PROVIDENCE, R.I. [Brown University] — The effects of spinal cord injuries are complex and multifaceted. People lose not only the ability to control the movement of their limbs, but also the ability to receive sensory feedback from them. Both are critical to generate the coordinated movement involved in walking.
Now, a team of researchers from Brown University, Rhode Island Hospital, and VA Providence Healthcare has shown progress in restoring two-way communication across a damaged site of the spinal cord. In a study in Nature Biomedical Engineering, the researchers report results from a clinical trial involving three people who had lost the use of their legs following complete spinal cord injuries.
The participants received electrical stimulation of the spinal cord from electrode arrays implanted both above and below their injury sites. The study found that stimulation below the injury could partially restore muscle control in lower extremities, while stimulation above the injury enabled participants to understand where their legs were located in space as they walked, with the assistance of physical therapists, on a treadmill.
"This is the first time that simultaneous motor stimulation and sensory feedback have been demonstrated in people with complete spinal cord injuries," said David Borton, an associate professor of engineering at Brown and a biomedical engineer at the VA Center for Neurorestoration and Neurotechnology. "This is an important step toward the goal of fully bridging the gap created by a spinal lesion. By providing both motor activation and simultaneous sensory feedback, we are making progress toward restoring coordinated movements and functional independence."
The federally supported research is a collaboration between scientists and clinicians from Brown's Institute for Biology, Engineering and Medicine and Carney Institute for Brain Science , Rhode Island Hospital's neurosurgery department, the VA Center for Neurorestoration and Neurotechnology, and other institutions.
"We are incredibly grateful to the participants who volunteered for this study without expectation of long-term benefit to themselves," said Borton, the study's corresponding author. "Their generosity paves the way for future research aimed at fully restoring functions lost to devastating spinal cord injuries."
Dr. Jared Fridley, chief of spinal neurosurgery at the University of Texas at Austin, who contributed to the research while he was a neurosurgeon at Rhode Island Hospital, said the research demonstrates what is possible when engineering innovation is integrated with clinical neuroscience.
"By simultaneously restoring motor activation and meaningful sensory feedback, we're moving beyond isolated function toward coordinated, purposeful movement," Fridley said. "That's a critical step if neurotechnology is going to translate into real-world independence for people living with severe spinal cord injury."
Stimulation above and below the injury site
For the study, the team recruited three volunteers who were paralyzed from the waist down after spinal cord injuries. Surgeons placed small electrode arrays above and below the injury site along the participants' spinal cords at the beginning of the two-week, in-hospital study. Both arrays delivered patterned electrical stimulation to the spinal cord that can mimic the natural electrical pulses that travel through a healthy spinal cord.
Prior research by Borton and others had shown that this kind of patterned spinal stimulation can be used to drive control of muscles in non-human primates , which paved the way for translational studies in humans. But no one had previously attempted combining motor stimulation below the injury with sensory stimulation above it in the same individual.
"Most spinal cord injuries impair the sensory system — meaning people don't receive usable sensory feedback from below the lesion," Borton said. "That feedback is essential for coordinated movements like walking. Without it, people depend on their eyes for feedback on where their legs are in space, which is not ideal. The goal of this work was to combine stimulation for muscle activation with stimulation for sensory feedback, and to do those two things at the same time, taking the critical step for real world implementation of neurotechnologies."
To start, the researchers worked with trial participants to fine-tune electrical stimulation of nerves in the spinal cord responsible for muscle movements involved in walking. The participants used a control device that enabled them to personally adjust the stimulation patterns in their spines. The "DJ board," as the research team calls it, has an array of knobs and sliders that allowed the participants to control which parts of the spinal cord received stimulation, along with the speed and intensity of the stimulation. The participants used the device to zero in on the patterns that generated flexion and relaxation of leg muscles.
