Abstract
Background/Aim: An update on the incidence, risk factors, clinical sequalae, and management of postoperative pancreatic fistula (POPF) following cytoreductive surgery (CRS) and hyperthermic intraperitoneal chemotherapy (HIPEC). Patients and Methods: Retrospective analysis of prospectively collected data from the St George CRS/HIPEC database. Results: Sixty-five (5.7%) out of 1,141 patients developed a POPF. Patients with POPFs were older, had a higher peritoneal cancer index, longer operation time, and required more units of blood intraoperatively. Splenectomy and distal pancreatectomy were significant risk factors for developing POPFs. While there was no effect on overall long-term survival in POPF patients, they did suffer higher rates of Clavien–Dindo grade 3/4 complications, in-hospital deaths, and longer hospital length of stay. Of the 65 POPF patients, 23 were taken back to theatre, 48 required radiological drains and 7 underwent endoscopic retrograde cholangiopancreatography. Conclusion: There are multiple risk factors for developing POPFs that are non-modifiable. While POPFs are associated with increased postoperative morbidity, long-term survival does not appear to be affected.
- Cytoreductive surgery
- heated intraperitoneal chemotherapy
- peritoneal carcinomatosis
- postoperative pancreatic fistula
- splenectomy
- distal pancreatectomy
The prognosis of peritoneal carcinomatosis (PC) has improved markedly over the past 30 years (1). Patients were previously resigned to palliative surgery and/or palliative chemotherapy. However, a combination of aggressive cytoreductive surgery (CRS) and heated intraperitoneal chemotherapy (HIPEC) has gained increased support and has become the mainstay of treatment in many centres around the world (2). Its main criticism has been the perceived high mortality and morbidity rates associated with the procedure. Major complication rates at high volume centres range between 1-34% (3-7). Schwarz et al. summarised this issue such that “the risk/benefit ratio must be considered and adapted for each patient to fit a stringent selection process for patients at an acceptable risk of mortality” (2).
As acceptance has increased worldwide, the focus of studies has shifted towards refining this selection process by examining clinical and treatment-related risk factors to help reduce both morbidity, mortality and hospital stay. There have been limited reports on the incidence, risk factors, and outcomes related to postoperative pancreatic fistulas (POPFs) in CRS/HIPEC. Data published from our unit in 2010 found that intraoperative time greater than nine hours, transfusion of greater than six units of blood intraoperatively and splenectomy were all independent risk factors for developing POPFs (1). In 2016, Schwarz et al. published data demonstrating that peritoneal cancer index (PCI) greater than 20 and intraoperative time greater than 550 min (9.16 h) were risk factors for developing POPFs, yet failed to show an association with splenectomy (2). Conversely, some studies have argued that an operative time greater than nine hours and the number of units of blood transfused intraoperatively are likely surrogate markers for a more extensive CRS procedure rather than true risk factors (1, 2, 7). The variance in risk factors for POPFs has been further highlighted by both Mahadi et al. and Downs-Canner et al. who found no correlation with length of surgery, PCI, blood requirements intraoperatively, age, gender, type of primary or American Society of Anaesthesiologists score (ASA) (8, 9).
This update aimed to further examine the discrepancies found in the literature regarding the clinical risk factors and treatment of POPFs following CRS/HIPEC. To our knowledge, this is the largest study to date that examines POPFs in a cohort of patients undergoing CRS/HIPEC at a single centre, while outlining both clinical and treatment-related risk factors.
Patients and Methods
Selection criteria and Study criteria. In this study, standardized clinical data of 1,141 patients was retrospectively reviewed from the prospectively maintained St George Hospital CRS/HIPEC database. It includes all patients undergoing CRS with or without HIPEC from February 1996 to July 2017. All patients within this database had a preoperative workup which included a combination of computed tomography (CT) and positron emission tomography (PET) scans, diagnostic laparoscopy, biopsy confirmed PC and review by a multidisciplinary team (including medical oncologists, treating surgeons and radiologists). Patients were deemed fit for surgery if they had no extra abdominal metastasis, good performance status (WHPS ≤2) (10) and were under the age of 80 years. Clinical, treatment, and non-modifiable risk factors were considered and analysed for potential impact on POPFs.
