Genome-wide 5-hydroxymethylcytosine in circulating cell-free DNA reflects DNA methylation dynamics for non-invasive detection of early hepatocellular carcinoma

BACKGROUND: Early hepatocellular carcinoma (HCC) remains difficult to diagnose because of its insidious and asymptomatic onset. Liquid biopsy provides a promising non-invasive approach for early tumor detection. 5-Hydroxymethylcytosine (5-hmC), the oxidation product of 5-methylcytosine (5-mC), has emerged as an informative epigenetic mark reflecting tumor-associated alterations during hepatocarcinogenesis. This study aimed to identify and validate a circulating cell-free DNA (cfDNA) 5-hmC-based signature for the early detection of HCC and to evaluate its diagnostic and prognostic potential. METHODS: Using an XGBoost-based ensemble learning framework, we identified cfDNA 5-hmC markers that defined an early HCC 5-hmC signature. The signature was assessed in training and validation cohorts using receiver operating characteristic (ROC) analysis, with the area under the curve (AUC) as the primary performance metric. RESULTS: We defined a 19-gene cfDNA 5-hmC signature (eHMS), which showed high diagnostic accuracy for early HCC across multiple precancerous states, with AUCs of 92.3% (healthy controls vs. early HCC), 82.5% [benign liver lesions (BLL) vs. early HCC], 78.8% [chronic hepatitis B (CHB) vs. early HCC], 81.3% (liver cirrhosis vs. early HCC), and 89.4% when pooling all non-HCC, further validated in an external cfDNA cohort (AUC =93.5%). When traced to tissue 5-mC, the 5-mC-related eHMS (MeHMS) score distinguished tumors from adjacent tissues with an AUC of 99.9% and showed improved discriminatory performance compared with alpha-fetoprotein (AFP, AUC =71.5%). Moreover, the signature may reflect early metabolic reprogramming and provide prognostic value by stratifying patients for survival and disease progression. CONCLUSIONS: Our findings indicate that the eHMS represents a promising biomarker for liquid biopsy in early HCC surveillance.