Sei sulla pagina 1di 8

Chapter 25

G. Tsiotos J. Tsiaoussis

Pancreatic fistulas (PFs) and intestinal fistulas (IFs)


are troublesome, occasionally significant, and not uncommon sequelae of necrotizing pancreatitis (NP).
They account for increased morbidity and sometimes
mortality, and prolonged hospital stay, and they are
costly both financially and with regard to resources.
Incidence varies between 5% and over 50% among
published series, but there is a definite decreasing
trend recently across the literature. The wide variation in incidence reflects not only different levels of
expertise among the authors, but also the striking
lack of a universally accepted and applied definition
of PF specifically in the context of NP. Although recently a consensus definition and staging of postoperative PF was published [1], there has been no similar
unifying attempt in the setting of NP, as is our topic.
The decreased incidence of PF/IF in the recent literature, in addition to improved surgical expertise, certainly reflects in part the recent change in the overall
management strategy of NP, as will be discussed below.
Higher imaging precision has led to more accurate
diagnoses by the delineation of fine, but crucial anatomic details of both PF and IF. In addition, advanced
technology and refined operative and interventional
or minimally invasive techniques have contributed to
an improved outcome in these patients.
In this chapter we will discuss the pathogenesis
and management of PF and IF separately, but prior to
this, it is essential to briefly outline a very significant
change in the management scheme of NP that has
taken place during the last decade, which has crucial
implications in both the incidence and the treatment
of PF and IF.

Pancreatic
and Intestinal Fistulas

Modern Management of NP
and its Implications
Since it has been recognized that the early peak of
mortality in the biphasic mortality pattern of NP is
due to the overwhelming systemic inflammatory response syndrome (SIRS; not sepsis), whereas the later
second peak is due to sepsis, two major components
of modern management have emerged: (1) very aggressive hemodynamic, ventilatory, metabolic, and
nutritional support and avoiding operative treatment
in the early phase, and (2) delayed operative treatment
(where necessary) for as long as possible. This approach, which has been substantiated by cornerstone
clinical studies [2,3] and is now the preferred management strategy in patients with NP [4], has led to
optimized hemodynamics early after NP, much fewer
reoperations for debridement (usually just one), essentially no gauze packing, and placement of fewer
drains. As will be discussed in detail below, these factors have substantially decreased the incidence of NP/
IF.

Pathogenesis
Although the pathogenesis of PF/IF is multifactorial,
the most common factor in their development is the
presence of pancreatic parenchymal necrosis, as this
results in the disruption of small or large pancreatic
ducts with subsequent extravasation of exocrine secretions into the retroperitoneum [5]. Operative necrosectomy and local drainage allow for external
egress of these extravasated secretions and the potential for a pancreaticocutaneous fistula. The concurrent pancreatic and peripancreatic inflammatory
process may also lead to stenosis of the pancreatic
ducts, which represents a substantial element for the
chronicity of fistulas. The importance of pancreatic
parenchymal necrosis as the main risk factor in the
pathogenesis of PF is stressed by the finding in one
study that all patients who developed pancreaticocu-

281

282

Pancreatic and Intestinal Fistulas

taneous fistulas had proven pancreatic parenchymal


necrosis, while none of the patients with peripancreatic retroperitoneal fat necrosis with an inflamed but
viable pancreas developed a PF [6].

Compromised Blood Supply


Compromised blood supply (in the form of vascular
thrombosis) to the colon and the duodenum has been
postulated as a pathogenetic mechanism for the formation of gastrointestinal fistulas in particular. Enzyme-rich fluids and inflammatory products released
or produced early in the course of the necrotizing
process can dissect throughout the retroperitoneal
tissues and into the transverse mesocolon to involve
the vascular supply to the colon or duodenum, with
consequent vascular thrombosis. The subsequent
ischemia (if extensive enough) may lead to segmental
colonic or duodenal necrosis and eventually the formation of a gastrointestinal fistula. Colonic ischemia
may occur as a result of a low-flow state [7] caused by
inadequate initial resuscitation or as a consequence of
the hemodynamic response to the sepsis syndrome.
This speculation would explain the tendency for fistulas to arise from the left transverse colon and the
splenic flexure, where collateral flow is more compromised in low-flow states. However, the modern aggressive hemodynamic resuscitation in the early
phase of NP (see above) has minimized a low-flow
state as a cause of a fistula.

