The Muscle Cuff Regenerative Peripheral Nerve Interface for the Amplification of Intact Peripheral Nerve Signals

This manuscript provides an innovative method for developing a biologic peripheral nerve interface termed the Muscle Cuff Regenerative Peripheral Nerve Interface (MC-RPNI). This surgical construct can amplify its associated peripheral nerve's motor efferent signals to facilitate accurate detection of motor intent and the potential control of exoskeleton devices.

Robotic exoskeletons have gained recent acclaim within the field of rehabilitative medicine as a promising modality for functional restoration for those individuals with extremity weakness. However, their use remains largely confined to research institutions, frequently operating as a means of static extremity support as motor detection methods remain unreliable. Peripheral nerve interfaces have arisen as a potential solution to this shortcoming; however, due to their inherently small amplitudes, these signals can be difficult to differentiate from background noise, lowering their overall motor detection accuracy. As current interfaces rely on abiotic materials, inherent material breakdown can occur alongside foreign body tissue reaction over time, further impacting their accuracy. The Muscle Cuff Regenerative Peripheral Nerve Interface (MC-RPNI) was designed to overcome these noted complications. Consisting of a segment of free muscle graft secured circumferentially to an intact peripheral nerve, the construct regenerates and becomes reinnervated by the contained nerve over time. In rats, this construct has demonstrated the ability to amplify a peripheral nerve's motor efferent action potentials up to 100 times the normal value through the generation of compound muscle action potentials (CMAPs). This signal amplification facilitates high accuracy detection of motor intent, potentially enabling reliable utilization of exoskeleton devices.

In the United States alone, approximately 130 million people are affected by neuromuscular and musculoskeletal disorders, resulting in over $800 billion in annual economic impact^1,2. This group of disorders is typically secondary to pathology within the nervous systems, at the neuromuscular junction, or within the muscle itself³. Despite the variety of pathologic origins, the majority share some degree of extremity weakness^1,3. Unfortunately, this weakness is often permanent given the limitations in neural and muscle tissue regeneration, especially in the setting of severe trauma^4,5,6.

Extremity weakness treatment algorithms have classically focused on rehabilitative and supportive measures, often relying on harnessing the capabilities of the remaining intact limbs (canes, wheelchairs, etc.)⁷. This strategy falls short, however, for those whose weakness is not limited to a single extremity. With recent innovations in robotic technologies, advanced exoskeleton devices have been developed that restore extremity functionality to those living with extremity weakness^{8,9,10,11,12,13}. These robotic exoskeletons are often powered, wearable devices that can assist with initiation and termination of movement or maintenance of limb position, providing a varying amount of force that can be individually tailored for the user^{8,9,10,11,12,13}. These devices are classified as either passive or active depending on how they provide motor assistance to the user: active devices contain electrical actuators that augment power to the user, whereas passive devices store energy from the user's motions in order to release it back to the user when necessary¹⁴. As active devices have the ability to increase a user's power capabilities, these devices are utilized far more frequently in the setting of extremity weakness^[14].

In order to determine motor intent in this population, modern exoskeletons commonly rely on pattern recognition algorithms generated from either electromyography (EMG) of distal limb muscles^8,15,16,17 or surface electroencephalography (sEEG) of the brain^18,19,20. Despite the promise of these detection modalities, both options have significant limitations that preclude widespread utilization of these devices. As sEEG detects microvolt-level signals transcranially^18,19,20, criticisms frequently focus on the inability to differentiate these signals from background noise²¹. When background noise is similar to the desired recording signal, this produces low signal-to-noise ratios (SNRs), resulting in inaccurate motor detection and classification^22,23. Accurate signal detection additionally relies on stable, low-impedance scalp contact²¹, which can be significantly affected by the presence of coarse/thick hair, user activity, and even sweating^22,24. In contrast, EMG signals are several magnitudes larger in amplitude, facilitating greater motor signal detection accuracy^15,18,25. This comes at a cost, however, as nearby muscles can contaminate the signal, decreasing the degrees of freedom able to be controlled by the device^16,17,25 and an inability to detect deep muscle motion^25,26,27,28. Most importantly, EMG cannot be used as a control method when there is significant muscle compromise and complete absence of tissue²⁹.

