Abstract 3244: <i>cfTrack-methyl</i>: A personalized approach using cfDNA methylomes for ultra-sensitive MRD detection

Abstract Methylation alterations are pervasive in cancer, making cfDNA methylomes a promising tool for cancer detection and minimal residual disease (MRD) monitoring. However, cfDNA methylation-based MRD detection faces unique challenges. Unlike genomic mutations, methylation signals are prone to noise from non-tumor-related factors, such as heterogeneous cell types and environmental influences. This noise introduces confounding signals and dilutes tumor-specific methylation markers. While patient-specific tumor methylation profiles offer personalized insights, they often contain inherent noise that compromises their performance. Existing approaches failed to address this challenge, as they either (1) disregard personal tumor methylation profiles, yielding suboptimal performance, or (2) fail to account for noise, treating all personal markers as equally informative. These limitations underscore a crucial need for a novel mca and evaluates their prevalence across population cancer methylome data. Population prevalence quantifies the cancer relevance of noisy personal markers, filtering out non-tumor-related signals and prioritizing true tumor markers. With these purified personal markers, cfTrack-methyl achieves sensitive and specific MRD detection, even at ultra-low tumor fractions. In the titration series between fragmented liver tumor and WBC gDNA, the detection limit of cfTrack-methyl was calculated as &amp;lt; 0.01% with 100% sensitivity and 100% specificity. In simulated data with plasma spike-ins from various cancer types, cfTrack-methyl achieved a median AUC of 99.6% at a tumor fraction of 0.01% when tumor tissue was informed and maintained robust performance with a median AUC = 100.0% at a tumor fraction of 0.01% when tumor tissue was unavailable and pre-treatment plasma was used. Further, in liver cancer patients with complex comorbidities (cirrhosis or HBV), cfTrack-methyl achieved comparable detection limits, with a median AUC = 99.9% at a tumor fraction of 0.01%. In all scenarios, cfTrack-methyl outperformed the MRD detection using general cancer markers. In conclusion, cfTrack-methyl delivers unprecedented sensitivity for MRD detection in cfDNA at low tumor fractions across diverse clinical scenarios. By integrating personal and population data, this approach maximizes the potential of cfDNA methylomes for ultra-sensitive MRD detection. Ongoing studies using serial samples from patients with surgically resected tumors aim to further validate this personalized strategy. Citation Format: Li Shuo, Weihua Zeng, Xiaohui Ni, Chun-Chi Liu, Yonggang Zhou, Mary L. Stackpole, Wenyuan Li, Arjan Gower, Kostyantyn Krysan, Andrew Melehy, Angela H. Yeh, Megumi Yokomizo, Preeti Ahuja, Lopa Mishra, Kirti Shetty, Yazhen Zhu, Hsian-Rong Tseng, Clara Lajonchere, Daniel H. Geschwind, David S. Lu, Steven S. Raman, William Hsu, Clara E. Magyar, Samuel W. French, Denise R. Aberle, Steven-Huy B. Han, Edward B. Garon, Vatche G. Agopian, Wing H. Wong, Steven M. Dubinett, Xianghong Jasmine Zhou. cfTrack-methyl: A personalized approach using cfDNA methylomes for ultra-sensitive MRD detection [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 3244.