Autodigestion
Autodigestion of adjacent organs as a cause of IF in
the course of NP has more of a theoretical background.
According to this hypothesis, the extravasated exocrine secretions may result in transmural necrosis of
the stomach and small intestine in a way similar to
peripancreatic fat necrosis. This concept for the development of gastrointestinal tract fistulas seems much
less likely because, unlike fat, the stomach and small
intestine have a much better vascular supply and,
thus, associated protective mechanisms [6].

Choice of Operative Technique


The development of fistulas may be related to the
choice of operative technique, since repeated local
trauma to the surface of an organ, as might occur
with repeated open packing of the lesser sack or dur-

G. Tsiotos J. Tsiaoussis

ing planned relaparotomies, may lead to intestinal


wall erosions. To reduce this possibility, covering of
the exposed viscera and major blood vessels with a
form of nonadherent interface before applying the intra-abdominal gauze packing has been proposed [5].
In addition, recent studies indicate that necrosectomy
followed by closed packing or by closed continuous
lavage may lead to a lower incidence of PF/IF formation [8,9], as this approach requires fewer intra-abdominal interventions for the repeated removal of
pancreatic necrotic material.
It is interesting, however, to note that the discussions about individual techniques and the comparisons among them tend to become obsolete, since todays management strategy of patients with NP
consisting of aggressive nonoperative initial management followed by necrosectomy as late as possible (removal of well-demarcated necrotic tissue without
compromising viable viscera usually 1month after
the onset of NP) leads to a more accurate distinction
between viable and necrotic tissue and a more complete necrosectomy, with a lower risk of leaving nonviable infected debris behind. Thus, a much lower
number of relaparotomies is required, with a lower
incidence of adjacent organ injury and, as a result, a
lower frequency of gastrointestinal fistulas.

Choice of Operative Approach


to the Retroperitoneal Space
The choice of operative approach to the retroperitoneal space during necrosectomy could predispose to
fistula development as a result of adjacent organ injury. The lesser sac can be approached through the
transverse mesocolon, the gastrocolic omentum, or
the gastrohepatic omentum. Because of the inflammatory process, the stomach and transverse colon
may have been densely adherent to the inflammatory
mass. Consequently, accessing the lesser sac through
an avascular area of the mesocolon to the left of the
ligament of Treitz seems quicker and safer, avoiding
any inadvertent injury to the adjacent organs. Fernandez-del Castillo et al. recognized the fact that no
colonic fistula developed in their series and attributed
this to their preferred access via the mesocolon [3].

Pressure Necrosis
Another iatrogenic mechanism of fistula formation
may be from pressure necrosis of a segment of bowel
from an adjacent drain and such a mechanism could

Chapter 25

reflect the late development of certain colonic, smallbowel, or gastric fistulas. To avoid this, the peripancreatic drains should not be positioned directly on the
duodenum or ascending colon when placed from the
patients right side, or on the descending colon when
placed from the left side. The ideal placement of the
drains is behind the splenic flexure of the colon and
below the lower pole of the spleen, thereby exiting the
abdominal wall in the left anterior axillary line. Also,
large, hard, stiff sump drains should be avoided as an
additional means of prevention of gastrointestinal fistula.
Minimally invasive techniques (percutaneous or
endoscopic drainage) and minimally invasive surgery
(retroperitoneoscopic debridement) are alternatives
to open surgery in select cases, promising lower morbidity including lower incidence of PF [10]. These innovations have not been popularized yet and only a
small number of series (with highly selected patients)
have been published [11]. With the currently available
experience, laparoscopy-assisted necrosectomy might
be followed by a higher trend for significant injuries
to intra-abdominal viscera including a higher incidence of PF/IF, as has been reported (2060%) [9].

