From Single-cell Insights to Clinical Relevance: An M2 Macrophagebased Prognostic Model for Osteosarcoma

M2 macrophages are closely associated with an immunosuppressive tumor microenvironment (TME) and may influence drug response, but their clinical relevance in osteosarcoma (OS) remains to be comprehensively elucidated. This study developed an M2-related model to predict the risk, immune response, and drug sensitivity for patients with OS.

Bulk RNA-seq data (TARGET-OS), microarray data (GSE21257), and scRNA-seq data (GSE162454) were obtained and analyzed. Single-cell data were processed using the Seurat package for cell annotation and cellular heterogeneity characterization. Intercellular communication networks were inferred using the CellChat R package. Next, based on M2 macrophage-associated genes identified through ssGSEA, we developed a four-gene prognostic model using WGCNA and LASSO Cox regression analysis. Prognostic performance of the four-gene model was evaluated by using Kaplan-Meier (KM) survival analysis and time-dependent ROC curves. Immune infiltration was assessed by ssGSEA, ESTIMATE, and MCP-counter, while drug sensitivity was predicted using oncoPredict.

The scRNA-seq analysis identified myeloid and osteoblastic cells as the dominant cell populations in OS, with M2 macrophages exhibiting extensive intercellular crosstalk. M2 macrophage activity scores were computed for samples in the TARGET-OS via ssGSEA. Based on these scores, WGCNA identified the M2 module as a key module, which comprised 93 genes. Among these modular genes, four genes were selected by LASSO Cox regression analysis to establish a four-gene RiskScore. High- -risk patients showed worse survival (p < 0.05), which was also observed in the independent GSE21257 cohort. The high-risk group also exhibited lower ImmuneScore and reduced infiltration of T cells, B cells, dendritic cells (DCs), and macrophages. The RiskScore was correlated with predicted IC50 values for multiple drugs, including AZD8055_1059, suggesting a potential link between the M2 macrophage-related model and in silico drug sensitivity profiles.

This study developed an M2 macrophage-related risk model based on LPAR5, MS4A4A, TNFSF8, and VSIG4, which was associated with survival outcomes, TME features, and predicted drug response profiles in OS.

This study developed an M2 macrophage-related four-gene model that was closely related to the immune microenvironment features, drug sensitivity, and survival outcomes in OS. These findings offer preliminary insights into risk stratification and therapeutic treatment for OS.