Journal Logo


Targeted Reinnervation for Transhumeral Amputees: Current Surgical Technique and Update on Results

Dumanian, Gregory A. M.D.; Ko, Jason H. M.D.; O’Shaughnessy, Kristina D. M.D.; Kim, Peter S. M.D.; Wilson, Christopher J. M.D.; Kuiken, Todd A. M.D., Ph.D.

Author Information
Plastic and Reconstructive Surgery: September 2009 - Volume 124 - Issue 3 - p 863-869
doi: 10.1097/PRS.0b013e3181b038c9


Prosthetic function for upper extremity transhumeral amputees is poor. Body-powered technology that uses bicycle cables to transfer the energy of the chest and shoulder girdle to move a prosthetic limb is the standard of care. Motorized prostheses controlled with electromyographic signals are available. However, only one prosthetic “joint” can be manipulated at a time, and it is not intuitive to use the residual biceps and triceps to control a prosthetic hand or wrist. These limitations have prevented routine activities from being performed in a usable time frame; the function of conventional transhumeral prostheses is generally poor.

Recently, a new surgical procedure termed “targeted reinnervation” was introduced.1–3 Amputated upper extremity nerves that had previously innervated the arm are transferred to reinnervate nearby, otherwise functionless muscle segments. The muscles are considered functionless, as they are no longer connected distally to an active joint. After successful neurotization, the newly reinnervated muscle serves to amplify the electric signal from the amputated nerve. The electromyographic signal is detected transcutaneously and used to improve the control of a myoelectric prosthesis. The nerve transfers are performed so as to not injure the intact musculocutaneous (biceps) and proximal radial (triceps) signals. The system is intuitive, because the biceps and triceps electromyographic signals control the prosthetic elbow joint flexion and extension, whereas the median and distal radial innervated muscles (after successful transfers) once again control the opening and closing of the prosthetic hand.

The first generation of this surgical procedure was performed 5 years ago and reported recently.4 Refinements in the procedure have allowed the technique to be applicable to patients with more proximal transhumeral amputations who do not have a brachialis muscle. In this report, we document our current surgical technique and present our results with this procedure.


Anatomical Dissections

Eleven brachial plexus dissections were performed to better understand the complex anatomy of the radial nerve in the posterior upper arm.

Clinical Study

This study describes the results of the first six patients with transhumeral amputation to receive targeted reinnervation as described below. Follow-up of at least 1 year is available for each of these patients.

Surgical Procedure

The goal of this procedure is to transfer the median nerve to the motor nerve entering the medial head of the biceps, and the distal radial nerve to the motor nerve entering the lateral head of the triceps. For patients with long residual limbs, the ulnar nerve is transferred to the motor nerve of the brachialis muscle. Another goal is to maximize the amplitude and separation of electromyographic signals for prosthesis control. Electromyographic amplitude is maximized by removing subcutaneous fat over the muscles of interest. Electromyographic signal separation is maximized by placing this subcutaneous fat between the muscles of interest to create spatial separation. Both of these goals are realized by creation of an adipofascial flap as described below. The adipofascial flap also decreases the chance of the original motor nerve reinnervating the target muscle segment.

In the holding area, the outlines of the biceps and triceps are carefully marked. The lack of a distal insertion for these muscle groups allows the muscle bellies to spin around the humerus and to be difficult to palpate with the patient asleep. Under general anesthesia in the supine position with the arm circumferentially prepared, an anterior incision over the biceps muscle is made from the inferior edge of the deltoid toward the end of the amputation (Fig. 1, above, left). Dilute epinephrine solution (without lidocaine) is used to tumesce the tissues, improve hemostasis, and open tissue planes. Skin flaps are elevated medially and laterally just under the dermis, analogous to a face lift skin flap, for 2 to 3 cm in each direction. Then, the exposed fat and fascia is elevated as a proximally based adipofascial flap (Fig. 1, above, right). The raphe between the heads of the biceps are opened widely, exposing the musculocutaneous nerve with its branches to the heads of the biceps and the continuation of the nerve that includes the motor branch to the brachialis muscle (Fig. 1, center, left). Dissection medial to the biceps is performed to identify the median nerve adjacent to the brachial artery. After adequate distal mobilization (Fig. 1, center, right), the median nerve is transected distally and passed into the space between the heads of the biceps next to the musculocutaneous nerve. The motor branch to the medial head of the biceps is transected close to the muscle and cephalad to any bifurcations of the nerve as it enters the muscle. The medial head motor branch is then buried in the lateral biceps to prevent it from reinnervating the medial biceps. After freshening the median nerve to unaffected fibers, the median nerve is coapted to the motor nerve of the medial head of the biceps, catching the epimysium of the muscle to improve the strength of the coaptation. Typically, this is performed without a microscope and with 5-0 Prolene sutures (Ethicon, Inc., Somerville, N.J.). The motor nerve to the lateral head of the biceps is carefully protected. The adipofascial flap is then sewn in between the heads to decrease the chance of aberrant reinnervation and to improve the distance between the muscles axially for later electromyographic signal separation (Fig. 1, below).5,6 Another benefit of raising the adipofascial flap is to decrease the subcutaneous tissue thickness in the radial direction over the biceps musculature for later electromyographic signal detection.

