Allogeneic Anti-BCMA CAR T Cells Are Superior to Multiple Myeloma-derived CAR T Cells in Preclinical Studies and May Be Combined with Gamma Secretase Inhibitors

Multiple myeloma remains an incurable plasma cell malignancy despite the rapidly evolving treatment landscape. Chimeric antigen receptor T cells targeted against BCMA have recently shown great promise in relapsed refractory multiple myeloma; however, all patients ultimately still progress from their disease. Lack of CAR T-cell persistence, impaired T-cell fitness in autologous CAR T-cell products and the presence of an immunosuppressive bone marrow (BM) microenvironment are contributory factors to treatment failure. We generated anti-BCMA CAR T cells from healthy donors (HD) and patients with multiple myeloma at different stages of disease to compare their T-cell profile, fitness, and cytotoxic activity in preclinical studies. We also used an ex vivo assay with multiple myeloma BM biopsies from distinct genomic subgroups to test the efficacy of HD-derived CAR T cells in a clinically relevant model. HD volunteers showed increased T-cell counts, higher CD4/CD8 ratio, and expanded naïve T-cell population compared with patients with multiple myeloma. After anti-BCMA CAR T-cell production, patients with relapsed multiple myeloma had lower frequencies of CAR+ T cells, decreased central memory phenotype, and increased checkpoint inhibitory markers compared with HD-derived products, which compromised their expansion and cytotoxicity against multiple myeloma cells in vitro. Importantly, HD-derived CAR T cells efficiently killed primary multiple myeloma cells within the BM microenvironment of different multiple myeloma genomic subgroups and their cytotoxic activity could be boosted with gamma secretase inhibitors. In conclusion, allogeneic anti-BCMA CAR T cells are a potential therapeutic strategy for patients with relapsed multiple myeloma and should be further developed in the clinic. Significance: Multiple myeloma is an incurable cancer of the plasma cells. A new therapy with anti-BCMA CAR T cells — the patient's own T cells genetically engineered to find and kill myeloma cancer cells — has shown encouraging results. Unfortunately, patients still relapse. In this study, we propose to use T cells from HD volunteers, which have a stronger T-cell fitness, higher cancer killing capacity, and are ready to be administered when needed.


Introduction
Multiple myeloma is a plasma cell malignancy with a heterogeneous genomic profile that remains incurable despite the use of therapies such as alkylating cells, and not on hematopoietic stem cells, which makes it an ideal cancer associated antigen to target by immunotherapy (14,(16)(17)(18). Different anti-BCMA CAR T-cell products have been developed over recent years and several clinical trials have shown impressive response rates to anti-BCMA CAR T-cell therapy in patients with relapsed refractory multiple myeloma (19)(20)(21). However the median duration of response observed in these trials ranged from 8.8 to 22 months following CAR T infusion.
While the mechanisms of CAR T-cell resistance in multiple myeloma are still under investigation (22,23), downregulation of BCMA expression and lack of CAR T-cell persistence, due to CAR T-cell differentiation and exhaustion, are potential contributory factors. Clinical trials of autologous anti-BCMA CAR T cells in multiple myeloma have shown differences in T-cell profiles between responders and nonresponders, with T-cell exhaustion and terminal differentiation compromising the efficacy of the CAR T-cell products (24)(25)(26). Therefore an alternative strategy is to use allogeneic healthy donor (HD)-derived CAR T cells since they are generated from young HDs, whose T cells are likely to have a more favorable phenotype, superior immune fitness and cytotoxic ability (27).
In this study, we compared the T-cell profile and activity of anti-BCMA CAR T cells derived from HDs with those generated from patients with multiple myeloma of different genomic subtypes and disease stages, including patients who were eligible for CAR T-cell clinical trials. The cytotoxicity of anti-BCMA CAR T cells was evaluated in primary bone marrow (BM) cultures from patients with multiple myeloma enabling the potential immunosuppressive effects of the BM microenvironment to be explored. In addition, the effects of enhancing BCMA expression, using a gamma secretase inhibitor (GSI), on CAR T-cell cytotoxicity were investigated.

Patients and HD Volunteers
Peripheral blood and BM samples from HD volunteers and patients with multiple myeloma were obtained from the Haemato-Oncology Tissue Bank at King's College London under the terms of the research ethics protocol reference HR-17/18-5515 in accordance with the Declaration of Helsinki and after approval by an institutional review board and with the Human Tissue Authority license number 12223. Written informed consent forms were signed by all the patients prior to sample collection.

