Preliminary efficacy of cognitive multisensory rehabilitation for neuropathic pain in chronic spinal cord injury: a phase I randomized controlled trial

Abstract

Neuropathic pain affects 69% of adults with spinal cord injury (SCI) and responds poorly to current treatments. Emerging evidence suggests that impaired mental body representations (MBR) may contribute to neuropathic pain. The parietal operculum and insula, key regions for pain perception and body awareness, show disrupted connectivity after SCI. We aimed to identify changes in brain function, neuropathic pain, and sensorimotor function after Cognitive Multisensory Rehabilitation (CMR) vs. usual care. CMR is a physical therapy which improves MBR. In this Phase I randomized delayed-treatment trial, 26 adults with SCI-related neuropathic pain were randomized to either 6 weeks of immediate CMR (group A), with 6-week + 1-year follow-up; or a 6-week observational period + 6-week CMR + 1-year follow-up (Group B). We assessed neuropathic pain intensity, sensorimotor function, resting-state and task-based fMRI. Neuropathic pain significantly decreased after CMR in group A + B (Numerical Pain Rating Scale, highest pain change: -4.92 ± 2.92 points, d = -1.68; average pain − 4.12 ± 2.23 points, d=-1.84), and sensorimotor function improved, while no changes followed the observational period. Benefits were maintained at one year. Post-CMR fMRI revealed stronger resting-state parietal operculum connectivity, and enhanced activation in pain- and MBR-related brain regions during a toe sensation task. Results suggest preliminary CMR efficacy for neuropathic pain reduction and functional improvement.

Clinical Trial registration: ClinicalTrials.gov, NCT04706208; first registration 08/01/2021; first participant enrolled 09/01/2021.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-44859-w.

Introduction

About 308,620 Americans have a spinal cord injury (SCI), causing sensorimotor impairments and mental body representations (MBR) impairments^1–4. MBR are internal representations of body awareness and visuospatial body maps. Body awareness refers to an awareness of body sensations, positional awareness of body parts, and whole-body awareness⁵. Visuospatial maps help a person recognize the body’s spatial location at each moment to help guide movements. MBR are an integral part of the sensorimotor system in the brain: peripheral multisensory information converges in the multimodal integration network (which includes parietal operculum and insula) to form body awareness. These areas continuously interact with the posterior parietal cortex, where dynamic visuospatial body maps are formed⁶. Because of reduced/absent sensory information post-SCI, the brain cannot construct MBR correctly, impacting sensorimotor functional recovery because the brain does not recognize the body’s spatial location to guide movements^1,3.

MBR deficits are thought to generate and/or maintain chronic neuropathic pain^1,2,7,8, which occurs in 69% of adults with SCI⁹. Pain medications have limited effects, and some carry high risks for addiction and adverse effects¹⁰. New interventions are needed, but neuropathic pain-related brain mechanisms are unclear. Studies have shown inconsistent results as to which brain regions and networks influence neuropathic pain in this heterogeneous population^11–13. While the causal relationship between MBR impairments and neuropathic pain remains to be established, several lines of evidence suggest a potential link: parietal operculum (i.e., secondary somatosensory cortex, parts OP1/OP4) and insula are key areas for body awareness and pain perception^14,15. The present study tests the hypothesis that interventions improving MBR may reduce neuropathic pain through reactivating brain function and strengthening networks of these two key brain areas.

Cognitive Multisensory Rehabilitation (CMR)¹⁶ is such a candidate. Our prior work demonstrated disrupted resting-state OP1/OP4 connectivity in adults with chronic stroke compared to resting state OP1/OP4 connectivity in healthy adults⁶. A 6-week CMR intervention activated OP1/OP4 and insula;^6,17–19 and strengthened the resting-state OP1/OP4 connectivity in adults with chronic stroke, concurrent with upper limb sensorimotor improvements sustained at least 1 year⁶ Reduced pain and/or improved sensorimotor function after CMR has been reported in adults with stroke and with shoulder impingement syndrome^6,20–22. An adult with cortical blindness and tetraplegia improved visual and sensorimotor function after CMR, > 2 years after cortical damage²³. Phantom limb pain was alleviated after CMR in a person with amputation²⁴.

More information about CMR in adults with SCI can be found in the protocol paper²⁵ and briefly below and in the method section. The CMR therapist guides the patient by asking questions or guiding them with kinesthetic motor imagery (i.e., imagining the feeling in the body, rather than visualizing the body or body movement) during multi-sensorimotor discrimination exercises to help them restore MBR^6,16,23. For example, a CMR therapist asks patients to focus on past memories of feeling body positions or feeling textures and relate that memory to the feeling in the current exercise of, e.g., differentiate one leg position relative to the other leg; or differentiate textures under the foot. Adults with SCI-related neuropathic pain often have an altered perception of size, dimension, body weight, pressure, or touch in painful and/or sensorimotor-impaired body parts, which provides a strong rationale for using CMR to treat neuropathic pain and sensorimotor function²⁶. Because neuropathic pain occurs in adults regardless of their injury level or SCI completeness, CMR can also be provided to adults with severe sensorimotor impairments. In sum, we hypothesize that CMR, by improving MBR and potentially strengthening OP1/OP4 network connectivity, may reduce neuropathic pain and improve sensorimotor function in adults with SCI.

