Prognosis and Treatment of Malignant Brenner Tumor: A Machine Learning Approach

Background: Malignant Brenner Tumor (MBT) is a rare subtype of epithelial ovarian tumors. It is characterized by a challenging diagnosis and variable prognosis. Our aim was to assess the influence of various treatment approaches on survival and to identify predictive factors using machine learning (ML) techniques to enhance the understanding of MBT. Methods: This study used the Surveillance, Epidemiology, and End Results (SEER) database, covering the years 2000-2020, for its analysis. To determine prognostic factors, we employed Cox regression analysis and developed predictive models for 5-year survival using five ML algorithms. The accuracy and reliability of these ML models were assessed using a validation approach that incorporated the area under the curve (AUC) of the receiver operating characteristic (ROC) curve. Additionally, we conducted a Kaplan-Meier survival analysis as part of our investigation. Results: The study population comprised 221 patients, with a median age of 64 years. Among the patients, 64.7% had tumors > 10 cm. Multivariate Cox regression analysis showed that unmarried status (hazard ratio [HR]: 2.4, 95% confidence interval [CI]: 1.3–4.3) and the presence of metastasis (HR: 4.4, 95% CI: 2.3–8.6) were significant poor prognostic factors. Random Forest Classifier achieved the highest AUC of 0.8280. Metastasis has emerged as the most important factor for prognosis, highlighting the need for tailored approaches based on individual risk profiles in patients with MBT. Conclusion: Our study highlights metastasis as the most important prognostic factor in MBT, with chemotherapy showing a limited impact on survival. We developed a novel personalized ML tool to predict survival and support tailored treatment decisions.