Anti-BCMA CAR T-Cell Production
Peripheral blood mononuclear cells (PBMC) were isolated using Histopaque 1077 density gradient media (Sigma, catalog no. RNBG8589) and centrifuged at 2,000 rpm for 30 minutes without any brake. PBMCs present in the cellular buffy coat were collected, washed twice with sterile PBS, and frozen down for future processing and analysis.
After 48 hours, activated T cells were transduced with a lentiviral construct encoding the BCMA-5 R2 CAR (28) containing an anti-BCMA scFv with 4-1BB and CD3z domains, and the media further supplemented with recombinant human IL2. On day 5, human CD3/CD28 beads were removed from each sample using an EasySep magnet and the individual samples were plated in complete X-Vivo 15 media with human IL2. Cells were cultured for 9 days and frozen on day 14. Untransduced T cells (UT) from each of the donors were used as negative controls and underwent the same manufacturing steps as the corresponding anti-BCMA CAR T cells with the exception of the lentiviral transduction step.

BM Sample Processing
Fresh BM samples were transferred into 50 mL Falcon tubes and incubated with Pharm lysis buffer (BD Biosciences, catalog no. 555899) for 15 minutes at room temperature. The samples were then centrifuged and washed twice with sterile PBS, before passing through a 100-μm mesh (Miltenyi Biotec). The samples were then incubated in RPMI1640 complete media (with 10% FBS, 1× penicillin/streptomycin, and 1× l-glutamine) and stored in the tissue culture incubator at 37°C, 5% CO 2 . Overall cell viability and multiple myeloma cell percentage were then measured by FACS as described below. BM samples were always used fresh and the available volume varied from sample to sample, which limited the number of samples used in specific assays.

Ex Vivo BM Cytotoxicity Assay
BM samples were stained with a multiple myeloma antibody panel to quantify BM cell viability and the percentage of multiple myeloma primary cells using the following mouse monoclonal anti-human antibodies: anti-CD56
Multiple myeloma primary cells present in patient BM samples were identified using the following gating strategy: side scatter/CD45, CD138 pos /CD38 high , CD38 high /CD45 low and CD56/CD19 neg . The anti-BCMA CAR T-cell killing percentage was quantified as: [(% multiple myeloma cell lysis cocultured with anti-BCMA CAR T cells − % multiple myeloma cell lysis cocultured with UT cells)/% spontaneous multiple myeloma cell lysis].

Data Availability Statement
The data generated in this study are available upon request from the corresponding author.

HDs have Higher T-Cell Counts, CD4/CD8 T-Cell Ratio, and Naïve T-Cell Phenotype Compared with Patients with Multiple Myeloma
PBMCs were collected from 9 HDs and 16 patients with multiple myeloma, their immune profile assessed by flow cytometry and the antitumor efficacy of anti-BCMA CAR T-cell products generated from them compared.  Fig. 1C).

Anti-BCMA CAR T Cells Generated from HDs have a Higher Transduction Rate and Increased Proportion of Central Memory Cells Compared with CAR T Cells Derived from Patients with Relapsed Multiple Myeloma
To evaluate differences in phenotype and efficacy between HD-derived and multiple myeloma-derived CAR T cells, a second-generation fully human anti-BCMA CAR construct (ref. 28; Fig. 2A) was used to generate anti-BCMA CAR T cells from HDs and patients with multiple myeloma, subcategorized into newly diagnosed (ND), early relapsed and late relapsed multiple myeloma patients. FACS analysis of anti-BCMA CAR T cells after 14 days of manufacturing showed a higher percentage of CAR-transduced cells in HD-derived samples compared with late relapsed multiple myeloma samples (64.5 ± 4% vs. 45 ± 5.2%, P < 0.05), although CAR expression as measured by mean fluorescence intensity (MFI) was similar between the different groups ( Fig. 2B-D). The CD4/CD8 ratio of the final CAR T-cell product was not significantly different between any of the groups (Fig. 2E). HD CAR T cells showed a higher percentage of central memory CD8 T cells compared with rMM CAR T cells (39.7 ± 4% vs. 20.5 ± 7%, P < 0.05) which may have an impact on their cytotoxic capacity.
The proportions of stem cell memory, effector memory, and effector cells did not appear to be significantly different between HD, newly diagnosed multiple myeloma and relapsed multiple myeloma CAR T-cell products ( Fig. 2F and G). (Continued) and age < 30 (HD), newly diagnosed multiple myeloma patients not yet exposed to any treatment regimen, and relapsed multiple myeloma patients, who have undergone several lines of treatment as described. The third group is subdivided into early relapse (≤3 prior lines of therapy) and late relapse (>3 prior lines of therapy). B, T-cell counts per 10,000 live PBMCs after PBMC isolation from peripheral blood: HD (n = 9), newly diagnosed MM (n = 4), early relapsed MM (n = 9), and late relapsed MM (n = 4). C, CD4/CD8 T-cell ratios from the four groups analyzed above.

