Functional network connectivity impairments and core cognitive deficits in schizophrenia

Abstract

Cognitive deficits contribute to functional disability in patients with schizophrenia and may be related to altered functional networks that serve cognition. We evaluated the integrity of major functional networks and assessed their role in supporting two cognitive functions affected in schizophrenia: processing speed (PS) and working memory (WM). Resting-state functional magnetic resonance imaging (rsfMRI) data, N = 261 patients and 327 controls, were aggregated from three independent cohorts and evaluated using Enhancing NeuroImaging Genetics through Meta Analysis rsfMRI analysis pipeline. Meta- and mega-analyses were used to evaluate patient-control differences in functional connectivity (FC) measures. Canonical correlation analysis was used to study the association between cognitive deficits and FC measures. Patients showed consistent patterns of cognitive and resting-state FC (rsFC) deficits across three cohorts. Patient-control differences in rsFC calculated using seed-based and dual-regression approaches were consistent (Cohen's d: 0.31 ± 0.09 and 0.29 ± 0.08, p < 10^-4 ). RsFC measures explained 12-17% of the individual variations in PS and WM in the full sample and in patients and controls separately, with the strongest correlations found in salience, auditory, somatosensory, and default-mode networks. The pattern of association between rsFC (within-network) and PS (r = .45, p = .07) and WM (r = .36, p = .16), and rsFC (between-network) and PS (r = .52, p = 8.4 × 10^-3 ) and WM (r = .47, p = .02), derived from multiple networks was related to effect size of patient-control differences in the functional networks. No association was detected between rsFC and current medication dose or psychosis ratings. Patients demonstrated significant reduction in several FC networks that may partially underlie some of the core neurocognitive deficits in schizophrenia. The strength of connectivity-cognition relationships in different networks was strongly associated with network's vulnerability to schizophrenia.

Keywords: effect size; processing speed; resting-state functional connectivity; working memory.

Figure 1

Mega‐analytic results. Mega‐analytic Forest plots for group differences between individuals with schizophrenia and healthy controls, for resting‐state functional connectivity (rsFC) measures, with (a) the seed‐based analysis approach; and (b) the dual regression analysis approach, for each resting‐state network's functional connections. Enclosed rectangular boxes, with different colors, separate the functional connections for each resting‐state network. Abbreviations: AN, auditory network, a₁/a₂, left/right primary and association auditory cortices; AttN, attention network, f₁/f₂, left/right middle frontal gyrus, and p₁/p₂, left/right superior parietal lobule; DMN, default mode network, r₁, posterior cingulate/precuneus, r₂, bilateral temporal–parietal regions and, r₃, ventromedial frontal cortex; ECN, executive control network, r₁, anterior cingulate cortex and r₂, bilateral medial frontal gyrus; FPN, fronto‐parietal network, f₁/f₂, left/right frontal area (inferior frontal gyrus) and p₁/p₂, left/right parietal area (inferior parietal lobule); SN, salience network, r₁, anterior cingulate cortex and r₂/r₃, left/right insula; SMN, sensorimotor network, m₁/m₃, left/right motor area and m₂, supplementary motor area; and VN, visual network, v₁, medial visual areas, v₂, occipital visual areas, and v₃, lateral visual areas [Color figure can be viewed at http://wileyonlinelibrary.com]

Figure 2

Mega‐analytic Forest plots for group differences between individuals with schizophrenia and healthy controls, from resting‐state functional connectivity (rsFC) measures for resting‐state subnetwork. Each subnetwork consisted of two functional connections between two regions of interest ROIs (connections from the first region to the second region and vice versa). The abbreviations used are the same as in Figure 1; additional prefixes (l/r: left/right) or suffixes (1, 2, 3) are used to represent the subnetwork components

Figure 3

The statistical t‐map showing the patient‐control differences in the functional connectivity (FC) for each pair of regions of interest (ROIs) in between‐network connectivity analysis. The t values for the ROI pairs, statistically significant after Bonferroni multiple comparisons correction, were displayed. These ROI pairs showed reduced FC in patient sample [Color figure can be viewed at http://wileyonlinelibrary.com]

Figure 4

Correlation analysis between canonical correlation coefficients obtained using resting‐state functional connectivity (rsFC) measures as a set of variables with cognitive measures as another set of variables versus Cohen's d effect size (ES) with schizophrenia. The first column shows the correlation between the canonical correlation coefficients obtained using rsFC measures with processing speed versus the ES (Cohen's d) for diagnosis. The second column shows the correlation between the canonical correlation coefficients obtained using rsFC measures with working memory versus ES (Cohen's d) of diagnosis. The first to third rows show information for the full combined sample (a,b), the control sample (c,d), and the patient sample (e,f) respectively. The correlation analysis was performed between 17 resting‐state subnetworks, with each network comprising of two connections. Positive linear relationships were significant (p < .05), except for working memory in the control sample