Fig. 1.
Fig. 1.:
(Above, left) Marking of the midline raphe between the medial and lateral heads of the biceps. (Above, right) Adipofascial flap raised to expose the biceps muscle and its raphe. (Center, left) The medial and lateral heads of the biceps are separated to reveal the musculocutaneous nerve. Vessel loops are passed around the motor nerves entering the two muscle bellies. A nerve seen between these two motor nerves and without a vessel loop is the continuation of the musculocutaneous nerve as it innervates the brachialis muscle. (Center, right) Dissected median and ulnar nerves draped over the biceps muscle. The vessel loops seen are still passing around the motor nerves to the medial and lateral head of the biceps as in center, left. (Below) An adipofascial flap in place between the heads of the biceps muscle.

The radial nerve transfer is performed in a manner analogous to the median nerve transfer. With the patient still in the supine position, full extension of the shoulder easily reveals the posterior aspect of the arm. An incision over the raphe between the long and lateral heads of the triceps is made and an identical proximally based adipofascial flap is elevated (Fig. 2, above). Blunt dissection between the heads of the triceps reveals motor nerves entering the lateral head and the radial nerve traveling distally. The ability to split the triceps is variable. In some patients, a dissection around the lateral head triceps rather than between the heads will be necessary to expose motor nerve branches, and this dissection is similar to that of a lateral arm flap. No branches to the long head of the triceps are seen with this straight posterior approach, as the motor branches to the long head leave the radial nerve separate and more proximal from the motor nerves to the lateral head. A motor branch traveling to the lateral head is selected for its size and distal entry into the muscle. This motor nerve is followed proximally and transected off of the radial nerve (Fig. 2, below). The radial nerve proper is identified and followed distally toward the amputation stump, transected, and cut back to healthy fascicles. A coaptation between the distal radial nerve and the motor nerve to the lateral head of the triceps is performed, again including epimysium in the bite of suture to improve the strength of the repair. The adipofascial flap is inserted between the heads of the muscle, and the incision is closed.

Fig. 2.
Fig. 2.:
(Above) The patient is still supine on the operating room table, and the shoulder is now fully extended to reveal the posterior surface of the arm. An incision between the heads of the triceps is made, and a similar proximally based adipofascial flap is raised. (Below) A right-angle clamp demonstrates a large motor nerve innervating the lateral head of the triceps that is about to be clipped and divided. The distal radial nerve is seen immediately posterior to the right-angle clamp.


Anatomical Dissections

Targeted reinnervation for the radial nerve would require that the motor nerves to the long head of the triceps be proximal and distant from the lateral head motor nerves. Our findings were consistent with other cadaveric dissection series.7 The motor nerves to the long head leave the radial nerve 3 to 4 cm proximal to the exit site of the lateral head motor nerves (Figs. 3 and 4) and travel medially in the upper arm, away from the lateral head nerves.

Fig. 3.
Fig. 3.:
Anatomical dissection of the radial nerve branches in the right arm. The hand is to the left, and the humerus is removed. Motor branches to the lateral head exit the nerve and travel quite separately from the motor branches to the medial and long heads.
Fig. 4.
Fig. 4.:
Anatomical diagram of the course of the motor branches to the triceps seen from the anteroposterior view.

Clinical Study

There were no wound issues or other significant complications. One patient developed an increase in his phantom limb pain after surgery. The pain diminished to preoperative levels within 2 months. The patients were able to return to using a conventional myoelectric prosthesis within 1 to 2 months after surgery, when surgical edema subsided. The electrodes of the prosthesis where adjusted to use the intact lateral biceps and long head of the triceps muscle segments. Of note, the surface electromyographic signals were noticeably larger in most of the patients after surgery because of the removal of the subcutaneous adipose tissue.

