Multiple myeloma is a blood-borne cancer with its cell-of-origin in plasma cells or progenitors of plasma cells. In contrast to the extensively studied biology of myeloma cells, the bone marrow microenvironment for myeloma cells remains an uncharted field and poorly understood [2]. Current therapies of myeloma display increasing efficacy in reducing the tumor burden but nonetheless often inevitably lead to resistance and disease relapse, which makes myeloma, among many other blood and solid tumors, treatable but not yet curable [3]. Changes in the myeloma bone marrow microenvironment has been implicated in myeloma pathogenesis, therapy resistance and relapse [4-6]. Therefore, better understanding of the bone marrow microenvironment and the interaction of myeloma cells with such microenvironment might reveal new aspects of myeloma pathogenesis and facilitate development of therapeutic interventions correspondingly. By utilizing cutting-edge technology of single-cell RNA-seq (scRNA-seq), de Jong et al. [1] provided detailed and exhaustive analyses of the human non-hematopoietic as well as hematopoietic niches in the bone marrow of myeloma patients and age-matched non-cancer control individuals, which covered nearly all resident cell types in the bone marrow including myeloma cells, diverse immune cells, and particularly stromal cells. Across these cell types residing in the bone marrow that potentially constitute the microenvironment for myeloma cells, mesenchymal stromal cells (MSCs) are of particular interest, at least partly because they actively secrete cytokines (e.g. IL-6) that have been implicated in myeloma tumorigenesis [7,8].

To specifically define and analyze the non-hematopoietic bone marrow niche, de Jong et al. [1] excluded hematopoietic cells, erythroid lineage cells and plasma cells using flow cytometry sorting and simultaneously validated the identity of non-hematopoietic cells by gating for combinations of cell surface markers. The RNA-seq analyses of these non-hematopoietic cells revealed MSCs as five closely related clusters (MSC1-5), endothelial cells, sinusoidal endothelial cells, and osteolineage cells. Uniform manifold approximation and projection (UMAP) of myeloma patients and non-cancer control individuals showed no discernible differences for all the non-hematopoietic cell types with the exception that MSC1 and MSC2 are dramatically enriched in MM patients. This is especially remarkable for MSC1 which constitute 0.4% of total MSCs in non-cancer control individuals that increased to 24% and 33% of total MSCs for hyperdiploid-associated myeloma and translocation-associated myeloma, respectively. Interestingly, UMAP showed a more separable and more delineated contour of MSC1/2 for translocation-associated myeloma than hyperdiploid-associated ones, for which the implication was unknown and warrants further investigation. Interestingly, Ranked Gene Set Enrichment Analysis (GSEA) showed a dramatic enrichment of genes in the ‘TNFα signaling via NFκB’ pathway in MSC1/2 as compared to MSC3/4/5, which mostly comprise inflammatory genes such as chemokines, cytokines, cytokine receptors, etc., suggesting that MSC1/2 represents a myeloma-specific inflammatory signature population. In this sense, de Jong et al. [1] coined “inflammatory MSC” (iMSC) for MSC1/2. The emergence of iMSCs suggested that myeloma has an inflammatory bone marrow microenvironment, which was further confirmed by associated increase of protein levels of the cytokine IL-6, and the chemokines CXCL8 and CCL2 in the bone marrow plasma. That myeloma bone marrow has an inflammatory stromal microenvironment is the major finding in this study. Nonetheless, it’s worth noting that there was no correlation between iMSC percentage and protein levels of inflammatory cytokine/chemokines, suggesting that either a small fraction of iMSCs secrete such cytokine/chemokines or different subsets of iMSCs secrete different cytokine/chemokines, among other possibilities.

A tantalising possibility is that iMSCs might co-localize in the bone marrow with myeloma cells in close vicinity or even direct contact. Testing this possibility with fluorescent immunohistochemistry necessitates defining specific markers for iMSCs. However, unfortunately scRNA-seq analyses failed to identify any markers that were exclusively specific to iMSCs, among a number of genes dramatically enriched in them. de Jong et al. [1] thus exploited this gene expression enrichment and validated the scRNA-seq data by flow cytometry analyses that CD44 was greatly enriched in iMSCs. So, by combining CD44 with a pan-MSC marker CXCL12, they designated CXCL12⁺CD44⁺ cells as a surrogate for iMSCs. de Jong et al. [1] found that CXCL12⁺CD44⁺ iMSCs co-localized with CD138⁺ myeloma cells in bone marrow biopsies from myeloma patients but not that from non-cancer control individuals. However, this interpretation should be taken with caution because myeloma patients had vastly increased numbers of both iMSC and CD138⁺ myeloma/plasma cells in their bone marrow so that detection of their co-localization could be merely a result of mathematical chance. The definitive answer to this interesting possibility of iMSC/myeloma cell co-localization might be attainable by leveraging spatial transcriptomics or the state-of-the-art Stereo-seq [9-12].

