Greater sustained sensorimotor function recovery and neuropathic pain reduction with Cognitive Multisensory Rehabilitation compared to adaptive fitness in adults with spinal cord injury: a pilot clinical trial

1 Introduction

Spinal cord injury (SCI) imposes a severe dual burden on individuals, characterized not only by profound sensorimotor deficits but also by debilitating neuropathic pain, affecting about 69% of this population.(1–4) Traditional rehabilitation focuses heavily on functional motor recovery. Concurrent management of neuropathic pain is often reliant on pharmacological treatments that are frequently insufficient and can often cause significant adverse reactions. Consequently, there is an urgent need for effective, non-pharmacological interventions that address both functional recovery and pain to improve quality of life for adults with SCI.

A critical, yet often overlooked, driver of both poor sensorimotor recovery and neuropathic pain is the disruption of the brain’s ‘mental body representations’ (MBR). MBR encompasses whole-body awareness, and the brain’s ability to locate body parts in space via visuospatial body maps.(3–7) When SCI interrupts sensory and motor signals, the brain areas responsible for these maps—specifically the parietal operculum and insula for body awareness, the posterior parietal cortex for visuospatial body maps within the sensorimotor network, and the posterior cingulate cortex for self-location and body ownership—lose their normal input.(8–11)

We have previously demonstrated that SCI leads to weaker sensorimotor network connectivity and maladaptive neuroplasticity, compared to non-injured adults.(9) Without accurate MBR, the brain cannot locate the (paralyzed) limbs, thus, severely limiting motor recovery (e.g., walking), even when residual muscle strength exists, because the brain does not know how or where to activate those muscles. This same maladaptive cortical reorganization is a well-documented contributor to neuropathic pain onset and maintenance.(3,4,12,13) As the parietal operculum and insula are key regions for both MBR and pain,(14,15) restoring MBR is expected to improve sensorimotor recovery and reduce neuropathic pain.

Our previous pilot clinical trial demonstrated that Cognitive Multisensory Rehabilitation (CMR)─a physical therapy approach designed specifically to restore MBR─ holds significant promise, increasing brain activity and connectivity in areas relevant for MBR and sensorimotor function.(10,16) The clinical sensorimotor recovery after CMR is consistent with other studies in stroke populations.(8,17–20) Our earlier case studies further showed that CMR helped restore vision, sensory, and motor function in a person with cortical blindness and quadriplegia, and, relieved phantom limb pain after amputation.(21,22)

Based on these findings, we investigated whether 8 weeks of CMR could yield greater sensorimotor recovery than an active control intervention. We selected adaptive fitness as the control because it is a standard, community-accessible exercise program, commonly provided to adults with SCI following clinical neurorehabilitation. While systematic reviews show that fitness training improves cardiovascular health and upper limb strength, these programs, excluding interventions needing more hands-on assistance like body-weight supported treadmill training, typically focus on non-paretic limbs and do not explicitly target MBR restoration or neuropathic pain relief.(23–25).

Because MBR disruption occurs regardless of pain status, we recruited a broad SCI cohort to target this underlying mechanism for sensorimotor recovery, rather than restricting inclusion to those with neuropathic pain. Still, we prospectively tracked neuropathic pain as an important secondary outcome. We hypothesized that CMR would restore MBR and thereby reduce neuropathic pain in participants presenting with it. We therefore aimed to determine if 8 weeks of CMR or adaptive fitness (1) improved sensory and motor function (primary) and reduced neuropathic pain when present (secondary), and (2) enhanced brain activity and connectivity related to sensorimotor function and body awareness.

2 Materials and methods

3 Results

Fig. 1 displays the CONSORT flow diagram. Of 27 recruited participants with SCI, 3 were excluded (baclofen pump, cognitive impairment from brain injury, inability to self-transfer), and 8 withdrew during the study for various reasons [e.g., illnesses unrelated to the study (n=2), i.e., acute ulcer wound, COVID-19; the baseline fitness level of participants was too high to participate in the study (n=3); or were not responding to communication after consenting (n=3)]. We collected MRI scans for 13 participants at baseline. Reasons for not having an MRI were either a bullet in the body or a heating risk of a Songer cable as part of a cervical fusion.

