Abstract 7018: Application of HiTIMED and cell type adjusted epigenome-wide models identifies driver tumor DNA methylation alterations in multiple cancers

Abstract DNA methylation (DNAm) is an epigenetic mark crucial in lineage specification and cellular differentiation, resulting in distinct patterns by cell type. These patterns can serve as biomarkers in deconvolution approaches inferring cell proportions from composite DNAm signals in biospecimens like blood and tumor tissue. Adjusting for cellular heterogeneity is critical to avoid confounding by cell type. The HiTIMED (hierarchical tumor immune microenvironment epigenetic deconvolution) algorithm provides high-resolution tumor microenvironment profiling, up to 17 cell types, including tumor, myeloid, and angiogenic compartments. We analyzed bladder, prostate, and uterine cancers, applying HiTIMED on the tumor (n=1, 341) and normal adjacent tissue (n=105), adjusting for age, sex, and cell types. DNAm data from The Cancer Genome Atlas (TCGA) was used, and HiTIMED was applied to deconvolute various cell proportions. Linear model comparisons to identify significantly differentiated loci with/without cell type adjustment are summarized in Table 1. We demonstrate greater than 99% reduction in both hypermethylated and hypomethylated loci in all cancer types after adjustment for cell type. Our study demonstrates HiTIMED's applicability to deconvolve the tumor microenvironment in multiple cancer types and demonstrates the importance of adjusting for cell type in epigenome-wide association studies to identify differentially methylated loci, offering greater biological insight into the etiology of specific cancer types. Significant differentially methylated loci in tumor versus normal adjacent tissue, FDR 0. 05 Cancer type Linear model Hypermethylated in tumor (logFC 2. 0) Hypomethylated in tumor (logFC -2. 0) Bladder cancer n = 21, normal adjacent tissue n = 413 tumor Model 1 (Cancer Status, Age, Sex) 4495 19, 936 Model 2: (Cancer Status, Sex, age, six cell types) 27 39 Prostate adenocarcinoman = 50, normal adjacent tissuen = 499 tumor Model 1 (Cancer Status, Age) 8988 2207 Model 2: (Cancer Status, age, six cell types) 2 4 Uterine carcinoman = 34 normal adjacent tissuen = 429 tumor Model 1 (Cancer Status, Age) 11, 698 16, 249 Model 2: (Cancer Status, age, six cell types) 250 8245 Citation Format: Irma M. Vlasac, Ze Zhang, Lucas A. Salas, Brock C. Christensen. Application of HiTIMED and cell type adjusted epigenome-wide models identifies driver tumor DNA methylation alterations in multiple cancers 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 (6Suppl): Abstract nr 7018.