Hepatocellular carcinoma (HCC) is one of the most common malignant tumors in the world. However, current biomarkers that discriminate HCC from liver cirrhosis (LC) are important but are limited. More reliable biomarkers for HCC diagnosis are therefore needed. Serum from HCC patients, LC patients and healthy volunteers were analyzed using NMR and LC/MS-based approach in conjunction with random forest (RF) analysis to discriminate their serum metabolic profiles. Thirty-two potential biomarkers have been identified, and the feasibility of using these biomarkers for the diagnosis of HCC was evaluated, where 100% sensitivity was achieved in detecting HCC patients even with AFP values lower than 20 ng/mL. The metabolic alterations induced by HCC showed perturbations in synthesis of ketone bodies, citrate cycle, phospholipid metabolism, sphingolipid metabolism, fatty acid oxidation, amino acid catabolism and bile acid metabolism in HCC patients. Our results suggested that these potential biomarkers identified appeared to have diagnostic and/or prognostic values for HCC, which deserve to be further investigated. In addition, it also suggested that RF is a classification algorithm well suited for selection of biologically relevant features in metabolomics.
Keywords: hepatocellular carcinoma; liquid chromatography-mass spectrometry; metabolomics; nuclear magnetic resonance; random forest.
© 2013 UICC.