The following manuscript describes a novel method for developing a biologic, closed loop neural feedback system termed the composite regenerative peripheral nerve interface (C-RPNI). This construct has the ability to integrate with peripheral nerves to amplify efferent motor signals while simultaneously providing afferent sensory feedback.
Recent advances in neuroprosthetics have enabled those living with extremity loss to reproduce many functions native to the absent extremity, and this is often accomplished through integration with the peripheral nervous system. Unfortunately, methods currently employed are often associated with significant tissue damage which prevents prolonged use. Additionally, these devices often lack any meaningful degree of sensory feedback as their complex construction dampens any vibrations or other sensations a user may have previously depended on when using more simple prosthetics. The composite regenerative peripheral nerve interface (C-RPNI) was developed as a stable, biologic construct with the ability to amplify efferent motor nerve signals while providing simultaneous afferent sensory feedback. The C-RPNI consists of a segment of free dermal and muscle graft secured around a target mixed sensorimotor nerve, with preferential motor nerve reinnervation of the muscle graft and sensory nerve reinnervation of the dermal graft. In rats, this construct has demonstrated the generation of compound muscle action potentials (CMAPs), amplifying the target nerve's signal from the micro- to milli-volt level, with signal to noise ratios averaging approximately 30-50. Stimulation of the dermal component of the construct generates compound sensory nerve action potentials (CSNAPs) at the proximal nerve. As such, this construct has promising future utility towards the realization of the ideal, intuitive prosthetic.
Extremity amputations affect nearly 1 in 190 Americans1, and their prevalence is projected to increase from 1.6 million today to over 3.6 million by 20502. Despite documented use for over a millennium, the ideal prosthetic has yet to be realized3. Currently, there exist complex prosthetics capable of multiple joint manipulations with the potential to reproduce many motor functions of the native extremity4,5. However, these devices are not considered intuitive as the desired prosthetic motion is typically functionally separate from the input control signal. Users typically consider these "advanced prosthetics" difficult to learn and therefore not suitable for everyday use1,6. Additionally, complex prosthetics currently on the market do not provide any appreciable degree of subtle sensory feedback for adequate control. The sense of touch and proprioception are vital to carrying out daily tasks, and without these, simple acts such as picking up a cup of coffee become burdensome as it relies entirely on visual cues7,8,9. For these reasons, advanced prostheses are associated with a significant degree of mental fatigue and are often described as burdensome and unsatisfactory5,10,11. To address this, some research laboratories have developed prosthetics capable of providing a limited degree of sensory feedback via direct neural interaction12,13,14,15, but feedback is often limited to small, scattered areas on the hands and fingers12,13, and sensations were noted to be painful and unnatural at times15. Many of these studies unfortunately lack any appreciable long-term follow-up and nerve histology to delineate local tissue effects, while noting interface failure on the scale of weeks to months16.
For this population, the ideal prosthetic device would provide high fidelity motor control alongside meaningful somatosensory feedback from the individual's environment throughout their lifetime. Critical to the design of said ideal prosthetic is the development of a stable, reliable interface that would allow for simultaneous transmission of afferent somatosensory information with efferent motor signals. The most promising of current human-machine interfaces are those that interact with the peripheral nervous system directly, and recent developments in the field of neuro-integrated prosthetics have worked towards bridging the gap between bioelectric and mechanical signals17. Current interfaces utilized include: flexible nerve plates14,15,18, extra-neural cuff electrodes13,19,20,21,22,23, tissue penetrating electrodes24,25,31,32, and intrafascicular electrodes26,27,28. However, each of these methods has demonstrated limitations with regards to nerve specificity, tissue injury, axonal degeneration, myelin depletion, and/or scar tissue formation associated with chronic indwelling foreign body response16,17,18,19. More recently, it has been postulated that a driver behind eventual implanted electrode failure is the significant difference in Young's moduli between electronic material and native neural tissue. Brain tissue is subject to significant micromotion on a daily basis, and it has been theorized that the shear stress induced by differences in Young's moduli causes inflammation and eventual permanent scarring30,31,32. This effect is often compounded in the extremities, where peripheral nerves are subject to both physiologic micromotion and intentional extremity macromotion. Due to this constant motion, it is reasonable to conclude that utilization of a completely abiotic peripheral nerve interface is not ideal, and an interface with a biologic component would be more suitable.
