Objective: To construct a prediction model of clinically relevant postoperative pancreatic fistula(CR-POPF) in patients with pancreatic cancer, and verify the prediction efficiency and calibration degree of the model. Methods: The clinical data of 115 patients with pancreatic cancer in our hospital from January 2021 to December 2023 were collected. All patients underwent pancreaticoduodenectomy(PD). According to the presence or absence of CR-POPF after operation, they were divided into CR-POPF group(24 cases) and non-CR-POPF group(91 cases). Clinical data such as body mass index(BMI), total abdominal fat area(TFA), main pancreatic duct diameter(MPDD), extracellular volume(ECV) fraction, and abdominal drainage fluid amylase on postoperative day 1(ADFA1) were collected. Logistic regression analysis was used to screen the influencing factors of CR-POPF in patients with pancreatic cancer, and a CR-POPF prediction model was constructed. The receiver operating characteristic(ROC) curve was drawn to evaluate the predictive value of the model, and the Hosmer-Lemeshow goodness of fit test was used to test the validity of the prediction model. Results: Univariate analysis showed that BMI, TFA, MPDD, ECV fraction and ADFA1 were associated with CR-POPF occurrence in pancreatic cancer patients(P<0.05). Binary Logistic regression analysis showed that TFA, MPDD, ECV scores, ADFA1 were independent influencing factors for CR-POPF in pancreatic cancer patients(P<0.05). The CR-POPF prediction model of pancreatic cancer was constructed based on TFA, MPDD, ECV score and ADFA1. The goodness of fit test showed that the model had good discrimination ability(χ2=1.472, P=0.993). The ROC curve analysis showed that the area under curve predicted by the model was 0.986(95% CI 0.970-1.000), the sensitivity was 95.83%, and the specificity was 89.01%. Conclusion: TFA, MPDD, ECV fraction and ADFA1 are influencing factors for the occurrence of CR-POPF in patients with pancreatic cancer. The CR-POPF model established based on these factors has good prediction efficiency and has great significance for early clinical diagnosis and treatment. |