• Download figure
• Open in new tab

Fig. 1.

CONSORT flow diagram.

Sixteen adults were randomized to the interventions. The group had an average age of 52 ± 8 years (range of 38-64 years), 6 women, and 12 men; 25% African American, 75% White non-Hispanic adults, and 19% living in rural areas. They were 13 ± 10 years after SCI (range of 1-29 years). Six adults had tetraplegia, one had Brown-Sequard, and 10 had paraplegia. There were no significant baseline differences in AIS or NRS outcomes between the groups. The demographic and clinical information of the 16 participants are detailed in Table 1.

All 8 participants completed the CMR intervention and all 8 completed the 3-month follow-up phase. In the fitness group, 8 participants started the 8-week adapted fitness intervention and one person withdrew during the intervention phase because of pneumonia. Of the remaining 7 that completed the 8-week intervention, 6 participants completed the 3-month follow-up phase (1 did not respond to communication at the time of the follow-up testing). All participants averaged 135 min/week (3 sessions of 45min/week) of intervention throughout the 8 weeks. On occasion, missed sessions due to illness were made up the following week. The final analysis included 15 participants in total: 8 participants in the CMR group and 7 people in the fitness group, with n=6 for the 3-month follow-up in the fitness group.

3.1 Clinical outcomes

Clinical outcomes for Aim 1 are reported in Table 2 with effect sizes for post-intervention and 3-month follow-up vs baseline, reflecting the dual burden of SCI across sensorimotor function and neuropathic pain.

CMR yielded greater sensorimotor improvements than adaptive fitness post-intervention. AIS effect sizes were large for CMR (touch: Cohen’s d=1.54; pinprick: d=1.83; lower limb motor function: d=1.32) and small-to-moderate for adaptive fitness (touch: d=0.49; pinprick: d=0.53; lower limb motor function: d=0.74, Fig. 2). NRS effect sizes were also larger for CMR (core: d=2.19; upper limb: d=0.69; lower limb: d=0.74) than adaptive fitness (core: d=0.73; upper limb: d=0.34; lower limb: d=0.00, Fig. 3).

• Download figure
• Open in new tab

Fig. 2. AIS exam.

Bar graphs show average left and right scores for touch, pinprick sensation, and lower limb strength at baseline, post-intervention, and 3-month follow-up for the CMR and adaptive fitness groups.

• Download figure
• Open in new tab

Fig. 3. NRS exam.

Bar graphs show scores for core strength, upper limb and lower limb function at baseline, post-intervention, and 3-month follow-up for the CMR and adaptive fitness groups.

At 3-month follow-up compared to baseline, CMR maintained large AIS effect sizes (touch: d=1.73; pinprick d=2.07; lower limb motor function: d=1.28, Fig. 2) and moderate-to-large NRS effect sizes (core: d=1.74; upper limb: d=0.79; lower limb: d=0.78). Adaptive fitness group showed moderate AIS effect sizes (touch: d=0.56; pinprick d=0.74; lower limb motor function: d=0.63) and moderate-to-large NRS effect sizes (core: d=0.62; upper limb: d=0.85; lower limb: d=0.65, Fig. 3).

Notably, 53% of participants across both groups (n=4 in each group) achieved maximum upper limb NRS scores at baseline, reflecting 7 individuals with paraplegia, and 1 person with quadriplegia. The person with quadriplegia had a dislocated rather than a fractured spine, which may have contributed to a complete motor recovery, yet the participant still had moderate pinprick sensory impairment throughout the trunk at baseline. In sum, this expected ceiling effect constrained the observable upper limb change scores across the cohort.

Neuropathic pain results across the full cohort (n=15), showed a tendency toward divergent outcomes between the two interventions. However, interpretation is limited by the small sample size, given that only 10 out 15 participants had neuropathic pain at baseline, and with large NPRS standard deviations, reflecting substantial inter-individual variability (baseline highest neuropathic pain severity range 3-10/10).

