Massive and irreparable posterior-superior rotator cuff tears present a difficult treatment problem. Chronic cuff tears can lead to tendon retraction and fatty atrophy of the rotator cuff muscles. Full-thickness rotator cuff tears involving more than one tendon can lead to superior migration of the humeral head in relation to the glenoid, altered glenohumeral forces, and ultimately to rotator cuff arthropathy1-4. Hemiarthroplasty5,6 components designed to compensate for rotator cuff tear arthropathy provide pain relief but are less reliable for restoring shoulder function in patients with irreparable tears of two or more tendons. Reverse total shoulder arthroplasty7-9 is indicated for low-demand elderly patients with rotator cuff arthropathy, and the procedure restores function in addition to providing pain relief. Patients with glenohumeral joint arthropathy from an irreparable rotator cuff tear are not generally treated with tendon transfer as this treatment option does not address the joint degeneration. However, one study10 demonstrated good results with use of the reverse total shoulder prosthesis in patients with an irreparable rotator cuff tear and no glenohumeral arthritis. Despite this, there are concerns regarding the longevity of reverse total shoulder implants and limited possibilities for salvage after implant failure. As a result, transfer of the latissimus dorsi muscle is considered a surgical option in the treatment of younger patients without glenohumeral arthritis and with functional deficits caused by an irreparable posterior-superior rotator cuff tear. However, the results and expectations for this procedure have been reported to be variable11–23.
Latissimus dorsi transfer provides a large, vascularized tendon that closes the cuff defect and exerts an external rotational moment, allowing more effective action of the deltoid muscle. In its native location, the latissimus dorsi muscle contributes to internal rotation, retroversion, and abduction of the shoulder joint. When the latissimus dorsi tendon is transferred to the greater tuberosity, the muscle’s internal rotator torque is removed and the function of the muscle changes into an external rotator24,25. This external rotation is accomplished by either a synergistically active tendon transfer or a tenodesis effect16. Electromyographic studies have suggested that in some cases the transfer is truly active whereas in others the patient cannot actively synchronize latissimus dorsi muscle activity with supraspinatus and infraspinatus muscle activity11,18,26. In either case, the improved balance between the anterior and posterior soft-tissue structures of the shoulder is believed to be biomechanically important to serve as a balanced fulcrum as described by Burkhart27. Furthermore, this balance across the glenohumeral joint is believed to allow the deltoid muscle to contribute to shoulder motion more effectively27.
The purpose of the present study was to critically examine the outcomes of latissimus dorsi tendon transfers for the treatment of irreparable rotator cuff tears and to present an evidence-based synthesis of the available literature. Specifically, we sought to answer the following questions:
What are the demographics of patients treated with latissimus dorsi tendon transfer surgery for an irreparable rotator cuff tear?
What are the expected functional outcomes and pain relief after latissimus dorsi tendon transfer for irreparable rotator cuff tears?
What variables portend a successful outcome?
What are the complications and what is the frequency of complications associated with this procedure?
We searched the MEDLINE and EMBASE computerized literature databases from January 1950 through March 2011. Articles were identified by using an electronic search of keyword terms and their respective combinations (Table I). Studies were included in this systematic review if they were written in the English language and all patients had been followed for a minimum of twenty-four months. We excluded review articles, case reports, technique articles without reported patient data, and studies in which the inclusion criteria were not explicit. Two authors performed the initial search, and three independently reviewed the references of the qualifying papers and selected the appropriate studies on the basis of the above criteria. If one or more authors selected a paper, it moved to the next phase. In the final phase of review (elimination by full-text review), there was no disagreement about which papers would be ultimately included.
We obtained 608 unique articles from a combined MEDLINE and EMBASE search with use of the criteria listed in Table I (Fig. 1). We initially excluded 561 articles according to the title because it indicated irrelevance to the topic in question or that the article was a review, editorial, or commentary. An additional thirty-two articles were eliminated after review of the abstract. We eliminated articles on the basis of the abstract only if it indicated that the article was an editorial, letter, or review article. If there was any question about inclusion, we reviewed the full text. Our next step of the process was to review the full text of the remaining fifteen articles, of which five failed to meet our inclusion and exclusion criteria. Thus, this systematic review left ten articles for analysis11,13,15-22. The references of these articles were manually searched for other potential articles of interest. The references were screened in the same manner as the articles from the original search. No additional articles that met the inclusion criteria were identified from the references.
