Thought-Controlled Hand Works in Proof-of-Principle Study

Janis C. Kelly

February 26, 2015

Efforts to replace hand function lost because of brachial plexus injuries advanced with a proof-of-principle report by Austrian researchers. The new data indicate that free-functional muscle transfer and/or nerve transfer can provide the electromyographic signals needed for a patient to control a myoelectric hand prosthesis.

"In effect, brachial plexus avulsion injuries represent an inner amputation, irreversibly separating the hand from neural control. Existing surgical techniques for such injuries are crude and ineffective and result in poor hand function. The scientific advance here was that we were able to create and extract new neural signals via nerve transfers amplified by muscle transplantation. These signals were then decoded and translated into solid mechatronic hand function," lead author Oskar C. Aszmann, MD, director of the Christian Doppler Laboratory for Restoration of Extremity Function at the Medical University of Vienna, Austria, said in a journal news release.

Dr Aszmann, together with engineers from the Department of Neurorehabilitation Engineering of the University Medical Center Göttingen, Germany, reported successful outcomes in three patients in an article published online February 24 in the Lancet. The researchers used a new approach that combines selective nerve and muscle transfers, elective amputation, an advanced robotic prosthesis, and an intensive rehabilitation program using virtual reality to help patients attain cognitive control of the prosthesis. All of the patients had previously undergone brachial plexus reconstruction for injuries that included lower root avulsions.

The study results are encouraging because they provide evidence that the new approach can provide additional neural inputs into prosthetic systems, according to Simon Kay, MD, consultant plastic surgeon at Leeds Teaching Hospitals in the United Kingdom, who carried out the United Kingdom's first hand transplant. (Most myoelectric prostheses, including the devices used with these three patients, rely on transcutaneous muscle action potentials or surface electromyographic signals for control.)

However, Dr Kay, who was not involved in the study, told Medscape Medical News that this method cannot overcome the inherent limitation of a "choke point," where signals from the cerebral cortex are sent to the device.

Dr Kay said, "It seems highly unlikely that muscle action potentials on reinnervated muscles will ever have adequate volume of distinct signals intuitively controllable to allow the precise and nuanced control of a multiaxial multimotor dynamic limb. That kind of high-speed multichannel system will require higher-order control more centrally, and that can be best sought from direct representation of brain activity. Our current brain–world artificial interfaces are crude, but research in implantable arrays and in transcranial sensing shows potential."

The current case series included three patients. Patient 1 suffered avulsion of the right C8–T1 and nerve crush injuries at the costoclavicular space in a motor vehicle accident. All three cords using the remaining three roots (C5–C7) had been reconstructed using multiple autologous nerve grafts. An attempt to restore finger flexion using a free-functioning muscle transfer coapted to the brachialis motor nerve was unsuccessful. At 10 years postaccident, the patient had shoulder and elbow function but no useful hand function and suffered numerous unintentional burn injuries as a result of impaired sensory feedback.

Patient 2 suffered root avulsion injuries to the left C7–T1 and a stretch traction injury to the infraclavicular plexus in a free-fall climbing accident. Brachial plexus reconstruction and nerve transfers had returned shoulder and elbow function, but the patient had no useful hand function and also suffered intense phantom pain.

Patient 3 suffered a left brachial plexus injury of all five roots with avulsion of roots C7–T1 and other injuries in a motor vehicle accident. Multiple autologous nerve grafts reconstructed roots C5–C6 and restored shoulder and elbow function, but there was little activity below the elbow, and the patient had an internal rotation contracture of the shoulder with minimal range of motion.

Because all prior attempts at biological reconstruction had failed for these patients, they were considered candidates for the experimental bionic reconstruction. All had stiff, insensate hands. Eligibility requirements for the reconstruction required that they have neither useful sensation in their hand nor unstable shoulder nor an inability to lift the forearm against resistance.

In addition, reconstruction with a prosthetic hand required that the patient have at least two cognitively separate electromyographic signals in the forearm. Patient 2 had undergone early reinnervation of the forearm flexors with nerve transplantation, using the brachialis muscle branch. The recovered motor activity was not enough to move the patient's injured hand, but there was enough myoelectric activity to control a prosthesis. Patients 1 and 3 had fatty degeneration and fibrosis because of the delay after plexus reconstruction. Both were treated with free-functioning muscle transfer of the gracilis muscle. In patient 1, the forearm flexors and the forearm extensors were unable to move the damaged hand against gravity, but electromyographic activity in these two compartments was strong enough to control the prosthesis. Patient 3 underwent free-functioning muscle transfer with coaptation to the deep branch of the radial nerve to create one of the two signals needed for prosthesis control, and residual pronator teres activity provided the necessary opposing signal.

Before amputation of the damaged hand, all three patients underwent a transitional period of cognitive training, using virtual reality to learn how to activate the muscles that would be needed to control the prosthesis. The authors write, "The patients could practice the different functions of the prosthesis through virtual rehabilitation before actual fitting. Once confident in the virtual environment, they were fitted with a hybrid hand, in which a prosthetic hand was attached to a splint-like device fixed to their remaining hand. As crude as it seems, the device provided direct proof for patients that better hand function could be achieved using the prosthesis than with their denervated hand."

Finally, patients underwent amputation of the denervated hand and were fitted with either a Michelangelo Hand or a Myobock Hand (Otto Bock HealthCare, Germany). The most sensitive skin surface was used to cover the stump to provide better fitting and feedback.

The researchers measured outcomes using the Action Research Arm Test; the disabilities of the Arm, Shoulder and Hand questionnaire; visual analogue pain scales; and the Short Form-36 Health Survey (German version) for changes in quality of life.

The authors write, "Pre-interventional testing showed that all patients had dismal hand function. The patients did not use the impaired hand in daily life, even when bimanual tasks were specifically requested."

At 3 months after amputation, all three patients showed significantly better hand function, as well as improved physical functioning and mental health. The patients had begun to do bimanual tasks, and pain was significantly reduced for patient 2. Patients were able to accomplish many routine activities of daily living and were wearing their prostheses for 8 to 12 hours per day.

"The patients also had increased psychological wellbeing, which shows the importance of being able to interact with two hands with the environment for self-confidence and enjoyment of human interaction and social activities. Mental health also improved in two patients (and remained the same in the third), which reinforces our decision to select patients who had had great psychological harm as a result of injury," the authors write.

Dr Kay, who coauthored a linked comment, noted that patients' use of motorized prostheses tends to decrease over time because the devices are heavy, need power, are noisy, and are hard to repair. Dr Kay said that a wish-list for future bionic prostheses might include lighter materials, alternative power supply (eg from body movements), silent hydraulic actuators, and a processor controller that learns.

"Just as the human learns to use the machine, so the machine should learn its role by experience," Dr Kay said.

The authors, Dr Kay, and Dr Wilks have disclosed no relevant financial relationships.

Lancet. Published online February 25, 2015. Article abstract, Comment abstract


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: