The Latarjet procedure, a modification of the Bristow procedure, has been indicated for treatment of anterior glenohumeral instability in patients with associated glenoid bone loss of up to 40% as well as in patients who have previously undergone anterior shoulder stabilization procedures that were unsuccessful1,2. Recently, some surgeons have considered use of the Latarjet procedure after an initial glenohumeral dislocation even if no glenoid bone loss is present.
Multiple previous reports have included comments on the outcomes of the Latarjet or the Bristow procedure that indicated a recurrent instability rate of between 0% and 14%3-8. Complications of the Bristow procedure were described well by Young and Rockwood and include recurrent anterior and posterior instability, arthrosis, osseous nonunion, implant failure or loosening, and neurovascular injury9. Other authors have specifically noted neurologic injuries after anterior stabilization procedures10-13. Ho et al.10 reported a neurologic deficit in 8.2% of patients after open anterior stabilization for recurrent shoulder instability, and Richards et al.12 identified eight patients with brachial plexus injury after Putti-Platt, Bristow, or modified Bristow procedures. Bach et al.11 and Maquieira et al.13 described rarer nerve injuries affecting the ulnar and suprascapular nerves, respectively. However, differences between the Bristow and Latarjet techniques as well as the heterogeneous nature of prior studies confound attempts to estimate the prevalence of complications following the Latarjet procedure.
Despite the reported success of the Latarjet procedure for treatment of glenoid bone loss and recurrent anterior glenohumeral instability, there is a paucity of literature related specifically to complications of this procedure. Allain et al.6 reported a 7% complication rate in ninety-five patients treated with the Latarjet procedure; the complications included infection, frozen shoulder, and humeral fracture (after manipulation). Burkhart et al. reported a 5% complication rate in 102 patients; the five complications involved hematoma formation, loose implants, and fibrous union of the bone graft1. Neurologic complications were not documented in either of these two studies. Given the surgical anatomy involved as well as the complexity of the procedure, documentation of neurologic injury is relevant and important.
The purpose of our study was to highlight all complications including infection, recurrent glenohumeral instability, and neurologic injury following the Latarjet procedure. We attempted to determine the prevalence of each complication, identify possible risk factors, and evaluate the consequences of each complication on the outcome of the Latarjet procedure.
Forty-seven patients (forty-eight shoulders) who were treated with the Latarjet procedure for anterior glenohumeral instability between January 2005 and January 2010 and participated in a supervised rehabilitation protocol formulated by the senior author were eligible for inclusion in the study. Patients were eligible if the treatment was performed for anterior shoulder instability secondary to glenoid bone loss of <30% or if it was performed after a prior arthroscopic or open capsule and labral (Bankart) repair for instability was unsuccessful. Patients were excluded from the study if they had symptoms of multidirectional glenohumeral instability or volitional instability, had undergone a prior arthroplasty procedure, or were not compliant with the postoperative rehabilitation protocol.
The mean age of the patients was thirty years (range, fifteen to sixty years); thirty-nine of the shoulders were in male patients and nine were in female patients. Thirty-five (73%) of the forty-eight shoulders had undergone prior surgery for glenohumeral instability. Overall, the mean number of surgical procedures prior to the Latarjet procedure was 1.23, with twelve shoulders (25%) having undergone more than one prior surgical procedure and seventeen (35%) having undergone a prior open stabilization procedure.
Glenoid bone loss was estimated with use of three-dimensional computed tomography. The Latarjet procedure was recommended if the tangential length of the region of anterior glenoid bone loss was greater than the maximal glenoid radius as described by Gerber and Nyffeler14.
Surgical Technique
All patients were examined under anesthesia to confirm the preoperative diagnosis of shoulder pathology made on the basis of the physical examination and imaging studies. The patient was positioned in the beach chair position with use of a TENET table (Smith & Nephew, Andover, Massachusetts) and SPIDER arm holder (Smith & Nephew) for arm positioning. If a prior arthroscopic soft-tissue stabilization procedure had been performed, arthroscopic examination was performed to assess the quality of the soft tissues and to assess any bone involvement that had not been appreciated on preoperative imaging.
