Abstract 1955: Blood-based mapping of the personalized tumor epigenomic landscape

Abstract Background: Mapping the epigenetic landscape of tumors provides insights into regulatory and functional aspects beyond somatic mutations, offering valuable information that can influence clinical decisions and benefit the patient journey. Given the challenges of obtaining tissue biopsies, especially in the longitudinal monitoring or progression setting as tumors evolve due to selective pressure from treatment, the advantages of non-invasive liquid biopsies to characterize tumors through blood are highly beneficial. Here, we demonstrate that the epigenetic landscape of a patient’s tumor can be evaluated in plasma using an epigenetics circulating tumor DNA assay, providing a promising approach for non-invasive tumor profiling. Methods: We collected paired tumor and plasma samples from lung, breast and colorectal cancer patients, and applied an analytically validated next-generation sequencing epigenomics hybrid capture assay to measure DNA methylation across over 1.3 million CpG loci. A total of 44 patients (10 colorectal, 13 breast, 21 lung) with circulating tumor DNA (ctDNA) detected at a level of 0.5% or greater were included in the analysis. For each patient, we identified personalized tissue differentially methylated regions (DMRs) as the set of regions that were highly methylated in tissue but showed low methylation in an independent set of 197 cancer-free plasma samples. For each region from a personalized tissue DMR list, we computed the expected count in plasma as a function of tumor fraction and assay efficiency. Results: Across the 44 analyzed tissue samples, the median number of personalized tissue DMRs identified was 1, 133, with a range of 136 to 4, 220 and a standard deviation of 850. For paired tissue and plasma samples from the same patient the Spearman correlation between the expected and observed counts in plasma was high across all samples (Mean=0.687, SD=0.120). The correlation between expected and observed signals at personalized tissue DMRs was calculated for all possible plasma and tissue sample pairs, regardless of whether the samples were from the same patient. Notably, for 43 out of 44 plasma samples, the highest correlation between expected and observed plasma counts occurred with the plasma’s matched tissue sample (Mean difference=0.196, SD=0.096). Conclusions: These data demonstrate a strong concordance between a tumor tissue's unique epigenetic landscape and the epigenetic signal in plasma. Furthermore, the finding that each plasma sample correlated most with its paired tissue sample over other pairings indicates that the assay can reliably measure personalized tumor epigenetic features in plasma. This supports the use of an epigenetics-only ctDNA assay for monitoring patients and guiding treatment decisions. Citation Format: Elena Zotenko, Rachel Gittelman, Alan Selewa, Noam Vardi, Tam Banh, Gleb Martovetsky, Oscar Westesson, Emily Tsang, Sara Wienke, Drew Kennedy, Meromit Singer, Stefanie Mortimer, Darya Chudova. Blood-based mapping of the personalized tumor epigenomic landscape [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 1955.