Associations of brain connectivity with disease progression and cognitive dysfunction in autosomal‐dominant Alzheimer disease depend on imaging modality: Neuroimaging / multi‐modal comparisons. (7th December 2020)

• Record Type: Journal Article
• Authors: Dicks, Ellen; Vermunt, Lisa; Tijms, Betty M.; van Der Flier, Wiesje; Barkhof, Frederik; Tanenbaum, Aaron; Strain, Jeremy F.; Morris, John C.; Bateman, Randall J.; Benzinger, Tammie L.S.; Gordon, Brian A.
• Is Part Of: Alzheimer's & dementia. Volume 16(2020)Supplement 4
• Page Start: n/a
• Page End: n/a
• Publication Date: 2020-12-07
• Subjects: Alzheimer's disease -- Periodicals; Alzheimer Disease -- Periodicals; Dementia -- Periodicals; Démence; Maladie d'Alzheimer; Périodique électronique (Descripteur de forme); Ressource Internet (Descripteur de forme); 616.83
• DOI: 10.1002/alz.045942
• Languages: English
• ISSNs: 1552-5260

Abstract

Abstract: Background: In the early stages of Alzheimer disease (AD) brain connectivity becomes disrupted. Brain connectivity can be measured based on resting‐state fMRI, DTI and structural MRI (i.e., grey matter (GM) covariance). It currently remains unclear, whether the timepoint of brain connectivity disruptions and associations with cognitive dysfunction depend on the imaging modality used. Here, we investigated for each modality in autosomal‐dominant AD, at which time to estimated symptom onset (EYO) mutation carriers (MC) and non‐carriers (NC) start to diverge and their association with cognition. Method: From the Dominantly Inherited Alzheimer Network (DIAN) we included all individuals who had one resting‐state fMRI (n=205), DTI (n=68) and/or structural MRI scan (n=434) available. Preprocessing and analysis are described in Fig.1. We calculated the network measures clustering, path length and their normalized versions (i.e., gamma and lambda). For each modality we used Bayesian mixed effects models with EYO as predictor to determine the first timepoint in the disease process, at which MC and NC start to diverge. We additionally assessed in MC the relationship between network measures and cognitive function (MMSE and logical memory, delayed recall), correcting for age. Result: Across the EYO, GM‐based network measures showed divergences between MC and NC starting ‐8.1 years before symptom onset, followed by fMRI‐based network measures 0.7 years after symptom onset (Fig.2). For DTI, we found no significant differences, but MC showed slightly decreased clustering values around ‐10 years before symptom onset compared to NC. Both GM‐ and DTI‐based network measures showed associations with the MMSE, with the strongest effects for lower MMSE with lower gamma (standardized β±SE(DTI)=0.44±0.04; standardized β±SE(GM)=0.59±0.06; all p<0.001). GM‐ and fMRI‐based network measures were associated with logical memory performance, with the strongest effects for worse performance with lower lambda in fMRI‐based networks (standardized β±SE=0.16±0.07; p<0.05) and lower gamma for GM‐based networks (standardized β±SE=0.49±0.06; p<0.001). Conclusion: GM‐based network measures showed the earliest differences between mutation carriers and non‐carriers, followed by fMRI. GM‐based network measures were associated with MMSE and logical memory, DTI with MMSE and fMRI with logical memory. This suggests that different modalities may capture specific processes in AD. Grant: P50AG005681.