The first arthroscopic Latarjet fixation method introduced by Lafosse et al. [6] required the use of two cannulated screws. However, fixation using metal screws can result in screw loosening and graft non-union. Coracoid graft fracture during screw insertion and impingement of the screw head with the humeral head due to an inappropriate screw angle can also be complications.

Because of these complications, surgeons have devised a procedure using cortical bone fixation buttons. Boileau et al. [10,11] introduced a method of passing two suture strands from the coracoid process to the posterior cortex of the glenoid through the bone tunnel and fixing the strands using the cortical buttons. In a study of 136 patients, Boileau et al. [10,11] reported that bone union was achieved in 95% of patients and that no secondary surgery was required to remove the implant. Xu et al. [12] introduced a procedure using a cortical bone fixation button and an anti-rotation suture anchor; they reported that 98% of 102 patients achieved bone union after an average follow-up of 40.3 months. Similarly, Castricini et al. [13] devised a procedure using four cortical bone fixation buttons.

Controversy surrounds the optimal fixation method for the graft. In a biomechanical study using a cadaver, Provencher et al. [14] reported that there were no significant differences in the maximum tensile strength and average failure strength between the metal screws and the cortical bone fixation buttons. However, Williams et al. [15] reported that the total load at failure and maximum cycle displacement for the cortical bone fixation button were significantly lower than that for a metal screw. Hardy et al. [16] analyzed 236 patients who had metal screw fixation and 72 patients who had cortical button fixation. These researchers found a significantly lower instability recurrence of 2.5% with screw fixation compared to 8.3% in patients fixed with cortical buttons, demonstrating the superiority of metal screws in fixation. However, no patient needed revision surgery with recurrence of instability after using the cortical fixation buttons. Rather, the re-operation rate was higher (5.9%) in patients with metal screws, which was due to complications such as metal screw irritation and protrusion [16]. Therefore, each fixation method has pros and cons and is selected carefully according to the surgeon's preference and the patient's condition.

The patient is placed in a modified beach chair position, and a surgical drape is applied to sufficiently expose the center of the sternum. The patient's upper body angle should be 30°–45°, which is lower than the general beach chair position. We use six arthroscopic portals. As with any other arthroscopy procedure, the arthroscopic Latarjet procedure requires an arthroscopic portal hole in the appropriate location and thorough use of the portal. These are crucial for successful surgery (Fig. 2). Some of these portals are unfamiliar; however, since the unfamiliar ones are anatomically safe, these are made with greater confidence as surgeons gain more experience. Since there are many procedures performed on the medial side of the conjoined tendon, anatomical knowledge is essential before performing surgery. Unlike other arthroscopic procedures, the portals in the arthroscopic Latarjet procedure do not use cannulas and move freely during the operation.

The posterior portal, the arthroscopic examination viewing portal, for the introduction of the arthroscope is made parallel to the articular surface. After the arthroscopic examination, the anterior portal is made on the lateral side of the coracoacromial ligament. Instruments are inserted through the anterior portal; and the rotator interval, remnant anterior labral tissue, and joint capsule are debrided to create sufficient space and prevent interference between the glenoid and the transferred coracoid process. Marking the areas in the 2 and 5 o'clock positions from the expected location of the coracoid process graft attachment using an electric cautery device helps determine the location. Next, the lateral portal is made between the conjoined tendon and the subscapularis tendon in a direction parallel to the upper part of the subscapularis tendon. Ease in checking the location of the coracoid process is achieved by releasing the lateral and inferior surfaces of the coracoid process through the lateral portal using electric cautery.

When the arthroscopy is moved to the lateral portal, trimming the anterior glenoid bone surface by inserting the burr through the anterior portal is easy. This process not only provides decortication for bone union but is also an important task in matching the shape of the inferior part of the coracoid and the anterior glenoid surface to achieve perfect congruency (Fig. 3).

With the lateral portal used as the viewing portal, the inferior portal is created in the long axis of the conjoined tendon, which is near the axillary pouch at the surface anatomical perspective. Using the inferior portal as the working portal, the soft tissue of the lateral and superior surface of the coracoid, including the coracoacromial ligament, is released. Inserting a Wissinger rod into the anterior portal assists in the exposure of the superior surface of the coracoid and in securing sufficient space by sustaining the deltoid and thoracic muscles.

After moving the arthroscopy to the inferior portal, the soft tissues on the superior and inferior surfaces of the coracoid are released as much as possible using the lateral portal as a working portal. Especially because the inferior surface of the coracoid will be attached to the anterior glenoid at a later stage of the operation, decortication can be performed in advance. Afterwards, the medial portal is made using a spinal needle in a position that can be entered in parallel to the glenoid articular surface. Using the medial portal as the working portal, the medial side of the coracoid is exposed. Since the musculocutaneous nerve is located medially to the coracoid, the pectoralis minor muscle is detached with caution so as not to damage the nerve. All soft tissue on the slope of the coracoid is removed, and the superior surface is exposed to the border of the coracoclavicular ligament base. The medial and lateral sides of the conjoined tendon are also released to facilitate the transfer of the coracoid.