None of the studies included a nonoperative control group or a comparison group that had undergone an alternate surgical treatment. Although some rudimentary statistical analyses were used, no investigators employed multivariate or stratified statistical analyses to control for confounding factors or to account for bias. We used data from each individual study to assess demographic patient characteristics. In cases in which the outcomes data were similar between studies, we pooled the results for the purposes of generating summary outcomes through the use of frequency-weighted values. Frequency-weighted values represented the mean value for a study weighted by the included number of patients from each given study. These frequency-weighted means and standard deviations were used to generate p values by using the number of “subjects” as the number of studies. The statistic generated for each variable was a value within which 95% of the studies fell. This represents the likelihood that the studies would demonstrate different results by chance alone 5% of the time (i.e., an alpha of 0.05).
A power analysis revealed that, to detect a change in motion of 10° in any plane after surgery, a minimum of eighty-five patients had to be assessed preoperatively and postoperatively, assuming equal variance, with a type-I error of 0.05 and a certainty of 0.90.
Source of Funding
There was no external funding source for this study.
Ten studies that fulfilled all inclusion and exclusion criteria were included11,13,15-22. Eight studies were published as Level-IV evidence11,13,15-17,19,20,22; one, as Level-III21; and one, as Level-II18. No study included a comparison group of patients treated without latissimus dorsi tendon transfer, and all studies evaluated a uniform cohort of patients who had undergone latissimus dorsi tendon transfer for an irreparable rotator cuff tear. Eight studies were retrospective13,15-18,20-22, and two were prospective11,19. No study was performed at more than one institution, and all ten studies were published between the years 1992 and 201011,13,15-22. The years in which the surgery was performed were recorded in all studies, and the years were between 1990 and 2005 in all11,13,15-22. Indications for tendon transfer consistently included an irreparable posterior-superior rotator cuff tear in a patient with chronic pain and limitation of shoulder function. The studies varied with regard to their inclusion11,15-17,19,20 or exclusion13,18,21,22 of patients with subscapularis insufficiency. The authors of several studies13,16,17,22 specifically noted an unsuccessful six-month period of nonoperative treatment with physical therapy and nonsteroidal anti-inflammatory medication before the tendon transfer was considered. An external rotation deficit was specifically noted as an indication in two studies16,22.
Demographics
There was a total of 309 patients (range, ten to seventy-five per study) and 313 shoulders (range, twelve to seventy-seven per study) at baseline in the ten studies11,13,15-22. The preoperative demographic data are presented in Table II. The frequency-weighted mean age (and standard deviation) was 58.7 ± 4.1 years (range, thirty-two to seventy-seven years). Nine studies included demographic data on sex, with 188 males (70.4%) and seventy-nine females (29.6%) represented11,13,15,16,18-22. Nine studies indicated whether the dominant extremity had undergone the surgery, with 210 surgical procedures (76.6%) performed on the dominant side, sixty-two (22.6%) done on the nondominant side, and two (0.8%) performed bilaterally11,13,15,16,18-22. Nine studies included data on the number of patients who had undergone previous ipsilateral shoulder surgery (100 of 242; 41.3%)11,13,16-22. Of the 100 patients who had undergone prior surgery, eighty-two had been treated with an attempt at rotator cuff repair.
Surgical Technique
In nine studies, the latissimus dorsi tendon transfer was the exact or a modified version of the original procedure described by Gerber et al. (dual incision)26. One of these studies20 included a single patient who had undergone a combined latissimus dorsi and teres major tendon transfer. The authors of only one study utilized a modified L’Episcopo technique17 and transferred the latissimus dorsi and teres major via a single incision.
