Impact of study partner type on primary endpoint volatility in a phase 3 registration trial of mild‐to‐moderate Alzheimer’s disease

Abstract Background In Alzheimer’s disease (AD) clinical trials, participants are required to enroll with a study partner (SP) who ensures compliance and completes outcomes assessing participant cognitive and functional ability. The SP role is commonly filled by spouses and other family members. Spouse and non‐spouse SPs differ in numerous ways, including average time in contact with the participant. These differences may impact volatility of SP‐reported assessments, which may in turn impact trial integrity. We quantified the impact of SP type on the variance of SP‐reported Disability Assessment for Dementia (DAD) scores. Method We analyzed data from the Bapineuzumab 301 (ApoE e4 non‐carriers) and 302 (ApoE e4 carriers) phase 3 trials in mild‐to‐moderate AD, provided by the trial sponsor, Janssen Research & Development, LLC for use in these analyses. We compared the sample variances of end‐of‐study DAD scores for spousal and non‐spousal SPs in each trial. 95% confidence intervals (CIs) for the corresponding parameters were computed via bootstrapping. We used similar methods to compare the variances in the difference between baseline and end‐of‐study DAD scores for SP types. Result Among the 1,232 and 1,121 participants analyzed from trials 301 and 302, 818 (66.4%) and 837 (74.7%) had spousal SPs at baseline, respectively (Table 1). On the other hand, 665 (665/973 = 68.3%) and 697 (697/924 = 75.4%) spousal SPs in each trial completed the end‐of‐study visit, respectively. In trial 301, the variance of end‐of‐study DAD scores for spousal SPs were observed to be 14% lower compared to non‐spousal SPs (spousal/non‐spousal = 630.41/746.55 = 0.84, 95% CI = [0.71, 1.03]). In trial 302 this relative decrease was observed to be 9%, though with less precision (spousal /non‐spousal = 610.85/673.86 = 0.91, 95%CI = [0.77, 1.12]). Neither trial had significant differences in spousal vs non‐spousal variances of the change‐from‐baseline DAD scores (Trial 301: spousal/non‐spousal = 378.10/425.17 = 0.91, 95%CI = [0.66, 1.22]; Trial 302: spousal/non‐spousal = 354.38/352.22 = 1.01, 95%CI = [0.77, 1.35]). DAD variance over time is depicted in Figure 1. Differences in the linear trend of DAD variance did not significantly differ between SP types in either trial (Trial 301 Est.: ‐1.74, 95%CI: [‐3.58, 0.23]; Trial 302 Est.: 0.94, 95%CI: [‐5.83, 4.31]). Conclusion SP type was not associated with variation in SP‐reported study endpoints in this study.