To address this need for a biologic component, our laboratory developed a biotic nerve interface termed the Regenerative Peripheral Nerve Interface (RPNI) to integrate transected peripheral nerves in a residual limb with a prosthetic device. RPNI fabrication involves surgically implanting a peripheral nerve into an autologous free muscle graft, which subsequently revascularizes and reinnervates. Our lab has developed this biologic nerve interface over the past decade, with success in amplifying and transmitting motor signals when combined with implanted electrodes in both animal and human trials, allowing for suitable prosthetic control with multiple degrees of freedom2,34. In addition, we have separately demonstrated sensory feedback through the use of peripheral nerves embedded in dermal grafts, termed the Dermal Sensory Interface (DSI)3,35. In more distal amputations, using these constructs simultaneously is feasible as motor and sensory fascicles within the target peripheral nerve can be surgically separated. However, for more proximal level amputations, this is not feasible due to intermingling of motor and sensory fibers. The Composite Regenerative Peripheral Nerve Interface (C-RPNI) was developed for more proximal amputations, and it involves implanting a mixed sensorimotor nerve into a construct consisting of free muscle graft secured to a segment of dermal graft (Figure 1). Peripheral nerves demonstrate preferential targeted reinnervation, thus sensory fibers will re-innervate the dermal graft and motor fibers, the muscle graft. This construct thus has the ability to simultaneously amplify motor signals while providing somatosensory feedback36 (Figure 2), allowing for the realization of the ideal, intuitive, complex prosthetic.
All animal experiments are performed under the approval of the University of Michigan's Committee on the Use and Care of Animals.
NOTE: Donor rats are allowed free access to food and water prior to skin and muscle donation procedures. Euthanasia is performed under deep anesthesia followed by intra-cardiac potassium chloride injection with a secondary method of bilateral pneumothorax. Any strain of rat can theoretically be utilized with this experiment; however, our laboratory has achieved consistent results in both male and female Fischer F344 rats (~200-250 g) at two to four months of age. Donor rats must be isogenic to the experimental rats.
1. Preparation of the dermal graft
2. Preparation of the muscle graft
3. Common peroneal nerve isolation and preparation
4. C-RPNI construct fabrication
Construct fabrication is considered unsuccessful if rats develop an infection or do not survive surgical anesthesia. Previous research has indicated these constructs require approximately three months to revascularize and reinnervate2,3,17,36. Following the three-month recovery period, construct testing can be pursued to examine viability. Surgical exposure of the constructs after three months will reveal revascularized muscle and skin if successful (Figure 3). At times, the free muscle and dermal grafts can consist solely of scar tissue, and/or the nerve will not be attached to the construct; these findings indicate an unsuccessful attempt. However, if successful, gentle squeezing of the common peroneal nerve with forceps proximal to the construct will result in visible muscle contraction (Video 1). Histological analysis of constructs should demonstrate viable skin, nerve, and muscle (Figure 4). Immunostaining will also reveal motor and sensory nerve reinnervation to their neuromuscular junctions and sensory end organs, respectively (Figure 5). If the common peroneal nerve does not reinnervate those tissues, immunostaining will not demonstrate any individual nerve fibers within the construct with the exception of the implanted nerve itself.
Electrophysiologic testing can be performed on these constructs in vivo (Figure 6); previous research has been conducted at 3 and 9 months following C-RPNI fabrication36 (Table 1). Following maximal stimulation with a hook electrode at the proximal common peroneal nerve just distal to its takeoff from the sciatic nerve, compound muscle action potentials (CMAPs) can be measured at the muscle component with visible muscle contraction. The type of electrode used at the muscle can vary according to preference, but epimysial patch, epimysial pad, and bipolar probe electrodes have been used successfully in this research. The average CMAP amplitude recorded at the muscle was 8.7 ± 1.6 mV at 3 months and 10.2 ± 2.1 mV at 9 months. The average conduction velocity was 10 ± 1.2 m/s at 3 months and 9.5 ± 0.6 m/s at 9 months. In comparison, CMAPs generated by physiologic EDL muscle typically range from 10-18 mV37. Following stimulation at the dermal component of the C-RPNI, compound sensory nerve action potentials (CSNAPs) were produced at the proximal common peroneal nerve, with average CSNAP amplitude measuring 113.7 ± 35.1 µV at 3 months and 142.9 ± 63.7 µV at 9 months. Figure 7 illustrates single and summation CMAP and CSNAP signals obtained during electrophysiologic testing in a graphical format.
