The inconsistent nature of symptomatic pancreatico-jejunostomy anastomotic strictures

Abstract

Background

Pancreatico-jejunostomy strictures (PJS) after pancreatiocoduodenectomy (PD) are poorly understood.

Methods

Patients treated for PJS were identified from all PDs (n =357) performed for all indications in our practice (2002 to 2009). Technical aspects of the original operation, as well as the presentation, management and outcomes of the resultant stricture were assessed.

Results

Seven patients developed a symptomatic PJS for an incidence of 2%. ‘Soft’ glands and small ducts (≤3 mm) were each present in 3/7 of the original anastomoses. Pancreatic fistula occurred in 6/7. The latency period to stricture presentation averaged 41 months. Diagnosis of PJS was confirmed by secretin magnetic resonance cholangio-pancreatography (MRCP). Therapeutic endoscopic retrograde cholangiopancreatography (ERCP) was attempted – each unsuccessfully – in four patients. All patients required operative correction of their PJS by takedown/revision of the original pancreatico-jejunal anastomoses (PJA) (n =4) ± a modified Puestow (n =2). One patient's PJS was completely inaccessible due to dense adhesions. Another patient's stricture recurred and was successfully revised with a stricturoplasty. At a mean follow-up of 25 months, all are alive, but only 4/7 are pain free.

Conclusion

A symptomatic PJS appears to be independent of original pathological, glandular or technical features but pancreatic fistulae may contribute. Secretin MRCP is diagnostically useful, whereas ERCP has been proven to be therapeutically ineffective. Durable resolution of symptoms after surgical revision is unpredictable.

Introduction

Pancreaticoduodenectomy (PD) is now commonly performed for a full array of indications ranging from benign conditions such as chronic pancreatitis, to pre-malignant lesions like intraductal papillary mucinous neoplasm (IPMN), to full-fledged malignancy.¹ Although in-hospital mortality is now just a fraction of what it once was, down to under 5% in specialist's hands, operative complications are still common. These occur in approximately 30–50% of patients,^2–4 and at different temporal points in the recovery period. Significant immediate-term adverse outcomes include clinically relevant pancreatic fistula (CR-POPF; the most common problem – seen in approximately 15% of patients),^5,6 delayed gastric emptying (DGE), haemorrhage and various infections.² Intermediate-term ramifications include both endocrine and exocrine insufficiency, which occur unpredictably (around 25% of the time each) and have significant physiological consequences.^7,8

While this morbidity profile has been extensively recognized, described and studied, less is understood regarding longer-term sequelae, including structural complications such as anastomotic strictures. This problem is particularly germane given the expanded indications for, and improving survival after, PD. Recent literature estimates the incidence of bilioenteric anastomotic strictures to be between 2.5% and 8%.^9,10 Yet, little has been published regarding the long-term patency and natural history of pancreatico-jejunal anastomoses (PJA). In a limited review of PD performed for benign disease only, one study reports a pancreatico-jejunostomy stricture (PJS) incidence of 5%,¹⁰ whereas another, more-extensive investigation claims a 11% rate in patients exclusively limited to a diagnosis of chronic pancreatitis.¹¹ While other surgically-induced strictures (bilioenteric, entero-enteric) can be addressed through minimally-invasive endoscopic or radiographic means, management of a PJS is challenging in that the altered anatomy of the post-PD reconstruction often precludes successful employment of such techniques.

As so many PJAs (∼60%) are constructed in the setting of malignancy with its attendant short-term lifespan (roughly 2 years), the natural history of a PJA over time is not well understood. Herein, we describe an experience with the presentation, diagnosis, management and outcomes of post-PD PJS performed for the full spectrum of pathological indications.

Methods

This series was accrued over an 8-year period (2002–09) at a multidisciplinary, tertiary-care medical centre with a high-volume practice in pancreato-biliary surgery. Pancreaticoduodenectomy was performed by two fellowship-trained pancreatic surgical specialists (M.P.C. and C.M.V.) for a full range of benign, premalignant and malignant indications. Two patients' index PD operations were performed elsewhere by other surgeons antecedent to 2002. Pancratico-jejunostomy (PJ) was always performed (including the two operations performed by others) as the form of enteric reconstruction in all patients with the exception of four pancreatico-gastrostomies (PG) which were excluded from analysis. PJA in the setting of central pancreatectomy was also excluded because of this limited cohort (<10 patients).

