Integrative single-cell and machine-learning analysis identifies LGALS1 as a macrophage-associated diagnostic and prognostic biomarker in hepatocellular carcinoma

Abstract

Background Hepatocellular carcinoma (HCC) is characterized by marked heterogeneity and an immunosuppressive microenvironment in which tumor-associated macrophages contribute to disease progression. This study aimed to identify macrophage-associated biomarkers with diagnostic, prognostic, and translational relevance in HCC. Methods Single-cell RNA-sequencing datasets were integrated with bulk transcriptomic and clinical data from TCGA-LIHC, GEO, and ICGC cohorts. Macrophage markers were intersected with tumor-associated differentially expressed and survival-associated genes. Consensus clustering, immune and mutation analyses, and multiple machine-learning algorithms were applied. Model performance was evaluated in TCGA-LIHC and three external cohorts. LGALS1 was further assessed through survival, treatment-response, immune-infiltration, single-cell localization, virtual-knockout, and in vitro loss-of-function analyses. Results A 16-gene macrophage-associated signature stratified HCC into subtypes with distinct overall survival, clinicopathological features, mutation patterns, and predicted immune-evasion characteristics. Ridge regression showed the most consistent diagnostic performance, with an area under the receiver operating characteristic curve of 0.988 in TCGA-LIHC and 0.895–0.924 in external cohorts. XGBoost-based SHAP analysis identified LGALS1 as a highly informative feature. Higher LGALS1 expression was associated with adverse survival outcomes, advanced disease features, sorafenib non-response, and greater macrophage infiltration. LGALS1 was enriched in monocyte/macrophage and fibroblast populations. Exploratory analyses in non-HCC immunotherapy cohorts suggested associations with immune checkpoint inhibitor outcomes. LGALS1 silencing reduced HCC-cell proliferation, colony formation, migration, and invasion. Conclusions Integrative single-cell and multi-cohort analyses identified LGALS1 as a macrophage-associated candidate biomarker in HCC. These findings support its potential value in tissue classification and risk assessment and warrant prospective HCC-specific validation and macrophage-focused mechanistic investigation.