Long Thoracic Nerve Decompression and Transfer: Single-Center Case Series of Clinical and Electromyographic Outcomes

Abstract

Background:

Long thoracic nerve (LTN) decompression is considered in recalcitrant scapular winging secondary to chronic LTN palsy. Nerve transfer, typically from the thoracodorsal nerve (TDN), is suggested if, despite adequate decompression, intraoperative nerve stimulation demonstrates no improvement. Literature concerning transfer is scarce. To evaluate and compare these 2 procedures’ clinical and electrical outcomes, we performed a single-center, retrospective case series of all LTN decompression patients with or without transfer for chronic LTN palsy, examining postoperative adapted Medical Research Council (MRC) grades as a primary and electromyography (EMG) stimulation thresholds as a secondary outcome.

Methods:

We identified 11 decompression-only and 6 patients undergoing additional transfer over an 8-year period, confirmed with preoperative serratus anterior EMG. Decompression involved proximal and distal neurolysis, with transfer, typically the lateral branch of TDN, reserved for irresponsiveness to intraoperative stimulation following decompression. Adapted pre- and postoperative serratus anterior MRC values were evaluated using a 2-tailed Student t test.

Results:

Preoperative adapted MRC grades for all 17 patients was 0; at median 12-month follow-up, this reached 3. The decompression-only preoperative median was 0 and final grade 3; for the transfers, these were 0 and 3.5 respectively, which were insignificantly different. However, time to first recovery, the first clinical evidence of serratus anterior contraction, was significantly different between the decompression-only cohort, at 3 weeks, and transfer, 7 months. Preoperative EMG thresholds were 1.0 mA pre- and 0.1 mA postoperatively; they did not impact final adapted MRC grades.

Conclusions:

We conclude nerve transfers achieve comparable long-term outcomes where decompression alone did not improve intraoperative nerve stimulation.

Introduction

While a majority of long thoracic nerve (LTN) palsies recover spontaneously, with a mean duration of 16 months, at least a quarter of patients experience little relief with conservative management and require surgical intervention. ¹

Surgical interventions preserving the serratus anterior muscle include decompression/neurolysis alone or in combination with nerve transfer. Tendon transfer surgery can be employed, but does not replicate the role of the serratus anterior in producing synchronized scapulothoracic motion. Noland et al ² organize these options into a novel treatment algorithm, recommending serial monthly electromyography (EMGs) for 6 months postinjury and offering surgery to those with no clinical or electromyographic improvement at 6 months. Firstly, a proximal LTN decompression at the level of the middle scalene is followed by intraoperative nerve stimulation. Those without response undergo distal decompression at the chest and, if still unresponsive, a thoracodorsal to LTN supercharge end-to-side transfer. Those that respond to stimulation are not immediately offered distal decompression nor nerve transfer in the same sitting, but clinically followed up for 3 months and offered transfer if no clinical recovery is evident.

Existing literature concerning outcomes is mainly limited to case series primarily investigating LTN decompression, utilizing most commonly distal decompression, proximal decompression or a combination. Accepting heterogeneity in definitions, a multitude of case series in various institutions consistently report long-lasting “excellent” or “good” outcomes in over 80% of their cohorts, together with dramatic improvements in active shoulder flexion and abduction to near-normal angles, plus resolution of scapular winging.^{3 –6} These findings are similar whether proximal or distal decompression is performed, and the case series indicate a more favorable prognosis if performed within twelve months of injury.

By contrast, case series concerning LTN transfer is sparse. Uerpairojkit et al ⁷ describe 5 patients with traumatic C5 and C6 nerve root avulsions undergoing thoracodorsal to LTN nerve transfer as part of their brachial plexus reconstructive surgery. Two excellent, 2 good, and 1 fair result were reported at a mean of 28 months’ follow-up. Promising results were also reported by Noland et al, ² with full active shoulder flexion regained in their 3 patients at 2.5 months postoperatively.

Our case series aims to add to the growing body of evidence concerning the long-term outcomes of LTN neurolysis and, in particular, report on our center’s experience with LTN nerve transfer surgery using an adapted postoperative Medical Research Council (MRC) grading as our primary outcome measure. Finally, we also evaluate the preoperative prognostic utility of EMG and share our preoperative and intraoperative parameters required for nerve stimulation to guide future practice.