Accounting for this variability and the fact that 5 participants did not have pain at baseline (3 in the CMR group, 2 in the adaptive fitness group), CMR showed small post-intervention effect sizes for pain reduction (highest NPRS: d=0.48; average NPRS: d=0.46) and small-to-moderate effects at 3-month follow-up, compared to baseline (highest NPRS: d=0.63; average NPRS: d=0.31). CMR also reduced severity of neuropathic pain characteristics (DN4) with a moderate effect size sustained at follow-up (d=0.59 at both timepoints).

The adaptive fitness group showed a tendency toward increased neuropathic pain post-intervention (highest NRPS, d=−0.19) and neuropathic pain characteristics (DN4, d=−0.38), but a slight decrease in average pain (NPRS, d=0.38). At 3-month follow-up compared to baseline, highest NPRS was unchanged (d=0.00), average NPRS slightly decreased (d=0.41), neuropathic pain characteristics slightly worsened in severity (d=−0.41).

In the CMR group, 2 participants took neuropathic pain medication. One reduced Lyrica by 88% at the 3-month follow-up, along with 2-point reduction in highest neuropathic pain. The other participant reduced medical cannabis by 50% while maintaining the Gabapentin dosage, with no change in pain intensity levels. All medication dosage changes were authorized by their treating physicians. In the fitness group, 2 out of the 4 participants on pain medication increased Gabapentin by 50-66% (one reporting more tingling, the other a 2-point reduction in highest pain at follow-up). Another person reduced Hydrocodone by 29% post-intervention but increased it by 12.5% at follow-up (compared to the baseline dosage), while keeping Amitriptyline stable and pain slightly worsening by one point. The last participant maintained stable Gabapentin and Baclofen dosages across the study with no pain changes.

Beyond our sensorimotor and neuropathic pain indices, we evaluated a comprehensive suite of secondary clinical measures to capture the broader functional and psychosocial impacts of the interventions. These outcomes were analyzed across the full cohort (n=15) except for SCI-FI/AT, which has a separate scoring system for paraplegia and quadriplegia.

3.1.1 Function

Both groups showed large PSFS effect sizes, for self-identified functional goal attainment (CMR: d=1.99, adaptive fitness: d=1.07), sustained at follow-up (CMR: d=1.95; adaptive fitness: d=1.81).

For SCI-FI/AT paraplegia, CMR (n=5) showed large effect sizes for basic mobility (d=1.47), self-care (d=1.41), and fine motor function (d=1.38), with moderate improvements in ambulation (d=0.61), all maintained at follow-up (d=1.37; d=1.41; d=1.27; d=0.61, respectively). The adaptive fitness results (n=4 at post-intervention; n=3 at follow-up) were more varied and should be interpreted cautiously given the small sample size. Basic mobility improved moderately post-intervention (d=0.78) but decreased to baseline levels at follow-up. Self-care declined after the fitness intervention, then improved at follow-up (d=0.58). Fine motor function was unchanged after fitness (d=0.00) but improved at follow-up (d=0.58). Finally, ambulation improved post-intervention (d=0.50) and further at follow-up (d=1.09).

For adults with quadriplegia (n=3 per group), CMR showed large improvements in SCI-FI/AT basic mobility (d=1.16) and self-care (d=1.16) at both timepoints, versus moderate-to-large fitness improvements (selfcare: d= 0.58; basic mobility: d=0.87). All 3 CMR participants had maximum fine motor scores at baseline (d=0.00) precluding improvement. Fitness showed large fine motor function gains (d=1.00) at both timepoints. Ambulation improved in both groups, with moderate CMR effect sizes (d=0.58) and large fitness effect sizes (d=1.12) at both timepoints.

Mobility, as indexed by the CHART-S, reflected gains following CMR (d=1.18) compared to adaptive fitness (d=0.37), with moderate effect sizes in both groups at the 3-month follow-up (d=0.76 for CMR; d=0.55 for fitness).