Pancreatic Fistulas
Definition and Incidence
The lack of a widely accepted definition of this complication in the setting of NP has contributed in part
to the major discrepancy in its reported incidence
among published series. Criteria such as amylase level
in the excreted fluid, daily output of the fistula, and
its duration vary among studies. Despite the varying
definitions, a PF can be conceptualized anatomically
as an abnormal communication between a pancreatic
duct (major or minor) and the skin (pancreaticocutaneous fistula or external fistula), or between the pancreatic duct and peritoneal or pleural cavity, or another hollow viscus (internal fistula). Because of its
clinical significance we will deal with external fistulas in this chapter. The presence of infected NP (versus sterile NP) seems to favor the development of PF,
as PF is far more common (up to 76%) in the former
group [6,12].

Diagnosis and Imaging


The diagnosis of the presence of a PF can be easily
made by measuring the amylase activity of the fluid

Pancreatic and Intestinal Fistulas

excreted through a drain tube; this is higher than


1000IU/dl and usually up to a few thousands of IU/dl.
Amylase activity levels of a few hundreds of IU/dl
generally do not reflect a PF. The diagnosis of a PF
should be followed by the precise delineation of its
anatomic details. Complete mapping of the PF and its
relationship to the pancreatic ductal system determines its prognosis and dictates the management options.
The two issues of paramount importance to be
studied and looked for by imaging are:
1. Communication of the PF with the main pancreatic
duct or one of its minor branches.
2. Integrity of the pancreatic duct downstream (i.e.,
between the ductal disruption and the sphincter of
Oddi). The possibilities are the following:
a. proximal stenosis (i.e., stenosis of the main
pancreatic duct between its disruption and the
sphincter of Oddi, in which case the PF is fed
primarily, but not exclusively, by the portion of
the pancreatic duct of the distal pancreas)
b. disconnected pancreatic duct (i.e., no communication between the PF and the proximal pancreatic duct, in which case the PF is exclusively fed
by the portion of the pancreatic duct of the distal pancreas. This distal portion of the pancreas
is then an isolated pancreatic segment draining
solely through the fistula)
c. normal pancreatic duct (i.e., the PF is fed by
a rather small disruption of the pancreatic duct,
but its proximal and distal portions are in continuity and there is no stenosis along the length of
the duct).
Imaging techniques that can be employed in order to
extract this fine information include: (1) contrast sinogram, (2) contrast-enhanced computed tomography (CECT), (3) magnetic resonance cholangiopancreatography (MRCP), and (4) endoscopic retrograde
cholangiopancreatography (ERCP).
The water-soluble contrast sinogram is the first
imaging study that should be performed and it may
well be the only one required. It is very easy to perform, dynamic, low-cost, and noninvasive. It may reveal a communication of the PF with the main or a
side pancreatic duct. In cases of right or mid-body
pancreatic necrosis, the sinogram may show filling of
the distal pancreatic duct without opacification of the
proximal pancreatic duct or the duodenum. A spiral
CECT scan with thin cuts may also demonstrate the
presence and anatomy of a PF, but there is no comparative study that favors CECT scan over sinogram.

283

284

Pancreatic and Intestinal Fistulas

MRCP has been utilized increasingly to demonstrate


pancreatic ductal anatomy and certainly does have a
role in the imaging of PF as they relate to the ductal
system. It should be kept in mind, however, that although CECT and MRCP represent today the most
modern modalities of cross-sectional anatomy and
are readily available in most institutions, they may
not necessarily provide more information pertinent
to the precise anatomy of a PF compared to an expertly performed sinogram. Interestingly, it has been
shown, for example, that although MRCP is capable
of identifying major pancreatic ductal injuries, its reliability to discriminate subtle anomalies in anatomy
or to demonstrate a communication between the pancreatic duct and a pseudocyst is not high [13]. ERCP is
generally reserved when sinogram and cross-sectional modalities have not provided all of the necessary
anatomic information, or when an endoscopic therapeutic procedure is contemplated after sinogram has
precisely demonstrated the anatomy.

