Optimal Dual-Targeted CAR Construct Simultaneously Targeting Bcma and GPRC5D Prevents Bcma-Escape Driven Relapse in Multiple Myeloma

Multiple myeloma (MM) remains generally incurable, calling for the development of novel treatment strategies such as chimeric antigen receptor (CAR) T cell therapy. Most clinically tested CAR T cell therapies for MM target B cell maturation antigen (BCMA), but despite high response rates, many patients relapse (Raje N. NEJM 2019). BCMA negative-low MM cells are implicated as a reservoir preceding relapse (Brudno J. JCO 2018; Cohen A. JCI 2019). Our aims are to (1) evaluate whether upfront simultaneous targeting of an additional antigen such as G protein-coupled receptor class C group 5 member D (GPRC5D; Smith EL. Sci Trans Med 2019) can mitigate BCMA escape-mediated relapse in MM, and (2) compare dual targeting strategies to identify an optimal approach. Dual targeting for CD19/CD22 malignancies has been investigated, and multiple approaches are feasible; however, approaches have yet to be comprehensively compared head to head. Here, we compare 2 parallel production and 3 single-vector dual targeting strategies (Fig. 1A). To enhance clinical translatability, all strategies are built on the BCMA(125)/4-1BBζ CAR (BCMA scFv 125; Smith EL. Mol Ther 2018), which is currently under multi-center clinical investigation (NCT03430011; Mailankody S. ASH 2018). We confirmed that all dual targeted approaches lyse, proliferate, and secrete polyfunctional cytokines specifically in response to BCMA and GPRC5D mono- and dual-positive cell lines and/or primary patient MM aspirate samples. Activity in vivo was confirmed using the bone marrow-tropic OPM2 MM model (endogenously BCMA+GPRC5D+). In all experiments MM cells (2 x 106) were injected IV into NSG mice and engrafted/expanded for 14 days before treatment. A high dose of all dual targeted CAR T cell approaches (3 x 106 CAR+) induced long-term disease control (median overall survival (mOS) BCMA(125) non-signaling del control 32d vs other groups mOS not reached; p < 0.05). Prevention of latent BCMA escape-mediated relapse was evaluated by re-challenge of previously treated long-surviving mice with 2 x 106 OPM2 BCMA CRISPR KO (OPM2BCMA KO) cells at day 100 without re-treatment. While mice previously treated with BCMA(125)/41BBζ CAR T cells succumbed to OPM2BCMA KO disease, dual targeted approaches prevented OPM2BCMA KO growth (mOS BCMA mono-targeted arm 37d post re-challenge vs other groups mOS not reached; p < 0.05). To better recapitulate human MM and distinguish among dual targeting approaches, we modeled established BCMA heterogeneous disease by spiking 5-10% OPM2BCMA KO into bulk OPM2WT cells for injection. Each OPM2 population was modified to express distinct luciferases for simultaneous in vivo monitoring by bioluminescent imaging (BLI). Treatment with a moderate (5 x 105) dose of CAR T cells eradicated OPM2WT cells in all groups, but anti-GPRC5D CARs with CD28 co-stimulation, whether included within a mixed T cell population or in a bicistronic construct (Fig. 1A ii, iv), failed to control OPM2BCMA KO cells (Fig. 1B). Correspondingly, 4-1BB-only containing CAR T cells had increased in vivo expansion (2.1-4.1-fold increase CAR T cell BLI at day 7 over CD28 containing groups; p < 0.05). As this result is likely from greater activation-induced cell death in the CD28-containing approaches that was not rescued by 4-1BB, we later compared 4-1BB-only containing approaches (Fig. 1A i, iii, v). These 3 dual targeting approaches effectively controlled OPM2WT disease at moderate (1 x 106 CAR+) and low (2.5 x 105 CAR+) doses. However, when using a sub-therapeutic dose (2.5 x 105 CAR+) in the OPM2BCMA KO-spiked model, the tandem scFv-single stalk design was least effective in controlling OPM2BCMA KO disease (Fig 1C). At a dose that is sub-therapeutic to control OPM2WT disease (1 x 105 CAR+), the bicistronic dual 4-1BB design (Fig. 1A iii) was more effective in eradicating tumor compared with the parallel production approach (6-fold difference tumor BLI at day 28; p < 0.05). These results indicate that upfront dual targeting of BCMA/GPRC5D with CAR T cells can mitigate BCMA escape-mediated relapse in a model of MM. While parallel infusion of separate BCMA- and GPRC5D-targeted CAR T cells is effective, a single bicistronic vector encoding two 4-1BB-containing CARs avoids the practical challenges of parallel manufacturing, and uniquely may provide superior anti-MM efficacy. Figure Disclosures Fernandez de Larrea: Takeda: Consultancy, Honoraria, Research Funding; Celgene: Consultancy, Honoraria, Research Funding; Janssen: Consultancy, Honoraria; Amgen: Consultancy, Honoraria, Research Funding. Brentjens:JUNO Therapeutics: Consultancy, Patents & Royalties, Research Funding; Celgene: Consultancy. Smith:Celgene: Consultancy, Patents & Royalties, Research Funding; Fate Therapeutics and Precision Biosciences: Consultancy.