In order to advance the development of robotic exoskeletons, consistent and accurate detection of motor intent of the intended user is required. Interfaces that utilize the peripheral nervous system have arisen as a promising interface technique, given their relatively simple access and functional selectivity. Current peripheral nerve interfacing methods can be invasive or non-invasive and typically fall within one of three categories: extraneural electrodes^30,31,32,33, intrafascicular electrodes^34,35,36 and penetrating electrodes^37,38,39,40. As peripheral nerve signals are generally on the level of microvolts, it can be difficult to differentiate these signals from similar amplitude background noise^41,42, which reduces the overall motor detection accuracy capabilities of the interface. These low signal-to-noise (SNR) ratios often worsen over time secondary to worsening electrode impedance⁴³ produced from either degradation of the device^39,43, or local foreign body reaction producing scar tissue around the device and/or local axonal degeneration^37,44. Although these shortcomings can generally be resolved with reoperation and implantation of a new peripheral nerve interface, this is not a viable long-term solution as foreign-body-associated reactions would continue to occur.

To avoid these local tissue reactions generated from peripheral nerves' interaction with abiotic interfaces, an interface incorporating a biologic component is necessary. To address this shortcoming, the Regenerative Peripheral Nerve Interface (RPNI) was developed to integrate transected peripheral nerves in the residual limbs of those with amputations with prosthetic devices^45,46,47,48. Fabrication of the RPNI involves surgical implantation of a transected peripheral nerve into a segment of autologous free muscle graft, with revascularization, regeneration, and reinnervation occurring over time. Through the generation of milli-volt level compound muscle action potentials (CMAPs), the RPNI is able to amplify its contained nerve's micro-volt level signal by several magnitudes, facilitating accurate detection of motor intent^45,48,49. There has been considerable development of the RPNI over the past decade, with notable success in amplifying and transmitting efferent motor nerve signals in both animal^50,51 and human⁴⁷ trials, facilitating high accuracy prosthetic device control with multiple degrees of freedom.

Individuals with extremity weakness but intact peripheral nerves would similarly benefit from high accuracy detection of motor intent through peripheral nerve interfaces in order to control exoskeleton devices. As the RPNI was developed for integration with transected peripheral nerves, such as in persons with amputations, surgical modifications were necessary. Building from experience with the RPNI, the Muscle Cuff Regenerative Peripheral Nerve Interface (MC-RPNI) was developed. Consisting of a similar segment of free muscle graft as in the RPNI, it is instead secured circumferentially to an intact peripheral nerve (Figure 1). Over time, it regenerates and becomes reinnervated through collateral axonal sprouting, amplifying and translating these efferent motor nerve signals to EMG signals that are several orders of magnitude larger⁵². As the MC-RPNI is biologic in origin, it avoids the inevitable foreign body reaction that occurs with peripheral nerve interfaces currently in use⁵². Furthermore, the MC-RPNI confers the ability to control multiple degrees of freedom simultaneously as they can be placed on distally dissected nerves to individual muscles without significant cross-talk, as has been previously demonstrated in RPNIs⁴⁹. Finally, the MC-RPNI can operate independent of distal muscle function as it is placed on the proximal nerve. Given its advantages over current peripheral nerve interfaces, the MC-RPNI holds substantial promise for providing a safe, accurate, and reliable method of exoskeleton control.

All animal procedures and experiments were carried out with the approval of the University of Michigan's Institutional Care and Use of Animals Committee (IACUC). Male and female Fischer F344 and Lewis rats (~200-300 g) at 3-6 months of age are most frequently utilized in experiments, but any strain can theoretically be utilized. If utilizing donor rats instead of autologous muscle grafts, donor rats must be isogenic to the experimental strain. Rats are allowed free access to food and water both pre- and post-operatively. Following terminal endpoint evaluations, euthanasia is performed under deep anesthesia with intra-cardiac potassium chloride injection followed by a secondary method of bilateral pneumothorax.

