Temporal variability in outcomes of identical regimens across newly diagnosed myeloma trials: a systematic review

Cross-trial comparisons in oncology are inherently problematic yet commonly performed. Diagnostic changes in 2014 have led to the inclusion of patients previously with smoldering myeloma being reclassified as having myeloma. Informative censoring may also affect results. We aimed to assess the variability in progression-free survival outcomes for identical therapeutic regimens across multiple myeloma trials in which bortezomib, lenalidomide, and dexamethasone (VRd) or lenalidomide/dexamethasone (Rd) without transplant served as either a control or intervention arm and identify factors contributing to this variation.

This was a systematic review of randomised phase II/III trials including VRd or Rd arms without the use of transplant. Three databases (MEDLINE/PubMed, Embase, and the Cochrane Central Register of Controlled Trials) were searched between database inception to March 18, 2026, for randomised controlled trials (RCTs) involving patients with multiple myeloma (MM). ClinicalTrials.gov was additionally for completed studies with reported results. The search strategy incorporated both free-text and MeSH terms related to MM and applied filters for RCTs published in English. Kaplan-Meier curves were reconstructed to assess progression-free survival (PFS) and differential censoring patterns. As a sensitivity analysis, excess censored control patients were modeled after the longest-surviving patients, while excess censored intervention patients were treated as experiencing an event. Trials were categorised by enrolment timing relative to 2014 diagnostic criteria changes.

Seven VRd trials and seven Rd trials were included. Median PFS with VRd varied from 33.6 months to 54.3 months. The two studies with longest PFS (IMROZ: 54.3 months; CEPHEUS: 52.6 months) enrolled after diagnostic reclassification. Rd PFS ranged from 18.3 to 31.9 months. Differential censoring was present in most trials and mostly favoured control arm, though statistical differences if present, were preserved after adjustment for all trials other than DETERMINATION.

VRd and Rd demonstrate substantial outcome variability across trials, likely reflecting diagnostic reclassification, varying proportions of patients censored, improvements in supportive care, imaging advances, and stage migration rather than true therapeutic differences. These findings highlight the problematic nature of cross-trial comparisons and have implications for clinical interpretation. Future studies should explore this phenomenon using more novel regimens and in larger datasets.

None.