Management
PFs are quite often complicated by fluid and electrolyte abnormalities, malnutrition, skin erosion, and
less often hemorrhage and sepsis. The initial management of PF is conservative and its fundamental principles are the following:
1. Provision of optimal drainage to avoid intra-abdominal fluid collections.
2. Maintenance of fluid and electrolyte balance.
3. Treatment of local infection.
4. Optimization of nutritional status by parenteral or
preferably enteral feeding.
5. Skin care.
An additional appealing line of PF management is the
reduction of pancreatic secretion (and thus PF output) by possible administration of octreotide. The
role of octreotide in the prevention and the treatment
of PF has been studied in the postpancreatectomy setting (for tumor or chronic pancreatitis), but not in the
NP setting. It is fair to note that five of nine prospective randomized studies demonstrate a favorable effect of octreotide over placebo, whereas the remaining four did not [14]. Again, all nine studies included
patients who had undergone an elective pancreatic
operation and not patients with PF in the context of
NP. Our experience with the use of octreotide in an
effort to accelerate the closure of inflammatory PF
has not been encouraging [6]. It seems unlikely that

G. Tsiotos J. Tsiaoussis

this issue will be definitively resolved in the near future; today, the use of octreotide in this setting is not
evidence-based and should be reserved for use only
within a clinical protocol.
Spontaneous closure of a PF should be the primary
therapeutic goal. Compared to postoperative PF, postNP PF tends to have less chances of spontaneous closure (53% vs 86%) and longer duration for those that
do selfresolve (22weeks vs 11weeks) [15]. Although
good nutritional status, optimal drainage, and absence of local and systemic infection certainly provide
a favorable background for spontaneous PF closure,
the single most important factor determining prognosis and dictating definitive management is the pattern of ductal disruption that has given rise to the
development of the PF. Failure of spontaneous closure
is generally due to anatomic factors such as downstream ductal obstruction and disconnected duct
syndrome (isolated pancreatic tail). For example, recognition of an intact pancreatic duct without a downstream ductal obstruction indicates a high possibility
of spontaneous closure. On the contrary, surgery is
necessary when the PF is associated with a leak from
the pancreatic duct that is not joined with the gastrointestinal tract. Indeed, no such PF closed after a
mean of 26weeks of aggressive medical therapy, and
all of these patients required surgical intervention
[15].
In general, when a PF persists for more than
2weeks and its daily output remains essentially unchanged over this period (i.e., without significant,
meaningful decrease), a sinogram should be performed, followed possibly by MRCP. The choice
among further treatment options depends on the specific findings:
Communication of the PF with a Minor (Side)
Pancreatic Duct
Nonoperative management is justified and the chances are that such a PF will eventually close, even after a
few months, provided that the duct drains with no
stenosis (stricture) toward the main pancreatic duct
and to the duodenum.
Communication of the PF with the Main
Pancreatic Duct that has a Proximal Stenosis
Spontaneous closure is unlikely because the pancreatic duct distal to the stricture is preferentially draining to the PF. These patients are ideal for endoscopic
bridging of the proximal and distal parts of the pancreatic duct (traversing the stricture, as well as the

Chapter 25

ductal disruption feeding the PF) and stent placement. Small-diameter (57mm) stents with variable
length across the site of the ductal stricture and disruption may be used. Success rates are high (75100%)
[16,17]. Suboptimal stent placement, stent migration,
or stent occlusion may require repetition of the technique to achieve complete closure of the PF (up to five
stent placements in one patient have been reported for
PF resolution). Interestingly, even when the stent only
traverses the stricture, but does not bridge the site of
ductal leakage, the outcome may be successful [18].
The duration of endoscopic therapy to close the PF
varies from a few days to several weeks, and once it has
been rendered, stents are retrieved after 1014days.

Pancreatic and Intestinal Fistulas

of PF in the vast majority of patients with appropriately defined fistulas and pancreatic ductal anatomy.
However, in the minority in whom this approach fails,
an operation is mandatory. For patients with no disconnected pancreatic segment, a side-to-side pancreaticojejunostomy at the area of the origin of the fistulous tract is usually the optimal operative option.

