THE AGONIST-ANTAGONIST MYONEURAL INTERFACE (AMI) IN THE SURGICAL REVISION OF UPPER LIMB AMPUTATION FOR THE PROPRIOCEPTIVE FEEDBACK AND ADVANCED CONTROL OF BIONIC POLYARTICULATED PROSTHESIS

Currently, in Europe, at least 100,000 people suffer from major upper limb amputations, with an annual increase of 2%. Advanced poly-articulated prostheses, which interpret user intentions from electromyographic (EMG) signals recorded via surface electrodes on residual muscles, are commercially available. This non-invasive signal collection method can be affected by various factors such as temperature, sweat, electromagnetic fields, and movement artifacts. Additionally, these prostheses allow for isolated joint movements but not for the simultaneous and coordinated movement of multiple joints. Motor control is challenging and not intuitive, especially for patients with very proximal amputations (trans-humeral or shoulder disarticulation) who lack sufficient residual muscles to decode the EMG signal for multiple degrees of freedom. Proper control of these devices can only be achieved with visual feedback or other types of incidental feedback (such as auditory cues or socket pressure). A significant limitation and reason for the abandonment of today’s prostheses is the lack of sensory feedback, increasing the cognitive load required to perform even simple tasks. This contributes to the high drop-out rate of myoelectric prostheses (about 39%). Research is now focused on restoring sensory feedback, particularly proprioceptive feedback, which is crucial for controlling grasping and manipulation functions and for efficient motor planning and execution without visual feedback. Without proprioception, the position and velocity of the prosthesis must be inferred visually, and forces in prosthesis-environment interactions can only be indirectly perceived through the limb-prosthesis connection. Currently, invasive strategies for proprioception restoration are a significant challenge. Among the various methods described, the agonist-antagonist myoneural interface (AMI) is the most innovative. The AMI is a surgical construct and neuroprosthetic interfacing strategy that involves grafting two residual muscles to create an agonist-antagonist pair within the residual limb, allowing bidirectional communication with an external prosthesis through sensors, stimulators, and a control system, thereby creating a neural control loop. In the AMI tissue construct, a primary effector muscle (the agonist) is biomechanically linked to a counter muscle (the antagonist) so that contraction of the agonist causes proportional stretching of the antagonist. This dynamic relationship activates both afferent and efferent neural pathways, mimicking a natural biological system. Multiple AMIs could potentially control multiple degrees of freedom within an external prosthesis. The AMI procedure is crucial for restoring limb proprioception, increasing volitional control of the prosthesis, and preventing or reversing residual limb atrophy. Despite its complexity, implementing the AMI procedure for restoring proprioceptive feedback in upper limb amputations is a significant step, especially since it has never been applied to these patients before. The project is structured in two phases. First, the AMI procedure for upper limb amputation was developed on human cadaveric models. Then, the technique was applied to amputees of the upper extremity for the first time. A transhumeral amputation at the distal third of the humerus was simulated on cadaveric models. Biceps and triceps muscles were isolated, and the bone stump was exposed and prepared to function as a pulley for the sliding of the biceps-triceps construction sutured with nonabsorbable stitches. Some issues emerged, such as the difficulty of associating this procedure with the implantation of an osseointegrated prosthesis, often indicated for patients with trans-humeral amputation. A technical solution involved creating a pulley system on the distal side of the residuum. Moreover, the simulated amputation used healthy bone and muscle tissues, allowing for residual limb shortening as needed, thus limiting the technique to patients with sufficient muscle tissue and bone stock. In July 2022, the first surgical procedure was performed on M.D.P., a young man who had a motorcycle accident resulting in a right transhumeral amputation and proximal paralysis of the residual right upper limb in July 2021. The patient also suffered from brachial plexus palsy. Since less than 12 months had passed since the trauma, plexus repair was possible, along with a targeted reinnervation operation (TMR) and an AMI for proprioceptive recovery. The patient underwent clinical and instrumental tests, including neurological evaluation, EMG, right shoulder X-ray, and brachial plexus MRI, confirming a post-traumatic injury with axonotmesis of the right brachial plexus (C5 and C6 root injury). The surgical procedure involved brachial plexus repair with sural nerve grafts, TMR, and a connection through tenomyodesis between the biceps and triceps tendons (AMI procedure). Eight months postoperatively, there were no complications. In May 2023, the patient underwent a joint neurological and orthopedic clinical assessment and was satisfied with the progress, particularly greater shoulder mobility and the absence of phantom limb pain. There was an objective improvement in right upper limb abduction movements, with visible contractions of the deltoid, previously absent. Electromyographic studies showed signs of reinnervation in the deltoid, short head of the biceps brachii, and brachialis muscles. The patient's post-operative course was uneventful, with regular wound healing and no complications. Radiographs showed stable bony anatomy over time. The brachialis and the short head of the biceps, reinnervated through TMR, functioned effectively. Consequently, the AMI construct between the brachialis and biceps from the anterior compartment and the triceps from the posterior compartment successfully restored upper limb proprioceptive feedback. Preoperatively, the patient could perceive but not move the amputated limb. Post-surgery, the patient experienced natural sensations of joint movement throughout the phantom joint space, associated with voluntary AMI muscle activation. The construct is stable, allowing various fixation strategies based on distal tissue availability, and our dermal-adipose graft effectively preserved construct excursion. The significant improvement in proprioceptive feedback demonstrates the procedure's efficacy, consistent with outcomes reported in lower limb amputations. No significant muscular atrophy was observed in the residual limb, possibly due to the preservation of natural firing patterns from repetitive proprioceptive feedback during voluntary or reflexive AMI muscle contractions. Additionally, our research confirms that osseointegration is a beneficial surgical method for amputees, offering more adaptability than traditional socket prostheses. Osseointegrated prostheses are a promising treatment for selected patients, enhancing functional outcomes, quality of life, and playing a crucial role in the prosthetic integration process.