HD-Derived anti-BCMA CAR T Cells Have a Less Dysfunctional Phenotype and Show Superior Cytotoxicity Compared with Late Relapsed Multiple Myeloma-Derived CAR T Cells In Vitro
Given that significant differences were observed between HD and late relapsed multiple myeloma populations in the above analysis, and late relapsed multiple myeloma is the most relevant group for CAR T-cell treatment, we assessed the expression of the immune checkpoint molecules PD-1, TIGIT, LAG3, and TIM3 on HD-derived and late relapsed multiple myeloma-derived CAR T cells by flow cytometry. As shown in Fig. 3A-B, PD-1, TIGIT, TIM3, and LAG3 expression by MFI was not significantly different between HD and late relapsed multiple myeloma CAR T-cell products. However, the percentage of PD-1expressing CAR T cells was significantly increased in late relapsed multiple myeloma compared with HD CAR T-cell products (Fig. 3C, P < 0.05). Coexpression of immune checkpoints on T cells was also compared between HD and late relapsed multiple myeloma CAR T-cell products. Trends toward higher double positivity were noted, only achieving significance for PD-1 + LAG3 + cells (46.57 ± 5.1%) in late relapsed multiple myeloma compared with HD CAR T cells (26.0 ± 4.2%, P < 0.05; Fig. 3D and E). CD38 + CD101 + T cells, which characterize a permanently dysfunctional population of T cells, previously linked to post-stem cell transplant relapse in multiple myeloma (33), were present in 22.02 ± 7.6% of late relapsed multiple myeloma CAR T cells compared with 6.39 ± 0.9% in HD products (NS; Fig. 3E).
The cytotoxic properties of HD and late relapsed multiple myeloma anti-BCMA CAR T cells were compared using a 7-day in vitro rechallenge assay with U266 cells as the target (Fig. 3F). HD CAR T cells showed superior cytotoxicity compared with late relapsed multiple myeloma CAR T cells (54% vs. 12% target cell lysis, P < 0.05) and greater expansion against U266 cells (418.5 ± 124 vs. 48.4 ± 24 T cells, P < 0.05; Fig. 3G and H).

HD-Derived Anti-BCMA CAR T Cells Efficiently Target Multiple Myeloma Primary Cells in Ex Vivo Assays including BM Microenvironment Constituents from Patients with Multiple Myeloma
To evaluate the activity of HD-derived anti-BCMA CAR T cells in a clinically relevant model, we designed a whole BM ex vivo cytotoxicity assay using BM samples from patients with multiple myeloma to reflect the immunosuppressive effects of the BM tumor microenvironment (Fig. 4A). We tested 11 multiple myeloma samples from different patient subgroups (including those with standard and high-risk cytogenetics), age, disease stage, and prior treatment regimens as well as one age-matched normal control (MM#20; Fig. 4B). Primary multiple myeloma cells were identified within the BM using a gating strategy as described in Fig. 4C and ranged from 1% to 16% of total nucleated cells.
HD-derived anti-BCMA CAR T cells efficiently targeted multiple myeloma primary cells in a dose-response manner as shown in the representative example (Fig. 4D), with 52.5% cell lysis of MM#16 at an E:T ratio of 10:1 compared with 2.8% lysis using UT cells from the same donor. Across the different multiple myeloma samples (n = 11), the specific tumor lysis of HD-derived anti-BMCA CAR T cells ranged from 6.9% to 72.8% (Fig. 4E). The specificity of anti-BCMA CAR T cells toward multiple myeloma primary cells was confirmed by demonstrating no significant lysis of granulocytes, macrophages/monocytes, or T cells in the BM samples following coculture ( Fig. 4F and G).
The impact of the tumor microenvironment on CAR T-cell function was investigated by assessing the relationship between CAR T cell-mediated multiple myeloma cell lysis and the proportion of granulocytes, T cells, and monocytes/ macrophages in BM (Fig. 4H-J). No direct correlation between T cells, monocyte/macrophage, or granulocyte percentage and tumor lysis was observed (Fig. 4I). However, there was a significant difference in granulocyte percentage (63% vs. 48%, P < 0.05) between BM samples shown to have low versus high tumor lysis (defined as tumor lysis below or above mean of 32.9%; Fig. 4J).