1. Experimental preparation of the rat

1. Anesthetize the experimental rat utilizing a solution of 5% isoflurane in oxygen at 0.8-1.0 L/min in an induction chamber. Once adequate anesthesia is achieved and confirmed with the absence of corneal reflex, place the rat on a rebreather nose cone with isoflurane lowered to 1.75%-2.25% for maintenance of anesthesia.
2. Inject a solution of 0.02-0.03 mL Carprofen (50 mg/mL) in 0.2 mL of sterile saline with 27 G needle in the subcutaneous plane between the shoulder blades for peri- and post-operative analgesia.
3. Apply sterile eye ointment to both eyes to prevent corneal ulcers while anesthetized.
4. Using an electric razor, shave the lateral portion of the bilateral lower limbs, extending from the hip joint, over the thigh, and to the dorsal surface of the paw.
5. Sterilize the surgical site by first wiping with an alcohol prep pad, followed by povidone-iodine solution application, ending with a final cleanse with a new alcohol prep pad to remove the residual povidone-iodine solution. Repeat this alternating cleansing process three times to maintain sterility.
   NOTE: This can be a dermatological irritant; ensure the majority of the solution is removed.

2. Preparation of the muscle graft

1. Place the rat on a heating pad underneath a surgical microscope with an intraoral body temperature probe of choice for body temperature monitoring. Maintain isoflurane at 1.75%-2.25% and oxygen at 0.8-1.0 L/min.
2. Make a longitudinal incision along the anterior aspect of the desired donor hindlimb extending from just above the ankle to just below the knee with a #15 scalpel.
3. Dissect through the underlying subcutaneous tissue using sharp iris scissors to expose the underlying musculature and distal tendons just proximal to the ankle joint. Tibialis anterior (TA) is the largest and most anterior of the muscles; the extensor digitorum longus (EDL) muscle can be found just deep and posterior to this muscle. Isolate the EDL muscle and its distal tendon from the surrounding musculature.
4. Ensure isolation of the correct tendon by inserting both tines of a forceps or iris scissor underneath the distal tendon just proximal to the ankle joint. Exert upward pressure on the tendon by opening either the forceps or iris scissors. This motion should produce a simultaneous extension of all of the toes simultaneously. If isolated ankle dorsiflexion, ankle eversion, or single toe dorsiflexion occurs, the wrong tendon has been isolated.
5. Perform a distal tenotomy of the EDL muscle at the level of the ankle with sharp iris scissors and dissect the muscle free from surrounding tissues working proximally towards its tendinous origin.
6. Once the proximal tendon is visualized, perform a proximal tenotomy utilizing sharp iris scissors to free the graft.
7. Trim both tendinous ends of the muscle graft and cut to the desired length with sharp iris scissors.
   NOTE: Grafts measuring 8-13 mm have been utilized with success; however, the most common length utilized is 10 mm.
8. On one side of the muscle graft, make a longitudinal incision along the entire trimmed length to facilitate placement of the nerve within the muscle graft and provide contact of the nerve with endomysium.
9. Place the prepared muscle graft in a saline-moistened gauze to prevent tissue desiccation.
10. Close the skin overlying the donor site with 4-0 chromic suture in a running fashion.

