per-
formed for nonpancreatic pathologic findings. In addition,
patientswithelevatedalkalinephosphataselevelswerefound
to have a significantly lower incidence of fistula after pan-
creaticoduodenectomy in our series. In addition, there was
no correlation with fistula formation and patients having an
albumin level of 3 g/dL or less. However, only 49 patients
(20.9%) had albumin levels of 3 g/dL or less. Therefore, par-
enteral nutrition was needed only in patients who had
persistent high-volume fistulae. The development of a pan-
(REPRINTED) ARCH SURG/ VOL 141, JUNE 2006
578
Downloaded from www.archsurg.com on September 9, 2008
©2006 American Medical Association. All rights reserved.
creatic fistula increases the length of the postoperative
hospital stay, as seen in this study and others.48,49
There are several reasons why delayed gastric emp-
tying occurs in patients undergoing pancreaticoduode-
nectomy.49 Delayed gastric emptying occurred in only
6.0% of patients in this series. We believe that this is due
to the fact that we perform a classic pancreaticoduode-
nectomy that includes a distal gastrectomy. In addition,
the use of metoclopramide after surgery may have also
helped. Although metoclopramide increases the tone and
amplitude of the antral stomach, it has an effect on the
motility of the entire stomach in addition to increasing
the tone of the lower esophageal sphincter. Metoclopra-
mide is used to prevent impaired gastric motility after duo-
denectomy and reflux secondary to gastric stasis. Be-
cause partial pancreatectomy can have an ulcerogenic
effect,1 proton pump inhibitors are used after surgery.
Intra-abdominal abscess occurred in 14 patients (6.0%)
and wound infections occurred in 13 (5.5%). Intra-
abdominal abscesses are mainly due to leakage from the
pancreaticoenteric anastomosis.49 Leaks from the hepati-
coenteric and gastroenteric anastomosis cause fewer intra-
abdominal abscesses. Intra-abdominal abscesses are
thought to be associated with increased mortality.49 This
was not true in this series. All intra-abdominal ab-
scesses were managed successfully with maintenance of
intraoperatively placed drains or new percutaneous drains.
In conclusion, this study, which is the largest study
of pancreaticogastrostomy in the literature, shows the
safety of pancreaticogastrostomy. Mortality is low, re-
peated surgery is also low, and other complications of
pancreatic fistulae, delayed gastric emptying, intra-
abdominal abscesses, wound infections, hemorrhage, bile
leaks, repeated operations, and cardiac complications are
also low, further supporting the safety of pancreatico-
gastrostomy. Pancreaticogastrostomy remains a viable
option for management of the pancreatic remnant after
pancreaticoduodenectomy.
Accepted for Publication: January 14, 2006.
Correspondence: Gerard V. Aranha, MD, FRCSC, De-
partment of Surgery, Loyola University, Stritch School
of Medicine, 2160 S First Ave, EMS Bldg, Room 110-
3236, Maywood, IL 60153 (garanha@lumc.edu).
Previous Presentation: This paper was presented at the
113th Scientific Session of the Western Surgical Asso-
ciation; November 9, 2005; Rancho Mirage, Calif; and is
published after peer review and revision. The discus-
sions that follow this article are based on the originally
submitted manuscript and not the revised manuscript.
REFERENCES
1. Trede M, Schwall G. The complications of pancreatectomy. Ann Surg. 1988;207:
39-47.
2. Braasch JW, Gray BN. Considerations that lower pancreaticoduodenectomy
mortality. Am J Surg. 1977;133:480-484.
3. Edis AJ, Kiernan PD, Taylor WF. Attempted curative resection of ductal carci-
noma of the pancreas: review of Mayo Clinic experience, 1951-1975. Mayo Clin
Proc. 1980;55:531-536.
4. Goldsmith HS, Ghosh BC, Huvos AG. Ligation versus implantation of the pancreatic
duct after pancreaticoduodenectomy. Surg Gynecol Obstet. 1971;132:87-92.
5. Papachristou DN, Fortner JG. Pancreatic fistula complicating pancreatectomy for
malignant disease. Br J Surg. 1981;68:238-240.
WWW.ARCHSURG.COM
 6. Di Carlo V, Chiesa R, Pontiroli AE, et al. Pancreaticoduodenectomy with occlu-
sion of the residual stump with Neoprene injection. World J Surg. 1989;13:
105-111.
