A GSVA based gene set synergizing with CD4+T cell bearing harmful factors yield risk signals in HBV related diseases via amalgamation of artificial intelligence

Genes encoding chemokines and extracellular matrix (ECM) play pivotal roles in chronic HBV infection (CHB), HBV related fibrosis (HBV-LF) and hepatocellular carcinoma (HBV-HCC). The landscape and potential of these genes in prognosis across diseases stages have not been fully and systemically understood. In this study, we defined an HBV-LF associated gene set comprised of chemokines and ECM related genes directly induced by initial HBV infection through GSVA algorithm that named as CLST (C stands for CXCL9, CXCL10, CCL19 and CCL20; L for LUM; S for SOX9 and SPP1; T for THBS1, THBS2) and evaluated its biomarker values in CHB and HBV-LF. Enrichment scores (ES) of CLST was subsequently observed synergized with activated CD4+T cells (aCD4) highly related to T helper cell 17 (TH17) associated genes and immune checkpoints and addressed as risk signals due to bearing harmful prognosis factors in tumor tissues of patients with HBV-HCC. Dual higher enrichment score (ES) of CLST and aCD4 in HBV-HCC patients exhibited worse overall survival (OS). Feature genes specific to these two gene sets showed promising clinical relevance in early-stage of HBV-HCC definition and OS prediction incorporating laboratory parameters via artificial intelligence (AI) systems. Finally, a novel mechanistic insight into the issue was proposed that PEG IFN- as an immunotherapy through modulating CLST signal in treatment responders and these immune signals down-regulation could be beneficial for HBV related diseases control and prevention. Together, our study provides GSVA and AI derived immunogenomic prognosis signatures and clinical utility of these signals will be benefit for HBV related diseases cure.