Development and validation of an eight-gene signature based predictive model to evaluate the prognosis of hepatocellular carcinoma patients: a bioinformatic study.

BACKGROUND

Hepatocellular carcinoma (HCC) is a malignant tumor with a poor prognosis, however, biomarkers for the prognostic assessment of HCC remain suboptimal. Consequently, we aimed to develop a reliable tool for prognostic estimation of HCC.

METHODS

Differentially expressed genes (DEGs) between HCC and adjacent normal tissues in 3 Gene Expression Omnibus (GEO) datasets were identified, followed by hub gene selection and least absolute shrinkage and selection operator (LASSO) Cox regression to develop a prognostic gene signature. Kaplan-Meier survival analysis, univariate and multivariate Cox regression, time-dependent area under the curve (AUC), and integrated value of time-dependent AUC (iAUC) were used to assess the relationship between predictors and clinical outcomes in the training and validation datasets. Then we built nomograms including gene signature and clinicopathological factors to forecast the probability of death. Moreover, we performed quantitative real-time PCR (qPCR) to compare the expression of prognostic genes between HCC and adjacent normal tissues. Finally, the relationship between prognostic genes and tumor microenvironment (TME) was investigated using immune cell infiltration algorithms and single cell transcriptomic database.

RESULTS

Eight prognostic genes (, , , , , , , and ) were finally identified to construct the gene signature. Each patient's risk score was calculated according to the gene signature. Patients with high-risk scores showed worse outcomes in the training set [hazard ratio (HR) =3.404, P<0.001]. Risk score, age, body mass index (BMI), and TNM stage were identified as independent prognostic factors for overall survival (OS) in the training set. The nomogram including risk score and other independent prognostic factors showed better performance as opposed to the clinicopathological model. In the validation dataset, we obtained the similar results as well. Moreover, we found a close relationship between risk score and immune cell infiltration. Patients with high-risk scores had elevated expression of immune checkpoint genes, indicating that these patients may be more suitable for immunotherapy.

CONCLUSIONS

We have established and validated an eight-gene based prognostic model, which could be an effective tool for the prognostic evaluation of HCC patients.