Restoration of proprioceptive and cutaneous sensation using Regenerative Peripheral Nerve Interfaces (RPNIs) in humans with upper-limb amputations

Summary

Without meaningful and intuitive sensory feedback, even the most advanced prosthetic limbs remain insensate and impose an enormous cognitive burden during use. The Regenerative Peripheral Nerve Interface (RPNI) can serve as a novel bidirectional motor and sensory neuroprosthetic interface. In previous human studies, RPNIs demonstrated stable high-amplitude motor EMG signals with excellent signal-to-noise ratio for prosthetic control. In addition, RPNIs can treat and prevent post-amputation pain by mitigating neuroma formation. In this study, we investigate whether electrical stimulation applied to RPNIs could produce appreciable proprioceptive and/or tactile sensations in two participants with upper-limb amputations. Stimulation of the RPNIs resulted in both participants reporting proprioceptive sensations in the phantom hand. Specifically, stimulation of Participant 1’s median nerve RPNI activated a flexion sensation in the thumb or index finger, whereas stimulation of the ulnar nerve RPNI evoked a flexion sensation of the ring or small finger. Likewise, stimulation of one of Participant 2’s ulnar nerve RPNIs produced a sensation of flexion at the ring finger distal interphalangeal (DIP) joint. In addition, stimulation of Participant 2’s other ulnar nerve RPNI and the median nerve RPNI resulted in perceived cutaneous sensations that corresponded to each nerve’s respective dermatome. These results suggest that RPNIs have the potential to restore proprioceptive and cutaneous sensory feedback that could significantly improve prosthesis use and embodiment.

Introduction

Modern robotic technology has provided amputation patients the ability to mimic basic movements and function of the human upper extremity.¹ Prosthetic users rely primarily on visual and auditory feedback to improve prosthetic control and function.² These necessary adaptations impose high cognitive load during prosthetic use and are known risk factors for prosthesis abandonment.³ The ability to provide proprioceptive and cutaneous feedback is essential to optimize functional performance and embodiment of prosthetic limbs^4–6. Some neural interfaces have demonstrated the potential to evoke meaningful sensory feedback to enhance prosthetic use^7–11 but none are able to reliably provide both motor and sensory modalities through one interface. The Regenerative Peripheral Nerve Interface (RPNI) is a biologic nerve interface that transduces neural signals by allowing a residual peripheral nerve to reinnervate a free skeletal muscle graft. RPNIs demonstrate high-amplitude motor EMG signals for prosthetic control and provide sufficient signal specificity for independent movements of artificial fingers.¹² In this study, we sought to characterize the potential afferent sensory capabilities of the RPNI with the overall goal of developing a reliable bidirectional prosthetic interface.

Materials and Methods

The Institutional Review Board at the University of Michigan approved this study and each participant provided written and informed consent. Electrode implantation surgery was performed under an FDA investigational device exemption.

Participant 1 (P1) was a 30-year-old male who previously sustained a traumatic amputation of the right hand resulting in right wrist disarticulation. Subsequently, P1 underwent resection of symptomatic median, ulnar, and dorsal radial sensory nerve neuromas in his distal forearm; one RPNI was created on each of these nerves (Fig. 1). In 2018, he underwent implantation of eight indwelling bipolar electrodes; one each in the median and ulnar nerve RPNIs, and six into intact forearm muscles associated with finger and wrist movements.¹² Figure 2 shows an example of a bipolar electrode implanted in P1’s median nerve RPNI.

Fig. 1: Illustrations of RPNI surgical creation and electrode placement.

(A) P1 had three RPNIs created, one on each of the median, ulnar, and radial nerves. A bipolar electrode was surgically implanted into each of the median and ulnar RPNIs one year after RPNI surgery (black rectangle). (B) P2 had four RPNIs created, one on each of the median and radial nerve and two on the ulnar nerve. Bipolar electrodes were surgically placed in each ulnar RPNI and the median RPNI.

Fig. 2: Surgical electrode implantation for P1.

(A) Example of a 30.5 cm bipolar electrode wire prior to implantation for P1. Rectangular inset shows the 5 mm positive and negative surfaces of the electrode contacts with a 10 mm gap in between contacts. The bipolar electrode lead length for P2 was 60 cm (not shown). (B) (Top) Implanted wire insertion into an individual RPNI (dashed line). (Bottom) Final connector setup after all electrodes have been implanted.