3.1.2 Body awareness and motor imagery

KVIQ showed large CMR effect sizes at both timepoints (d=1.02; d=1.07) while fitness showed moderate improvements post-intervention (d=0.73) but a decline at follow-up (d=−0.60).

CMR produced immediate and sustained FAS improvements (d=0.61; d=0.52), whereas fitness had minimal impact (d=−0.13; d=0.10, respectively). PAS improved moderately in both groups post-intervention (CMR: d=0.60; fitness: d=0.54) but there were only small improvements compared to baseline at follow-up (CMR: d=0.25; fitness: d=0.34).

BARQ-R (body tension, emotional body awareness, awareness of breathing) showed no immediate post-intervention effects, but there were moderate CMR improvements (d=0.59) and a small fitness improvement (d=0.43) at follow-up.

Pertaining to interoception, the eight domains of MAIA-2 showed small improvements immediately post-CMR (ranging from d=0.11 to 0.43), except for attention regulation (d=−0.42), growing to small-to-large improvements at follow-up (ranges of d=0.11 to 1.14; except for attention regulation: d=−0.12). Fitness only improved in 3 subdomains with small-to-moderate effect sizes (d=0.25 to 0.77), while 5 subdomains worsened (d=−0.38 to −1.37). This pattern persisted at follow-up (3 improved: d=0.04- 0.15; 5 subdomains worsened compared to baseline levels: d=−0.24 to −0.66).

3.1.3 Spasticity, autonomic function, sleep, and psychosocial wellbeing

Spasticity frequency and intensity improved comparably in both groups post-intervention and at follow-up (ranges d=0.38 to 0.65) except for spasm frequency in the CMR group at follow-up, which was unchanged compared to baseline (d=0.00).

Autonomic function improved moderately with CMR at both timepoints (d=0.68) while fitness showed slight worsening post-intervention that nearly resolved to baseline at follow-up (d=−0.55; d=−0.18).

Both groups also experienced small improvements in sleep quality (PSQI; CMR: d=0.34 and 0.12; fitness: d=0.31 and 0.22, respectively post-intervention and at 3-month follow-up).

Psychosocial wellbeing outcomes presented a more nuanced profile. Trait anxiety decreased substantially more with CMR (d=1.53; d=1.59) than fitness (d=0.56; d=0.72) at both timepoints. TSK improved in both groups at 3 months follow-up (CMR: d=0.75, fitness: d=0.78), but state anxiety, PHQ-9, and MSES showed no meaningful changes in either group.

3.2 Brain imaging outcomes

For aim 2, baseline results confirmed the validity of our tasks and region-of-interest selections. Resting state seeds (insula, parietal operculum, and post-cingulate cortex) connected with sensorimotor areas (pre-and postcentral gyri), pain perception and body awareness regions (parietal operculum and insula), and visuospatial body map areas (superior parietal lobe, supramarginal gyrus, and angular gyrus). Task-based fMRI similarly showed baseline activation in these same brain regions.

Concurrent with clinical improvements, resting-state fMRI showed increased brain connectivity after CMR compared to baseline (Fig. 4, Table 3), specifically, stronger connectivity between the right parietal operculum (as region-of-interest) and the right precentral gyrus (Fig. 4A), and between the right posterior cingulate cortex (as region-of-interest) and the left precentral gyrus, left insula, and left parietal operculum (parts OP1/OP4, (Fig. 4B). No connectivity changes were found in the fitness group.

• Download figure
• Open in new tab

Fig. 4. Pre-post CMR changes in connectivity strength.

(A) Stronger parietal operculum network connectivity after 8 weeks of CMR: right parietal operculum (ROI) to right precentral gyrus (PreCG). (B) Stronger posterior cingulate cortex network connectivity after CMR: right posterior cingulate cortex (ROI) to left precentral gyrus (PreCG), left insula (INS), and left parietal operculum (OP1/OP4).

