A Novel Approach to Upper Extremity Amputation to Augment Volitional Control and Restore Proprioception

Upper extremity amputation is among the oldest known surgical procedures in medical history, with many of its technical principles having first been elucidated by Hippocrates. Despite the passage of more than two millennia, relatively little has changed in the operative approach to upper limb sacrifice. An estimated 58,000 patients in the United States currently suffer from upper extremity limb loss at either the above elbow (AEA) or below elbow (BEA) level, and the prevalence of upper limb amputation is expected to rise to approximately 95,000 patients by 2050.

Normal function of the upper limb is enabled through the dynamic interplay of multiple muscle groups acting in concert. Manual dexterity is a remarkably orchestrated biomechanical process that is dependent upon a complex feedback loop involving the central and peripheral nervous systems and the musculoskeletal system. In their native state, the muscles of the upper extremity exist in a balanced agonist/antagonist situation in which volitional activation of one muscle leads not only to its contracture, but also to passive stretch of its opposite. Changes in muscle tension manifest through this interaction of agonist and antagonist units lead to stimulation of specialized receptors within the muscle fibers (e.g., muscle spindle fibers and Golgi tendon organs) that transmit joint position information to the cerebral cortex. Such feedback, in conjunction with cutaneous sensory information from skin mechanoreceptors, provides us with a sense of limb proprioception that ultimately enables high fidelity limb control, even in the absence of visual feedback.

Unfortunately, the standard operative approach to upper limb amputation at either the AEA or BEA level obliterates many of the dynamic relationships characteristic of the uninjured upper extremity. Initial surgical exposure is typically accomplished through a fishmouth-pattern incision, followed by progressive transection of muscles, vessels, nerves and bone at the level of the incision. Tissues distal to the site of structural transection are discarded, regardless of whether or not there may be viable segments, and the proximal residual muscles are layered over the distal transected bone in order to provide insulation to this exposed osseous surface. The surrounding skin is then advanced over the bone/muscle construct in order to achieve definitive closure. The rudimentary approximation of discordant tissues in the distal limb in this approach results in a disorganized scar mass in which normal dynamic muscle relationships are destroyed. The uncoupling of native agonist/antagonist muscle pairings results in isometric contraction of residual muscle groups upon volitional activation, producing incomplete, unbalanced neural feedback to the brain that results in aberrant perception of residual limb position. Such disturbed feedback not only leads to impaired limb function with prostheses, but also manifests as pathological sensory perception of the extremity in the form of phantom limb and phantom pain symptoms.

To date, the limitations of these approaches have been tolerated due to the fairly simplistic goal of upper limb amputation: to provide a stable, padded surface for mounting a prosthesis. Historically, upper limb prostheses have afforded amputees the opportunity to recover at least some measure of upper limb function. However, such devices have generally not been able to recapitulate the complex biomechanics of the human upper limb due to limited ranges of motion and lack of feedback control. These limitations have resulted in reported upper limb prosthesis rejection rates ranging from 23% to 45%, including both body-powered and myoelectric devices.

However, the capabilities of modern prostheses are now expanding remarkably. Technological advances including increasingly miniaturized electronics, wireless communications and ever-refined positional sensors have enabled prosthetic developers to create next-generation bionic limbs with greatly enhanced degrees of freedom over prior models. Even more advanced prostheses are currently being developed that have the potential to offer sensory feedback - both tactile and positional - in a manner never before seen. Such prosthetic devices, while not yet available commercially, are presently being studied in experimental settings. For example, the Defense Advanced Research Projects Agency (DARPA) recently issued a request for proposals under the Hand, Proprioception and Touch Interfaces (HAPTIX) Program incorporating an upper limb prosthesis including six degrees of freedom at the wrist, thumb and all digits, 10 pressure sensors capable of providing sensory feedback, and joint angle and velocity sensors capable of providing joint position data.

Despite these technological advances in prosthesis development, surgical methods regarding management of the residual limb have not kept pace with these enhanced prosthesis capabilities. Classic techniques of upper limb amputation do not provide innervated interfaces that can serve as relays for complex prosthetic control; without such biological actuators in the residual limb to provide afferent and efferent conduits for information exchange, next generation prostheses are of little use. Stated another way, next generation prosthetic devices currently incorporate drivers and sensors capable of providing far more enhanced functionality than ever before seen, but standard approaches to limb amputation do not deliver a way to effectively link these prosthetics to their intended beneficiaries. An evolution in the manner in which upper limb amputations are performed - one that will provide a biological interface that will allow upper limb amputees to take advantage of the enhanced capabilities offered by the remarkable prostheses currently under development - is now required.