The C-RPNI serves to amplify a nerve's inherent microvolt signal, and previous research has demonstrated sufficient amplification from the microvolt to millivolt level38. Therefore, if a construct does not provide that level of amplification, it is not considered successful. If either the dermal, muscle, or both components of the C-RPNI fail, testing would result in recordings that mimic the stimulation signal utilized. For the muscle component specifically, a suboptimal result (but one that is still considered operational) would be one that has CMAP amplitude and conduction velocity in the range that falls between the signal stimulation value and that of physiologic EDL muscle. Additionally, these signals can become attenuated and lack the characteristic CMAP waveform (Figure 8A). Suboptimal results at the level of the dermal component can occur but are difficult to quantify given that rats cannot express the quality of sensation they experience. These suboptimal results usually involve dampening of the waveform with significant background noise (Figure 8B). However, if there is significant scarring or callusing of the skin graft, or minimal graft survived, no CSNAPs will be appreciated at the proximal common peroneal nerve regardless of stimulation value.
Figure 1: Illustrative schematic of the C-RPNI construct. The common peroneal nerve can be seen secured between the top dermal layer and bottom muscle layer. This construct is secured to the femur periosteum proximally and distally via EDL's tendinous junctions. Please click here to view a larger version of this figure.
Figure 2: A pictorial representation of the C-RPNI in a patient with a trans-radial amputation. The user forms a desired motor intention at the cerebral level (e.g., pincer grasp), which is transmitted as an efferent motor signal to the C-RPNI via the implanted peripheral nerve. This signal generates a compound muscle action potential (CMAP) at the muscle component, which is recorded by implanted electrodes and recognized by the prosthetic device, generating the desired motion. Sensors on the device's fingertips recognize the amount of pressure generated, and relay that information to an electrode implanted in the dermal component of the C-RPNI. These signals activate the corresponding sensory end organs, generating an afferent compound sensory nerve action potential (CSNAP) transmitted through the peripheral nerve to the sensory cortex. An example signal generated at each component is pictured within the blue boxes pictured next to each component. Please click here to view a larger version of this figure.
Figure 3: C-RPNI in vivo. (A) A C-RPNI immediately following fabrication and at (B) 3 months post-construction at time of electrophysiologic testing. The muscle component is the deep layer of the construct and the dermal, the superficial. Muscle tissue is marked by (M), dermis (D), and common peroneal nerve (N). Please click here to view a larger version of this figure.
Figure 4: C-RPNI histology 6 months. C-RPNI H&E at 6 months in (A) cross-section and (B) longitudinal section. Muscle noted by (M), dermis (D), and nerve (N). Please click here to view a larger version of this figure.
Figure 5: Immunostaining of the C-RPNI. (A) Representative example of a cross-section of muscle tissue, with red arrows identifying neuromuscular junctions. A higher magnification of the central neuro-muscular junction (NMJ) is pictured at the bottom-right. (B) Close-up of a neuromuscular junction noted in the sample. For (A) and (B), red staining (alpha-bungarotoxin) indicates presence of cholinergic receptors in muscle tissue; blue (neurofilament 200) specifies presence of neurofilaments within neuronal tissue; and green (choline acetyltransferase) notes specifically motor neuron presence. (C) Representative example of an iDISCO image focusing on the dermal junction, with red arrows marking sensory neurons (white) entering the dermis. (D) On-lay view of iDISCO demonstrating multiple sensory neurons (white, neurofilament 200). Please click here to view a larger version of this figure.
Figure 6: Electrophysiologic testing schematic. The top image is an illustration of the standard electrode arrangement for testing the C-RPNI constructs. There is a patch and/or probe electrode placed on both the muscle and dermal components of the C-RPNI, with a double hook electrode placed at the common peroneal nerve proximally. The bottom image is an in vivo example of the testing arrangement on a rat subject. Please click here to view a larger version of this figure.