The PJA was always performed in an end-to-side, duct-to-mucosa fashion in one (31%) or two (69%) layers. The inner ductal anastomosis was constructed with interrupted 5-0, 6-0 or 7-0 absorbable sutures placed in a radial fashion, whereas the outer layer approximated the pancreatic capsule to the serosa of the jejunum with interrupted 3-0 silk sutures. A PJ anastomotic stent (usually a short and internalized, but sometimes a long and externalized, 5-F paediatric feeding tube) was infrequently placed in situations perceived at high risk for the development of a pancreatic fistula. Prophylactic octreotide was administered in selected cases with recognized risk factors,^12,13 and fibrin glue was rarely applied (fewer than five patients).

Specific patient characteristics and results were accrued from an Institutional Review Board (IRB)-approved prospective database. Patients who developed a PJS that required revision were identified from all PDs performed in that time period. Previously developed risk factors for CR-POPF were analysed with the supposition that fistula development may eventually lead to stricture formation.^6,11 Data were not sought on those asymptomatic patients with sub-clinical (e.g. ductal dilation, glandular atrophy) or clinical (diabetes) evidence of PJ stenosis. Technical aspects of the original operation, and the subsequent presentation and management of the resultant stricture, were assessed.

Investigation of symptomatic abdominal pain included biochemical analysis for evidence of pancreatitis, as well axial imaging, including both computed tomography and magnetic resonance cholangio-pancreatography (MRCP) with secretin administration. Operative interventions were undertaken in those patients with debilitating abdominal pain and evidence of a stenotic anastomosis from secretin MRCP analysis who failed less-invasive treatment modalities (enzyme supplementation, narcotics and endoscopic interventions). Short- and long-term outcomes of revision operations were analysed. Post-repair follow-up was performed via longitudinal visits in the surgeon's clinic with emphasis on recurrence of abdominal pain (and its nature), as well as the new-onset or acceleration of exocrine insufficiency or diabetes. Exocrine insufficiency was measured by the presence of steatorhea, persistent weight-loss or the need for additional enzymatic supplementation. Faecal elastase measurement has not been employed.

Diagnosis of PJS was secured in all cases through the use of MRCP with secretin in the setting of clinically-relevant pain. MR examinations were performed in either a 1.5T (GE Twin or HDX, GE Healthcare; Vision, Symphony, or Espree, Siemens) or 3T (GE HDX) clinical scanner, using a phased-array coil, with the patients in the supine position. The secretin MRCP (sMRCP) protocol includes standard T1-weighted and T2-weighted images. A baseline thick-slab MRCP image was obtained followed by the intravenous injection over 1 min of secretin at a weight-dependent dose (0.2 mcg/kg). MRCP thick-slab images are then obtained every 30 s for 10 min. Finally, three-dimensional dynamic gradient-echo images are acquired before and after the intravenous administration of a bolus of gadolinium (0.1 mmol/kg) during the arterial, portal and delayed venous phase.

MR examinations were interpreted by one of six MRI fellowship-trained radiologists in direct conjunction with the referring surgeon. The imaging findings for both the routine MRCP and the Secretin-enhanced MRCP examinations were assessed. Anatomic variants (i.e. pancreas divisum, dominant dorsal duct and aberrant ductal communications) and ductal pathology including filling defects, stenosis or obstruction, on standard and secretin-enhanced MRCP images was recorded. Similarly, non-ductal pancreatic pathology including the presence of solid/cystic masses, atrophy or extrapancreatic disease was discerned. Symptoms associated with secretin administration were also elicited. PJS was defined by the presence of a fixed filling defect at the anastomotic site, along with post-obstructive ductal dilation, side-branch enhancement and/or decreased functional excretion into the jejunal drainage limb.

Results

Patients

Original operation

Over the 8-year study period, 357 PDs were performed in this specialty practice for the following diagnoses: pancreatic adenocarcinoma (n =146; 41%), other periampullary malignancies (n =64; 18%), chronic pancreatitis (n =50; 14%), cystic lesions (n =43; 12%), benign lesions (n =29; 8%), neuroendocrine tumours (n =11; 3%), metastatic lesions (n =7; 2%) and other indications (n =7; 2%). Seven patients, with a wide variety of original diagnoses, eventually presented with a PJS, which manifest with symptomatic abdominal pain, for an overall rate of 2% (Table 1). Excluding the two patients originally resected by other surgeons, the incidence of symptomatic stricture formation in our practice was 1.4% (5/357). Among the seven patients who went on to develop PJS, four were noted at the time of original surgery to have ‘firm’ glands, while three were noted to have ‘soft’ glands. Additionally, the pancreatic duct was small (≤3 mm) in just 3/7 patients. The original PJA was constructed over a short, internalized stent in 16% of the overall series and in 2/7 of patients with an ultimate PJS. A CR-POPF occurred in 6/7 of the original operations.

