Constructing a prognostic model for hepatocellular carcinoma based on bioinformatics analysis of inflammation-related genes.

BACKGROUND

This study aims to screen inflammation-related genes closely associated with the prognosis of hepatocellular carcinoma (HCC) to accurately forecast the prognosis of HCC patients.

METHODS

Gene expression matrices and clinical information for liver cancer samples were obtained from the Cancer Genome Atlas (TCGA) and the International Cancer Genome Consortium (ICGC). An intersection of differentially expressed genes of HCC and normal and GeneCards yielded inflammation-related genes associated with HCC. Cox regression and the minor absolute shrinkage and selection operator (LASSO) regression analysis to filter genes associated with HCC prognosis. The prognostic value of the model was confirmed by drawing Kaplan-Meier and ROC curves. Select differentially expressed genes between the high-risk and low-risk groups and perform GO and KEGG pathways analyses. CIBERSORT analysis was conducted to assess associations of risk models with immune cells and verified using real-time qPCR.

RESULTS

A total of six hub genes (C3, CTNNB1, CYBC1, DNASE1L3, IRAK1, and SERPINE1) were selected using multivariate Cox regression to construct a prognostic model. The validation evaluation of the prognostic model showed that it has an excellent ability to predict prognosis. A line plot was drawn to indicate the HCC patients' survival, and the calibration curve revealed satisfactory predictability. Among the six hub genes, C3 and DNASE1L3 are relatively low expressed in HCCLM3 and 97H liver cancer cell lines, while CTNNB1, CYBC1, IRAK1, and SERPINE1 are relatively overexpressed in liver cancer cell lines.

CONCLUSION

One new inflammatory factor-associated prognostic model was constructed in this study. The risk score can be an independent predictor for judging the prognosis of HCC patients' survival.