Single-cell and bulk transcriptomic profiling of M2-like tumor-associated macrophages reveals a prognostic signature and an inflammation–metabolism crosstalk program in multiple myeloma

BACKGROUND: M2-like tumor-associated macrophages (TAMs) contribute to multiple myeloma (MM) progression and treatment resistance, yet their internal heterogeneity remains poorly understood. We used single-cell RNA sequencing (scRNA-seq) with bulk transcriptomic cohorts to resolve MM-specific M2-TAM subpopulations and develop a prognostic signature. METHODS: scRNA-seq data from MM and normal bone marrow (GSE124310, GSE278230) were processed after quality filtering and Harmony batch correction. M2 macrophages were sub-clustered; MM-restricted subclusters (MM-M2) were identified by Ro/e analysis and SCENIC. Intercellular signaling was characterized via CellChat and CellCall; trajectories were reconstructed with Monocle 2. hdWGCNA co-expression modules were intersected with differentially expressed genes (GSE6477), and candidates were evaluated across 101 configurations of ten machine learning algorithms to build a consensus prognostic model (MRGs). GSE9782 served as training; GSE57317 as validation. Immune infiltration was estimated with ESTIMATE and six deconvolution tools; drug sensitivity via CTRP and PRISM. RESULTS: Seven M2 subclusters were nearly absent in normal marrow but enriched in MM (MM-M2), each with a distinct transcription factor profile. Ligand-receptor signaling involving MM-M2 cells, plasma cells, and stromal cells was markedly stronger in MM. MRGs stratified patients by survival in both cohorts: high-risk patients showed an immune-depleted microenvironment; low-risk patients had greater immune infiltration. Drug sensitivity also differed between risk groups. MM-M2 macrophages showed a hypoxia-associated transcriptional pattern, with higher hypoxia scores and lower oxidative-phosphorylation scores, that correlated with inflammatory pathway scores at both the single-cell and patient levels. CONCLUSION: This study identified a previously unrecognized MM-specific M2-TAM subpopulation and derived a prognostic signature (MRGs) predicting survival, reflecting immune microenvironment composition, and nominating candidate therapeutic compounds for more precise MM risk stratification.