3. Common peroneal nerve isolation and preparation

1. Mark the surgical incision, which will extend from a line ~5 mm from the sciatic notch, extending to just inferior to the knee joint. Ensure that this marking is inferior to, and angled away from, the femur that can be palpated below.
2. Incise through the skin and subcutaneous tissues along the marked incision line with a #15 blade. Carefully incise through the underlying biceps femoris fascia, taking care not to extend through the entire depth of the muscle as the sciatic nerve lies just below.
3. Utilizing blunt-tipped small scissors or a hemostat, dissect carefully through the biceps femoris muscle.
   NOTE: The sciatic nerve travels in this space underlying the biceps, oriented in approximately the same direction as the incision marked on the skin. There are three notable sciatic nerve branches: sural (most posterior and smallest of the nerves), tibial (typically most anterior, but this nerve always dives deep to the knee joint), and common peroneal (typically located between tibial and sural, always travels above the knee joint).
4. Identify the common peroneal (CP) nerve and carefully isolate it from the surrounding nerves using a pair of micro-forceps and micro-scissors. Remove any surrounding connective tissue from the middle 2 cm of the nerve. Take care not to crush the CP nerve with forceps in this process, as crush injury can alter endpoint results.
5. Over the central-most portion of the freed CP nerve, perform an epineurial window by removing 25% of the epineurium along the length of the nerve that matches the desired length of the muscle graft.
6. To perform this hold the proximal epineurium with micro-forceps, cut into the epineurium immediately underlying with micro-dissection scissors, and remove ~25% of the epineurium traveling distally along the nerve. Take care to remove this segment in one piece, as multiple attempts can cause irregular epineurial removal, increasing the risk of nerve injury. 
   NOTE: The nerve tissue underlying the epineurium will have a goo-like texture; noting this quality of nerve ensures the correct tissue plane has been removed.

4. MC-RPNI construct fabrication

1. Remove the muscle graft from the saline-moistened gauze and place it under the central portion of the CP nerve where the epineurial window was created. Rotate the nerve 180° so the epineurial window section contacts intact muscle and does not underlie the eventual suture line.
2. Using an 8-0 nylon suture, suture the epineurium of the CP nerve both proximally and distally to the muscle graft within the groove created in step 2.8 using simple interrupted sutures to secure epineurium to endomysium.
   NOTE: Place these stitches, ensuring the muscle is at normal resting length. Stretching or compressing the muscle too much can impact regeneration and signaling capabilities later on.
3. Circumferentially wrap the edges of the muscle graft surrounding the now-secured nerve and suture in place utilizing simple interrupted 8-0 nylon stitches (~4-6 depending on length).
4. Once hemostasis is achieved, close the biceps femoris fascia over the construct with 5-0 chromic suture in running fashion.
5. Close the overlying skin in running fashion with a 4-0 chromic suture.
6. Cleanse the surgical area with an alcohol prep pad and apply antibiotic ointment.
7. Terminate inhalational anesthetic and place the rat in a clean cage isolated from cage mates and allow to recover with food and water ad lib.
8. Once the rat has suitably recovered, place it back with cage mates in a clean cage.
   NOTE: These constructs require maturation of three months at the minimum to produce adequate nerve signal amplification.

MC-RPNI surgical fabrication is considered a peri-operative failure if rats do not survive emergence from surgical anesthesia or develop an infection within a week of the operation. Previous research has indicated a 3 month maturation period will result in reliable signal amplification from this constructs^42,45,48,49. At that time or thereafter, surgical exposure of the constructs and evaluation can occur. If MC-RPNI fabrication was successful, revascularized muscle should be readily visible at the original MC-RPNI implantation site (Figure 2B). Successful MC-RPNIs will additionally contract following proximal nerve stimulation (Video 1). At times, significant scarring and atrophic muscle graft can be present (Figure 2C), indicating a failure of revascularization/regeneration typically secondary to too large of a graft, improper handling, or peri-operative tissue injury. These atrophic grafts commonly have some degree of contraction upon proximal nerve stimulation but produce lower signal amplification. Overall, it is considered a fabrication failure if, upon exposure, the MC-RPNI is found dislodged from the nerve or has no contraction upon proximal nerve stimulation.

Histological analysis of these constructs should demonstrate viable nerve and muscle tissue without any significant fibrosis or scarring (Figure 3). Immunohistochemistry can also be performed to confirm the presence of innervated neuromuscular junctions with neurofilament as a general nerve marker in combination with alpha-bungarotoxin as a marker for postsynaptic acetylcholine receptors (Figure 4). If the target implanted nerve fails to innervate the muscle component of the MC-RPNI, immunostaining would not demonstrate any collateral motor nerve sprouts traversing the construct, nor any innervated neuromuscular junctions.