7. Suc B, Msika S, Fingerhut A, et al. Temporary fibrin glue occlusion of the main
pancreatic duct in the prevention of intra-abdominal complications after pancre-
atic resection: prospective randomized trial. Ann Surg. 2003;237:57-65.
8. Strasberg SM, Drebin JA, Mokadam NA, et al. Prospective trial of a blood supply–
based technique of pancreaticojejunostomy: effect on anastomotic failure in the
Whipple procedure. J Am Coll Surg. 2002;194:746-760.
9. Lillemoe KD, Cameron JL, Kim MP, et al. Does fibrin glue sealant decrease the
rate of pancreatic fistula after pancreaticoduodenectomy? results of a prospec-
tive randomized trial. J Gastrointest Surg. 2004;8:766-774.
10. Warshaw AL, Thayer SP. Pancreaticoduodenectomy. J Gastrointest Surg. 2004;
8:733-741.
11. Willams JG, Bramhall SR, Neoptolemos JP. Purse-string pancreatico-
jejunostomy following pancreatic resection. Dig Surg. 1997;14:183-186.
12. Murr MM, Nagorney DM. An end-to-end pancreaticojejunostomy using a me-
chanical purse-string device. Am J Surg. 1999;177:340-341.
13. Landen S. Consolidation of a friable pancreas for pancreaticojejunal anastomosis.
Dig Surg. 1998;15:297-298.
14. Roder JD, Stein HJ, Bottcher KA. Stented versus nonstented pancreaticojeju-
nostomy after pancreaticoduodenectomy: a prospective study. Ann Surg. 1999;
229:41-48.
15. Sakorafas GH, Freiss H, Balsiger BM, Buchler MW, Sarr MG. Problems of re-
construction during pancreatoduodenectomy. Dig Surg. 2001;18:363-369.
16. Sutton CD, Garcea G, White SA, et al. Isolated Roux-loop pancreaticojejunos-
tomy: a series of 61 patients with zero postoperative pancreaticoenteric leaks.
J Gastrointest Surg. 2004;8:701-705.
17. Papadimitriou JD, Fotopoulos AC, Smyrniotis B, et al. Subtotal pancreaticoduo-
denectomy: use of a defunctionalized loop for pancreatic stump drainage. Arch
Surg. 1999;134:135-139.
18. Ingebrigtsen R, Langfeldt E. Pancreaticogastrostomy. Lancet. 1952;2:270-271.
19. Dill-Russell AS. Pancreaticogastrostomy. Lancet. 1952;1:589-590.
20. Wells C, Shepherd JA, Gibbon N. Pancreaticogastrostomy. Lancet. 1952;1:
588-589.
21. Silverstone M. Pancreaticoduodenectomy and pancreaticogastrostomy: a five-
year survival with notes on the metabolism. Br J Surg. 1956;44:299-302.
22. Strauch GO. The use of pancreatogastrostomy after blunt traumatic pancreatic
transection: a complete and efficient operation. Ann Surg. 1972;176:16-18.
23. Millbourn E. Pancreatico-gastrostomy in pancreatico-duodenal resection for car-
cinoma of the head of the pancreas or the papilla of vater. Acta Chir Scand. 1958;
116:12-27.
24. Flautner L, Tihanya T, Szecseny A. Pancreaticogastrostomy: an ideal comple-
ment to pancreatic head resection with preservation of the pylorus in the treat-
ment of chronic pancreatitis. Am J Surg. 1985;150:608-611.
25. Madiba TE, Thomson SR. Restoration of continuity following pancreaticoduoden-
ectomy. Br J Surg. 1995;82:158-165.
26. Icard P, Dubois F. Pancreaticogastrostomy following pancreaticoduodenectomy.
Ann Surg. 1988;207:253-256.
27. Pikarsky AJ, Muggia-Sullam M, Eid A, et al. Pancreaticogastrostomy after pan-
creatoduodenectomy: a retrospective study of 28 patients. Arch Surg. 1997;
132:296-299.
28. Delcore R, Thomas JH, Pierce GE, Hermreck AS. Pancreaticogastrostomy: a safe drain-
age procedure after pancreaticoduodenectomy. Surgery. 1990;108:641-647.