CMR also increased task-based brain activity compared to baseline: Imagining feeling the left leg, activated the pre- and postcentral gyri, and superior parietal lobe to a greater extent post-CMR vs baseline (Fig. 5, Table 3). The whole-body imagery task produced greater activation post-CMR in the right pre- and postcentral gyrus, bilateral supramarginal gyrus, superior parietal lobe, and angular gyrus (Fig. 6A, Table 3).

• Download figure
• Open in new tab

Fig. 5. Pre-post CMR changes in brain activity during leg kinesthetic imagery.

Post-CMR, enhanced brain activity was observed in the right pre- and postcentral gyri, and right superior parietal lobe.

• Download figure
• Open in new tab

Fig. 6. Pre-post changes in brain activity during whole-body imagery for CMR and adaptive fitness.

(A) Post-CMR increased activation in the supplementary motor area (SuppMotor), right pre-and postcentral gyrus (PreCG, PostCG), bilateral supramarginal gyrus (SMG), superior parietal lobe (SPL), and angular gyrus (AG). (B) Post-adaptive fitness, only smaller areas of the left supramarginal gyrus (SMG) and superior parietal lobe (SPL) were activated.

In the adaptive fitness group, we saw much smaller activation increases in the left supramarginal gyrus and superior parietal lobe compared to baseline during the whole-body imagery fMRI task (Fig. 6B, Table 3).

4 Discussion

In this pilot clinical trial, our aims were to determine if 8 weeks of CMR or adaptive fitness (1) improved sensory and motor function (primary) and reduced neuropathic pain when present (secondary), and (2) enhanced brain activity and connectivity related to sensorimotor function and body awareness.

The results provide promising preliminary evidence that CMR may address the dual burden of sensorimotor deficits and chronic neuropathic pain in adults with SCI. Both interventions yielded clinical benefits, but CMR produced greater sensorimotor function improvements (reflected by large effect sizes for AIS and NRS) than fitness. These gains should be interpreted in light of injury variability. For instance, upper limb NRS effect sizes were constrained by the fact that 8 participants (7 with paraplegia, 1 with quadriplegia) achieved a maximum score for this item at baseline. The remaining two participants with paraplegia did not reach maximum scores on NRS upper limb function due to a right frozen shoulder in one, and a high thoracal lesion (T4) limiting trunk balance during upper limb activities in the other.

Both groups showed large PSFS effect sizes and improvements on nearly all SCI-FI/AT measures for adults with quadriplegia. For adults with paraplegia, self-reported ease of performing daily functions was markedly better in the CMR group vs the fitness group. The CMR group experienced large improvements for basic mobility, selfcare, and fine motor function, and moderate improvement for ambulation across both time points. The fitness group showed more varied results, with no lasting basic mobility improvements, moderate self-care and fine motor function gains at follow-up, and large ambulation gains at follow-up.

These results show the importance of using both objective and subjective measures when assessing motor function. Objective measures do account for adaptive equipment use, whereas self-reported daily function does, which can produce divergent results. For example, the participants in the fitness group did not improve in objective measures as much as the CMR group for lower limb function; yet, participants with paraplegia self-reported on average that walking with adaptive equipment became easier after the fitness intervention and at follow-up.

Sensory function also improved more after CMR than after adaptive fitness. Given how closely MBR is integrated with sensorimotor function, it follows that both sensory and motor function improve when whole-body awareness, awareness of the relationship between body parts, and spatial body awareness are restored. Across our pilot studies, CMR consistently produces both sensory and motor function gains, which is encouraging, as sensory recovery has been difficult to achieve with other neurorehabilitation treatments.

