Purpose: This study aimed to establish a reliable criterion for early drain removal after pancreaticoduodenectomy (PD) based on predictive factors of clinically relevant postoperative pancreatic fistula (CR-POPF) available on postoperative day 3 (POD3).
Methods: A total of 300 consecutive patients who underwent PD with pancreaticojejunostomy at our hospital from 2011 to 2015 were analyzed retrospectively. CR-POPF was defined as POPF grade B or C according to the definition by ISGPF. Clinicopathological factors available on or before POD3 were analyzed to identify predictors of CR-POPF. Using obtained predictors, we developed a criterion for no CR-POPF and internally validated its relevance in 100 consecutive patients.
Results: The incidence rates of CR-POPF, severe complications (Clavien-Dindo ≥ grade IIIa), and postoperative mortality were 35%, 9.6%, and 0.3%, respectively. Multivariate analysis showed that drain amylase (d-AMY) levels ≥ 350 IU/l on POD3, C-reactive protein (CRP) levels ≥ 14 mg/dl on POD3, preoperative endoscopic retrograde biliary drainage, and no portal vein resection were significant predictors of CR-POPF. Using the strongest predictors (i.e., d-AMY and CRP), we established a criterion for no CR-POPF: d-AMY levels < 350 IU/l and CRP levels < 14 mg/dl on POD3. The incidence rates of CR-POPF were 6%, 38%, and 88% in patients who fulfilled both of (n = 149), each of (n = 74), and none of (n = 77) the two factors, respectively. In the internal validation cohort, the positive predictive value of CR-POPF was 89%.
Conclusions: A simple two-factor criterion available on POD3 after PD has a reliable predictive ability. In patients who fulfill this criterion, early drain removal is considered safe.
Keywords: C-reactive protein; Drain; Pancreaticoduodenectomy; Postoperative pancreatic fistula.