The goal for each patient was the ability to detect distinct muscle segments under cortical control of the musculocutaneous, median, proximal radial, and distal radial nerves. The goal in each operation was to leave undisturbed innervation to the lateral head of the biceps (musculocutaneous) and the medial head of the triceps (proximal radial). Two nerve transfers were performed on each of six patients, for a total of 12 nerve coaptations, and each of these transfers yielded electromyographic signals separate and distinct from adjacent musculature. Therefore, after the targeted reinnervation procedure, each of the patients now had four separate muscles controlled by four different nerves that could be used for signal detection and prosthetic limb control. Furthermore, the electromyographic signal from these muscles was physiologically correlated to the functions controlled in the prosthesis, making operation more intuitive for the amputee.

Initial signs of reinnervation were detected between 8 and 12 weeks. Good muscle recovery that produced electromyographic signals of sufficient magnitude to operate a prosthesis developed within 5 to 7 months. All of the patients were successfully fitted with new prostheses. Small modifications to the programming of the Boston Digital Arm prostheses (Liberating Technologies, Holliston, Mass.) were made so that all electromyographic signals could be recorded and used to simultaneously operate the elbow and hand. For these patients, the remaining lateral biceps and medial triceps were used to provide electromyographic control of only the motorized elbow. Electromyographic signals from the medial biceps (reinnervated with the median nerve) were used for hand closing, and the electromyographic signals from the lateral triceps (reinnervated by the distal radial nerve) were used for hand opening. Their prostheses also contained a motorized wrist rotation unit that was operated with a pull switch built into the shoulder harness of the prosthesis (a standard control mechanism for motorized prostheses). Thus, each of the motors of the prostheses had separate controllers and could be operated independently.

After prosthetic fitting, occupational therapy was required to train the patients to use the new control systems. All patients had good basic control of the prostheses and were able to operate the elbow and hand within a few sessions. After a few weeks of therapy and use, they were all able to demonstrate simultaneous control of the elbow and hand; however, they usually operated only one joint at a time for practical use.

Comparisons between preoperative and postoperative function have been documented previously for patients before and after targeted reinnervation using identical prostheses.4 Illustrative videos of postoperative function in three of these six patients are presented. The videos demonstrate the smooth coordination of both prosthetic elbow and terminal device function after targeted reinnervation (see Videos, Supplemental Digital Content 1, 2, and 3, which demonstrate prosthetic function after targeted reinnervation in patients A, B, and C,,, and, respectively).

Supplemental Digital Content 1.
Supplemental Digital Content 1.:
This video demonstrates prosthetic function after targeted reinnervation in patient A,
Supplemental Digital Content 2.
Supplemental Digital Content 2.:
This video demonstrates prosthetic function after targeted reinnervation in patient B,
Supplemental Digital Content 3.
Supplemental Digital Content 3.:
This video demonstrates prosthetic function after targeted reinnervation in patient C,


The first-generation procedure for targeted reinnervation was successful but required a long residual limb and an intact brachialis muscle. The motor branch to the brachialis was swung laterally to reach the distal radial nerve. This surgical procedure was fairly demanding, and the two nerves could only be coapted with some difficult maneuvering. In our first three patients reported, one transfer to the brachialis was not successful, as the distal radial nerve appeared scarred because of a more proximal and undetected brachial plexus injury.

Targeted reinnervation has been extremely reliable in achieving successful muscle neurotization. All 12 nerve coaptations in this study have provided independently recordable electromyographic signals. The success of this peripheral nerve surgery is attributable to two factors. First, a large proximal nerve cut back to normal-appearing fascicles is coapted to a small motor branch without any tension. Termed “hyperreinnervation,” in rats this nerve size mismatch has been shown to improve the quality and dependability of reinnervation.8 Second, the reinnervated muscles just need to act as a nerve signal amplifier and produce an electromyographic signal. Strength and muscle excursion are not required. In standard nerve transfer surgery, the quality of muscle strength and movement postoperatively are much more stringent criteria for success.

