Abstract
Background: The distinction between pancreatic ductal adenocarcinoma (PDAC) and benign pancreatic lesions remains challenging. This study aimed to evaluate the utility of computed tomography (CT) imaging features and clinical characteristics in differentiating PDAC from benign pancreatic lesions. Methods: In this retrospective study, a total of 97 patients with PDAC and 90 patients with benign pancreatic lesions were included. Various imaging features and clinical characteristics were assessed. Univariable and multivariable logistic regression analyses were conducted, and receiver operating characteristic (ROC) curves and their corresponding areas under the curve (AUCs) were assessed. The optimal cut-off value for D-dimer was determined using the Youden index. The DeLong test was employed to compare the AUCs of the ROC curves between different prediction models. Results: The clinical and radiologic models achieved AUCs of 0.86 and 0.85, respectively. Moreover, the combined model demonstrated superior predictive performance compared to either model alone. This overall model included two significant clinical predictors (D-dimer and CA19-9) and three radiological predictors (lymph node enlargement, pancreatic atrophy, and cystic components). It yielded an AUC of 0.92 (95% CI: 0.88-0.95), with a sensitivity of 83.5% and specificity of 82.2%. In addition, the optimal cut-off value of D-dimer for differentiating PDAC from benign pancreatic lesions was found to be 0.84 mg/L. Conclusions: The overall model including clinical and radiologic variables (e.g., serum D-dimer, CA19-9, lymph node enlargement, pancreatic atrophy, and cystic components) demonstrated higher sensitivity and specificity in differentiating PDAC from benign pancreatic lesions. Serum D-dimer may serve as a valuable adjunctive biomarker in the diagnosis of pancreatic cancer and may further enhance the diagnostic performance of CA19-9 when used in combination.