An EMT-Based Five-Gene Prognostic Model and Single-Cell Analysis of SPP1 Expression in Hepatocellular Carcinoma

Abstract Background and Objective: Hepatocellular carcinoma (HCC) remains a major cause of cancer-related mortality, and reliable prognostic biomarkers are still needed. Epithelial–mesenchymal transition (EMT) is involved in HCC progression, and EMT-related genes may serve as candidates for prognostic modeling. This study aimed to construct an EMT-based five-gene prognostic model for HCC and to investigate the cellular expression pattern of SPP1 using single-cell and immunohistochemical analyses. Methods: EMT-related genes were obtained from the dbEMT database and intersected with differentially expressed genes and WGCNA hub genes in the TCGA cohort. Candidate prognostic genes were screened by univariate Cox and LASSO regression, and an EMT-based five-gene prognostic model was constructed and validated in TCGA and GSE14520 datasets using survival analysis, C-index, and hazard ratios (HRs). Functional enrichment analysis and immune cell infiltration assessment were performed to compare high- and low-risk groups. Single-cell analysis using GSE189903 was conducted to investigate the cellular distribution of SPP1. Immunohistochemistry (IHC) was performed on human HCC tissues and paired adjacent and normal liver tissues to preliminarily assess SPP1 protein expression. Results: The EMT-based five-gene prognostic model, consisting of MMP1, EZH2, SPP1, CCNA2, and MYCN, showed modest predictive performance in both training and validation cohorts. High-risk tumors were enriched in proliferation, cell cycle, and MYC target pathways, accompanied by immune microenvironment dysregulation. Single-cell analysis showed that SPP1 was expressed in multiple cell types, with relatively higher expression in myeloid cells (including macrophages, mast cells, and neutrophils) and detectable expression in hepatocarcinoma cells. Preliminary IHC analysis showed that SPP1 protein was elevated in HCC tissues compared with adjacent and normal liver tissues, with positive signals observed in hepatocellular carcinoma cells and some stromal areas, consistent with the single-cell finding. Conclusion: The EMT-based five-gene prognostic model showed modest predictive performance for survival risk in HCC patients. SPP1 was expressed in myeloid cells and hepatocarcinoma cells in single-cell data and was elevated in HCC tissues by IHC. However, the cellular source and functional role of SPP1 in HCC progression remain to be determined. These findings are descriptive and hypothesis-generating.