Sample size estimates for biomarker-based outcome measures in clinical trials in autosomal dominant Alzheimer's
David M Cash1, Katy E Morgan2, Antoinette O'Connor3
1Dementia Research Centre, UCL Queen Square Institute of Neurology, First floor, 8-11 Queen Square, London, WC1N 3AR, UK; UK Dementia Research Institute, 6th Floor, Maple House, Tottenham Court Road, London W1T 7NF, UK.
Introduction:
Alzheimer disease (AD)-modifying therapies are approved for treatment of early-symptomatic AD. Autosomal dominant AD (ADAD) provides a unique opportunity to test therapies in presymptomatic individuals.
Methods:
Using data from the Dominantly Inherited Alzheimer Network (DIAN), sample sizes for clinical trials were estimated for various cognitive, imaging, and CSF outcomes. Sample sizes were computed for detecting a reduction of either absolute levels of AD-related pathology (amyloid, tau) or change over time in neurodegeneration (atrophy, hypometabolism, cognitive change).
Results:
Biomarkers measuring amyloid and tau pathology had required sample sizes below 200 participants per arm (examples CSF Aβ42/40: 47[95 %CI 25,104], cortical PIB 49[28,99], CSF p-tau181 74[48,125]) for a four-year trial in presymptomatic individuals (CDR=0) to have 80 % power (5 % statistical significance) to detect a 25 % reduction in absolute levels of pathology, allowing 40 % dropout. For cognitive, MRI, and FDG, it was more appropriate to detect a 50 % reduction in rate of change. Sample sizes ranged from 250 to 900 (examples hippocampal volume: 338[131,2096], cognitive composite: 326[157,1074]). MRI, FDG and cognitive outcomes had lower sample sizes when including indivduals with mild impairment (CDR=0.5 and 1) as well as presymptomatic individuals (CDR=0).
Discussion:
Despite the rarity of ADAD, presymptomatic clinical trials with feasible sample sizes given the number of cases appear possible.
