Prognostic index construction and analysis of immune infiltration combined with RNA binding protein and transcription factor in hepatocellular carcinoma / 中华肝胆外科杂志

ABSTRACT

Objective:To comprehensively analyze the prognostic prediction value of RNA binding protein, transcription factor gene expression and immune infiltration in hepatocellular carcinoma (HCC).Methods:Common gene sets associated with RNA-binding proteins and transcription factors were screened in TCGA ( n=365) , GSE54236 ( n=78) and GSE14520 ( n=221) datasets. Univariate Cox regression was used for primary screening. The survival regression model was constructed by LASSO-Cox. And a complex index [CIRT=(score-min)/max] was calculated. According to the median of CIRT, the HCC patients were divided into CIRT high group ( n=182) and CIRT low group ( n=182). The differences of prognosis, immune infiltration between the two groups were analyzed. Results:Of 37 prognostically relevant RNA binding protein and transcription factor genes were identified. The prognosis prediction model based on seven selected genes was determined by stepwise regression. Patients in the CIRT high group exhibited a lower percentage of macrophages in M1 ( P=0.032), macrophages in M2 ( P=0.009), resting mast cell ( P<0.001), activated NK cells ( P=0.007), and resting memory CD4 + T cells ( P<0.001), while patients in the CIRT low group showed a lower level of resting dendritic cells ( P=0.048), macrophages in M0 ( P<0.001), neutrophils ( P=0.049), follicular helper T cells ( P=0.004) and regulatory T cells ( P=0.001). GSEA analysis has shown that CIRT high groups were highly enriched in cell cycle, DNA repair pathways in TCGA and GSE14520. In the TCGA cohort, the CIRT low group had better overall survival than the CIRT high group. Analysis of 5-year follow-up data in the TCGA cohort showed that CIRT had a good predictive value for long-term survival of patients with liver cancer (area under receiver operating characteristic curve was 0.71). Conclusion:A novel prognostic index and classifier based on RNA-binding protein expression, transcription factors and immune expression profiles were developed and cross-cohort validated. CIRT could be used as an independent predictor.