Recognition of the increased need for effective neural interfaces for prosthetic limbs can be seen in the expanding number of efforts in this sphere over the past decade. Initial efforts to provide high-resolution control of distal prostheses were focused primarily on indirect and direct brain interfaces, either through placement of electroencephalographic scalp sensors or implantable parenchymal electrodes, respectively. However, such endeavors have been plagued by poor resolution, inconsistencies in signal acquisition and, in the case of implantable devices, progressive foreign body reactions leading to impulse degradation over time.

As the limitations of brain interfaces have become more evident, focus has shifted instead to peripheral control loci. Efforts in this vein have included direct peripheral nerve interfaces including interposed sieves and cuffs designed to transduce electrical signals directly from individual nerve fascicles to distal prostheses. Such monitors have, however, shown little clinical promise due to progressive nerve compression secondary to scarring, as well as to significant neurological crosstalk and interference in biological models.

The most promising efforts regarding peripheral nerve interface development are now within the realm of biological systems. These models consist of configurations in which native tissues are innervated with distal nerve endings to create biological actuators for distal prosthesis control and feedback. The two leading models in this sphere are as follows:

• Targeted Muscle Re-innervation (TMR): Pioneered by Dumanian and Kuiken et al, TMR is a technique whereby a series of nerve transfers is used to re-innervate specific target muscles to create additional prosthesis control sites after distal limb amputation.
• Regenerative Peripheral Nerve Interfaces (RPNI): Championed by Cederna et al, RPNI offers an alternative version of an innervated biological interface. An RPNI is a surgical construct that consists of a non-vascularized segment of muscle that is coapted to a distal motor or sensory nerve ending.

While both TMR and RPNIs have demonstrated promise in offering improved functionality to patients who have already undergone amputation, neither technique has been incorporated into a fundamental redesign of the way in which amputations are performed in the first place; in all cases of clinical implementation of TMR or RPNIs reported to date in the literature, these techniques have been employed to further optimize the functionality of patients who had already experienced limb loss.

The clinical protocol proposes a reinvention of the manner in which upper limb amputations are performed, building upon several of the principles established in the work already performed in the realm of TMR and RPNIs. As elaborated below, the core innovation is the utilization of distal limb tissues that would ordinarily be sacrificed in the course of standard lower limb amputations to provide the substrate for natively innervated pairings of agonist/antagonist muscles capable not only of intuitive, volitional motor activation but also proprioceptive feedback. Conceptually, this idea consists of physical linkage of biologically opposed muscles (e.g., the biceps and triceps) such that when neurologically triggered contraction of one muscle is effected, simultaneous stretch of its partner is also achieved, resulting in observable motion of the dyad and stimulation of standard proprioceptive pathways. The investigators have named this construct the agonist-antagonist myoneural interface (AMI).

The manner in which this dynamic agonist/antagonist muscle concept may be operationalized clinically depends upon whether or not intact, innervated and vascularized native muscles are present at the time of operative intervention. Over the past five years, the research group has developed experimental models for a variety of clinical scenarios through a series of preclinical investigations in both murine and caprine populations.

If healthy native muscle is available as a reconstructive substrate, coaptation of the distal ends of disinserted agonist/antagonist pairs may be incorporated into the design of the residual limb; when coupled with a native synovial canal as a gliding interface, a pulley-like system can be established to provide a dynamic muscle construct. Construction of AMIs in a rat model have demonstrated preservation of construct muscle bulk, viability over time and production of graded afferent signals in response to ramp and hold stretches in a manner similar to native muscle architecture. Furthermore, performance of an amputation at the transtibial level with incorporation of AMI construction in a goat model has demonstrated clear coupled motion of the agonist-antagonist pair in the presence of both natural neural commands and artificial muscle stimulation.

Based on these proof of concept animal studies, the investigators hypothesize that the AMI offers the potential to provide a biological relay for volitional control that is superior to other neural interface strategies, with the additional benefit of being able to restore limb proprioception. When coupled with an appropriately adapted next-generation prosthesis, the AMI thus may provide the first biological mechanism to achieve true closed-loop neural interactivity with a mechanical limb.

The investigators here propose a three-year, prospective, controlled assessment of the functional and somatosensory advantages of the modified amputation model in an upper extremity scenario. The investigators believe this model has the potential to provide upper limb amputees with a biological interface that offers not only unprecedented, high-resolution motor control of prostheses, but also is highly intuitive and capable of restoring limb proprioception. If manifest, these augmented capabilities may result in improved functionality and overall health outcomes, including more robust return to work status and diminished psychological strain.