Training With A Non-Invasive Brain-Machine Interface, Combined With Virtual Reality And Assisted Robotic Locomotion, Induces Significant Motor Recovery And Partial Reversal Of Widespread Cortical Atrophy In Asia A Paraplegics: A Randomized, Controlled, Single-Center Trial

Twelve years ago, the Walk Again Project (WAP) introduced a multi-stage neurorehabilitation protocol, the Walk Again Neurorehabilitation Protocol (WANR), which combines virtual reality training with robotic gait systems, both controlled by an EEG-based, non-invasive brain-machine interface (BMI). Training with the WANR led to significant partial neurological and functional recovery in spinal cord injury (SCI) patients. Yet, the neural mechanisms underlying such a recovery remain unknown. Here, we report on the results obtained with an adapted version of the WANR protocol in a larger, randomized, controlled, assessor-blinded, two group clinical trial aimed at evaluating the concurrent morphological and functional brain changes that take place during motor recovery in SCI. A total of 19 ASIA A, SCI patients (age 21 to 59; lesion time 13 months to 25 years) were initially allocated in two groups: a control classical neurorehabilitation (NR, n=9), and an experimental group (WANR, n=10). Later on, four patients that had finished the NR protocol joined the WANR group. Altogether, 14 patients were trained with the WANR (12 completed 9 months of training and two completed 5 months). When compared to the NR group, the WANR group exhibited a significant motor recovery, measured by the Lower Extremity Motor Score (LEMS) and the Walking Scale for Spinal Cord Injury (WISCI), which was detected at 5 months (mean{+/-}SD LEMS= 1.83{+/-}1.19 and WISCI= 5.92{+/-}2.78), but peaked at 9 months (mean{+/-}SD LEMS= 2.75{+/-}1.54 and WISCI= 9.75{+/-}2.14). Overall, 50% of these WANR-trained patients made a transition from ASIA A to ASIA C during this period. Longitudinal brain imaging analysis revealed that chronic SCI induced a widespread reduction in cortical thickness, including a substantial bilateral atrophy (between 15-22%, or 0.4-0.8mm) in many cortical regions in the insula, temporal, frontal, parietal, and even the occipital lobes. These findings are consistent with a variety of cognitive impairments observed in 40-64% of SCI patients. Moreover, they reinforce the hypothesis that SCIs trigger an acceleration of the normal processes of brain aging, which could explain the higher risk that SCI patients have of developing neurodegenerative disorders. Training with the WANR induced a very significant process of cortical plasticity, represented by a significant reversal of cortical atrophy (up to 1mm in the temporal lobe and insula) and an increase in both functional connectivity strength in cerebellar regions and in eigenvector centrality in executive and integrative cortical regions, like the angular gyrus, precuneus, and middle frontal gyrus. Meanwhile, sensorimotor hubs experienced a centrality reduction. This suggests a functional shift from damaged motor loops towards potential compensatory cortical networks. Accordingly, we propose a new mechanism for BMI-induced clinical improvement and raise the hypothesis that the WANR could induce neurologic recovery in other neurological conditions that produce cortical atrophy. Our findings also suggest that BMI training, coupled with virtual reality, and vigorous exercise or robotic-assisted walking, may also help mitigate or delay the normal process of brain aging in elderly patients. Our findings also categorically indicate that there is neither clinical, nor ethical justification for employing highly invasive cortical implants to treat paraplegics, given that non-invasive protocols, like the WANR, may suffice to induce significant neurological and functional gains in those patients with a few months of non-invasive neurorehabilitation.