Outcomes
Compilation of the data from all ten studies11,13,15-22 showed that a total of 258 patients (range, thirteen to sixty-seven per study) and 262 shoulders (range, thirteen to sixty-nine per study) were included in the postoperative analyses (Table III). Seventeen percent of the patients were lost to follow-up. Patients were followed for a frequency-weighted mean of 45.5 ± 3.7 months (range, twenty-four to 126 months). Preoperative and postoperative adjusted Constant scores were reported in six studies13,16,17,19,21,22. Patients had a frequency-weighted mean adjusted Constant score of 45.9 preoperatively compared with 73.2 postoperatively (p < 0.001). The authors of one study reported only postoperative adjusted Constant scores, noting a mean of 73 (range, 35 to 97)15. The authors of six studies reported a preoperative and postoperative pain score on a scale of 0 (maximum pain) to 15 (no pain)13,16,17,19,21,22. Patients had a frequency-weighted mean pain score of 4.8 preoperatively compared with 12.1 postoperatively (p < 0.001). The authors of two studies reported modified University of California, Los Angeles (UCLA) scores, with a frequency-weighted mean of 8.7 preoperatively compared with 20.7 postoperatively (p < 0.001)11,20. In the two studies in which Subjective Shoulder Value scores were reported, the frequency-weighted mean was 27.1 preoperatively compared with 66.5 postoperatively (p = 0.161)13,16. In two studies (n = 33), the investigators asked the question: “If you could go back in time, would you have surgery again?”; they reported twenty-nine (88%) affirmative responses15,20. Three studies (n = 57) specifically addressed satisfaction, with forty-five patients (79%) stating that they were satisfied with their outcome18,20,22.
With regard to functional assessment on physical examination, the authors of all ten studies11,13,15-22 reported preoperative and postoperative active forward elevation and those of nine studies13,15-22 reported active external rotation with the arm at the side. Active forward elevation improved from a frequency-weighted mean of 101.9° preoperatively to 137.4° postoperatively (p < 0.001), and active external rotation improved from a frequency-weighted mean of 16.8° to 26.7° (p < 0.001). The authors of five studies reported preoperative and postoperative active abduction, noting a preoperative frequency-weighted mean of 91.4° compared with 130.7° postoperatively (p ≤ 0.001)13,15,16,19,21. The studies were inconsistent with regard to reporting active external rotation with the arm abducted and in internal rotation (Table III). Preoperative and postoperative dynamometer measurements of strength in abduction were reported in five studies13,16,19,21,22, in which the frequency-weighted mean strength improved from 1.9 kg preoperatively to 3.2 kg postoperatively (p < 0.001).
Radiographs
In three studies16,19,22, the authors evaluated preoperative and postoperative osteoarthritis radiographically using the Samilson and Prieto score28. The frequency-weighted mean preoperative osteoarthritis score was 0.9 compared with 1.3 postoperatively (p = 0.095). Four studies compared acromiohumeral distance, with the authors noting a frequency-weight mean of 5.6 mm preoperatively compared with 5.0 mm postoperatively (p = 0.213)13,17,19,22. Three studies (n = 70) evaluated the change in superior migration of the humeral head after surgery according to the criteria defined by Gerber15; no change was noted in twenty-five patients (36%), a decrease in superior migration was noted in five (7%), and an increase in superior migration was found in forty (57%)11,15,21.
Complications/Reoperations
The authors of all studies reported on complications11,13,15-22, which included one infection (0.4%), seven cases of neuropraxia (2.7%), nine tears of the transferred tendon (3.4%), two failures of deltoid repair (0.7%), two hematomas (0.7%), and four cases of wound dehiscence (1.5%). The overall reported complication rate was 9.5% (twenty-five of 262).
Data on reoperations were reported in all studies11,13,15-22. Reoperations included revision repair of the transferred latissimus dorsi tendon (two, 0.8%), reverse total shoulder arthroplasty (two, 0.8%), shoulder arthrodesis (two, 0.8%), subscapularis repair (three, 1.1%), distal clavicle excision (three, 1.1%), deltoid muscle repair (two, 0.8%), rotator cuff debridement without infection (one, 0.4%), wound revision (two, 0.8%), and irrigation and debridement for infection (one, 0.4%). The overall rate of reoperations in all studies was 6.9% (eighteen of 262).
