Utilizing personalized tumor-specific methylation haplotypes of circulating tumor DNA for monitoring minimal residual disease in hepatocellular carcinoma patients after curative resection

BackgroundDetection of minimal residual disease (MRD) post-treatment is essential for timely intervention to prevent recurrence. We hypothesized that DNA methylation biomarkers, commonly used in cancer early detection, could monitor MRD and dynamically track tumor burden changes in blood. Thus, we are investigating their feasibility as prognostic indicators for MRD and compare different strategies head-to-head.MethodsWe aimed to prospectively enroll 200 hepatocellular carcinoma (HCC) patients undergoing radical surgery. With two-year post-operative surveillance, tumor, para-tumoral tissue and a series of blood samples were collected. MRD status in samples was measured by GutSeer assay, a targeted DNA methylation sequencing assay for early detection of gastrointestinal cancers. We applied two approaches to detect MRD: the tumor-naïve approach directly used the GutSeer model, while the tumor-informed approach, TORNADO, was developed based on individualized tumor-specific methylation haplotypes.ResultsTo date, 73 early-stage HCC patients who underwent surgical resection were recruited. Within a median follow-up of 11.77 months, 19 patients experienced relapse (26%). One month after surgery, 19.18% to 6.9% patients remained positive detected by TORNADO and GutSeer, respectively. It indicates the sensitivity of GutSeer assay in tracking dynamic change of tumor burden and detecting MRD. Cox regression analyses showed that MRD positive status significantly correlates with recurrence in both approaches (p < 0.01). Furthermore, TORNADO model demonstrated superior recurrence risk stratification compared to the tumor-naïve approach (p < 0.05, Delong’s test).ConclusionsIn this pilot study, we demonstrated that ctDNA methylation serves as a significant prognostic factor of MRD status and disease relapse for early-stage HCC patients after curative resection. The tumor-informed method demonstrated enhanced performance compared to the tumor-naïve approach.Clinical trial identificationNCT06178809.Legal entity responsible for the studyThe authors.FundingHas not received any funding.DisclosureM. Xu, H. Chen, C. Ma, M. Su, Q. He, Y. Zhang, Z. Su: Financial Interests, Personal, Full or part-time Employment: Singlera Genomics Inc. R. Liu: Financial Interests, Personal, Leadership Role: Singlera Genomics Inc. All other authors have declared no conflicts of interest.