1. ORIGINAL ARTICLE
Postoperative Pancreatic Fistulas
Preventing Severe Complications and Reducing Reoperation
and Mortality Rate
Sergio Pedrazzoli, MD, FACS,* Guido Liessi, MD,† Claudio Pasquali, MD,* Roberto Ragazzi, MD,‡
Mattia Berselli, MD,* and Cosimo Sperti, MD*
them. In this article, we will report our experience with the
Background: Postoperative pancreatic ﬁstula (POPF) is responsible for
treatment of pancreatic ﬁstulas.
severe complications and death in patients who underwent pancreatic
The reported success rate of the conservative treatment of
surgery. The reported success rate of conservative treatment is around
a POPF is about 80%.4 Patients with a high output ﬁstula due to
80%. Therefore up to 20% of patients usually need surgical treatment that can
anastomotic or pancreatic leak and with signs of severe sepsis or
be repeated in some. Uncontrolled sepsis and massive hemorrhage are the main
hemorrhage cannot be managed by other means and should
causes for mortality in this setting.
undergo laparotomy.4 The reoperation rate and the mortality rate
Method: Four hundred forty-ﬁve patients underwent surgery for pancreatic
vary greatly, ranging from 2.8% to 65.8% and from 0% to 22%,
diseases (January 1993–August 2007); 70 of them developed a POPF. An
respectively.2,3,9,13–17 After stating that “Pancreatic ﬁstula no
early aggressive treatment based on interventional radiology was applied to
longer seems to be a major problem,”2 Markus Buchler from
all patients. The drain’s track and/or percutaneous approach was used to
Heidelberg reported that pancreatic ﬁstula is still a problem that
insert catheters into the peripancreatic ﬂuid collection/s or abscess/es. The
in his recent experience prompted 23 completion pancreatecto-
position of catheters was veriﬁed at least once a week. Surgery was
mies with a mortality rate of 39.1%.18
performed in case of failure of conservative approach.
The postoperative management of surgical drains may be
Results: Conservative treatment (approach by drain’s track in 49, percu-
one of the keys to decrease the complication rate and their
taneous in 16, mixed in 2) was successful in 67 patients. A patient under
severity. Prophylactic drains after pancreatic surgery allow mon-
dialysis had the drains inserted during an emergency surgery for perito-
itoring of the occurrence of intra-abdominal bleeding, the detec-
nitis 6 days after surgery; a second patient underwent repeated surgical
tion and drainage of a pancreatic, biliary, or enteric ﬁstula.
debridement, and a third patient underwent a procedure on the abdominal
However surgical placement of drains may result in an increased
wall to separate a POPF from a colonic ﬁstula. No patient with diagnosed
risk of intra-abdominal infections due to infection through the
drain track. Drains left for more than 4 days are reported to
Conclusions: Early aggressive interventional radiology allowed managing
increase signiﬁcantly the pancreatic ﬁstula rate and the rate of
conservatively 95.7% of POPF preventing severe complications and avoid-
intra-abdominal infections.19 Drains left in place for several days
may give also other complications, like enteric ﬁstula due to
(Ann Surg 2009;249: 97–104) decubitus of the surgical drain/s. It has been also suggested that
prophylactic drains after pancreatic surgery are useless.20
We conducted a retrospective study to see if our policy to
remove or to exchange the prophylactic drains within 5 to 8 days
after surgery prevented further surgery and mortality in POPF
P ostoperative pancreatic ﬁstula (POPF) has been for many years
the main concern for surgeons performing pancreatic resec-
tions because it was responsible for an increased mortality rate
and length of hospital stay.1 Although some authors report that PATIENTS AND METHODS
in recent years the mortality rate due to a pancreatic ﬁstula is
substantially decreased,2,3 the impact on the postoperative com- Data Collection
plication rate is still signiﬁcant. The reported ﬁstula rate after We reviewed the medical records of 445 consecutive patients
pancreaticoduodenectomy (PD) ranges from 0% to 2%2,3 to more that underwent surgical treatment for pancreatic or periampullary
than 20%,4,5 after distal pancreatectomy (DP) from 0% to disease by our surgical team between January 1 1993 and August 31
60%,6 – 8 after central pancreatectomy from 0% to 40%,9 and 2007. During this period, our surgical activity was performed in 3
after enucleation of an endocrine or cystic neoplasm from different periods:
25% to 50%.10 –12 1. January 1993–December 1995: The activity was performed as
The occurrence of pancreatic ﬁstulas draws 3 different Consultant for pancreatic surgery of a Unit of General Surgery in
questions: how to prevent, how to diagnose, and how to treat Castelfranco Veneto.
2. January 1996 –August 2003: Unit of Special Surgical Pathology
in Castelfranco Veneto.
From the *Clinica Chirurgica IV, University of Padova; †Radiology, General 3. September 2003–August 2007: IV Surgical Clinic, Padua.
Hospital, Castelfranco Veneto (Treviso); and ‡Radiology, Azienda Osped-
aliera, Padova, Treviso, Italy. Clinical presentation, operative treatment, and postoperative
Reprints: Sergio Pedrazzoli, MD, FACS, IV Surgical Clinic, Department of course were abstracted from the clinical records. The review in-
Medical and Surgical Sciences, University of Padua, Via Giustiniani, 2, 35128 cluded patient demographics (age and sex), surgical data (type of
Padova, Italy. E-mail: email@example.com.