Figure 7: Typical C-RPNI electrophysiologic signaling. (A) A single CMAP signal recorded at the muscle component following a 5.00 mA signal applied to the CP nerve. (B) 24 CMAPs generated by 5.00 mA stimulation at the nerve. (C) A single CSNAP signal recorded from the proximal CP nerve following dermal component stimulation at 900 µA. (D) A series of CSNAPs recorded from the proximal CP nerve following increasing stimulation at the dermal component from 500 µA to 1000 µA. Please click here to view a larger version of this figure.
Figure 8: Abnormal C-RPNI signaling. (A) A series of CMAPs obtained while ramping CP nerve stimulation from 0.2 to 4 mA. Waveforms peak at different points and fail to return to baseline, possibly indicating defective electrodes or inadequate overall construct function. (B) Summation of CSNAPs obtained while stimulating dermal component, ramping 0.1 to 5 mA. These findings can occur for a multitude of reasons, including malfunctioning electrode(s), dermal graft scarring, and/or nerve damage. Please click here to view a larger version of this figure.
3 Month Data | CMAP Data (Stimulate CP nerve and record from muscle graft) | CSNAP Data (Stimulate skin graft and record from CP nerve) | |||||
Rat ID Number | Construct Weight (g) | Stimulation Amplitude (mA) | Conduction Velocity (m/s) | V Peak-to-Peak (mV) | Stimulation Amplitude (mA) | Conduction Velocity (m/s) | V Peak-to-Peak (µV) |
4607 | 0.087 | 4.17 | 11.3 | 10.3 | 18 | 11.1 | 121 |
4608 | 0.15 | 1.65 | 11.1 | 17.1 | 7.7 | 6.5 | 136 |
4611 | 0.113 | 8.3 | 9.6 | 11.2 | 10 | 10 | 121 |
4613 | 0.116 | 3.18 | 10 | 9.6 | 1.44 | 8.3 | 134 |
4614 | 0.189 | 3 | 10.8 | 9.6 | 7.39 | 9 | 151 |
4616 | 0.122 | 5.2 | 9.4 | 14.9 | 1.8 | 9.1 | 100 |
4620 | 0.118 | 2.91 | 7.6 | 7.4 | 8.7 | 10 | 219 |
9 Month Data | CMAP Data (Stimulate CP nerve and record from muscle graft) | CSNAP Data (Stimulate skin graft and record from CP nerve) | |||||
Rat ID Number | Construct Weight (g) | Stimulation Amplitude (mA) | Conduction Velocity (m/s) | V Peak-to-Peak (mV) | Stimulation Amplitude (mA) | Conduction Velocity (m/s) | V Peak-to-Peak (µV) |
4687 | 0.238 | 1.35 | 9.6 | 18.2 | 0.99 | 11 | 181 |
4688 | 0.131 | 1.08 | 10 | 8.8 | 1.11 | 8 | 132 |
4689 | 0.26 | 1.26 | 9.6 | 21.8 | 1.9 | 8.6 | 237 |
4690 | 0.192 | 4.2 | 8.3 | 12.8 | n/a | n/a | n/a |
4691 | 0.213 | 1.38 | 10 | 18.6 | 6.6 | 8 | 153 |
4693 | 0.178 | 1.11 | 9.6 | 15.1 | 8.7 | 8.3 | 306 |
Table 1: Electrophysiologic testing of C-RPNIs at 3- and 9-months post-construction. To obtain CMAPs, a recording electrode was placed on the muscle with a stimulating electrode on the proximal common peroneal nerve. A series of stimulations increasing in amplitude was applied to the nerve until maximal CMAP values were obtained and results recorded. A similar methodology was applied to the dermal component but with the recording electrode placed on the nerve and stimulating electrode on the dermis. For the sensory evaluation of rat 4690 at 9 months, the dermal graft was found to be too scarred to allow for testing.