An anterior incision was made, and the deltopectoral interval was identified and opened. A sharp Hohmann retractor was placed above the coracoid process, and the coracoacromial ligament was incised off its insertion on the acromion. The coracohumeral ligament beneath the coracoacromial ligament was released from the lateral edge of the coracoid process. The pectoralis minor was detached from the superomedial aspect of the coracoid with use of electrocautery until the coracoclavicular ligaments were visible. A coracoid osteotomy was performed with use of a 90° sagittal saw, cutting in a medial-to-lateral direction just anterior to the coracoclavicular ligaments.
The coracoid was prepared by flattening its undersurface with a saw and burr without detaching the coracoacromial ligament. In most cases, a 2.5-mm drill was used to create two holes 1 cm apart. In the other five shoulders, a cannulated screw system was to be used and the coracoid was drilled initially with a guidewire and subsequently with a 3.2-mm drill.
A subscapularis-splitting approach to the glenoid was performed by means of a horizontal incision at the junction of the upper two-thirds and lower one-third of the muscle. The plane between the subscapularis muscle and the capsule was bluntly developed. A blunt Hohmann retractor was placed inferiorly in this interval to protect the axillary nerve, which had been identified prior to the subscapularis split. A vertical capsulotomy was performed, and the lateral capsule was tagged with number-2 nonabsorbable suture for later repair to the coracoacromial ligament. A blunt Hohmann retractor and a large Steinmann pin were used to retract the lower and upper aspects of the subscapularis, respectively. The Steinmann pin was inserted into the coracoid base beneath the upper aspect of the subscapularis and was angled superiorly to retract the muscle. A Fukuda retractor was placed into the glenohumeral joint to expose the anterior aspect of the glenoid. The labrum and medial capsule at the anteroinferior aspect of the glenoid were debrided, and the osseous bed was prepared with use of an osteotome.
A 2.5-mm drill was used to create the inferior screw hole, starting approximately 5 to 7 mm medial to the glenoid rim; this hole was drilled as parallel as possible to the glenoid surface and penetrated the posterior cortex. The appropriate screw length was determined by adding the depth measurements made in the drill hole in the glenoid and the coracoid. A 3.5-mm fully threaded screw was first passed through the inferior screw hole in the coracoid process and then through the glenoid bone tunnel. A 2.5-mm drill was then used to create the superior screw hole through the coracoid and the glenoid, and a 3.5-mm cortical screw of the appropriate length screw was inserted. Any lateral overhang of bone into the joint was debrided with a burr to create a smooth surface; soft tissue was removed but the coracoacromial ligament was preserved. Five shoulders had a larger cannulated screw, rather than a 3.5-mm screw, placed over Kirschner wires.
The remaining coracoacromial ligament was sutured to the lateral capsule with number-2 nonabsorbable suture, and the lateral aspect of the subscapularis was closed with number-2 suture as well. Closure of the skin was performed in a standard fashion. The patient's arm was placed in a sling and abduction pillow before leaving the operating room.
Data Collection and Clinical Evaluation
Patients were seen in the office preoperatively and at two, six, twelve, and sixteen to twenty weeks postoperatively; the final evaluation was usually performed at one year postoperatively. Radiographs were obtained at the two-week, twelve-week, and sixteen to twenty-week postoperative visits. A computed tomography (CT) scan was ordered at fours months postoperatively to assess union of the coracoid to the glenoid. All patients were evaluated with use of a modification of the rating scale developed by Rowe (see Appendix), the Subjective Shoulder Value score (SSV), and the American Shoulder and Elbow Surgeons (ASES) self-assessment score15,16. Shoulder range of motion (forward flexion and external and internal rotation in abduction and adduction) was documented with use of scapular stabilization. An examination for anterior glenohumeral instability was performed as described by Jobe et al.17. A patient who was asymptomatic at the final evaluation was instructed to return for further follow-up as necessary.
The mean duration of clinical follow-up in the office for the patients with at least six months of follow-up was 9.4 months (range, six to fifty-five months). Forty-four of the forty-seven patients were also available for a final phone interview at a mean of 39.3 months (range, six to sixty-two months) postoperatively to assess the subjective clinical outcome. No additional subjective complications were reported during the final phone interview.