29. Kapur BML, Misra M, Seenu V, Goel AK. Pancreaticogastrostomy for reconstruc-
tion of pancreatic stump after pancreaticoduodenectomy for ampullary carcinoma.
Am J Surg. 1998;176:274-278.
30. Ihse I, Axelson J, Hansson L. Pancreaticogastrostomy after subtotal pancreatec-
tomy for cancer. Dig Surg. 1999;16:389-392.
31. Morris DM, Ford RS. Pancreaticogastrostomy: preferred reconstruction for Whipple
resection. J Surg Res. 1993;54:122-125.
32. O’Neil S, Pickleman JR, Aranha GV. Pancreaticogastrostomy following
pancreaticoduodenectomy. World J Surg. 2001;25:567-571.
33. Aranha GV. Pancreaticogastrostomy. Oper Tech Gen Surg. 2001;3:54-60.
34. Tripodi AM, Sherwin CF. Experimental transplantation of the pancreas into the
stomach. Arch Surg. 1934;28:345-346.
35. Person EC Jr, Glenn F. Pancreaticogastrostomy: experimental transplantation of
the pancreas into the stomach. Arch Surg. 1939;39:530-550.
36. Waugh JM, Clagett OT. Resection of the duodenum and head of pancreas for
carcinoma: an analysis of thirty cases. Surgery. 1946;20:224-232.
37. Mackie JA, Rhoads JE, Park D. Pancreaticogastrostomy: a further evaluation.
Ann Surg. 1975;181:541-545.
38. Yeo CJ, Cameron JL, Sohn TA, et al. Six hundred fifty consecutive pancreatioc-
duodenectomies in the 1990s: pathology, complications and outcomes. Ann Surg.
1997;226:248-260.
(REPRINTED) ARCH SURG/ VOL 141, JUNE 2006
579
Downloaded from www.archsurg.com on September 9, 2008
©2006 American Medical Association. All rights reserved.
39. Mason GR. Pancreaticogastrostomy as reconstruction for pancreaticoduode-
nectomy: review. World J Surg. 1999;23:221-226.
40. Bartoli FG, Arnone GB, Ravera G, Bachi V. Pancreatic fistula and relative mor-
tality in the malignant disease after pancreaticoduodenectomy: review and sta-
tistical meta-analysis regarding 15 years of literature. Anticancer Res. 1991;11:
1831-1848.
41. Yeo CJ, Cameron JL, Maher MM, et al. A prospective randomized trial of pancre-
aticogastrostomy versus pancreaticojejunostomy after pancreaticoduodenectomy.
Ann Surg. 1995;222:580-592.
42. Choi SH, Moon HJ, Heo JS, Jott JW, Kim YI. Delayed hemmorrhage after
pancreaticoduodenectomy. J Am Coll Surg. 2004;199:186-191.
43. Sohn TA, Yeo CJ, Cameron JL, et al. Resected adenocarcinoma of the pancreas:
616 patient results, outcomes and prognostic indicators. J Gastrointest Surg.
2000;4:567-579.
44. Aranha GV, Hodul P, Golts E, Oh D, Pickleman J, Creech S. A comparison of pan-
creaticogastrostomy and pancreaticojejunostomy following
pancreatiocduodenectomy. J Gastrointest Surg. 2003;7:672-682.
45. Oussoultzoglou E, Bachellier P, Bigourdan J-M, Weber J-C, Nakano H, Jaeck D.
Pancreaticogastrostomy decreased relaparotomy caused by pancreatic fistula af-
ter pancreaticoduodenectomy compared with pancreaticojejunostomy Arch Surg.
2004;139:327-335.
46. Munoz-Bongrand N, Sauvanet A, Denys A, Sibert A, Vilgrain V, Belghiti J. Con-
servative management of pancreatic fistula after pancreaticoduodenectomy with
pancreaticogastrostomy. J Am Coll Surg. 2004;199:198-203.
47. Lin JW, Eng M, Cameron JL, Yeo CJ, Raill TS, Lillemoe KD. Risk factors and out-
comes in post-pancreaticoduodenectomy pancreaticocutaneous fistula. J Gas-
trointest Surg. 2004;8:951-959.
