Untargeted plasma metabolomics for risk prediction of hepatocellular carcinoma: A prospective study in two Chinese cohorts

Characterization of metabolic perturbation prior to hepatocellular carcinoma (HCC) may deepen the understanding of causal pathways and identify novel biomarkers for early prevention. We conducted two 1:1 matched nested case-control studies (108 and 55 pairs) to examine the association of plasma metabolome (profiled using LC-MS) with the risk of HCC based on two prospective cohorts in China. Differential metabolites were identified by paired t tests and orthogonal partial least-squares discriminant analysis (OPLS-DA). Weighted gene coexpression network analysis (WGCNA) was performed to classify metabolites into modules for identifying biological pathways involved in hepatocarcinogenesis. We assessed the risk predictivity of metabolites using multivariable logistic regression models. Among 612 named metabolites, 44 differential metabolites were identified between cases and controls, including 12 androgenic/progestin steroid hormones, 8 bile acids, 10 amino acids, 6 phospholipids, and 8 others. These metabolites were associated with HCC in the multivariable logistic regression analyses, with odds ratios ranging from 0.19 (95% confidence interval [CI]: 0.11-0.35) to 5.09 (95% CI: 2.73-9.50). WGCNA including 612 metabolites showed 8 significant modules related to HCC risk, including those representing metabolic pathways of androgen and progestin, primary and secondary bile acids, and amino acids. A combination of 18 metabolites of independent effects showed the potential to predict HCC risk, with an AUC of 0.87 (95% CI: 0.82-0.92) and 0.86 (95% CI: 0.80-0.93) in the training and validation sets, respectively. In conclusion, we identified a panel of plasma metabolites that could be implicated in hepatocellular carcinogenesis and have the potential to predict HCC risk.