Abstract 3712: A computational method for quantifying circulating tumor DNA (ctDNA) in plasma using methylation profiling data

Abstract Background: Abnormal DNA methylation is a hallmark of tumorigenesis and cancer development, where patients exhibit distinct methylation patterns compared to healthy individuals, referred to as tumor-specific methylation haplotypes. These haplotype markers are widely used for cancer screening and diagnosis. Accurate quantification of ctDNA in cell-free DNA (cfDNA) is important in evaluating the analytical performance of clinical assays. Conventional methods generally employ 120-nt DNA probes to capture both the Watson and Crick strands, assuming that CpGs are either fully methylated or fully unmethylated, and therefore are unbiased. With sufficient sequencing depth, ctDNA quantification can be performed by calculating the proportion of tumor-related methylation haplotype reads. However, in enrichment-capture methods, where DNA probes are designed to specifically target tumor-related signals, quantification of ctDNA can be challenging due to varying enrichment efficiencies across numerous probes. Here, we present a novel computational method for precise quantification of ctDNA in plasma based on methylation profiling data, where tumor-related methylation signals were enriched experimentally. Methods: We constructed an Enrichment Factor Model (EFM) based on the thermodynamic characteristics of probes that target methylation haplotypes from a panel. The enrichment factor is determined by comparing the true values and observed values generated from multiple reference samples with different haplotype frequencies captured by that panel. This enrichment factor remains stable when applied to unknown samples captured by the same panel, and therefore the model can be used to predict the ctDNA amount in any unknown clinical samples. Results: 16 reference samples with a variety of haplotype frequencies were generated by mixing the 0% and 100% HCT116 at different ratios. Using 12 reference samples, we were able to calculate the enrichment factor. The remaining 4 samples were used to validate the EFM, accurately predicting their values with an R2 of 0.99 and a mean squared error (MSE) of 0.0011. We further applied the model to 59 colorectal cancer (CRC) cases and 103 healthy control clinical plasma samples. The estimated ctDNA fraction closely aligned with clinical tumor stages, with higher fractions consistently observed in late-stage patients. Additionally, a clear distinction was observed between cases and controls with a p-val of 4.73 E-6. Conclusion: We developed a computational method to accurately quantify ctDNA fractions using methylation profiling data, specifically optimized for experimental methods that enrich tumor-related signals. Our method offers a reliable tool for ctDNA quantification in cancer detection and monitoring applications. Citation Format: Ping Lan, Xiaowen He, Ying Mei, Yangjunyi Li, Nina Guanyi Xie, Liyuan Zhao, Zhihui Xu, Chen Zhu, Yujie Chen, Min Li, Baoliang Zhu, Xiaohui Wu, Xueguang Sun. A computational method for quantifying circulating tumor DNA (ctDNA) in plasma using methylation profiling data [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 3712.