Carfilzomib 56mg/m2 Twice-Weekly in Combination with Dexamethasone and Daratumumab (KdD) Versus Daratumumab in Combination with 8 Cycles of Bortezomib and Dexamethasone (DVd); A Matching-Adjusted Indirect Treatment Comparison

Introduction: Lenalidomide is increasingly used for the treatment of newly diagnosed multiple myeloma. However, most patients experience relapse during treatment or after prolonged exposure to lenalidomide, therefore, there is an emerging unmet need for new, highly efficacious lenalidomide-sparing regimens at relapse. CANDOR is an ongoing phase 3 randomized clinical trial (RCT) that compares the lenalidomide-free KdD regimen with carfilzomib 56mg/m2 twice-weekly with dexamethasone (Kd) in patients with relapsed and/or refractory multiple myeloma (RRMM). Based on the primary analysis, KdD was associated with a 37% reduction in the risk of progression or death compared to Kd (HR: 0.63, p=0.001).(Dimopoulos et al. 2020 Lancet) Besides the trial-based comparator Kd, DVd is considered as an additional relevant comparator. DVd is the most similar lenalidomide-free triplet regimen comprising a proteasome inhibitor, daratumumab, and dexamethasone. DVd was assessed in the phase 3 CASTOR RCT and demonstrated superior PFS versus 8 cycles (24 weeks) of bortezomib in combination with dexamethasone (Vd) (HR: 0.31, p<0.001).(Weisel et al. 2019 ASH) In the absence of a direct head-to-head trial comparing these treatments and an assessment suggesting that network meta-analysis is not feasible, matching-adjusted indirect comparisons (MAIC) were used (Signorovitch et al. 2010) to assess the efficacy of KdD relative to DVd. MAIC leverages all available data and compares outcomes across balanced populations.

Method: Baseline characteristics and PFS data from the CANDOR and CASTOR trials were used in the analyses. Outcomes were compared in three steps. First, after aligning patient selection criteria of both trials, KdD patients in CANDOR were matched to the average characteristics of DVd patients in CASTOR using matching-adjusted indirect treatment comparisons methodology. The matching variables reflected tumor stage, disease burden, and treatment history, and were generally consistent with validated prognostic markers used in stratifying patients into different risk groups. Second, the published Kaplan-Meier PFS curves for DVd were digitized to generate virtual patient-level data that reconstructed the curves. Third, Cox regression models were fitted to the matched CANDOR data and the reconstructed virtual patient-level CASTOR data to estimate the efficacy of KdD versus DVd. The outcome measure was the hazard ratio (HR) for PFS. Outcomes were also compared for the first 24 weeks and for the period beyond 24 weeks to assess the impact of stopping Vd in DVd after 24 weeks on the HRs. Scenario and subgroup analyses were conducted to assess the robustness of results.

Results: KdD patients in the CANDOR trial who were proteasome inhibitor-refractory (n=102) were removed from the data so that the KdD population was aligned with that of DVd in the CASTOR trial. Following the matching procedure, the effective sample size for KdD reduced to 82. In matched populations, PFS at 12 months was 73.1% for KdD and 59.9% for DVd. The MAIC results showed that treatment with KdD decreases the risk of progression or death versus DVd, with an overall HR of 0.61 (95% CI: 0.45-0.82). The time-dependent analysis demonstrated similar PFS for the two treatments for the first 24 weeks and a larger benefit for KdD over DVd after 24 weeks. Similar results were obtained in scenario analyses; in bortezomib-exposed patients the overall HR was 0.46 (95% CI: 0.33-0.64).

Conclusions: After adjusting for cross-trial differences, KdD was associated with a significant reduction in the risk of progression or death compared with DVd. Although based on an approach in agreement with methods guide, the results are associated with inherent limitations due to the decreased sample size for KdD, and due to known (longer follow-up for DVd, different duration of proteasome inhibition) or unobserved differences in the trials (potential unobserved heterogeneity in patient characteristics). A comparison of overall survival was not considered due to immature data in CANDOR. The results are in line with those of the head-to-head ENDEAVOR trial that directly compared carfilzomib with bortezomib (HR for PFS for Kd vs Vd 0.53; 95% CI: 0.44–0.65) supporting the clinical validity of this study. The present analysis suggests that KdD improves outcomes compared with DVd in patients with RRMM and may provide a rationale for a preferential treatment.