Copyright © 2009 by Lippincott Williams & Wilkins
surgery), pathologic diagnosis (pancreatic adenocarcinoma, ampul-
ISSN: 0003-4932/09/24901-0097 lary tumor, chronic pancreatitis, bile duct tumor, benign and malig-
DOI: 10.1097/SLA.0b013e31819274fe nant cystic neoplasms, intraductal papillary mucinous neoplasms,
Annals of Surgery • Volume 249, Number 1, January 2009 97
2. Pedrazzoli et al Annals of Surgery • Volume 249, Number 1, January 2009
islet cell tumor, duodenal tumor, and others), postoperative morbid- gram was performed at least 7 days after surgery, and the drain
ity, and mortality rates and length of hospital stay. The amount, and replaced with 8 –12 Fr pigtail or malecot catheters by an interven-
the amylase content, of the ﬂuid coming from surgically placed tional radiologist. The number of catheters was chosen on the basis
drains and registered during the postoperative period was also of the number of secondary collections visualized by the ﬁstulo-
collected. Pancreatic ﬁstula was deﬁned according to the criteria of gram. If a pancreatic ﬁstula was diagnosed, one of the catheters was
Yeo et al21: “pancreatic ﬁstula was diagnosed when more than 50 inserted as soon as possible through the ﬁstula into the bowel’s
mL of drainage ﬂuid with an amylase concentration of more than lumen, or as close to the disrupted pancreatic duct as possible22
3-fold the normal upper limit of serum was obtained on or after (Figs. 1, 2). Aim of the radiologic procedure was obtaining a thin
postoperative day 10, or when pancreatic anastomotic disruption and straight external ﬁstula that closes spontaneously after removal
was shown radiologically.” The severity of POPFs was classiﬁed of the drain. The persistence or appearance of purulent material from
according to the ISGPF grading.5 the drains, or of patient’s septic conditions, meant that the ﬁstula
During the study period, our policy with surgical drains was track was not adequately drained and a ﬁstulogram was immediately
as follows: in presence of clear ﬂuid with normal amylase content, performed. Whenever an US or a CT allowed detecting a peripan-
drains were removed as soon as possible, usually within 5 to 8 days. creatic ﬂuid collection or abscess after removal of surgical drains, or
Whenever an amylase rich or unclear ﬂuid was detected, a ﬁstulo- distant from drains still in place, a percutaneous drainage was
FIGURE 1. POPF in a patient who underwent
PPPD for pancreatic cancer. A, The fistulogram
through the surgical drain shows an irregular
fluid collection. B, Two pigtail catheters are
placed in the cavity. C and D, The fistulogram
after 7 and 14 days shows an unchanged cavity.
FIGURE 2. A, A pigtail catheter is inserted into
the bowel through the fistula. B, A second pig-
tail is put just in front of the fistula. C, One
week later the fluid collection is much de-
creased. D, A further week later, the catheters
can be removed safely.
98 © 2009 Lippincott Williams & Wilkins
3. Annals of Surgery • Volume 249, Number 1, January 2009 Postoperative Pancreatic Fistula
performed. A ﬁstulogram performed once or twice a week allowed
verifying that an adequate drainage was obtained (Figs. 1, 2). Oral TABLE 1. Distribution of the Surgical Procedures During
feeding was maintained whenever possible in these cases: ﬁstula the 3 Periods of the Study
excluded from the alimentary tract, transected pancreatic stump, First Second Third
Roux en Y pancreaticojejunostomy. Octreotide (0.1 mg s.c. each 8 Surgical Procedure Period Period Period Total
hours) was administered only when the ﬁstula output was 200 mL Pancreaticoduodenectomy 9 (1) 69 (7) 116 (11) 194 (19)
of nonpurulent ﬂuid.
Whipple 4 12 16 32
Surgical Technique PPPD* 5 56 96 157
Pylorus preserving pancreaticoduodenectomy (PPPD) with PPPD† 0 1 4 5
isolated Roux loop pancreaticojejunostomy was the procedure of Distal pancreatectomy 11 (2) 54 (6) 45 (12) 110 (18)
choice whenever possible. A Whipple (PD) with isolated Roux loop Spleen preserving 7 14 10 31
pancreaticojejunostomy was performed when pylorus preservation Nonspleen preserving 3 38 29 70
was contraindicated. A pancreaticogastrostomy was performed Subtotal nonspleen preserving 0 1 3 4
mainly for intraductal papillary mucinous neoplasm patients. Stent- Subtotal spleen preserving 1 0 1 2
ing of the pancreatic anastomosis and ﬁbrin glue sealant was not Appleby procedure 0 1 2 3
used during the study period. Central pancreatectomy (CP), duode- Total pancreatectomy 1 8 6 15
num preserving pancreatic head resection (DPPHR), and enucleation
DPPHR 5 (2) 8 (4) 8 (3) 21 (9)
were performed as previously described.23–25 Pancreaticojejunos-
tomy was usually performed as previously described26 Standard With pancreaticojejunostomy 3 7 6 16
procedures were used for distal pancreatectomy, subtotal left pan- With pancreaticogastrostomy 0 1 0 1
createctomy (75%–90%), and pancreatico/cysto-jejunostomy. Spleen Without anastomosis‡ 2 0 2 4
was preserved whenever possible. Central pancreatectomy 1 (0) 17 (7) 13 (9) 31 (16)
Enucleation 2 (0) 20 (4) 9 (1) 31 (5)
Statistical Analysis Pancreaticojejunostomy 2 (0) 11 (0) 10 (0) 23 (0)
The 2 test was used to compare categorical variables. Stu- Frey 0 4 9 13
dent t test was used to compare continuous variables; results are Partington-Rochelle 0 3 1 4
presented as mean SD unless otherwise indicated. Binary logistic
Puestow 0 2 0 2
regression analysis was made considering the presence or absence of
ﬁstula as dependent variable, whereas sex, surgery type, and histol- Pedrazzoli50 2 2 0 4
ogy as predictors. A P value 0.05 was considered statistically Pancreaticojejunostomy§ 1 (0) 0 1 (0) 2 (0)
signiﬁcant. The analysis was performed with SPSS statistical soft- Cystojejunostomy 1 (0) 7 (0) 6 (0) 14 (0)
ware release 13.0 (SPSS Inc, Chicago, IL). Remaking 1 (0) 1 (1) 0 2 (1)
Total duodenectomy 0 0 1 (0) 1 (0)
Resection of minor papilla 0 0 1 (0) 1 (0)
Patient Characteristics Total 34 (5) 195 (29) 216 (36) 445 (70)
From January 1993 to August 2007, 445 consecutive patients The number of pancreatic ﬁstulas reported within brackets.
