CAR T Cells Contend with Myeloma in the Bone Marrow Microenvironment

B-cell maturation antigen (BCMA)–targeted CAR T-cell therapies are one of the most promising new treatments for multiple myeloma (MM), with two recent approvals for autologous T-cell products: idecabtagene vicleucel (Ide-cel) and ciltacabtagene autocel (Cilta-cel). Response rates of 73% and 97%, respectively, have been seen in patients with relapsed or refractory (R/R) disease (1, 2). Impressive rates of 27-month progression-free survival (PFS) and overall survival at 54.9% and 70.4%, respectively, in patients treated with Cilta-cel were reported (3). However, late relapses are still occurring, and in the absence of much longer follow-up, MM is still considered an incurable disease. Thus, the focus remains on maximizing PFS and enabling patients to have extended treatment-free periods with good quality of life.

The duration of response to BCMA-targeted CAR T-cell therapies has been linked to the depth of remission (1, 2). Additionally, the loss of CAR T-cell persistence has been shown to precede disease progression in most patients treated with Ide-cel (1). Factors influencing CAR T-cell persistence include the proportion of less-differentiated naïve and stem cell memory CD8 T cells (CD45RO⁻CD27⁺CD8⁺) within the apheresis product, which are capable of robust expansion (4). An immune response to the CAR construct may conversely limit CAR expansion. The development of anti-CAR antibodies was associated with treatment failure in patients given a second infusion of Ide-cel (1). Although antigen-negative tumor escape is a recognized cause of relapse following CD19-targeted CAR T-cell therapy, to date, only case reports have described the loss of BCMA on malignant plasma cells leading to disease progression (5). The role of the tumor microenvironment in permitting response or driving resistance to CAR T cells in MM has not yet been explored in depth.

In this issue of Blood Cancer Discovery, Dhodapkar and colleagues performed a high-dimensional single-cell analysis of the MM tumor microenvironment (TME) following BCMA targeting CAR T-cell therapy (6). They investigated the impact of tumor and hematopoietic cell lineages on PFS in patients treated with an experimental BCMA CAR T-cell therapy (clinical trial NCT02546167; refs. 4, 6). In this phase I study, autologous BCMA CAR T cells (anti-BCMA scFv, 4-1BB, CD3ζ) were given to patients in three cohorts. Cohort 1 patients received a dose of 1 to 5 × 10⁸ CAR T cells without prior lymphodepletion, cohort 2 received lymphodepletion with 1.5 g/m² cyclophosphamide and a cell dose of 1 to 5 × 10⁷ CAR T cells, cohort 3 received the same cyclophosphamide lymphodepletion regimen but a higher dose of 1 to 5 × 10⁸ CAR T cells (4). Cells were infused over 3 days with split dosing. The overall response rate was 48% (12 of 25) in all patients treated, with higher response rates seen in cohort 3 (64%, 7 of 11; ref. 4). Although the response rate observed in this trial and with this BCMA-directed CAR T-cell product was lower than that reported for Ide-cel or Cilta-cel, the focus of this study was on the patients whose disease did respond to therapy.

Responding patients were split into two groups with those who remained in remission for >6 months being classified as having a long PFS and those who experienced disease progression in under 6 months having a short PFS. Bone marrow was sampled pretreatment and then again at day 28 and month 3 following infusion. Utilizing CITE-seq, single-cell transcriptomics, T-cell receptor (TCR) sequencing, and mass cytometry, the authors investigated differences in bone marrow cell populations between the two groups (6). They identified associations with PFS not only in MM cells and CAR T cells but also in non–CAR-expressing T cells, myeloid cells, and dendritic cells residing in the bone marrow (Fig. 1).

In terms of the primary tumor, baseline plasma cells from patients with a short PFS had a gene-expression signature consistent with less mature B cells. Malignant cells from long PFS patients expressed a gene signature consistent with more mature plasma cells; they also expressed higher levels of interferon response genes and of the antigen target, BCMA. Thus, antigen density on the tumor may have played a role in enhancing PFS. On the T-cell side, CAR T cells from patients with a short PFS had higher expression of lytic granule genes at day 28 compared with CAR T cells from patients with a longer PFS. This suggests that having a higher proportion of terminally differentiated effector T cells by day 28 may be detrimental to the durability of action.