Video 1: Muscle contraction within a C-RPNI. A pair of forceps can be seen to the left of the video gently squeezing the proximal common peroneal nerve. This results in contraction of the muscle component of a 3-month-old C-RPNI that is visible to the viewer. Please click here to view this video (Right click to download).
The C-RPNI is a novel construct that provides simultaneous amplification of a target nerve's motor efferent signals with provision of afferent sensory feedback. In particular, the C-RPNI has unique utility for those living with proximal amputations as their motor and sensory fascicles cannot easily be mechanically separated during surgery. Instead, the C-RPNI utilizes the inherent preferential reinnervation properties of the nerve itself to encourage sensory fiber reinnervation to dermal sensory end organs and motor fibers to neuromuscular junctions.
As C-RPNI fabrication relies on the reinnervation abilities of the target nerve, careful handling of the nerve is paramount during the procedure. During dissection, avoid direct manipulation of, and trauma to, the target nerve. If the nerve must be handled, it is recommended to manipulate epineurium or surrounding connective tissue instead. Although our laboratory has not encountered neuroma formation within this construct, theoretically, significant nerve trauma could increase the risk. An additional key step in the process is the harvesting of the dermal grafts. All epidermal tissue must be removed from the hindpaw graft as retained epidermis can increase the risk of infection and inclusion cysts during the healing process. Furthermore, the dermal graft must be adequately thinned to promote imbibition and revascularization throughout the graft and avoid significant ischemia and necrosis.
Although the majority of studies conducted with the C-RPNI have been performed on the common peroneal nerve, any mixed sensorimotor nerve could be substituted. A pure motor or pure sensory nerve could be utilized, but the results are difficult to predict and would likely result in either largely muscle or dermal reinnervation, respectively. With regards to the muscle graft, as long as epimysium is removed from the portion contacting the nerve, any muscle graft similar in size could be utilized as long as it contained tendinous or fascial tissue at either end suitable for anchoring to nearby periosteum. For the dermal graft, glabrous tissue is specifically used due to the potential for hair growth following grafting. Non-glabrous skin was previously attempted, but due to the difficulty of removing individual hair follicles, all resultant constructs had significant hair growth, inflammation, and scarring following the three-month maturation period. Additionally, other rat species can be employed, but Lewis and Fischer rats are recommended for this experiment as many other rat species will self-mutilate secondary to nerve transection39,40.
Given the delay between procedure and results, it is difficult to know ahead of time if any alterations must be made to the method. Infection is a theoretical risk rarely encountered by our laboratory, but if infection occurs, it is typically responsive to antibiotics. Occasionally, rats chew on their incisions causing dehiscence, and this can be treated with washout, debridement, and re-closure. If, after three months at time of exposure, the construct is found to be non-functional and/or scarred, there are several potential causes. At times, if the nerve is not secured correctly to the construct with at least three sutures, the nerve can tear from the construct with ambulation. Additionally, the muscle and/or dermal grafts can necrose, causing failure. Typically, this is a result of either repeated infection, the dermal graft being too thick, or the muscle too damaged at harvest to recover properly. Additionally, if the muscle is not secured to periosteum at resting length, contraction can be impaired causing inadequate signals during testing. At times, the construct will appear viable but fail to produce adequate CMAPs/CSNAPs on testing (5-10% of constructs, on average). This could be secondary to failure in equipment, elevated electrode impedance, or significant skin callusing. Skin callusing can dampen and completely block signal transduction if the dermis is not thinned properly during fabrication. If any of the preceding described events are seen frequently during the testing process, one must return to the protocol and make appropriate alterations. In our laboratory's experience with over 90 successful C-RPNI constructs, our failure rate is <5% and typically attributed to surgical error during fabrication.