underwent resective surgical treatment for pancreatic or periampul- *With isolated Roux loop pancreaticojejunostomy.
lary disease by our surgical team. Palliative and explorative surgery †
for pancreatic or periampullary cancer were excluded. A POPF was Wide atypical head resection without interruption of the Wirsung duct included in
diagnosed, according to the criteria of Yeo et al,21 in 70 patients. §
Triple palliative procedure, including GEA and hepaticojejunostomy, for aborted PD.
The mean age of patients with POPF was 57.5 14.1 years (range ¶
Stenosis of the anastomosis after PD.
28 – 87), whereas it was 59.3 13.9 years (range 14 – 86) in 375
patients without pancreatic ﬁstula (P 0.52). Twenty-ﬁve (35.7%)
of the patients with POPF were male and 45 (64.3%) were female, The distribution of the surgical procedures during the 3
whereas 205 (54.7%) of the patients without POPF were male and periods of the study is reported in Table 1. Seventy patients (15.7%)
170 (45.3%) female (P 0.005). developed a POPF, with a median peak amylase level of 12.514 U/L
(range 3–272.400) in the abdominal drainage ﬂuid. The incidence of
Operative Results POPFs was unchanged during the 3 periods of the study.
A PPPD was performed in 162 patients with 5 pancreati- The development of a POPF depended to a considerable
cogastrostomies, a PD in 32, a DP in 110 (spleen was preserved degree on the procedure performed (Table 1) and on the pathology
in 33), a CP in 31, a DPPHR in 21 (4 of them underwent a wide (Table 3). The rate of POPF was 9.8% (19/194) for PD and PPPD,
atypical head resection without interruption of the Wirsung duct 16.4% (18/110) for DP, 16.1% (5/31) for enucleation, 42.9% (9/21)
and were included in DPPHR group), an enucleation in 31, a for DPPHR, and 51.6% (16/31) for CP ( 2 52.5, P 0.0001). A
pancreatico-cysto/-jejunostomy in 41, a total pancreatectomy in low POPF rate (6.4%) was registered for the 47 patients operated for
15, a total duodenectomy, and a resection of the minor papilla in chronic pancreatitis, whereas it was high for serous cystadenomas
1 each (Table 1). (37.9%), mucinous cystadenomas (29.4%), and endocrine pancreatic
tumors (21.7%), and an increased ﬁstula rate was registered also for
Postoperative Outcomes bile duct cancer (18.2%), duodenal cancer, and mixed benign dis-
Two hundred ﬁfty-ﬁve patients (57.3%) had an uneventful eases (16.7%) ( 2 23.3, P 0.003).
postoperative course. The overall morbidity and mortality rate was After binary logistic regression analysis, only procedure (P
42.7% (190/445) and 3.1% (14/445), respectively. No patient with 0.03) and pathology (P 0.02) remained statistically signiﬁcant on
POPF died. The cause of death for the 14 patients without POPF is the development of a POPF, whereas the difference was no more
reported in Table 2. signiﬁcant for sex (P 0.47).
© 2009 Lippincott Williams & Wilkins 99
4. Pedrazzoli et al Annals of Surgery • Volume 249, Number 1, January 2009
TABLE 2. Cause of Death in 14 Patients
Patient Year Age Sex ASA Surgical Procedure Complication Cause of Death PO (d)
1 1995 70 F 3 PD Pentalobar pneumonia ARDS 42
2 1997 65 M 3 PPPD AHG, Bleeding pseudo-aneurysm of the RHA MOF 42
3 1999 64 M 3 DP MI MI 12
4 2000 70 M 3 PPPD GI hemorrhage, Respiratory failure Uncontrolled bleeding 15
5 2000 78 M 3 PPPD Duodenojejunostomy leak, sepsis MI 17
6 2001 51 F 1 DP Pulmonary embolism Pulmonary embolism 1
7 2003 62 F 1 PPPD Cardiac tamponade Cardiac tamponade 1
8 2004 72 M 3 TP Early massive hemorrhage MOF 141
9 2005 65 F 2 PD Delayed massive hemorrhage Uncontrolled bleeding 12
10 2006 64 F 3 Palliative DP* Hepatic insufﬁciency MOF 43
11 2006 67 F 2 TP Necrosis of segments IV–V due to an Massive peritoneal bleeding 57
12 2007 78 M 3 PD Peritoneal bleeding MOF 29
13 2007 69 M 2 PPPD Hemorrhage from PJ MI 12
14 2007 71 M 1 TP Acute hemorrhagic gastritis Massive peritoneal bleeding 0
A silent POPF was found at autopsy of patient 9. A POPF was excluded by surgery and/or autopsy in all the other patients.
