ceRNA network development and tumor-infiltrating immune cell analysis in hepatocellular carcinoma

Hepatocellular carcinoma (HCC) is among the primary causes of cancer deaths globally. Despite efforts to understand liver cancer, its high morbidity and mortality remain high. Herein, we constructed two nomograms based on competing endogenous RNA (ceRNA) networks and invading immune cells to describe the molecular mechanisms along with the clinical prognosis of HCC patients. RNA maps of tumors and normal samples were downloaded from The Cancer Genome Atlas database. HTseq counts and fragments per megapons per thousand bases were read from 421 samples, including 371 tumor samples and 50 normal samples. We established a ceRNA network based on differential gene expression in normal versus tumor subjects. CIBERSORT was employed to differentiate 22 immune cell types according to tumor transcriptomes. Kaplan-Meier along with Cox proportional hazard analyses were employed to determine the prognosis-linked factors. Nomograms were constructed based on prognostic immune cells and ceRNAs. We employed Receiver operating characteristic (ROC) and calibration curve analyses to estimate these nomogram. The difference analysis found 2028 messenger RNAs (mRNAs), 128 micro RNAs (miRNAs), and 136 long non-coding RNAs (lncRNAs) to be significantly differentially expressed in tumor samples relative to normal samples. We set up a ceRNA network containing 21 protein-coding mRNAs, 12 miRNAs, and 3 lncRNAs. In Kaplan-Meier analysis, 21 of the 36 ceRNAs were considered significant. Of the 22 cell types, resting dendritic cell levels were markedly different in tumor samples versus normal controls. Calibration and ROC curve analysis of the ceRNA network, as well as immune infiltration of tumor showed restful accuracy (3-year survival area under curve (AUC): 0.691, 5-year survival AUC: 0.700; 3-year survival AUC: 0.674, 5-year survival AUC: 0.694). Our data suggest that Tregs, CD4 T cells, mast cells, SNHG1, HMMR and hsa-miR-421 are associated with HCC based on ceRNA immune cells co-expression patterns. On the basis of ceRNA network modeling and immune cell infiltration analysis, our study offers an effective bioinformatics strategy for studying HCC molecular mechanisms and prognosis.