Pooled Modeling of Disease Progression in Two BACE1 Inhibitor Verubecestat Trials in Prodromal to Moderate Alzheimer’s Disease

The BACE inhibitor verubecestat (MK-8931) at doses of 12 and 40 mg demonstrated cognitive and functional decline relative to placebo in a Phase 3 trial of prodromal AD (APECS, NCT01953601) and no effect on progression in a Phase 3 trial of mild-to-moderate AD (EPOCH, NCT01739348). Disease progression modeling was used to investigate the temporal pattern of the cognitive and functional drug effects and to explore covariate, dose- or exposure-dependency in the pooled data from both trials. The key objective was to determine whether the rate of progression was slowed by BACEi but obscured by an early detrimental drug effect. Disease progression models for the clinical efficacy outcomes (ADAS-cog11, CDR-SB) were developed using similar methodology to an established ADAS-cog disease progression model (Ref). Parameters describing placebo response were first estimated. Akaike Information Criteria (AIC) were then used to select among potential drug effects (effects on progression rate and/or early detrimental offset) and to test for dependencies on dose, drug exposure, and patient characteristics. A simulation exercise confirmed that a true 25% or 50% reduction in the progression rate with treatment could be reliably identified in the setting of the early detrimental effect. Results were generally concordant across ADAS-cog11 and CDR-SB. An early symptomatic (detrimental offset) response was identified with verubecestat; no effect on rate of disease progression was supported. The magnitude of the drug effect offset was estimated 1.8 for ADAS-cog11 and 0.33 for CDR-SB. Covariates were generally concordant across measures with disease severity at baseline identified for both progression rate and detrimental offsetting. Additionally, age, weight, geographic region, and education effects were identified on progression. Threshold analysis of the detrimental offsetting effect suggested that the effect was present in less severe disease but absent in the most advanced AD patients (ADAS-cog11 baseline>24 and CDR-SB baseline > = 12). No dose or AUC dependency on the detrimental offsetting effect of verubecestat was found, indicating that the maximum effect is achieved at 12 and 40 mg doses. There was no evidence supporting a slowed rate of disease progression by BACEi after accounting for the early detrimental drug effect.