*Endocrine pancreatic tumor with liver metastases and bleeding from the invaded stomach.
PD indicates Whipple pancreaticoduodenectomy; PPPD, pylorus preserving pancreticoduodenec-tomy; DP, distal pancreatectomy; TP, total pancreatectomy; ARDS, acute
respiratory distress syndrome; MOF, multiple-organ failure; AHG, acute hemorrhagic gastritis; RHA, right hepatic artery; MI, myocardial infarction; PJ, pancreaticojejunostomy.
tional radiology. A second patient had a simultaneous POPF and
TABLE 3. Distribution of POPF according to the pathology a colonic ﬁstula after PPPD and colonic resection for colon
Pathology POPF No POPF Total % PF cancer. A simple procedure on the abdominal wall with local
anesthesia allowed separating the 2 ﬁstula tracks that then healed
Pancreatic cancer 17 119 136 12.5
Cancer of Vater’s papilla 1 17 18 5.5 The different treatments of the 70 POPFs are reported in
Duodenal cancer 1 5 6 16.7 Table 5. The surgical drain was replaced by the interventional
Bile duct cancer 2 9 11 18.2 radiologist in 51 patients (72.9%), and 2 of them (2.9%) also
Pancreatic metastasis* 3 8 1† 11 1† 25.0 underwent a percutaneous approach. The percutaneous approach
Adenoma of Vater’s papilla 0 1 1 0 only was used in 16 patients (22.9%). A TPN was maintained until
IPMN 3 28 1† 31 1† 9.4 an adequate oral feeding was obtained. This was achieved within 10
Serous cystadenoma 11 18 29 37.9 days in 42/70 patients (60%), whereas in 28, TPN was maintained
Mucinous cystadenoma 5 12 17 29.4 for a mean of 21.5 days (median 20, range 11–71 days). Sepsis was
Mucinous cystadenocarcinoma 0 4 4 0 registered in 22/70 patients (31.4%) 14 of whom had associated
delayed gastric empting. All patients were on oral feeding before
Endocrine pancreatic tumor 18 65 83 21.7
ﬁstula closure. Antibiotic treatment was maintained until clear ﬂuid
Papillary solido cystic tumor 1 3 4 25.0
came from angiographic drains and any sign of infection disap-
Chronic pancreatitis 3 44 47 6.4 peared.
Pancreatic pseudocyst 0 16 16 0 A further pancreatic ﬁstula was diagnosed at autopsy in a
Mixed benign diseases 5 25 30 16.7 patient (Table 2, patient 9) that died of GI and peritoneal bleeding in
Total 70 375 445 15.7 a few minutes after an otherwise uneventful course. A CT scan
*Renal cancer (5) and colon cancer (6).
performed the evening before because of an unexplained increase in
One patient with 2 diseases. white-cell count had shown only a small (2 3 cm) peripancreatic
Postoperative hospital stay was signiﬁcantly longer (t 8.87,
The distribution of Grade A, B, C POPF according to the P 0.0001) for patients with POPF (mean: 29.8 16.7, median:
deﬁnitions of the ISGPF5 among the different surgical procedures is 25, range 10 –95 days) than for patients without POPF (mean: 15.3
reported in Table 4. Type A POPFs were observed in 29/70 (41.4%), 11.5), median: 12, range: 6 –102).
type B in 38/70 (54.3%), and type C in only 3 patients (4.2%) (1
after DPPHR, and 2 after PPPD). Three patients underwent surgical
treatment, but only 1 was operated after failure of a prolonged DISCUSSION
conservative treatment, whereas the 2 others underwent surgery Pancreatic surgery is becoming more prevalent and safer,
because of an acute abdomen or to facilitate conservative treatment. with a mortality rate well below 5% in high volume centers, but it
Surgical treatment was performed after a DPPHR in a patient due to is still associated with signiﬁcant morbidity, even in the most
the presence of infected necrosis in the area of the resected pancre- experienced hands.2,3,13–15,17,19,20 POPF remains the most problem-
atic head. In a patient under dialysis for renal insufﬁciency, an acute atic and feared common complication.
abdomen 6 days after a PPPD prompted an emergent explorative POPFs can be approached by 3 different aspects: how to
laparotomy. A small pancreatic ﬁstula was found and treated suc- prevent,2,3,7,8,27,28 how to diagnose,5,29,30 and how to treat
cessfully with angiographic catheters positioned as for interven- them.31
100 © 2009 Lippincott Williams & Wilkins
5. Annals of Surgery • Volume 249, Number 1, January 2009 Postoperative Pancreatic Fistula
analogues in reducing complications associated with pancreatic
TABLE 4. Distribution of Pancreatic Fistulas According to surgery did not show any effect on the rate of clinical leak.27
the Surgical Procedures and the ISGPF Grading5 Somatostatin analog prophylaxis was seldom used in our patients.
