Multivariate analysis of cognitive profiles in Alzheimer's disease

peer reviewedThe neuropsychological profiles of patients with early Alzheimer’s disease (AD) appear to be heterogeneous. In this study, we examined whether this heterogeneity corresponds to the existence of cognitively distinct subtypes of AD or rather to impairments along a continuum of performances in different cognitive domains. A large group of 187 AD patients recruited in the European project NEST-DD performed a neuropsychological battery. A factor analysis of cognitive performance identified three factors, which respectively reflected attentional/instrumental function, declarative memory and executive function. Three clustering methods were applied on the factor scores in order to explore the existence of separate groups. The clustering methods indicated that cognitive profiles among the patients were sufficiently variable to identify clusters, but there was continuity between clusters rather than clear-cut subtypes. Moreover, clusters corresponded to various combinations of relatively impaired and preserved functions, suggesting multidimensional distribution within a large population of patients. Finally, clusters of cognitive profiles were characterized by different levels of metabolism in brain regions commonly (but variably) involved or relatively preserved in AD