Abstract 3666: Multi-cancer early detection using metrics of DNA methylation based epigenetic instability

Abstract Cancers present significant changes in DNA methylation, which has proven to be highly useful in cell-free DNA (cfDNA) based cancer detection. Cancer epigenomes are marked by a high degree of intra- and inter-tumor epigenetic variation, indicative of the high level of epigenetic instability. However, the nature of the regions with high epigenetic instability and its utility as biomarker has not been explored. Here we developed a novel methodology and metric to measure the degree of epigenetic perturbation, termed the Epigenetic Instability Index (EII), for multiple cancer screening through cfDNA. Our novel methodology provides a sample specific score of epigenetic perturbation in relation to the expected normal state as opposed to only looking at CpG sites that change between samples. Through machine learning, we have elucidated 269 CpG-island regions which sufficiently capture the epigenetic instability in cancers. We have built classifier models using the EII metrics of these 269 regions and demonstrate that they can efficiently identify breast and lung cancer cases from cfDNA methylation data. Particularly, the models can differentiate even Stage IA of NSCLC with ~75% sensitivity at 95% specificity and early-stage breast cancer at ~68% sensitivity and 95% specificity. The EII metrics perform significantly better than metrics that measure absolute DNA methylation levels. Our studies highlight the potential of measuring epigenetic instability from cfDNA in any given sample in addition to traditional methods of methylation measurement for non-invasive screenings via liquid biopsies. Citation Format: Sara-Jayne Thursby, Zhicheng Jin, Steve B. Baylin, Malcolm Brock, Thomas Pisanic, Hariharan Easwaran. Multi-cancer early detection using metrics of DNA methylation based epigenetic instability [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 3666.