Grade Grade Grade The usefulness of stenting the pancreatic anastomosis is still to be
Surgical Procedure A B C Total proved.32,33 We never stented the anastomosis during the study
Pancreaticoduodenectomy 6 11 2 19 period, and we avoided using ﬁbrin sealant after completing a
Whipple 1 1 0 2 negative prospective randomized trial.34
We applied a pragmatic approach to the diagnosis of POPFs: the
PPPD* 5 10 2 17
criteria of Yeo et al,21 were applied “a posteriori” and therefore we may
PPPD† 0 0 0 0
have missed some biochemical (grade A) POPFs. Collected retrospec-
Distal pancreatectomy 7 13 0 20 tive data were insufﬁcient to use other criteria for the deﬁnition of a
Spleen preserving 2 4 6 POPF. However, our incidence of POPF is within the range of the
Nonspleen preserving 2 8 10 reported incidence for the different surgical procedures (Table 1)
Subtotal non spleen preserving 3 0 3 and different pathologies (Table 3) included in the study.2–12
Subtotal spleen preserving 0 1 1 We analyzed all POPFs we have diagnosed during a
Appleby procedure 0 0 0 14.7-year period, independently on the surgical procedure that
DPPHR 6 2 1 9 was responsible for them. This pattern corresponds better to the
With pancreaticojejunostomy 6 2 1 9 variety of clinical situations encountered in the surgical practice.
With pancreaticogastrostomy 0 0 0 0 The main results are zero mortality, zero completion pancreate-
Without anastomosis‡ 0 0 0 0
ctomies, and a 4.2% reoperation rate for 70 consecutive POPFs.
In spite of the large number of POPFs included in the study, our
Central pancreatectomy 5 11 0 16
results compare favorably with similar data collected from the
Enucleation 4 1 0 5 literature (Table 6).2,13–16,20,28,29,33,35– 44
Remaking pancreaticojejunostomy§ 1 0 0 1 The best way to prevent severe complications from a POPF is
Total 29 38 3 70 to obtain a perfect drainage without stasis and consequent infection
*With isolated Roux loop pancreaticojejunostomy. of the pancreatic juice. Prophylactic drains after pancreatic surgery
With pancreaticogastrostomy. allow monitoring of the occurrence of intra-abdominal bleeding and
Wide atypical head resection without interruption of the Wirsung duct included in the detection and drainage of a pancreatic ﬁstula.19,20 However, the
Stenosis of the anastomosis after PD.
results and the management of surgical drains are widely debated.
Some authors reported that drains are useless,20 some others sug-
gested a very early (1–3 days),45,46 or early (4 days) removal,19
Several studies tried to reduce the POPF ﬁstula rate with many others maintain the drains in place for 7 or more days.30 The
octreotide and/or stenting the anastomosis and/or using the ﬁbrin main concerns of maintaining the drains in place are an increased
glue sealant. A meta-analysis on the value of somatostatin and its rate of intra-abdominal abscess and infected abdominal collec-
TABLE 5. Distribution of Pancreatic Fistulas According to the Surgical Procedures
Surgical Procedure Percutaneous Drain Substitution Mixed* Surgery Total
Pancreaticoduodenectomy 3 14 0 2 19
Whipple 0 2 0 2
PPPD† 3 12 2 17
PPPD‡ 0 0 0 0
Distal pancreatectomy 7 13 0 0 20
Spleen preserving 1 5 6
Non spleen preserving 6 4 10
Subtotal nonspleen preserving 0 3 3
Subtotal spleen preserving 0 1 1
Appleby procedure 0 0 0
DPPHR 1 6 1 1 9
With pancreaticojejunostomy 1 6 1 1 9
With pancreaticogastrostomy 0 0 0 0 0
Without anastomosis§ 0 0 0 0 0
Central pancreatectomy 4 11 1 0 16
Enucleation 1 4 0 0 5
Remaking pancreaticojejunostomy¶ 0 1 0 0 1
Total 16 49 2 3 70
*Mixed percutaneous and drain substitution.
With isolated Roux loop pancreaticojejunostomy.
Wide atypical head resection without interruption of the Wirsung duct included in DPPHR group.
Stenosis of the anastomosis after PD.
© 2009 Lippincott Williams & Wilkins 101
6. Pedrazzoli et al Annals of Surgery • Volume 249, Number 1, January 2009
TABLE 6. Incidence, Reoperation, and Mortality Rate of POPF
Author (Year) Procedures N° N° N° % CoPa Mortality Rate %
Bottger et al. (1999)35 PD 186 20 NR 10
Buchler et al. (2000)2 PD, PPPD, DPPHR 331 7 1 14.3 0 0
Conlon et al. (2001)20 PD, PPPD, DP 179 18 NR 0
Bassi et al. (2001)36 PD 150 16 1 6.2 0 0
Adam et al. (2001)37 PD, PPPD, DPPHR, DP 345 34 8 23.5 0 12
Imaizumi et al. (2002)33 PD, DPPHR, CP 148 11 NR 9.1
Sarr et al. (2003)38 PD, DP, CP 275 56 NR 1.8
Gueroult et al. (2004)39 PD 282 38 7 18.4 7 7.9
Lin et al. (2004)13 PD, PPPD 1891 216 17 7.9 2 1.4
de Castro et al. (2005)14 PD 459 41 27 65.8 9 14.6
Kazanjian et al. (2005)15 PD, PPPD 437 55 3 5.5 0 1.8
Muscari et al. (2005)16 PD, PPPD 300 50 NR 22
Shinchi et al. (2006)29 PD, PPPD 207 29 0 0.0 0 0
Satoi et al. (2006)40 PD 198 27 2 7.4 2 14.8
Vanounou et al. (2007)28 PD, PPPD 227 60 3 5.0 0 0
Crippa S et al. (2007)41 CP 100 44 0 0.0 0 0
Kollmar et al. (2007)42 PD, PPPD, CP, DPPHR, CJ 391 98 5* 33.3 NR NR
Liang et al. (2007)43 PD 100 32 0 0.0 0 3.1
Murakami et al. (2008)44 PD, PPPD† 150 11 0 0.0 0 0
Present series PD, PPPD, DP, CP, DPPHR, E, SP, PCJ 445 70 3 4.2 0 0
*Five of 67 patients; the fate of 324 patients (83 POPFs) was not reported.