The authors did not only focus on the direct interaction between tumor cells and CAR T cells. In terms of endogenous immunity, markers of terminal differentiation were present in the baseline bone marrow T cells in patients with a short PFS, while a more diverse TCR repertoire was present in patients with a longer PFS. At day 28 following infusion, long PFS patients had a higher proportion of T cells in the bone marrow compared with patients with a short PFS. On the antigen-presenting side of endogenous immunity, the authors also found that a higher proportion of dendritic cells expressing Clec9a in the bone marrow following CAR T-cell infusion was associated with a longer PFS. Shorter PFS was also associated with a higher proportion of BAFF- and PD-L1–expressing myeloid cells, and with expression of myeloid-derived suppressor cell-associated genes. Taken together, these data suggest that endogenous immunity, including antigen presentation, and the natural T-cell receptor repertoire play a role in sustained antitumor responses following BCMA CAR T-cell therapy.

This is the first study to systematically examine the hematopoietic cell lineages associated with the duration of response to BCMA-targeted CAR T cells in the TME. The authors show that endogenous T cells, myeloid cells, and dendritic cell populations may influence or be associated with the duration of response to BCMA CAR T cells. This implies that response is not just a direct interaction between effector and target cells, but rather that a favorable TME is required for sustained antitumor activity. It raises several intriguing questions, such as whether myeloid cells are inhibiting BCMA CAR T cells. This could be via expression of inhibitory ligands such as PD-L1, or by providing prosurvival signals to plasma cells via the BAFF/BCMA pathway. Dendritic cells, which are potent antigen-presenting cells, may be playing an important role in priming non–CAR-expressing T cells. It also raises the possibility of whether endogenous T cells have a direct antitumor effector function. Further work is required to delineate the mechanisms by which endogenous bone marrow immune cells could affect BCMA CAR T-cell function.

Of note, most patients in this study received single-agent cyclophosphamide as lymphodepletion or no lymphodepletion at all prior to cell infusion. In contrast, fludarabine and cyclophosphamide chemotherapy is typically given prior to the licensed BCMA CAR T-cell products. Lymphodepletion is administered to create a favorable environment for the CAR T cells to expand and will affect the relative proportion of bone marrow cell populations present. Fludarabine likely alters the relationship between myeloid cells, endogenous T cells, and dendritic cells, with an overall effect of reducing endogenous immune responses.

The results of this study may also not be generalizable to other BCMA CAR T cells due to differences in manufacturing processes, BCMA binding affinity, and other features of the CAR construct that can affect cell activation, cytotoxicity, and persistence. Beyond hematopoietic cell lineages, other TME cells may influence the outcome. Bone marrow mesenchymal stromal cells and cancer-associated fibroblasts derived from myeloma patients have been shown to impair CAR T cell–mediated killing in in vitro experiments (7, 8), but the effects of these cells were not captured in this experimental design. This current study excluded patients who were primary refractory from the analysis, and it is not known whether similar mechanisms that lead to a shorter duration of response may also contribute to primary refractory cases.

Careful dissection of the immune networks at play in orchestrating the antitumor effects of CAR T cells will be instrumental in designing the next generation of immunotherapies. To date, the role of endogenous immunity and epitope spreading has been unclear in the setting of CAR T cells, but the study by Dhodapkar and colleagues provides the first hints (and the data set) that endogenous immunity may play a role in sustaining antitumor responses that are initiated by direct antitumor CAR T cells.

C.E. Graham reports nonfinancial support from Celgene and grants from Servier outside the submitted work. M.V. Maus reports personal fees from 2Seventy Bio outside the submitted work; in addition, M.V. Maus has a patent for CAR T cells for multiple indications, including multiple myeloma pending.