Pancreatitis After Pancreatoduodenectomy Predicts Clinically Relevant Postoperative Pancreatic Fistula
- 299 Downloads
Morbidity after pancreas resection is still high with postoperative pancreatic fistulas (POPF) being the most frequent complication. However, exocrine insufficiency seems to protect from POPF. In clinical practice, patients showing increased postoperative systemic amylase concentrations appear to frequently develop POPF. We therefore retrospectively examined the occurrence of systemic amylase increase after pancreas resections and its association with the clinical course.
Patients and Methods
Perioperative data from 739 consecutive pancreas resections were assessed in a prospectively maintained SPSS database. Serum and drain amylase concentrations were determined by routine clinical chemistry. POPFs were graded into A–C according to ISGPF definitions.
In patients with reduced serum amylase (n = 89) on day 1 after pancreatoduodenectomy, clinically relevant POPFs were not observed. In patients with normal serum amylase concentrations, clinically relevant POPFs occurred in 9 %, while in 39 % of the patients with more than three times elevated amylase concentrations, a clinically relevant postoperative fistula was observed (p < 0.001). Systemic hyperamylasemia detected on postoperative day 1 after pancreatoduodenectomy was further a good predictor for clinically relevant POPFs (AUROC = 0.797, p < 0.001).
Patients with a high risk for developing clinically relevant POPFs can be identified on the first postoperative day by determining serum amylase.
KeywordsPancreatoduodenectomy Distal pancreatectomy Pancreatic fistula Postoperative pancreatitis
Pancreatic surgery has always been challenging and associated with complications. Despite huge progress in perioperative management and treatment of complications, morbidity rates still range from 30 % up to 60 %.1–3 The most common but also most inconvenient complication is development of postoperative pancreatic fistula (POPF) which was precisely defined in 2005 by an international group of experts.4 Known reasons for POPF development after pancreatoduodenectomy are soft pancreatic tissue and a small pancreatic duct size while after distal pancreatectomies these factors are associated with lower fistula rates.2,5 After pancreatic head resection, a hard pancreatic tissue, which is mostly found in patients with chronic pancreatitis and pancreatic adenocarcinoma, is associated with a reduced susceptibility for POPF formation.
Surprisingly, prospective randomized trials indicated that the surgical techniques applied appear to be less important for POPF formation after pancreatoduodenectomy as well as after distal pancreatectomy.6,7 Therefore, beneath technical and mechanical aspects, there seem to be other reasons and risk factors involved in the development of POPF after pancreas resections. An inflammation due to mechanical damage of the pancreatic parenchyma induced by the surgical procedure together with exocrine sufficiency of the pancreatic gland may be one factor of importance. Similar to acute pancreatitis, the damage of the pancreatic parenchyma is also likely to be associated with increased serum amylase concentrations. To investigate this hypothesis, we explored the occurrence of increased systemic amylase concentrations on the first three postoperative days and its correlation with the complications after pancreas resections.
Perioperative data of all patients treated with pancreatoduodenectomy or distal pancreatectomy at our institution between August 2001 and July 2014 were continuously collected in a SPSS database, which was approved by the local ethics committee. Pancreas resections were mostly conducted by one of four experienced pancreatic surgeons. After pancreatoduodenectomy, the pancreatic anastomosis was performed as either a duct-to-mucosa pancreaticojejunostomy or a pancreatogastrostomy in a standard fashion as described elsewhere.7 In a subset of patients, external pancreatic duct drainage was used to protect the pancreaticojejunostomy. In patients that underwent DP, the pancreas was either transected by scalpel, high-energy device, or varying staplers. If the transection was performed by stapler, no additional closure of the pancreatic remnant was performed. In all other patients, the duct was additionally occluded by a non-absorbable suture. In a subset of patients, the pancreatic stump was additionally covered by the second jejunal loop. The jejunum remained closed, and in that case, the parenchyma-to-serosa suture was either performed by trans-pancreatic sutures, single stitches, or running sutures. At the end of the operation, a soft silicon drain was placed at the pancreatic anastomosis. Serum and drain amylase levels were determined daily.
The data analysis included demographic factors, results of the histopathological examination, and factors related to the operation or the postoperative course. Postpancreatectomy pancreatitis was defined by serum amylase. With the upper normal serum amylase at our institution being 53 U/l, patients showing values more than three times this value (serum amylase of >159 U/l) were defined as having developed postpancreatectomy pancreatitis (measured on postoperative day 1 (POD1)). Development of postoperative pancreatic fistula (POPF) was recorded according to the ISGPF classification.4 In 3.3 % of the patients after PD resp. DP, an postoperative drain needed to be placed additionally. Due to the retrospective manner of the study, it could not be clarified if these patients were primarily misclassified or had other reasons for abscess formation. For further subgroup analyses, four groups were built according to their serum amylase levels on day 1 after surgery: the first group was defined as below-normal serum amylase ranging from 0 to 12 U/l, the second group had normal serum amylase values from 13 to 53 U/l, the third group had elevated serum amylase concentration but no pancreatitis (54–158 U/l), and the last group was defined as postpancreatectomy pancreatitis with at least three times elevated serum amylase of higher than 159 IU/l.
