Postoperative day 1 combination of serum C-reactive protein and drain amylase values predicts risks of clinically relevant pancreatic fistula. The "90-1000" score.

ABSTRACT

BACKGROUND: Several scoring systems predict risks of clinically relevant postoperative pancreatic fistula after pancreatectomy, but none have emerged as the gold standard. This study aimed to evaluate the accuracy of postoperative day 1 drain amylase and serum C-reactive protein levels in predicting clinically relevant postoperative pancreatic fistula compared with intraoperative pancreatic characteristics. METHODS: Patients who underwent pancreatectomy between 2017 and 2019 were included prospectively. Cutoff values were determined using receiver operating characteristic curves, and a score combining postoperative day 1 drain amylase and serum C-reactive protein was tested in a multivariate logistic regression model to evaluate clinically relevant postoperative pancreatic fistula risk. RESULTS: A total of 274 pancreatic resections (182 pancreaticoduodenectomies and 92 distal pancreatectomies) were included. The pancreatic gland texture was "soft" in 47.8% (n = 131), and 55.8% (n = 153) had a small size main pancreatic duct (≤3 mm). Clinically relevant postoperative pancreatic fistula occurred in 58 patients (21.2%). Drain amylase ≥1,000 UI/L and serum C-reactive protein ≥90 mg/L were identified as the optimal cutoffs to predict clinically relevant postoperative pancreatic fistula. On multivariate analysis these cutoffs were independent predictors of clinically relevant postoperative pancreatic fistula after both pancreaticoduodenectomies (drain amylase P < .001, serum C-reactive protein P = .006) and distal pancreatectomies (drain amylase P = .009, serum C-reactive protein P = .001). The postoperative day 1 "90-1000" model, a 2-value score relying on these cutoffs, significantly (P < .001) outperformed intraoperative pancreatic parenchymal characteristics in predicting clinically relevant postoperative pancreatic fistula after both pancreaticoduodenectomies and distal pancreatectomies. A postoperative day 1 "90-1000" score = 0 had a negative predictive value of 97% and 94%, respectively, after pancreaticoduodenectomy and distal pancreatectomies. CONCLUSION: A combined score relying on postoperative day 1 values of drain amylase and serum C-reactive protein levels was accurate in predicting risks of clinically relevant postoperative pancreatic fistula after pancreatectomy.