Lack of shoulder mobility and persistent shoulder instability are common and debilitating problems in adult patients with brachial plexus injury. Upper-extremity function is substantially hindered by the lack of shoulder mobility even if the elbow, wrist, and fingers are functional. Recovery of rotator cuff and deltoid muscle function is often incomplete or absent in patients with brachial plexus injury, regardless of whether or not nerve reconstruction was performed. The loss of function is usually manifested by lack of shoulder external rotation and poor abduction and flexion1,2. Traditionally, reduction of inferior shoulder subluxation, recovery of active elbow flexion, and improvement of pain are interpreted as clinical success. However, most patients with brachial plexus injury maintain an internally rotated shoulder position (“hand on belly”) because the lack of shoulder external rotation3 limits the functional use of the upper extremity.
Options for improving shoulder function and stability include fusion and tendon transfers. Shoulder fusion reliably stabilizes the glenohumeral joint; however, it is irreversible and shoulder motion relies entirely on scapulothoracic motion, which many patients with brachial plexus injuries do not have4-7. Fusion is consequently contraindicated in patients with limited scapulothoracic motion or paralysis of the periscapular muscles4,5. Saha8 and Goldner9 both recommended tendon transfer for a paralytic shoulder and used shoulder fusion only as a last resort.
The purpose of the present study was to evaluate the early outcome of shoulder tendon transfer in patients with brachial plexus injury and to determine the factors associated with favorable outcomes. Our hypothesis was that tendon transfer about the shoulder in patients with brachial plexus injury resulting in a paralytic shoulder improves shoulder function, particularly external rotation.
A retrospective review of all adult patients with traumatic brachial plexus injury who underwent shoulder reconstructive surgery including transfer of the lower portion of the trapezius muscle was performed. The inclusion criterion for this procedure was persistent shoulder dysfunction either two years after nerve reconstruction or more than one year after the original injury. Exclusion criteria for the procedure included active medical conditions that precluded clearance for anesthesia, active infection, complete periscapular muscle paralysis, and unwillingness to follow the necessary postoperative immobilization and rehabilitation protocol.
Fifty-two adult patients with brachial plexus injury treated with transfer of the lower portion of the trapezius, with or without additional tendon transfers, between October 2007 and June 2009 were identified. This cohort included forty-one men and eleven women with a mean age of twenty-seven years (range, twenty to sixty-nine years). The right upper extremity was involved in twenty-two patients (and was dominant in twenty-one), and the left upper extremity was involved in thirty (and was dominant in three). Twelve patients had a C5-6 injury, twenty-two had a C5-7 injury, five had a C5-8 injury, and thirteen had a C5-T1 injury. Thirty-six patients had undergone a previous unsuccessful shoulder nerve reconstruction as part of the global brachial plexus reconstruction. The remaining sixteen patients had not undergone nerve reconstruction involving the shoulder. Most of the injuries resulted from accidents involving motorcycles (twenty patients), other motor vehicles (eighteen), snowmobiles (four), and personal watercraft such as Jet Skis (four). The remaining patients were injured during sports (two), were struck by a heavy object (two), or fell from a height (two). Thirty-seven patients were college students, and eleven of these were involved in competitive sports activities. Three of the remaining patients were office workers, two were not employed, four were construction workers, two were truck drivers, two were soldiers, and two were farmers.
Preoperative Evaluation and Planning
Patient information, including the medical history and the results of the preoperative and postoperative physical examinations performed by the senior author (B.E.), was obtained retrospectively from the available medical records. The shoulder examination was used to determine the presence of inferior glenohumeral subluxation, stiffness, and the range of motion. The severity of shoulder stiffness was determined by measuring passive shoulder motion in all directions and comparing the results with those in the contralateral, normal shoulder. Capsular release might be necessary if a difference of >30° existed between the injured and the normal shoulder in any direction, especially with external rotation. Inferior shoulder subluxation was evident in all patients with complete atrophy of the deltoid on clinical examination. The shoulder was grossly inferiorly subluxated because of the lack of the dynamic function of the deltoid and rotator cuff muscles. In most instances, there was no need to pull down on the arm to determine the degree of inferior subluxation because the shoulder was spontaneously inferiorly subluxated. The examiner could easily reduce the shoulder by applying an upward-directed force on the humerus through the flexed elbow. One patient had traumatic anteroinferior dislocation of the shoulder, and the glenoid and humeral head could be easily palpated because of substantial atrophy of the paralyzed muscle. Palpation of the humeral head showed it to be situated in an anteroinferior position on the glenoid surface, and this was confirmed by an axillary shoulder radiograph.