Data analyses were performed with SPSS (IBM SPSS Statistics, last version used: 23.0; Armonk, NY, USA, IBM Corp.). Descriptive statistics as well as Fisher’s exact test and Pearson’s chi-squared test were performed in addition to non-parametric Kruskal-Wallis and Mann-Whitney-U test when appropriate. Predictive values were calculated by receiver operating characteristics (ROC) and expressed as area under the curve (AUC). Univariate analysis and multivariate binary logistic regression with calculation of odds ratios were conducted to discover independent risk factors for development of a postpancreatectomy pancreatitis.
A total of 863 patients underwent pylorus-preserving pancreatoduodenectomy or distal pancreatectomy between August 2001 and July 2014 at our institution. In 739 patients, serum amylase was available for postoperative day 1 (POD1). These 739 patients were included in our analyses.
The median age of the patients, consisting of 46 % men and 54 % women, was 65 years (range 15–89). History of alcohol abuse was reported in 12 % of the patients, and 23 % of the patients reported former or ongoing smoking. An American Society of Anesthesiologists score of 3 or higher existed in 33 % of the patients.
Indications for resection
38.2 % (283)
31.5 % (233)
2.4 % (18)
4.3 % (32)
8.8 % (65)
Distal bile duct cancer
7.2 % (53)
2.3 % (17)
5.5 % (41)
7.6 % (56)
16 % (118)
Other benign lesion
9.6 % (71)
Other malignant lesion
2.6 % (20)
2 % (15)
Occurrence of Postpancreatectomy Pancreatitis (PPP)
Perioperative parameters depending on development of postpancreatectomy pancreatitis
Variable [n = 739]
Serum amylase POD1 < 159 [n = 483]
Serum amylase POD1 ≥ 159 [n = 256]
Time of operation [min]
All [n = 737]
PD [n = 560]
DP [n = 177]
Stay on ICU [days]
All [n = 203]
PD [n = 149]
DP [n = 54]
Hospital stay [days]
All [n = 235]
PD [n = 188]
DP [n = 47]
Blood loss [ml]
All [n = 317]
PD [n = 240]
DP [n = 77]
Days until removal of drain
All [n = 690]
PD [n = 532]
DP [n = 158]
Median serum amylase on POD1 depending on surgical aspects and patient-related factors after PPPD
n = 561
Median serum amylase (95 % CI) U/l on POD1
Surgical technique [n = 561]
Preop. biliary stenting [n = 552]
Pancreatic anastomosis [n = 556]
BMI [n = 556]
Association of PPP with POPF
A/B POPF development dependent on serum amylase on POD1
Lower serum amylase (0–12 U/l)
Normal serum amylase (13–58 U/l)
Elevated serum amylase (54–158 U/l)
Postpancreatectomy pancreatitis (>159 U/l)
A: after PPPD [n = 561]
B: after PD [n = 178]
Association of Type of Resection and Meaning of Serum and Drain Amylase
PPP Predicting POPF (ROC)
Risk Factors and Clinical Relevance of Postpancreatectomy Pancreatitis in Patients After PD
Univariate and multivariate analyses for independent risk factors for postpancreatectomy pancreatitis (PPP) development after PPPD
95 % CI
95 % CI
Coronary heart disease
Additionally, postpancreatectomy pancreatitis on POD1 seems to be a trigger for POPF but not a motor of postoperative complications since the occurrence of pancreatic fistulas but not the occurrence of postpancreatectomy pancreatitis is associated with postoperative complications such as infectious complications, re-interventions, or death.
Mortality after pancreatic surgery has been significantly reduced throughout the past decades. This was achieved by an optimized perioperative management as well as more subtle surgical procedures reducing the surgical trauma. However, the morbidity after pancreas resections remains high and approximately every second patient experiences any complication throughout the clinical course. Despite intense research focusing on the pancreatoenteric anastomosis, the rate of the most common complication, the postoperative pancreatic fistula, has remained unaltered.8,9 Apparently, in addition to the technique of the pancreatoenteric anastomosis, other still undefined factors must be involved in inducing the leakage of pancreatic enzymes from the pancreatoenteric anastomosis causing postoperative pancreatic fistulas.10–13
The only other known condition to cause pancreatic fistulas apart from pancreas resections is acute or chronic pancreatitis, where pancreatic tissue turns necrotic inducing the leak of pancreatic enzymes.14,15 In analogy to pancreatitis, a local inflammatory process occurring in the area of the pancreatic anastomosis after pancreas resection could induce increased systemic amylase concentration and subsequently impaired healing of the anastomosis. This is especially possible since inflammation and acute postoperative pancreatitis has scarcely been described as one risk factor for development of POPF.16–18
To investigate the effect of postoperative pancreatitis on the clinical course after pancreas resection, we defined postoperative pancreatitis as an elevation of serum amylase of more than three times the upper normal limit. The other factors defined by the Revised Atlanta Classification for acute pancreatitis could not be utilized since all patients suffer from abdominal pain and no data was available for characteristic findings of acute pancreatitis on contrast-enhanced computed tomography on the first postoperative day after pancreas resection.