iMSC might either represent a cell activation state that is transient, stochastic, and reversible or arise as a result of lineage differentiation that is stable and programmed. Consistent with Ranked Gene Set Enrichment Analysis (GSEA) which showed a marked enrichment of genes in the ‘TNFα signaling via NFκB’ pathway in MSC1/2, de Jong et al. [1] found that IL-1β protein level was selectively increased in myeloma bone marrows plasma [TNF protein level was too low for reliable measurement]. In vitro, both IL-1β and TNF induced an iMSC-like transcriptome in ex vivo-expanded bone marrow MSCs from non-cancer control individuals as well as of myeloma patients, suggesting that iMSC might be a cell activation state rather than a differentiation stage. Moreover, the activation of iMSC transcriptome were more pronounced for MSC1 than MSC2 according to the expression levels of iMSC-enriched inflammatory genes. Consistently, pseudotime analysis of the trajectory of transcriptional changes across MSC1 to MSC5 also suggested that MSC2 is an intermediate state of inflammatory stimulation that eventually leads to MSC1. In this sense, it’s arguable that it would be more appropriate to ascribe iMSC exclusively to MSC1 rather than to a collection of both MSC1 and MSC2. However, collectively such evidence still renders it speculative that iMSCs in myeloma patient bone marrow might be indeed induced by such cytokines in vivo. Other cytokines might instead participate in or even be responsible for this process.

de Jong et al. [1] divided the hematopoietic cells in the bone marrow niche into CD38^- and CD38⁺ cells for each cell type, as relevant to recent clinical trials using anti-CD38 antibodies [13]. The hematopoietic counterpart of the bone marrow microenvironment comprises a wide spectrum of various lymphocytes and subtypes, diverse myeloid cells, as well as plasma/myeloma cells. Although the myeloma bone marrow incurred relative expansion of few subsets of these cells (i.e. IFN-responsive T cells, CD8⁺ stem cell memory T cells (T_scm) cells and a subset of NK cells), the overall abundance and distribution of the hematopoietic cells in the bone marrow was not altered, which is in sharp contrast to that of the non-hematopoietic niche where MSCs displayed dramatic changes. Given that the immune cells are plausible sources of IL-1β and TNF that were implicated in activating iMSCs in vitro, de Jong et al. [1] found that TNF expression was significantly increased in certain NK and effector T cells as well as in the IFN-responsive T cells and T_scm cells that showed selective enrichment in the myeloma bone marrow. This implies that these two subsets of T cells might contribute to the activation of iMSC through secreting TNF. However, the TNF protein level cannot be reliably measured in the bone marrow plasma, leaving this possibility loose ends. In contrast, although IL-1β protein level was shown to be enriched in the myeloma bone marrow plasma, it showed no differential transcription in myeloid cells that were the dominant source of IL-1β among all the cells analyzed. In conclusion, this reinforces that whether IL-1β or TNF contribute to the activation of iMSC in vivo remains an open question. Reliable measurement of TNF in the bone marrow plasma, identification of elusive cell types that mediate the increased IL-1β, and surveying other potential cytokine as in vivo activators will help resolve this problem.

Last but not least, de Jong et al. [1] addressed the question as to whether the myeloma bone marrow inflammation landscape persisted after induction therapies of myeloma. In order to include samples that technically preclude scRNA-seq analysis for transcriptome analysis, they performed bulk RNA-seq analysis instead and reproducibly confirmed that the iMSC transcriptome signature was retained in bulk RNA-seq analysis. Surprisingly, inflammatory iMSC transcriptome, increased cytokine/chemokine levels such as IL-1β, as well as enrichment of differentially expressed inflammatory genes in the ‘TNFα signaling via NFκB’ pathway all persisted after successful induction therapies, even up to the point that patients were minimal residual disease negative. The mechanisms for such ‘memory’ of inflammatory microenvironment in the myeloma bone marrow remain to be unveiled.

In summary, this study provided a thorough single-cell RNA-seq analysis of the bone marrow microenvironment (including the hematopoietic as well as the non-hematopoietic counterparts) in the myeloma context and found that the myeloma bone marrow is associated with inflammation as best manifested by the emergence of ‘inflammatory MSCs’. Myeloma bone marrow inflammation persisted even after successful induction therapies, which might be of clinical significance that future studies need to uncover. While bearing these important findings in mind, the author would bring into attention that several shortcomings of this study. First, the numbers of samples from myeloma patients and non-cancer control individuals were limited, which might compromise the robustness of the findings. Second, bone marrow aspirates might be sampled from different bones or different sites of the bones: for the non-cancer control individuals it was taken from the sternum or femur heads whereas for the myeloma patients it was not specified but usually it is taken from the pelvic bones in the clinic. Third, the non-cancer control individuals were nonetheless not healthy individuals. They had either cardiac diseases or osteoporosis and therefore might be under drug treatments with anti-inflammatory effects. All these confounding factors could skew the observations or interpretations of the findings. Thus, more following work with better control over these aspects needs to be done to reproduce and reexamine the observations /conclusions made by de Jong et al. [1].

This study posed a series of unprecedented addressable questions that warrant further investigation. First, as discussed above, the cytokines that directly activate iMSCs are yet to be definitively identified, as well as the cell source of such cytokines. Second, upon the advent of the cutting-edge technologies of spatial transcriptomics (esp. Stereo-seq), iMSCs need to be spatially resolved in order to survey whether they co-reside with myeloma cells or other cells in the bone marrow microenvironment. This will also help define the spatial cell landscape of the myeloma bone marrow microenvironment in general. Third, are iMSCs supportive, suppressive, or largely neutral with to myeloma progression? Investigation into whether iMSCs are present in MGUS and smoldering myeloma may shed new light on this question. More mechanistic studies, especially in mouse models of myeloma, will enable manipulation of iMSC and other components in the microenvironment possible. Most importantly, the significance of iMSCs needs to be understood in the genetic (e.g. hyperploid-driven vs translocation-driven myeloma) as well as clinical settings (treatment vs relapse). All these endeavors may expand the horizon of our understanding of myeloma and proceed collectively step by step towards its curability.