Impact of protocol design on between scale discrepancies in early AD clinical trials

Abstract

AbstractBackgroundWe previously presented the impact of protocol design on the MMSE changes observed in the screening period. Protocols requiring inclusion criteria be met at Screening alone had significantly more large MMSE changes between Screening and Baseline than did protocols where the criteria needed to be met at both Screening and Baseline. [Kott,2020]. In the current analysis we explore the impact of protocol design on the presence of between scale change discrepancies at Baseline and 6 months after randomization.MethodData were pulled from multi‐national clinical trials in early Alzheimer's Disease (AD) where MMSE and ADAS‐Cog and/or CDR were collected at Screening, Baseline and 6 months after baseline. Subjects were categorized into 2 groups depending on whether inclusion criteria were required at Screening alone or at Screening and Baseline. We defined discordance as occurring when at least 2 of the 3 instruments showed a clinically meaningful change from the prior visit but the changes were in opposite directions. Chi‐square test was used to compare the distribution of discordances between the protocols at Baseline and at month 6.ResultOur dataset consisted of 4,237 subjects with data available from Screening, Baseline and month 6. At Baseline, we saw significantly more instances of discordance from Screening in studies requiring criteria be met at Screening alone (5%) vs 2% when the requirement needed to be met at both Screen and Baseline (chi2 = 12.5, P < 0.001). At month 6, no difference between the protocol types was identified – (12.2% vs. 12.4%, chi2 = 0.01, p = 0.912).ConclusionOur analyses indicate a significant impact of protocol design on the presence of between scale change discrepancies. Protocols requiring criteria be met at Screening only had a significantly increased number of discordances at Baseline but this difference disappeared by month 6. Among the potential explanations for such between scale discordances, score manipulations to comply with inclusion criteria needs to be considered. These findings should be considered when designing protocol inclusion criteria. Further research is necessary to understand the impact of these discrepancies on drug‐placebo separation.