To date, only one recent study of D’Amico et al. (2026)(63) examined the effects of cervical transcutaneous spinal cord stimulation combined with rehabilitation, finding that 55% of participants with chronic SCI increased total AIS sensory scores by more than 5.19 [the minimum clinically important difference, MCID, for acute and subacute SCI(64) populations; there is no current MCID available for chronic SCI].(63) Their study design did not include a control group or follow-up measures.(63) In comparison, our CMR group improved total AIS sensory scores by more than 7 points, at both post-intervention and at 3-months follow-up: touch improved post-CMR by 7.50±4.88 and 8.06±4.66 points at follow-up. Pinprick improved by 8.50±4.64 and 9.88±4.78 points, respectively. These are the largest improvements we have encountered in literature, sustained for at least 3 months. Notably, direct score comparisons are limited by differences in injury profile: D’Amico et al. (2026)(63) enrolled only participants with cervical incomplete lesion (with majority C4-C6; AIS C-D), whereas our cohort included 44% cervical incomplete lesions (C1-C7; majority of this subgroup AIS D). Nonetheless, our prior study in 26 adults with SCI-related neuropathic pain (where 53% had AIS A) demonstrated that CMR is also effective in persons with complete SCI (AIS A).(16) In our present study, fitness improved touch by 2.14±4.38 and 2.50±4.47 points; and pinprick 2.00±3.81 and 3.50±4.72 points, at the two timepoints.

Furthermore, CMR produced consistent small-to-moderate reductions in neuropathic pain and symptom severity (DN4) while the fitness group showed no change or slight exacerbation of these symptoms (except for a small decrease in average pain). With only 10 participants with pain (5 per group) these findings are preliminary. However, our prior study in 26 adults with SCI-related neuropathic pain demonstrated significant pain reduction after 6 weeks of CMR (highest NPRS: almost 5-point reduction; average NPRS: 4-point reduction; 35% pain-free) sustained at 1-year follow-up in a subgroup of 14 participants who agreed to be tested.(16)

Clinical improvements observed after CMR coincided with enhanced brain connectivity and activation. Our earlier study demonstrated that adults with SCI have weaker connectivity networks and less brain activation in areas responsible for sensorimotor function and MBR than healthy adults.(9) In this study, following 8 weeks of CMR, resting-state connectivity strengthened between the right parietal operculum and right precentral gyrus; and between the right posterior cingulate cortex (as region-of-interest) and the left precentral gyrus, left insula, and left parietal operculum (parts OP1/OP4).(65,66)

The posterior cingulate cortex plays a key role in self-localization, as identified in an out-of-body fMRI study,(11) though the intraparietal-retrosplenial-hippocampal connections appear more relevant in that context. Given our small sample size, we cautiously speculate that the observed connectivity seen in our study between the posterior cingulate cortex and precentral cortex represents self-localization of the ‘actual body’/‘of the self’ being linked to motor areas to guide movements, analogous to how the parietal operculum, insula, and posterior parietal cortex collects, integrates, and transmits multisensory information for motor guidance.

Imagery of feeling the left leg increased activation in the pre- and postcentral gyri and superior parietal lobe, while the whole-body imagery task activated the right pre- and postcentral gyrus, bilateral supramarginal gyrus, superior parietal lobe, and angular gyrus to a greater extent post-CMR vs baseline. Consistent with our prior CMR brain imaging findings, the brain not only processes and interprets sensory information from the affected areas, but integrates multisensory and visuospatial information into a coherent conscious spatial body awareness to accurately guide movements.(4,8–10,67)

The fitness group showed markedly smaller activation increases compared to baseline, limited to the left supramarginal gyrus and superior parietal lobe during whole-body imagery.

While causal relationships cannot be established, the concurrent improvements in MBR-related networks and clinical outcomes support our central hypothesis that targeting MBR can address both sensorimotor deficits and neuropathic pain. SCI-related sensory deficits create a mismatch between expected and received sensations in the brain, disrupting MBR.(3–5) Given the brain is wired for survival, when unable to locate the (paralyzed) body parts, the brain signals this incongruence through spasms and neuropathic pain. We consistently observe that when participants regain awareness of their body parts’ location, spasms resolve and neuropathic pain diminishes.