Table 1.

Presentation and management of pancreatio-jejunal strictures after pancreaticoduodenectomy

Patient	Original diagnosis	Stricture presented	Presenting symptoms	Operative intervention	Follow-up	Outcome
1	Serous cystadenoma	120 months	Episodic epigastric and back pain; steatorrhoea	Revision of PJA	40 mo	Persistent abdominal pain
2	Pancreatic NE tumour	16 months	Postprandial pain; steatorrhoea; weight loss	Revision of PJA	35 mo	Symptom free
3	Chronic pancreatitis	8 months	Epigastric pain; nausea; biochemical pancreatitis	Revision of PJA	57 mo	Persistent pancreatitis
4	Ampullary adenocarcinoma	34 months	Epigastric and back pain; weight loss biochemical pancreatitis;	Revision of PJA – modified Puestow	14 mo	Symptom free
5	Chronic pancreatitis	62 months	Postprandial epigastric pain; biochemical pancreatitis	Revision of PJA – modified Puestow	18 mo	Symptom free
6	Gastro-intestinal stromal tumour	17 months	Episodic epigastric and back pain	Exploration w/ intra-operative ERCP – pancreas/stricture inaccessible	12 mo	Persistent episodic abdominal pain
7	Duodenal fibrosis	29 months	Episodic epigastric pain; biochemical pancreatitis	Revision of PJA	18 mo	Recurrent stricture, pancreatitis
7 (Same Patient)	Pancreatitis	18 months	Biochemical pancreatitis	Trans-jejunal stricturoplasty	17.5 mo	Symptom free

PJA, pancreatico-jejunal anastomoses; ERCP, endoscopic retrograde cholangiopancreatography.

Pancreaticojejunostomy stricture

The latency period from original resection to clinical presentation of PJS was a median time of 18 months with a single 10-year outlier (range: 8–120 months). All seven patients had abdominal pain typical of obstructive pancreatopathy, with either episodic or post-prandial epigastric abdominal pain. Five patients had biochemical evidence of pancreatitis (Table 1). Narcotic use was common (5/7). Strictures were definitively diagnosed using secretin MRCP in all patients, and minimally invasive therapeutic intervention via endoscopic retrograde cholangiopancreatography (ERCP) was attempted in four patients. However, each of these were unsuccessful for one of two reasons: (i) the PJA was inaccessible because of the altered anatomy of the pancreaticobiliary limb used for PD reconstruction, or (ii) if successfully identified, the PJS was unable to be cannulated owing to either the tight/complete degree of stenosis or there were technical challenges of ERCP with such unorthodox positioning.

All patients required operative intervention with takedown/revision of the original PJA (n =4), and/or a ‘modified Puestow’ procedure (n =2). Revised anastomoses were performed in one or two layers, after a short segment (∼1 cm) of remnant pancreas was resected – usually revealing a somewhat dilated pancreatic duct. For the ‘modified Puestow’ patients, the new anastomosis was extended longitudinally along the anterior pancreatic duct for a centimetre or two to assure wide patency. There was sufficient redundancy of the pancreaticobiliary jejunal limb such that it could be reused for the new anastomosis without requiring a newly-fashioned Roux-en-Y conduit. An anastomotic stent was used in four instances. Median estimated blood loss for these revisions was 175 mL. There were no mortalities, two post-operative complications (UTI, prolonged ileus) and a median LOS of 8 days (range: 5–14).

In one patient (no. 6, Table 1), the pancreas and PJA were completely inaccessible because of dense adhesions (17 months later). An intra-operative ERCP was attempted via the pancreaticobiliary drainage limb, but was again unsuccessful in identifying the stenosed PJ orifice. The proposed anastomotic revision was abandoned after a transverse colectomy was required for colonic ischaemia induced by the dissection and the post-operative recovery was uneventful. The patient continues to suffer from episodic abdominal pain.

Follow up

At a median follow-up time of 18 months (range: 12 to 57 months), all patients are alive, but only 4 out of 7 patients are completely pain free. None have evidence of declining exocrine (use of pancreatic enzyme supplementation and steattorhea) or endocrine (accelerated diabetes) function. However, two of these symptomatic patients have demonstrated overt biochemical pancreatitis. One of these two patients (no. 7, Table 1) had a PJS recur a second time 18 months after a primary repair, and was successfully managed with a stricturoplasty achieved through the anterior wall of the jejunal limb. An externalized stent was employed and interrogated radiographically 6 weeks later – showing no evidence of any stenosis or diminished flow across the anastomosis. The patient is now symptom free.

