Hepatocellular carcinoma (HCC) remains a therapeutic challenge due to tumor microenvironment heterogeneity and metabolic reprogramming. DHRS3, a retinol-metabolizing enzyme, exhibits paradoxical roles across cancers, but its specific function in HCC is poorly understood.This study integrated pan-cancer data from TCGA and GTEx, conducted single-cell and spatial transcriptomic analyses, and carried out in vitro assays and in vivo xenograft models for validation. A machine learning framework was used to develop the DHRS3-cholesterol metabolic prognostic signature (DCMPS) by combining DHRS3 expression with cholesterol-related genes.DHRS3 was upregulated in HCC and associated with favorable prognosis. It enhanced anti-tumor immunity by promoting MHC-I/II antigen presentation and inhibited metastasis via SEMA4A signaling and endothelial-mesenchymal transition (EndMT) suppression. Conversely, DHRS3 drove proliferation through cholesterol metabolism, reversible by atorvastatin (ATO). ATO synergized with all-trans retinoic acid (ATRA) to suppress tumor growth and enhance immune cell infiltration. The DCMPS model outperformed TNM staging in prognostic prediction.DHRS3 acts as a dual-functional immune-metabolic regulator in HCC. The DCMPS signature serves as a robust prognostic biomarker and may guide the application of ATRA-ATO combination therapy, offering a rationale for a novel precision medicine strategy in HCC management.
Hu et al. (Fri,) studied this question.