Safety and Preliminary Efficacy Results from a Phase Ib/II Study of Cobimetinib As a Single Agent and in Combination with Venetoclax with or without Atezolizumab in Patients with Relapsed/Refractory Multiple Myeloma

Introduction: MAPK pathway mutations are present in more than 50% of patients (pts) with relapsed/refractory multiple myeloma (RRMM; Xu et al. Oncogenesis 2017; Kortum et al. Blood 2016). While MEK inhibitors have demonstrated limited single-agent activity in RRMM, combination with BCL-2 inhibition with or without PD-L1 inhibition may improve efficacy, by shifting the apoptotic balance towards cell death and increasing CD8+ T cell recognition and enhancing the anti-tumor immune response. We conducted a Phase Ib/II study of cobimetinib (C) as a single agent and in novel combinations with venetoclax (V) with or without atezolizumab (A) in pts with RRMM (NCT03312530).

Methods: Pts with RRMM who had received 3–5 prior therapies, including a proteasome inhibitor and an immunomodulatory drug, were enrolled. Safety run-in (SRI) cohorts, SRI 1 and SRI 2, evaluated the safety of C+V and C+V+A, respectively. Pts were then randomized 1:2:2 to C (Arm A), C+V (Arm B), or C+V+A (Arm C). Pts received C 60 mg PO daily on Day (D) 1–21 (Arm A), or C 40 mg PO daily on D1–21 and V 800 mg PO daily on D1–28 (SRI 1, Arm B) and A 840 mg IV on D1 and D15 (SRI 2, Arm C) on 28-day cycles. Responses were assessed using IMWG 2016 criteria. Pharmacokinetic (PK) analyses were performed. Exploratory biomarker assessments included cytogenetics, mutational analysis, and PD-L1 and BCL-2 family expression (RNAseq).

Results: As of May 6 2020, 49 pts were enrolled (SRI 1, n=6; SRI 2, n=6; Arm A, n=6; Arm B, n=16; Arm C, n=15). Median age was 65 years (range 44–79), 63% were male, 94% had ECOG PS 0–1, and 47% had ISS II/III disease. Median prior lines of therapy was 4 (range 3–5), with prior ASCT in 43% and prior daratumumab in 41% of pts. 24% of pts had high-risk cytogenetics (del(17p), t(4;14), t(14;16)), 20% were t(11;14) positive, 51% had RAS mutations (KRAS, NRAS, BRAF, NF-1), and 31% had high PD-L1 expression.

The most common adverse events (AEs) of any grade (Gr; C/C+V/C+V+A) were diarrhea (33%/82%/91%), nausea (17%/50%/67%), anemia (17%/46%/57%), neutropenia (0/32%/57%), thrombocytopenia (0/27%/33%), blood creatine phosphokinase increase (17%/32%/24%), and rash (50%/14%/33%). The most common Gr 3–4 AEs were neutropenia (0/14%/38%), anemia (0/23%/24%), thrombocytopenia (0/18%/24%), and pneumonia (0/14%/14%). The most common serious AE was pneumonia (0/23%/14%). Two AEs of tumor lysis syndrome (TLS) were reported, with 1 meeting Howard criteria for laboratory TLS (t(11;14) pt on C+V, no dose interruption). Treatment discontinuation due to AEs was noted in 17%, 18%, and 14% of C, C+V, and C+V+A pts, respectively. At a median follow-up of 13.4 months (mo), there were 27 (55%) deaths overall (4 [67%]/14 [64%]/9 [43%]). The leading cause of death was progressive disease (PD; 2 [50%]/11 [79%]/7 [78%]). Two treatment-related deaths were observed (C+V: pneumonia; C+V+A: general physical health deterioration in the setting of PD). Overall survival (OS) for C, C+V, and C+V+A was 12.9, 12.4, and 23.3 mo respectively, and comparable with historical OS in 3L+ pts (Usmani et al. Oncologist 2016; Kumar et al. Leukemia 2017). No clinically relevant drug–drug interactions were identified between C, V, and A based on preliminary PK analysis.

The overall response rate (ORR; ≥partial response) was 0% (0/6) for C, 27% (6/22) for C+V, and 29% (6/21) for C+V+A (Figure A). Compared with the non-t(11;14) pts, responses to C+V and C+V+A were higher among t(11;14) pts, with an ORR of 50% (2/4) and 100% (5/5), respectively. Median duration of response was 11.5 mo for C+V (range 2–18) versus 5.1 mo for C+V+A pts (range 2–15), with individual pt characteristics likely contributing to the observed difference. Prolonged disease stability and durable responses were noted in a subset of pts, irrespective of t(11;14) and/or RAS mutation status (Figure B/C). At data cut-off, 4 pts remain on treatment, 3 of whom are t(11;14) positive and 1 who is t(11;14) negative and RAS wild-type but BCL2/BCL2L1 (BCL-X_L) high (log2≥2.3).

Conclusions: C, C+V and C+V+A demonstrated manageable safety and tolerability in heavily pretreated pts with RRMM. Preliminary efficacy results show moderate activity of the combinations in all-comers and higher activity in t(11;14) pts, supporting a biomarker-driven approach in the development of V in MM. Further biomarker analyses may provide insight into the pt subgroups that may benefit from combined MAPK and BCL-2 inhibition, and the contribution of PD-L1 inhibition.