Cytoreductive Surgery + HIPEC. As a single centre study, all patients were operated on by a single surgical team and the HIPEC treatment also was standardised throughout the population group. The technique used for CRS was based on the Sugarbaker approach (4, 11) and included primary tumour removal, gastrointestinal tract resections, resection of tumour nodules and peritonectomy. The aim of CRS was complete macroscopic disease removal. Upon entering the abdomen, the PCI score was calculated using established guidelines (12) to accurately score macroscopic disease from 0-39. Completeness of the macroscopic resection was graded at the completion of surgery using the CC score; CC0 indicates no residual macroscopic disease; CC1 indicates no nodule over 2.5 mm in diameter remains; CC2 indicates that nodules between 2.5 mm and 2.5 cm remain; and CC3 indicates that nodules over 2.5 cm in diameter remain. After macroscopic cytoreduction is complete but prior to repair of serosal tears or intestinal anastomosis, HIPEC was introduced using the colosseum technique. HIPEC used included mitomycin C, cisplatin and doxorubicin. No closed HIPEC techniques were used. At least one surgical drain was inserted prior to closure.
Postoperative management and complications. Postoperatively, patients were treated in accordance with a predefined protocol (13) and complications were stratified according to the Clavien–Dindo classification (14). Pancreatic fistulas were diagnosed according to the guidelines set out and updated in 2016 by the International Study Group of Pancreatic Fistula (ISGPF) (15, 16). The diagnosis of POPF was made if any drain output had an amylase recording of more than three times the upper limit of normal, on or after postoperative day three. The 2016 update has redefined ‘grade A’ as a simple biochemical leak if it fulfils the criteria as stated above. ‘Grade B’ requires a change in the postoperative management (drains remain for more than 3 weeks which are repositioned through endoscopic or percutaneous procedures; signs of infection without organ failure; or angiographic procedures for bleeding). Lastly, ‘grade C’ criteria included POPFs that led to reoperation, single or multiple organ failure and/or death. In our study, drain amylase was only measured in patients with clinical symptoms and/or abnormal drain output (volume and appearance), therefore limiting our diagnosis to clinically relevant fistulas (mostly grades B and C). Duodenal and gastric fistulas (perforation and/or anastomotic leak) were not classified as pancreatic fistulas to minimise confounding biochemical results.
Data collection and statistical analysis. Continuous variables were compared using Student’s t, Wilcoxon rank-sum, one-way analysis of variance ANOVA, and/or Kruskal–Wallis-tests as appropriate. Where necessary, log- transformation of data was performed to achieve normal distribution. Differences between proportions derived from categorical data were compared using Pearson’s χ2- or Fisher’s exact test where appropriate. Data are reported as median with inter-quartile range (IQR) unless denoted otherwise. Serum and body fluid (drain) amylase and lipase levels were extracted from hospital medical records, with local cut-offs being used for elevated/normal levels.
To assess the impact of POPFs on long-term patient outcomes, POPF stratified survival times were plotted using the Kaplan–Meier method and differences in survival were calculated using the log-rank test. Overall survival was defined as time from surgery to date of death or last follow-up. Recurrence-free survival was defined as time from date of surgery to recurrence at any site or death, whichever occurred first. Only patients with complete data were included in the survival analyses. Finally, uni- and multi-variate logistic regression analyses were performed to determine other clinical predictors of POPFs. Factors that were significant upon univariate analysis were taken forward for multivariate analysis. All p-values <0.05 were regarded as statistically significant and all analyses were performed using R Statistical Packages (17).
Results
Patient factors. A total of 65 POPFs were evaluated from the prospective database of 1,141 patients. These patients were further stratified by age, sex, histopathologic diagnosis, ASA and Eastern Cooperative Oncology Group Performance Status (ECOG) as demonstrated in Table I. Of these factors, only sex and age were statistically significant, with POPFs occurring more frequently in older patients (mean 58.6 vs. 53.5 years, p=0.002) and in males compared to females (8.2% vs. 3.7%, p=0.001).
Preoperative factors. During preoperative planning, patient albumin, haemoglobin and tumour markers were recorded and analysed. Of these factors, only a higher CA125 was associated with POPFs (mean 87.8 vs. 64.3, p=0.02).
Operative factors. Stratifying patients by their operative characteristics showed that POPF patients had higher PCIs (25 vs. 16, p<0.001), more units of blood transfused intraoperatively (8.5 vs. 4.8 units, p<0.001) and longer operations (10 vs. 8.2 h, p<0.001) as demonstrated in Table II. POPF rates were higher in patients that underwent HIPEC (6.4% vs. 0%, p<0.001) and had a CC score greater than 0 (p<0.001), but did not vary in patients that underwent early postoperative intraperitoneal chemotherapy (EPIC) (Table III). Intraoperatively, surgical interventions of splenectomy (12% vs. 1.7%, p<0.001), gastrectomy (12.5% vs. 5%, p=0.002), distal pancreatectomy (52% vs. 3.6%, p<0.001) and/or pancreas stripping (12.5% vs. 5.1%, p=0.004) were all associated with higher rates of POPF development (Table III).