Electrophysiologic testing can be performed on these constructs at any time following maturation, with results published demonstrating stable signals specifically in the MC-RPNI at 3 months⁵² and up to 3 years in RPNIs in human subjects⁴⁷. Electrophysiologic testing schematics can vary according to the area of interest and equipment available (Figure 5), but evaluations are most commonly performed with the provision of maximal stimulation to the proximal nerve with a hook electrode followed by a recording of compound muscle action potentials (CMAPs) generated at the MC-RPNI (Table 1). Recording electrodes can vary according to user preference, but epimysial patch/pad, epimysial bipolar probe, and penetrating bipolar electrodes have been used experimentally with success. The average compound nerve amplitude (CNAP) recorded at the CP nerve following more proximal nerve stimulation was 119.47 µV ± 14.87 µV. Average CMAP amplitude recorded at the MC-RPNI following similar proximal CP nerve stimulation was 3.28 mV ± 0.49 mV, resulting in amplification of the nerve signal from 11-87x, with an overall average amplification factor of 31.8 ± 7.70. These generated CMAP waveforms are similar in appearance to native muscle, further supporting that they have become reinnervated by their contained nerve (Figure 6B).

To ensure MC-RPNI fabrication does not cause negative functional impact, electrophysiologic and muscle force testing can be performed on distally-innervated muscle. The majority of testing has been performed on the ipsilateral EDL muscle as it is easily accessible for testing and is innervated by the common peroneal nerve (the contralateral EDL is harvested for MC-RPNI fabrication and thus not assessed). CMAPs generated by physiologic EDL muscle following proximal CP nerve stimulation typically range from 20-30 mV⁵². When performing this test on rats with implanted MC-RPNIs, EDL CMAPs are not significantly different, averaging 24.27 mV ± 1.34 mV. Additionally, when comparing generated CMAP waveforms between these two groups, they are remarkably similar (Figure 6C). As an additional measure of distally-innervated muscle function, muscle force testing of the muscle of interest can be pursued (Table 2). Following proximal CP nerve stimulation, the average EDL maximal tetanic force generated in MC-RPNI subjects is 2451 mN ± 115 mN, similar to the average force of 2497 mN ± 122 mN obtained from EDL muscle in control subjects⁵².

The overall purpose of the MC-RPNI is to amplify its contained nerve's microvolt-level signal by several magnitudes, increasing the SNR ratio and thus facilitating accurate detection of motor intent. This amplification has been demonstrated to occur in a reliable fashion in the range of 10-20 times⁵², with more recent experiments achieving amplification factors of over 50 times; therefore, if a construct does not provide a similar level of amplification, it is considered sub-optimal. Sub-optimal results can typically be attributed to problems at the level of the muscle graft in the MC-RPNI, as incomplete regeneration and thus reinnervation can result in lower than standard CMAP, lowering the overall amplification abilities of the construct. The generated waveform is typically attenuated, with a noticeably abnormal appearance. If the muscle graft completely fails, the signal measured at the muscle component can either be non-existent (secondary to significant scar tissue) or mirror the CNAP generated at the upstream nerve.

Figure 1: Illustrative schematic of the MC-RPNI. The target peripheral nerve can be seen in yellow within the surrounding muscle graft. The MC-RPNI is able to amplify its contained nerve's motor efferent action potentials on the level of microvolts through the generation of compound muscle action potentials (CMAPs) several magnitudes larger. This facilitates the detection of motor intent that is easily differentiated from background noise. Please click here to view a larger version of this figure.

Figure 2: MC-RPNI in vivo. The MC-RPNI is fabricated using an autologous extensor digitorum longus (EDL) muscle graft harvested from the contralateral limb. It is then circumferentially secured to the common peroneal nerve, with an example MC-RPNI outlined in white (A) at the time of initial fabrication. This same MC-RPNI is again pictured in (B) at the time of endpoint evaluation 3 months following. The MC-RPNI has similar coloration as the surrounding muscle and has retained a good portion of the volume. An example of an atrophic muscle graft is shown in (C). The MC-RPNI has a similar appearance to surrounding scar and connective tissue and has lost considerable volume. Please click here to view a larger version of this figure.