Discussion

Recovery after PD can be viewed in a continuum of three temporal phases: immediate, intermediate and long term. There is a vast body of literature on peri-operative mortality^14,15 as well as short-term complications^3,4,16 for PD. Foremost among these is the most feared and difficult to manage complication – pancreatic anastomotic failure.¹⁷ Other problems such as permanent endocrine and exocrine dysfunction usually manifest within the first month of recovery. Conversely, little has been described of the longer-term complications of PD, and specifically those relating to the biliary and pancreatic anastomoses. This is partly because of the fact that many patients with malignant diagnoses do not live long enough after PD for there to be adequate numbers to study. Furthermore, many patients may live remotely from their regional centres of specialization, and others can be lost to follow-up from their original surgeons.

However, with the increased safety and better outcomes of PD, the operation has found a wider range of applications beyond malignancies – including premalignant conditions such as IPMN, neuroendocrine tumors and gastrointestinal stromal tumours, as well as benign maladies such as chronic pancreatitis. As a result, there are more patients to study and follow with improved life spans, and thus the longer-term morbidity of PD is becoming evident. House et al. reported on biliary strictures in their large series of PD for all indications, whereas Reid-Lombardo et al. published similarly on biliary and pancreatic strictures in those who underwent PD for benign disease exclusively.^9,10 The latter series found four PJS for a rate of 3.3%, two of which were rectified surgically. While there have been other descriptions of PJS in the literature, they have been largely limited to single-case reports describing therapeutic techniques.^18–20 Most recently, Morgan et al. reported of a large cohort of pancreatic head resections performed for pancreatitis with a significant PJS rate (11%) over a 56-month follow-up.¹¹

This current series, evaluating PD performed for a variety of indications – both benign and malignant – expands on this knowledge base. We believe it is necessary to consider this problem in the differential diagnosis for all patients who present with chronic abdominal pain after a partial pancreatectomy with PJ reconstruction. The rate of PJS resulting in symptomatic pain in this series is low, under 2%, with a mean time to presentation with the complication of 41 months. This number is certainly skewed by one patient, but even with the exclusion of this data point, the mean is still close to 26 months. Occurrence of PJS appears to be independent of original pathological, glandular and technical features – but perhaps is most affected by the development of clinically-relevant post-operative pancreatic fistulae which induce aggressive local inflammation and an accentuated repair response marked by development of fibrosis. Morgan's group has suggested that an original diagnosis of chronic pancreatitis may similarly contribute to anastomotic stenosis. That analysis, where PJA revisions were also performed for intractable pain, came to similar conclusions in that specific population as we have with broader indications for the original pancreatic resection presented herein. That is, long-term pain relief following anastomotic revision is dubious.¹¹ Our rate of PJS in patients suffering from chronic pancreatitis was not as impressive (4%, 2/49). However, another patient (no. 7, Table 1) suffered chronic pancreatitis of his pancreatic remnant as a result of his first PJS, and this required a revision via a stricturoplasty, which has provided definitive pain relief.

The primary limitation of this single institutional experience is that this series did not address the scenario of sub-clinical evidence of anastomotic stenosis, which probably occurs more frequently than we are aware. This could be completely asymptomatic, or might manifest subtly as new-onset or progressive diabetes or exocrine insufficiency. It is not our policy to screen for such findings with regular imaging after all PDs to interrogate the characteristics of either the pancreatic duct or the glandular architecture. However, it is not uncommon to see evidence of mild ductal dilation or even radiographical features of chronic pancreatitis in those patients who have been serially followed with imaging – perhaps for surveillance for recurrence of malignancy. In this series, patients who developed complications from a PJS were readily identified because of the predominant symptom of abdominal pain, and/or the development of biochemical pancreatitis.

While, similar to the Morgan series, all patients treated in this series had overt pain as an indication of their stenosis, it could be argued that an operative approach could also be valuable for those patients who manifest with these other, less obvious and debilitating, symptoms. Generally, operative consideration has thus far been tempered by the fact that these scenarios have less invasive, medical management strategies available. However, it is conceivable that earlier surgical intervention for such strictures may ameliorate declining pancreatic function. Secretin MRCP proved a useful tool for diagnosis as it allows for a ‘functional’ test of the remnant pancreas, and it is more discriminating for ductal stenosis than is traditional static MRCP.²¹ Furthermore, ongoing follow-up of all patients who had an original PD is limited and sometimes haphazard given the large territory our regional referral centre serves. We typically follow these patients with clinical visits through the first year and address complications on an as-needed basis thereafter. Therefore, the true incidence of anastomotic stricture formation in this series may be underappreciated.