Additionally, fMRI studies have described a phenomenon known as ‘discomplete SCI’, in which no clinical evidence of fiber tract function is detectable through clinical sensory and motor examination, but where residual fibers may still possess the potential to influence perception and function^27,28. Notably, 48% of adults classified as having complete SCI─as determined by the International Standards for Neurological Classification of Spinal Cord Injury (INSCI) exam, also called ASIA Impairment Scale (AIS)─ showed brain activation in the primary somatosensory cortex (S1) during a big toe stimulation fMRI task, either by human touch²⁷ or with a plastic tool²⁸. This finding demonstrates the presence of ‘sensory discomplete SCI’^27,28. The clinical significance of ‘discomplete SCI’ remains uncertain. Nevertheless, Lee et al. (2015)²⁹ used fMRI to study passive toe movements before spine decompression and fixation, discovering that all participants (100%) who exhibited sensorimotor cortex activation with passive toe movement prior to surgical spine decompression and fixation improved on AIS scores (from AIS A to B or C) post-surgery²⁹. In contrast, only 43% improved if they had ipsilateral sensorimotor cortex or frontal lobe activation, and just 9% improved if no brain activation was present before decompression²⁹.

Therefore, our main aim was to identify changes after CMR, in brain function, neuropathic pain, sensorimotor function, SCI-related and pain-related outcomes in adults with SCI-related neuropathic pain and with AIS A-D.

Results

Twenty-eight adults with SCI were consented between January 2021-December 2022. Figure 1 shows the CONSORT study flow chart. Two participants did not start the baseline testing or intervention. One participant did not meet the inclusion criteria for neuropathic pain intensity (the highest neuropathic pain was less than 4/10). The other participant contracted COVID-19 prior to the start of the study, necessitating hospitalization, and therefore did not continue the study.

Fig. 1.

CONSORT flow diagram.

Supplementary Table S1 shows the demographic and clinical characteristics data of enrolled adults with SCI (n = 26). Supplementary Table S2 shows the baseline clinical and behavioral outcomes in adults with SCI. All participants with SCI completed rehabilitation in acute/outpatient rehabilitation centers after their SCI incident. They did not receive rehabilitation (other than the CMR intervention) during the study.

All 26 adults with SCI completed the CMR intervention (3x/week for 6 weeks, 18 sessions total) and pre-post assessments. There were no study-related adverse events. We added 1-year follow-up assessment as a protocol modification after the 6-week assessment was over, and, because this was not communicated at the start of the study, we believe that is the reason why only 54% of participants (n = 14) agreed to the 1-year follow-up or parts of it. The 46% attrition rate at 1-year follow-up limits our ability to draw definitive conclusions about long-term maintenance of treatment effects.

The 26 participants were 52 ± 15 years old, 18 ± 16 years post-SCI (range 1–56 years), 7 were women (27%), 8% were of diverse race, 8% of diverse ethnicity (Hispanic), and 12% lived in a rural area. Fourteen participants had complete paraplegia, eight had incomplete paraplegia, two had incomplete tetraplegia, and two had Brown-Sequard Syndrome. 46% of participants used a manual wheelchair; 12% a combination of manual and power wheelchair; 19% a power wheelchair, and 23% walked with or without assistance.

Clinical outcomes

We conducted the repeated-measures ANOVA analysis on the highest, average, and lowest pain scores, which test the effects of CMR interventions on pain score changes while adjusting for their pre-CMR measures.

In addition, for all clinical variables, including pain, we added two-sample t test to compare the pre-CMR measures between groups A & B for all clinical variables. The Benjamini-Hochberg’s (BH) False Discovery Rate-adjusted (BH-adjusted) p-values ranged from 0.36 to 0.99, indicating no difference between the two groups at the combined baseline.

The participants’ highest baseline neuropathic pain was 7.81 ± 1.33 (CI₉₅ = [7.30–8.32]), assessed with the Numeric Pain Rating Scale (NPRS)³⁰. Given that pain levels were not significantly different pre-versus post-observation period in group B (8.50 ± 1.29 points vs. 7.81 ± 1.33 points, p = 0.38 see Supplementary Table S3), we used Group B’s post-6-week observation as ‘baseline’ for Group A + B combined analysis. After CMR (in combined group A + B), the highest neuropathic pain was reduced to 2.88 ± 2.92 (CI₉₅ = [1.76-4.00]). Nine participants were pain-free (34.62%). At 1-year follow-up (n = 14), four participants were pain-free, four other participants had no pain for average pain intensity, four maintained their average pain of 2/10, and two participants scored 5 and 6 respectively on average pain, related to a urinary tract infection or a surgery at the time of the 1-year follow-up, causing a temporary increase in highest and average neuropathic pain levels. They were included in the analysis.

At baseline, 20 of 26 participants used pain medication. Following CMR, two participants stopped all pain medications, three discontinued opioids, and one reduced opioid use by 85%. In total, 12 of these 20 participants (60%) either reduced their weekly medication dosage by 8% to 66%—including opioids—with physician approval or eliminated the need for pain medication after CMR.

Supplementary Table S3 shows the means and standard deviations of clinical assessments tested across all time points and includes within- and between-group statistical analyses. During the first 6 weeks, we found significant differences between group A (immediate CMR, n = 12) and group B (observation period, n = 14) for highest, average, and lowest neuropathic pain³⁰; average nociceptive pain; and sensorimotor function, tested with the International Standards for Neurological Classification of SCI or ASIA Impairment Scale (ISNCSCI or AIS)³¹ neurological exam.

Group A improved in sensorimotor function and had significant neuropathic³⁰ and nociceptive³² pain reduction and pain interference on daily life, mood, and sleep³² after CMR. Group B did not show changes in neuropathic pain during the observation period³⁰ (Fig. 2, left), ISNCSCI³¹ (Fig. 3, left), pain interference, or nociceptive pain levels³² (Supplementary Fig. S1, top).

Fig. 2.