Intestinal Fistulas
Definition and Incidence

As with PFs, the incidence of IFs varies widely among


studies (143%) [10]. However, in the case of Ifs, this
wide variation is not primarily due to variability of
definition, but rather reflects differences in the techCommunication of a PF with a Disconnected
nique of the initial necrosectomy and further opera Pancreatic Duct
Nonoperative therapy is doomed to failure and endo- tive debridements. It is especially with IFs where the
scopic stent placement is highly unlikely to be suc- changed management scheme of NP (i.e., delayed first
cessful since there is no communication between the necrosectomy with much lower number of subsequent
proximal and the distal portions of the pancreatic debridements required) has resulted in a recent sigduct; the distal isolated pancreatic segment is drain- nificant reduction in their incidence.
ing exclusively via the PF. Operative treatment should
IFs should be conceptualized anatomically in upbe planned after the surrounding inflammation has per-gut fistulas and colonic fistulas. This distinction
ceased and the general condition of the patient is im- is clinically relevant because the former have a generproved (usually it is after several weeks or even a few ally milder course and tend to close nonoperatively, as
months before operative treatment takes place). Sur- opposed to the latter, which can be associated with
gical options include distal pancreatectomy with or significant morbidity and often require operative
without splenectomy (realistically, the latter is techni- management. The pathogenesis of IF has been already
cally hard given the recent extensive retropancreatic discussed in detail earlier in this chapter.
inflammation and the fibrosis resulting from NP)
The diagnosis of the presence of IF is easily susand distal pancreaticojejunostomy using a defunc- pected, solely by the nature of the fluid coming out
tionalized Roux-en-Y loop. A more rarely utilized via the drain tube (or the incision). Low-viscosity, bilthird option is a fistulojejunostomy, provided that the ious fluid, higher-viscosity, green-brownish fluid,
fistulous tract has well matured. Operative manage- and obvious fecal material obviously reflect duodement, when indicated, is generally successful (>90%), nal/proximal jejunal, small intestinal, and colonic fisbut is associated with a not insignificant mortality tulas, respectively. Amylase activity level of the IF
(6%) [19]. Failures are due to inadequate resolution of fluid is high, but certainly far lower than the levels
the inflammatory process of the NP.
associated with PFs; it is usually a few hundreds of
IU/dl. When amylase activity level from an IF is in the
range of thousands of IU/dl, one should suspect a coCommunication of the PF with a Normal Main existing PF draining through the same fistulous
tract.
Pancreatic Duct
Although the main pancreatic duct has no strictures,
it does have a disruption, but its portions proximal
and distal to this disruption are in continuity. Enough
Imaging
time should be provided for spontaneous closure. If
this does not take place, endoscopic stent placement Water-soluble contrast infusion through the fistulous
should be performed and this is expected to be the tract under fluoroscopy (sinogram) is the first imagdefinitive therapy.
ing modality to perform. This will show the length
Minimally invasive endoscopic techniques for and width of the fistulous tract, it will light up the
stent placement today represent the definitive therapy hollow viscous where the tract originates from (and

285

286

Pancreatic and Intestinal Fistulas

thus define whether it is duodenum, proximal or distal small bowel, or colon), and will demonstrate the
presence or absence of intestinal stenosis or obstruction distal to the intestinal disruption and the source
of the fistula. A CECT scan should always be performed primarily to rule out an undrained fluid collection that often coexists at the time that the IF is
first diagnosed.

Management
As with PF, the principles of IF management are: provision of excellent drainage, optimization of fluid,
electrolyte, and nutritional status, treatment of local
infection, and skin care. In patients with upper-gut
fistulas, nutrition should be provided either parenterally or enterally distal to the site of the IF using a nasojejunal tube or a tube placed operatively.
A lot of the decision-making regarding further
management depends upon the source of the fistula
(upper gut versus colonic) and on the specific anatomic details of the IF:
1. Every undrained fluid collection seen in CECT
should be well drained by percutaneously placed
tubes under radiologic guidance. This will alleviate
systemic infection and also simplify the fistulous
tract by providing a direct communication between
the intestine and the skin without pooling of intestinal content into the surrounding tissues.
2. The presence of a luminal obstruction or stenosis
distal to the luminal disruption feeding the IF essentially guarantees failure of nonoperative management. On the contrary, absence of a distal stenosis
justifies nonoperative management for long time.
3. The pattern of the fistulous tract is of paramount
importance. Narrow and long fistulas are far more
likely to close spontaneously compared to the wide
and short ones. Prolonged nonoperative management is the preferred option for the former, whereas
it is not justified for the latter.
4. The role of IF output is important and twofold. A
high output (i.e., >200ml/day persistently) may be
first associated with hard-to-maintain electrolyte
balance, and second with distal luminal stenosis
and/or a wider and shorter fistulous tract. It is for
these reasons that a persistently high-output IF is
less likely to close spontaneously.
Provided that no absolute contraindication (i.e., distal
luminal obstruction) is present for nonoperative management, the prognosis of IF is closely related to the

