Abstract 3454: Cell-free DNA methylation and fragmentomic signatures identify tissue damage and predict cancer risk

Nicholas Cheng,Kimberly Skead,Tom Oullette,Scott Bratman,Daniel De Carvalho,David Soave,Philip Awadalla

DOI: https://doi.org/10.1158/1538-7445.am2024-3454

IF: 11.2

2024-03-23

Cancer Research

Abstract:Cell-free DNA (cfDNA) epigenetic and fragmentomic profiling has emerged as prominent non-invasive approaches for early cancer detection and subtyping. However, owing to difficulties in observing the early development of human malignancies, most cancer biomarker and evolution studies to date have primarily examined biologics following a diagnosis. Utilizing cfDNA as a screening tool for early disease detection requires assessment of blood plasma samples collected from asymptomatic individuals prior to the diagnosis of cancers. Here, we leverage the Ontario Health Study (OHS), a prospective longitudinal population cohort, to assess cfDNA profiles in blood plasma collected from participants that developed breast (n=476), prostate (n=342) and pancreatic (n=32) cancers throughout the study follow up time between 2009 and 2019. We performed cell-free methylated DNA immunoprecipitation and high-throughput sequencing (cfMeDIP-seq) on baseline plasma samples from incident cancer cases up to eight years prior to diagnosis, in addition to matched controls (n=404) with no history of cancer through follow-up. Genome-wide plasma cfDNA differential methylation analysis revealed significantly hypermethylated regions, particularly among CpG island and shore regions, were predictive of early breast cancers, and overlapped hypermethylated regions in solid cancer tissues relative to adjacent normal tissues and peripheral blood leukocytes. Using a repeated cross-validation strategy we found that as few as 200 hypermethylated regions can identify individuals with breast cancer in blood up to seven years prior to diagnosis among discovery set samples (AUC=0.725) and are highly generalizable to samples processed in separate batches (AUC = 0.650). Remarkably, integrating biomarker hypermethylated regions with fragmentomic features including fragment length and 5' tetranucleotide motif proportions further increased accuracy for detecting individuals with early prostate cancers (AUC=0.804). In addition, we observed shared genome-wide cfDNA methylome changes associated with aging and heavy alcohol consumption in cfDNA. We identified increased shedding of tissue-specific methylation markers from liver and pancreatic tissue among individuals consuming more than three drinks a week relative to non-drinkers, reflective of potential signatures of tissue damage. Citation Format: Nicholas Cheng, Kimberly Skead, Tom Oullette, Scott Bratman, Daniel De Carvalho, David Soave, Philip Awadalla. Cell-free DNA methylation and fragmentomic signatures identify tissue damage and predict cancer risk [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular s); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl) nr 3454.

oncology