Statistical Analysis
The Fisher exact test was used to calculate p values associated with comparisons between categorical variables (prevalence of a complication), and paired and unpaired t tests were used to calculate p values associated with comparisons between continuous variables (mean values). For categorical variables, a logistic regression model was used to determine the odds ratio for the occurrence of a complication. The Hosmer-Lemeshow test was used to test the goodness of fit for each significant variable in the regression analysis to determine whether the result provided an accurate representation of the observed values. Finally, for each variable for which a significant difference was not found, a power analysis was performed to determine whether the sample size would have been adequate to demonstrate the existence of a difference with 80% probability if a difference had in fact existed. A p value of <0.05 was considered significant for all comparisons.
Source of Funding
There was no external source of funding.
The objective and clinical outcomes for the entire study population are given in Tables I and II. Forward elevation remained unchanged from the preoperative to the final postoperative assessment, averaging 149°. External rotation, however, decreased by 10.3° (p < 0.05).
According to the modified Rowe scoring system, all shoulders were graded as “poor” preoperatively (because of glenohumeral instability). Postoperatively, twenty-six were graded as “excellent,” nine as “good,” nine as “fair,” and one as “poor”; 78% had either excellent or good results (Table III). The mean SSV improved significantly from 35.9% preoperatively to 81.5% at the final postoperative assessment. At the time of the final follow-up, the mean ASES score on the operatively treated side was 86.3, which was 90% of the value of 96.2 on the contralateral side.
Twenty-nine (60%) of the forty-eight shoulders underwent postoperative CT scanning with coronal and sagittal reconstruction to assess for osseous union and incorporation of the coracoid graft. The CT scans demonstrated incorporation and union of the graft in twenty-one (72%) of these twenty-nine shoulders at a mean of 6.5 months postoperatively (Figs. 1 and 2). Three of the eight shoulders without osseous union were symptomatic for recurrent glenohumeral instability and required an additional surgical procedure, and only these three cases were identified as complications because they resulted in clinical instability. The other five cases (four of which involved a fibrous nonunion) were asymptomatic and were not counted as a complication because they did not affect the patient outcome.
Complications of the Latarjet Procedure
We divided complications into three major categories: infection, recurrent glenohumeral instability, and neurologic injury.
Infection
The postoperative infection rate was 6% (three of forty-eight shoulders); all three cases involved a superficial infection. Two of the three shoulder injuries involved a Workers’ Compensation claim, and none of the patients smoked. All three shoulders had undergone at least one prior surgical procedure, with two having undergone multiple surgical procedures that included open stabilization. All three shoulders underwent irrigation and debridement in the operating room under sterile conditions. One patient presented with a hematoma at two weeks postoperatively, had subsequent positive cultures for Enterobacter aerogenes and coagulase-negative Staphylococcus aureus, and was successfully treated with intravenous antibiotics. The second patient presented at four weeks postoperatively and had negative cultures but was treated with intravenous vancomycin for two weeks, which resulted in resolution of the symptoms. The third patient presented at six weeks postoperatively with a draining, purulent wound and was determined to have a superficial infection that did not penetrate the deep fascial layers. The symptoms resolved after intravenous administration of vancomycin for one week followed by dicloxacillin for two weeks after positive cultures for coagulase-negative Staphylococcus aureus.
Instability
The rate of recurrent glenohumeral instability was 8% (four of forty-eight shoulders). Three of the four shoulder injuries involved a Workers’ Compensation claim, and two of the patients smoked. All four shoulders had undergone a prior open stabilization procedure, with two having undergone multiple open procedures. Furthermore, either 4.0-mm or 4.5-mm cannulated screws had been used for coracoid fixation in all four of the shoulders. Two failures were identified within eight months of the Latarjet procedure. One of these patients had continued pain and subjective glenohumeral instability without frank dislocation, and CT scans and radiographs demonstrated a loose inferior screw with nonunion of the coracoid graft at seven months postoperatively. This patient underwent autogenous iliac crest bone-grafting with subscapularis reconstruction seven months after the primary Latarjet procedure. The other early failure was identified at eight months postoperatively; the patient had continued pain and subjective instability without frank dislocation, and CT scans and clinical examination demonstrated nonunion of the graft with resorption and subscapularis insufficiency. This patient underwent removal of the screws and a pectoralis major transfer, without bone-grafting because the glenoid defect was not exceptionally large.
