Pre-Diagnostic Plasma Metabolites are Associated with Incident Hepatocellular Carcinoma: A Prospective Analysis.

Abstract

Despite increasing incidence of hepatocellular carcinoma (HCC) in vulnerable populations, accurate early detection tools are lacking. We aimed to investigate the associations between pre-diagnostic plasma metabolites and incident HCC in a diverse population. In a prospective, nested case-control study within the Southern Community Cohort Study (SCCS), we conducted pre-diagnostic liquid chromatography-mass spectrometry metabolomics profiling in 150 incident HCC cases (median time to diagnosis 7.9 years) and 100 controls with cirrhosis. Logistic regression assessed metabolite associations with HCC risk. Metabolite set enrichment analysis identified enriched pathways, and random forest classifier was used for risk classification models. Candidate metabolites were validated in the UK Biobank (N=12 incident HCC cases and 24 cirrhosis controls). In logistic regression analysis, seven metabolites were associated with incident HCC (Meff p<0.0004), including N-acetylmethionine (OR=0.46, 95% CI=0.31-0.66). Multiple pathways were enriched in HCC, including histidine and coenzyme A metabolism (FDR p<0.001). Random forest classifier identified ten metabolites for inclusion in HCC risk classification models, which improved HCC risk classification compared to clinical covariates alone (AUC=0.66 for covariates vs. 0.88 for 10 metabolites plus covariates; p<0.0001). Findings were consistent in the UK Biobank (AUC=0.72 for covariates vs. 0.88 for four analogous metabolites plus covariates; p=0.04), assessed via nuclear magnetic resonance spectroscopy. Pre-diagnostic metabolites, primarily amino acid and sphingolipid derivatives, are associated with HCC risk and improve HCC risk classification beyond clinical covariates. These metabolite profiles, detectable years before diagnosis, could serve as early biomarkers for HCC detection and risk stratification if validated in larger studies.