LAUSR

Longitudinal neuroimaging studies aid our understanding of recovery mechanisms in moderate-to-severe traumatic brain injury (TBI); however, there is a dearth of longitudinal functional connectivity research. Our aim was to characterize longitudinal functional connectivity patterns in two clinically-important brain networks, the frontoparietal and default mode networks, in moderate-to-severe TBI. This inception cohort study of prospectively-collected longitudinal data used resting-state fMRI to characterize functional connectivity patterns in the frontoparietal and default mode networks. Forty adults with moderate-to-severe TBI (mean ± standard deviation; age=39.53±16.49 years, education=13.92±3.20 years, lowest Glasgow Coma Scale score=6.63±3.24, sex=70% male) were scanned at approximately 0.5, 1-1.5, and 3+ years post-injury. Seventeen healthy, uninjured participants (mean ± standard deviation; age=38.91±15.57 years, education=15.11±2.71 years, sex=29% male) were scanned at baseline and approximately 11 months afterwards. Group independent component analyses and linear mixed-effects modeling with linear splines that contained a knot at 1.5 years post-injury were employed to investigate longitudinal network changes, and associations with covariates, including age, sex, and injury severity. In TBI patients, functional connectivity in the right frontoparietal network increased from approximately 0.5 to 1.5 years post-injury (unstandardized estimate=0.19, standard error (SE)=0.07, p=0.009), contained a slope change in the opposite direction, from positive to negative at 1.5 years post-injury (estimate=-0.21, SE=0.11, p=0.009), and marginally declined afterwards (estimate=-0.10, SE=0.06, p=0.079). Functional connectivity in the default mode network increased from approximately 0.5 to 1.5 years (estimate=0.15, SE=0.05, p=0.006), contained a slope change in the opposite direction, from positive to negative at 1.5 years post-injury (estimate=-0.19, SE=0.08, p=0.021), and was estimated to decline from 1.5 to 3+ years (estimate=-0.04, SE=0.04, p=0.303). Similarly, the left frontoparietal network increased in functional connectivity from approximately 0.5 to 1.5 years post-injury (estimate=0.15, SE=0.05, p=0.002), contained a slope change in the opposite direction, from positive to negative at 1.5 years post-injury (estimate=-0.18, SE=0.07, p=0.008), and was estimated to decline thereafter (estimate=-0.04, SE=0.03, p=0.254). At approximately 0.5 years post-injury, patients showed hypoconnectivity compared to healthy, uninjured participants at baseline. Covariates were not significantly associated in any of the models. Findings of early improvement but a tapering and possible decline in connectivity thereafter suggest that compensatory effects are time-limited. These later reductions in connectivity mirror growing evidence of behavioural and structural decline in chronic moderate-to-severe TBI. Targeting such declines represents a novel avenue of research and offers potential for improving clinical outcomes.