Predictive Factors
In reviewing the literature, we found that certain preoperative variables were determined in individual studies to influence overall outcome. As a result, these variables were individually assessed in this systematic review.
Primary Versus Revision Surgery
In four studies16,19,21,22, the authors conducted a subgroup analysis comparing outcomes in patients treated with primary reconstruction (n = 147) with those in patients who had a revision reconstruction after a previous failed rotator cuff repair (n = 47). In all four studies, the investigators utilized the Constant score as their instrument of evaluation, reporting postoperative scores; in two of these studies, the authors also reported preoperative scores. The mean frequency-weighted adjusted preoperative Constant score was 50.0 for patients treated with primary reconstruction compared with 44.1 for those treated with revision surgery (p = 0.790). Postoperatively, the frequency-weighted Constant score improved to 76.0 for those treated with primary reconstruction compared with 60.1 for those managed with revision surgery (p = 0.051). The Constant score improved by an average of 26.7 points after primary reconstruction compared with 15.1 points after revision surgery (p = 0.004).
Subscapularis Insufficiency
In two studies16,19 (n = 101), the authors specifically evaluated the preoperative status of the subscapularis tendon, with a preoperative lift-off test in one16 and with magnetic resonance imaging (MRI) in the other19, and correlated this to preoperative and postoperative function. The subscapularis was intact in seventy-three patients and deficient in twenty-eight patients. The mean frequency-weighted adjusted preoperative Constant score was 52.6 in patients with subscapularis insufficiency compared with 41.3 in those with an intact subscapularis (p = 0.580). Postoperatively, the frequency-weighted Constant score improved to 67.9 in those with subscapularis insufficiency compared with 66.0 in those with an intact subscapularis (p = 0.903). The Constant score improved an average of 15.3 points in those with subscapularis insufficiency compared with 24.7 points in those with an intact subscapularis (p = 0.343). Gerber et al.16 found that the mean pain score (on a scale of 0 to 15) improved from 6 points preoperatively to 12 points postoperatively (p < 0.0001) in patients with an intact subscapularis, whereas it improved from 6 to 9 points (p < 0.05) in those with subscapularis insufficiency. While active forward elevation and abduction improved similarly in both groups, active external rotation improved from 21° preoperatively to 31° postoperatively (p = 0.007) in those with an intact subscapularis compared with a decrease from 28° preoperatively to 21° postoperatively (p > 0.05) in those with subscapularis insufficiency19. In an additional study15, the author reported only postoperative adjusted Constant scores in patients with a deficient subscapularis (n = 4, Constant score = 48), as determined with MRI, and in those with an intact subscapularis (n = 12, Constant score = 82). The author also stratified preoperative and postoperative active forward elevation on the basis of subscapularis tendon status. Patients with an intact subscapularis had a mean preoperative and postoperative active forward elevation of 82° and 149°, respectively. In contrast, patients with an insufficient subscapularis had a mean preoperative and postoperative active forward elevation of 85° and 91°.
Status of the Teres Minor Tendon
In three studies13,21,22, the authors attempted to correlate the preoperative status of the teres minor on MRI with postoperative outcomes. In one study13 (n = 22), the teres minor status was assessed with MRI and fatty atrophy was graded as Goutallier29 Stage 0 in five cases (23%), Stage 1 in six (27%), Stage 2 in four (18%), Stage 3 in six (27%), and Stage 4 in one (5%). The authors noted that fatty infiltration of the teres minor muscle of Stage 2 or lower was associated with better preoperative and postoperative active external rotation and active forward elevation as well as an improved adjusted Constant score compared with those associated with fatty infiltration of Stage 3 or higher. Additionally, eleven patients (50%) had a partial tendon tear and two patients (9%) had a full-thickness tear of the teres minor. The presence of a partial or complete tear of the teres minor was associated with worse preoperative adjusted Constant scores (52% versus 61%, p = 0.04) and active external rotation (6° versus 32°, p = 0.007) but was not associated with worse postoperative outcomes. Another study21 (n = 47) included eleven patients with fatty infiltration of the teres minor of Stage 3 or higher as seen on MRI, and the authors noted more “inferior results,” defined as a lack of clinical improvement in the Constant score as compared with the preoperative score, in this group compared with those with fatty infiltration of Stage 2 or lower (p = 0.018). Similarly, the authors of a third study22 noted that Stage-3 or 4 fatty infiltration of the teres minor as seen on a computed tomography (CT) arthrogram was associated with preoperative and postoperative active external rotation deficits (p < 0.05) and lower postoperative adjusted Constant scores (p < 0.05) as compared with patients with fatty infiltration of Stage 2 or lower.
