Keywords
pancreaticopleural fistula - pancreatitis - pleural effusion - endoscopic retrograde
cholangiopancreatography
Introduction
Pancreaticopleural fistula (PPF) is a relatively rare complication following pancreatic
duct disruption, characterized by an amylase-rich fluid accumulation in the pleural
space.[1] PPF is usually associated with acute/chronic pancreatitis (CP), trauma, or surgery.
The incidence of PPF is very low, occurring in approximately 0.4% of CP and around
1% of acute pancreatitis (AP).[2] An abnormal communication to the pleural space from posterior pancreatic duct disruption
or pancreatic pseudocyst extension into the pleural cavity is often identified. Diagnostic
dilemmas due to thoracic symptoms result in delayed diagnosis, as initial efforts
tend to be directed toward finding a thoracic pathology.[3] Minimally invasive endoscopic intervention is usually attempted before invasive
surgical management, by utilizing the endoscopic retrograde cholangiopancreatography
(ERCP) technique, with endoscopic sphincterotomy and main pancreatic duct (MPD) stenting
(passive transpapillary drainage) to ensure physiological outflow of pancreatic juice
into the duodenum. However, data on endoscopic management for this entity is scarce,
and evidence-based treatment algorithms are required. In this study, we present our
experience of endoscopic management of symptomatic PPF.
Materials and Methods
Patients with PPF were identified from departmental database between 2018 and 2022.
Their case records were reviewed for demographic details, clinical presentation, natural
history, progression of the disease, treatment strategies, and outcome.
Cases Definitions
Diagnosis of pancreatitis, clinical and morphological categorization, and definitions
of local and systemic complications were based on the 2012 revised Atlanta classification.
Patients with symptomatic pleural effusion for more than 3 weeks, fluid amylase levels
more than 1,000 U/L, underlying pancreatic disease, and no other causes of pleural
effusion were diagnosed with PPF.[4]
[5] Technical success of endoscopic therapy was defined as successful deep cannulation
of MPD and detection of leak. Clinical success was defined as clinical or radiological
improvement of pleural effusion after endotherapy. PPF without clinical signs or those
not associated with pancreatic inflammatory disease (acute or CP) were excluded from
the study.
Analyses of Cross-Sectional Imaging
Computed tomography (CT) or magnetic resonance cholangiopancreatography (MRCP) images
were reviewed in all patients (MRCP—6, contrast-enhanced computed tomography [CECT]–4)
to determine the site of ductal obstruction, extent of fluid collections (intra-abdominal
and/ or pleural), and presence of PPF.
Management
Conservative treatment of pancreatitis (nasogastric or nasojejunal feeding along with
intravenous fluid therapy and analgesia) was initially done in all the patients as
per international guidelines (Working Group International Association of Pancreatology
(IAP)/American Pancreatic Association (APA) Acute Pancreatitis Guidelines, 2013).
Catheter drainage of pleural fluid was done in symptomatic cases and percutaneous
or endoscopic ultrasound-guided (internal) drainage of intra-abdominal collections
was done when indicated. The decision to use interventional treatment for PPF was
based on symptomatology and cross-sectional imaging results (CECT and/or MRCP; [Fig. 1A–C]; [Fig. 2A–C]).
Fig. 1 Contrast-enhanced computed tomography images (A–C) showing the peripancreatic collection in the lesser sac extending toward the left
pleural cavity. The red arrow indicates the lesser sac collection and the yellow arrow
indicates the left-sided pleural effusion.
Fig. 2 Magnetic resonance imaging T2-weighted images (A–C) showing the peripancreatic collection in the lesser sac tracking toward the right
pleural cavity. The green arrow indicates the lesser sac collection and the orange
arrow indicates the right-sided pleural effusion.
ERCP was performed under conscious sedation using intravenous midazolam (0.1 mg/kg)
and ketamine (1 mg/kg) after obtaining informed consent. All procedures were done
using carbon dioxide insufflation with duodenoscope (TJF 180V, Olympus Corporation,
Tokyo, Japan). In all patients, transpapillary route was used to attempt documentation
of contrast-leak site, assess morphology and integrity of MPD, opacification of upstream
duct, and attempt to place stent in MPD to bridge the leak.
