Association between composite scores of domain-specific cognitive functions and regional patterns of atrophy and functional connectivity in the Alzheimer's disease spectrum

Abstract

Background: Cognitive decline has been found to be associated with gray matter atrophy and disruption of functional neural networks in Alzheimer's disease (AD) in structural and functional imaging (fMRI) studies. Most previous studies have used single test scores of cognitive performance among monocentric cohorts. However, cognitive domain composite scores could be more reliable than single test scores due to the reduction of measurement error. Adopting a multicentric resting state fMRI (rs-fMRI) and cognitive domain approach, we provide a comprehensive description of the structural and functional correlates of the key cognitive domains of AD.

Method: We analyzed MRI, rs-fMRI and cognitive domain score data of 490 participants from an interim baseline release of the multicenter DELCODE study cohort, including 54 people with AD, 86 with Mild Cognitive Impairment (MCI), 175 with Subjective Cognitive Decline (SCD), and 175 Healthy Controls (HC) in the AD-spectrum. Resulting cognitive domain composite scores (executive, visuo-spatial, memory, working memory and language) from the DELCODE neuropsychological battery (DELCODE-NP), were previously derived using confirmatory factor analysis. Statistical analyses examined the differences between diagnostic groups, and the association of composite scores with regional atrophy and network-specific functional connectivity among the patient subgroup of SCD, MCI and AD.

Result: Cognitive performance, atrophy patterns and functional connectivity significantly differed between diagnostic groups in the AD-spectrum. Regional gray matter atrophy was positively associated with visuospatial and other cognitive impairments among the patient subgroup in the AD-spectrum. Except for the visual network, patterns of network-specific resting-state functional connectivity were positively associated with distinct cognitive impairments among the patient subgroup in the AD-spectrum.

Conclusion: Consistent associations between cognitive domain scores and both regional atrophy and network-specific functional connectivity (except for the visual network), support the utility of a multicentric and cognitive domain approach towards explicating the relationship between imaging markers and cognition in the AD-spectrum.

Keywords: Alzheimer’s disease spectrum; Cognitive domain score; Cortical atrophy; Multicenter cohort; Resting-state functional connectivity; Visuo-spatial cognitive deficits.

Fig. 1

Regional gray matter volume significantly differed between diagnostic groups in AD-spectrum, for the (A) MCI and (B) AD diagnostic groups respectively. Voxel-wise multiple comparisons are thresholded with p < 0.01, FDR corrected, cluster size ≥ 50 voxels, 0.40 ≤ d ≤ 1.70. Red voxels show clusters of significantly reduced gray matter volume in patients with MCI and AD compared to HC subgroup. Statistical maps are superimposed on a rendering of the Montreal Neurological Institute template brain. MNI coordinates and corresponding t values are provided in Supplementary Table 1. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 2

Regional gray matter volume is associated with the (A) visuo-spatial, (B) executive, (C) working memory, (D) memory, and (E) language domains respectively. Figure (F) and (G) shows the association of gray matter volume with the clock drawing and clock copy subtest of visuospatial function respectively. Voxel-wise multiple comparisons are thresholded at p < .01, FDR corrected for only figures a-e, cluster size ≥ 50 voxels, 0.38 ≤ d ≤ 0.70. Figures F and G are displayed at p < .001, 0.38 ≤ d ≤ 0.50 uncorrected for multiple comparisons. Red voxels show clusters of significant association between gray matter volume and cognitive domain scores. Statistical maps are superimposed on a rendering of the Montreal Neurological Institute template brain. MNI coordinates and corresponding t values are provided in Supplementary Table 2. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 3

Network-specific resting-state functional connectivity significantly differed between diagnostic groups in AD-spectrum, for the (A) executive, (B) default mode and (C) language networks respectively. Significance is reported at p < .05 FDR corrected for the executive and default mode networks, and at p < .01 uncorrected for the language network accordingly. Cluster size ≥ 20 voxels, 0.30 ≤ d ≤ 0.80. Red voxels represent group resting-state networks, yellow voxels show clusters of significant difference between the patient and healthy control subgroups on network-specific functional connectivity. Statistical comparison was restricted to the corresponding networks only by functional masks determined from the whole sample (see Section 2.3). Statistical maps are superimposed on a rendering of the Montreal Neurological Institute template brain. MNI coordinates and corresponding t values are provided in Supplementary Table 3. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 4

Network-specific resting-state functional connectivity is associated with the (A) executive, (B) working memory, (C) memory, and (D) language functions respectively. Significance is reported at p < .05 FDR corrected for the executive and memory functions, and at p < .01 uncorrected for the working memory and language functions accordingly. Cluster size ≥ 20 voxels, 0.24 ≤ d ≤ 0.50. Red voxels represent group resting-state networks, yellow voxels show clusters of significant association between network-specific functional connectivity and cognitive domain scores. Association was restricted to the corresponding networks only by functional masks determined from the whole sample (see Section 2.3). Statistical maps are superimposed on a rendering of the Montreal Neurological Institute template brain. MNI coordinates and corresponding t values are provided in Supplementary Table 4. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)