Multivariate analysis. Fifteen clinical and treatment variables were analysed using multivariable regression analysis, including age, sex, histopathology, preoperative and intraperative factors. Of these, only two risk factors were found to be sigificant for POPFs - splenectomy (odds ratio 3.83, 95%CI=1.75-8.94, p=0.001) and distal pancreatectomy/pancreas stripping (odds ratio 20.27, 95%CI=9.24-46.1, p<0.001).
Postoperative course. Of the 65 POPF patients, 4 (6%) had grade A fistulas, 31 (48%) had grade B, while the remaining 30 (46%) patients suffered from grade C fistulas. Twenty-three (35%) were taken back to theatre, 61 (94%) were treated with subcutaneous octreotide, 51 (78%) were given an intravenous proton pump inhibitor (PPI), 57 (88%) were commenced on total parenteral nutrition (TPN), 7 (11%) underwent ERCPs and 48 (74%) required radiological drain insertion. Median time to drainage from the initial surgery was 14 days (range=2-51 days) and median drain removal time was 30.5 days (range=5-96 days). Seven patients were discharged with drains in situ, which were removed in the community. Postoperatively, patients who developed a POPF had a longer hospital admission (mean 43.1 vs. 26 days, p<0.001) and were more likely to have Clavien– Dindo grade III/IV complications (p<0.001).
Overall and recurrence-free survival. While POPF patients had significantly higher rates of in-hospital deaths (7.7% vs. 1.5%, p<0.001), the difference in median overall long-term survival between POPF and non-POPF patients (51 vs. 59 months respectively, p=0.485) was not statistically significant. Similarly, recurrence-free survival did not differ significantly between the groups (18 vs. 19 months respectively, p=0.301). One-, 3-, and 5-year survival rates for POPF vs. non-POPF patients approached 81.5% vs. 86.9%, 60.1% vs. 62.9% and 45.6% vs. 48.8%, respectively, as shown in Figure 1 and Figure 2. Figure 3 and Figure 4 demonstrate the overall and recurrence-free survival of patients by primary diagnosis based on the development of POPF.
Discussion
Classically a terminal illness, PC has transformed into a multifaceted disease amenable to surgery. CRS with HIPEC offers major survival benefits for PC and has become widely accepted as the mainstay of treatment. Due to the gruelling nature of the operation together with the side effects of HIPEC, major CRS is associated with significant complications. Of significance, POPFs are associated with unfavourable postoperative outcomes and a lengthy recovery. This study aimεδ to examine POPFs post CRS/HIPEC to ascertain risk factors and characterise their impact on prognosis and postoperative outcomes.
In this study, 65 patients (5.7%) were diagnosed with POPF, which is comparable to the lower end of the spectrum of previous studies where rates have varied from 4.8% to as high as 33% (1, 2, 7, 18, 19). Inconsistent definitions used prior to the 2005 ISGPF guidelines and subsequent updates severely limit our ability to draw on old literature for reliable conclusions with regards to POPF formation rates, prompting our detailed analysis.
On univariate analysis, three clinical and seven treatment-related variables were significant for POPF development. The clinical risk factors included increased age, higher preoperative CA125 and higher PCI. Treatment variables included longer operative time, greater blood transfusion requirements, receiving HIPEC, incomplete cytoreduction, splenectomy, gastrectomy, distal pancreatectomy, and pancreas stripping. On multivariate analysis, however, these results varied. Only two treatment-related variables were significant for patients to develop POPF – splenectomy (p=0.001) and distal pancreatectomy/pancreas stripping (p<0.001). Interestingly, no clinical variables were significant on multivariate analysis. These findings are in contrast to previous studies which identified the following independent risk factors on multivariate analysis: cisplatin dosage for HIPEC higher than 240 mg, transfusion of more than 6 units of blood, PCI >20, and operative time ≥9 h (1, 2, 18, 19).