Participant 2 (P2) was a 53-year-old female whose right hand required a partial hand amputation after an IV extravasation injury resulting in progressive contracture and loss of functionality. Consequently, she underwent a distal transradial amputation. One RPNI was created on each of the median and radial nerves, and an intraneural dissection of the ulnar nerve was performed to create two ulnar nerve RPNIs (Fig. 1). One year after the RPNI surgery, P2 elected to undergo implantation of indwelling bipolar electrodes. Eight electrodes were implanted, one in each median and ulnar RPNIs and five in intact forearm muscles.¹²

Each bipolar electrode was implanted by creating a small 3–4 mm window in the muscle component of the RPNI or intact muscle belly with scissors. The electrode was inserted bluntly into the substance of the muscle and fastened by placing an absorbable stitch to secure the proximal wire in order to reduce motion at the electrode-muscle interface. The wire was passed proximally using a tendon passer instrument to exit percutaneously to a housing unit that was affixed to the skin with adhesive. To prevent infection, the electrode wire, connectors, and exit sites were cleaned with 70% isopropyl alcohol. All external components were covered with a soft dressing and transparent tape. Cleaning and dressing change was performed every 3 days.

Study patients participated in experimental stimulation sessions approximately 1 month following indwelling electrode placement and once per month thereafter as dictated by our FDA-approved IDE protocol. P1 was implanted for 12 months and had 7 stimulation sessions. P2 remains implanted at the time of writing and has had 15 stimulation sessions. RPNIs on the median and ulnar nerves were stimulated for 3–5 seconds with a biphasic square wave using a human-grade stimulator (DS7A, Digitimer, Ft. Lauderdale, FL, USA, used for P1; Neuro Omega, Alpha Omega, Alpharetta, GA, USA, used for P2). Stimulation parameter settings ranged from 20–100 Hz frequency, 100 or 200 µs pulse width, and 1–4 mA amplitude. Parameters were set at low values and linearly incremented one at a time until subjects reported a subjective sensory perception. Sensory perception thresholds were measured by adjusting the amplitude intensity parameter and fixing the frequency at 20 Hz and pulse width at 200 µS for P1 and 100 µS for P2. Increments of 0.1 mA were applied until subject reported perception of sensation. To test the effects of frequency, frequency was set at 20 Hz and adjusted by increments of 20 Hz with the amplitude intensity set just above perception threshold and the pulse width fixed at values mentioned earlier. Subjects reported on perceived sensations after each increment.

Participants were blinded to the electrode contact associated with each RPNI and which contact was stimulated for each trial. They reported where they felt the stimulation on a sketched drawing of an ipsilateral hand and arm, and on the researcher’s ipsilateral hand [See Video (online), which demonstrates the method of stimulation and reporting]. RPNIs were stimulated in a pseudo-random order. The quality, location, and associated stimulation parameters were recorded. Using MATLAB (MathWorks), linear regression models were fitted to test for increasing or decreasing trends in sensory perception thresholds across time.

Results

Anatomically appropriate proprioceptive sensations were reported when stimulating P1’s median and ulnar nerve RPNIs. P1 reported a sensation of flexion in his phantom thumb or index finger when stimulating the median nerve RPNI. When stimulating the ulnar nerve RPNI, a flexion sensation was felt in his phantom small or ring finger (Fig. 3A and Table 1). P2 reported both proprioceptive and cutaneous sensations. Stimulation of P2’s ulnar nerve RPNI 2 invoked a proprioceptive sensation at the distal interphalangeal joint of the phantom ring finger. In addition to proprioceptive sensations, stimulation of P2’s median and ulnar nerve RPNIs evoked sensations consistent with the dermatome of each nerve, respectively. Stimulation of the median nerve RPNI produced cutaneous sensations described as tingling at the base of her phantom thumb. Similarly, stimulation of ulnar nerve RPNI 1 produced a tingling cutaneous sensation along the ulnar aspect of the small finger and palm (Fig. 3B and Table 1).

Fig. 3: Sensory map of invoked sensory modalities from electrical stimulation.

(A) P1’s sensory map during RPNI stimulation. Orange and blue arrows indicate movement occurred in the referred phantom limb. (B) P2’s sensory map showing location of proprioceptive and cutaneous sensations during RPNI stimulation. Cutaneous sensations were reported in the dermatomes corresponding to the peripheral nerve that was stimulated.

Table 1.