Materials and Methods

This is a single-center consecutive retrospective case series evaluating, in adults with chronic LTN palsy, the results of LTN decompression, with or without nerve transfer, on postoperative adapted MRC findings and time of first improvement. This case series has been reported in line with the PROCESS 2020 Guidelines. ⁸ All patients were treated at our home institution with date limits set between 2012 and 2019, with patients identified through the hospital IT database system coding all nerve transfer operations, followed by manual extraction and anonymization of cases involving LTN decompression or transfer.

Inclusion criteria were any adult patient with a chronic LTN palsy confirmed through nerve conduction studies (NCS) and EMG analysis, independent of etiology, with symptomatic scapular winging undergoing LTN decompression with or without nerve transfer. Exclusion criteria included pediatric patients, patients with acute LTN repair, patients undergoing concomitant tendon transfer as part of scapular winging and patients unable to cooperate with clinical examination.

All patients underwent preoperative clinical examination to grade serratus anterior function on an adaptation of the Medical Research Council (MRC) scale used by our unit as follows. In the sitting position, patients flexed their arm to approximately 130° and the clinician forced the scapula into retraction and the arm into extension. The patient was asked to resist the clinician, who graded the serratus anterior power as described on Table 1. ⁹ Preoperative electrical evaluation generally formed part of a wider brachial plexus evaluation beyond solely the LTN. At the discretion of the individual neurophysiologist, patients were positioned sitting or lateral. At least 2 prominent digits of serratus anterior were electrically evaluated, identified through palpation and occlusion of adjacent intercostal spaces, and concentric EMG needle placement into the belly along the mid-axillary line. Patients were asked to perform voluntary supramaximal contraction of serratus anterior by pushing their hand against a wall. The neurophysiologists defined active denervation as the presence of fibrillation potentials and positive sharp waves on needle insertion and during voluntary contraction, and chronic denervation as presence of polyphasic motor units and diminished recruitment. As meaningful amplitudes of LTN in NCS are difficult to obtain, we chose to document the threshold to stimulate serratus anterior in milliamps, which was felt to be most objective and least susceptible to interobserver variation. To define threshold required to stimulate serratus anterior, either surface or needle stimulation of the LTN, at Erb’s point or more distally, at the discretion of the individual neurophysiologist. Incremental increases in current were applied to the LTN until definitive visual contraction or twitching of serratus anterior was visualized; the minimum current in milliamps represented the threshold required to stimulate. Clinical documentation was used to retrieve adapted MRC grades and diagnostic results.

Table 1.

The Adapted MRC Grading Employed by Our Unit to Quantify Serratus Anterior Muscle Activity.

Adapted MRC grade	Description
5	Holds against maximal resistance
4	Holds against moderate resistance
3	Cannot hold against resistance but moves through full range of motion without winging
2	Scapular winging observed when patient adopts test position of scapular retraction and arm flexion.
1	Contractile activity observed
0	No contractile activity observed

Note. MRC = Medical Research Council.

Operative Technique

All procedures were performed using microscopic magnification by a consultant specialized in nerve surgery. The point of maximal Tinel’s was marked out preoperatively. Intravenous antibiotics were administered under the operating surgeon’s discretion. Under general anesthetic, patients were positioned in supine position with a standard anterior or mid-axillary line longitudinal incision. The LTN was identified and protected with vascular loops, with its individual branches isolated. Intra-operative nerve stimulation was performed prior to decompression to identify and record minimum threshold current amplitudes required for muscular contraction. Thereafter, distal LTN neurolysis was performed, commencing with serratus anterior and neurolyzing the segmental branches that innervate each slip of muscle. Care was taken to look for points of tether or compression, particularly around the Tinel’s point(s) marked out preoperatively. The neurolysis was extended proximally into the axilla. If, despite adequate neurolysis, no stimulation was elicited, the decision to proceed to nerve transfer was made.

Where nerve transfer was then performed, the donor nerve of choice was the lateral branch of the thoracodorsal nerve, identified through the same approach and running almost parallel to the LTN on the deep surface of the latissimus dorsi muscle (technique first described in 2009 by Uerpairojkit et al). ⁷ Provided sufficient amplitude was achieved on nerve stimulation, the lateral branch of the thoracodorsal nerve was released by neurolysis, transferred and coapted with 8-0 or 9-0 Nylon under microscope magnification with fibrin glue augmentation (Tisseal, Baxter Inc.)

