Umibecestat in the API Generation program: Worsening in RBANS and/or CDR on treatment reverses after wash‐out

Abstract Background Recruitment in the API Generation program and treatment with umibecestat were terminated in July 2019 after detection of an early signal of mild worsening in some measures of cognitive function with umibecestat. Participants were followed to assess the reversibility of these and other observed effects. We will summarize the key clinical results for umibecestat vs placebo during treatment and follow up. Method Following randomization to umibecestat or placebo, clinical assessments were scheduled at month 3, month 6 and every 6 month thereafter. After discontinuation of study medication, participants underwent a full evaluation within 3 months of the last dose. Data will be presented for visits on‐ and off‐treatment. For visits off‐treatment, results will be compared for visits occurring before vs after wash‐out of umibecestat (> 31 days after last dose). Analysis of cognitive scales in relation to age, sex, genotype and amyloid status, and a safety summary will also be presented. Result At termination about 650 participants had reached week 26. The final evaluation was performed on most of the 1561 ongoing participants (45% after wash‐out). Worsening was detected in a group‐level comparison at month 3 and month 6 in both studies for the two doses of umibecestat (effect size 0.2‐0.3 for RBANS Total, Immediate and Delayed Memory indices). Analysis from the visits conducted after wash‐out indicated that key cognitive findings seen before wash‐out were reversible: we saw no more between‐group differences in RBANS and no more imbalance in the proportion of participants having a >14‐point decrease in total RBANS or > 1‐point increase in CDR‐SB after wash‐out. Similar findings were also seen in homozygotes without elevated amyloid. Conclusion The negative effect of umibecestat vs placebo on RBANS and CDR‐SOB was detected as early as month 3 and was detectable throughout the treatment period, but was no longer present after wash‐out of umibecestat. This effect seems to be independent of the presence of brain amyloid. Confirmation of the contemporary reversibility of worsening in measures of cognition is an important result to inform development of future prevention therapies for Alzheimer’s disease (AD).