CoPa indicates completion pancreatectomy; PD, pancreaticoduodenectomy; PPPD, pylorus preserving pancreaticoduodenectomy; DPPHR, duodenum
preserving pancreatic head resection; DP, distal pancreatectomy; CP, central pancreatectomy; CJ, cistojejunostomy; E, enucleation; SP, subtotal left
pancreatectomy; PCJ, pancreatico/cisto-jejunostomy.
tions,19 and the perforation of a hollow viscus due to decubitus of due to ﬁstulization of a pancreaticojejunostomy 6 days after a PPPD
the surgical drain. However, not all POPFs are drained by the was treated, after an adequate peritoneal cleaning, by intraoper-
surgical drains left in place and a percutaneous, or surgical, ap- ative positioning of the angiographic drains in the same way as
proach may become necessary. the POPFs treated conservatively, and, after then, had an un-
In our series, the time of removal of the surgical drains was eventful course. A communication with the lumen of the pancre-
not established a priori, but the drains were removed within 5 to 8 aticojejunal anastomosis, or with the pancreatic duct for stump
days, and, in any case, as soon as a clear ﬂuid with low amylase ﬁstulas, was demonstrated in almost all our patients. This kind of
content was seen in a patient without any sign of infection. This treatment was applied in the same way during the 3 periods of the
allowed draining the majority of our POPFs through the drain’s study, with the same success rate.
track (Table 5), to reduce the number of percutaneous approaches, Unfortunately the appearance of a ﬁstula track adequately
and prevent severe complications needing surgical treatment. Too sealed-off from the peritoneal cavity needs at least 6 to 8 days. This
early removal of peripancreatic drains may delay the diagnosis of means that an early change of the surgical drains is not advisable.
POPF, eliminate an easy way to drain it through the drain’s track,
The use of computed tomography scans and ﬁstulogram postoper-
and prompt the need of a percutaneous or surgical approach. The
atively to guide interventional radiology to drain undrained areas
standardized early removal of pancreatic drains45,46 may be one of
and to reposition malfunctioning drains was one of the key stones
the contributing factors to the 23 completion pancreatectomies
performed between 2001 and 2006 in Heidelberg with a mortality in reducing mortality after pancreatic surgery.31,49 In our expe-
rate of 39.1%.18 rience, interventional radiology was used not only for percutaneous
Management of peripancreatic drains similar to ours was aspiration and drainage in symptomatic patients31,49 but also to use
reported by Kazanjian et al,15 with a very low reoperation and extensively the existing surgical drains as an access route to postoper-
mortality rate and no completion pancreatectomy. Sterile substitu- ative intra-abdominal ﬂuid collection/s or abscess/es.47,48 A percutane-
tion of surgical drains with angiographic catheters is very simple and ous access was used alone (16) or as an adjunct (2) in only 18/70
useful,47,48 and reduces the probability of an infection through the patients (Table 5).
drain track. Furthermore, draining the ﬁstula at its origin through the In our series, the rate of severe (grade C) POPF5 was quite
disrupted pancreatic anastomosis22 (Figs. 1, 2), or in close proximity low 3/70 (4.2%). However, the incidence of true severe (grade C)
to the disrupted Wirsung duct, reduces stagnation and consequent POPF is largely dependent on the attitude of the surgical team, as
infection of the pancreatic juice. After obtaining a thin and straight surgeons that chose more frequently to perform a reoperation in
external ﬁstula, it closes spontaneously after removal of the drain. patients with POPF will have an higher rate of grade C POPFs and
Also our patient who underwent emergency surgery for peritonitis vice versa. In fact reoperation is the only No/Yes clue between grade
102 © 2009 Lippincott Williams & Wilkins
7. Annals of Surgery • Volume 249, Number 1, January 2009 Postoperative Pancreatic Fistula
B and C POPF. A different grading system, based on more objective 17. Aranha GV, Aaron JM, Shoup M, et al. Current management of pancreatic
data, will be useful. ﬁstula after pancreaticoduodenectomy. Surgery. 2006;140:561–569.
A POPF was diagnosed at autopsy in a patient (Table 2, patient 18. Muller MW, Friess H, Kleeff J, et al. Total pancreatectomy – Renaissance of
a formerly abandoned surgical procedure. Ann Surg. 2007;246:966 –975.
9) that died of massive GI and peritoneal bleeding in a few minutes after
19. Kawai M, Tani M, Terasawa H, et al. Early removal of prophylactic drains
an otherwise uneventful course. A CT scan performed the evening reduces the risk of intra-abdominal infections in patients with pancreatic head
before because of an unexplained increase in white-cell count had resection. Prospective study for 104 consecutive patients. Ann Surg. 2006;
shown only a small (2 3 cm) peripancreatic ﬂuid collection. The 244:1–7.
patient died before any further diagnostic procedure or treatment was 20. Conlon KC, Labow D, Leung D, et al. Prospective randomized clinical trial
attempted, and therefore excluded from our POPF series. of the value of intraperitoneal drainage after pancreatic resection. Ann Surg.