Figure 3: MC-RPNI histology. (A) H&E of a MC-RPNI cross-section with M indicating the muscle component, and N, the nerve. (B) Cross-section of the ipsilateral distally-innervated EDL muscle in a rat with a MC-RPNI. (C) Cross-section of EDL muscle in a control rat without a MC-RPNI. Please click here to view a larger version of this figure.

Figure 4: Immunostaining of the MC-RPNI. The image at the top left shows a longitudinal section of a MC-RPNI specimen with nuclei noted in blue (DAPI) and nerve tissue in green (neurofilament). A close-up of another MC-RPNI is shown at the bottom right with multiple neuromuscular junctions present (alpha-bungarotoxin in red for acetylcholine receptors). Please click here to view a larger version of this figure.

Figure 5: Endpoint electrophysiologic evaluation setup. Electrophysiologic testing requires at a minimum of three electrodes: (1) a grounding electrode - not pictured; (2) a nerve stimulating bipolar electrode; and (3) a bipolar recording electrode. In this setup, a bipolar stimulating hook electrode can be seen in white to the right of the image placed on the common peroneal nerve. The recording bipolar probe electrode is placed on the distal MC-RPNI. Signals are then recorded from the MC-RPNI following proximal nerve stimulation at the hook electrode until maximal CMAPs are achieved. Please click here to view a larger version of this figure.

Figure 6: Standard electrophysiologic waveforms. These graphs depict typical waveforms captured during electrophysiologic analysis of a rat with an implanted MC-RPNI following proximal CP nerve stimulation. (A) In blue, a CNAP (*) recorded from the CP nerve proximal to the MC-RPNI is pictured. The system artifact is indicated with a (**). (B) The representative CMAP recorded from the MC-RPNI following the generated CNAP in (A). (C) The resultant CMAP waveform recorded from the ipsilateral distally-innervated EDL muscle. Please click here to view a larger version of this figure.

Table 1: Endpoint electrophysiologic analysis of MC-RPNIs. A selection of results obtained from rats undergoing endpoint analysis at 3 (Rats 1-9) and 6 (Rats 10-12) months post-fabrication. Following proximal common peroneal (CP) nerve stimulation, compound nerve action potentials (CNAPs) were recorded at the downstream CP nerve, and compound muscle action potentials (CMAPs) were recorded at the downstream MC-RPNI. The amplification factor for each test can be seen in the column at the right. Note: for Rats 10-12, the CNAP proximal to the MC-RPNI was unable to be measured given anatomical limitations that resulted from fabricating the MC-RPNI too close to the CP nerve's takeoff from the sciatic nerve. The average CNAP amplitude recorded was 119.47 µV ± 14.87 µV while the average CMAP amplitude was 3.28 mV ± 0.49 mV, producing an average amplification factor of 31.8 ± 7.70.

Table 2: Muscle force analysis of rats with implanted MC-RPNIs. Muscle force testing was conducted on the ipsilateral extensor digitorum longus (EDL) muscle to determine if the MC-RPNI had any impact on distally-innervated muscle function. Following proximal CP nerve stimulation, force tracings were recorded and active force was calculated relevant to the test of interest. L_o was defined as the optimal muscle resting length that produced maximal force. Average maximal twitch force recorded from rats with implanted MC-RPNIs was 722.0 mN ± 32.11 mN and average maximal tetanic force recorded was 2451 mN ± 115 mN, similar to values obtained from control animals (maximal twitch: 822.2 mN ± 41.11 mN; maximal tetany: 2497 mN ± 122 mN).

Video 1: MC-RPNI contraction following proximal nerve electrical stimulation. Following proximal nerve electrical stimulation provided by the hook electrode at the right, visible muscle contraction of the MC-RPNI can be seen in the center. Please click here to download this Video.

The MC-RPNI is a novel construct that allows for amplification of an intact, peripheral motor nerve's efferent action potentials in order to accurately control an exoskeleton device. Specifically, the MC-RPNI confers a particular benefit to those individuals with extremity weakness caused by significant muscle disease and/or absence of muscle where EMG signals cannot be recorded. Reducing already compromised muscle function would be devastating in this population; however, the MC-RPNI has the ability to provide this nerve signal amplification without detriment to distally-innervated muscle⁵² (Table 1 and Table 2). In those individuals with muscle-based or lower motor neuron disease, peripheral sensory nerves are typically unaffected by the disease process⁵³. As sensation is preserved, it is imperative to keep the nerve in continuity and avoid injury, and the MC-RPNI appears to avoid any detriment to distally-innervated targets as a whole based on histology (Figure 3), immunohistochemistry (Figure 4), and evaluation of muscle function (Table 2).

