RPS3 predicts poor overall survival in HBV-related hepatocellular carcinoma patients: a data-mining with LASSO-regression algorithm.

Abstract

This analysis aimed to investigate the candidate biomarkers associated with overall survival (OS) in hepatocellular carcinoma (HCC) patients. In the GSE14520 dataset, candidate parameters were selected and included in the Cox regression and Nomogram models through bioinformatic enrichment methods and LASSO analysis, survivor functions of candidate biomarkers were also assessed. Complement and coagulation cascades including 36 differential expressed genes (DEGs) and ribosome pathway including 27 DEGs were significantly enriched (both p < 0.05 and adjusted p < 0.05). LASSO model, Cox regression and nomogram analysis indicated that RPS3, together with BCLC and TNM staging, were significantly associated with OS in HCC patients. Validated in the GEO series, TCGA and Human Protein Atlas (HPA) datasets, RPS3 mRNA and RPS3 protein were significantly upregulated in tumor tissues compared to that in nontumor tissues (all p < 0.05). Upregulation of RPS3 has been linked to high alpha fetoprotein (AFP), advanced tumor stages and multinodular (all p < 0.05). After adjusting AFP, tumor stage and multinodular, log rank analysis revealed that HCC patients with high RPS3 had unfavorable OS compared to those with low RPS3 (all p < 0.05). RPS3 upregulation in tumors might contribute to unfavorable OS in HCC patients.