An argument for factorial designs in ALS trials

Background: ALS is uncommon. Design of high-efficiency phase II trials is of paramount importance to more quickly identify promising treatments appropriate for phase III trials in ALS.  Objective: To examine the value of the n-way factorial trial design for ALS. Methods: Using values of ALSFRS-R slope mean and variance as inputs, power analyses of factorial trials that use change in ALSFRS-R slope as outcome were performed, and further examined using simulation by outcome-weighted resampling of the PRO-ACT dataset. Arguments in favor of the factorial design are presented. Results: Assuming ALSFRS-R mean slope and standard deviation to be -1 and 0.84 points/month respectively, and an expected treatment benefit of 20% reduction in slope, a single-treatment futility-design randomized controlled trial (RCT) of 250 patients rejects 60% of ineffective treatments while accepting 95% of effective treatments for further testing. With a factorial design, a similar-sized RCT can realistically test 2 to 4 treatments simultaneously, translating to an efficiency of 62 to 125 patients per treatment, without needing to resort to historical controls. Necessary assumptions include (a) absence of pharmacokinetic and pharmacodynamic interactions between treatments, (b) preclinical safety of combinations, and (c) low prior probability of efficacy of treatments (reasonable in the ALS field). Advantages of factorial designs include (a) improved recruitment, because only a minority of patients (1/4th to 1/16th), or even none (without much loss of efficiency) need receive only placebo or sham treatment, (b) reduced overhead costs, and (c) greater investigator control over design and execution. Challenges include (a) constraints on choices of treatments that can be simultaneously tested, (b) potential confounding by unexpected interactions, and (c) need for agreement between study sponsors. Conclusions: Factorial trial designs have been successfully used to test multiple treatments simultaneously in other diseases, and may have utility in ALS.