One of the other two shoulders that failed was continuing to undergo nonoperative treatment at nineteen months postoperatively, and the remaining shoulder underwent revision surgery at forty-two months. The nonoperatively treated patient had continued pain and instability without frank dislocation after being released to full activity; associated radiographs demonstrated loose screws. A CT scan was ordered, but the patient did not return for follow-up, although he did respond to the phone interview and indicated that he was continuing nonoperative management. The remaining patient did well for approximately three years until she incurred multiple falls and trauma to the shoulder after joining the Army Reserve. A CT scan demonstrated complete resorption of the graft with loosening of the screws. She underwent autogenous iliac crest bone-grafting and did well for approximately one additional year until she fell again and redislocated the shoulder. At the time of the final follow-up, she had not undergone a repeat revision procedure for glenoid reconstruction.
Neurologic Injury
The rate of neurologic injury was 10% (five of forty-eight shoulders). Two of the five neurologic injuries involved the musculocutaneous nerve, one involved the radial nerve, and two involved the axillary nerve. None of the injuries were associated with a Workers’ Compensation claim, and one patient was a smoker. Three of the five shoulders had undergone prior surgical procedures, with only one of the shoulders having undergone a prior open procedure. Both patients with musculocutaneous nerve involvement had sensory neurapraxia (decreased sensation in the lateral antebrachial cutaneous nerve distribution) that resolved within two months postoperatively. One of these patients underwent electromyography and nerve conduction velocity testing (EMG/NCV) that revealed focal neuropathy. The patient with radial nerve involvement also experienced sensory neurapraxia that resolved within two months. One of the two patients with axillary nerve injury continued to have dysesthesias at five years postoperatively despite having normal strength. The other patient had axillary, medial cord, and suprascapular neuropathy without denervation documented on EMG/NCV, and this patient was subjectively improving, with minimal continued numbness and weakness of the deltoid, at fourteen months postoperatively.
Clinical Effect of the Complications
The overall complication rate was 25% (twelve of forty-eight shoulders). According to the modified Rowe score, the results were excellent or good in 42% of the shoulders with a complication compared with 91% of the shoulders without a complication. The mean postoperative SSV and ASES scores in the shoulders with a complication (65.5% and 69.9, respectively) were both significantly worse than those in the shoulders without a complication (87% and 91.8). The percentage of shoulders that had undergone surgical procedures prior to the Latarjet procedure and the mean number of prior procedures were higher in the group with a complication (83% and 1.58 procedures, respectively) than in the group without a complication (69% and 1.11 procedures), although these differences were not significant (Table II). Furthermore, there was a trend toward a higher prevalence of prior multiple surgical procedures including an open stabilization procedure in the shoulders with a complication. However, comparisons between the shoulders with and without prior surgery should be interpreted with caution; because of the small size of the group without prior surgery, it does not necessarily constitute a representative sample of the population of shoulders that have not undergone prior surgery.
Three variables were determined to be significant predictors of the occurrence of a complication following the Latarjet procedure according to the multivariate analysis (Table IV). A Workers’ Compensation claim and the use of 4.0 or 4.5-mm cannulated screws both had an odds ratio of 12 (p = 0.0260), and age had an odds ratio of 1.075 per year (p = 0.0188). Thus, the odds of sustaining a complication following the Latarjet procedure were twelve times greater when the injury was associated with a Workers’ Compensation claim or when cannulated screws were used, and the odds increased 7.5% for every one-year increase in age. Testing of the goodness of fit of the model for age with use of the Hosmer-Lemeshow method and testing of the power of the other two variables determined that age was the only variable that was a good predictor of the occurrence of a complication. However, Workers’ Compensation status and use of cannulated screws both had a power of 0.759 (nearly 80%); thus the likelihood of a type-II error (beta) was close to 20%, the level at which we would have concluded that these variables were good predictors of the occurrence of a complication. All other variables, including smoking (power = 0.374), prior surgery (power = 0.046), prior open surgery (power = 0.460), and sex (power = 0.341), had low power and thus large type-II errors, indicating that the sample size was inadequate to classify these variables as good predictors.