There are reports of using balloon dilatation with or without stenting to alleviate stricture-related symptoms.¹⁰ Of the 122 patients in the series from the Mayo Clinic, four PJS were reported. Two patients were managed operatively with revision of the PJA, whereas the remaining two were managed endoscopically, and all four patients were symptom free at follow-up. However, these minimally-invasive measures proved unsuccessful in our patients. In fact, all patients went on to require operative (or attempted operative) revision of their PJAs. In one patient (no. 6, Table 1), even intra-operative ERCP access directly into the enteric limb was futile after the pancreas was inaccessible from dense adhesions.

Operative revisions of these strictures can be technically demanding and require flexibility on the surgeons part. The first challenge is navigating to the pancreas through what can be a field of dense adhesions and scar tissue, particularly when a pancreatic fistula originally occurred. In one patient, we were unsuccessful at even safely identifying the pancreas. In fact, the blood supply to the transverse colon was compromised and required a partial colectomy. If the PJA is in fact identified, complete dissection of the strictured limb-pancreas complex off the underlying porto-splenic venous confluence may be possible. In that situation, a complete revision may be undertaken by resecting back on the pancreas over a centimetre from the stricture. This usually identifies a moderately dilated pancreatic duct suitable for another end-to-side duct-to-mucosa apposition. In other cases, the degree of ductal dilation is minimal and the reconstruction to another small duct may be unsatisfying. In such a circumstance, a ‘modified Puestow’ might be possible by longitudinally opening the duct on its anterior surface for a few centimetres. Usually there is enough redundancy in the end of the jejunal limb to appose it as a hood over the pancreatic duct trough. However, in some patients, pancreatitis induced by the obstructing stricture may make the strictured anastomotic complex ‘frozen’. If this is encountered, a trans-jejunal stricturoplasty can be entertained. Finally, if a markedly dilated obstructed duct (>6 mm) is the circumstance, longitudinal drainage procedures (a traditional Puestow or a side-to-side pancreaticogastrostomy) are options.²²

PG offers an alternative method of reconstruction after pancreaticoduodenectomy and is favoured by some surgeons over pancreaticojejunostomy, especially for soft glands. Putatively, this may be less prone to long-term stricture formation – particularly if the remnant pancreas is introduced into the posterior stomach wall for a distance of a few centimetres without a direct apposition of the pancreatic duct to the gastric mucosa (as described by Bassi et al.²³). We have a limited and unfavourable experience with this technique, with three out of four patients developing symptomatic post-operative ductal stenosis. In each circumstance, the pancreatic tissue and duct was obscured by overgrowth of the gastric mucosa. Literature about rates of this complication after PG is lacking, however, there are papers describing both endoscopic²⁴ and surgical^22,25 solutions to this dilemma. Another series of 61 PGs indicates that stricture is a significant factor in the development of post-operative pancreatic exocrine insufficiency.²⁶

Finally, a technically satisfying revision operation does not guarantee long-term symptom relief. Just over half of our patients are pain free after a mean follow-up of 2 years. This unpredictable outcome may reflect inaccurate diagnosis of the source of the original abdominal pain. Pancreatitic pain could be attributed to stricturing of the pancreatic duct remote from the anastomotic site. However, this scenario was not recognized on any of our patients on either their pre- or post-operative MRCP imaging. Causes of abdominal pain beyond clinically evident pancreatitis should be explored pre-operatively. Patients with recurrent pain after a stricture revision should be worked-up de novo for another PJS, as we had one patient who indeed benefited from another operative endeavour. However, once a stricture originally occurs, obstructive pancreatitis is induced in what was most often normal pancreas at the index procedure. As pointed out by Morgan, the contribution of such fibrotic pancreatitic changes to chronic abdominal pain remains elusive.¹¹ While the stricture may contribute to episodic pancreatitis events, chronic, enduring pain is most likely from the resultant fibrosis which may not manifest biochemically.

Conclusions

Occurrence of PJS is rare and appears to be independent of original pathological, glandular or technical features; however, pancreatic fistula likely predispose to PJS by inducing an inflammatory reaction with aggressive scar formation. Secretin MRCP provides a useful, non-invasive tool to identify these strictures, whereas ERCP has proven therapeutically ineffective. Operative solutions (PJA revision, remnant resection and ductal obliteration) are often required, and require technical flexibility. As the incidence of this complication is still small, it is difficult to draw absolute conclusions, although it appears that durable resolution of symptoms after surgical revision is unpredictable. The number of longer-term survivors after PD is likely to grow in the next several years, providing further opportunity to study long-term complications.

Conflicts of interest

None declared.