Introduction

The Latarjet procedure, originally described in 1954, involves open coracoid osteotomy, graft transfer, and fixation to the inferior margin of the glenoid from the 2 o’ clock to the 6 o’ clock position using a horizontal split of the subscapularis [1].

The procedure affords optimum results also in the long term [2–4] and is considered as the gold standard for the treatment for anterior recurrent instability in patients with large bone defects and in contact sport athletes [5].

Similar to rotator cuff tears, which used to be repaired exclusively with open techniques and are now effectively managed with minimally invasive approaches [6, 7], instability appears to be increasingly amenable to arthroscopic treatment thanks to advances in diagnostic and surgical techniques [8–10].

Technical and scientific advances have enabled Laurent Lafosse’s group to develop the arthroscopic Latarjet procedure, which aims to combine the excellent results of the original open approach with the clear advantages of arthroscopy, for example, low invasiveness, greater accuracy in graft placement, easier identification, and treatment for associated lesions (SLAP, posterior labral lesions), lower risk of postoperative stiffness, less pain, better cosmetic result, earlier mobilization, and prompter recovery [11, 12].

Although its medium-term outcomes appear to be excellent, it is a complex and demanding procedure that involves a learning phase and should be performed only by expert surgeons [11]. However, given the recent development of the approach, the available literature regards a small number of patients and short follow-up periods.

The aim of this study was to analyze the experience of a surgical team by evaluating operative time, complications, and clinical outcomes with a view to describing the learning phase of the arthroscopic Latarjet procedure in quantitative terms.

Materials and methods

The first 30 consecutive arthroscopic Latarjet procedures, performed by a surgical team of three from February 2011 to June 2012, were reviewed. They involved 29 men and one woman, who accounted for 35 % of 85 patients operated on for anterior instability; the other 55 (65 %) subjects received an arthroscopic Bankart procedure. Clinical follow-up data were available for all patients at a mean interval of 13 months (range 6–22). All patients gave their informed consent to be included in the study.

The indication for the arthroscopic Latarjet procedure was based on age, number of dislocations, recurrences, extent of glenoid bone defect (>15–20 %), and humeral bone loss.

Mean patient age at the time of surgery was 32 years (range 21–52). The right shoulder was affected in 12 (40 %) patients and the dominant arm in 17 (56 %); about 24 (80 %) patients practiced sports (20 % at a competitive level). The mean number of preoperative dislocations was 8 (range 1–60); four patients (13 %) had recurrent dislocation after arthroscopic Bankart (n = 3) or Putti Platt (n = 1) repair.

Standard X-ray views and either CT (Pico method) [13] or MRI scans were obtained preoperatively. A Hill–Sachs lesion was detected in 24 (80 %) subjects and glenoid bone loss in all 30 patients.

All 30 operations were carried out with the patient in the beach-chair position by a single surgical team, who routinely perform arthroscopic shoulder surgery in the lateral decubitus position. The seven-portal technique described by Lafosse and colleagues (Fig. 1) and dedicated instrumentation (Depuy Mitek, Wokingham, UK) were used for all procedures with the patient under interscalene block and general anesthesia. The operation can be divided into 5 stages [11, 12]:

Fig. 1
figure 1

Portal position. The beach-chair position enables conversion to the open procedure using the same instrumentation

Full size image
  1. 1.

    Joint evaluation and exposure: the joint is examined through the posterior portal for associated lesions of the cuff, labrum, Bankart lesions, and Hill–Sachs lesions. The anteroinferior labrum, the middle glenohumeral ligament, and part of the inferior glenohumeral ligament are resected from the anterior portal. Both sides of the subscapularis tendon must be exposed to achieve a correct tendon split. Opening of the rotator interval allows visualization of the lateral side of the coracoid and release of the coracoacromial ligament.

  2. 2.