Most patients with brachial plexus injury have paralyzed deltoid and rotator cuff muscles. However, if any trace of active motion in the shoulder was detected, a detailed examination of the rotator cuff and deltoid muscles was performed10. In addition, planning for potential tendon transfer included a detailed examination of the periscapular muscles to evaluate their availability and strength.
Radiographs were obtained to assess the congruency of the glenohumeral joint, the presence of osteoarthritis, and the presence of any previous fracture or implants. Patients with brachial plexus injury may have signs of cartilage wear in the glenohumeral joint, yet their symptoms appear to be the result of the inferior glenohumeral joint subluxation (Fig. 1). The patients describe an inferior pulling pain that is not localized to the area of the joint line, but rather occurs more globally about the deltoid area, and that improves or resolves with manual reduction of the glenohumeral joint. For this reason, tendon transfers are not contraindicated in the presence of glenohumeral joint wear.
Types of Tendon Transfers
Historically, transfer of the upper portion of the trapezius muscle to the proximal end of the humerus is the most common transfer performed for adult patients with brachial plexus injury (including patients who undergo spinal accessory nerve transfer as part of the brachial plexus reconstruction)8,11. In most of these patients, the upper portion of the trapezius is preserved and hypertrophies over time12. Most reports indicate that this procedure results in only modest improvement of shoulder abduction and/or flexion, and shoulder external rotation is not expected to improve with this transfer13,14.
We recently described the use of a single tendon transfer involving the lower portion of the trapezius muscle to improve shoulder external rotation, and this procedure had a promising outcome3,15. However, we later began to perform multiple tendon transfers to restore rotator cuff and deltoid muscle function with the aim of improving the global function of the shoulder. In this case, our preference was to transfer the levator scapulae to the supraspinatus, the lower portion of the trapezius to the infraspinatus, the upper portion of the serratus anterior to the subscapularis, the teres major to the teres minor, and the upper and middle portions of the trapezius to the proximal end of the humerus to restore the function provided by the middle portion of the deltoid (abduction). If the latissimus dorsi was available for transfer, it was used in a pedicled muscle transfer to the anterior aspect of the shoulder to restore anterior deltoid function.
Five of the patients in the present study underwent a single transfer of the lower portion of the trapezius muscle, and the remaining thirty-nine underwent multiple tendon transfers. All of the multiple tendon transfers included transfer of the upper and middle portions of the trapezius to the proximal lateral end of the humerus. In addition, twenty-six transfers of the levator scapulae to the supraspinatus, ten transfers of the upper portion of the serratus anterior to the subscapularis insertion, eleven bipolar transfers of the latissimus dorsi and one bipolar transfer of the pectoralis major to the anterior portion of the deltoid, and nine transfers of the teres major to the teres minor were performed.
Six patients had previously undergone spinal accessory nerve transfer, which resulted in paralysis of the lower portion of the trapezius in all six patients and paralysis of the middle portion of the trapezius in three. Restoration of shoulder external rotation in these patients was achieved by performing a flip transfer of the major and minor rhomboids to the infraspinatus in four patients and by transfer of the remaining trapezius to the infraspinatus in two patients. These six patients were excluded from the outcome measurements so that the results would be focused on the patients who underwent transfer of the lower portion of the trapezius muscle with or without other transfers.
Surgical Technique
The details of the surgical technique for transfer of the lower portion of the trapezius muscle through either a single extended incision15 or a minimally invasive two-incision method have been previously reported16. Essentially, the lower portion of the trapezius is exposed and the interval between the lower and middle portions of the trapezius is identified and developed. The tendinous insertion of the lower portion of the trapezius is detached from the medial aspect of the spine of the scapula. The tendon is lengthened by attachment of an Achilles tendon allograft fixed to the infraspinatus insertion.
The main purpose of additional tendon transfers is to attempt to restore rotator cuff and deltoid functions. The additional transfers may include the levator scapulae to the supraspinatus, the upper portion of the serratus anterior to the subscapularis, the teres major (if available) to the teres minor, the latissimus dorsi (if available) to the anterior portion of the deltoid, and the upper and middle portions of the trapezius to the deltoid.