Latissimus dorsi tendon transfer has been considered to be a means of restoring function and providing pain relief in patients with irreparable tears involving the infraspinatus and supraspinatus tendons in the absence of substantial glenohumeral arthritis. In our systematic review, patients who underwent latissimus dorsi transfer were younger (mean age, 58.7 years) than patients treated with reverse shoulder arthroplasty (mean age, seventy-one years) for irreparable rotator cuff tears in the absence of glenohumeral arthritis in a study by Mulieri et al.10. In no study in this systematic review did the authors specifically consider outcomes of latissimus dorsi tendon transfer on the basis of patient age, and it remains unclear whether age is an independent risk factor for outcome. Similarly, although the large majority of patients who underwent this procedure were men (70%), no study demonstrated a difference in outcome or complications based on sex. It is unclear whether this demographic difference is due to an increased number of irreparable tears in men, differences in patient selection, or some other unstudied variable. Importantly, most patients had the latissimus dorsi transfer on the dominant side (77%), and a large proportion of patients (41%) had had previous failed rotator cuff surgery. No study demonstrated a difference in outcomes based on whether the procedure had been done on the dominant extremity; however, it is unlikely that any study generated the power necessary to evaluate this variable. Of note, the reporting of preoperative occupation and activity was inconsistent among the studies. As a result, the impact of these variables on the determination of surgical indications for, and outcomes of, latissimus dorsi tendon transfer for irreparable rotator cuff tears are unclear.
In general, latissimus dorsi tendon transfer led to significant improvements in both function and pain relief; however, as evidenced by the data presented, patients and physicians should not expect a return to normal function or complete elimination of pain. Furthermore, patients can be counseled that, on average, they can expect a gain of approximately 35° in active forward elevation, a gain of approximately 10° in active external rotation, and an approximately 70% increase in abduction strength, but not a return to normal active motion or strength. Despite postoperative improvements, no study compared latissimus dorsi tendon transfer with any other treatment option. As a result, it is not possible to determine, without using historical controls, whether latissimus dorsi tendon transfer is a preferable treatment method in young, active patients with irreparable posterior-superior rotator cuff tears. We cannot, on the basis of this systematic review, describe how the results of a tendon transfer might compare with those of an alternative treatment, such as partial rotator cuff repair, augmented rotator cuff repair, or reverse shoulder arthroplasty, in a similar patient population. Furthermore, as a result of the limited numbers of patients, we are unable to determine the preferred technique for latissimus dorsi tendon transfer (single-incision versus dual-incision) or whether the additional transfer of the teres major influences outcomes in a clinically relevant manner. This review shows that latissimus dorsi transfer does not eliminate the potential for glenohumeral joint arthropathy and the data presented demonstrate mean worsening of the osteoarthritis grade and acromiohumeral distance as well as increasing superior migration of the humeral head in >50% of shoulders after this transfer. However, as no study compared progression of osteoarthritis in patients who had undergone latissimus dorsi tendon transfer with the same variable in patients treated with another operative or nonoperative modality, it is unclear whether the procedure ultimately slows, increases, or has no effect on the natural progression of rotator cuff arthropathy.