    Subscapularis split: the anterolateral portal is established parallel to the superior margin of the subscapularis tendon and the scope is passed through it to visualize the conjoint tendon and the lateral portion of the subscapularis. Further portals are also made: an inferior portal at the apex of the anterior axillary fold; an anteroinferior portal midway between the inferior and the anterolateral portal; and a medial portal. With a switching stick passed through the posterior portal, the area where the subscapularis split is to be performed is identified. Another switching stick passed through the medial portal helps protect the conjoint tendon and the plexus. The intra- and extra-articular spaces are easily seen from the anterolateral portal. The split is prepared and the anterior wall of the glenoid neck is carefully exposed and abraded with a burr, to enhance graft take.

  3. 3.

    Coracoid graft harvesting: the scope is moved to the anterolateral or the anteroinferior portal to visualize the coracoid; the pectoralis minor is released and a fresh portal established with a needle over the coracoid, to insert the drill guide. The drill holes must be made at the junction of the lateral 2/3 and the medial 1/3, not too close to the coracoid tip. The drill guide is removed after passing two K-wires through the coracoid, then the K-wires are overdrilled. The holes are tapped and the top hats inserted into the fragment using a flexible Chia wire. The base of the coracoid is burred from different portals and the osteotomy is completed. Using a double cannula, the coracoid screw is passed over the Chia wire and advanced through the top hats, thus securing the bone fragment. Any spikes on the edge and surface irregularities are trimmed or smoothed with a burr, to ensure perfect contact between bone graft and host bone.

  4. 4.

    Graft transfer: the graft is mobilized through the subscapularis split using the double cannula, and appropriate graft position on the anterior glenoid neck is selected with the switching stick passed through the posterior portal.

  5. 5.

    Graft fixation: 2-K-wire holes are drilled through the coracoid and the glenoid using the double cannula. The inferior hole is made first, ensuring that it is parallel to the glenoid surface (the arm must be pushed backward to do so). A cannulated screw is loaded over the K-wire and threaded into the glenoid. The same procedure is performed for the second screw. The screws must be tightened carefully, to avoid damaging the graft. Graft and screw position are checked through the posterior portal. Any further graft trimming can be done with the burr.

Joint evaluation disclosed 4 (13 %) associated lesions (2 SLAP lesions, 1 posterior Bankart lesion, and 1 partial cuff tear) which were treated in the course of the procedure. Two screws were applied in all patients but one, who required a single screw for graft fixation.

The time taken by the whole operation and by each stage was recorded during the procedure or was calculated later based on the video recording. The 5 operative stages were timed as follows: (1) joint evaluation and exposure: from the start of the operation until the change from the posterior to the anterolateral portal for the assessment of anterior structures; (2) subscapularis split: until split completion; (3) coracoid graft harvest: until completion of the coracoid osteotomy; (4) coracoid transfer: until graft placement; and (5) graft fixation: until the end of the procedure.

Patients wore an abduction brace for 3 weeks, during which time they were permitted pendulum exercises and elbow mobilization; they then began passive and active exercises aimed at recovering the complete ROM and progressive strengthening exercises. Patients were recommended to undergo a thorough clinical and X-ray examination before resuming sports activities (after at least 3 months).

An X-ray examination was performed postoperatively to assess graft placement.

Clinical evaluation (pre- and postoperative) was performed with the Rowe score, where scores from 100 to 90 are considered excellent, those from 89 to 75 are good, scores from 74 to 51 are fair, and those <50 are poor [14].

Patients rated themselves as very satisfied, satisfied, not very satisfied, and not satisfied with the procedure.

External rotation beside the body was measured with a goniometer, recording any complications.

To analyze the learning curve, patients were divided into the first 15 (group A) and the last 15 (group B), and Rowe scores, satisfaction, complications, operative time, and graft placement data were compared.

Statistical analysis was performed by standard parametric or nonparametric methods, as appropriate. A value of p < 0.05 was considered significant.

