Transcriptomic profiling of CD8⁺ T cells across distinct stages of chronic HBV infection and hepatitis B virus-related acute-on-chronic liver failure: development of a short-term prognostic model

Background & aims: Hepatitis B virus (HBV) infection affects 254 million people worldwide, and some patients progress to hepatitis B virus-related acute-on-chronic liver failure (HBV-ACLF), whose pathogenesis remains incompletely understood. CD8⁺ T cells mediate viral control and liver injury in HBV-ACLF. This study aimed to systematically characterize the transcriptomic trajectory of CD8⁺ T cells from chronic HBV infection to HBV-ACLF. Methods We enrolled 11 healthy controls (HCs), 58 patients with chronic hepatitis B (CHB), and 25 patients with HBV-ACLF. CD8⁺ T cells were immunomagnetically sorted and profiled by Bulk RNA-seq. Analyses included differential gene expression, deconvolution, gene set scoring, and qRT-PCR. Logistic regression was used to identify independent prognostic factors for 28-day outcomes. Results Compared with CHB, 659 differentially expressed genes (DEGs) were identified in HBV-ACLF (624 up, 35 down), with enrichment in interferon, TNFα signaling and inflammatory response pathways. Gene set scoring and qRT-PCR demonstrated CD8⁺ T cells from HBV-ACLF exhibited a terminal state. Deconvolution analysis revealed that Tn, Tscm, and Tem subsets were significantly decreased in HBV-ACLF. Logistic regression identified ENTPD1 (OR = 2.018, 95% CI: 1.279–4.162, P < 0.001 ) and HLA-DRA (OR = 1.691, 95% CI: 1.081–3.187, P = 0.019 ) as independent risk factors for poor short-term prognosis. MELD-Na alone showed limited predictive performance (AUROC = 0.603), whereas combining it with ENTPD1 (AUROC = 0.971, P = 0.004 ) or HLA-DRA (AUROC = 0.890, P = 0.007 ) significantly improved prognostic performance. Conclusions CD8⁺ T cells in HBV-ACLF display a terminal phenotype characterized by concurrent immune activation and exhaustion. Combining MELD-Na with ENTPD1 or HLA-DRA substantially improves short-term prognostic performance for HBV-ACLF.