If MPD disruption was identified, pancreatic sphincterotomy (with sphincterotome,
Fusion OMNI Sphincterotome FS-OMNI-35–480, Cook Endoscopy Inc., North Carolina, United
States) was performed and a pancreatic plastic stent (5 Fr/7 Fr/10 Fr; Zimmon Pancreatic
Stent, Cook, Endoscopy Inc., North Carolina, United States) was placed to bridge the
leak ([Fig. 3A–C]). MPD disruption site (head, body, tail) and diameter were taken into consideration
for choosing stent length and diameter. In cases where leak was not detected, pancreatic
sphincterotomy and plastic stenting of downstream duct were done. Repeat ERCP was
done after 4 weeks of index procedure, to document the status of leak. Persisting
leaks were managed by stent replacement after 3, 6, 12, or 24 months or until no contrast
leakage was identified. In cases of CP, clearance of calculi and stricture dilatation
was also done. Peripancreatic fluid collections (if present) were managed with either
endoscopic ultrasound-guided or percutaneous drainage prior to ERCP.
Fig. 3 Pancreatogram images (A–D) show a contrast leak from the distal body near the site of transgastric Percutaneous
drainage (PCD); pancreatic duct stent was placed bridging the leak site.
Follow-Up
Patients were followed up with symptom and signs analysis and serial ultrasonography
for any recurrence of ascites/pleural effusion after removal of the MPD stent.
Statistical Analysis
Descriptive statistics were used for presentation, clinical features, and interventions.
Categorical data are analyzed as frequencies. Medians and ranges were used to analyze
nonparametric continuous variables.
Results
A total of 842 (502 males) patients with pancreatitis were treated in our department
between 2018 and 2023. Postinflammatory PPF was diagnosed in 10/842 (1.2%) patients
(mean age 33.6 ± 15.4 years, 6 males). The etiology of pancreatitis in the study group
was alcohol-related in six and idiopathic in four patients. The mean duration of illness
was 16.5 ± 21.01 weeks, while respiratory symptoms were present for 5.1 ± 2.9 weeks.
The commonest presenting symptoms were abdominal pain (10/10), pleuritic pain (8/10),
and dyspnea (6/10). Four patients had a history of intercostal drainage due to respiratory
discomfort, while two required recurrent therapeutic pleural taps prior to admission.
Fever was observed in three patients ([Table 1]).
Table 1
Demographics and clinical presentation
|
No. of patients (n = 10)
|
Percentage (%)
|
Age (mean ± SD) years
|
33.6 ± 15.4 years
|
|
Etiology
|
|
Alcoholic
|
6
|
60
|
Idiopathic
|
4
|
40
|
Duration of illness (mean ± SD)
|
16.5 ± 21.01 months
|
|
Duration of respiratory symptoms (mean ± SD)
|
5.1 ± 2.9 weeks
|
|
Clinical presentation
|
|
Abdominal pain
|
10
|
100
|
Pleuritic pain
|
8
|
80
|
Dyspnea
|
6
|
60
|
Severe dyspnea (requiring ICD)
|
4
|
40
|
Fever
|
3
|
30
|
Abdominal distension
|
2
|
20
|
Recurrent therapeutic pleural tap (at least 2 times/week)
|
2
|
20
|
Abbreviations: ICD, intercostal drainage; SD, standard deviation.
Majority of patients had left-sided pleural effusion (8/10), while 2/10 had only right-sided
pleural effusion. Six patients had CECT features suggestive of CP. Among the CP patients,
three had downstream calculi and two had downstream strictures. All patients had pleural
fluid amylase levels more than 2,000 IU/mL ([Table 2]). Endoscopic ultrasound-guided transmural drainage was done in one patient prior
to transpapillary drainage, as patient had gastric outlet obstruction. Pancreatic
duct leak was documented in 7/10 patients (70%); site was genu in 57.1% cases, tail
28.5%, and body 14.2%. Among 3 patients, where leak was not demonstrated, one patient
had resolution of leak, while remaining two had resolution after successful treatment
of downstream stricture. Pancreatic duct leak was bridged in three patients (42.8%).
Pancreatic sphincterotomy and MPD stenting were done in all patients. Technical success
was achieved in 70%. Successful resolution of pleural effusion (clinical success)
was obtained in all the patients ([Table 2]). The mean number of ERCP was 4.1 (range: 2–12). Two patients developed post-ERCP
pancreatitis, which were mild and improved on conservative management. There was no
difference in course and outcome in terms of site of ductal disruption and time of
resolution of leaks between AP and CP patients.
