Responsiveness of magnetic resonance imaging and neuropsychological assessment in memory clinic patients

Background: Scales of global cognition and behavior, often used as endpoints for intervention trials in Alzheimer's disease (AD) and mild cognitive impairment (MCI), are insufficiently responsive (i.e., relatively insensitive to change). Large patient samples are needed to detect beneficial drug effects. Therefore, magnetic resonance imaging (MRI) measures of cerebral atrophy have been proposed as surrogate endpoints. Objective: To examine how neuropsychological assessment compares to MRI in this respect. Methods: We measured hippocampal atrophy, cortical thickness, and performance on neuropsychological tests in memory clinic patients at baseline and after two years. Neurologists rated the patients as cognitively normal (n = 28; Clinical Dementia Rating, CDR = 0) or as impaired (n = 34; CDR > 0). We administered five tests of memory, executive functioning, and verbal fluency. A composite neuropsychological score was calculated by taking the mean of the demographically corrected standard scores. MRI was done on a 3 Tesla scanner. Volumetric measurements of the hippocampus and surrounding cortex were made automatically using FreeSurfer software. Results: The composite neuropsychological score deteriorated 0.6 SD in the impaired group, and was virtually unchanged in the normal group. Annual hippocampal atrophy rates were 3.4% and 0.6% in the impaired and normal cognition groups, respectively. Estimates of required sample sizes to detect a 50% reduction in rate of change were larger using rate of hippocampal atrophy (n = 131) or cortical thickness (n = 488) as outcome compared to change scores on neuropsychological assessment (n = 62). Conclusion: Neuropsychological assessment is more responsive than MRI measures of brain atrophy for detecting disease progression in memory clinic patients with MCI or AD.