G. Tsiotos J. Tsiaoussis

part of the gastrointestinal tract that the IF originates


from.
Duodenal Fistulas
Duodenal fistulas are very rare [20]. When present,
they originate from the medial aspect of the duodenum as a result of extension of the pancreatic inflammatory process and duodenal wall necrosis. If well
controlled, despite their potential for a high initial
output, most close spontaneously. In patients where
the duodenal fistula is diagnosed during the first or a
subsequent necrosectomy, it is reasonable to proceed
at that setting with pyloric exclusion (pyloric stapling
with a noncutting device and gastrojejunostomy), or
simple tube duodenostomy and excellent periduodenal drainage. Duodenal fistulas manifested and diagnosed prior to, or following necrosectomy should be
controlled by interventional radiology means (percutaneous drainage). In patients where a duodenal fistula persists (longer than 23months with essentially
unchanged daily output) despite optimal nonoperative management, a Roux-en-Y duodenojejunostomy
at the site of the duodenal wall defect is usually therapeutic. Needless to say, such a procedure should be
delayed enough in relation to the time of onset of NP
so that the inflammatory process has completely
ceased and a safe operation can be performed.
Fistulas of the Small Intestine
In general, enteric fistulas tend to close spontaneously
(more often than their duodenal counterparts). Plenty
of time should be allowed (months) before the surgeon decides to proceed with an operation, provided
(as previously emphasized) that the fistula and the intestine are well studied and there is no distal stenosis.
In patients with a persisting high-output enteric fistula, operative management consists of resection of
the fistulous tract, usually with the corresponding
bowel segment and an anastomosis between the proximal and the distal bowel. Resection of the fistulous
tract and oversewing of the bowel wall defect where
the fistula originated from, although appealing, may
be realistically more technically challenging and may
in fact compromise the diameter of the bowel at that
point.
Colonic Fistulas
Colonic fistulas are associated with more severe forms
of NP, and the mortality of NP among patients with a
concomitant colonic fistula is higher, highlighting

Chapter 25

the severity of this condition [21]. The portions of the


colon more frequently involved in colonic fistulas are
the transverse because of its close proximity to the
lesser sac and the splenic flexure because of its vascular pattern and the subsequent suboptimal blood supply in low-flow states. Colonic fistulas not only reflect
a more severe episode of NP, but also predispose patients to further comorbidity in and of themselves.
This is why it has been highly recommended to proceed with immediate fecal diversion [12] as soon as a
colonic fistula is diagnosed.
The increased morbidity due to a colonic fistula is
usually secondary to undrained intra-abdominal fecal collections around the colonic wall defect; this
may give rise to systemic sepsis. Sufficient and optimal percutaneous drainage cannot be always achieved
due to the high viscosity and particulate nature of the
fecal material. The combination of these factors usually dictate an urgent operation, during which proximal fecal diversion (loop ileostomy preferably, or loop
colostomy) should be the first priority, followed by excellent debridement and some type of colectomy (depending on the site of the colonic defect and the extent of colonic wall necrosis).
It is important to realize, however, that it is not the
mere presence of a colonic fistula that translates to
operative treatment, but rather the concomitant presence of undrained fecal material pooled around the
colonic necrosis and causing sepsis. If such conditions
are proved by CECT scan not to be present and the
patient is not septic, a colonic fistula may in fact be
conceptualized as a colostomy and no immediate action needs to be taken. In such a setting, where colonic fistulas are clinically asymptomatic, well controlled, have a low output, and there is no distal
colonic obstruction, they can be treated conservatively, and spontaneous closure may ensue.