The MC-RPNI relies on the concept of collateral axonal sprouting of the contained peripheral nerve, a concept readily demonstrated in both prior research⁵² and also in the well-described surgical technique of end-to-side neurorrhaphy^54,55. To ensure adequate reinnervation of the muscle graft during MC-RPNI fabrication and avoid negative impact to distally-innervated targets, meticulous handling of the nerve is imperative. During dissection of the nerve, trauma can be avoided through concise handling of either the epineurium or connective tissue only. However, the potential for nerve injury in MC-RPNI fabrication is the highest during the epineurial window step. To avoid sharp transection of nerve fibers, it is recommended to perform this step only under a high-power surgical microscope (at least 5x) after several opportunities for practice on non-experimental rats. This step can take several attempts to master, and it is not recommended to first perform this step on rats intended for experimental analysis. Theoretically, neuroma-in-continuity is a complication that could occur following MC-RPNI fabrication, especially in the presence of significant nerve trauma. However, this complication has not been encountered over the many years in development.

The majority of experiments performed with the MC-RPNI have been performed on the common peroneal nerve given its relative ease of access as well as evaluation of distally-innervated targets. Theoretically, any peripheral nerve with a motor component could be substituted. Pure sensory axons could be utilized as muscle tissue does have sensory components (spindle fibers, Golgi tendon organs, etc.), but these experiments have not been conducted thus far, and the results are difficult to predict. For the muscle graft component of the MC-RPNI, grafts range from 20-150 mg depending on graft length and age of the rat, and any similar size muscle graft can be successfully utilized. Muscle graft regeneration relies in part on the ability to revascularize, and large/thick grafts are more likely to undergo necrosis and fibrosis, affecting overall signaling capability⁵⁶. Research performed specifically on RPNIs has indicated successful muscle regeneration and maintenance of signal amplification in grafts up to 300 mg⁵⁶. With regards to rat breed, Lewis and Fischer are recommended as the majority of other rats used for experimental purposes are known to self-mutilate secondary to nerve injury^57,58.

Overall, current experiences with MC-RPNI fabrication have produced a failure rate of <5%. The most common construct failures seen are typically attributed to the muscle graft segment, whereupon on the exposure they are noted to be either atrophic or dislodged from the nerve. Dislodged MC-RPNIs usually result from inadequate suturing at the time of fabrication, leading to "opening" of the circumferentially wrapped muscle graft and eventual partial extrusion of the contained nerve. However, these MC-RPNIs usually retain some degree (albeit reduced) of signal amplification capabilities as a portion of the graft still remains secured to the nerve. Atrophic MC-RPNIs are obvious on exposure as they lack the typical skeletal muscle appearance, often undistinguishable from scar tissue with light pink to grey/white coloration (Figure 2C). Atrophy of the muscle tissue can result from many factors, including infection, too large/thick of a muscle graft, acute blood loss anemia, muscle and/or nerve injury during fabrication, as well as the failure of the epineurial securing sutures causing pistoning of the graft on the nerve, reducing revascularization. On electrophysiologic testing, atrophic MC-RPNIs typically produce little to no signal amplification; if using high-sensitivity electrodes, recordings of the underlying nerve's CNAP can be recorded through the atrophic muscle. If significant atrophy is noted in multiple experimental subjects, one must return to the protocol and determine which steps require adjustment. Of course, if no signals are recorded when performing evaluations, it is important to troubleshoot and not assume the construct is a failure. Troubleshooting the device set up is paramount, as lack of signals can be secondary to damaged electrodes (recommend impedance <16 Ω), incorrect electrode configuration, or even inadequate proximal nerve stimulation (some nerves require 0.5-5 mA electrical stimulation to begin producing CMAPs at the downstream MC-RPNI).