Within our cohort, 440 patients (39%) underwent a splenectomy, of which 53 patients (12%) developed a POPF (p<0.001). Saxena et al. and Kusarmura et al. share similar findings, establishing it as an independent risk factor for developing a POPF (1, 18). Of note, the rate of POPFs in splenectomies for other indications is significantly lower, with laparoscopic splenectomy POPF rates ranging between 0-1% (20). This suggests that CRS/HIPEC must have other inherent factors contributing to this higher complication rate.
Although limited by sample size (Saxena n=271, Schwarz n=118, Kusarmura n=265) (1, 2, 18), other studies have previously demonstrated an association between POPFs and longer operative time and/or quantity of blood transfused. These factors likely represent surrogate indicators for a more extensive cytoreductive procedure, increasing the likelihood of dissection within the vicinity of the pancreas and thus precipitating pancreatic injury. Despite our results indicating that these factors are not significantly associated with the development of POPF, further research may be warranted.
The role of HIPEC in increasing the rate of POPFs should be further delineated. In our study, 1016 patients (89%) received HIPEC, with all 65 POPF patients falling within this treatment group. While significant for POPFs on univariate analysis, HIPEC did not yield statistical significance on our multivariate analysis. This is likely limited by the small comparison group size (n=125). As Saxena et al. outlined, “the survival benefits of HIPEC are likely to outweigh any increase in pancreatic morbidity that may result from this procedure” (1), and the relatively low incidence in small population sizes make studying POPF in non-HIPEC patients difficult.
The treatment of POPFs remains quite controversial in the literature. In our cohort, we often used a combination of octreotide, intravenous PPI and/or TPN. While somatostatin analogues have garnered widespread attention for their role in accelerating closure of POPFs (21, 22), the role of TPN in improving outcomes remains debatable. Goh et al. found that 92% of their cohort of 72 patients with pancreatic fistulas could be treated non-operatively while maintaining an enteral diet. This is supported by Klek et al. who found that enteral feeding is associated with “lower incidence of infection, higher 30-day fistula closure rates, and shorter time to closure of postoperative pancreatic fistula as compared with total parenteral nutrition” (23).
Despite previous reports that most pancreatic fistulas can be managed conservatively (24, 25), 7 (11%) of our POPF patients required ERCP and 23 (35%) were taken back to theatre. Our POPF patients experienced higher rates of in-hospital deaths, yet there was no impact on overall long-term survival. In other high-volume centres, procedure-related mortality rates range from 1 to 6% (1) which is comparable to ours.
Whilst a previous study by Downs-Canner et al. suggests that patients with POPF have a higher 1-year disease recurrence rate in CRS (9), our data did not support any changes in recurrence-free survival. The discrepancy may be due to confounding factors associated with POPF such as difficulty of resection and/or incomplete cytoreduction, increasing the likelihood of recurrence.
Conclusion
As CRS/HIPEC for PC remains relatively limited to specialist centres and often requires a radical approach. Reducing postoperative complications, hospital length of stay, and mortality is essential for its implementation globally. We presented the largest single centre study looking into POPFs post CRS/HIPEC. Splenectomy and distal pancreatectomy/ pancreatic stripping appear to be the key risk factors for the development of POPFs. In contrast, operative duration, units of blood given intraoperatively, and PCI>20 were not found to be statistically significant risk factors despite previous reports. No reliable or modifiable preoperative risk factors have been identified thus far. Our findings highlight the importance of surveillance and appropriate postoperative care in patients undergoing splenectomy and/or pancreatic operations as part of major cytoreductive surgery as they are prone to developing POPFs. Surveillance may be routinely implemented in such high-risk cases by measuring drain amylase οn day 3 postoperatively to identify leaks early and manage accordingly, limiting morbidity from POPFs. Notably, whilst POPFs are clearly associated with significant postoperative morbidity, recurrence-free survival and overall long-term survival do not appear to be affected.
Footnotes
* These Authors contributed equally to this study.
Authors’ Contributions
All Authors contributed to the writing of this article. Prof Morris and Dr Alzahrani were the principal surgeons and supervisors in this study and contributed to the study design. Drs Matar and Meares prepared the manuscript, while Drs Gauci and Rao collected and assembled the data. Dr Fisher performed the statistical analysis. Dr Alshahrani reviewed the manuscript.
Conflicts of Interest
The Authors have no affiliations with or involvement in any organisation or entity with any financial or non-financial interest in the subject matter or materials discussed in this manuscript.
- Received August 30, 2021.
- Revision received September 28, 2021.
- Accepted September 29, 2021.
- Copyright © 2021 International Institute of Anticancer Research (Dr. George J. Delinasios), All rights reserved.