Nerve Transfer Surgery to Restore Upper-limb Function After Cervical Spinal Cord Injury

All individuals with spinal cord injury (SCI) referred to an interdisciplinary clinic for potential nerve transfer undergo a standardized clinical examination, according to the International Standards for Neurological Classification of Spinal Cord Injury (ISNCSCI), to classify the neurological level of injury. Additionally, individuals receive detailed sensorimotor and electrodiagnostic evaluations, and their upper-limb goals are carefully explored. They are then monitored for spontaneous upper limb motor recovery over the next 2.5-3 months, at which point they are re-evaluated with electrodiagnostic testing. Individuals who are deemed candidates for nerve transfer surgery will be screened and invited to participate in the study. Participation in the study will not, in any way, influence clinical care.

After receiving written consent, outcome measurement will take place pre-operatively (approximately 5-6 after injury) and at 3-month intervals post-operatively through 24 months after surgery. Primary and secondary outcomes are measured for each limb undergoing nerve transfer, so participants can be included in the study even if they receive double nerve transfer on only one limb. Outcome measurement will be performed at GF Strong Rehabilitation Hospital (Primary site; Vancouver), the International Collaboration on Repair Discoveries (ICORD; Vancouver), Sunnybrook Hospital (Toronto), Kingston Health Science Centre (Kingston), and Dalhousie University (Halifax). All study data will be stored in a REDCap database, which will be shared between all participating institutions. We will carefully document any adverse events during the peri-operative and post-operative rehabilitation phases. Adverse events will be classified according to the CTCAE Grading System.

Study outcomes in the surgical (experimental) group will be compared with those of a historical control group derived from a multicenter cohort study (PMID: 35652348). Participants in the surgical cohort will be matched 1:1 with historical controls based on age, sex, and injury characteristics. Given the invasiveness of the intervention, randomization and blinding are not possible. Therefore, our team used a non-blinded, non-randomized design - features that reduce the level of evidence provided by the data. Every attempt has been made to minimize bias, including using an objectively measured primary outcome and adding a control group to facilitate comparison to those without nerve transfer. There are biases and potential confounding effects of using historical controls, including differences in measurement techniques and sociocultural and health system factors.

A linear mixed-effects model will be used to determine whether nerve transfer improves functional hand outcomes from baseline to 24 months post-surgery. Analyses will be conducted at the limb level, with fixed effects for time (baseline, 24 months) and a random intercept for participant to account for any correlation between limbs contributed by the same participant. To quantify the within-participant improvement in our primary outcome, the adjusted mean change in the surgical group from pre-operative to 24 months post-surgery will be estimated. We will also compare the mean change in GRASSP in the surgical group to the change in GRASSP of the historical control cohort between 6 and 12 months post-injury. We believe this comparison is justified as individuals with motor complete injuries reach neurological stability around 1 year post-injury. 95% confidence intervals and two-sided p-values will be reported for both analyses (α = 0.05).

We will also perform an exploratory analysis to identify baseline factors associated with upper-limb function at 24 months after nerve transfer surgery. This analysis will be restricted to the surgical cohort and will use the 24-month GRASSP score as the outcome. Prespecified predictors include participant age, neurological level of injury, and baseline donor nerve health. Donor nerve health will be summarized as the number of donor nerves demonstrating PSA on baseline needle EMG (range 0-2 per limb). These analyses are intended to identify candidate predictors to inform future multivariable clinical prediction model development.

The trial is powered based on the within-participant change in total GRASSP from the pre-operative baseline to 24 months post-surgery (two-sided α = 0.05). Because no paired data are available for this interval on the total GRASSP score, we based the calculation on external data from the GRASSP version II sensation subtest, which reported a paired effect size d = 0.69 on motor-complete SCI patients from baseline (~10 days post-injury) to 12 months post-injury (PMID: 31573454). Under these assumptions, the required number of participants with 24-month post-injury data to achieve 95% power is 29 (calculated using the pwr R package); assuming a 90% retention rate, the recruitment target increases to 33 participants. To account for the possibility that total GRASSP may be more variable than the sensation subtest, we inflated the target by 20%, resulting in a planned sample of 40 participants. Power was calculated at the participant level, allowing for the inclusion of up to 80 surgical limbs.