Highest, average, and lowest neuropathic pain rating scores. The line graphs represent the weekly neuropathic pain intensity ratings during the 6-week observation period (Group B, n = 14, in the first 6 weeks, graph on the left) and during the pre-post CMR intervention (group A + B receiving CMR, n = 26, graph on the right) with the one-year follow-up testing included (n = 14).

Fig. 3.

ISNCSCI exam. The bar graphs represent the touch and pinprick sensation as well as lower limb motor function during the 6-week observation period (Group B, n = 14, in the first 6 weeks, bar graph on the left) and during the pre-post CMR intervention (group A + B receiving CMR, n = 26, bar graph on the right) with the one-year follow-up testing included (n = 14).

The repeated-measures ANOVA, adjusted for baseline values, was conducted on the longitudinal pain score data. The repeated-measures ANOVA analysis demonstrated that CMR reduced neuropathic pain intensity in both groups; however the magnitude of pain reduction did not significantly differ between groups A and B (highest pain = 2.64, p = 0.11; average pain = 2.48, p = 0.12; lowest pain = 2.29, p = 0.13). Additionally, two-sample t-tests also indicated that changes pre-post CMR in all other clinical variables assessed were similar between groups A and B, and that the magnitude of these changes pre-post CMR did not significantly differ between groups A and B (Benjamini-Hochberg’s (BH) False Discovery Rate-adjusted p-values range between 0.095 and 0.95). Consequently, we report these results for groups A and B combined.

The within-group analysis of pre-post CMR (Group A + B, Fig. 2, right) showed significant reductions in the highest (change score of -4.92 ± 2.92, large effect size Cohen’s d=-1.68, CI₉₅=[-6.10;-3.74], BH-adjusted p < 0.0001); average (-4.12 ± 2.23, d = 1.84, CI₉₅=[-5.02;-3.21], BH-adjusted p < 0.0001), and lowest neuropathic pain intensity levels (-2.31 ± 2.07, d=-1.11, CI₉₅ = [-3.15;-1.47], BH-adjusted p < 0.0001) on the NPRS³⁰; significant reduction in neuropathic pain interference on activity (-4.69 ± 2.83, d = -1.66, CI₉₅ = [-5.83;-3.55], BH-adjusted p < 0.0001), mood (-3.96 ± 2.89, d = -1.37, CI₉₅ = [-5.13;-2.79], BH-adjusted p < 0.0001), and sleep (-4.42 ± 3.38, d = -1.31, CI₉₅ = [-5.79;-3.06], BH-adjusted p < 0.0001),³² and average nociceptive pain levels³² (-3.05 ± 2.04, d = -1.50, CI₉₅ = [-4.04;-2.07], BH-adjusted p < 0.0001); all sustained at 1-year follow-up (Supplementary Fig. S1, bottom).

Group A + B (n = 26) experienced significantly improved sensation to light touch after CMR as assessed by the ISNCSCI³¹ neurological exam (improvement of 8.81 ± 5.37 points, d = 1.64, CI₉₅ = [6.64;10.98], BH-adjusted p < 0.0001), as well as for pinprick sensation (7.50 ± 4.89 points, d = 1.53, CI₉₅ = [5.52;9.48], BH-adjusted p < 0.0001), and for lower limb muscle strength (3.87 ± 2.80 points, d = 1.38, CI₉₅ = [2.73;5.00], BH-adjusted p < 0.0001). This was also sustained at 1-year follow-up (Fig. 3, right).

After CMR, participants also had significant reduction in spasm frequency (-1.06 ± 1.11 points, d = -0.95, CI₉₅ = [-1.6;-0.50], BH-adjusted p = 0.002) and severity (-0.83 ± 0.92 points, d=-0.90, CI₉₅ = [-1.29;-0.37], BH-adjusted p = 0.003), measured with the Penn Spasm Frequency Scale³³. They also had reduced SCI secondary conditions³⁴ (-8.59 ± 4.37 points, d = -1.96, CI₉₅ = [-10.84;-6.34], BH-adjusted p < 0.0001) (Supplementary Table S3).

At the start of the study, all participants completed the Patient Specific Functional Scale³⁵, by self-identifying three goals related to daily life activities important to them but that were difficult to perform because of neuropathic pain. Goal achievement was rated between 0 (unable to do the activity) and 10 (able to do the activity). The participants had significant improvements in these self-identified goals after (average improvement across the three goals of 5.88 ± 2.56, d = 2.29, CI₉₅ = [4.82;6.94], BH-adjusted p < 0.0001, Supplementary Table S3).

We also assessed the Spinal Cord Injury-Functional Index/Assistive Technology (SCI-FI/AT)^36,37 in 17 out of the 26 participants (because this scale was added later in the study), which measures functional performance across basic mobility, self-care, fine motor function, and ambulation, with separate questions for adults with paraplegia and tetraplegia. We demonstrated significant improved function on self-care in adults with tetraplegia (n = 3) after CMR. Adults with paraplegia (n = 14) improved significantly on basic mobility (improvement of 7.64 ± 5.68 points, d = 1.35, CI₉₅ = [4.36;10.92], BH-adjusted p = 0.0005), self-care (5.36 ± 5.31 points, d = 1.01, CI₉₅ = [2.29;8.43], BH-adjusted p = 0.004), fine motor function (3.07 ± 4.12 points, d = 0.75, CI₉₅ = [0.69;5.45], BH-adjusted p = 0.021) which reflected improvements in balance so they could have their arms and hands free for these activities; and ambulation (2.86 ± 3.30 points, d = 0.87, CI₉₅ = [0.95;4.76], BH-adjusted p = 0.01) after CMR (Supplementary Table S3).