In conclusion, complete prevention of POPFs after pancreatic 2001;234:487– 494.
surgery is still a dream. Therefore, facing a patient at risk for POPF, 21. Yeo CJ, Cameron JL, Maher MM, et al. A prospective randomized trial of
pancreaticogastrostomy versus pancreaticojejunostomy after pancreaticoduo-
we need to be ready to do an early diagnosis, and to manage the denectomy. Ann Surg. 1995;222:580 –592.
POPF with a proper aggressiveness, if we will decrease the mortality 22. Boverie JH, Remont A. Percutaneous management of ﬁstulas in the digestive
rate close to 0. Early interventional radiology aimed to capture tract. In: Dondelinger RF, Rossi P, Kurdziel JC, et al, eds. Interventional
pancreatic juice as close to its origin as possible may prevent Radiology. Stuttgart, Germany: Georg Thieme Verlag;1990:746 –752.
reoperative surgery in most of our patients. Furthermore, using the 23. Sperti C, Pasquali C, Ferronato A, et al. Median pancreatectomy for tumors
surgical drain/s as an access route minimized the need of a percu- of the neck and body of the pancreas. J Am Coll Surg. 2000;190:711–716.
taneous procedure. 24. Pedrazzoli S, Sperti C, Pasquali C. Pancreatic head resection for noninﬂam-
matory benign lesions of the head of the pancreas. Pancreas. 2001;23:309 –
25. Pasquali C, Sperti C, Baratella P, et al. Enucleoresezione di tumori neuroen-
The authors thank Dr. Daniela Basso of Laboratory Medicine docrini pancreatici. Venticinque anni di esperienza. Suppl Tumori. 2005;4:
of Azienda Ospedaliera, Padova for her assistance with statistical S59 –S60.
analysis, and Mrs. Lazzarin Tania and Callegari Simona for their 26. Pedrazzoli S, Sperti C, Pasquali C. An easier technique for end to end
contribution in retrieval of clinical records. pancreaticojejunostomy. HPB Surgery. 1996;9:141–143.
27. Connor S, Alexakis N, Garden OJ, et al. Meta-analysis of the value of
somatostatin and its analogues in reducing complications associated with
REFERENCES pancreatic surgery. Br J Surg. 2005;92:1059 –1067.
1. Griffanti Bartoli F, Arnone GB, Ravera G, et al. Pancreatic ﬁstula and relative 28. Vanounou T, Pratt WB, Callery MP, et al. Selective administration of
mortality in malignant disease after pancreaticoduodenectomy. Review and prophylactic octreotide during pancreaticoduodenectomy: a clinical and cost-
statistical meta-analysis regarding 15 years of literature. Anticancer Res. beneﬁt analysis in low- and high-risk glands. J Am Coll Surg. 2007;205:546 –
2. Buchler MW, Friess H, Wagner M, et al. Pancreatic ﬁstula after pancreatic
¨ 29. Shinchi H, Wada K, Traverso LW. The usefulness of drain data to identify a
head resection. Br J Surg. 2000;87:883– 889. clinically relevant pancreatic anastomotic leak after pancreaticoduodenec-
3. Rosso E, Bachellier P, Oussoultzoglou E, et al. Toward zero pancreatic ﬁstula tomy. J Gastrointest Surg. 2006;10:490 – 498.
after pancreaticoduodenectomy with pancreaticogastrostomy. Am J Surg. 30. Reid-Lombardo KM, Farnell MB, Crippa S, et al. Pancreatic anastomotic
2006;191:726 –732. leakage after pancreaticoduodenectomy in 1,507 patients: a report from the
4. Alexakis N, Sutton R, Neoptolemos JP. Surgical treatment of pancreatic pancreatic anastomotic leak study group. J Gastrointest Surg. 2007;11:1451–
ﬁstula. Dig Surg. 2004;21:262–274. 1459.
5. Bassi C, Dervenis C, Butturini G, et al. Postoperative pancreatic ﬁstula: an 31. Sohn TA, Yeo CJ, Cameron JL, et al. Pancreaticoduodenectomy: role of
international study group (ISGPF) deﬁnition. Surgery. 2005;138:8–13. interventional radiologists in managing patients and complications. J Gas-
6. Balzano G, Zerbi A, Cristallo M, et al. The unsolved problem of ﬁstula after trointest Surg. 2003;7:209 –219.
left pancreatectomy: the beneﬁt of cautious drain management. J Gastrointest 32. Roder JD, Stein HJ, Bottcher KA, et al. Stented versus nonstented pancre-
Surg. 2005;9:837– 842. aticojejunostomy after pancreatoduodenectomy. A prospective study. Ann
7. Knaebel HP, Diener MK, Wente MN, et al. Systematic review and meta- Surg. 1999;229:41– 48.
analysis of technique for closure of the pancreatic remnant after distal 33. Imaizumi T, Harada N, Fukuda A, et al. Stenting is unnecessary in duct-to-
pancreatectomy. Br J Surg. 2005;92:539 –546. mucosa pancreaticojejunostomy even in the normal pancreas. Pancreatology.
