Construction of the Predictive Model for Lymph Node Metastasis in Patients with Epithelial Ovarian Cancer: Based on 18F-FDG PET/CT Radiomics Technology

Background Accurate preoperative assessment of lymph node metastasis in patients with epithelial ovarian cancer is of great significance for formulating treatment plans and assessing prognosis. Radiomics technology has been used as a non-invasive means to predict lymph node metastasis in various cancers before surgery, but there are few research reports on its application effects in the field of gynecological cancer. Objective To construct a predictive model for lymph node metastasis in patients with epithelial ovarian cancer using radiomics technology based on 18F-fluorodeoxyglucose positron emission tomography with integrated computed tomography (18F-FDG PET/CT) . Methods A total of 98 patients with epithelial ovarian cancer admitted to the Department of Gynecology, the Second Xiangya Hospital of Central South University from September 2020 to December 2022 were selected, according to the lymph node metastasis status, they were divided into lymph node metastasis group (n=65) and non-lymph node metastasis group (n=33), into training set of 68 cases and validation set of 30 cases in a 7∶3 ratio at the same time. All patients' clinical characteristics were analyzed, and the lymph node metastasis status was used as the result label for model construction. Results The lymph node metastasis rate in patients with epithelial ovarian cancer was 66.3% (65/98) in this study. There was a statistically significant difference in the level of human epididymal protein 4 (HE4) and location of the primary lesion between lymph node metastasis group and non-lymph node metastasis group (P<0.05). Multivariate Logistic regression analysis showed that, HE4 level, location of the primary lesion, and Radscore were predictive factors for lymph node metastasis in patients with epithelial ovarian cancer (P<0.05). The clinical predictive model was constructed using HE4 level and location of the primary lesion, the combined predictive model was constructed using HE4 levels, location of the primary lesion, and Radscore. Delong's test showed that, the combined predictive model had an AUC of 0.80 (95%CI=0.70-0.90) for predicting lymph node metastasis in the training set of patients with epithelial ovarian cancer, which was higher than that of the clinical predictive model (AUC=0.73, 95%CI=0.61-0.85) (P=0.042). The calibration curve showed that, the combined predictive model passed the calibration test (P=0.990), had good discrimination ability. Decision curve (DCA) analysis showed that, both the clinical and combined predictive models had good predictive performance, but the net benefit of the combined predictive model was higher. Conclusion We successfully constructed the combined predictive model for lymph node metastasis in patients with epithelial ovarian cancer using radiomics technology based on 18F-FDG PET/CT, with high robustness, good discrimination ability, and relatively high net benefit, which can be used for clinical doctors to formulate individualized treatment plans for patients and assess patients' prognosis.