Summary
PFs and IFs are notorious complications of NP, but
their incidence has decreased due to the recent change
of management scheme for NP consisting of aggressive initial hemodynamic support and delayed necrosectomy, which has led to a reduction in the number
of reoperations required. Precise delineations of the
fistulous tract in relation to the pancreatic ductal system and the pattern of pancreatic duct disruption dictate the prognosis and the appropriate management
option for a PF. Many PFs close spontaneously, some
require advanced endoscopic intrapancreatic procedures (ductal stent placement), and few require opera-

Pancreatic and Intestinal Fistulas

tive intervention. Most upper-gut fistulas close spontaneously provided that there is no distal intestinal
stenosis, whereas colonic fistulas are generally troublesome, may lead to sepsis, and require urgent operative treatment.

References
1. Bassi C, Dervenis C, Butturini G, Fingerhut A, et al (2005)
Postoperative pancreatic fistula: an international study group
(ISGPF) definition. Surgery 138:813
2. Mier J, Luque-de Leon E, Castillo A, et al (1997) Early versus
late necrosectomy in severe necrotizing pancreatitis. Am J
Surg 173:7175
3. Fernandez-Del Castillo C, Rattner DW, Makary MA, Mostafavi A, McGrath D, Warshaw AL (1998) Debridement and
closed packing for the treatment of necrotizing pancreatitis.
Ann Surg 228:676684
4. Uhl W, Warshaw A, Imrie C, Bassi C, et al (2002) IAP Guidelines for the surgical management of acute pancreatitis. Pancreatology 2:565573
5. Bradley EL III (1993) A clinically based classification system
for acute pancreatitis. Arch Surg 128:586590
6. Tsiotos GG, Smith CD, Sarr MG (1995) Incidence and management of pancreatic and enteric fistulas after surgical
management of severe necrotizing pancreatitis. Arch Surg
130:4852
7. Albridge MC, Francis ND, Glazer G, Dudley HAF (1989)
Colonic complications of severe acute pancreatitis. Br J Surg
76:362367
8. Buchler MW, Gloor B, Muller CA, et al (2000) Acute necrotizing pancreatitis: treatment strategy according to the status
of infection. Ann Surg 232:619626
9. Werner J, Feuerbach S, Uhl W, Buchler MW (2005) Management of acute pancreatitis: from surgery to interventional
intensive care. Gut 54:426436
10. Connor S, Alexakis N, Raraty MGT, et al (2005) Early and
late complications after pancreatic necrosectomy. Surgery
137:499505
11. Connor S, Ghaneh P, Raraty MGT, et al (2003) Minimally
invasive retroperitoneal pancreatic necrosectomy. Dig Surg
20:270277
12. Ho HS, Frey CF (1995) Gastrointestinal and pancreatic
complications associated with severe pancreatitis. Arch Surg
130:817823
13. Nealon WH, Walser E (2005) Surgical management of
complications associated with percutaneous and/or endoscopic management of pseudocyst of the pancreas. Ann Surg
241:948960
14. Alexakis N, Sutton R, Neoptolemos JP (2004) Surgical treatment of pancreatic fistula. Dig Surg 21:262274
15. Howard TJ, Stonerock CE, Sarkar J, et al (1998) Contemporary treatment strategies for external pancreatic fistulas.
Surgery 124:62733
16. Boerma D, Rauws E, van Gulik T, Huibregtse K, Obertop H,
Gouma D (2000) Endoscopic stent placement for pancreaticocutaneous fistula after surgical drainage of the pancreas.
Br J Surg 87:15061509
17. Costamagna G, Mutignani M, Ingrosso M, et al (2001) Endoscopic treatment of postsurgical external pancreatic fistulas.
Endoscopy 33:317322

287

288

Pancreatic and Intestinal Fistulas


18. Kozarek R, Ball T, Patterson D, et al (1997) Transpapillary
stenting for pancreaticocutaneous fistulas. J Gastrointest
Surg 1:357361
19. Voss M, Ali A, Eubanks E, Pappas T (2003) Surgical management of pancreaticocutaneous fistula. J Gastrointest Surg
7:542546

G. Tsiotos J. Tsiaoussis
20. Sakorafas GH, Tsiotos GG, Sarr MG (1999) Experience with
duodenal necrosis: a rare complication of acute necrotizing
pancreatitis. Int J Pancreatol 25:139142
21. Kriwanek S, Gschwantler M, Beckerhinn P, et al (1999) Complications after surgery for necrotizing pancreatitis: risk factors and prognosis. Eur J Surg 165:952957

Potrebbero piacerti anche