Post-Operative Care

Patients returned for regular in-person follow-up appointments, where clinical adapted MRC grading of muscle power was regularly assessed and recorded, forming the basis of our data collection. We defined first clinical improvement as the first clinical evidence of serratus anterior contraction elicited clinically.

The stages of rehabilitation following motor nerve transfer surgery have been previously described by our department. ¹⁰ In summary, the shoulder was protected in a sling with elbow and wrist active and passive exercises until wound review, generally between 1 and 2 weeks following surgery. Both cohorts were advised on scar massage. Once wound healing was verified, the simple decompression cohort was permitted range of motion as tolerated to avoid nerve tethering. To prevent inadvertent disruption to the coaptation site, the transfer cohort were permitted initial passive shoulder and elbow range of motion exercises under the guidance of a physiotherapist. The first sign of re-innervation is suggested by tenderness on deep muscle palpation; once this is elicited, typically but variably in 8-12 weeks postoperatively, gravity-assisted exercises are commenced; once adapted MRC stage 2 power is reached, gravity-eliminated exercises can then be introduced, with constant visual feedback to allow patients to activate the recipient muscle using the donor nerve and to avoid persistent activation of periscapular muscles. All patients were advised clinical scapular winging will generally persist until good control has been achieved.

Data Handling and Statistical Analysis

On confirming a patient meeting inclusion and exclusion criteria, baseline data (age at surgery, gender, mechanism of injury, concomitant injuries, comorbidities, time from injury to surgery) and surgery type (neurolysis only vs transfer) were extracted. The clinical correspondence in the same database contained both pre- and intraoperative EMG findings, namely, active versus chronic denervation observed in serratus anterior, the threshold required for serratus anterior stimulation, in mA, plus preoperative and postoperative adapted MRC grades, scaled from 0 to 5, including follow-up times, in months.

Missing data were encountered in our data collection and, in view of a small sample set inherent with infrequently executed procedures plus the low level of missing data, an available-case analysis approach was performed; that is, patients missing data for a particular outcome were discounted from statistical evaluation of this particular outcome, but included in other outcomes where such data existed.

To statistically compare adapted MRC grades and electrical stimulation thresholds between cohorts (neurolysis only vs transfer; shorter (under 12 months) vs longer (over 12 months) time to surgery; and active vs chronic denervation), a 2-tailed Student t test was used with a significance level of 0.05 (GraphPad).

Results

Twenty-one patients met the inclusion and exclusion criteria. Four were lost to clinical follow-up with preoperative but no postoperative data, leaving 17 patients; 5 females and 12 males. Baseline data are summarized in Table 2. While no patient underwent additional tendon transfer/bony procedures (a specific exclusion criterion in our case series), 4 had additional nerve procedures within the same sitting: 1 cubital tunnel decompression, 1 brachialis-to-medial head of triceps nerve transfer, 1 suprascapular nerve decompression and 1 suprascapular and greater auricular nerve decompression with sural to spinal accessory nerve grafting.

Table 2.

Baseline Characteristics of Patients Included in This Case Series.

Patient demographics (total 17 patients)
Age at time of surgery in years	Median age 38 years Range 16-71 years
Gender	5 females 12 males
Mechanism of injury	6 idiopathic 4 trauma 3 iatrogenic (1 mastectomy, 1 hiatus hernia dissection, 1 laparoscopic pyeloplasty for calculus) 3 exertional 1 Parsonage-Turner syndrome
Concomitant injuries of the 4 traumatic etiology patients	1 from road traffic accident requiring liver and bowel resections No other concomitant injuries found
Comorbidities	1 type I diabetes 1 previous squamous cell carcinoma of the tongue 1 angioedema 1 previous hysterectomy The remainder had no significant comorbidity
Time from injury to surgery in months	Median 12 months Range 3-84 months
Type of surgery	11 neurolysis only 6 additional nerve transfer
Additional procedures	1 additional cubital tunnel decompression 1 brachialis-to-medial head of triceps nerve transfer 1 suprascapular nerve decompression 1 suprascapular and greater auricular nerve decompression with sural-to-spinal accessory nerve grafting