Four of the five shoulders that underwent fixation of the coracoid graft with either 4.0 or 4.5-mm cannulated screws (rather than 3.5-mm cortical screws) had recurrent glenohumeral instability postoperatively (Figs. 1 and 2). Three of these four shoulders had a postoperative CT scan (made at three, four, and thirty-six months) available for review, and none of these scans showed union of the coracoid graft. In addition, CT scanning performed postoperatively in the group that was not considered to have a complication revealed four shoulders in this group that did not have osseous union (at four, four, five, and six months postoperatively). Fixation in all four of these shoulders had been performed with use of 3.5-mm noncannulated screws, and all four shoulders remained asymptomatic at the time of the latest follow-up.
Four (33%) of the twelve shoulders with a complication were in patients who were smokers, compared with three (8%) of the thirty-six shoulders without a complication, but this difference did not reach significance (p = 0.0552).
Latarjet originally described the transfer of the coracoid process to the anterior aspect of the glenoid in 195418. The current study revealed a 25% rate of complications including infection, recurrent glenohumeral instability, and neurologic injury following the Latarjet procedure. To our knowledge, only Allain et al. previously reported the rate of infection associated with this procedure6. An infection requiring reoperation occurred in 6% of the shoulders in our study, and all patients recovered after administration of intravenous antibiotics. The only notable finding was that all three shoulders with an infection had undergone prior surgery, with two having undergone multiple procedures including prior open stabilization.
Although the rate of recurrent instability in our series was greater than that reported by others1,6, it approaches the recurrence rate reported for the Bristow procedure and other open anterior stabilization procedures2-7,9. In our study, a higher rate of instability was found in patients with cannulated screw fixation (all four cases), a history of smoking (three of the four), and a Workers’ Compensation claim (two of the four). Our study indicated that the prevalence of any complication, not just instability, was twelve times greater if patients had cannulated screw fixation or if they had Workers’ Compensation claims. The power of this study was not sufficient to demonstrate these differences with statistical significance. However, increased age increased the risk of a complication, with every year of additional age increasing the odds of having a complication by 7.5%. Goodness-of-fit modeling demonstrated that age was a good predictor for occurrence of a complication. Other studies of the Latarjet procedure have shown that age and Workers’ Compensation status can negatively impact the outcome. However, to our knowledge, the use of cannulated screws, which has been advocated and used as a method for coracoid graft fixation, has not been previously reported as a potential cause of failure.
Biomechanically, cannulated screws are weaker than noncannulated screws; however, we did not find any breakage at the time of follow-up. Therefore, we postulate that fixation with cannulated, partially threaded screws was inferior to that with noncannulated, fully threaded screws. The thread depth of the cannulated screws is less than that of the noncannulated screws, which can affect purchase in the native scapula and may result in less compression of the graft.
A 10% rate of neurologic injury (five of forty-eight shoulders) was found in the present study compared with a maximum of 8% in previous studies19-21. Neurologic injury following stabilization procedures is thought to be caused by traction, patient malpositioning, and inadvertent suturing10,21. Ho et al. attributed neurologic injury to traction, as the majority of the injuries in their study were mixed in nature and 40% were diffuse plexopathies10. Additionally, Burge et al. demonstrated that the shortest nerves (the axillary nerve and musculocutaneous nerves) are at the highest risk for a traction injury22. In contrast to other studies, we found primarily sensory deficits resulting from injuries to the musculocutaneous and radial nerves. Also, these transient nerve injuries did not affect long-term outcomes. This differs from our findings regarding the axillary nerve injuries, which did result in longer-term deficits.
The limitations of this study are those inherent to retrospective studies. Studies of this nature almost always underestimate the rate of complications because complications are not necessarily accurately recorded in the patient records. Despite utilization of computerized records, the actual complication rate may be even higher than 25% because of the retrospective study design.
Summary
The Latarjet procedure is useful for treatment of anterior instability of the glenohumeral joint, including revision anterior stabilization in cases of recurrent instability with glenoid deficiency as well as primary anterior stabilization in shoulders with glenoid bone loss. However, the complication rate may be greater than that previously reported, especially with regard to neurologic injury.