LAUSR

There is a high rate of inconformity between clinical staging and surgical-pathologic staging in endometrial cancer. Many patients with advanced endometrial cancer are preoperatively understaged and thereby do not receive the optimal therapy. Here, we aimed to develop a predictive model or biomarker for preoperative diagnosis of advanced endometrial cancer via multivariate logistic regression analysis. In this study, 259 eligible patients were included, and 195 patients were assigned to the training dataset and 64 patients to validation dataset. Age, menopause status, sterilization situation, parity, body mass index, hypertension, diabetes mellitus, tumor size, and ovarian malignancy algorithm (ROMA) index were included as predictive variables, and the binary outcome was advanced endometrial cancer or not. When the P value was set as less than 0.01 in forward stepwise regression, only ROMA index was retained. The odds ratio of being positive ROMA index was 15.531 times that of negative value. The area under receiver operating characteristic curve was 0.790 in the training dataset and 0.776 in the validation dataset. The decision curve analysis curve showed that the prediction by ROMA index added more net benefits for almost all threshold probabilities. Therefore, ROMA index is an effective predictor for advanced endometrial cancer before surgery. Since ROMA index is a standard, measurable, and reliable laboratory test, it can be used as a reference tool for gynecologists to design the appropriate therapeutic schedule for patients with high-stage endometrial cancer before surgery.