Results

Comparison of mean preoperative (27, range 20–30) and follow-up (90, range 75–100) Rowe scores indicated a highly significant improvement (p < 0.005; Wilcoxon test). Based on this scale, clinical outcome was excellent in 21 (70 %) patients and good in 9 (30 %). The postoperative ROM showed an average difference of 12° in external rotation (53° vs. 65°) compared with the contralateral shoulder. The mean time until return to work was 30 days (range 2–12 weeks), and the mean time until return to sports was 4 months (range 2–6). At 13 months, patient satisfaction with the operation was high, with 21 (70 %) very satisfied and 9 (30 %) satisfied patients.

There were 3 (10 %) complications (graft fracture), two in the first patient group and one in the second (p = 0.543, chi-square test). Two fractures occurred within 2 days of the operation; these included the patient who received a single screw due to a technical mistake involving excessively low positioning of the graft. In the third case, the patient reported a direct trauma to the shoulder at 30 days (Fig. 2). All 3 patients, who were aged 41, 49, and 51 years, respectively, underwent open revision surgery and osteosynthesis. Age >40 years was found to be associated with complications (p = 0.002; Fisher’s exact test).

Fig. 2
figure 2

CT scan documenting early graft fracture

Full size image

The graft was positioned medially in 2 (7 %) patients; it was flush with the glenoid in 23 (76 %) cases (3–5 o’clock) (Fig. 3); it was too high in 2 (7 %) and too low in one (3 %) case. The remaining 2 patients were those who experienced early fracture.

Fig. 3
figure 3

Correct graft placement documented on a X-rays and b arthroscopic image

Full size image

There were no neurological, vascular, or septic complications.

Two patients required late (1 year) screw removal due to intolerance to fixation devices; the procedure was performed arthroscopically.

There were no cases of recurrent dislocation.

Groups A and B appeared to be homogeneous at baseline, without significant differences in gender (p = 1, Fisher’s exact test), age (p = 0.663, t test), affected side (p = 0.136, chi-square test), or preoperative Rowe score (p = 0.174, Mann–Whitney U test). Comparison of their postoperative results highlighted no significant difference in terms of Rowe score change (p = 0.775, Mann–Whitney U test), patient satisfaction (p = 0.256, chi-square test), complications (p = 1, Fisher’s exact test), or graft placement (p = 0.505, Mann–Whitney U test).

Operative time decreased significantly (p = 0.001, t test) from 132 min in group A to 99 min in group B (Fig. 4). Phase duration also decreased significantly (except in stage 4), as follows: stage 1 (A 37 min, B 27 min); stage 2 (A 37 min, B 27 min); stage 3 (A 35 min, B 27 min); and stage 5 (A 15 min, B 9 min) (all p < 0.05, t test). The duration of phase 4 was unchanged (17 min; p = 0.767, t test) (Table 1).

Fig. 4
figure 4

Operative time as a function of case number

Full size image
Table 1 Comparison of operative time in the two patient groups
Full size table

Discussion

The Latarjet procedure is a key treatment option for anterior instability, whose clinical and imaging presentation is increasingly variable due to diagnostic and technological advances.

Even though arthroscopic Bankart repair is quite widespread and provides excellent results [15, 16], it does entail a risk of recurrence, which appears to be related to factors such as age, involvement in contact sports, glenoid and humeral bone loss, and complex soft-tissue injury (e.g., HAGL lesion) [17–19].

The Latarjet procedure is the gold standard for the treatment for chronic anterior instability in patients with large bone deficits [5].

Compared with the free bone graft, it also benefits from the sling effect of the conjoint tendon crossing the subscapularis in the framework of the triple blocking effect described by Patte et al. [20].

Despite the fact that it has become standardized, the open Latarjet procedure is however not devoid of risks and complications.

Allain et al. and Burkhart et al. [2, 21] reported complication rates of 5–7 % that included infection, frozen shoulder, humeral fracture (after manipulation), hematoma formation, graft loosening, and fibrous union of the bone graft. More recently, a 25 % rate of short-term complications has been described by Shah et al. [22] (6 % superficial infection, 8 % recurrent glenohumeral instability, 10 % neurological injury), with a 12 % rate of revision surgery and a risk of complications that increased with age.