Surveys confirmed that body awareness and imagery improved more with CMR than fitness. For example, at 3-month follow-up, the CMR group showed moderate-to-large effect sizes for BARQ-R and MAIA-2’s subdomains of noticing, not worrying, emotional awareness, self-regulation and body listening─no moderate or large effect sizes were seen in the fitness group. This confirms prior findings in a larger SCI cohort showing improved BARQ-R (n=69), and MAIA-2’s subscale ‘not worrying’ (n=43) after CMR or Qigong.(68) By helping participants correctly identify sensations of body positions, touch, pressure, and other multisensory variables, CMR reduces the brain’s alarm response: sensations are recognized rather than flagged as unknown, diminishing neuropathic pain signaling.(69)

Participants reported that they were more attuned to their body signals, and reacted to them faster, e.g., when they felt they had a urinary tract infection. Nevertheless, none of the scales related to body awareness or interoception directly assessed SCI-specific MBR. We are currently developing a new SCI-BodyMap to address this gap.(70,71)

The clinical and neuroimaging findings parallel participant reports of increased body awareness after CMR─such as being able to accurate sense where there paralyzed limbs were in space, as well as the contact of the soles of the feet with the floor─which subsequently facilitated transfers and activities of daily living. Crucially, reports from participants in the CMR group encompassed benefits across several domains: sensory, motor, autonomic, and cognitive function. Here are some examples below:

[woman in her 50s] “I think of parts of my body that I have not thought about in over 20 years and it’s interesting. I find myself touching my feet and legs a bit differently and recognizing they are in fact still there. Sensation in the feet returns, also sensation on the soles of the feet. Transfers feel easier.”

[man in his 50s] “The right big toe has woken up, and I have more and more ongoing sensation in the buttocks. This is different from other therapies. I am now able to get out of a low sports car without help. My hips move better and feel more free. I am able to get in the RV up the stairs and then sit down with weight equal[ly distributed] in my feet. I continue to walk faster with smart crutches.”

[man in his 30s] “My right foot is not dragging anymore. I use a new way of walking now, more natural, faster and smoother, and I am not fatigued. I stand more on top of my feet. It was a no-brainer that it [CMR] was worth it.”

[man in his 30s] “I used to lean on the right side all the time, now I am sitting straight. My left hand function is better now. I can hold the bottle with my left hand; I could not do that as well before. My handgrip is normal now. Before it would slip (grip was too weak) and that is not happening now. I stood up with a walker; I could not do that before for more than 20 years. Transfers are better.

Before I could only transfer from right to left and now, I can also transfer from left to right. Neuropathy is half of what it used to be.”

[woman in her 60s] “For the first time I was able to touch my left foot with my hand in crossed leg position, without back support by leaning my trunk a little forward and using the right foot as a support.”

[[man in his 30s] “It is an interesting journey. The CMR helps me also with my depression and clear-thinking problem. I am at a point where I start getting more clarity, feeling awake, and clear thinking, like a runner’s high. I no longer feel “crushing” pain on the right side. This week, I could feel my cat on my body on the left side during the night in an area I am usually not able to feel.”

[man in his 50s]: “I did not know that internal map was incongruent with reality; it’s very revealing; a 180-degree paradigm shift. Fine tuning what needs to be finetuned. Very interesting stuff. I had no clonus in the left leg during sessions last week. I do not have restless leg syndrome anymore at night, and that used to wake me up before. I can now transfer from a lower surface to my chair whereas previously, the treatment table needed to be significantly higher. Overall, transfers are better now.”

[woman in her 50s] “(week 2): I can cut my nails with both hands now which I was not able to do before. (week 5): all my goals are reached: I can now bike 12 miles, clip toenails (before it would slip), I can climb a ladder, I can snap with both hands (left better than right). Feet and hands are not going numb anymore, and at night when I lay on my side, my legs do not go numb anymore. My hands and feet do not feel cold anymore.”

Participants also applied what they learned in CMR to daily life, such as thinking about being grounded; knowing where the feet are; improved transfers and body weight shifting; greater attentiveness to leg and foot sensations; better awareness of when to reposition in the wheelchair.

For the fitness group, feedback included noticeable balance improvements, which helped with walking, increased strength, having more energy, and easier transfers. One person would have preferred one-on-one workouts with a personal trainer (instead of oversight by a trainer), but overall, the group enjoyed working out in the adaptive fitness gym.