The standard patient positioning during surgery is the semilateral position, with the operative side facing up (Fig. 2). An inverted U-shaped incision is created beginning midway between the spine and the medial aspect of the scapula, just distal to the spine of the scapula; this is extended proximally, laterally, and then distally over the lateral aspect of the arm to the level of the deltoid insertion (Fig. 3). The lower portion of the trapezius is first detached from the medial aspect of the spine of the scapula, elongated with an extension of the posterior deltoid fascia, and dissected medially, to the level of its origin, from the spine (Fig. 4). The spinal accessory nerve is identified and protected throughout (Fig. 5). The upper and middle portions of the trapezius are harvested after the lower portion of the trapezius is detached. The most anterior fibers of the upper portion of the trapezius are detached from the clavicle and an acromial osteotomy is performed, leaving the upper trapezius attachment (Fig. 5). Dissection of the trapezius is then performed more medially to allow full mobilization of this portion of the muscle. The undersurface of the acromion is debrided to expose bleeding cancellous bone. The muscle is then pulled medially and the shoulder is pulled laterally, which allows full exposure of the superomedial corner of the scapula. This corner of the scapula includes the attachment of the levator scapulae (posteromedially) and the upper two slips of the serratus anterior muscle (anterolaterally) (Fig. 6). These two muscles are detached with a small osseous insertion and are mobilized for transfer (Fig. 7). The osseous insertion of the levator scapulae is attached to the supraspinatus insertion with transosseous sutures. The osseous insertion of the upper portion of the serratus anterior is attached to the subscapularis insertion with transosseous sutures. The teres major, if available, is exposed and its tendon is detached from its insertion on the humerus, mobilized, and inserted on the teres minor tendon (Figs. 8 and 9). Once these deeper transfers to the rotator cuff muscles are completed, transfer of the upper and middle portions of the trapezius to the proximal end of the humerus is performed. The shoulder is placed in approximately 80° of abduction during this transfer. It is important not to abduct the shoulder excessively in patients who have undergone previous intercostal nerve transfer to prevent potential injury to the transferred nerves. The proximal end of the humerus is exposed, and the area distal to the greater tuberosity is debrided to expose bleeding bone. The osseous insertion of the trapezius is then transferred to the prepared osseous area and fixed in placed with two or three partially threaded 3.5-mm cortical screws with washers. Mobilization and transfer of the latissimus dorsi to restore anterior deltoid function is performed as described by Itoh et al. (Fig. 10)17. All surgical procedures in the present series were performed by a single surgeon (B.E.).
Postoperative Management
The shoulder is held in abduction and external rotation for eight weeks postoperatively with use of either a brace or a shoulder spica cast. The degree of abduction and external rotation is determined at the time of the surgery. Progressive adduction of the shoulder is then permitted over the following two weeks. Active-assisted shoulder range-of-motion exercises are performed for six weeks, followed by active shoulder motion and swimming exercises for six weeks. Strengthening exercises are then started, and the patient is allowed unrestricted activities at six months postoperatively.
At the time of the latest follow-up, the patients were asked about their overall satisfaction with the shoulder surgery and whether they would have undergone it again. Active shoulder range of motion was measured and compared with the preoperative findings. The hand-on-belly position was considered to represent shoulder rotation of –80° (internal rotation of 80° from the neutral shoulder position). We considered the glenohumeral joint to be stable if the tendon transfer reduced the inferior glenohumeral subluxation, as determined both subjectively and objectively. Subjectively, patients were asked whether the painful feeling associated with the inferior glenohumeral subluxation had improved or resolved. Objectively, physical examination was performed to determine the reduction of the shoulder by gross inspection, palpation, and gentle inferior pulling on the humerus to evaluate inferior glenohumeral subluxation. In the one patient with true anteroinferior glenohumeral dislocation, additional abduction-external rotation testing of the shoulder was performed to determine its stability. Preoperative and postoperative radiographs were also compared to determine the reduction of the inferior subluxation of the glenohumeral joint (Figs. 1 and 11).
Pain (on a visual analog scale ranging from 0 to 10, with 0 indicating no pain and 10 the worst possible pain), the Subjective Shoulder Value (SSV)10, and the Disabilities of the Arm, Shoulder and Hand (DASH) score18 were assessed preoperatively and postoperatively.