Participants reported markedly improved daily life function: 12 out of 20 participants stood up for the first time with minimal assistance, and 17 out of 18 participants with balance problems at baseline had better balance after CMR, which participants self-reported and which could also be observed during clinical testing. They did not need to hold onto the table anymore with their arms and could move more freely while sitting or changing positions, or while doing transfers. Weight shift during transfers was facilitated by increased feeling and awareness of their leg position in space, the pelvis position, and body weight distribution. Participants reported that their movements were more controlled. One person reported improved wheelchair-driving ability. Three participants reported better walking, and one reported walking faster. Another participant reported that her legs and feet felt connected to her body for the first time in 16 years. A person with tetraplegia reported that instead of thinking “open the hand”, she now more consciously and intentionally thought about what the hand would feel, when, for example, touching the couch with her palm, which improved her hand function.

Participants reported significant reduction in trait anxiety³⁸ (-5.04 ± 7.20 points, d = -0.70, CI₉₅ = [-8.01;-2.07], BH-adjusted p = 0.003), state anxiety³⁸ (-5.04 ± 8.09 points, d = -0.62, CI₉₅ = [-8.38;-1.70], BH-adjusted p = 0.008), mood (PHQ-9³⁹, -1.89 ± 3.16 points, d = -0.53, CI₉₅ = [-2.98;-0.38], BH-adjusted p = 0.02), kinesiophobia⁴⁰ (-2.60 ± 4.58 points, d = -0.57, CI₉₅ = [-4.49; -0.71], BH-adjusted p = 0.014). They also improved in quality of life⁴¹ regarding physical health (8.20 ± 17.21 points, d = 0.48, CI₉₅ = [1.10; 15.30], BH-adjusted p = 0.033) and psychological health (7.24 ± 15.80 points, d = 0.46, CI₉₅ = [0.72;13.76], BH-adjusted p = 0.039) (Supplementary Table S3).

Brain imaging

After CMR, resting-state fMRI revealed stronger OP1/OP4 network connectivity between the right OP1/OP4 and left OP1/OP4, angular and supramarginal gyri (Supplementary Table S4; Fig. 4A), compared to baseline. Additionally, the right-toe sensation fMRI task elicited greater brain activation relative to baseline in the bilateral postcentral gyrus (sensory function), angular and supramarginal gyri, superior parietal lobe (part of the posterior parietal cortex, related to visuospatial body maps); left insula (pain processing, body awareness)^6,42, and frontal operculum (Fig. 4B). These brain regions extend beyond simple stimulus detection or attention, as confirmed by self-report and the clinical testing of touch and pin prick sensation. Because the sensory task was standardized and unpredictable for participants (i.e., the participant did not know when or where the tester would stroke the feet or hands), the observed increases in brain activation suggest that, after CMR, the brain not only processes sensory information from the affected areas (represented in the postcentral gyrus), but also integrates and identifies these sensations in relation to other body parts (insula, angular and supramarginal gyri, superior parietal lobe)─reflecting high-order processing of multisensory and visuospatial information.

Fig. 4.

Pre-post CMR intervention differences: (A) A stronger parietal operculum network connectivity is seen in adults with SCI (n = 20) after 6 weeks of CMR with stronger connections between the right parietal operculum (ROI) and the left parietal operculum (parts OP1/OP4), left angular gyrus (AG), and left supramarginal gyrus (SMG). (B) Increased task-related activation is seen for right toe sensation in adults with SCI-related neuropathic pain (n = 21) after the CMR intervention, in line with increased sensation, evaluated with ISNCSCI³¹ testing. The activation was stronger in bilateral sensorimotor areas and areas important for body awareness and visuospatial body maps. Legend: AG: angular gyrus; INS: insula; FO: frontal operculum; postCG: postcentral gyrus; SMG: supramarginal gyrus; SPL: superior parietal lobe.

At baseline, most participants with SCI (64%) could not feel a small towel gently brushing the big toe during the toe sensation fMRI task. In those that had some sensation, toe sensation was altered in three participants (13%), i.e., they either felt some tingling in the feet or located the sensation elsewhere in the foot (e.g., heel). Three participants (13%) correctly felt one big toe, and two participants (9%) correctly felt the sensation in both big toes at baseline.

After CMR, 50% of participants who had absent sensation at baseline regained some sensation: seven participants correctly felt sensation in both toes (32%) vs. two participants (9%) at baseline. Of those who could not correctly identify light touch sensation in both toes after CMR, four participants had altered sensation (18%); and four participants correctly felt sensation in one toe (18%). Seven participants had absent toe sensation (32%) vs. 14 participants at baseline. Post-CMR assessment by the ISNCSCI³¹ neurological exam also showed improved touch sensation in the lower limbs. Most participants reported that with improved sensation, they could now feel their feet in daily life, i.e., the feet positions on the footplates of the wheelchair (and not falling off anymore). They were now also aware of weight shifts over the soles of their feet during transfers, which increased their confidence to bear weight on their legs, thereby improving transfers’ flow and speed.

Discussion

To investigate preliminary efficacy of CMR in restoring MBR, reducing neuropathic pain, and improving sensorimotor function, we conducted a Phase I randomized delayed-treatment trial, in which 26 adults with SCI-related neuropathic pain and AIS A-D were randomized to either 6 weeks of immediate CMR (group A), with a 6-week and 1-year follow-up; or a 6-week observational period, followed by a 6-week CMR intervention and 1-year follow-up (Group B). We assessed NPRS³⁰ weekly to monitor neuropathic pain. MRI, ISNCSCI exam³¹ and questionnaires were assessed at baseline and after the first and second 6-week periods. The 1-year follow-up included ISNCSCI³¹ exam and behavioral questionnaires.