Like other authors, who have reported their experience with and results of procedures originally performed by open surgery [23, 24], Lafosse and co-workers have proposed an arthroscopic Latarjet procedure to combine the advantages of the original technique with those of arthroscopy. In their 2010 study, 62 of 100 patients assessed prospectively were available for direct follow-up at 18 months; of these, 80 % rated their outcome as excellent, 18 % as good, and 2 % as disappointing. At 26 months, the 35 available patients reported 91 % excellent results and 9 % good results. The difference in external rotation was 18° compared with the contralateral limb [11]. There were 4 perioperative complications: hematoma formation (n = 2), intraoperative graft fracture (n = 1), and transient musculocutaneous nerve palsy that recovered fully (n = 1).

Although it has been reported that the arthroscopic Latarjet procedure involves a steep learning curve [11], to our knowledge, it has never been quantified objectively.

Similar learning curves have been described for other orthopedic procedures. Konan et al. [25] described a learning curve in 30 patients undergoing hip arthroscopy on the basis of operative time. Guttmann et al. [26] found a substantial reduction in operative time for arthroscopic rotator cuff repair as surgical experience increased and quantified the learning curve as the first 10 cases.

In the present study, comparison of groups A and B highlighted the absence of significant differences in terms of clinical score, patient satisfaction, radiographic appearance, and complications, reflecting the effectiveness of the procedure even at the time of its first performance. Operative time then decreased progressively and significantly as the familiarity with the procedure increased and the knowledge of the technique and of regional anatomy improved. The shorter operative time makes the arthroscopic approach safer and suitable for larger patient numbers.

Increasing surgical experience significantly reduced the duration of four of the five operative stages, even though duration does not correlate with complexity, each stage being highly demanding and characterized by specific problems. Intraoperative bleeding is a critical factor because it correlates with a longer procedure and entails possible risks. In our opinion, retrospective analysis of total and partial operative time provides for a more critical and reasoned approach to the procedure and makes it possible to assess the scope for its quantitative improvement.

None of the 30 operations described here entailed neurological, vascular, or septic complications, demonstrating the technique’s feasibility and reproducibility despite its difficulty and complexity.

The clinical results were excellent (70 %) or good (30 %) and the postoperative course appeared to be simpler and better tolerated compared with one of the open technique, although comparisons between the two approaches are currently not available. An additional benefit of the arthroscopic approach was that it also allowed identification and treatment for associated lesions in 13 % of patients. Graft placement was perfect in 76 % of cases. The postoperative ROM showed an average 12° difference in external rotation compared with the contralateral shoulder.

The three cases of graft failure correlated significantly with age >40 years, a finding that clearly requires further investigation. Graft fracture is a rare complication, and the 10 % rate in our series appears to be only partially attributable to technical errors.

Late screw removal has also been reported by Lafosse and Boyle (4 cases) [11]. In our series, it involved two patients (7 %) of the first group, who moreover had been operated on when the instrumentation still envisaged screws of a single size (36 mm). This complication may therefore be ascribed to technical factors that are susceptible of being addressed.

The short follow-up is clearly a weakness of the study, whose main objectives were to analyze the learning curve and early complications, to gauge the safety and reproducibility of the approach.

The arthroscopic Latarjet procedure appears to be the natural evolution of the open operation. Clearly, too few operations have been performed, and published reports involve short follow-up periods; however, the technique has the potential to couple the effectiveness of the original open technique and the advantages of arthroscopy. It is also expected to combine good preliminary results with an incredible opportunity for expert surgeons to hone their skills and broaden their knowledge, particularly where the anterior space and the relations with the brachial plexus are concerned.

Although the technique is standardized, hence reproducible, it should only be performed by surgeons with solid experience in arthroscopy and shoulder surgery. The present data suggest that the learning curve for this approach is roughly 15 cases.

We feel that training in the procedure should be maximized by providing formal fellowships, video training modules, and courses involving operations on cadavers before performance on patients.

Finally, extreme caution is required in analyzing critically the results and the different experiences, establishing whether any complications are related to the early part of the learning curve and to the difficulties of the procedure: this will ensure increasingly safe and effective patient management.