AUTHOR=Bernier Rachel Anne , Roy Arnab , Venkatesan Umesh Meyyappan , Grossner Emily C. , Brenner Einat K. , Hillary Frank Gerard 

TITLE=Dedifferentiation Does Not Account for Hyperconnectivity after Traumatic Brain Injury

JOURNAL=Frontiers in Neurology

VOLUME=Volume 8 - 2017

YEAR=2017

URL=https://www.frontiersin.org/journals/neurology/articles/10.3389/fneur.2017.00297

DOI=10.3389/fneur.2017.00297

ISSN=1664-2295

ABSTRACT=Objective: Changes in functional network connectivity following TBI have received increasing attention in recent neuroimaging literature. This study sought to understand how disrupted systems adapt to injury during resting and goal-directed brain states. Hyperconnectivity has been a common finding and dedifferentiation (or loss of segregation of networks) is one possible explanation for this finding. We hypothesize that individuals with TBI will show dedifferentiation of networks (as noted in other clinical populations; see Dennis & Cabeza, 2011) and such dedifferentiation will be associated with worse cognitive functioning. 

Methods: Graph theory was implemented to examine functional connectivity in 19 individuals with moderate/severe TBI and 14 healthy controls (HC). Using a functional brain atlas derived from 83 functional imaging studies (Power et al., 2011), graph theory was used to examine network dynamics and determine whether dedifferentiation accounts for changes in connectivity. Regions-of-interest were assigned to one of three groups: task-positive, default mode, or other. Relationships between these metrics were then compared with performance on neuropsychological tests. 

Results: Hyperconnectivity is a function of greater within-network connectivity. Network strength within networks that showed differences between TBI and HCs were correlated with performance on five neuropsychological tests typically sensitive to deficits commonly reported in TBI. Hyperconnectivity within the default mode network (DMN) during task was associated with better performance on Digit Span Backward, a measure of working memory (R2(18) = 0.28, p = 0.02). In other words, increased differentiation of networks during task was associated with better working memory. Hyperconnectivity within the task-positive network during rest was not associated with behavior. Negative correlation weights were not associated with behavior. 

Conclusion: The primary hypothesis that hyperconnectivity occurs through increased segregation of networks, rather than dedifferentiation, was not supported. Results suggest that the relationship between increased connectivity and cognitive functioning may be both state (rest or task) and network dependent. Enhanced connectivity within the DMN predicted better working memory performance. High cost network hubs were identical for both rest and task, and cost was negatively associated with performance on measures of psychomotor speed and set-shifting, but not associated with basic attention and working memory.