Individual Variation in Functional Brain Network Topography is Linked to Schizophrenia Symptomatology

Abstract

Resting-state fMRI (rsfMRI) demonstrates that the brain is organized into distributed networks. Numerous studies have examined links between psychiatric symptomatology and network functional connectivity. Traditional rsfMRI analyses assume that the spatial organization of networks is invariant between individuals. This dogma has recently been overturned by the demonstration that networks show significant variation between individuals. We tested the hypothesis that previously observed relationships between schizophrenia-negative symptom severity and network connectivity are actually due to individual differences in network spatial organization. Forty-four participants diagnosed with schizophrenia underwent rsfMRI scans and clinical assessments. A multivariate pattern analysis determined how whole-brain functional connectivity correlates with negative symptom severity at the individual voxel level. Brain connectivity to a region of the right dorsolateral prefrontal cortex correlates with negative symptom severity. This finding results from individual differences in the topographic distribution of 2 networks: the default mode network (DMN) and the task-positive network (TPN). Both networks demonstrate strong (r = ~0.49) and significant (P < .001) relationships between topography and symptom severity. For individuals with low symptom severity, this critical region is part of the DMN. In highly symptomatic individuals, this region is part of the TPN. Previously overlooked individual variation in brain organization is tightly linked to differences in schizophrenia symptom severity. Recognizing critical links between network topography and pathological symptomology may identify key circuits that underlie cognitive and behavioral phenotypes. Individual variation in network topography likely guides different responses to clinical interventions that rely on anatomical targeting (eg, transcranial magnetic stimulation [TMS]).

Keywords: psychosis; psychosocial treatments; schizophrenia.

© The Author(s) 2020. Published by Oxford University Press on behalf of the Maryland Psychiatric Research Center. All rights reserved. For permissions, please email: journals.permissions@oup.com.

Figures

Fig. 1.

Multivariate distance matrix regression identifies brain voxels whose functional connectivity varies with negative symptom severity. MDMR procedure: A. rsfMRI and SANS are collected from each participant. B. For each participant, a functional connectivity map is generated to an individual voxel. C. Voxel-wise temporal correlations between participants are used to generate a Pearson’s correlation r and a distance metric d. This is repeated for all participants to generate a matrix of between subject distances. D. The distance matrix is centered and an ANOVA-like test is used to generate an F-statistic to assess the relationship between a predictor variable (SANS score) and dissimilarities in functional connectivity at that voxel. E. This process is repeated for every voxel. This results in a whole-brain map of how significantly functional connectivity is related to emotional intelligence. Permutation testing then identifies whole-brain significant clusters in connectivity–SANS score relationships. F. In our sample of 44 participants with schizophrenia or schizoaffective disorder, we identified the bilateral subregions of the DLPFC (Brodmann Area 9) as regions where resting-state connectivity correlated significantly with SANS score. In this image, connectivity is thresholded at a voxel-wise level of P < 0.005 and extent threshold of P < 0.05.

Fig. 2.

In a follow-up analysis, maps of connectivity to the right dorsolateral prefrontal cortex (DLPFC) subregion were generated in all subjects and then this seed-based connectivity map was correlated with Scale for the Assessment of Negative Symptoms (SANS) scores to identify locations where increasing connectivity to DLPFC corresponds to more severe symptoms and decreased connectivity corresponds to less severe symptoms. Here, we observe: (A) The higher an individual’s SANS score, the more this DLPFC subregion was functionally connected to the task-positive network. (B) The higher an individual’s SANS score, the less this subregion was functionally connected to the default mode network. Color bar = T-statistic.

Fig. 3.

Negative symptom severity is linked to network topography. A. Plot of SANS score (y-axis) and functional connectivity between the rDLPFC ROI and the DMN network. B. Plot of SANS score (y-axis) and functional connectivity between the rDLPFC ROI and TPN network. Notably, both graphs demonstrate correlations that cross the reference line of null functional connectivity, meaning that the network association of this region appears to shift from being a member of the DMN to a member of the TPN as one moves along the distribution from less symptomatic individuals to more symptomatic individuals. Network topography correlates were highly significant (P < 0.001) for both networks.

Fig. 4.

Default mode network (DMN) topography in relation to negative symptom severity. The spatial organization of the DMN in the dorsolateral prefrontal cortex (DLPFC) is linked to negative symptom severity. Specifically, the rostral-caudal extent of the DMN in a DLPFC subregion is aligned with a gradient of symptom severity. For participants with the least-severe negative symptoms, the spatial distribution of the DMN in this region extended anteriorly. For participants with the greatest negative symptom burden, the distribution of DMN was more posterior with the task-positive network instead of extending into this region (not shown).