48. Aranha GV, Hodul PJ, Creech S, Jacobs W. Zero mortality after 152 consective
pancreaticoduodenectomies with pancreaticogastrostomy. J Am Coll Surg. 2003;
197:223-232.
49. Yeo CJ. Management of complications following pancreaticoduodenectomy. Surg
Clin North Am. 1995;75:913-924.
DISCUSSION
Michael B. Farnell, MD, Rochester, Minn: Dr Aranha and his
colleagues have been proponents of the use of pancreaticogas-
trostomy for reconstruction following pancreaticoduodenec-
tomy for a number of years. Their experience with 235 con-
secutive patients undergoing pancreaticogastrostomy for a variety
of indications, to my knowledge, is the largest reported to date.
The hypothesis in this retrospective analysis of prospectively
collected data is that pancreaticogastrostomy is a safe recon-
struction for a variety of periampullary conditions.
I believe that Dr Aranha and his colleagues’ data support
their conclusions that pancreaticogastrostomy is a safe alter-
native to pancreaticojejunostomy. The morbidity and mortal-
ity rates in this series are commendable. Pancreatic leak oc-
curred in 13.6% and delayed gastric emptying in only 6.0% of
patients. The mean length of stay was 9 days, and only 3 pa-
tients required reoperation. There were only 2 postoperative
deaths, for a mortality rate of 0.9%. Also, Dr Aranha pointed
out in his presentation, pancreatic fistula was correlated with
the consistency of the pancreas and is consistent with large se-
ries that one sees reported for pancreaticoduodenectomy. These
results are outstanding and are comparable to results obtained
in other high-volume centers employing more conventional re-
construction techniques consisting of either invagination or duct-
to-mucosa pancreaticojejunostomy.
The theoretic physiologic and technical advantages es-
poused by proponents of pancreaticogastrostomy include lack of
enzyme activation and alkaline milieu in the stomach, which may
prevent marginal ulceration, the superb blood supply to the stom-
ach, which may facilitate healing of the anastomosis, and naso-
gastric decompression, which may allow for lack of tension on
the anastomosis, unlike a jejunal limb, which may distend with
a pancreaticojejunostomy. The merits of these advantages, I am
sure, can be debated by proponents of pancreaticojejunal recon-
struction, many of whom report outstanding results comparable
to those we have heard from Dr Aranha today. Pancreatic recon-
struction has been analyzed in a prospective, randomized fash-
WWW.ARCHSURG.COM
ion by Yeo et al (Ann Surg. 1995;222:580-592), and in their study,
the morbidity and mortality was no different in pancreaticogas-
trostomy vs pancreaticojejunostomy.
Dr Aranha’s rich experience reported today and the surgi-
cal literature support the observation that either of these anas-
tomotic techniques are excellent for reconstruction with com-
parable safety. In other words, it doesn’t matter whether one
does pancreaticogastrostomy or pancreaticojejunostomy. What
really matters is that it is done well.
Heretofore, the durability of pancreatic anastomosis has not
been a high priority, unfortunately, and this is due to the poor
prognosis of many patients undergoing the operation. As the
indications for pancreaticoduodenectomy have broadened and
pancreatic surgeons are operating on more patients with cys-
tic neoplasms, the prognosis is improving and patients are liv-
ing for a longer period of time.
This leads me to my first question for Dr Aranha and that
has to do with patency of the anastomosis and preservation of
endocrine and exocrine function. The anastomosis with pan-
creaticogastrostomy is easily accessible endoscopically, and I
wondered if Dr Aranha has any experience with assessing pat-
ency and durability of his anastomosis and preservation of en-
docrine and exocrine function?
Second, in my practice I reconstruct using the duct-to-
mucosa technique, and should a leak concur, once it is well
controlled, I am comfortable feeding the patients and then ul-
timately dismissing them even with the leak persisting. Pre-
sumably with a leak following pancreaticogastrostomy, there
is extravasation from the stomach. Are you able to feed these
patients and are you able to dismiss them from the hospital, or
do you have to wait until it is completely healed?
Last, I was impressed with your low incidence of delayed
gastric emptying. In the manuscript, your protocol for post-
operative management included both erythromycin on day 4
and Reglan [metoclopramide hydrochloride] on day 5 when
the NG [nasogastric] tube was removed. To what extent do you
feel that this pharmacologic regimen is responsible for your very
low 6.0% incidence of delayed gastric emptying?
