Abstract 5892: Optimized enzymatic conversion method for improved methylation detection in low-input cfDNA samples

Background: Cell-free DNA (cfDNA) is widely utilized in liquid biopsy-based early cancer screening, where DNA methylation serves as a predominant detection marker via next-generation sequencing. Enzymatic Methyl-Seq (EM-Seq) preserve cfDNA integrity and significantly improved methylation marker detection. However, with its expanded application, several limitations have been identified, including insufficient protection of 5-methylcytosine (5mC) in samples below 10 ng, leading to diminished methylation signals and inconsistencies across different input amounts. Additionally, uneven CH conversion results in background noise akin to bisulfite methods. These issues are critical as clinical cfDNA samples of early-stage cancer often fall below the 10 ng threshold, indicating great room for improvement in enzymatic conversion. Methods: To enhance the compatibility of enzymatic conversion methods with low-input cfDNA samples, we have made a series of improvements, including refining oxidation and deamination reaction systems. We evaluated these optimizations by testing on cfDNA from healthy donors and two cancer patients (colorectal cancer, CRC and gastric cancer, STAD). Results: The optimized enzymatic conversion method significantly improved library quality, increasing average mapped ratios by 4.24% and 7.53% for healthy samples and for cancer samples, with library complexity rising from 38.04% to 75.52% consequently. The average CpG sites protection efficiency at 1 ng input is nearly equivalent to that at 30 ng input (98.11% vs. 99.27%), and shows a notable improvement over the original method (96.49% vs. 99.11%). The CH non-conversion rate decrease from 0.13% to 0.065%, significantly reducing the background noise. Consequently, more CpG sites are captured with equivalent sequencing depths. Samples with inputs ranging from 1 to 30 ng showed greater consistency in methylation beta values, with an average Pearson's correlation of 0.93, compared to 0.87 for the original method. Sensitivity analysis indicated that optimized method detected more methylation haplotypes in artificially pooled samples with varying cancer cell spike-in ratios: 589 vs. 379 at 0.05%, 935 vs. 502 at 0.1%, 3,326 vs. 1,236 at 0.5%, and 6,318 vs. 2,126 at 1%, enabling detection at lower cancer cell abundances. Using healthy samples as a reference, the optimized method consistently identified more differential haplotypes in CRC and STAD samples at 1, 5, and 30 ng cfDNA inputs, with the disparity between methods decreasing as input increased, confirming its advantages in low-input conditions. Conclusion: Our optimized enzymatic conversion method provides superior library quality, increased complexity, reduced CH background and enhanced 5mC protection in low-input samples. These improvements enhance the sensitivity for detecting aberrant methylation signals in low-input cfDNA. Citation Format: Hong Xu, Quan Wang, Nan Zhang, Zhuoran Jiang, Lei Song, Ming Yang, Jing Zhao, Yue An, Congchong Wei, Yuting Liu, Feng Xu, Min Li, Jing Liu, Baoliang Zhu, Xiaohui Wu, Xueguang Sun. Optimized enzymatic conversion method for improved methylation detection in low-input cfDNA samples [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 5892.