Estimating treatment effects in clinical studies of Alzheimer's disease: What are estimands and why do we need them?

In clinical studies, it is scientifically important, and a regulatory expectation, that objectives be translated into key clinical questions by specifically defining treatment effects to be estimated. Estimands are part of a structured framework, as presented in International Council for Harmonisation (ICH) E9(R1), by which study objectives are linked to a suitable study design and tools for estimation. Estimands are constructed using five attributes: treatment, population, variable (or endpoint), population-level summary for the variable, and intercurrent events (ICEs). ICEs occur after treatment initiation and can affect the existence or interpretation of the measurements. In Alzheimer's disease (AD), potential ICEs include additional AD medication use, discontinuation of treatment, and death. We describe estimands in recent clinical studies of anti-amyloid therapies, including gantenerumab, lecanemab, and donanemab, in AD and use the evoke and evoke+ studies of semaglutide in early-stage symptomatic AD as examples of estimand application in AD trials.