Identification of Causal Plasma Proteins in Hepatocellular Carcinoma via Two-Sample Mendelian Randomization and Integrative Transcriptomic‒Proteomic Analysis.

abstractHepatocellular carcinoma (HCC) poses a serious threat to human health. Identifying diagnostic and therapeutic targets from plasma proteins is essential for early HCC detection and treatment. Large-scale plasma proteomics data and Mendelian randomization (MR) offer opportunities for this type of research. Proteomic data from protein quantitative trait loci were obtained from the UK Biobank Pharma Proteomics Project. Two-sample MR was used to evaluate the causal association between plasma protein levels and HCC. Sensitivity, directionality, and colocalization analyses were conducted to validate the causal associations identified by MR. Robustness was further tested through colocalization analysis. The single-cell dataset GSE166635 was used to assess the expression of genes corresponding to proteins identified via MR. Potential drugs targeting these proteins were evaluated using the DSigDB/DrugBank databases and molecular docking. Seven proteins with potential causal associations with HCC were identified: ASS1, B2M, FUOM, GABARAPL1, ST8SIA1, STOML2, and KRT8. Among these genes, ASS1, KRT8, and STOML2 presented the strongest effects in the colocalization analyses. In a single-cell liver cancer dataset, ASS1, KRT8, and STOML2 were expressed mainly in hepatic progenitor cells and malignant cells, with KRT8 predominantly found in hepatic progenitor cells and playing a role in the oncogenesis of malignant liver cells. Potential drugs targeting ASS1 include arginine, aspartic acid, and citrulline; for KRT8, ambroxol, diltiazem, and amikacin; and for STOML2, chlortetracycline, chlorzoxazone, and dirithromycin. This study identified several novel causal plasma proteins, providing new insights into the early diagnosis and treatment of HCC. SIGNIFICANCE: In this study, we identified several causal proteins in HCC using UK Biobank Pharma Proteomics Project proteomic data via two-sample MR. We performed colocalization and sensitivity analyses, utilized single-cell RNA sequencing data for validation, and discovered potential drugs through molecular docking.