Plasma metabolomic profiling for hepatocellular carcinoma diagnosis and microvascular invasion prediction.

Abstract

Altered metabolites are pivotal in hepatocellular carcinoma (HCC) development. This study employed untargeted metabolomic analysis to identify novel biomarkers for early HCC detection and explore their functions. Plasma samples were collected from 138 HCC patients, 69 patients with benign hepatic lesions, and 35 healthy donors. These samples were divided into a discovery set of 171 and a validation set of 71, and analyzed using ultra high performance liquid chromatography mass spectrometry. Through paired t-tests and orthogonal partial least-squares discriminant analysis, nine metabolites with significant predictive value were selected out and incorporated into a model for HCC diagnosis. Area under curves for the discovery set, the validation set, and all samples were 0.97, 0.95, and 0.96, respectively. The satisfactory diagnostic performance was maintained regardless of the China liver cancer (CNLC) staging. Additionally, this model demonstrated better diagnostic performance than alpha-fetoprotein (AFP) when comparing HCC to controls in different CNLC stages. The metabolite pathway enrichment analysis showed that alterations in plasma bile acids were associated with cirrhosis. Univariate and multivariate analyses indicated that the ratio of L-Serine and Sarcosine was an independent predictor for microvascular invasion (MVI). An integrated analysis of metabolomic data with transcriptomic data from the Cancer Genome Atlas revealed that the low expression of alanine glyoxylate aminotransferase (AGXT) and glycine amidinotransferase (GATM) was more likely related to MVI. To sum up, our research findings may offer valuable insights into HCC metabolic alterations and contribute to a better characterization of HCC.