Assessing the prognostic impact of intermediate outcomes on survival time in relapsed/refractory multiple myeloma (RRMM) patients (pts) via a robust and clinically interpretable quantitative procedure.

e17743 Background: Assessing the impact of intermediate outcomes on overall survival (OS) is crucial in RRMM. The conventional approach, using the hazard ratio (HR), is difficult to interpret clinically as hazard is not a risk measure. When the proportional hazards assumption is violated, the HR lacks clinical meaning as a contrast measure between two groups. An attractive alternative is to use the mean survival time at study end (restricted mean survival time [RMST]). The difference and ratio of two RMSTs can be used to quantify the prognostic impact of intermediate outcomes on OS by landmark (LM) analysis. This procedure is robust (model-free) and yields a clinically meaningful interpretation. Methods: 669 RRMM pts randomized to bortezomib or dexamethasone in the phase III APEX trial (NCT00048230) were considered for inclusion. Pts alive at a pre-specified LM time (0.5 yrs from randomization) were assessed for OS until trial end (N = 568). Quality of response (QOR, best response by LM), was categorized ...