Although study variability made it difficult to systematically identify predictors of poor outcome, the data suggest that poor functional outcomes are more likely after revision surgical procedures, in patient in whom the subscapularis is insufficient, and in those with advanced teres minor fatty muscle atrophy. As patients who underwent revision surgery had less of an improvement in the Constant score compared with those who underwent primary surgery, there likely exists a cohort of patients with rotator cuff tears who would benefit from latissimus dorsi transfer before rotator cuff repair is attempted. However, identifying this population is not possible on the basis of the available literature. Unfortunately, subscapularis insufficiency was evaluated by different methods (physical examination16 versus MRI19) in the studies in this review so it is difficult to compare these results in a systematic fashion. Although studies were somewhat mixed with regard to their findings of Constant score improvement based on the status of the subscapularis, the studies15,16 that investigated shoulder motion demonstrated significantly decreased active forward elevation and active external rotation in patients considered to have subscapularis insufficiency. This finding helps to support the belief that the subscapularis is important for centering the humeral head in the glenoid during motion in the horizontal (active forward elevation) and frontal (active forward elevation) planes in patients who undergo latissimus dorsi tendon transfer16. The integrity of the teres minor was assessed with preoperative imaging in only three studies13,21,22, and all demonstrated better functional outcomes and active external rotation in patients with fatty muscle atrophy of Stage 2 or lower. Future studies of latissimus dorsi tendon transfer should include preoperative MRI data regarding the quality of the subscapularis and the teres minor, and outcomes data should be stratified by these variables. Furthermore, future studies should be designed to conduct multiple regression analyses to evaluate the combined impact of these variables on outcome as they may be additive.
It is important to note that not all studies included postoperative evaluation with MRI or ultrasound, so the rate of repeat tendon tears or lack of healing may be underreported. Furthermore, no study investigated the preoperative quality of the latissimus dorsi muscle as a variable that could influence functional outcome or tearing of the transferred tendon. We recommend that authors of future studies attempt to determine the integrity of the repair using follow-up MRI or ultrasounds. Although the authors of all studies reported on reoperations, the reoperation rate is very strongly influenced by the duration of follow-up in individual studies. On the basis of the data available, only two patients (0.7%) underwent conversion to a reverse total shoulder arthroplasty, at mean 46.2 months postoperatively. Further follow-up will be necessary to determine the long-term reoperation rate as it is evident, even with this short-term follow-up, that joint arthropathy progresses despite the tendon transfer.
This study has a number of weaknesses. As is typical of any systematic review, the inherent weaknesses of each individual study (retrospective design, case series, short-term follow-up, etc.) translate into limitations of this review. The methodologies of the papers reviewed did not provide controls for bias, confounding, or chance as the articles were mainly descriptive in nature. As such, our study is simply an observational review of mostly observational studies, but it does provide an overview of the literature available. Weaknesses inherent to the individual studies are not improved by aggregating them. Additionally, the studies reviewed involved different methods of outcomes evaluation, and most did not present raw data for each patient included in the analysis. Furthermore, although we utilized a systematic methodology that we believe to be reproducible, it is possible that different search terms and different search engines would have provided additional studies that would have met our inclusion criteria.
This systematic review describes the demographic factors that characterize patients undergoing latissimus dorsi tendon transfer for irreparable posterior-superior rotator cuff tears. Compiled data and frequency-weighted means demonstrated improvement in shoulder function, motion, strength, and pain relief after tendon transfer. Importantly, patients and surgeons must be willing to accept that this is a salvage procedure and should not expect an outcome of “normal” function or complete pain relief. Although this systematic review provides valuable information, no study included a nonoperatively treated group or a comparison group treated with a different surgical procedure and the mean duration of follow-up was only 45.5 months. Thus, it remains challenging to draw meaningful conclusions regarding the long-term outcomes and benefits of this procedure compared with those of other treatment modalities for irreparable posterior-superior rotator cuff tears.
Disclosure: None of the authors received payments or services, either directly or indirectly (i.e., via his or her institution), from a third party in support of any aspect of this work. None of the authors, or their institution(s), have had any financial relationship, in the thirty-six months prior to submission of this work, with any entity in the biomedical arena that could be perceived to influence or have the potential to influence what is written in this work. Also, no author has had any other relationships, or has engaged in any other activities, that could be perceived to influence or have the potential to influence what is written in this work. The complete Disclosures of Potential Conflicts of Interest submitted by authors are always provided with the online version of the article.