Reorganization of brain networks after spinal cord injury: a qualitative synthesis of the literature

Introduction: Changes in brain organization and structure have been associated with spinal cord injury (SCI) and have been extensively studied (Freund et al. 2013; Nardone et al. 2013). On the other hand, our understanding of brain connectivity following SCI is significantly lower, with studies appearing during the last decade. In the present study we attempt a qualitative synthesis of relevant peer-reviewed literature. Methodology: We performed a search in Pubmed, Scopus and ScienceDirect using the terms “brain/cortical connectivity/network” and “spinal cord injury”. We further tried to identify relevant articles through retrieved original and review papers. In our synthesis, we included only original studies estimating brain connectivity on SCI patients. We excluded reviews, brain activation and intrinsic spinal cord connectivity studies, theoretic mentions of brain networks and studies on animals (with one exception in the absence of a longitudinal study on humans). Results: Changes in neural function and brain connectivity have been shown to appear even during the early stages of the chronic condition and to correlate with the degree of neurological impairment (Hou et al. 2014a; 2014b). A study on a mixed group of complete/incomplete SCI patients within the first month post-injury with fMRI depicted altered spontaneous resting-state brain activation in almost all cortical and sub-cortical sensorimotor areas (Zhu et al. 2016). An fMRI study on SCI patients at mean two months post-injury showed, not only structural changes such as important gray matter atrophy at the sensorimotor cortical areas and pathways (Hou et al. 2014a) but network alterations as well (Hou et al. 2014b). Decreased inter-hemispheric resting-state functional connectivity (FC) was calculated between sensorimotor cortices bilaterally and increased intra-hemispheric resting-state FC within the sensorimotor cortex, premotor area, supplementary motor area (SMA), as well as within other nodes of the motor pathways such as the thalamus and cerebellum. Another study also demonstrated increased resting-state FC between primary motor areas and SMA or basal ganglia (Min et al. 2015b). This intra-hemispheric increase also correlated with higher degree of motor impairment (Hou et al. 2014b). On the other hand, somatosensory components of the sensorimotor network showed decreased connectivity in SCI patients compared to healthy controls (Min et al. 2015b). Predictors of good versus poor neurological recovery at 6 months post-injury included increased resting-state FC between the primary motor cortex and SMA and premotor cortex for those patients that showed good motor recovery (Hou et al. 2016). Graph properties of resting-state networks, as measured by fMRI in patients with incomplete cervical SCI, did not present significant changes except for greater path lengths compared to healthy individuals (Min et al. 2015a). FC changes appear to be dynamic post-injury procedures. While this has not been directly demonstrated in a human longitudinal study, evidence comes from a study on rhesus monkeys (Rao et al. 2016). Resting-state fMRI recordings at pre-injury and 4, 8 and 12 months post-injury showed initially increased FC between multiple major sensorimotor nodes such as primary sensorimotor cortices, SMA and putamen (Rao et al. 2016), a finding that possibly corresponds to the increased intra-hemispheric FC reported by Hou et al. (2014b) in humans. Gradually FC tended to approach baseline levels in most areas at 12 weeks post-injury. During chronic phases of complete SCI, the continued disruption of sensorimotor pathways causes reorganization of resting-state functional networks, namely an overall decrease in FC (Oni-Orisan et al. 2016). Primary motor and somatosensory areas show decreased FC between them and adjacent cortical sensorimotor nodes both intra- and inter-hemispherically. Deeper connectivity is also altered, as an increase of FC between left primary somatosensory area and bilateral thalami was also shown (Oni-Orisan et al. 2016). Furthermore, the importance of residual reciprocal sensorimotor communication (in an incomplete injury) has been demonstrated in a case of measuring an increase of resting-state FC in fMRI after rehabilitation sessions even a decade post-injury (Chisholm et al. 2015). The brain network properties of chronic complete SCI patients have also been investigated with high-resolution EEG recordings and using graph analysis (Astolfi et al. 2006a; 2006b; De Vico Fallani et al. 2006). Statistical comparison of SCI patients and healthy subjects showed a higher degree of local efficiency of brain networks of SCI patients during attempted (paralyzed) foot movements, suggested to be a compensative mechanism (De Vico Fallani et al. 2007; Sinatra et al. 2009). Cingulate motor area (CMA) was identified as an important information hub and time-varying estimation of connectivity further showed larger cortical networks for the SCI patients and greater involvement of the parietal cortex around motor imagery onset (Astolfi et al. 2007; 2008; 2009; Sinatra et al. 2009). On the other hand, SCI patient brain networks seem to show less communication redundancy in the higher EEG spectra (and supposedly less tolerance for dysfunction) compared to healthy individuals, presenting suppressed longer alternative inter-cortical region pathways, as a negative result of the spinal trauma (De Vico Fallani et al. 2009). In lower spectra (such as theta band), no difference in redundancy was shown and higher degree of communication between closest cortical areas in SCI patient networks (De Vico Fallani et al. 2011). The sensorimotor networks of SCI patients and healthy individuals share similar patterns of connectivity but some new functional interactions were identified as unique to SCI patients (Mattia et al. 2009), namely inflow of information from the ipsilateral CMA and SMA to the superior parietal cortex (SPC) and information exchange between bilateral primary motor foot area and SMA. These occurrences could be attributed to both adaptive and maladaptive organization effects after the injury (Mattia et al. 2009; Sinatra et al. 2009; Astolfi et al. 2010; De Vico Fallani et al. 2010). Conclusion: Large-scale, multi-modal, longitudinal studies on SCI patients are needed to understand how brain network reorganization is established and progresses through the course of the condition. Simultaneous analysis of brain and spinal cord activations and interactions could also shed further light (Vahdat et al. 2015). The expected insight holds great clinical relevance in neurofeedback based neurorehabilitation and the design of connectivity-based Brain-Computer Interfaces for SCI patients (Athanasiou et al. 2016).