Combined PD-1 and Tim-3 Blockade Improves the in Vitro Anti-Myeloma Activity of T Cells of Patients on Talquetamab

Background: In the recent phase I/II MonumenTAL-1 trial (NCT03399799, NCT04634552), the CD3xGPRC5D bispecific antibody (BiAb) Talquetamab (Tal) showed ≥70% response rate at the recommended phase 2 doses (RP2D) in heavily pretreated multiple myeloma (MM) patients (Chari A et al., NEJM 2022). However, there is still a need to identify tumor intrinsic and immune microenvironmental determinants of response depth and durability as well as devising strategies to overcome resistance to improve outcomes for MM patients.

Method: We identified 40 patients who received Tal monotherapy at the RP2D (16 pts 400ug/kg weekly and 24 pts 800ug/kg biweekly) at Mount Sinai and had bone marrow (BM) samples collected at baseline, cycle 3 day 1 (C3D1) and progression of disease (POD). We performed whole exome sequencing (WES), RNA sequencing and immunohistochemistry (IHC) on tumor samples to look for GPRC5D loss, mutations, RNA and protein expression at baseline and POD. We performed spectral flow using a panel of 39 markers to immunophenotype and measure cytokine production of the immune microenvironment at different treatment time points. We further established an in vitro assay to measure the cytotoxic capability of patient T cells obtained from different time points against the GPRC5D-expressing MM1S cell line in the presence of Tal with and without nivolumab and sabatolimab, targeting PD-1 and Tim-3, respectively.

Results: The 40 patients had a median age of 66.8 years and 50% were female. They received a median of 6 prior therapy lines (range: 2-17), with 83% being triple-class refractory and 48% penta-drug refractory. Moreover, 20% had received anti-BCMA CART prior to Tal but none were treated with another BiAb. 48% of patients had high-risk cytogenetics and 25% had extramedullary disease at baseline. Patients were followed for a median of 33.5 months and had a median progression-free survival (PFS) of 10.4 months. 83% of patients achieved a partial response (PR) or better, and 40% of patients had achieved a complete response (CR).

RNA sequencing on 25 baseline BM samples revealed that patients with a deeper response (≥VGPR) and longer response (PFS ≥11 months) had a higher median GPRC5D RNA expression (p=0.025 and p=0.0013, respectively). IHC of 26 baseline BM samples showed a wide range of GPRC5D staining from 0% up to 90% of total CD138+ plasma cells in the BM, and 3 out of 4 baseline BMs of non-responders had no GPRC5D expression. IHC of 6 POD samples showed that only 3 out of 6 had no GPRC5D expression. GPRC5D RNA expression and % of GPRC5D positive MM cells by IHC were positively correlated: R² = 0.658; p=0.002).

Out of 25 baseline exome samples, only 1 patient had a mono-allelic GPRC5D loss. Out of 5 POD exome samples, only 1 patient had a concurrent mono-allelic GPRC5D loss along with 2 GPRC5D mutations p.W199* and p.E197fs.

When looking at the baseline BM immune microenvironment, we found that patients with a longer PFS had a higher percentage of CD8+ T effector memory (T_em) cells (p=0.003), while those with a shorter PFS had more CD8+ T_EMRA (p=0.03). Furthermore, patients with a longer PFS, had a trend towards a higher percentage of gamma-delta T cells in the BMs at C3D1. We next co-cultured whole BM or whole PB from baseline, C3D1 and POD with MM1S cell line at a 1:1 ratio along with Tal and found that the cytotoxic capability of T cells did not correlate with clinical response. Yet, we found that the in vitro cytotoxic capability of T cells does decrease over time, whereby it was lower at C3D1 and POD compared to the T cells obtained from patients at baseline. In order to overcome this T cell exhaustion, we added nivolumab, sabatolimab or the combination of the two and found that T cell killing capability is significantly increased by up to two folds at baseline and C3D1 but not at POD. Of note, the improvement in T cell killing with the addition of Sabatolimab was positively correlated with the fold change increase of Tim-3 after in vitro stimulation of patients’ T cells (R²= 0.615; p=0.005).

Conclusion: Response and resistance to Tal stem from both tumor intrinsic and extrinsic factors. There is a correlation of GPRC5D RNA expression with depth and duration of response. Presence of CD8+ T_em cells at baseline resulted in longer PFS, whereas more CD8+ T_EMRA cells predicted shorter PFS. While our in vitro model has limitations, it shows that targeting the checkpoint markers PD-1 and Tim-3 may improve potency of BiAbs.