Abstract A002: In-silico screening of race-specific endometrial cancer methylation markers using machine learning techniques

Abstract

Abstract Targeted molecular therapy is a promising strategy for cancer treatment. Despite the benefits of these therapies, racial disparities in cancer therapy efficacy persist. To close this racial gap, there is a need to find race-specific biomarkers at the molecular level. Since abnormal DNA methylation has been linked with cancer and has a great potential to transform the diagnosis and treatment of cancer, the main objective of this study was to combine differential DNA methylation and machine learning techniques to identify race-specific methylation signature biomarkers in endometrial tumor samples from White and Black women. We performed genome wide Differentially Methylated CpG (DMCs) analysis in tumors from White patients (TWP) samples compared to tumors from Black patients (TBP) samples using Endometrial Carcinoma (EC) methylation and clinical data from The Cancer Genome Atlas (TCGA). We used supervised and unsupervised machine learning techniques to filter and select signature biomarkers. To evaluate the prognostic capacity of these genes, we performed univariate and multivariate survival analyses using Cox proportional hazards regression model from survival R package. By combining differential methylation CpGs and machine learning techniques, we reduced 704 differentially methylated CpGs to reliable race-specific epigenetic signature genes (ESGs) and Core- epigenetic signature genes (core-ESGs). An area under the curve (AUC) of 0.98 was achieved based on these signature genes. We further found that ESGs showed statistically significant overall survival differences between two racial groups. In addition, comparison of the mutational landscape showed differences in mutation frequencies in the ESGs between the two groups. By combining differential methylation analysis and machine learning techniques, we identified novel race-specific epigenetic signature genes in Black and White women with endometrial cancer. We further showed the potential of these gene in predicting survival differences, mutation, and transcriptomic differences in Black and White women. We propose that combined methylation and somatic mutation profiling of these novel race-specific signature genes can be used to develop potential therapeutic targets for diagnosis and treating endometrial cancer. Citation Format: Huma Asif, Grace Foley, Julie Kim. In-silico screening of race-specific endometrial cancer methylation markers using machine learning techniques [abstract]. In: Proceedings of the AACR Special Conference on Endometrial Cancer: Transforming Care through Science; 2023 Nov 16-18; Boston, Massachusetts. Philadelphia (PA): AACR; Clin Cancer Res 2024;30(5_Suppl):Abstract nr A002.