Adaptive design of VALOR, a phase III trial of vosaroxin or placebo in combination with cytarabine for patients with first relapsed or refractory acute myeloid leukemia.

Abstract

TPS201 Background: Median overall survival (OS) for patients with relapsed or refractory AML remains short (3-6 mos) with current treatments. Vosaroxin (formerly voreloxin) with cytarabine had a promising activity/tolerability profile in phase II in this population (N=69): median OS 7.1 mos, combined CR rate 29%, current median leukemia-free survival 14.4 mos, 30-day all-cause mortality 3% (Stuart Chem Found Symp 2010), supporting phase III trial design. VALOR (NCT01191801) is a phase III, randomized, controlled, double-blind, multinational trial of the efficacy and safety of vosaroxin and cytarabine versus placebo and cytarabine in patients with first relapsed or refractory AML; OS is the primary endpoint. VALOR enrollment initiated in December 2010. Methods: Adaptive Design: Base case assumed 40% improvement in median OS, from 5 mos for the placebo + cytarabine arm to 7 mos for the vosaroxin + cytarabine arm (hazard ratio, HR=0.71); 375 OS events are required to provide 90% power at a 2-sided alpha of 0.05 with 1 interim look allowed at 50% OS events. The interim result is partitioned into zones based on conditional power: futility, unfavorable, promising, favorable and efficacy. The Data and Safety Monitoring Board (DSMB) can recommend stopping early for futility or efficacy or completing per base case if the interim result is either unfavorable or favorable. However, if the result falls into the promising zone, the DSMB can recommend a one time sample size adjustment, increasing the final number of events and sample size by 50% (from 450 to 675 evaluable patients) in order to detect a smaller but relevant treatment effect (HR > 0.71) (Mehta, Pocock, Stat Med 2010). For example, if the HR is 0.77 rather than 0.71, the 50% increase in sample size boosts the conditional power within the promising zone from approximately 70% to slightly greater than 90%. A key advantage of VALOR’s design is the 50% increase is recommended only if the interim result demonstrates that the increase substantially reduces the risk of failing to detect a clinically meaningful OS benefit. This design satisfies the statistical and operational requirements of the FDA Draft Guidance on Adaptive Design.