[Feasibility of identifying POLE-mutated endometrial carcinoma through a pathological features-based scoring system]

Objective: To assess the utility of histopathological and immunohistochemical characteristics in identifying POLE-mutated (POLEmut) endometrial carcinoma (EC). Methods: A total of 1 541 EC cases that underwent molecular classification at Peking University Third Hospital from January 2018 to December 2024 were included. Cases were categorized into POLEmut and non-POLEmut groups (including p53-abnormal, mismatch repair-deficient, and no specific molecular profile subtypes). A comparative analysis of histopathological features, mismatch repair (MMR) protein expression, and p53 immunostaining patterns was performed. A multivariable logistic regression-derived prediction model for POLEmut status was developed and internally validated. Results: POLEmut tumors showed significantly higher frequencies of prominent peritumoral lymphocytes (PTLs) and tumor-infiltrating lymphocytes (TILs), high-grade nuclei, bizarre nuclei, clear cytoplasm, and ambiguous morphology compared with non-POLEmut tumors (all P<0.05). Multivariable analysis identified severe PTLs/TILs infiltration, high-grade nuclei, ambiguous morphology, and clear cytoplasm as independent positive correlates for POLE mutation status, while aberrant p53 expression and MMR protein loss were negative correlates. The scoring system showed robust discrimination, with an area under the curve (AUC) of 0.867 in the training set (n=1 319) and 0.873 in the test set (n=222). At the ≥3- point cutoff, it provided a sensitivity of 80.8%, specificity of 66.3%, and negative predictive value (NPV) of 96.3% in the test set. Calibration analysis indicated good overall performance (Brier score=0.045), though local miscalibration was observed in intermediate-and low-risk subgroups. Conclusions: This POLEmut scoring system achieves high sensitivity and high NPV for initial screening of POLEmut EC. Despite a relatively low positive predictive value requiring confirmatory sequencing, it effectively enriches candidates needing POLE gene sequencing in resource-limited settings. Future multicenter validation is warranted to assess its clinical utility.