Abstract 981: Robustness of fragmentation-based cell-free DNA approaches to clonal hematopoiesis

Abstract

Abstract INTRODUCTION: Clonal hematopoiesis (CH) is the result of the clonal expansion of hematopoietic stem cells which acquire and pass on somatically gained mutations, including SNVs as well as arm level losses and gains, defined here as mosaic deletions and amplifications (mDAs). The proportion of blood cells in circulation containing CH alterations increases with age and environmental exposures such as smoking, leading to a higher prevalence in many high-risk cancer screening populations. CH remains a major confounder for mutation-based liquid biopsy tests evaluating circulating cell-free DNA (cfDNA), since these assays are unable to discriminate between mutations originating from the tumor as opposed to CH. While the impact of CH mutations on fragmentation-based methods of detecting ctDNA appears to be limited, the effect of mDAs has not yet been explored. METHODS: To assess the impact of mDAs on screening efforts using whole genome sequencing (WGS) of cfDNA, we performed copy number analyses on cfDNA from plasma collected from 895 individuals consisting of high-risk individuals meeting the USPSTF guidelines for annual lung cancer screening using low-dose computed tomography. From a subset of 203 available donors (157 without cancer and 46 with lung cancer), matched genomic DNA from white blood cells (WBCs) were extracted from the isolated buffy coat and WGS was performed to evaluate whether ploidy in cfDNA plasma is derived from WBCs. Genome-wide fragmentation features (short to long ratios and fragment size densities), coverage-based statistics, and copy number (CN) statistics adjusting for mDAs were calculated (CNadj). Machine learning models to calculate the probability of cancer based on these features were then derived. RESULTS: 3.8% (95% CI 0.9%-6.8%) of individuals without cancer had an mDA in their WBCs that were also observed in the plasma (p < 0.01, permutation test). Fragmentation features were not different in regions with mDAs in individuals without cancer (p = 0.619; permutation test) but were altered in regions of tumor-derived chromosomal changes in patients with cancer (p < 0.01; permutation test). Coverage-based statistics were impacted by mDAs (p < 0.01; Wilcoxon test). Machine learning models fit with CNadj showed improved performance over CN. However, machine learning models fit with fragmentation features compensated for this difference overall showing robustness to mDAs (p = 0.45; bootstrap) demonstrating the robustness of cfDNA fragmentomes for detecting lung cancer. CONCLUSIONS: While coverage-based statistics of WGS analyses of cfDNA may be affected by mDAs, fragmentation-based approaches relying on cfDNA fragment length are robust to effects of CH. CNadj is an improved statistic for detecting tumor-derived chromosomal copy number changes in cfDNA, providing increased specificity by conditioning on fragment size distributions. Citation Format: Stephen Cristiano, Bahar Alipanahi, Jamie E. Medina, Lorenzo Rinaldi, Nicholas Dracopoli, Robert B. Scharpf, Alessandro Leal, Victor E. Velculescu, Jennifer Tom. Robustness of fragmentation-based cell-free DNA approaches to clonal hematopoiesis [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 981.