Circulating cell-free DNA methylation biomarkers for hepatocellular carcinoma risk prediction in HIV-positive Nigerian population.

Abstract

Hepatocellular carcinoma (HCC) is a leading cause of cancer mortality globally, with particularly high burdens among people living with HIV (PLWH) in low-resource settings like Nigeria. Effective early detection remains a major challenge due to limited access to imaging-based surveillance and the low sensitivity of current biomarkers such as alpha-fetoprotein (AFP). We conducted an epigenome-wide association study (EWAS) of circulating cell-free DNA (ccfDNA) methylation in a Nigerian cohort of HIV-positive individuals (n = 245), spanning HCC, cirrhosis (FibroScan ≥12.3 kPa), fibrosis (FibroScan ≥7.6 and <12.3 kPa), and HCC-free without fibrosis (FibroScan <7.6 kPa) groups. Using random forest modeling, we developed and evaluated a ccfDNA methylation classifier (ccfDNAmRF) for HCC risk prediction. We identified 73 CpG sites significantly associated with HCC (false discovery rate <0.01). The ccfDNAmRF model demonstrated strong discriminatory power, achieving 100% sensitivity and 80%-91% specificity for distinguishing HCC from cirrhosis, fibrosis, and HCC-free groups (area under the curve [AUC]: 92%-97%). Combining ccfDNA methylation risk scores with AFP further improved classification accuracy (AUC up to 98.5%). Notably, ccfDNA methylation patterns displayed clear dose-response relationships across the disease spectrum, supporting their utility for early-stage detection and risk stratification. Our findings highlight the promise of ccfDNA methylation biomarkers as a non-invasive, blood-based screening tool for improving early identification of HCC cases among PLWH.