When analyzing serum amylase concentrations on the first postoperative day, surprisingly, a substantial number of patients developed postpancreatectomy pancreatitis which confirmed previous findings using the arbitrary cutoff of a serum amylase greater 130 U/l.19 Patients with postpancreatectomy pancreatitis were characterized by reduced operation times and lower blood loss, which indicated that patients with postpancreatectomy pancreatitis were easy to resect. Consequently, preoperative biliary stenting as well as the histopathological diagnosis of ductal adenocarcinoma or chronic pancreatitis were favorable factors for not developing acute postpancreatectomy pancreatitis. Two additional factors modulating the risk for postpancreatectomy pancreatitis were pancreatogastrostomy as a risk factor and coronary heart disease as a protecting factor. This latter finding is in line with other observations of coronary heart disease being associated with a reduced rate of POPF after pancreas resections20 and smoking as a risk factor for both coronary heart disease and pancreatic cancer. This may be the connecting explanation for this finding. While pancreatogastrostomy had a higher rate of postpancreatectomy pancreatitis, however, the rate of pancreatic fistula appears to be comparable between pancreatojejunostomy and pancreatogastrostomy.8,21,22
Postpancreatectomy pancreatitis was associated with a substantial risk for POPF formation. However, our analyses indicated two situations, which questioned our current definition of postoperative pancreatic fistula. In contrast to the ISGPF definition defining POPF as “abnormal communication between the pancreatic ductal epithelium and another epithelial surface containing pancreas-derived, enzyme-rich fluid,” our data suggests that the absolute amylase content should not be a key criterion for assessing leaks associated to the pancreatoenteric anastomosis in patients with exocrine insufficiency.4 In these patients, no pancreatic fistulas occurred at all which is unlikely to occur in almost 89 patients. Therefore, modified definitions adopted from hepatic surgery, where bile leaks are defined as a bilirubin level in drains at least three times the serum bilirubin, may be more realistic for the goal of detecting leaks from the pancreatoenteric anastomosis.23 Vice versa, patients with postpancreatectomy pancreatitis after the third postoperative day all develop grade A fistulas in the absence of leaks from the anastomosis since elevated serum amylase also increases the peritoneal fluid amylase then detected in drains.
Another interesting observation is the apparent difference in the pathomechanism in pancreatoduodenectomy and distal pancreatectomy for POPF development. As serum amylase showed an excellent sensitivity and specificity for predicting POPF after pancreatoduodenectomy, this was not the case for distal pancreatectomy at all. Probably, the occlusion of the pancreatic remnant after distal pancreatectomy is less dependent on the healing processes but appears to rely more on mechanical stability. In contrast to that, the development of POPF after pancreatoduodenectomy appears to be more complex and also appears to be possibly negatively influenced by local inflammatory events more than mechanical effects induced by sutures. This view is supported by a subgroup analysis of patients that had pancreatic stump closure after distal pancreatectomy by an excluded jejunal loop with transpancreatic sutures which had no differences neither in POPF development nor in the frequency of PPP. A potential hypothesis for the observed postpancreatectomy pancreatitis and subsequent POPF formation could be the anatomy of the pancreatic duct system. In the pancreas, the branch ducts flow into the main duct from peripheral to central in a caudal-to-cephalic manner as in a quill. After pancreas resection, there may be branch ducts entering the surface of dissection which are not connected to the main duct. This subsequently could cause congestion of the pancreas juice in a specific pancreatic segment leading to localized pancreatic inflammation. However, further studies confirming this hypothesis need to be performed.
In summary, elevated pancreatic enzymes can frequently be detected in the blood on the first day after pancreas resection and are associated with the formation of postoperative pancreatic fistulas. This association is specific for pancreatoduodenectomy and indicates that to date unknown pathomechanisms occur during the formation of a pancreatic fistula that are independent of the surgical technique for reconstruction.
U.W. and C.K. designed the study; N.S. acquired the data; N.S. analyzed the data; C.K., O.S., and F.M. interpreted the data and performed statistical analysis. C.K. and U.W. wrote the manuscript. F.M. and U.H. helped revise the manuscript. All authors discussed the data and the analysis methods and contributed to the manuscript.
Compliance with Ethical Standards
No funding was received for this study.
Conflict of Interest
The authors declare that they have no conflicts of interest.
- 8.Crippa, S. et al. Pancreaticojejunostomy is comparable to pancreaticogastrostomy after pancreaticoduodenectomy: an updated meta-analysis of randomized controlled trials. Langenbecks Arch. Surg. 401, 427–437 (2016).Google Scholar
- 17.Ridolfi, C. et al. Morphohistological features of pancreatic stump are the main determinant of pancreatic fistula after pancreatoduodenectomy. BioMed Res. Int. 2014, 641239 (2014).Google Scholar
- 21.Bassi, C. et al. Reconstruction by pancreaticojejunostomy versus pancreaticogastrostomy following pancreatectomy: results of a comparative study. Ann. Surg. 242, 767–773 (2005)Google Scholar