Bone marrow adipose tissue is enriched in multiple myeloma cells

Abstract Introduction The role of the bone marrow microenvironment in multiple myeloma (MM) pathogenesis, progression and response to therapy has long been appreciated. However, while adipocytes constitute a significant amount of the bone marrow, most available data regarding interactions between adipocytes and myeloma cells is restricted to in vitro co-cultures of adipocyte- and myeloma cell lines. Here, we examined the bone marrow adipose tissue (BMAT) of newly diagnosed MM (NDMM), follow-up patients, precursor stages (monoclonal gammopathy of undetermined significance, MGUS), and non-tumor controls. Results First, we used biopsies from MGUS, NDMM, and follow-up patients to determine where MM cells localize in correlation to adipocytes. The distance between MUM1+ MM cells and the nearest adipocyte was measured and then compared to the distance between randomly distributed artificial MM cells and the same adipocytes. Our analyses revealed that in the majority of analyzed samples, MM cells localized significantly closer to adipocytes than a random distribution would imply (MGUS: 2/3, 66.7%; NDMM: 7/9, 77.8%; follow-up: 3/4, 75%). Next, we optimized a protocol for reliable and reproducible isolation of BMAT from freshly obtained iliac crest aspirates. Presence of BMAT was similar across the various groups (non-tumor control: 18/22, 82%; MGUS: 13/14, 93%; NDMM: 33/45, 73%; follow-up: 12/14, 86%). BMAT was subsequently enzymatically digested, disrupting adipocytes but at the same time releasing non-adipocytic cells residing within the BMAT, resulting in a single cell solution available for various downstream studies. Cells from this BMAT-fraction were compared to those in the remaining aspirate. Flow cytometric analyses revealed the enrichment of clonal MM cells in BMAT compared to respective aspirates in almost all samples analyzed (32 of 34) throughout MGUS, NDMM, and follow-up groups. Similarly, the frequency of non-clonal plasma cells was also higher in adipose tissue in all groups, including non-tumor controls. Transcriptomically, FABP5 gene was significantly increased in fat-associated MM cells, and metabolic Seahorse assays showed that non-adipocytic cells within the BMAT had an increased oxygen consumption rate and extracellular acidification rate, indicative of higher metabolic activity. In vitro, a 4-hour treatment of isolated fat-associated and aspirate-associated cells with bortezomib, melphalan, dexamethasone, or thalidomide revealed a high inter-patient heterogeneity, as in a subset of patients adipose-associated MM cells reacted less to treatment compared to MM cells from the aspirate (measured as percentage of viable MM cells after treatment). With flow cytometry we observed that the composition of the immune microenvironment was similar between BMAT and aspirate in non-tumor controls. However, while the frequency of NK cells, and especially CD16hiCD56lo cytotoxic NK cells, increased in the aspirates of MGUS, NDMM, and follow-up patients, this increase was absent from the BMAT of these patients. Discussion Combined, our data raise the possibility that BMAT might serve as a plasma cell-supporting sub-compartment within the bone marrow. The lack of increase of cytotoxic NK cells in BMAT in MGUS and MM patients (observed in the aspirate) suggests the appearance of a MM-tolerant immune niche within the BMAT, potentially influencing disease course and response to therapy. The specific molecular interactions between adipocytes, MM cells, and immune cells remain to be identified. Furthermore, as the amount (and presence) of BMAT is highly heterogenous per patient, the exact significance of fat-associated MM cells and their contribution to disease burden has to be assessed.