Our findings must be interpreted within several limitations. Most importantly, the small sample size(1) precludes definitive efficacy claims and warrants a larger controlled trial, particularly for neuropathic pain, given the wide variability in the present cohort; (2) prevented between-group neuroimaging comparisons and mediation analyses to establish causal brain-outcome links; and (3) precluded subanalyses by injury level (paraplegia vs quadriplegia) or other SCI-related variables. Regarding health disparities, 25% were of minority race, none were veterans, 19% lived rurally, 38% were women, and 25% had significant economic distress or income below the official poverty level. While we achieved some satisfactory levels of recruitment of patients with healthy disparities, we aim to recruit more veterans in future studies, given their higher prevalence of SCI compared to community members (in 2019, 42,000 or about 14% of all adults with SCI were veterans, numbers that are likely higher today).(72,73) A remotely delivered CMR version could also extend reach to more participants living in rural locations.

Despite these limitations, the large CMR effect sizes for objectively assessed sensorimotor recovery (AIS and NRS), alongside indications of reduced neuropathic pain and enhanced body awareness are clinically noteworthy. The small-to-moderate effect sizes in the adaptive fitness group support its viability as active control in a larger trial. Notably, this study demonstrated potential for improvements in sensation and movement even up to 29 years after SCI, consistent with our prior work showing gains up to 56 years post-SCI.(16)

Villiger et al. (2013)’s pilot study with virtual reality-augmented lower limb training combining action observation and training, reported reduced neuropathic pain beyond MCID in 5 of 9 adults with incomplete SCI (ASIA D).(74) While mirror therapy and augmented reality with mirror therapy, have been explored to create a perceptual illusion of movement (e.g. walking) for pain relief with varying results,(75–77) to our knowledge, CMR is the only therapeutic approach that consciously targets MBR restoration using the participants’ own body parameters, with consistent evidence of increased brain function after CMR. The therapist’s guidance in restoring MBR is central to these gains. Together, these preliminary results indicate that targeting maladaptive cortical changes may yield greater dual-burden recovery than standard adaptive fitness, strongly warrant larger randomized controlled efficacy trials.

5 Conflict of Interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

6 Author Contributions

Every author contributed to the study: Conceptualization (AVdW, LM); Funding acquisition (AVdW); Data curation (AVdW, SC, WD); Formal analysis and data interpretation (AVdW, LZ, TH, KL, BM, LM); Cognitive Multisensory Rehabilitation intervention (SB, JB); Recruitment (AVdW); Data collection (AVdW, SC, WD, BM); Supervision (AVdW); Validation (AVdW, SC, WD, LZ); Visualization (AVdW, LZ); Consultant with lived experience of SCI (SMS); PM&R screening for adaptive fitness readiness (RN); SCI study doctor (LM); Writing – original draft (AVdW); Writing – review & editing (AVdW, SC, SB, JB, WD, LZ, TH, BM, SMS, RN, LM, KL). We confirm that more than one author has directly accessed and verified the underlying data reported in the manuscript (AVdW, SC, JB, LZ).

7 Funding

The Paralyzed Veterans of America Research Foundation Grant #3187 funded this study (IRB no. STUDY00014710; ClinicalTrials.gov Identifier: NCT05167032). Additional support was provided by the National Center for Advancing Translational Sciences of the National Institutes of Health Award Number #UM1TR004405, the Biotechnology Research Center: P41EB015894, the National Institute of Neurological Disorders & Stroke Institutional Center Core Grants to Support Neuroscience Research: P30 NS076408; and the High-Performance Connectome Upgrade for Human 3T MRI scanner: 1S10OD017974. Image processing resources were provided by the Minnesota Supercomputing Institute (MSI) at the University of Minnesota. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. None of the funding sources had a role in study design, data collection, data analysis, data interpretation, or writing of the report. None of the authors have been paid to write this article. Authors were not precluded from accessing data in the study, and they accept responsibility to submit for publication.