Early diagnosis for α-fetoprotein (AFP) negative hepatocellular carcinoma (HCC) remains a critical problem. Metabolomics is prevalently involved in the identification of novel biomarkers. This study aims to identify new and effective markers for AFP negative HCC.
In total, 147 patients undergoing liver transplantation were enrolled from our hospital, including liver cirrhosis patients (LC, n=25), AFP negative HCC patients (NEG, n=44) and HCC patients with AFP over 20 ng/mL (POS, n=78). 52 Healthy volunteers (HC) were also recruited in this study. Metabolomic profiling was performed on the plasma of those patients and healthy volunteers to select candidate metabolomic biomarkers. A novel diagnostic model for AFP negative HCC was established based on Random forest analysis, and prognostic biomarkers were also identified.
15 differential metabolites were identified being able to distinguish NEG group from both LC and HC group. Random forest analysis and subsequent Logistic regression analysis showed that PC(16:0/16:0), PC(18:2/18:2) and SM(d18:1/18:1) are independent risk factor for AFP negative HCC. A three-marker model of Metabolites-Score was established for the diagnosis of AFP negative HCC patients with an area under the time-dependent receiver operating characteristic curve (AUROC) of 0.913, and a nomogram was then established as well. When the cut-off value of the score was set at 1.2895, the sensitivity and specificity for the model were 0.727 and 0.92, respectively. This model was also applicable to distinguish HCC from cirrhosis. Notably, the Metabolites-Score was not correlated to tumor or body nutrition parameters, but difference of the score was statistically significant between different neutrophil-lymphocyte ratio (NLR) groups (≤5 vs. >5, P=0.012). Moreover, MG(18:2/0:0/0:0) was the only prognostic biomarker among 15 metabolites, which is significantly associated with tumor-free survival of AFP negative HCC patients (HR=1.160, 95%CI 1.012-1.330, P=0.033).
The established three-marker model and nomogram based on metabolomic profiling can be potential non-invasive tool for the diagnosis of AFP negative HCC. The level of MG(18:2/0:0/0:0) exhibits good prognosis prediction performance for AFP negative HCC.