Determinants of response of the BCMA/GPRC5D-dual-targeting T-cell redirecting trispecific antibody JNJ-5322 in multiple myeloma

Abstract Introduction BCMA- and GPRC5D-targeting bispecific antibodies (BsAbs) have shown unprecedented clinical efficacy in patients with heavily pretreated MM. However, not all patients respond and eventually most will develop progressive disease. We hypothesized that dual-targeting T-cell immunotherapies may improve efficacy by addressing the issue of heterogeneous target expression and prevent the development of resistance due to antigen escape. Aims We evaluated preclinical activity of JNJ-5322, a trispecific antibody (TsAb; BCMAxGPRC5DxCD3), in MM cell lines and in fully autologous bone marrow (BM) samples from MM patients. We also investigated the potential impact of T-cell and tumor characteristics, and diverse components of the immunosuppressive microenvironment on the efficacy of JNJ-5322. Methods BM samples from newly diagnosed (ND), daratumumab-naive relapsed/refractory (DARA-naive RR), and daratumumab-refractory (DARA-R) MM patients were analyzed for tumor- and immune cell composition and incubated with JNJ-5322 (0.032-5 nM) or mono-targeting BsAb controls (BCMAxCD3, GPRC5DxCD3). MM cell lysis was assessed by flow cytometry after 48 hours. Luciferase-transduced MM cell lines were incubated with JNJ-5322 (0.008-25 nM) and T-cells from healthy donors or MM patients, including BsAb-exposed patients. MM cell lysis was assessed by bioluminescence assay after 48 hours. Results BCMA and GPRC5D expression on the surface of MM cells in BM samples was highly heterogeneous between patients (n=41). Some tumors had a predominantly BCMA- and GPRC5D- co-expressing tumor cell population (across all 41 samples: median 73%, IQR 42-83), while other tumors also had populations only expressing BCMA (median 6%, IQR 3-22) or GPRC5D (median 11%, IQR 4-17), or a population of double negative MM cells (median 4%, IQR 1-13). We first investigated the effect of JNJ-5322 with T-cells from healthy donors using 5 MM cell lines which model the heterogenous expression of BCMA and GPRC5D on MM cells. JNJ-5322 was significantly more effective than mono-targeting BsAbs in BCMA+GPRC5D+ MM cell lines, and also efficiently eliminated single-positive MM cell lines. T-cells from BsAb-exposed patients induced lower lysis of the MM cell line RPMI-8226 in comparison to T-cells from BsAb-naïve patients (i.e., ND, DARA-naïve RR and DARA-R patients). Additionally, JNJ-5322 effectively induced MM cell lysis in all 32 BM samples from MM patients (median lysis with 0.4 nM: 59%). This was accompanied by T-cell activation, cytokine secretion, degranulation and granzyme B release. BCMA- and GPRC5D-targeting BsAbs induced significantly lower MM cell lysis (median with 0.4 nM: 7% and 15%, respectively), T-cell activation and T-cell degranulation in BM samples in comparison to JNJ-5322, underscoring the benefit of dual-antigen targeting. There was no difference in JNJ-5322-mediated lysis of MM cells from ND, DARA-naïve RR (median of 4 prior lines of therapy), or DARA-R patients (median of 4 prior lines of therapy). Importantly, tumor cell target expression was a critical determinant of response, shown by superior MM cell lysis in samples with higher than median baseline proportion of GPRC5D/BCMA double-positive MM cells, when compared to lower GPRC5D/BCMA expression (25.6 fold shift in EC50). In addition, lower MM cell lysis, T-cell activation and T-cell degranulation were found in samples with a high baseline frequency of PD-1+ or TIGIT+ CD8+ T-cells. The presence of high-risk cytogenetic abnormalities del(17p), t(4;14), and/or t(14;16) did not influence JNJ-5322 efficacy. Finally, we compared the efficacy of JNJ-5322 versus the two mono-targeting BsAbs in combination. Simultaneously targeting BCMA and GPRC5D with JNJ-5322 resulted in enhanced MM cell lysis, compared to the combination of the mono-targeting BsAbs in both MM cell lines and in 32 patient BM samples (median lysis at 0.4 nM: 59% versus 20%, respectively). This highlights the benefit of simultaneously engaging both BCMA and GPRC5D with JNJ-5322 over combining BsAbs targeting either BCMA or GPRC5D, possibly by increasing the overall avidity of the interaction. Conclusion We show that dual-antigen targeting with JNJ-5322 is more effective than mono-targeting BsAbs alone or in combination. Tumor-related factors, such as BCMA/GPRC5D expression, and differences in the BM microenvironment, including the frequency of PD-1+ or TIGIT+ CD8+ T cells, contribute to the variability in ex vivo response to JNJ-5322.