Pancreatic Fistula, Predictors Factors and Post Operative Management

Background The aim of the PhD project was to evaluate ideal PF management strategies and on the base of assumed and shared clinical and biological PF concepts identify predictors factors during the post operative management after pancreticoduodenectomies. Prof. Claudio Bassi in our Institution published in 2010 a randomized clinical trial proposing an early removal of the surgical drains in post operative day 3 (POD3) in patients with low risk for pancreatic fistula. In that study, the dosage of the drain amylase in POD1 and POD3 was used to indentify a group of patients with high risk to develop a PF. The aim was to test the hypothesis that in patients with drains amylase < 5000U/L an early drain removal in POD3 was associated to lower rate of PF and abdominal complications. To consolidate the hypothesis that the value of 5000 U/L drain amylase could represent a cut off to identify group of risk patients, a international multicentre data collection was designed with the aim to collect more data among different “cultural” approach in the management of the patients and their surgical drain after pancreaticoduodenectomies. Patients and Methods Between January 2011 to June 2013 a total of 121 patients data undergoing to pancreaticoduodenectomy were collected among two Pancreatic Surgical Centre of tertiary University Teaching Hospitals in United Kingdom: Oxford Upper GastroIntestinal Surgical Centre at Oxford University Hospitals and Peninsula Hepatico-pancreatico-biliary Cancer Centre at Plymouth Hospitals NHS Trust. The data were collected through a standard form submitted for any suitable candidate patient who received a pancreaticoduodenectomy. Inclusion criteria were consisting in patients with periampullary tumors who underwent to standard pancreatic resections. Exclusion criteria were consisting in clinical suspect of postpancreatectomy hemorrhage (defined according to ISGPS and re-laparotomy within 72 hours from index operation). Discharge criteria: Absence of temperature (<37.5°C for >48 h), Adequate pain control with oral analgesics, Ability to take solid foods (at least 1000kcal/day), Passage of stools, Adequate mobilization, Absence of complications, Acceptance of discharge by patient. A list of complications and a protocol are formalized according with the literature and International Pancreatic Surgical Centre protocol preoperative and postoperative management. Results A total 121 non-consecutive pancreaticoduodenectomy cases were collected between the two institutions. Patients were divided into two groups based on the presence of a PF, which was defined according to the International Study Group on Pancreatic Fistulae (ISGPF): elevated output from the surgical drain on or after POD3 with an amylase level ≥3 times the upper normal serum value. All of them (%) had serum amylase on POD1 and POD3 recorded. Twenty-three of the 121 developed a PF (19 %): 7 type A, 14 type B, and 2 type C. Patients with a PF had a mean drain amylase on POD1 of 758 U/L and on POD 3 of 4283 compared to POD1 of 246U/L and POD3 187U/L in control patients (p<0.05). We had no patients with drain amylase > 5000 U/L on POD1. Discussion The development of the ISGPF grading system created a standardized classification scheme to permit enhanced research and communication among surgeons and scientists. Since the ISGPF classification was adopted, the drain amylase trend over the first post-operative days was object of numerous studies with the aim to find a predictive cut off value to better address the further management of the patients after their pancreaticoduodenectomy. In this scenario the Verona study founding that amylase content of the drainage fluid of more than 5000 units/L had high positive and negative predictive values for the prediction of clinical fistula was an excellent example. In our project we were meant to consolidate this hypothesis but we were unable to confirm the same Verona study cut off. One of the other major drawbacks was clearly the lack of number and the limited population and centres participating to the data collection. The ideal plan to identify predictor factors of PF would be analyse data from several centres, with different level of experience and different “local and cultural approach” at the management of this patients, to determine the independent predictive factors and then validate the results in another independent population, prospectively.