Preventing pancreatic fistula following pancreatectomy: meta-analysis of randomized clinical trials

Background: The majority of postoperative morbidity and mortality after partial pancreatectomy is attributable to postoperative pancreatic fistula. Several interventions to prevent postoperative pancreatic fistula have been assessed in randomized clinical trials but consensus on their efficacy is lacking. This is the first systematic review and meta-analysis of randomized trials focusing on strategies to prevent postoperative pancreatic fistula after all types of partial pancreatectomy. Methods: Eligible randomized clinical trials assessing postoperative pancreatic fistula following partial pancreatectomy were included from the Evidence Map of Pancreatic Surgery database (2005-2025). As primary endpoint, both the 2005 and 2016 International Study Group for Pancreatic Surgery definitions of grade B/C postoperative pancreatic fistula were accepted. Interventions shown to reduce postoperative pancreatic fistula in at least one randomized trial were identified, and meta-analysis undertaken for those with at least three available trials, irrespective of their outcome. Results: Overall, 193 RCTs were included with 29 408 patients from 24 countries. The pooled rate of postoperative pancreatic fistula grade B/C was 15.1%: 14.2% after pancreatoduodenectomy and 19.1% after left pancreatectomy. Overall, 24 trials identified 18 interventions as effective in reducing the rate of postoperative pancreatic fistula: 11 intraoperative and 7 postoperative. Eight interventions were eligible for meta-analysis, involving 46 trials. Following meta-analysis, three interventions remained effective: drain omission in left pancreatectomy, perioperative use of somatostatin analogues, and administration of perioperative corticosteroids. The certainty of evidence was high for drain omission, moderate for somatostatin analogues, and low for corticosteroids. Eleven trials focused specifically on high-risk patients and four on ongoing postoperative pancreatic fistula. No trial combined interventions. Conclusion: Omission of drains in left pancreatectomy, perioperative use of somatostatin analogues, and administration of perioperative corticosteroids are effective strategies to prevent postoperative pancreatic fistula after partial pancreatectomy. However, current evidence remains immature and further RCTs evaluating these interventions are warranted, as they are likely to change the conclusions of subsequent meta-analyses. Such trials could potentially focus on high-risk patient cohorts and/or assess combinations of interventions.