Drain management after pancreatoduodenectomy: risk-stratified dynamic algorithm.

Abstract

Background: The ability of drain fluid amylase (DFA) to predict pancreatic fistula (POPF) and safely guide early drain removal is unknown in high-risk pancreatoduodenectomy (HR-PD). The aim of this analysis was to develop a new risk-based algorithm for drain management after pancreatoduodenectomy.

Methods: All consecutive patients undergoing pancreatoduodenectomy (2019-2024) at two institutions were included. During surgery, patients were stratified into HR-PD, defined according to the International Study Group on Pancreatic Surgery score (soft gland and duct ≤ 3 mm), and non-HR-PD. Postoperative predictors (DFA, serum amylase/lipase, C-reactive protein) were retrieved on postoperative days (PODs) 1 to 5 and correlated to POPF development using a risk-based approach. Results were validated in a retrospective cohort (2015-2019) of pancreatoduodenectomies performed at the same institutions.

Results: Out of 825 screened patients, 788 were included. HR-PD (251, 32%) had higher POPF rates (50% versus 9%; P < 0.001) and worse outcomes. In HR-PD, DFA had a lower area under the curve for POPF for POD 1 (0.73 versus 0.85; P = 0.003) and POD 3 (0.74 versus 0.88; P = 0.002) compared with non-HR-PD, but similar for POD 5 (0.83 versus 0.86; P = 0.535), whereas serum amylase/lipase maintained a low area under the curve for POPF compared with non-HR-PD. The optimal criteria for POD 3 drain removal in non-HR-PD were POD 1 DFA < 200 UI/l, POD 3 DFA < 150 UI/l, and absence of elevated serum amylase/lipase on POD 1 (Youden's index 0.674). In HR-PD, the criteria for POD 5 drain removal were POD 5 DFA < 150 UI/l and POD 5 C-reactive protein < 150 mg/ml (Youden's index 0.593).

Conclusion: In patients with an intrinsic higher risk of POPF, early predictors are less reliable, and drain management protocols should be tailored. In non-HR-PD, an early removal policy can be recommended based on DFA and serum amylase/lipase. In HR-PD, drain removal may be safely deferred until POD 5 according to DFA and C-reactive protein.