Table 2
Imaging findings and endotherapy results
CT findings
|
No. of patients
|
Percentage (%)
|
Pleural effusion (right/left)
|
2/8
|
|
Features of CP
|
6
|
60
|
MPD stricture/calculi
|
2/3
|
|
Pleural fluid amylase (>2,000)
|
10
|
100
|
ERCP details
|
Endotherapy done
|
10
|
100
|
Leak detected
|
7
|
70
|
Site of leak (identified in 7 patients)
|
|
|
AP
|
2/3
|
66.6
|
CP
|
5/6
|
83.3
|
-Body (CP-1)
|
1
|
14.2
|
-Genu (AP-2, CP-2))
|
4
|
57.1
|
-Tail (CP-2)
|
2
|
28.5
|
Leak bridged
|
3
|
42.8
|
AP
|
1/3
|
33.3
|
CP
|
2/6
|
33.3
|
Sphincterotomy performed
|
10
|
100
|
PD stent placed
|
10
|
100
|
Technical success (detection of leak)
|
7
|
70
|
Clinical success (clinical improvement/ resolution of effusion)
|
10
|
100
|
Abbreviations: AP, acute pancreatitis; CP, chronic pancreatitis; CT, computed tomography;
ERCP, endoscopic retrograde cholangiopancreatography; MPD, main pancreatic duct.
Mean duration of endotherapy was 12.1 ± 9.4 months (10 ± 1.4 months in AP, 13.1 ± 11.4
months in CP) and mean duration for documented leak closure was 15.3 ± 10.4 weeks.
Strictures (2, 22.2%) resolved on multiple stent therapy ([Table 3]). Surgical intervention was required in one patient (despite ERCP documentation
of leak and successful MPD stenting) after resolution of PPF as patient developed
groove pancreatitis requiring laparotomy, gastrotomy, drainage of retro gastric phlegmon,
and loop gastrojejunostomy.
Table 3
Follow-up of Patients after Endotherapy
Follow-up and outcome (of 10 patients)
|
Duration
|
Median follow-up
|
39.3 ± 13.4 months
|
Chronic pancreatitis
|
13.1 ± 11.4 months
|
Acute pancreatitis
|
10 ± 1.4 months
|
Mean duration of endotherapy
|
12.1 ± 9.4 months
|
Mean duration for leak closure
|
15.3 ± 10.4 weeks
|
Association with local collection
|
10/10
|
Surgical intervention
|
1/10 (for GOO)
|
Recurrence rate
|
Nil
|
Mortality
|
Nil
|
Abbreviation: GOO, Gastric outlet obstruction
Long-term success of treatment with a median follow-up of 39.3 ± 13.4 months was seen
in all 10 cases.
Discussion
The current literature lacks clear guidelines defining an algorithm for performing
diagnostic and therapeutic procedures in patients with PPFs. Most of the available
data are in the form of individual case reports or case series. PPF represents both
a diagnostic and a therapeutic challenge. This is an uncommon complication associated
with AP, CP, and trauma to the pancreas.[6] The incidence of PPF is extremely low, occurring in approximately 0.4% of CP patients,
around 1% in AP. In our study, the incidence rate of PPF in patients with pancreatitis
was 1.2%. However, this may not reflect the true incidence, as our facility is a tertiary
referral center. Disruption of MPD occurs in over 80% patients with postinflammatory
pancreatic and peripancreatic fluid (PPF) collections during acute or CP. Typically,
PPF occurs following MPD disruption, when the pancreatic duct opens into the pleura,
or when a pseudocyst forms and communicates with the pleural cavity. The pancreatic
fluid, rich in proteolytic enzymes, disrupts fascial planes and flows through the
retroperitoneum, usually through the esophageal hiatus, into the pleural cavity. Occasionally
transdiaphragmatic communication may also be the route of fluid movement.[7]
[8] Internal pancreatic fistula forms following MPD disruption, and ERCP is considered
the gold standard method for diagnosing PD disruption, defined as extravasation of
contrast medium from the pancreatic ductal system.[4]
[9]
[10] On pancreatogram, partial disruption is recognized when PD opacification is seen
upstream to the point of disruption, or complete disruption when no PD can be seen
upstream to the point of disruption.[9]
[11]
[12]
There are no typical clinical features of postinflammatory PPF, making its diagnosis
difficult. A patient with history compatible with pancreatitis along with demonstration
of pancreatic ductal disruption with pleural effusion on imaging, and pleural exudate
showing high amylase levels is considered diagnostic of PPF.[4]
[5]
[7]
[8] Majority of patients present with dyspnea (65–76%) followed by abdominal pain, cough,
chest pain, and fever.[3]
[4] Predominance of chest symptoms may lead to delay in diagnosis and treatment. According
to Uchiyama et al, 68% of PPF present with dyspnea, abdominal pain, cough, and chest
pain.[13] In our study, chest symptoms (dyspnea, pleuritic pain) were present in 80% cases.
