Language production deficits occur early in the course of Alzheimer’s disease (AD); however, only few studies have focused on language functional networks in prodromal AD. The current study aims to uncover the extent of language alteration at a prodromal stage, on a behavioral, structural and functional level, using univariate and multivariate analyses. Twenty-four AD participants and 24 matched healthy controls underwent a comprehensive language evaluation, a structural T1-3D MRI and resting-state fMRI. We performed seed-based analyses, using the left inferior frontal gyrus and left posterior temporal gyrus as seeds. Then, we analyzed connectivity between executive control networks and language network in each group. Finally, we used multivariate pattern analyses to test whether the two groups could be distinguished based on the pattern of atrophy within the language network; atrophy within the executive control networks, as well as the pattern of functional connectivity within the language network; and functional connectivity within executive control networks. AD participants had language impairment during standardized language tasks and connected-speech production. Univariate analyses were not able to discriminate participants at this stage, while multivariate pattern analyses could significantly predict the group membership of prodromal patients and healthy controls, both when classifying atrophy patterns or connectivity patterns of the language network. Language functional networks could discriminate AD participants better than executive control networks. Most notably, they revealed an increased connectivity at a prodromal stage. Multivariate analyses represent a useful tool for investigating the functional and structural (re-)organization of the neural bases of language.
bioRxiv Subject Collection: Neuroscience