Article Abstract

Clinical validation of scoring systems of postoperative pancreatic fistula after pancreatoduodenectomy: applicability to Eastern cohorts?

Authors: Jae Seung Kang, Taesung Park, Youngmin Han, Seungyeon Lee, Jae Ri Kim, Hongbeom Kim, Wooil Kwon, Sun-Whe Kim, Jin Seok Heo, Seong Ho Choi, Dong Wook Choi, Song Cheol Kim, Tae Ho Hong, Dong Sup Yoon, Joon Seong Park, Sang Jae Park, Sung-Sik Han, Sae-Byeol Choi, Joo Seop Kim, Chang-Sup Lim, Jin-Young Jang


Background: Although several prediction models for the occurrence of postoperative pancreatic fistula (POPF) after pancreatoduodenectomy (PD) exist, all were established using Western cohorts. Large-scale external validation studies in Eastern cohorts that consider demographic variables including lower body mass index (BMI) are scarce. The purpose of this study was to externally validate POPF prediction models using nationwide large-scale Korean cohorts.
Methods: Nine tertiary university hospitals in the Republic of Korea participated. Patients’ preoperative characteristics, intraoperative factors, and pathologic findings were evaluated. POPF grades were determined according to the 2016 International Study Group on Pancreatic Surgery definition. Three POPF risk models (Callery, Roberts, and Mungroop) were selected for external validation.
Results: A total of 1,898 PD patients were enrolled. A non-pancreatic disease diagnosis [hazard ratio (HR), 1.856; 95% confidence interval (CI), 1.223–2.817; P=0.004), higher preoperative BMI (HR, 1.069; 95% CI, 1.019–1.121; P=0.006), and soft pancreatic texture (HR, 1.859; 95% CI, 1.264–2.735; P=0.002) were independent risk factors for clinically relevant POPF (CR-POPF). The area under the receiver operating characteristic curve (AUC) values were 0.61, 0.64, and 0.63 on the Callery, Roberts, and Mungroop models, respectively; all were lower than those published in each external validation study.
Conclusions: Western POPF prediction models performed less well when applied to Korean cohorts. Thus, a large-scale Eastern-specific and externally validated POPF prediction model is needed.