Statistical Analysis
A paired Student t test was used to compare the preoperative and postoperative outcome measures (DASH, SSV, and pain score). Regression analysis was performed to evaluate the difference in outcome on the basis of differences in age, vertebral level of the injury, and body mass index (BMI); this was done in an attempt to control for potential confounding by these factors. A p value of ≤0.05 was considered significant.
Source of Funding
No external source of funding was used for this study.
All patients were evaluated at a mean of nineteen months (range, twelve to twenty-eight months) postoperatively. All patients reported that the shoulder felt stable and that pain had improved or resolved after the tendon transfer reconstruction (Table I). Also, all patients with a preoperative radiographic finding of inferior glenohumeral subluxation had reduction of the glenohumeral joint. Mean shoulder external rotation improved substantially in all patients from –80° (no external rotation, the hand-on-belly position) preoperatively to a mean of 20° (range, –20° to 40°) postoperatively (p = 0.001).
Patients who had undergone an isolated transfer of the lower portion of the trapezius had marginal but nonsignificant improvement of shoulder abduction and flexion. Patients who had also undergone additional transfers had marginal improvements of shoulder flexion, from a mean of 10° (range, 0° to 20°) preoperatively to 60° (range, 30° to 110°) postoperatively, and in shoulder abduction, from a mean of 10° (range, 0° to 20°) to 50° (range, 20° to 70°) (p = 0.01 for both). Pain improved from a mean of 6 points (range, 3 to 8 points) preoperatively to 2 points (range, 1 to 3 points) postoperatively. The DASH score also improved from a mean of 59 points (range, 41 to 88 points) preoperatively to 47 points (range, 24 to 66 points) postoperatively (p = 0.001). This difference exceeded the minimally clinically important difference for the DASH score19. The SSV improved from a mean of 5% (range, 1% to 10%) to 40% (range, 20% to 60%) (p = 0.001).
Patients who were younger than thirty-five years of age at the time of surgery had significantly better outcomes compared with older patients (p = 0.01). There was a nonsignificant trend toward better outcomes in patients younger than fifty years of age compared with older patients (p = 0.33). Patients with an incomplete brachial plexus injury had significantly better outcomes compared with patients with a complete injury (p = 0.002). The better results in patients with incomplete nerve injury were expected because of the availability of more muscles about the shoulder for transfer. There was a nonsignificant trend toward worse outcomes in patients with a BMI of >35 kg/m2 compared with patients with a lower BMI and the same extent of injury.
There is little information to guide the surgeon on how to reconstruct the paralytic adult shoulder to improve function. In this study, tendon transfer was used to reconstruct the shoulder of patients with brachial plexus injury. This type of reconstruction did improve shoulder function, specifically external rotation. All patients who had transfer of the lower portion of the trapezius to the infraspinatus had improvement of shoulder external rotation. However, improvement of shoulder flexion and abduction was only marginal. Greater functional improvement was associated with greater availability of muscles for transfer in patients with less severe brachial plexus injury.
When reconstruction involving tendon transfer is contemplated in patients with brachial plexus injury, a thorough understanding of the basic biomechanics of the shoulder and accurate determination of the strength of each individual muscle available for transfer are critical. There are relatively few functional muscles available for transfer about the shoulder after brachial plexus injury. The function of the trapezius, levator scapulae, and rhomboids has been reported to be preserved or recovered in 96% of patients with brachial plexus palsy20. The trapezius innervation is from the spinal accessory nerve, whereas the rhomboids and levator scapulae are innervated by the dorsal scapular nerve with contributions from more proximal cervical nerves20. Division of multipennate muscles (such as the trapezius, serratus, and pectoralis major muscles) into their functional subcomponents, as described by Herzberg et al.21, can provide additional tendon transfer options.
Several biomechanical studies have indicated the importance of the rotator cuff as the primary dynamic stabilizer of the glenohumeral joint, and rotator cuff function is very important for restoration of the force couple of the shoulder that allows the deltoid muscle to function optimally22. For this reason, it has been our preference to restore both the rotator cuff force couple (the subscapularis and infraspinatus) and deltoid function. Effective shoulder abduction by the deltoid muscle requires that both the subscapularis and the infraspinatus be functioning appropriately, because these muscles are more important contributors to abduction than the supraspinatus and they provide additional rotational control.
