Design and sample size considerations for Alzheimer's disease prevention trials using multistate models

Area of Science:

• Neuroscience
• Biostatistics
• Clinical Trial Design

Background:

• Alzheimer's disease (AD) pathophysiology begins decades before cognitive symptoms.
• Early intervention is crucial for effective AD prevention.
• Designing preclinical AD trials presents challenges in sample size and follow-up duration.

Purpose of the Study:

• To apply a unifying multistate model for Alzheimer's disease preclinical progression.
• To address critical issues in designing primary and secondary prevention clinical trials for AD.
• To specify intervention effects during the long disease process.

Main Methods:

• Utilized a nonhomogeneous Markov multistate model for AD progression through preclinical states, mild cognitive impairment, and dementia.
• Employed transition probabilities from published cohort studies.
• Developed sample size methods considering preclinical state, endpoint, age-dependent rates, and intervention targets.

Main Results:

• AD prevention trials with clinical endpoints (MCI or dementia) require thousands of participants and 5+ years of follow-up.
• Large sample sizes are driven by the decades-long preclinical AD period, attrition due to mortality/loss to follow-up, and selection bias.
• A web application for sample size calculations is available.

Conclusions:

• Multistate models enhance accuracy in determining sample sizes for AD prevention trials by accounting for intervention timing.
• Innovative strategies are needed to design feasible AD prevention trials with manageable sample size and follow-up requirements.