Neuromodulation with transcutaneous spinal stimulation reveals different groups of motor profiles during robot-guided stepping in humans with incomplete spinal cord injury

Neuromodulation via spinal stimulation has been investigated for improving motor function and reducing spasticity after spinal cord injury (SCI) in humans. Despite the reported heterogeneity of outcomes, few investigations have attempted to discern commonalities among individual responses to neuromodulation, especially the impact of stimulation frequencies. Here, we examined how exposure to continuous lumbosacral transcutaneous spinal stimulation (TSS) across a range of frequencies affects robotic torques and EMG patterns during stepping in a robotic gait orthosis on a motorized treadmill. We studied nine chronic motor-incomplete SCI individuals (8/1 AIS-C/D, 8 men) during robot-guided stepping with body-weight support without and with TSS applied at random frequencies between 1 and up to 100 Hz at a constant, individually selected stimulation intensity below the common motor threshold for posterior root reflexes. The hip and knee robotic torques needed to maintain the predefined stepping trajectory and EMG in eight bilateral leg muscles were recorded. We calculated the standardized mean difference between the stimulation conditions grouped into frequency bins and the no stimulation condition to determine changes in the normalized torques and the average EMG envelopes. We found heterogeneous changes in robotic torques across individuals. Agglomerative clustering of robotic torques identified four groups wherein the patterns of changes differed in magnitude and direction depending mainly on the stimulation frequency and stance/swing phase. On one end of the spectrum, the changes in robotic torques were greater with increasing stimulation frequencies (four participants), which coincided with a decrease in EMG, mainly due to the reduction of clonogenic motor output in the lower leg muscles. On the other end, we found an inverted u-shape change in torque over the mid-frequency range along with an increase in EMG, reflecting the augmentation of gait-related physiological (two participants) or pathophysiological (one participant) output. We conclude that TSS during robot-guided stepping reveals different frequency-dependent motor profiles among individuals with chronic motor incomplete SCI. This suggests the need for a better understanding and characterization of motor control profiles in SCI when applying TSS as a therapeutic intervention for improving gait.