"Participants told us that using the DJ board was actually a lot of fun," said study lead author Jonathan Calvert, an assistant professor of neurological surgery at the University of California Davis who worked on the project as a postdoctoral researcher at Brown. "We gave them target leg positions and poses and they navigated the board until they found the correct stimulation patterns to achieve that pose. They really enjoyed being able to see their legs move again and having their own control through the interface."
The researchers then used data from the DJ board experiments to train a machine learning model developed by Thomas Serre, a professor of cognitive and linguistic sciences at Brown. The algorithm optimized the stimulation patterns, finding the most precise matches between desired muscle activity and stimulation parameters.
"The space of possible stimulations is huge — far too large to be efficiently searched by trial and error," said co-author Lakshmi Narasimhan Govindarajan, a recent Ph.D. graduate from Serre's lab who is now a postdoctoral researcher at MIT. "Machine learning provides an opportunity to more efficiently search and personalize stimulation patterns so they more precisely matched the muscle activity we were aiming for in each participant."
The researchers then used a similar process to optimize stimulation above the injury site to produce useful sensory feedback. Because the neural wiring of the sensory pathways to the brain have been severed, it's not currently possible to map electrical stimulation directly to the nerves associated with sensations of the legs or feet. Instead, the goal of this study was to test whether sensations generated by spinal cord stimulation above the lesion, networks that still connect sensation from other parts of the body to the brain, could be used to replace sensations of leg or foot movement.
"We used a sensory replacement approach where specific sensations are associated with specific actions or stimuli to enable participants to reinterpret sensory cues," Calvert said. "In this case, participants might feel a sensation in their chest or arm or back, but they can learn to associate those sensations with different joint angles in their legs."
The participants used the DJ board to select where on the body to receive sensory feedback, and the machine learning algorithm was used to correlate the electrical stimulation with different positions of participants' knee joints. To demonstrate that the feedback was useful, physical therapists placed the participants' legs in positions with varying degrees of knee bend. While blindfolded, the participants then used their arms to report how they thought their legs were positioned based on the sensation they received from electrical stimulation.
"We found that the participants could report the angle of their knee with high accuracy based on the intensity of sensations generated by stimulation," Calvert said. "That tells us that these sensations are providing sensory feedback that's useful in terms of knowing where their legs are in space at any point in time. The participants indicated that this type of sensory feedback could be very useful in their daily life such as transferring in and out of their wheelchair."
Combining movement and sensation
Having established that stimulation could produce both muscle control and sensory feedback, the team performed an experiment using both at the same time. Supported by a ceiling-mounted harness and aided by physical therapists, the participants performed walking movements on a treadmill. The experiment showed that the participants could simultaneously engage the appropriate muscles needed for walking while accurately reporting when their feet struck the ground.
The participants reported that the sensations they experienced were useful to them in coordinating their movements.
"[I] could tell when [my foot] hit based on feedback up to here [pointing to chest]," one participant said. "It wasn't like I could feel my foot hit the treadmill or anything like that, but it was close."
The results suggest that spinal stimulation has the potential to provide both motor control and useful sensory feedback to people with spinal cord injuries. That kind of feedback could be useful for future patients working to rehabilitate from spinal cord injuries.
"There's reason to believe that coordinated stimulation across an injury site could produce positive rehabilitation effects," Borton said. "That's not something we were able to fully explore in this study, but that we plan to pursue in future work."
No device-related adverse effects from the implanted electrodes or the electrical stimulation protocols were reported by the participants, which clears the way for additional clinical studies. The team plans to recruit new participants in a longer-term study testing spinal stimulation outside the hospital setting.
"We are excited by the potential of neurotechnology to supplement the long history of pharmaceutical-based approaches to helping people with spinal cord injury," Borton said.
The research was supported by funding from the Defense Advanced Research Projects Agency (D15AP00112, D19AC00015), Department of Veterans Affairs (I01RX002835, A9263A-S) and National Institutes of Health (5T32NS100663-04, S10OD02518).