The major improvement in this described targeted reinnervation procedure for transhumeral amputees is the handling of the distal radial nerve. Less nerve mobilization is required for performance of distal radial coaptation. The total movement of the distal radial nerve to reach the motor nerve of the lateral triceps can be as little as 1 cm. A second improvement is the initial raising of a proximally based, U-shaped adipofascial flap. The raised flaps improve the exposure for identification of the muscle raphes. Placement of the flaps between the muscle bellies decreases the chances for aberrant reinnervation. Finally, the flaps help to improve later electromyographic signal detection by greater separation of the muscle bellies from each other in an axial direction, and by surgically thinning the soft tissue overlying the muscle in the radial direction.

The anatomy of the radial nerve in the upper arm is confusing for a multitude of reasons. Other than humerus fractures and tumor excisions, this region of the body does not routinely undergo surgical exploration. The radial nerve is deep against the humerus, rendering it relatively inaccessible by the covering triceps. Finally, the spinning of the radial nerve around the triceps (as caused by the twisting of the arm in embryonic development) makes the course of the nerve difficult to conceptualize. A drawing with the arm “untwisted” shows that the radial nerve actually travels in a linear path down the posterior aspect of the arm. The motor nerves exit on the inferior aspect of the radial nerve as they travel to innervate the triceps (Fig. 5).

Fig. 5.
Fig. 5.:
Anatomical diagram of the course of the motor nerves to the triceps seen with the arm abducted and externally rotated to “straighten” the course of the radial nerve.

The motor nerve to the long head of the triceps is the first to leave the radial nerve and travels separately to its muscle belly. Its relative proximity to the axillary nerve has allowed it to be a useful nerve transfer for brachial plexus injuries.9 In these cases, the motor nerve to the lateral head stays intact to maintain active elbow extension.

The Boston Digital Arm was used for these patients because it was the only myoelectric prosthesis that allowed four separate electromyographic inputs for simultaneous control of prosthetic elbow and hand function. Because of the successes of this surgery, with time, we expect that additional companies will provide this functionality in their next generation of myoelectric arms.


The authors present their current surgical technique for targeted reinnervation for the transhumeral amputee. Refinements in technique are possible because of improved understanding of the radial nerve motor branches to the long and lateral heads of the triceps. The procedure can now be performed on patients with amputations at the midhumeral level but who still have good triceps and biceps function. Successful nerve transfers in six patients are attributed to large numbers of viable motor axons at the nerve coaptation site and to the relatively low requirement for active electromyographic signals in the target muscle.


1.Hijjawi J, Kuiken TA, Lipschutz RD, Miller LA, Stubblefield KA, Dumanian GA. Improved myoelectric prosthesis control accomplished using multiple nerve transfers. Plast Reconstr Surg. 2006;118:1573–1578.
2.Kuiken TA, Dumanian GA, Lipschutz RD, Miller L, Stubblefield KA. The use of targeted muscle reinnervation for improved myoelectric prosthesis control in a bilateral shoulder disarticulation amputee. Prosthet Orthot Int. 2004;28:245–253.
3.Kuiken TA, Miller LA, Lipschutz RD, et al. Targeted reinnervation for enhanced prosthetic arm function in a woman with a proximal amputation: A case study. Lancet 2007;369:371–380.
4.O’Shaughnessy KD, Dumanian GA, Lipschutz R, Miller L, Stubblefield K, Kuiken TA. Targeted reinnervation to improve prosthesis control in transhumeral amputees: A report of three cases. J Bone Joint Surg (Am.) 2008;90:393–400.
5.Kuiken TA, Lowery MM, Stoykov NS. The effect of subcutaneous fat on myoelectric signal amplitude and cross-talk. Prosthet Orthot Int. 2003;27:48–54.
6.Lowery MM, Stoykov NS, Kuiken TA. Independence of myoelectric control signals examined using a surface EMG model. IEEE Trans Biomed Eng. 2003;50:789–793.
7.Witoonchart K, Leechavengvongs S, Uerpairojkit C, Thuvasethakul P, Wongnopsuwan V. Nerve transfer to deltoid muscle using the nerve to the long head of the triceps, part I: An anatomic feasibility study. J Hand Surg (Am.) 2003;28:628–632.
8. Kuiken TA, Childress DS, Rymer WZ. The hyper-reinnervation of rat skeletal muscle. Brain Res. 1995;676:113–123.
9. Leechavengvongs S, Witoonchart K, Uerpairojkit C, Thuvasethakul P. Nerve transfer to deltoid muscle using the nerve to the long head of the triceps, part II: A report of 7 cases. J Hand Surg (Am.) 2003;28:633–638.

Supplemental Digital Content

©2009American Society of Plastic Surgeons