At baseline, our study participants reported severe neuropathic pain, with a highest neuropathic pain of 8 ± 1 points³⁰. Clinically, we reported preliminary CMR efficacy through significant neuropathic pain reduction, improved sensation, enhanced movements and function, and greater self-reported awareness of paralyzed body parts─all sustained at 1-year follow-up in the subset of participants assessed. Neuroimaging revealed stronger connectivity within the OP1/OP4 network and increased brain activation related to sensorimotor and MBR function during a right-toe sensation fMRI task after CMR. Further details of these main results are provided below.

Such high sustained pain reductions (i.e., almost 5-point NPRS reduction for highest; 4 points for average neuropathic pain; 35% of participants pain-free) were not seen in other non-pharmacological interventions:⁴³ A Cochrane study mentioned pain reduction of 0.2–0.9 points at > 6-week follow-up after neuromodulation; 0.5 points at 1–6 weeks follow-up after acupuncture, and max 2.5 points at 1–6 weeks follow-up after exercise⁴³. Our findings suggest clinically meaningful pain reduction that was sustained to 1-year follow-up in those participants who completed this assessment (n = 14). However, direct comparison to other interventions is difficult given differences in study design, particularly the lack of an active control group in the present Phase I trial, which prevents us from ruling out placebo, or attention.

One potential mechanism for the observed benefits is improved multisensory integration and MBR, as suggested by concurrent strengthening of brain connectivity and improved self-reported body awareness post-CMR compared to baseline. These results seem to confirm hypotheses put forth in a review by Vastano et al. (2022)². Formal mediation analyses would be needed to test whether brain changes account for clinical improvements after CMR.

Participants had spasm alleviation and significant lower limb muscle strength improvement, sustained at 1-year follow-up. Function (SCI-FI/AT)^36,37 was significantly improved for basic mobility, self-care, fine motor function, and ambulation in adults with paraplegia and self-care in adults with tetraplegia. Improvement in self-care is critical for adults with tetraplegia to regain more autonomy in daily life. Participants also reported standing for the first time with minimal assistance, better balance, and better transfers. When the soles of the feet and leg position in space can be felt, transfers happen more easily, and sensorimotor function can be maintained because participants may feel more comfortable moving.

The clinical/behavioral findings post-CMR were coupled with brain function improvements. Our resting-state fMRI results matched those found in our CMR study in stroke,⁶ confirming our hypothesis that focusing on restoring MBR has a modulating effect on this network¹⁴.

Secondly, after CMR, increased brain activation in relevant sensorimotor areas and parietal areas related to pain, and MBR were found during our toe sensation fMRI task. These brain imaging changes post-CMR compared to pre-CMR concurred with improved foot sensation in 50% of participants who had absent foot sensation at baseline, and with significantly improved touch sensation in the lower limbs, sustained for at least 1 year.

As established in ‘discomplete SCI’ research^27,28, finding the absence of clinically elicitable sensation and movement below the lesion during the ISNCSCI neurological exam³¹ (AIS A) does not necessarily indicate a complete severance of spinal cord nerves. Our study showed that even participants without baseline brain activation during sensory stimulation of the foot—all adults with AIS A—experienced sensorimotor improvements. Notably, we observed a greater rate of improvement in sensorimotor function (50%) compared to the 9% (2 out of 22 patients) reported by Lee et al. (2015)²⁹ in individuals without brain activation during a passive foot movement fMRI task. These results suggest that with targeted interventions such as CMR, it may be possible to improve the connection between the body and brain through the spinal cord.

To our knowledge, only one case series study reported changes in sensorimotor brain function in the foot and toe area during active toe flexion and ankle plantar flexion MRI tasks in 4 adults with SCI (AIS C or D) after 12 weeks of Lokomat-driven body weight-supported gait training with orthoses, 3x/week⁴⁴. In comparison, 53% of our participants were AIS A, 62% had no ankle movements, and 69% had no toe movements at baseline. After CMR, these percentages were reduced to 19% and 38%, respectively.

The current findings complement our CMR research in adults with stroke (i.e., through improving MBR and its relevant brain areas and networks)⁶, showing that CMR can be beneficial in multiple neurological disorders. One potential explanation for our findings is that improved body awareness and spatial mapping allow the brain to more accurately interpret sensory signals, reducing the likelihood of maladaptive pain responses. However, the specific mechanisms linking MBR improvements to pain reduction require further investigation.

In sum, our findings suggest that when the brain more accurately interprets sensory signals and restores MBR after CMR, incoming sensory signals may be less likely interpreted as pain and the brain may less likely induce muscle spasms. Accurate spatial body maps also improve the brain’s guidance of movements, thereby facilitating further sensorimotor recovery.

Our study has low racial and ethnic diversity, and no veterans were included in our sample. Given that 26% of adults with SCI receive care in a Veterans Health Administration setting, further CMR research in this population is warranted. Secondly, there is no MBR scale available for adults with SCI. We used the revised body awareness questionnaire⁴⁵, but this scale evaluates emotional reactions and internal body sensations, such as being tense or being aware of breathing. Thus, a new MBR scale that aims at assessing the specific impaired MBR seen in adults with SCI, is warranted, and is in progress^46,47. Thirdly, only four participants in our study had a cervical SCI. Thus, studies with a larger number of adults with tetraplegia are needed. Lastly, our participants span a broad range of time since injury and exhibit substantial variability in injury characteristics. While this broad inclusion was our intent to enhance generalizability, it also may introduce potential sources of variability in treatment responsiveness, and the results should be interpreted in this light. A future, larger trial should be conducted to allow for subgroup analyses to further examine these possible sources of variability.