Dr Aranha: You asked about whether I have studied the pan-
creatic duct. We have not as of yet. I am in the process of get-
ting IRB [institutional review board] approval to do such a study,
but I do have quality-of-life studies that we presented recently
in Durban, South Africa, at the International Surgical Society
Week. We studied 88 patients who had a pancreaticogastros-
tomy and 44 patients with pancreaticojejunostomy. Patients who
had the pancreaticogastrostomy took pancreatic enzymes for
a longer period than those who had the pancreaticojejunos-
tomy, suggesting that acid does inactivate amylase, and there-
fore, patients with pancreaticogastrostomy have more steator-
rhea. Overall, 40% of patients with pancreaticogastrostomy were
taking enzymes for more than 1 year after their Whipple pro-
cedure. However, 60% had stopped taking pancreatic en-
zymes.
The incidence of diabetes was 9% overall, but it was sig-
nificantly lower in those with pancreaticogastrostomy than in
those who have pancreaticojejunostomy. My feeling has al-
(REPRINTED) ARCH SURG/ VOL 141, JUNE 2006
580
Downloaded from www.archsurg.com on September 9, 2008
©2006 American Medical Association. All rights reserved.
ways been that diabetes occurred because the duct became ob-
structed, but this has not been studied in a prospective fash-
ion. There are other reasons for patients getting diabetes. The
amount of pancreas one removes may result in a decrease of
pancreatic polypeptide, and this may make the liver resistant
to insulin.
Your second question was, how do we treat leaks? If I have a
leak from the pancreaticogastrostomy, I will continue to feed the
patient. If the amount of fluid from the drain increases, I get a
drain study. Of the 23 leaks that we had, 6 patients had the pan-
creatic drain erode into the anastomosis. The drain was pulled
back, and the leak was closed. Also, one needs to get a CT scan
to make sure that there is no abscess that has to be drained. If the
patient can eat and the volume does not go up, we send the pa-
tient home with the drain until the fistula closes, at which time
we remove the drain. Fewer than 5 patients have had to stay in
the hospital and receive total parenteral nutrition.
Finally, your question about erythromycin and Reglan as
prokinetic agents. Erythromycin, as most of you know, is a mo-
tilin agonist. Motilin is concentrated in the duodenum, and there-
fore, loss of motilin is thought to be the reason for delayed gas-
tric emptying that occurs more in the pylorus-preserving
Whipple procedure. I think one of the reasons that we do not
have the same incidence of delayed gastric emptying is be-
cause we do the classic Whipple.
Lawrence J. Koep, MD, Phoenix, Ariz: The real advantage
of this drainage is access to the pancreas. We have continued
to do this despite the Baltimore data, like you in many cases,
whether it is papillary disease in the pancreatic duct or whether
it is pancreatitis, where it is essential to be able to get access to
that pancreatic duct as time goes on. The question of whether
you have been able to access this so far is really critical. We
have tried to do this. Early on we can do it up to about a year;
we can access that pancreatic duct, watching what is going on.
Then after that, we are having a lot of difficulty finding the pan-
creatic duct in the stomach because it seems like they lose it.
My question is, is this something we are going to be able to do?
Do you think that we will be able to access the anastomosis long
term and know what is happening to the pancreatic duct and
to exocrine function?
Dr Aranha: You asked if we have studied the patency of the
pancreatic duct after a pancreaticogastrostomy. We have not
as of yet. We did have 2 patients who had recurring attacks of
pancreatitis. On these patients, we did an upper GI [gastroin-
testinal] endoscopy and gave them secretin, and we were able
to identify the duct. I believe that over time, the gastric mu-
cosa may grow over the end of the pancreatic duct, but I have
not been able to prove this with endoscopic studies.
Your second question was in regard to patients with intra-
ductal papillary mucinous neoplasms. In patients who have a
Whipple procedure for an intraductal papillary mucinous neo-
plasm, a pancreaticogastrostomy is suggested in the manage-
ment of pancreatic remnant because the remnant can be ob-
served for changes by an endoscopic ultrasound. In this situation,
the Johns Hopkins group also supports pancreaticogastrostomy.
WWW.ARCHSURG.COM