Prognostic Molecular Stratification in Relapsed/Refractory Multiple Myeloma - Results of the Pomalidomide Mukseven (NCT02406222) Biomarker Trial

Background

Treatment of relapsed/refractory multiple myeloma (RRMM) remains challenging as durable remissions are achieved in patient sub-groups only. Identifying patients that are likely to benefit prior to or early after starting relapse treatments remains an unmet need. MUKseven is a trial specifically designed to investigate and validate biomarkers for treatment optimization in a 'real-world' RRMM population.

Design

In the randomized multi-center phase 2 MUKseven trial, RRMM patients (≥2 prior lines of therapy, exposed to proteasome inhibitor and lenalidomide) were randomized 1:1 to cyclophosphamide (500 mg po d1, 8, 15), pomalidomide (4 mg days 1-21) and dexamethasone (40 mg; if ≥75 years 20 mg; d1, 8, 15, 21) (CPomD) or PomD and treated until progression. All patients were asked to undergo bone marrow (BM) and peripheral blood (PB) bio-sampling at baseline, cycle 1 day 14 (C1D14, on-treatment) and relapse. For biomarker discovery and validation, IGH translocations were profiled by qRT-PCR, copy number aberrations by digital MLPA (probemix D006; MRC Holland), GEP by U133plus2.0 array (Affymetrix), PD protein markers by IHC and PB T-cell subsets by flow cytometry for all patients with sufficient material.

Primary endpoint was PFS, secondary endpoints included response, OS, safety/toxicity and biomarker validation. Original planned sample size was 250 patients but due to a change in UK standard of care during recruitment with pomalidomide becoming available, a decision was made to stop recruitment early.

Results

In total, 102 RRMM patients were randomized 1:1 between March 2016 and February 2018. Trial entry criteria were designed to include a real-world RRMM population, permitting transfusions and growth factor support. Median age at randomization was 69 years (range 42-88), 28% of patients had received ≥5 prior lines of therapy (median: 3). Median follow-up for this analysis was 13.4 months (95% CI: 12.0-17.5). 16 patients remained on trial at time of analysis (median number of cycles: 19.5; range 8-28).

More patients achieved ≥PR with CPomD compared to PomD: 70.6% (95% CI: 56.2-82.5%) vs. 47.1% (CI: 32.9-61.5%) (P=0.006). Median PFS was 6.9 months (CI: 5.7-10.4) for CPomD vs. 4.6 months (CI: 3.5-7.4) for PomD, which was not significantly different as per pre-defined criteria. Follow-up for OS is ongoing and will be presented at the conference.

High-risk genetic aberrations were found at following frequencies: t(4;14): 6%, t(14;16)/t(14;20): 2%, gain(1q): 45%, del(17p): 13%. Non-high risk lesions were present as follows: t(11;14): 22%, hyperdiploidy: 44%. Complete information on all high-risk genetic markers was available for 71/102 patients, of whom 12.7% had double-hit high-risk (≥2 adverse lesions), 46.5% single-hit high-risk (1 adverse lesion) and 40.8% no risk markers, as per our recent meta-analysis in NDMM (Shah V, et al., Leukemia 2018). Median PFS was significantly shorter for double-hit: 3.4 months (CI: 1.0-4.9) vs. single-hit: 5.8 months (CI: 3.7-9.0) or no hit: 14.1 months (CI: 6.9-17.3) (P=0.005) (Figure 1A). GEP was available for 48 patients and the EMC92 high-risk signature, present in 19% of tumors, was associated with significantly shorter PFS: 3.4 months (CI: 2.0-5.7) vs. 7.4 (CI: 3.9-15.1) for EMC92 standard risk (P=0.037).

Pharmacodynamic (PD) profiling of cereblon and CRL4^CRBN ubiquitination targets (including Aiolos, ZFP91) in BM clots collected at baseline and C1D14 is currently ongoing. Preliminary results for the first 10 patients demonstrate differential change of nuclear Aiolos (Figure 1C), with a major decrease in Aiolos H-scores in 7/10 patients from baseline to C1D14 and reconstitution at relapse.

T-cell PB sub-sets were profiled at baseline and C1D14 by flow cytometry. Specific sub-sets increased with therapy from baseline to C1D14, e.g. activated (HLA-DR+) CD4+ T-cells, as reported at last ASH. CD4+ T-cell % at baseline was associated with shorter PFS in these analyses in a multi-variable Cox regression model (P=0.005).

PD and T-cell biomarker results will be updated and integrated with molecular tumor characteristics and outcome.

Discussion

Our results demonstrate that molecular markers validated for NDMM predict treatment outcomes in RRMM, opening the potential for stratified delivery of novel treatment approaches for patients with a particularly high unmet need. Additional immunologic and PD biomarkers are currently being explored.

View largeDownload slide

View largeDownload slide

Close modal