Evaluation of external validity of the distal pancreatectomy fistula risk score (D-FRS) in a high-volume center.

Abstract

Background: To reduce the risk of Clinically Relevant Postoperative Pancreatic Fistula (CR-POPF) following distal pancreatectomy (DP), preoperative and intraoperative Distal Pancreatectomy Fistula Risk Scores (D-FRS) were developed. While these models have demonstrated strong internal discrimination, external validation is needed. Therefore, this study aims to evaluate the discrimination and calibration of both risk models in an external cohort of patients undergoing DP.

Methods: This retrospective cohort study included adult patients undergoing DP in a high-volume center (2020-2024). Preoperatively, all patients underwent a triple-phase CT scan measuring the pancreatic duct diameter (MPD, mm), neck thickness (mm), and late-early (L/E) phase attenuation ratio (L/E < 1 = soft texture). Preoperative D-FRS was calculated as the predicted probability based on MPD and neck thickness. Intraoperative D-FRS was calculated using MPD, neck thickness, body mass index (BMI, kg/m²), intraoperative time, and L/E ratio. CR-POPF was defined according to ISGPS criteria. Models' discrimination and calibration were assessed using the Area Under Curve (AUC) and calibration plot (ideal intercept = 0; slope = 1).

Results: A total of 521 patients were included, 58% of whom underwent laparoscopic DP. CR-POPF occurred in 128 (25%) patients. CR-POPF was significantly associated with a higher BMI (p = 0.019) but not with pancreatic duct diameter, thickness, operative time, or L/E ratio. Both preoperative and intraoperative D-FRS models demonstrated poor discrimination, with an AUC of 0.51 (95% CI: 0.45-0.56) and 0.52 (95% CI: 0.46-0.58), respectively. The preoperative D-FRS exhibited poor calibration, with an intercept of 0.342 and a slope of -0.052, while the intraoperative D-FRS showed an intercept of 0.892 and a slope of -0.008.

Conclusion: Both preoperative and intraoperative D-FRS had poor discrimination and calibration ability and tended to overestimate the risk of fistula. In our clinical context, D-FRS cannot be applied without further adjustment and recalibration.