Single Cell Multi-Omic Analysis of Tumor Microenvironment Evolution across the Disease Spectrum of Multiple Myeloma Identifies Differential Mechanisms of Immune Suppression

Background:

Multiple myeloma (MM) is preceded by the precursor states of monoclonal gammopathy of undetermined significance (MGUS) and smoldering multiple myeloma (SMM). The transformation of these precursors into myeloma is influenced not only by the evolution of tumor itself but also by the tumor microenvironment (TME). However, the specific molecular mechanisms and factors within the TME that drive this malignant progression remain largely unknown. Therefore, exploring the TME from non-malignant precursor stages to active myeloma could facilitate the development of novel therapeutic strategies aimed at slowing down this progression and possibly eradicating MM.

Methods:

We performed paired scRNA-seq, scTCR-seq and scBCR-seq on 148 freshly collected BM aspirate samples including 73 relapsed refractory (RR) RRMM, 16 newly diagnosed (ND) MM, 34 SMM, 21 MGUS and 4 normal BM (nBM) samples. BM aspirates were subjected to CD138+ selection (n=19) or processed as whole BM (WBM) (n=129, among which 73 samples were also enriched for CD138+ cells). cDNA libraries were prepared from enriched CD138+ (n=92) and WBM samples (n=129) followed by simultaneous 5’ gene expression (scRNA-seq) and T/B cell receptor sequencing (scTCR/BCR-seq). Single-cell data was processed as previously described (Dang et. al, Cancer Cell, 2023).

Results:

We profiled 656,371 high-quality cells including 344,665 CD138+ and 311,706 TME cells. We also obtained scTCR-seq and scBCR-seq data on 86,244 and 566,155 cells, respectively. Among them, 394,187 out of 652,399 cells had paired scRNA-seq data. Unsupervised clustering analysis revealed 15 different major TME cell types. Sub-clustering of each major cell type further resolved them into different subsets, resulting in a total of 94 TME cell subsets (Figure). Overall, we observed a decreased proportion of CD4 T cells, B cells and progenitors and the enrichment of CD8 T cells, monocytes, and neutrophils along the nBM-MGUS-SMM-NDMM-RRMM axis. Notably, the most profound changes were identified within the context of RRMM. In lymphoid cells, we observed a stepwise decrease in mature B cells during disease progression. For T and NK cells, the naïve-like subsets were more abundant in normal bone marrow and the precursors. In contrast, the effector/cytotoxic subsets were enriched in symptomatic MM (NDMM and RRMM). Immunoregulatory T cells, including CD4T_C2 (Treg), CD4T_C4 (Th17) and CD4T_C7 (Tfh), showed an increased proportion within CD4 T cells in MM, particularly in RRMM compared with precursors. Furthermore, we identified a small exhaustion-like CD8 T cell subset which were enriched in RRMM. In myeloid cells, the classical monocytes showed a high degree of transcriptional homogeneity while the non-classical and intermediate monocytes were more heterogeneous. We observed a unique CD16_Mono_C0 subset which downregulated a considerable number of cytokine/chemokine receptors, almost exclusively unique to MM. The macrophage subsets displaying high phagocytosis signature (Macro_C0/3) were enriched in precursors. The monocyte-like macrophages (Macro_C1/2/4), which showed an angiogenesis phenotype, showed greater abundance in RRMM. Neutrophils were found exclusively in symptomatic MM, demonstrating elevated expression levels of TGFB1. Among the neutrophil subsets, those expressing myeloid checkpoints such as PILRA, SIRPA, and VSIR (Neutro_C0/3/4) exhibited increased prevalence in RRMM. In contrast, subsets expressing ARG1 and MMP9 (Neutro_C1/2) were more enriched in NDMM. Within the dendritic cell population, a specific subset characterized by high TGFB1 and VEGFA expression (cDC2_C1) was exclusively present in MM. Additionally, we observed an enriched proliferative dendritic cell subset in MM.

Conclusions:

We observed a notable pattern of cellular changes along the nBM-MGUS-SMM-NDMM-RRMM axis, indicating a progressive increase in immune suppression primarily driven by myeloid cells. These observations provide valuable insights into immune cell dynamics, their functional characteristics, and potential roles during disease progression and highlight potential targets for further investigation, especially in the context of RRMM.