Predictors of T Cell Expansion and Clinical Responses Following B-Cell Maturation Antigen-Specific Chimeric Antigen Receptor T Cell Therapy (CART-BCMA) for Relapsed/Refractory Multiple Myeloma (MM)

Background: Relapsed/refractory (rel/ref) MM is associated with progressive immune dysfunction, including reversal of CD4:CD8 T cell ratio and acquisition of terminally-differentiated T cell phenotypes. BCMA-directed CAR T cells have promising activity in MM, but the factors that predict for robust in vivo expansion and responses are not known. In a phase 1 study of CART-BCMA (autologous T cells expressing a human BCMA-specific CAR with CD3ζ/4-1BB signaling domains) in refractory MM patients (median 7 priors, 96% high-risk cytogenetics), we observed partial response (PR) or better in 12/25 (47%) (Cohen et al, ASH 2017, #505). Recently, we demonstrated in CLL pts receiving CD19-directed CAR T cells that certain T cell phenotypes prior to generation of the CAR T product were associated with improved in vivo expansion and clinical outcomes (Fraietta et al, Nat Med 2018). We thus sought to identify pre-treatment clinical or immunological features associated with CART-BCMA expansion and/or response.

Methods: Three cohorts were enrolled: 1) 1-5 x 10⁸ CART cells alone; 2) cyclophosphamide (Cy) 1.5 g/m² + 1-5 x 10⁷ CART cells; and 3) Cy 1.5 g/m² + 1-5 x 10⁸ CART cells. Phenotypic analysis of peripheral blood (PB) and bone marrow (BM) mononuclear cells, frozen leukapheresis aliquots, and phenotype and in vitro kinetics of CART-BCMA growth during manufacturing were performed by flow cytometry. CART-BCMA in vivo expansion was assessed by flow cytometry and qPCR. Responses were assessed by IMWG criteria.

Results: Responses (≥PR) were seen in 4/9 pts (44%, 1 sCR, 2 VPGR, 1 PR) in cohort 1; 1/5 (20%, 1 PR) in cohort 2; and 7/11 (64%, 1 CR, 3 VGPR, 3 PR) in cohort 3. As of 7/9/18, 3/25 (12%) remain progression-free at 11, 14, and 32 months post-infusions. As previously described, responses were associated with both peak in vivo CART-BCMA expansion (p=0.002) as well as expansion over first month post-infusion (AUC-28, p=0.002). No baseline clinical or MM-related characteristic was significantly associated with expansion or response, including age, isotype, time from diagnosis, # prior therapies, being quad- or penta-refractory, presence of del 17p or TP53 mutation, serum hemoglobin, BM MM cell percentage, MM cell BCMA intensity, or soluble BCMA concentration. Treatment regimen given before leukapheresis or CART-BCMA infusions also had no predictive value. We did find, however, that higher CD4:CD8 T cell ratios within the leukapheresis product were associated with greater in vivo CART-BCMA expansion (Spearman's r=0.56, p=0.005) and clinical response (PR or better; p=0.014, Mann-Whitney). In addition, and similar to our CLL data, we found that a higher frequency of CD8 T cells within the leukapheresis product with an "early-memory" phenotype of CD45RO-CD27+ was also associated with improved expansion (Spearman's r=0.48, p=0.018) and response (p=0.047); Analysis of manufacturing data confirmed that higher CD4:CD8 ratio at culture start was associated with greater expansion (r=0.41, p=0.044) and, to a lesser degree, responses (p=0.074), whereas absolute T cell numbers or CD4:CD8 ratio in final CART-BCMA product was not (p=NS). In vitro expansion during manufacturing did associate with in vivo expansion (r=0.48, p=0.017), but was not directly predictive of response. At the time of CART-BCMA infusion, the frequency of total T cells, CD8+ T cells, NK cells, B cells, and CD3+CD56+ cells within the PB or BM was not associated with subsequent CART-BCMA expansion or clinical response; higher PB and BM CD4:CD8 ratio pre-infusion correlated with expansion (r=0.58, p=0.004 and r=0.64, p=0.003, respectively), but not with response.

Conclusions: In this study, we found that CART-BCMA expansion and responses in heavily-pretreated MM patients were not associated with tumor burden or other clinical characteristics, but did correlate with certain immunological features prior to T cell collection and manufacturing, namely preservation of normal CD4:CD8 ratio and increased frequency of CD8 T cells with a CD45RO-CD27+ phenotype. This suggests that patients with less dysregulated immune systems may generate more effective CAR T cell products in MM, and has implications for optimizing patient selection, timing of T cell collection, and manufacturing techniques to try to overcome these limitations in MM patients.