In conclusion, current therapies have limited success in neuropathic pain relief and functional recovery in adults with chronic SCI. Neuropathic pain-related brain mechanisms are unclear¹¹. New, effective interventions are needed. While some state that clinical recovery is slowed or stops at ± 2 years post-SCI⁴⁸, in this Phase I trial, CMR showed preliminary evidence of efficacy for neuropathic pain and sensorimotor improvements in adults with chronic SCI (1–56 years post-injury), with benefits maintained in those participants assessed at 1-year follow-up. Concurrently, we showed preliminary evidence of fMRI resting-state and task-based changes in terms of increased brain activation and connectivity after CMR. While these findings are promising, confirmation in larger, controlled trials with active control conditions (instead of an observation period group), and with more diverse populations, is essential before CMR can be recommended for clinical implementation in SCI populations.

Yet, there are promising translational considerations for this future clinical adoption. The dosage given in this study is similar to what patients would receive in an outpatient setting. Furthermore, while CMR is already established as a neurorehabilitation approach in Italy and other countries, 6-day basic courses are currently being organized at the University of Minnesota by the first author to facilitate US adoption. Physical and occupational therapists have reported anecdotally that CMR can seamlessly be integrated into conventional neurorehabilitation—respecting clinical milestones and expectations─while observing superior results. Therefore, we anticipate that CMR can be easily implemented in the US clinical care system once efficacy is definitively demonstrated.

Methods

Study design

MHealth/Fairview healthcare recruitment staff sent letters to patients with SCI. If they were interested, patients then contacted the investigators. We posted fliers in outpatient locations within the Minnesota Regional Spinal Cord Injury Model System. Community-dwelling adults with SCI were enrolled between 01/09/2021-08/31/2022. We employed a delayed-treatment design to ensure all participants received CMR while maintaining some experimental control through the observation period. By collecting data during the waiting (no-intervention) period and comparing it to the post-intervention data, researchers can directly assess the effect of the treatment on the same individual, reducing the influence of confounding variables like age, severity, or home environment. It is also a way to investigate new therapies without denying treatment to any participant. Delayed interventions are used regularly in rehabilitation studies^49–52. It has the benefit of randomization compared to a single-arm study or purely observational designs. However, we acknowledge this design limits our ability to control for placebo and non-specific treatment effects. The study was conceived as a Phase I trial primarily to assess feasibility and safety, with efficacy outcomes considered exploratory.

Clinical information was posted on January 8th, 2021, on ClinicalTrials.gov (NCT04706208), and the clinical trial enrollment began on January 9th, 2021. We completed the 1-year follow-up on December 1th, 2023. The study protocol is published²⁵.

The study followed the Declaration of Helsinki’s principles (2013). The University of Minnesota (UMN)’s Institutional Review Board approved the study (IRB#STUDY00008476). CMR sessions and the ISNCSCI³¹ exam occurred in Van de Winckel’s Brain Body Mind Lab at the University of Minnesota. MRI scanning was done at the UMN’s Center for Magnetic Resonance Research (CMRR). Questionnaires were completed over secure UMN Zoom by trained study staff.

Participants

We recruited 18–75 years old adults, with SCI > 3 months, medically stable with complete/incomplete traumatic/non-traumatic SCI, and with the highest level of SCI-related below-level diagnosed neuropathic pain in the prior week of ≥ 4/10 on the NPRS³⁰. The upper age limit minimized age-related brain atrophy for MRI data.

We excluded adults with severe cognitive, vision, or hearing impairments; major medical complications; pregnant women; or receiving concurrent rehabilitation that would influence outcomes. Although adults with complete tetraplegia can potentially benefit from CMR, we only included adults who could self-transfer with assistance to facilitate transfers on the scanner bed, and to ensure the ability to push the emergency MRI call button with the hand or wrist if needed. Participants provided eConsent through the secure UMN REDCap platform.

After consenting, study staff collected demographic, general health, and medical information. Participants self-reported their sex at birth (female-male) and gender identification (female, male, or ‘other’ with text option to specify their identification). Cognitive ability was screened with the Mini-Mental State Examination-Short version^53,54. They completed the Kinesthetic and Visual Imagery Questionnaire⁵⁵ because kinesthetic imagery is used in CMR. Participants completed the CMRR prescreening questionnaire, which was signed off by the CMRR’s Operations and Safety Director.

Randomization

Using computer-generated randomization, adults with SCI were allocated to Group A: 6 weeks CMR, 1:1, in-person, 3x/week, 45 min/session, followed by 6 weeks standard-of-care (no therapy) at home; or Group B: 6-week standard-of-care, followed by 6 weeks CMR because we anticipated that participants recruited and not offered an intervention would likely drop out. CMR dosage was similar to what patients receive in an outpatient clinic. The brain imaging analyst, biostatistician, and raters were blinded to group allocation. CMR sessions were logged and video-recorded for quality assurance of intervention content and delivery. CMR was minimal risk because of the low-intensity movements and kinesthetic imagery.

Procedures

At three time points (baseline, at 6 weeks and at 12 weeks), adults with SCI had an MRI (structural/resting-state/fMRI task), in-person ISNCSCI³¹ exam to assess sensorimotor function, and completed questionnaires on neuropathic pain³⁰, function³⁶, mood³⁹ and SCI-related symptoms^33–37. They continued routine SCI healthcare management, including prescribed/as-needed neuropathic pain medication. NPRS³⁰ was assessed weekly to monitor neuropathic pain. The 1-year follow-up included ISNCSCI³¹ exam and questionnaires.

Intervention

An experienced CMR-certified therapist provided the CMR intervention. More information about the CMR treatment can be found in the protocol paper²⁵ and is presented briefly here. The CMR therapist conducted a maximum of two CMR exercises per session that involved body parts that the participant described as painful. Exercises were done usually in sitting or lying position. If a person could stand, then the CMR therapist would also include exercises in standing position, e.g., to train the awareness of body weight shifts when transferring body weight from one foot to the other.