Our findings are summarized in Table 3. The median preoperative adapted MRC grading across all 17 patients was 0. The first improvement was noted at a median of 6 weeks (range 1 week-24 months). With a median time of injury to surgery at 12 months and at a median final follow-up of 12 months (range 3-60 months), the median adapted MRC improved to 3. Categorized by cohorts, the preoperative adapted MRC grading of the 11 neurolysis-only patients was 0, first improvement noted at 3 weeks and final adapted MRC was 3 at a median 12 months follow-up. Meanwhile, the preoperative adapted MRC grading of the 6 transfer patients was 0, first improvement noted at 7 months and final adapted MRC 3.5 with a median time of injury to surgery at 12 months and at a median 13 months follow-up. Accepting low patient numbers as a considerable limitation, a Student 2-tailed t test showed no significant difference between final adapted MRC grades between the 2 cohorts, but a significantly shorter time to first improvement in the decompression over the transfer cohort (P = .03).

Table 3.

Pre- and Postoperative MRC Grades and Follow-Up Periods in Months Categorized by Type of Surgery and Time to Surgery.

Patients	Number of patients	Median preop MRC grade from 0-5	Final median postop MRC grade from 0-5	Median postoperative final follow-up (months)	Median time to first improvement (months)
All patients	17	0	3	12	1.5
Categorization by surgery type
Neurolysis-only	11	0	3	12	0.75
Transfer	6	0	3.5	13	7
Categorization by time to surgery in months (all patients below neurolysis-only)
12 months	5	0	3	13	0.75
48 months	5	1	3	10.5	0.5

Note. MRC = Medical Research Council.

When categorized by time to surgery, 5 of the 11 neurolysis-only patients received their surgery at median 12 months (early group), with a preoperative adapted MRC of 0 and a postoperative of 3 at median 13 months follow-up with first improvement noted at 3 weeks. Five neurolysis-only patients received surgery at median 48 months (delayed group), and pre- and postoperative adapted MRC scores were 1 and 3. Respectively, at median 10.5 months follow-up with first improvement at 2 weeks. One of the 11 patients had no time-to-surgery data. Both early and delayed groups showed no substantial difference in time nor magnitude of improvement, but statistical analysis was not performed with a sample of this size. Similarly, 5 of our 6 transfer patients had a time to surgery under 12 months, so stratification was not undertaken in this cohort.

We were able to locate the predecompression EMG records for 16 patients and the postdecompression EMG records for 10. The intraoperative median threshold stimulation required to stimulate the serratus anterior muscle prior to decompression was 1.0 mA (range 0.1-5 mA), decreasing to 0.1 mA (range 0.04-0.5 mA) following decompression. When stratified by active versus chronic denervation cohorts, both cohorts showed broadly equal thresholds. The preoperative median adapted MRC grade for both active and chronic denervation cohorts was 0. At final follow-up, these improved to 3.5. Again, accepting low patient numbers as a considerable limitation, between the active and chronic denervation cohorts, a Student 2-tailed t test showed no significant difference. In preop adapted MRC grading, postop adapted MRC grading or time for first improvement.

Complications

Operative notes and clinic letters were reviewed retrospectively to identify any complications of surgery. No direct complications were identified with good wound healing in patients and improvement in the adapted MRC grade identified. One patient who had a nerve transfer, as part of a brachial plexus exploration (with multiple nerve transfers) did have a tight scar limiting movement with a referral made to local plastic surgery team for further management. Another patient did have persistent winging at 3 months but subsequently was lost to follow-up after this. Finally, 1 patient who had a neurolysis and suprascapular nerve neurolysis did report numbness and reduced function in the ulnar 2 digits with evidence of ulnar nerve involvement. This was evident at 2 years postoperation; however, the patient did not attend further follow-up appointments and therefore there has been no management of this to date.

Discussion

To our knowledge, this is the largest single-center case series reporting outcomes of LTN transfer.

Our case series supports the “distal first” modification of the algorithm as proposed by Noland et al, with our patient cohort exhibiting a median improvement of 3 adapted MRC grades compared with preoperative clinical findings. We believe that performing distal decompression first offers lower morbidity and greater ease compared with proximal decompression, ¹¹ reserving proximal decompression as part of axillary or even cervical exploration for nerve transfer surgery or where NCS and EMG data suggest a proximal site of LTN compression or a wider C5-7 pathology in the supraclavicular brachial plexus. We have organized our approach into a modified algorithm based on that originally proposed by Noland et al (Figure 1).

