The Bone Marrow Microenvironment of Multiple Myeloma Long-Term Survivors at Single Cell Resolution

To date, multiple myeloma (MM) remains an incurable disease with only a minor fraction of patients experiencing long-term remission (LTR) over 7 years after a single therapy line. Myeloma cells strongly depend on the interaction with their bone marrow microenvironment (BMME), but the molecular and cellular adaptations of the BMME to active MM disease and the role of the immune system in patients experiencing LTR remain poorly understood. In order to gain a global and detailed understanding of the BMME, we profiled over 290.000 BM resident cells from 11 MM patients in LTR 7 to 17 years after first-line therapy and 3 healthy donors using droplet-based single-cell RNA sequencing. Paired BM samples collected at initial diagnosis enabled us to analyze the changes from first diagnosis to the state of LTR in individual patients.

At initial diagnosis, we observed significant remodeling of the T cell, NK cell and myeloid compartments which was only partially reversible upon LTR. In- depth analysis of the CD8+ T- cell compartment revealed an unknown immunophenotype of myeloma-associated CD8+ T (MAT) cells expressing key mediators of T cell dysfunction such as NR4A2. The amino acid transporter LAT1 (SLC7A5), which is known to be critical to maintain the activation state of T cells, and the surface marker CD6 were specifically expressed by MAT cells. We validated the existence of this novel T cell immunophenotype in an independent group of 30 MM patients using FACS. The number of MAT cells was associated with myeloma cell burden indicating that MAT-cells might be an indirect marker for tumor load within the BM. The clinical and prognostic meaning of this population is currently under investigation. Within the myeloid compartment, we detected myeloma associated myeloid (MAM) cells at initial diagnosis that were only present in case of active disease. These MAM cells shared features of immunosuppression, inflammation and migration as well as chemotaxis hinting towards a phenomenon of immune cell recruitment to the site of disease.

At the LTR stage, 6 of 11 patients were still in complete remission (CR), while 5 patients presented with detectable disease activity after having achieved a CR. In the CR-group we observed a healthy-like state in the BMME but still detected a myeloma associated imprint even in minimal residual disease negative patients. Within the CD8+ T cell compartment, this imprint included a higher metabolic activity in the naïve T cell compartment as well as a higher grade of cytotoxicity within the effector T cell and NK cell compartment. These observations might reflect a state of active immunosurveillance in MM patients to maintain CR at the LTR state. In contrast, 5 LTR patients with detectable disease activity lost the CR associated immune signature approaching a BMME remodeling similar to initial diagnosis. Increasing disease activity over the next 2 years within this patient population showed that we captured a state of early relapse. This enabled us to describe programs specific to early relapse in comparison to the full- blown disease state. In this context, an increase in plasmacytoid dendritic cells (pDCs), key players in the production of interferons, was observed at the stage of early relapse hinting towards a role for pDCs in establishing the inflammatory changes in the BMME upon resurgence of disease.

Together, this study provides a comprehensive overview of the molecular and cellular patterns within the BMME that underlie active myeloma disease as well as LTR in MM. We describe novel immunophenotypes of T cells and myeloid cells associated with myeloma cell burden within the BM. At the stage of LTR, our results reveal how patients in CR approach a healthy-like state but still preserve an imprint of MM potentially associated with active immunosurveillance. Finally, this study deepens the understanding how BMME remodeling evolves from an early phase of relapse to a full-blown disease.