Cognitive profiles among dementia-free older adults and their associations with neuroimaging markers.

Prior studies have demonstrated the existence of cognitively-defined subgroups among dementia free older adults, however, it is unclear whether such subgroups are characterized by distinct neuroimaging measures of brain function and structure. To address this gap, the current study used latent profile analysis (LPA) to identify cognitively-defined subgroups in a sample of 167 (mean age = 69 years) dementia-free older adults with cognitive testing, amyloid PET, and multimodal brain MRI scans. The cognitive test scores covered the domains of episodic memory, executive function, language, and visuospatial processing. Linear regression models tested the associations between subgroup membership and neuroimaging measures, adjusting for age, sex, and years of education. Based on the LPA, three cognitive subgroups were identified: (1) high-average cognition (n = 61, 36%), (2) average cognition (n = 88, 53%), and low-average cognition (n = 18, 11%). Compared to the high-average group, the low-average group had lower volumes in cortical regions sensitive to Alzheimer's disease, lower global white matter microstructural integrity measured by diffusion tensor imaging, and higher global white matter hyperintensity burden. There were no group differences in global PET amyloid burden. Additionally, the high-average group tended to have higher resting-state functional connectivity within large-scale cognitive networks than the other two groups. These results suggest that cognitively-defined subgroups among older adults without dementia are associated with several measures of brain structure and function. Evaluating brain structure/function differences among dementia-free older adults may help identify individuals at greatest risk for future cognitive decline.