All CMR sessions were designed to improve MBR and reduce neuropathic pain, with a standardized reasoning of which type of MBR was impaired in a particular person (e.g. difficulty relating the foot and the hip; difficulty feeling the dimensions of the paralyzed legs). The treated body parts and choice of exercises were based on specific injury characteristics, specific altered MBR for that person, and functional activities that participants had a hard time executing because of the neuropathic pain. The CMR therapist always first inquired where the neuropathic pain was located; when the neuropathic pain would occur; what strategies the participant already used to decrease the neuropathic pain; and how the participant perceived the dimensions of his/her/their body.

The types of CMR exercises included identifying the leg position in relation to the pelvis and the upper body; identifying the dimension of the legs and pelvis; the sensation of the pelvis as a central body reference; the relationship between the left and right side of the body, or between the pelvis and the feet. Through the exercise, the participant learned to perceive those body parts better and to relate those body parts with the rest of the body and the therapist verified whether this increased body awareness changed their neuropathic pain perception.

Here is an example of an exercise: A participant who had complete sensory loss of the lower limb at baseline, had his eyes closed and was sitting on the treatment table. The therapist placed five pieces of carpet stacked under the left foot and 3 pieces of carpet stacked under the right foot and asked the participant to feel which foot was higher (Fig. 5). The CMR therapist provided strategies, additional questions, suggestions, or guided the participant through kinesthetic imagery to help the participant answer the question correctly, and through this exercise, restore their MBR. An example of a kinesthetic imagery guidance was as follows: the therapist asked the participant to remember a time before the SCI when he was riding a bike, and to remember the difference in the feeling in the foot when he was pushing with one foot at a time to pedal and compare that feeling to when the foot was resting on the ground, for example at a red light. Once the participant remembered this feeling (i.e., kinesthetic imagery), the therapist asked him if the current feeling (in the exercise) was more similar to having both feet on the ground, such as, at a standstill when talking to a friend, or having one foot ready to pedal. The participant responded, one foot ready to pedal. The therapists asked which one, and the participant responded correctly. Recalling a well-specified action from the past (that has a specific context, emotions, and a memory of a body movement) can help the participant access a more complete MBR, because the memory dates from before the SCI when the MBR were intact. Sometimes it can take a while before these stored memories are accessible again. But once remembered, this remembered feeling helped the brain access the stored intact MBR and integrate them into the current situation. Providing these strategies seems to be the key to restoring MBR. This approach is unique to CMR and is not found in other neurorehabilitation methods.

Fig. 5.

Example of a CMR exercise.

Certification in CMR is a 3-year post-educational specialization that can be completed in the Study Center for Cognitive Multisensory Rehabilitation (Centro Studi di Riabilitazione Neurocognitiva - Villa Miari) in Santorso, Italy. However, since the handbooks have not yet been translated to English, the first author (Van de Winckel) has been organizing CMR classes since 2021 for physical and occupational therapists in the community to bring the CMR curriculum to the United States. This course is certified and created in collaboration with the Center in Italy.

Clinical assessments

The primary clinical outcome (NPRS)³⁰ was the highest, average (i.e., pain intensity perceived most of the time), and lowest neuropathic pain intensity levels participants experienced in the prior week. The second primary clinical outcome was video-recorded sensorimotor testing with the ISNCSCI³¹ exam, performed by experienced physical therapists.

Secondary outcomes: NINDS-CDE International SCI Pain Basic Data Set Version 2.0³², which assesses average nociceptive and neuropathic pain, locations and types of pain (i.e., differentiating neuropathic pain from other types of pain), and interference of neuropathic pain with mood, activity, and sleep. We assessed spasm frequency and intensity (Penn Spasm Frequency Scale)³³, SCI-secondary conditions³⁴, SCI-FI/AT^36,37, and the Patient Specific Functional Scale³⁵. During the weekly phone check-ins, information on neuropathic pain, neuropathic pain medication dosage, healthcare use, adverse events if any, and perceived intervention effects was recorded into REDCap.

MRI assessments

Participants were scanned for 2 h on a Siemens 3-T Prisma scanner at CMRR. The protocol paper explains the structural and functional MRI acquisition in detail²⁵. In short, structural MRI acquisition included T1-weighted magnetization-prepared rapid acquisition with gradient echo (MPRAGE) [repetition time (TR) = 2.5s; echo time (TE) = 4.5ms; 0.8 mm isotropic voxels], and T2-weighted sampling perfection with application-optimized contrasts using different flip angle evolution (SPACE) [TR = 3.2s; TE=565ms; 0.8 mm isotropic voxels]. Resting-state and task fMRI scans were obtained with T2*-weighted multiband echo planar acquisition tipped 30 degrees relative to the anterior commissure–posterior commissure (AC-PC) plane according to auto-align software. This acquisition protocol was designed to measure whole-head blood-oxygen-level-dependent (BOLD)–contrast with optimal temporal and spatial resolution and to reduce signal dropout [TR = 0.8s; TE = 37ms; flip angle = 55degrees; 72 slices; multiband factor 8; 2 mm isotropic voxels].

For the resting-state fMRI imagery (12 min 10 s), we selected OP1/OP4 and insula as region-of-interest (ROI) based on their importance in sensorimotor function, pain, and body awareness^6,14,15. These brain areas are an integral part of the sensorimotor and visuospatial systems, as explained in the first paragraph of the introduction. During the resting-state fMRI scan, participants maintained eye fixation with a restful mind.