Anti-BCMA CAR T-Cell Cytotoxicity Does Not Correlate with BCMA Expression on Primary Multiple Myeloma Cells but can be Enhanced by Treatment with GSIs
Relative BCMA expression on primary myeloma cells (BCMA MFI on primary myeloma cells / BCMA MFI on T cells), ranged from 121 to 1,184 with a median of 549 (Fig. 5A). No direct correlation was found between relative BCMA expression on multiple myeloma cells and CAR T cell-mediated killing (R 2 = 0.34; Fig. 5B). BCMA is actively cleaved from the surface of multiple myeloma cells by the gamma secretase complex which results in reduced   Fig. 5F).

Discussion
Treatment with autologous anti-BCMA CAR T cells results in high overall response rates and deep remissions in relapsed refractory MM with a median duration of response between 8.8 and 22 months (19-21, 34, 35). Impaired T-cell fitness in patient-derived CAR T cells and an immunosuppressive BM microenvironment are thought to be the main causes for multiple myeloma relapse post CAR T-cell treatment. Patients with multiple myeloma are generally older with a median age of 72 years and have usually had multiple lines of treatment prior to being considered for CAR T-cell therapy. Both these factors have the potential to affect the fitness of their T cells. An alternative source of CAR T cells are young HDs (allogeneic) whose T-cell fitness is unlikely to be impaired.
Allogeneic CAR T cells offer the opportunity for patients with low T-cell counts who fail to generate an autologous CAR T-cell product to still be treated. In this study, we therefore compared the phenotype and antitumor activity of anti-BCMA CAR T cells derived from patients with relapsed refractory multiple myeloma with those derived from young HDs.
HDs were shown to have higher T-cell counts compared with patients with relapsed multiple myeloma. Furthermore, HDs were found to have a significantly higher CD4/CD8 T-cell ratio than patients with late relapse multiple myeloma.
In contrast, patients with untreated newly diagnosed multiple myeloma had a similar CD4/CD8 ratio to HDs suggesting that prior antimyeloma treatment may have an impact on the CD4/CD8 ratio. A decrease in CD4/CD8 ratio has previously been described in patients with multiple myeloma after treatment with daratumumab (36) and bortezomib (37). Dexamethasone has also been associated with lower CD4/CD8 ratio in premature infants (38). A higher CD4/CD8 T-cell ratio in the leukapheresis product has previously been shown to correlate with greater expansion of anti-BCMA CAR T cells in vivo (19,24) and better clinical responses (24). In line with these findings, some groups have modified their manufacturing process to generate CAR T-cell products with a predefined CD4/CD8 T-cell ratio (39) (NCT03430011) but it remains unclear whether this will result in improved clinical outcomes.
T-cell memory is also directly affected by age and MM treatment (36,40) and high levels of naïve and stem cell memory T cells in the leukapheresis product have been associated with higher expansion of anti-BCMA CAR T cells in vivo and better clinical responses (24,26). Our data demonstrates that at baseline HDs have a higher percentage of naïve T cells and lower proportion of effector T cells compared with patients with newly diagnosed multiple myeloma and relapsed multiple myeloma and importantly, postmanufacturing, have a higher percentage of central memory T cells in the final CAR T-cell product.
Clinical trials also suggest that high levels of exhaustion markers in autologous anti-BCMA CAR T cells correlate with poor clinical responses in patients with multiple myeloma (26). A detailed analysis of CAR T cells from patients with non-responder multiple myeloma revealed that these autologous products were mainly formed by activated effector T cells that expressed the checkpoint inhibitory markers, LAG3 and TIM3 (41). In fact, the expression of the checkpoint inhibitory molecules PD-1, TIM3, LAG3, and TIGIT has long been associated with lack of T-cell fitness and increased incidence of multiple myeloma relapse (33,(42)(43)(44)(45). We showed that PD-1 + T cells were significantly enriched in late relapsed multiple myeloma-derived CAR T cells compared with HDderived products, which may promote T-cell exhaustion and compromise their cytotoxic ability. PD-1 + LAG3 + double positive cells were also present in significantly higher numbers in late relapsed multiple myeloma-derived CAR T cells compared with HD-derived CAR T cells. These dysfunctional late relapsed multiple myeloma-derived CAR T cells were shown to proliferate less well and have reduced cytotoxicity against U266 multiple myeloma cells in vitro in comparison with HD-derived CAR T cells, which had a less exhausted phenotype.