Summary of stimulation parameters and perceived sensations

Participant	RPNI name	Nerve	Stimulation parameters	Perceived sensation
P1	Median RPNI	Median	3.0 mA, 20 Hz, 200 µS	Flexion at the phantom thumb or index finger
	Ulnar RPNI	Ulnar	1.0 mA, 20 Hz, 200 µS	Flexion at the small or ring finger
P2	Median RPNI	Median	2.0 mA, 100 Hz, 100 µS	Tingling near base of thumb
	Ulnar RPNI 1	Ulnar	1.5 mA, 100 Hz, 100 µS	Tingling near edge of small finger and palm
	Ulnar RPNI 2	Ulnar	1.5 mA, 100Hz, 100 µS	Flexion at DIP joint of ring finger

DIP – distal interphalangeal

Sensory perception thresholds were recorded across sessions in both participants (See Figure, Supplemental Digital Content 1, which shows A) P1’s sensory perception amplitude thresholds measured up to 271 days post-electrode implantation. Blue circles indicate amplitude thresholds for the ulnar nerve RPNI, while maize squares represent amplitude thresholds for the median nerve RPNI. Linear regression models were fitted and are displayed as maize and blue lines. There was no significant change in slopes for the median or ulnar nerve RPNI over time (p = 0.30, 0.83, respectively). (B) P2’s sensory perception amplitude thresholds measured up to 437 days post-electrode implantation. Blue circles and red diamonds represent amplitude thresholds for the two ulnar nerve RPNIs, while maize squares represent the median RPNI. Linear regression models were fitted and are displayed as maize, blue, and red lines. There was no significant change in the slope for the median RPNI (p = 0.45). However, ulnar nerve RPNIs 1 and ulnar nerve RPNI 2 showed a gradual change in slope (p < 0.05), INSERT HYPER LINK). In P1, the median and ulnar nerve RPNI perception thresholds did not change throughout the study period (271 days). In P2, the median RPNI perception thresholds also remained steady (437 days). However, P2’s ulnar nerve RPNI 1 and RPNI 2 sensory perception thresholds decreased gradually over time (p < 0.05), indicating that less stimulation was necessary to evoke sensory perceptions as the study progressed. Across sessions, P2 remained consistent with reporting the same sensations and perceived location when stimulating each RPNI. For P1, evoked sensations were different when comparing the first testing session to later testing sessions. Stimulating the ulnar RPNI at low amplitudes produced flexion sensations in the phantom index and middle finger, whereas stimulating at higher amplitudes yielded flexion sensations in the phantom ring and small fingers. In later sessions, stimulation at low or high amplitudes only provided flexions in the ring and small fingers.

Discussion

This preliminary report demonstrates that RPNIs can be used to mimic proprioceptive and cutaneous sensations in participants with upper-limb amputations. Electrical stimulation resulted in meaningful afferent percepts that were repeatedly experienced in the phantom hand. Our previous human study using ultrasound imaging revealed contractions of the RPNIs during volitional movements, indicating that efferent motor nerves have successfully reinnervated the muscle graft.¹² In this study, demonstration of proprioceptive sensation suggests that afferent sensory muscle spindle fibers have also reinnervated within the RPNI. For future studies, established methods of quantifying proprioception¹¹ will be utilized to explore the functional benefits of RPNI stimulation during physical use of a prosthesis.

For RPNIs facilitating cutaneous percepts, this may occur through direct afferent depolarization of free sensory nerve endings enclosed within the RPNI. We hypothesize that RPNIs provide physical and neurotrophic protection to regenerating cutaneous sensory nerve fibers which subsequently results in mitigation of neuroma formation.^13–16 Most notably, stimulation of P2’s two ulnar RPNIs created from intraneural dissection of the ulnar nerve demonstrated a separation of proprioceptive and cutaneous nerve fibers. This suggests that the ulnar nerve had been divided into its superficial (sensory) and deep (motor) branches, which allows the ability to perceive sensory feedback (afferent) and facilitate prosthetic control (efferent) independently.

These initial results encourage further investigation into the potential for RPNIs to provide naturalistic sensory feedback to enhance the use of an advanced prosthetic device. Previous studies have shown that a neuroprosthetic interface providing proprioceptive or cutaneous sensory feedback will improve a participant’s functional performance with a prosthetic limb.^7,10,11,17 Conceptually, RPNIs in the residual limb may be interfaced directly with existing force sensors built into a prosthetic device that would provide varying levels of stimulation. An increase in force detected on the fingertips of a prosthetic hand could result in increased electrical stimulation that is transduced by RPNIs into the perception of graded sensory feedback. Future studies will focus on combining the capture of efferent motor signals, while simultaneously stimulating RPNIs within the residual limb. Application of an implantable wireless system that concurrently transmits motor commands while receiving sensory feedback from prosthetic sensors will revolutionize prosthetic functionality and rehabilitation after limb loss.