OA-27: Single cell transcriptomics reveals the impact of tumor cell interactions with the bone marrow immune microenvironment in the progression of multiple myeloma

Introduction: Alterations within the bone marrow microenvironment (BMME) may contribute to the progression of multiple myeloma (MM) from its precursor stages. The significance of the BMME and the role of different immune cell types in the progression of myeloma could identify novel therapeutic approaches to prevent the progression of disease. Here we investigated the frequency of different immune cell types in the BMME at different stages of myeloma and their interactions with the tumor cells.

Methods: BM aspirates from smoldering multiple myeloma (SMM; n=10), newly diagnosed multiple myeloma (NDMM; n=23), and relapsed/refractory multiple myeloma (RRMM; n=22) patients were collected from the Indiana Myeloma Registry. BM mononuclear cells were isolated by Ficoll and underwent CD138 magnetic bead selection into CD138+ and CD138- fractions. CD138- samples underwent scRNA and TCR sequencing (10X Genomics) and CD138+ samples underwent multiomic analysis (10x Genomics). CD138- data was normalized, integrated, and clusters were annotated for cell type using Seurat (v4), resulting in 315,533 high quality immune cells. Significant proportional and gene expression differences between clusters at different disease stages were calculated. Quantitative analysis on intercellular communication networks were analyzed between single cell clusters of matched CD138+ cells and CD138- samples using CellChat (v1.6.0).

Results: Integration of all 55 CD138- samples resulted in 36 immune cell subclusters. Despite patient heterogeneity, significant changes in the immune cell proportions were identified as the disease progressed; a substantial decrease in CD4 memory T cells from NDMM to RRMM (P < 0.0001) (SMM=15.7%; NDMM=13.4%; RRMM=4%) whereas CD4 T effector cells significantly increased from NDMM to RRMM (P=0.005) (SMM=10.8%; NDMM=8.7%; RRMM=11.9%) and CD8 T MAIT cells decreased from NDMM to RRMM (P=0.002). CD56 natural killer cells increased (SMM=1.6%; NDMM=2.7%; RRMM=5.2%; P< 0.05) as did CD16 monocyte cell clusters (SMM=0.9%; NDMM=1.9%; RRMM=4.3%; P< 0.05). Regarding tumor-BMME intercellular communication, we observed known APRIL-BCMA and BAFF-BCMA/BAFF-TACI interactions between myeloid dendritic cells or monocyte clusters and tumor cell clusters. Our single cell data shows these interactions decreased from SMM to NDMM and RRMM (P < 0.05). In addition, RRMM samples had a significant increase in MIF-CD74/CXCR4 interactions between tumor cells and myeloid dendritic cells compared to SMM and NDMM cells (P < 0.05), which can contribute to the immune evasion of myeloma cells and limit the effectiveness of immune-based therapies.

Conclusions: Immune cell subgroups change as the disease progresses from SMM to NDMM and RRMM, as do their interactions with myeloma cells possibly resulting in increased resistance through immune evasion and MIF signaling mechanisms.