Indirect treatment comparison of belantamab mafodotin + pomalidomide + dexamethasone versus comparator regimens in lenalidomide-exposed relapsed/refractory multiple myeloma: A network meta-analysis

Abstract Introduction:The phase III DREAMM-8 study (NCT04484623) evaluated belantamab mafodotin, an antibody-drug conjugate targeting B-cell maturation antigen, plus pomalidomide (P) + dexamethasone (d; BPd) vs P + bortezomib (V) + d (PVd) in lenalidomide (len)-exposed patients with relapsed/refractory multiple myeloma (RRMM) who had ≥1 prior line of therapy (2L+). Improved progression-free survival (PFS; primary endpoint) was observed with BPd vs PVd (median PFS not reached for BPd vs 12.7 months for PVd; hazard ratio [HR] 0.52 [95% confidence interval: 0.37, 0.73], P<0.001). In the absence of head-to-head trials, an indirect treatment comparison (ITC) was used to compare PFS with BPd vs other regimens of interest for patients with len-exposed 2L+ RRMM. Methods: A systematic literature review identified randomized controlled trials (RCTs) (Jan 2008–Jan 2024) that reported PFS with regimens used in adults with 2L+ RRMM who had disease progression on/after their most recent therapy. Only RCTs that evaluated a regimen approved by the US Food and Drug Administration/European Medicines Agency, or which were of interest for health technology assessment, were included. In the len-exposed population, trials were linked by the treatment(s) they shared to form connected evidence networks (trials not part of the networks were excluded), and a fixed effects network meta-analysis (NMA) was conducted to compare PFS across RCTs. To validate the network structure, scenario analysis using an inverse probability of treatment weighting (IPTW) was performed that compared outcomes for patients receiving BPd (DREAMM-8) vs daratumumab (D) + Vd (DVd; DREAMM-7) and included in the network. Hazard ratios <1 indicated BPd was more effective than the comparator regimen, and 95% credible intervals (CrIs) that did not cross 1 indicated high probability that the treatment effect favored BPd. Deviance information criteria were used to estimate statistical fit for each model. Results: In the len-exposed population, the PFS network comprised 8 RCTs (including DREAMM-8) with comparator regimens: carfilzomib + dexamethasone (Kd), Kd + daratumumab (DKd), isatuximab + carfilzomib + dexamethasone (IsaKd), bortezomib + dexamethasone (Vd), DVd, PVd, and selinexor + Vd (SVd). The NMA found that BPd provided the longest PFS vs all included comparators (HR 0.29–0.86). Comparisons vs Vd (HR [CrI] 0.29 [0.20, 0.43]), high-dose (h)Kd (0.42 [0.26, 0.69]), SVd (0.46 [0.26, 0.83]), and PVd (0.52 [0.37, 0.73]) had 95% CrIs that did not cross 1. Comparisons vs Kd (HR [CrI] 0.58 [0.34, 1.01]), DKd (0.86 [0.46, 1.62]), IsaKd (0.73 [0.36, 1.47]), and DVd (0.73 [0.43, 1.25]) had 95% CrIs that crossed 1, but the trend favored BPd. The IPTW analysis using patient-level data showed BPd (DREAMM-8) provided longer PFS than DVd (DREAMM-7) (HR [CrI] 0.41 [0.25, 0.65]). When including results of the IPTW analysis of BPd vs DVd in the network in place of DREAMM-8 BPd vs PVd, BPd continued to provide the longest PFS vs all included comparators (HR 0.16–0.48), with 95% CrIs remaining below 1 for all comparisons except for BPd vs DKd (HR [CrI] 0.48 [0.22, 1.04]), indicating high probability of the treatment effect favoring BPd for these comparisons. Conclusions: In the absence of head-to-head randomized controlled trials, these ITC data suggested a high probability that PFS consistently favored BPd vs comparator regimens of interest in len-exposed patients with RRMM, with consistent findings in the IPTW analysis reducing uncertainty in the base-case NMA findings.