Figure 1.

Our modification of the algorithm originally proposed by Noland et al for treatment of chronic long thoracic nerve palsy. Distal decompression refers to decompression at the chest; proximal decompression refers to extension of decompression to axilla.

To clinically evaluate serratus anterior function, we use an adaptation of the Medical Research Council grading system, previously described in the literature, ⁹ as our primary clinical outcome measure, as this is clinician-directed, discrete and readily tested in everyday follow-up clinics, especially when the serratus anterior muscle location lends itself very poorly to objective evaluation with manometry. While the MRC adaptation allows specific evaluation of the abstract function of the serratus anterior muscle, given it does not cross a classical, synovial joint, we accept this is itself a source of inter-rater error. No single widely accepted grading system has been established. Previous case series mainly use improvement in shoulder range of motion, cosmetic assessment of scapular winging, presence of pain and then stratification into “excellent,” “good,” “fair,” and “poor,” which carry their own advantages and limitations. This heterogeneity in reporting hinders direct comparison between case series, but in the distal LTN decompression case series by Le Nail et al, ⁶ the authors define “excellent” as complete motor recovery, “good” as almost complete, “fair” as partial and “poor” as minimal, with 86.7% of cases classified as excellent or good with a 2-year time from injury to surgery. Taking adapted MRC grade 4 as good and adapted MRC 3 fair, only 8 of our 17 patients would be classified as “excellent or good,” a discrepancy probably attributable to the fact that nearly half of their patients had only partial paralysis of the serratus anterior muscle, clinically defined as incomplete lifting of the scapula, especially at the tip, whereas our cohort had a far more complete paralysis preoperatively with no patient scoring adapted MRC greater than 1 on preoperative review. Time to decompression before or after 12 months from presentation did not appear to substantially alter time to initial recovery nor adapted MRC grade of final recovery. Those that required additional nerve transfer achieved comparable final adapted MRC grades to the decompression-only cohort, suggesting this is a useful addition for those with refractory compression on intraoperative stimulation. However, we found that the time to first improvement is considerably longer for the transfer cohort, at 7 months, compared to the decompression-only cohort, at 3 weeks, despite previous case reports and series demonstrating minimal donor site morbidity following thoracodorsal nerve or latissimus dorsi harvest^12,13 and despite their slightly shorter time from injury to surgery. This is likely a function of their greater severity of starting compression and the known longer recovery times for nerve transfer for reinnervation. A significant limitation of our study is low patient numbers precluding reliable statistical tests, in view of the relative scarcity of this condition and procedure, but these are comparable in volume to other published case series concerning LTN intervention. Extremely small sample sizes, defined as n < 5, have been shown not to be absolute contraindications to performing t tests. ¹⁴ Similarly, adapted to our institution, the timing of follow-up of our patients was not in strict accordance to the algorithm as proposed by Noland et al Given the unique background of each patient’s LTN palsy and our geographically large catchment area, follow-up was tailored to individual patients’ needs, accounting for the substantial range in months. Despite these, substantial improvements in serratus anterior muscle activity were demonstrable, suggesting the algorithm can be reasonably adapted to each institution’s specific needs.

Our case series found no impact of the preoperative EMG findings (active vs chronic denervation) on eventual postoperative recovery of muscle function, a conclusion similar to that of Vastamäki et al, ¹⁵ where patients with complete denervation exhibited no detriment in patient-reported function, shoulder flexion and abduction, winging or strength compared with the partial denervation cohort. Indeed, the lack of prognostic utility of EMG was corroborated by Noland et al, ² attributing this to the difficulties of handling the serratus anterior muscle on EMG, a collection of multiple small, thin strips of muscle tissue that do not cross a joint to readily evaluate clinical movement and that can be significantly atrophied and displaced as part of scapular winging and moving away from a wholly EMG-based algorithm. As such, we did not exactly replicate the Noland et al algorithm of monthly preoperative EMGs for 6 months. However, we believe EMG remains important for purely diagnostic purposes and for confirmation of successful decompression intraoperatively.

Conclusion

Taken together, our data suggest that if nerve transfer in addition to decompression is ultimately necessary due to unfavorable intraoperative or postoperative responses to electromyographic stimulation following simple decompression alone, comparable final clinical results can be achieved.