Abdominal pain was observed in all the cases.
After documentation of pleural effusion on chest radiography, thoracocentesis is performed,
and elevated levels of amylase (> 2000 IU/mL) in the pleural fluid are considered
diagnostic of PPF.[4]
[14]
[15]
[16] PPF usually results in left-sided pleural effusion; however, right-sided and bilateral
effusions can occur in 19 and 14% of patients, respectively. In our study, 80% patients
had left-sided pleural effusion. Cross-sectional imaging (CECT/MRCP) identifies PPF
in approximately 70 to 80% of cases.[4]
[13]
[16]
[17]
[18]
Initial treatment includes conservative measures, that is, nil per oral, parenteral
nutrition, thoracocentesis, and octreotide infusion (used to decrease pancreatic fistula
output and closing time.[19]
[20]
[21] Even though conservative treatment can resolve the condition in 20 to 50% of cases,
prolonged treatment time, infection risk, and long duration hospitalization are limiting
factors.[15]
[19]
A recent addition in the management of PPF is endotherapy with either transpapillary
nasopancreatic drainage or MPD stenting.[5]
[10]
[22]
[23] Goals of therapy are to restore normal anatomy and attempt to close the leak. Pancreatic
sphincterotomy along with MPD stenting to bridge the disruption is determinant of
successful outcome when partial PD disruption is present.[12] ERCP confirms the diagnosis of PPF in 80% of cases and reveals a fistulous pathway
in approximately 59%.[12]
[24]
[25]
In the largest study on endotherapy of postinflammatory PPFs, 22 patients were treated
endoscopically,[26]
[27] and technical success was achieved in all cases. Clinical success was achieved in
21 (95.45%), and long-term success of endoscopic treatment was noted in 19 (86.36%)
patients. In our study, technical success (successful deep cannulation of MPD and
detection of leak) was 70%, and clinical and long-term success was achieved in all
(100%) patients. Though leak was bridged in only three patients, pancreatic sphincterotomy
and downstream stenting (when bridging was not possible) were successful in closing
the PPF. This is possibly due to reduction in downstream pressure gradient, facilitating
flow of pancreatic juice into duodenum. Although endotherapy is more effective in
cases of partial PD disruption (compared with total disruption), our study demonstrated
that pancreatic sphincterotomy with MPD stenting is beneficial even in cases of complete
PD disruption. However, it is important to highlight that patients with complete MPD
disruption often require transmural drainage, in addition to passive transpapillary
drainage, especially if there is pancreatic fragmentation (disconnected duct syndrome).
The strength of our study is the step-wise algorithm used for managing PPF patients,
which highlights the role of minimally invasive endoscopic therapy as a useful therapeutic
strategy after failure of medical therapy. Moreover, we have a long follow-up demonstrating
long-term efficacy of endoscopic treatment. We propose a step-wise algorithm for endoscopic
management of PPFs based on our experience ([Fig. 4]).
Fig. 4 Stepwise algorithm describing the approach to patients with pancreaticopleural fistula.
The main limitations are the relatively small number of patients, lack of randomization,
and single-center experience. Further studies with larger number of patients are needed
to formulate guidelines for the management of PPF.
Conclusion
PPFs create a diagnostic dilemma, as their symptoms mimic thoracic emergencies. Patients
with a history of pancreatitis or abdominal trauma with chest symptoms and pleural
effusion require a high index of suspicion for PPF. Cross-sectional imaging is essential
as it identifies the anatomy of duct disruption and fistula track in most cases. Early
restoration of ductal continuity with MPD stenting is very effective, if conservative
management fails.