Use of Model-Based Meta-Analysis to Inform the Design of Early Clinical Trials of Anti-Amyloid Beta Therapies in Alzheimer's Disease.

Abstract

To inform an efficient development of new investigational anti-amyloid beta (anti-Aβ) monoclonal antibodies (mAbs), a modeling-and-simulation-based strategy was proposed. A general modeling framework that links drug exposures to the time course of amyloid plaque removal and amyloid-related imaging abnormalities characterized by edema and effusion (ARIA-E) was developed based on publicly available data on aducanumab, lecanemab, and donanemab. A non-linear mixed effect model with shared model parameters described the dose response data from aducanumab, lecanemab, and donanemab studies after adjusting for different potency for different antibodies, which allowed the rate of amyloid plaque removal to vary by drug. A time-to-event model was developed to describe ARIA-E incidence. The model assumes that ARIA-E incidence rate is dependent on the rate of amyloid plaque removal with a drug-dependent scaling factor linking amyloid plaque removal rate and treatment-dependent hazard. Simulations of amyloid plaque removal and ARIA-E for a hypothetical anti-Aβ mAb based on certain assumptions and scenarios provided insights into possible outcomes. Overall, the meta-analysis of published data on existing anti-Aβ mAbs could be utilized to model exposure-response relationships and the time course of amyloid plaque removal and ARIA-E incidence of new anti-Aβ mAbs and to inform the design of early clinical trials for them.