8. Kuroki T, Tajima Y, Kanematsu T. Surgical management for the prevention 2002;2:116 –121.
of pancreatic ﬁstula following distal pancreatectomy. J Hepatobiliary Pan- 34. D’Andrea AA, Costantino V, Sperti C, et al. Human ﬁbrin sealant in
creat Surg. 2005;12:283–285. pancreatic surgery: is it useful in preventing ﬁstulas? A prospective random-
9. Roggin KK, Rudloff U, Blumgart LH, et al. Central pancreatectomy revisited. ized study. Ital J Gastroenterol. 1994;26:283–286.
J Gastrointest Surg. 2006;10:804 – 812. 35. Bottger TC, Junginger T. Factors inﬂuencing morbidity and mortalita after`
10. Pyke CM, van Heerden JA, Colby TV, et al. The spectrum of serous pancreaticoduodenectomy: critical analysis of 221 resections. World J Surg.
cystadenoma of the pancreas. Clinical, pathological and surgical aspects. Ann 1999;23:164 –172.
Surg. 1992;215:132–139. 36. Bassi C, Falconi M, Salvia R, et al. Management of complications after
11. Talamini MA, Moesinger R, Yeo CJ, et al. Cystadenoma of the pancreas. Is pancreaticoduodenectomy in a high volume centre: results of 150 consecutive
enucleation an adequate operation? Ann Surg. 1998;227:896 –903. patients. Dig Surg. 2001;18:453– 458.
12. Kiely JM, Nakeeb A, Komorowski RA, et al. Cystic pancreatic neoplasms: 37. Adam U, Makowiec F, Riediger H, et al. Pancreatic leakage after pancreas
enucleate or resect? J Gastrointest Surg. 2003;7:890 – 897. resection. An analysis of 345 operated patients. Chirurg. 2002;73:466 – 473.
13. Lin JW, Eng M, Cameron JL, et al. Risk factors and outcomes in postpan- 38. Sarr MG for the Pancreatic surgery Group. The potent somatostatin analogue
creaticoduodenectomy pancreaticocutaneous ﬁstula. J Gastrointest Surg. vapreotide does not decrease pancreas speciﬁc complications after elective
2004;8:951–959. pancreatectomy. A prospective multicenter, double-blinded, randomized, pla-
cebo-controlled trial. J Am Coll Surg. 2003;196:556 –565.
14. de Castro SM, Busch OR, Van Gulik TM, et al. Incidence and management
of pancreatic leakage after pancreaticoduodenectomy. Br J Surg. 2005;92: 39. Gueroult S, Parc Y, Duron F, et al. Completion pancreatectomy for
1117–1123. postoperative peritonitis after pancreaticoduodenectomy. Arch Surg.
15. Kazanjian KK, Hines OJ, Eibl G, et al. Management of pancreatic ﬁstulas
after pancreaticoduodenectomy. Results in 437 consecutive patients. Arch 40. Satoi S, Takai S, Matsui Y, et al. Less morbidity after pancreaticoduodenec-
Surg. 2005;140:849 – 855. tomy of patients with pancreatic cancer. Pancreas. 2006;33:45–52.
16. Muscari F, Suc B, Kirzin S, et al. Risk factors for mortality and intra- 41. Crippa S, Bassi C, Warshaw AL, et al. Middle pancreatectomy. Indications,
abdominal complications after pancreatoduodenectomy: multivariate analysis short- and long-term operative outcome. Ann Surg. 2007;246:69 –76.
in 300 patients. Surgery. 2005;139:591–598. 42. Kollmar O, Moussavian MR, Bolli M, et al. Pancreatojejunal leakage after
© 2009 Lippincott Williams & Wilkins 103
8. Pedrazzoli et al Annals of Surgery • Volume 249, Number 1, January 2009
pancreas head resection: anatomic and surgeon-related factors. J Gastrointest 47. Sacks BA, Vine HS, Bartek S, et al. Postoperative abscess drainage in
Surg. 2007;11:1699 –1703. patients with established sinus tracks or drains. Radiology. 1982;142:537–
43. Liang TB, Bai XL, Zheng SS. Pancreatic ﬁstula after Pancreaticoduodenec- 538.
tomy: diagnosed according to International Study Group Pancreatic Fistula 48. Kim YJ, Han JK, Lee JM, et al. Percutaneous drainage of postoperative
(ISGPF) deﬁnition. Pancreatology. 2007;7:325–331. abdominal abscess with limited accessibility: preexisting surgical drains as
44. Murakami Y, Uemura K, Hayasidani Y, et al. No mortality after 150 alternative access route. Radiology. 2006;239:591–598.
consecutive pancreatoduodenctomies with duct-to-mucosa pancreaticogastro- 49. Pellegrini CA, Heck CF, Raper S, et al. An analysis of the reduced morbidity
stomy. J Surg Oncol. 2008;97:205–209. and mortality rates after Pancreaticoduodenectomy. Arch Surg. 1989;124:
45. Muller MW, Friess H, Kleeff J, et al. Middle segmental pancreatic resection: An
¨ 778 –781.
option to treat benign pancreatic body lesions. Ann Surg. 2006;244:909–920. 50. Pedrazzoli S, Sperti C, Pasquali C. Pancreaticoduodenojejunostomy for
46. Berberat PO, Ingold H, Gulbinas A, et al. Fast track– different implications in chronic pancreatitis presenting with an inﬂammatory mass in the head of the
pancreatic surgery. J Gastrointest Surg. 2007;11:880 – 887. pancreas. Pancreas. 1995;11:289 –293.
104 © 2009 Lippincott Williams & Wilkins