Background: Targeted and immunotherapy are promising for the treatment of patients with hepatocellular carcinoma (HCC), however, individual survivals vary significantly due to tumor heterogeneity. This study aimed to identify targeted immunotherapy-related prognostic biomarkers and explore their correlations with tumor microenvironment (TME) and pre-treatment CT radiomics features to gain insights into HCC heterogeneity.

Methods: Targeted immunotherapy-related genes (TIGs) were identified from the intersection of GSE datasets. TIGs were analyzed through univariate Cox regression analyses and least absolute shrinkage and selection operator (LASSO)-Cox analyses to build a prognostic signature. A total of 518 HCC samples from GSE14520(n=221) and The Cancer Genome Atlas liver hepatocellular carcinoma (TCGA-LIHC) dataset(n=297) were used to develop and validate the prognosis signature. Kaplan-Meier survival curves were used to verify the reliability of the developed signature. Furthermore, the relationships between tumor microenvironment and the signature were explored by bioinformatics analysis. Finally, Weighted Gene Co-Expression Network Analysis (WGCNA) and visualization analysis was used to link imaging phenotypes to gene expression, decoding tumor heterogeneity.

Results: Four TIGs (LY86, MRC1, FADS1 and ADA2) were identified and used to establish the prognostic signature for HCC patients. Kaplan-Meier analysis showed that the signature divided patients into high- and low-risk groups with significantly difference survival (high-risk group: worse prognosis, P<0.05). Bioinformatics analysis revealed significant associations between the four-gene prognostic signature and specific patterns of immune cell infiltration within the TME. WGCNA identified the two key radiomics modules associated with the prognostic signature, and a color visualization of the hub radiomics features was presented, the heatmaps of radiomics features highlight the intra-tumoral vessels and hypo-enhancement region of the entire tumor area.

Conclusion: We developed and validated a four-gene prognostic signature based on TIGs in HCC. This signature not only predicts patient survival but also demonstrates significant links to the underlying immune microenvironment and distinct CT radiomics phenotypes, providing novel multi-modal insights into HCC heterogeneity that may facilitate personalized targeted and immunotherapeutic strategies.