Immune Profiling Reveals Distinct Features of Multiple Myeloma Subtypes Defined By Multi-Omics Network Analysis

Introduction: Multiple myeloma (MM) has a highly heterogeneous genomic landscape, challenging patient stratification and clinical management. By combining bulk tumor profiling with single-cell data of the tumor microenvironment (TME), we aimed to create a comprehensive multi-omics network model to dissect inter-tumor heterogeneity, which enabled us to characterize the immune cell landscape and correlate it with patient subgroups identified through multi-omics analysis of bulk tumor data, revealing significant features within each subgroup. Methods: We analyzed 185 samples collected at diagnosis from newly-diagnosed MM (NDMM) patients in MMRF CoMMpass, included in both our previous MM-PSN model (Bhalla et al, 2021) and the MMRF Immune Atlas (Pilcher et al, 2023). The CD138- fraction underwent 10X 3' single cell RNA seq (scRNA-seq), yielding 776,859 cells. Our dataset included 107,884 myeloid cells, 407,895 NK/T cells, 162,630 B/ erythroid cells, and 85,147 plasma cells. The analysis was performed using the R packages Seurat, CellphoneDB, pySCENIC, MOVICS and MOGONET. Results: We stratified 185 patients in the Immune Atlas cohort according to our multi-omics patient similarity network MM-PSN into the following groups: (i) hyperdiploidy (HD) and t(8;14) translocation of MYC (tMYC) (group 1, n = 105); (ii) translocations t(4;14) of MMSET (tMMSET) and t(14;16) of MAF (tMAF) (group 2, n = 46); and (iii) translocation t(11;14) of CCND1 (tCCND1) (group 3, n = 33). The overall cohort had a median progression-free survival (PFS) of 1082 days. The subgroups had the following median PFS: HD, 1094 days; tMMSET/tMAF, 634 days; and tCCND1, 1236 days. Group 1 had higher frequencies of hematopoietic stem cells (HSCs), plasmacytoid dendritic cells (pDCs) and B naïve cells (p<0.05), compared to groups 2 and 3. This suggests HD and tMYC may be associated with TME changes, influencing cytokine and growth factor production, as indicated by IL4R (B naïve) and GZMB (pDC) enriched signatures. Moreover, patients with HD and gain(1q) had higher plasma and B cell frequencies and the lowest CD8 T cell frequency compared to all other HD patients (p<0.001). Group 2, with tMMSET and tMAF, had higher frequencies of pro-B and large pre-B cells, and lower frequencies of CD4 and CD8 T naive cells (p<0.05), suggesting an enrichment of immature B cells and depletion of naive T cells, indicating broader immune dysregulation. Notably, patients with both tMMSET and gain(1q), who had the most aggressive cases with the lowest PFS and OS in the original MM-PSN study, showed higher frequencies of plasma cells and proliferating immature B cells (Ki67+) (p<0.001), as in HD with gain(1q). This supports and extends the MM-PSN findings associating gain(1q) with a more proliferative and aggressive myeloma clonotype. Finally, group 3 with tCCND1 had higher frequencies of CD4 naïve, CD8 naïve and CD56high NK cells, suggesting a larger pool of less differentiated T cells and potentially better T cell fitness. These findings align with the higher median PFS and better outcomes of tCCND1 in NDMM patients and may inform cellular therapies that leverage the improved T cell fitness in this subgroup. Conclusion: In this study, we enhanced our bulk tumor analysis of MMRF CoMMpass with TME profiling using scRNA data from the MMRF Immune Atlas. The analysis of patient subgroups revealed distinct tumor and immune landscapes. The presence of 1q gain correlated with a proliferative plasma cell compartment and was associated with lower survival rates and worse outcomes. These findings deepen our understanding of the TME's interaction with MM subtypes and provide a foundation for ongoing integrative analyses.