For allogeneic CAR T cells to be used in the clinic there would need to be further genetic modification, for instance knockout of the endogenous T-cell receptor (TCR) and CD genes to prevent graft-versus-host disease and allow the use of an anti-CD52 antibody as part of lymphodepletion to prevent CAR T-cell rejection, respectively (28). Of note HD-derived CAR T cells used in this study had not undergone such genetic manipulation; however, we do not anticipate these gene editing modifications to significantly alter the T-cell fitness of the final CAR T-cell product (28). In the ongoing UNIVERSAL trial of TCR KO  anti-CD19 CAR T cells were previously shown to have superior functionality over lymphoma patient-derived CAR T cells (47). Furthermore clinical trials of allogeneic HD-derived TCR KO /CD KO anti-CD19 CAR T cells (UCART19) in relapsed refractory B-acute lymphoblastic leukemia demonstrated significant CAR expansion and impressive clinical responses (48,49) supporting the hypothesis that CAR T cells derived from HDs may have superior functionality to products made from patients. Persistence of allogeneic CAR T cells, however, appears to be of shorter duration than with autologous CAR T cells presumably due to host immune system-mediated rejection (49). Graft-versus-host disease was rarely seen in these early trials.
Another important factor that may contribute to multiple myeloma relapse is the presence of multiple myeloma niches within the immunosuppressive environment of the BM. Multiple myeloma cells are known to be highly dependent on the BM microenvironment and the physical interaction between multiple myeloma cells and BM cells is known to promote cancer resistance against different therapeutics (50,51). Having demonstrated cytotoxicity against BCMA-positive myeloma cell lines in vitro, we evaluated the efficacy of HD-derived anti-BCMA CAR T cells in a clinically relevant model using ex vivo whole BM biopsies from patients with multiple myeloma, to take into account the genomic complexity and immunosuppressive cellular environment seen in these patients. We showed that HD-derived anti-BCMA CAR T cells specifically target multiple myeloma primary cells ex vivo demonstrating activity against primary multiple myeloma cells with a range of different genomic abnormalities. No differences were seen among the genomic risk subgroups, which corroborates what has been described across anti-BCMA CAR T-cell clinical trials (19,20,24,26). With the exception of tumor burden, anti-BCMA CAR T-cell activity did not correlate with any other biological factor intrinsic to multiple myeloma primary cells, such as expression of CD38, CD138, CD56, or CD45. Importantly, we showed no correlation between the level of BCMA expression on multiple myeloma primary cells and anti-BCMA CAR T cell-mediated cytotoxicity. This is in keeping with other studies that have shown no correlation between baseline BCMA expression on multiple myeloma cells and clinical response to anti-BCMA CAR T-cell therapy (19,20,24).
Taking advantage of our ex vivo BM model, we explored whether the presence of other BM microenvironmental cells affects anti-BCMA CAR T-cell activity. An interesting observation in our study was the finding of significantly higher granulocyte numbers in BM samples where low CAR T-cell cytotoxicity was seen compared with BM samples with high cytotoxicity, although a direct correlation between granulocyte numbers and cytotoxicity was not observed. Neutrophils, a subset of granulocytes, have been associated with cancer progression, metastasis, and poor prognosis in solid tumors (52). Tumor-associated neutrophils are known to express checkpoint inhibitory molecules such as PD-L1 and CD86 that can directly inhibit T cells (53,54), and a dysfunctional neutrophil profile has been detected in patients with multiple myeloma (55)(56)(57). In addition, high levels of neutrophils at diagnosis and high neutrophil to lymphocyte ratio are associated with poor prognosis and overall survival in patients with multiple myeloma (56)(57)(58). Further evaluation is therefore needed to clarify whether granulocytes or a particular subset of granulocytes may have a negative impact on anti-BCMA CAR T-cell function.
BCMA loss has also been shown to be a mechanism of relapse post anti-BCMA CAR T-cell therapy due to clonal selection of low antigen-expressing cells, allelic deletion or mutations of the BCMA gene (24)(25)(26)59). BCMA expression is a highly dynamic process and is regulated by the protease gamma secretase, which releases soluble BCMA into the bloodstream (60). We investigated, in our ex vivo BM assay, combination treatment of multiple myeloma with allo-