The fMRI sensation task (18 min 32 s) was chosen to generate activation of OP1/OP4 and insula and related brain areas. An investigator gently brushed the pads of the big toes and thumbs with a small towel, 6 times in each location. Participants were not made aware of where/when brushing would occur. They were asked afterward if/where they sensed any sensation on their hands or feet. This task tests residual sensory information reaching the somatosensory cortex^27,28. Strengthening of such responses post-CMR would be evidence of gains in signal transmission past the lesion.

Statistical analysis

Sample size

The sample size of n = 25 adults with SCI has > 80% power to detect a pre-post change of Cohen’s d = 0.6 at α = 0.05. This was the first CMR study in adults with SCI-related neuropathic pain; thus, direct power calculations were not possible. The present data will determine the effect size for larger clinical trials. Nevertheless, this sample size has sufficient power (~ 84%) to detect a minimally clinically significant difference of 1.8 points on the NPRS³⁰ for the pre-post changes between the two groups at a significance level of 0.05 given an estimate of 2.1 SD based on Estores et al. (2016)⁵⁶.

MRI

Neuroimaging data were preprocessed through the Human Connectome Project preprocessing pipeline. FMRI data were processed using the conn functional connectivity toolbox with established standardized controls for multiple comparisons (SPM family-wise error correction methods)²⁰. Data underwent realignment, scrubbing, artifact detection, and CompCor denoising. Primary analyses focused on voxel-wise whole-brain seed-based functional connectivity analysis of resting-state data using Pearson correlation coefficients, with OP1/OP4 and insula as the seed ROI. Task-based fMRI underwent the same preprocessing pipeline and voxel-wise brain activation was estimated for each subject using a general linear model.

Within-group pre-post changes in fMRI task-based activation and resting-state connectivity (with Fisher’s Z transformation) were tested with paired t-tests, and between-group differences of pre-post changes with 2 sample t-tests. We tested the effect of CMR on brain imaging outcomes using mixed-effects models implemented in FSL. Cluster-based correction for multiple comparisons was performed using a cluster-forming threshold of Z > 2.3 and a cluster-level significance threshold of p < 0.05 to control the family-wise Type I error rate.

Clinical assessments

Quantitative variables were summarized using descriptive statistics at each time point. Analyses were conducted with R. Primary analyses of the data collected during the 12 weeks of pre or post CMR intervention were based on intent-to-treat. However, the 1-year follow-up analysis was conducted per protocol, including only the 14 participants who consented to the 1-year follow-up extension. Participants who did not consent were not included in the 1-year follow-up analyses.

A three-way repeated-measures ANOVA analysis was conducted on the pain score data from all the time points using linear mixed-effects models to test if the CMR intervention has different effect on the rate of pain level changes over time between the two groups. The linear mixed effects model takes the form.

where is the observed pain score for subject i at time point t, is the overall mean, are the coefficients of the fixed main effects, two- and three-way interactions, is the subject-level random effect, and is the within-subject random error term. Significance of , as indicated by the F test for the three-way interaction in the repeated-measures ANOVA, would suggest that CMR intervention has a different impact on pain level changes between groups A and B.

Between-group comparison of changes in the first 6 weeks (Group B [observation period] vs. Group A [early CMR]), pre-CMR measures, and changes pre-post CMR were tested using 2-sample t-tests or Wilcoxon rank sum test; pre-post CMR changes (early+delayed CMR; Groups A + B combined) and pre-post observation group (Group B in first 6 weeks) were tested using paired t-tests. Benjamini-Hochberg’s False Discovery Rate procedure was used to adjust the p-values across clinical measures to control the overall type-1 error rates. A UMN Clinical and Translational Science Institute (CTSI) data integrity monitor and quality assurance reviewer monitored the study every 6 months.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

CMR

Cognitive multisensory rehabilitation

CMRR

Center for Magnetic Resonance Research (CMRR) of the University of Minnesota

AIS or ISNCSCI

ASIA Impairment Scale or International Standards for Neurological Classification of Spinal Cord Injury

fMRI

functional Magnetic Resonance Imaging

MBR

Mental body representations

NPRS

Numeric pain rating scale

OP1/OP4

Parts 1 and 4 of the parietal operculum

SCI

Spinal cord injury

SCI-FI/AT

Spinal Cord Injury Functional Index/Assistive Technology Short Forms

UMN

University of Minnesota

Author contributions

Every author contributed to the study: Conceptualization (AVdW, LM); Funding acquisition (AVdW, LM); Data curation (AVdW, SC, WD); Formal analysis and data interpretation (AVdW, LZ, TH, KL, CL, BM, LM); Cognitive Multisensory Rehabilitation intervention (SB); Recruitment (AVdW, LM, RW); Data collection (AVdW, SC, WD, BM); Supervision (AVdW, LM); Validation (AVdW, SC, WD, LZ); Visualization (AVdW, LZ); Writing – original draft (AVdW); Writing – review & editing (AVdW, SC, WD, SB, LZ, TH, CL, KL, BM, RW, RB, LM). We confirm that more than one author has directly accessed and verified the underlying data reported in the manuscript (AVdW, SC, WD, LZ).

Funding

The AIRP2-IND-30: Academic Investment Research Program (AIRP) University of Minnesota School of Medicine funded this study (IRB no. STUDY00008476; ClinicalTrials.gov Identifier: NCT04706208), first registration 08/01/2021. Additional support was provided by the National Center for Advancing Translational Sciences of the National Institutes of Health Award Number UL1TR002494, and 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.

Data availability

The datasets used and/or analysed during the current study available on Dryad: DOI: 10.5061/dryad.7h44j1007.

Competing interests

The authors declare no competing interests.