Joint Exposure‐Response Analysis of PET Amyloid Plaque Data from the BACE1 Inhibitor Verubecestat and Amyloid mAbs Trials in Prodromal to Moderate Alzheimer’s Disease

Abstract

AbstractBackgroundPositron‐emission tomography (PET) with amyloid radiotracers allows the quantification of pathological amyloid deposition in brain tissues. An exposure‐response (E‐R) analysis of individual study data of the β‐secretase‐1 inhibitor verubecestat and summary level data of four amyloid mAbs in patients with Alzheimer’s disease (AD) was conducted, to quantify the drug effects and plaque turnover rates and facilitate prediction of unstudied regimens.MethodIndividual (n = 188) verubecestat amyloid‐PET plaque data and exposures from APECS [Egan et.al, 2019] were pooled with summary. amyloid‐PET data and serum exposures from literature for aducanumab [Sevigny J, et al. 2016], donanemab [Lowe SL, et al. 2021, Lily ADPD, 2021], gantenerumab [Klein G, et al. 2019, Portron A, et al. 2020] and lecanemab [Swanson CJ, et al. 2021]. Amyloid plaque levels in SUVR units were converted to Centiloid [Klunk WE, et al. 2015]. All data were fit simultaneously in a joint model by nonlinear mixed‐effects modelingResultAn indirect response (turnover) model with verubecestat inhibiting plaque formation, and amyloid mAbs stimulating plaque removal well represented the available data (Figure 1). All plaque time course data from natural progression (control arms), β‐secretase‐1 inhibition and amyloid mAb treatment arms were well described by a joint model. The plaque turnover was estimated to be 0.00028 per day, corresponding to a half‐life of plaque of ∼ 6.8 years. Verubecestat AUC50 was estimated as 0.313 uM*h, commensurate to a 93.5% reduction in plaque formation at the typical 40 mg exposure and consistent with the >60% reductions in CSF Aβ40 and 42 observed in APECS. Aducanumab 10 mpk Q4W, donanemab 1400 mg Q4W, lencanemab 10 mpk Q2W, and gantenerumab 1200 mg SC Q4W were estimated to results in 13.5‐, 16.4‐, 16.6‐, and 12.6‐fold increases in plaque removal rate, respectively.ConclusionThe joint turnover model, linking drug exposure with amyloid plaque load, was adequate to describe natural progression and treatment response, and allows for estimation of the underlying turnover rate. This approach improves cross‐study comparisons and prediction of alternative regimens and therapeutic approaches by accounting for differences in baseline plaque load, dosing and titration regimens, and mechanism of action.