Methods commonly employed to amplify or record nerve signals include flexible nerve plates18, extra-neural cuff electrodes19,20,21,22,23, tissue penetrating electrodes24,25,31,32, and intrafascicular electrodes26,27,28, all of which have been associated with tissue injury, axonal degeneration, and/or scar tissue formation. This scarring is often attributed to chronic indwelling foreign body response29 and shear stress induced by differences in Young's moduli30. The C-RPNI, however, is a biologic construct and thus does not induce foreign body response in neural tissue. Additionally, its mechanical properties are several factors closer to neural tissue than electrodes. Histologic analysis of these samples has not demonstrated any significant degree of scar tissue formation in the nerve with chronic use, thus allowing the C-RPNI to interface with the nerve for extended periods in comparison to the methods listed above. Although this method is highly effective at amplification of efferent motor signals, it is limited with regards to sensory afferent signal production. We have measured and characterized signal transduction produced with mechanical and electrical stimulation of the dermal component of the C-RPNI36; however, these rat subjects cannot qualify the type or degree of sensations elicited from stimulation of this construct. As such, at this time it is impossible to know what kind of effect the C-RPNI is producing with regards to sensation. Future directions for this construct will include characterization of signals produced in the proximal nerve following specific provided stimuli (e.g., heat, pain, pressure, etc.) as well as correlation with somatosensory evoked potentials generated in the sensory cortex of the rodent brain. It is our laboratory's goal to establish a comprehensive foundation for C-RPNIs that will pave the way for clinical translation to human subjects.
The predecessor to the C-RPNI, the RPNI (regenerative peripheral nerve interface), consists of a free muscle graft attached to a transected nerve, with motor fibers reinnervating previously denervated neuromuscular junctions. The RPNI has demonstrated utility in human subjects, with several patients controlling advanced prosthetics from signals amplified by-and recorded from-these RPNIs34. Furthermore, these RPNIs have demonstrated beneficial treatment effects beyond prosthetic control, with several preliminary retrospective and prospective studies showing decreased neuroma formation, chronic pain, and phantom limb pain in those patients with extremity amputations. Despite these successes, a common complaint for those utilizing these advanced prosthetics, however, is the need to visualize the prosthetic during use as these prosthetics lack proprioception and provide minimal sensory feedback. The C-RPNI could be a solution to this common criticism by providing a way to deliver sensory feedback via the dermal component, leading to the realization of the much-desired, ideal prosthetic.
The authors have nothing to disclose.
The authors wish to thank Jana Moon for expert technical assistance. Studies presented in this paper were funded through an R21 (R21NS104584) grant to SK.
#15 Scalpel | Aspen Surgical, Inc | Ref 371115 | Rib-Back Carbon Steel Surgical Blades (#15) |
4-0 Chromic Suture | Ethicon | SKU# 1654G | P-3 Reverse Cutting Needle |
5-0 Chromic Suture | Ethicon | SKU# 687G | P-3 Reverse Cutting Needle |
6-0 Ethilon Suture | Ethicon | SKU# 697G | P-1 Reverse Cutting Needle (Nylon suture) |
8-0 Monofilament Suture | AROSurgical | T06A08N14-13 | Black polyamide monofilament suture on a threaded tapered needle |
Experimental Rats | Envigo | F344-NH-sd | Rats are Fischer F344 Strain |
Fluriso (Isofluorane) | VetOne | 13985-528-40 | Inhalational Anesthetic |
Micro Motor High Speed Drill with Stone | Master Mechanic | Model 151369 | Handheld rotary tool; kit comes with multiple fine grit stones |
Oxygen | Cryogenic Gases | UN1072 | Standard medical grade oxygen canisters |
Potassium Chloride | APP Pharmaceuticals | 63323-965-20 | Injectable form, 2 mEq/mL |
Povidone Iodine USP | MediChoice | 65517-0009-1 | 10% Topical Solution, can use one bottle for multiple surgical preps |
Puralube Vet Opthalmic Ointment | Dechra | 17033-211-38 | Corneal protective ointment for use during procedure |
Rimadyl (Caprofen) | Zoetis, Inc. | NADA# 141-199 | Injectable form, 50 mg/mL |
Stereo Microscope | Leica | Model M60 | User can adjust magnification to their preference |
Surgical Instruments | Fine Science Tools | Various | User can choose instruments according to personal preference or from what is currently available in their lab |
Triple Antibiotic Ointment | MediChoice | 39892-0830-2 | Ointment comes in sterile, disposable packets |
VaporStick 3 | Surgivet | V7015 | Anesthesia tower with space for isofluorane and oxygen canister |
Webcol Alcohol Prep | Coviden | Ref 6818 | Alcohol prep wipes; use a new wipe for each prep |