Integrative multi-omics landscape and diagnostic-therapeutic modelling of steatohepatitic hepatocellular carcinoma

BACKGROUND: Steatohepatitic hepatocellular carcinoma (SH-HCC) is a distinct subtype of hepatocellular carcinoma with poorly defined molecular features and limited diagnostic and therapeutic strategies. OBJECTIVE: This study aimed to delineate its multi-omics landscape, develop non-invasive diagnostic tools and identify actionable therapeutic targets. DESIGN: We integrated multimodal clinical and omics datasets from an overall cohort of 156 patients, including RNA-seq (n=69), spatial transcriptomics (n=36), spatial metabolomics (n=6), lipidomics (n=125) and serum proteomics (n=152), for comprehensive molecular profiling and diagnostic model construction. FABP4 was functionally validated in Sleeping Beauty-induced HCC mouse models. RESULTS: An SH-HCC index was established to reliably differentiate SH-HCC from conventional HCC. Drug response analysis suggested a potential reduction in responsiveness to sorafenib and immunotherapy-based regimens in SH-HCC, whereas transarterial chemoembolisation may remain an effective therapeutic option. Spatial transcriptomics revealed increased immune cell infiltration, a reinforced immune barrier and reduced tumour-immune cell communication. SH-HCC is characterised by a lipid-rich phenotype, and a diagnostic model based on 12 lipid species showed good performance in an independent cohort (area under the curve (AUC)=0.909). Serum proteomic model demonstrated good performance in the test cohort (AUC=0.828). Importantly, our preliminary findings suggest a potential therapeutic role for FABP4, as its genetic deletion suppressed tumour growth, reshaped the tumour immune microenvironment and enhanced the response to PD-1 blockade. CONCLUSION: This study provides the first multi-omics characterisation of SH-HCC, highlighting the SH-HCC index and non-invasive diagnostic models, with FABP4 as a potential target.