Abstract 4111: Whole-genome methylation profiling enhances cancer detection and tumor fraction quantification in liquid biopsies

Abstract Circulating tumor DNA (ctDNA) is a critical biomarker for non-invasive cancer diagnosis, prognosis, and therapeutic monitoring. However, accurately quantifying ctDNA content in liquid biopsy samples remains challenging due to its low abundance and interference from non-tumor DNA. Current approaches, such as targeted sequencing and mutation-based quantification, are limited by their dependence on known mutations and reduced sensitivity to low ctDNA fractions. Next-generation sequencing (NGS)-based methylation profiling offers a promising alternative by leveraging the distinct methylation patterns between tumor-derived and normal cell-free DNA (cfDNA). We developed a robust algorithm for the whole-genome methylation sequencing assay, PredicineEPICTM, as a multi-cancer detection solution enabling the quantification of ctDNA content in blood, urine, and cerebrospinal fluid samples. This approach focuses on comprehensive cancer-specific methylation markers to distinguish ctDNA fragments from background cfDNA molecules. We characterized methylation profiles of 800 simulated liquid biopsy samples with titrated tumor fractions (0.01%-50%) and applied machine learning models for analysis to evaluate ctDNA amounts. An independent set of 93 real-world samples from patients with lung, prostate, colorectal, ovarian, and bladder cancers was collected to validate the accuracy of the model. Performance was benchmarked against mutation panel-based reference tumor fractions. In the simulated dataset, our method achieved high accuracy (R2 = 0.92) and sensitivity in detecting ctDNA at tumor fractions as low as 0.05%. The tumor fractions estimated by this approach correlated strongly with variant allele frequency (p-value < 0.01). Compared to conventional mutation-based methods, our approach achieved a higher detection rate in liquid samples with low tumor burden. Notably, the whole-genome strategy offered deeper insights into epigenetic alterations and fragmentomics features associated with gene expression activity, which remain undetectable by mutation-based assays. This study highlights the potential of whole-genome methylation profiling as a highly precise and comprehensive tool for tumor fraction quantification in liquid biopsy samples. The demonstrated accuracy and clinical relevance underscore its applicability for early cancer detection and real-time disease monitoring. Citation Format: Hang Dong, Shuang Gan, Haoran Tang, Yue Zhang, Feng Xie. Whole-genome methylation profiling enhances cancer detection and tumor fraction quantification in liquid biopsies [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 4111.