Trapezius transfer, which is the most common transfer method described in the literature, was described by Hoffa in 1891, followed by Lewis in 1910 and Lange in 1911. Elongation of the trapezius tendon with fascia lata led to poor results, which were attributed to stretching and scarring8,14. The transfer of the trapezius muscle, along with its acromial insertion, to the proximal end of the humerus was reported by Bateman in 195511 and later modified by Saha8, whose name is now associated with the operation8,14. However, Saha originally defined three basic shoulder functions that needed to be reconstructed with available nonparalyzed local muscles. Thus, he was an advocate of multiple tendon transfers in an attempt to restore normal shoulder function, and rarely recommended trapezius transfer in isolation. However, most of his patients were younger than the patient population in our study and had shoulder weakness or paralysis as a result of poliomyelitis rather than traumatic brachial plexus injury.
Despite Saha’s recommendation for multiple tendon transfers to restore shoulder function in the paralytic shoulder of patients with polio, most of the published reports dealing with the Saha technique have actually involved isolated transfer of the upper portion of the trapezius8,23-25. Kotwal et al. reported that 60% of patients treated with the Saha technique had good results after transfer of the upper portion of the trapezius, and 19% had a poor outcome with persistent subluxation of the glenohumeral joint23. Aziz et al. reported a reduction in shoulder subluxation in all patients treated with transfer of the upper portion of the trapezius, with a mean gain of 40° of abduction and flexion24.
Rühmann et al. reported the outcome of transfer of the upper portion of the trapezius in adult patients with brachial plexus injury13,14. In their experience, the best results were achieved in patients with a functioning biceps, coracobrachialis, triceps, or pectoralis major, or any combination of these muscles. The better outcome in this group of patients was most likely related to the greater number of functioning muscles in this less severely injured patient cohort. However, the lack of shoulder external rotation was not addressed in that study or in any of the studies mentioned in the previous paragraph.
Tendon transfers involving the latissimus dorsi and the pectoralis major have also been reported in the literature26. Direct transfer of the latissimus dorsi tendon to the infraspinatus or supraspinatus is contraindicated in patients with a paralytic shoulder3,27. In the absence of subscapularis and deltoid function, the transferred latissimus dorsi functions as an inferior translator and extensor of the shoulder rather than as an external rotator. For this reason, transfer of the latissimus dorsi muscle as a pedicled flap to the anterior aspect of the shoulder is recommended to improve shoulder flexion17. Itoh et al. reported that the use of a pedicled latissimus dorsi flap to restore shoulder flexion in ten patients achieved good results17. All patients had a good shoulder contour, six had active flexion of >90°, and all but one had muscle power graded as 4/5. De Smet28 and Ferrier et al.29 used the same technique with similarly good results. Most of the patients in these three studies had isolated paralysis of the deltoid muscle and a partially innervated or completely normal rotator cuff.
Hou and Tai reported on seven patients with deltoid muscle paralysis treated with transfer of the upper portion of the pectoralis, either alone (four patients) or combined with transfer of upper portion of the trapezius (three patients)26. Isolated transfer of the upper portion of the pectoralis restored a mean of 40° of abduction. When transfer of the upper portion of the trapezius was added, however, abduction improved to 70° to 90°, with forward flexion of 60° to 150°. Lin et al. reported similarly good results in patients with brachial plexus injury30.
In the present study, we attempted restoration of rotator cuff function in addition to deltoid muscle function through multiple tendon transfers. The short-term outcome was very promising, with stabilization of the glenohumeral joint and improvement of external rotation in all patients. The improvement of shoulder abduction and flexion was marginal and more dependent on the level of injury and the number of muscles available for transfer.
There are several limitations to this study. First, it was retrospective in nature. Second, the population of patients with brachial plexus injury was heterogeneous, with tremendous variation in the nerve injuries, concomitant injuries to both soft tissue and bone, and a variable number of tendon transfers. This made it difficult to group patients for comparative studies. Third, the number of patients was relatively limited and there was no control group that underwent either no surgery or different types of tendon transfers. Finally, longer follow-up is needed to determine the long-term outcome of this type of reconstruction in paralytic shoulders.
Nevertheless, this study has demonstrated that shoulder reconstruction using these novel tendon transfers is an acceptable treatment option in patients with brachial plexus injury who have persistent shoulder dysfunction. Although improvement of shoulder external rotation is to be expected in all patients, only marginal improvements of shoulder flexion and abduction are to be expected. Better outcomes are expected in younger patients and in patients with fewer injured nerve roots and thus more muscles available for transfer.