Utilization of personalized tumor-specific methylation haplotypes of circulating DNA to monitor minimal residual disease in patients with hepatocellular carcinoma undergoing curative resection.

Abstract

e16245 Background: Minimal residual disease (MRD) is considered an essential factor leading to early relapse after radical surgery, which is challenging to be detected by conventional imaging. The majority of MRD studies in HCC are based on mutation detection. Simultaneously, in certain cancers such as lung and colorectal cancers, the use of methylation detection for MRD has been demonstrated to possess significant detection capabilities. In this study, we present a personalized methylation haplotype (MHP) method for MRD detection in HCC patients underwent curative resection. Methods: Tumor and paracancerous tissue samples were obtained from HCC patients undergoing surgical resections. WBC and plasma samples (T0) were collected before surgery, while plasma samples (T1) were collected in one month after surgery, and subsequent follow-up plasma samples were obtained every three months. All samples were tested using the GutSeer Panel. A personalized tumor-informed MRD detection approach, named TORNADO (Tumor-infORmed DNAm hAplotype DetectiOn), was then developed. Individual seed MHPs were pinpointed based on their presence in cancer tissue or plasma samples collected before surgery (T0 plasma), excluding those in para-cancer tissue and blood cells. The relapse risk score was determined by assessing the proportion of these MHPs detectable in follow-up evaluations. Results: We enrolled 39 early-stage HCC patients who have undergone surgical resections. All the patients have completed at least one year of regular clinical follow-up. Among them, 13 out of 39 patients (33.3%) were identified as MRD-positive based on predictions from T1. Of the MRD-positive patients, 6 experienced tumor relapse. In contrast, only 7.7% relapses were observed among the remaining 26 MRD-negative patients. When considering the T1 and T2 MRD cumulative risk score, 53.8% (21/39) of patients were classified as MRD-positive, and none of the MRD-negative patients experienced relapse. Regardless of whether based on T1 plasma alone or cumulative results, recurrence-free survival (RFS) was significantly correlated with MRD status. Kaplan-Meier analysis demonstrated that MRD-positive patients had significantly poorer RFS (p-value < 0.05). Cox regression analysis further revealed that MRD status was a significant variable for predicting recurrence-free survival (HR=16.00, 95% CI 2-130, P=0.009). Conclusions: The MRD status detected byTORNADOmethod emerged as a significant prognostic indicator for disease relapse in early-stage HCC patients undergoing curative resection, underscoring the clinical relevance of MHP-based MRD detection guided by tumor information. These discoveries pave the way for the development of effective postoperative cancer surveillance strategies specifically tailored for early-stage liver cancer.