Disease Modeling and Model-Based Meta-Analyses to Define a New Direction for a Phase III Program of Gantenerumab in Alzheimer's Disease.

Selecting the right dose is a significant challenge in designing clinical development programs, especially for slowly progressing diseases lacking predictive biomarkers of efficacy that may require long-term treatment to assess clinical benefit. Gantenerumab, a fully human monoclonal antibody (mAb) that binds to aggregated amyloid-beta, was tested in two 24-month phase III studies (NCT01224106, NCT02051608) in participants with prodromal and mild Alzheimer's disease (AD), respectively. Dosing in the first phase III study was suspended after a preplanned interim futility analysis in 2014. Subsequently, a dose-response relationship was observed in a subgroup of fast AD progressors that, together with contemporary aducanumab (another anti-amyloid-beta mAb) data, indicated higher doses may be needed for clinical efficacy. The gantenerumab phase III studies were therefore transformed into dose-finding, open-label extension (OLE) trials. Two exposure-response models were developed to support dose selection via simulations for the OLEs: a pharmacokinetics (PK)/PET (positron emission tomography) model describing amyloid removal using PET data from low-dose gantenerumab and high-dose aducanumab, and a PK/ARIA-E (amyloid-related imaging abnormalities-edema) model describing the occurrence of ARIA-E events leveraging an existing bapineuzumab model. Multiple regimens were designed to gradually up-titrate participants to the target dose of 1,200 mg gantenerumab every 4 weeks to mitigate the increased risk of ARIA-E events that may be associated with higher doses of anti-amyloid-beta antibodies. Favorable OLE data that matched well with model predictions supported the decision to continue the gantenerumab clinical development program and further apply model-based analytical techniques to optimize the design of new phase III studies.