Current methods of human-machine interfacing for exoskeleton use in those with extremity weakness typically rely on recordings obtained from either peripheral nerves or EMG from muscle tissue. As discussed previously, the MC-RPNI confers significant benefit with regards to exoskeleton control for those individuals with severely damaged or absent muscle tissue where EMG recordings are not possible²⁹. The MC-RPNI also offers an advantage over current peripheral nerve interfacing options, including extraneural electrodes^30,31,32,33, intrafascicular electrodes^34,35,36 and penetrating electrodes^37,38,39,40. As inherent nerve signals are commonly on the level of microvolts, the MC-RPNI has the ability to amplify these nerve signals over 30 times, facilitating accurate detection of motor intent from background noise and thus enabling reliable exoskeleton control. With chronic use, current electrode-based methods ultimately struggle to overcome complications inherent to material longevity in vivo and foreign body reaction, complications the MC-RPNI is able to avoid given its biologic origin. Over time, these foreign body reactions result in tissue damage, scar tissue formation, and eventual axonal demyelination and degeneration. Experiments conducted up to six months have not revealed any evidence of neuronal injury, scar, or fibrosis/degeneration of distally-innervated muscle tissue (Figure 3), and in combination with RPNI stability noted in human subjects over a three year observation period⁴⁷, it is reasonable to conclude that MC-RPNIs could successfully interface with peripheral nerves on the scale of years to decades.

The MC-RPNI is intended to be utilized for exoskeleton control in a variety of pathologies, including those arising at the level of the nervous system as well as the muscle itself. For example, muscle-based pathologies can include conditions ranging from trauma, muscular dystrophy, inflammatory myopathies, and myasthenia gravis. Despite the profound muscular damage and weakness that can result in these conditions^1,2,3, the majority have functioning lower motor neurons that would facilitate MC-RPNI reinnervation and detection of motor intent. For those conditions that result in widespread muscle disease (muscular dystrophy, etc.), it is certainly possible that the free muscle graft component could be affected, thus limiting the amplification potential. However, given that detection of even a single motor unit (10-400 µV)⁵⁹ can provide amplification of peripheral nerve signals, it is reasonable to assume the MC-RPNI would contain enough motor units within its smaller, defined area to facilitate exoskeleton control in this population. A significant limitation of the construct, however, is in those pathologies that result in significantly reduced upper and/or lower motor neurons, such as in stroke, spinal cord injury, spinal muscle atrophy (SMA), and amyotrophic lateral sclerosis (ALS). Without a suitable peripheral nerve fiber population to reinnervate the MC-RPNI, it cannot regenerate and provide signal amplification, leading to construct failure. Experiments are being performed to determine the minimum population of functional peripheral nerve fibers required for adequate MC-RPNI function.

The MC-RPNIs predecessor, the RPNI, has shown immeasurable success with accurate control of powered prosthetics in human subjects through the amplification and recording of signals generated from transected peripheral nerves. Most notably, it is able to do so on the scale of months to years without reoperation or recalibration of the prosthetic device. Common complaints with current methods of human-machine interfacing for exoskeleton control center on signal contamination from cross-talk and the need for frequent recalibration in EMG-reliant methods^26,27,28, and peripheral nerve interface instability over time necessitating secondary surgeries^37,39,44. The MC-RPNI, however, is able to avoid these complications given its biologic makeup as well as strategic placement capabilities. It is imperative to establish a thorough understanding of this construct in order to pave the way for use in human subjects and the eventual widespread utilization of accurate, reliable exoskeleton devices in those living with extremity weakness.

The authors have no disclosures.

The authors thank Jana Moon for her expert lab management and technical assistance and Charles Hwang for his imaging expertise. Experiments in this paper were in part funded through Plastic Surgery Foundation grants to SS (3135146.4) as well as the National Institute of Child Health and Human Development under Award Number 1F32HD100286-01 to SS, and the National Institute of Arthritis and Musculoskeletal and Skin Diseases of the National Institutes of Health under Award Number P30 AR069620.