After the exposure to the coracoid is completed, a superior portal is formed for the osteotomy. Since the superior portal is used not only for osteotome insertion but also for making a screw hole in the coracoid, the location and size of the portal should be considered carefully. With the inferior portal as the viewing portal, a guide is inserted through the superior portal to select the location to be drilled on the upper surface of the coracoid. Since the length and shape of coracoid processes differ between individuals, the use of preoperative 3D computed tomography is beneficial to determine the direction of the coracoid that can be most congruently attached to the joint when inserting a guide. In addition, not positioning the guide too far is important to avoid distal cortical bone fractures. This is accomplished by identifying the interface between the tendon and the bone. The guide is positioned 7 mm inward from the lateral surface of the coracoid, two proximal and distal guide wires are inserted along the guide, and drilling is performed along the guide wire. After drilling, a top hat is inserted in each hole to prevent the metal screw head from being inserted into the coracoid cortical bone and causing a fracture when pressure is applied during close contact with the anterior glenoid bone. The coracoid holding wire is then inserted; this plays an important role in helping to connect the coracoid positioning double cannula after the coracoid osteotomy. The coracoid holding wire passes through the proximal top hat and the inferior surface of the coracoid, returns to the distal top hat, and is withdrawn via the proximal and distal ends through the medial portal (Fig. 4).

Next, an osteotome is inserted through the superior portal to perform coracoid process osteotomy. If the burr is inserted through the lateral portal, decorticating the area in which the osteotomy is to be made can prevent fracture in an unexpected direction. When the osteotomy is performed completely, the arthroscope can be moved to the lateral portal, and two 3.5-mm coracoid screws are connected to the coracoid thorough a double cannula along the coracoid holding wire to manipulate the coracoid process (Fig. 5). The sharp boundaries of the osteotomized site and the inferior surface of the coracoid are trimmed to fit the shape of the glenoid and decorticated to facilitate bone union.

The lateral portal is used as the viewing portal and the inferior portal as the working portal. The separation location is generally acceptable for the boundary between the upper 2/3 and the lower 1/3 of the subscapularis tendon. However, separating the middle part of the subscapularis muscle does not affect the clinical outcome. Limiting separation to the muscle as much as possible is recommended. However, if the size of the graft is large or if the patient has bulky muscles, separating some tendinous parts of the muscle may be necessary. Since the axillary nerves are located medially, identifying the nerves to prevent damage is helpful. Separating the subscapularis muscle and securing a space through which the coracoid process will be transferred is important. We insert a silastic drain into the anterior portal to draw the superior part of the subscapularis and insert a Wissinger rod into the posterior portal to drag the inferior part downward (Fig. 6).

The coracoid graft connected to the double cannula is transferred through the separated subscapularis muscle to the anterior glenoid transplant site, which was marked previously. The Wissinger rod is inserted into the posterior portal and positioned parallel to the glenoid surfaces to determine the medial and lateral positions of the coracoid graft. Care must be taken not to under-hang or overhang the graft with joint surfaces.

Once the location of the coracoid graft has been determined, each guide pin is inserted through the coracoid screw located in the double cannula for fixation. The guide pin passes through the transferred coracoid graft, the anterior and posterior cortical bones of the glenoid, and the posterior skin of the scapula and is clamped outside the skin. This is to prevent the guide pin from pulling back during the removal of the cannulated drill. Once the guide pin is fixed, the coracoid screw is removed, and drilling is performed using a 3.2-mm cannulated drill bit through a double cannula. The length of the metal screw can be estimated by the depth of the drilling up to the posterior cortical bone. The 3.5-mm cannulated screw is inserted after drilling. After the two metal screws are partially inserted, insertion is completed by alternating compression of the two metal screws so that the coracoid graft can be properly compressed. Both the metal screw insertion and the glenoid and coracoid bone fixation status must be checked through the lateral portal (Fig. 7). This fixation assessment is to ensure that the fixation is parallel to the glenoid surface. After surgery, plain radiographs and computed tomography scans are used to confirm the location of the coracoid graft and the direction of the metal screw (Fig. 8).

Recent studies comparing mid- to long-term clinical outcomes and complication rates of Bankart repair to the Latarjet procedure showed that the Latarjet procedure was equivalent or superior to Bankart repair. According to a meta-analysis by Imam et al. [18], which included 3,275 shoulder joints, the infection rate of patients who underwent the Latarjet procedure was higher than that of those undergoing Bankart repair. However, the risk of recurrence or reoperation rate was higher in the Bankart repair group as the follow-up period lengthened.16 In a long-term follow-up study of adolescent patients with recurrent anterior shoulder dislocation, Waltenspül et al. [19] found that the treatment failure and re-operation rates were significantly higher in the patients who underwent Bankart repair; however, there was no difference in the Constant scores and subjective shoulder values between the two groups. Ernstbrunner et al. [20] comparing patients with a mean age of 47 years reported that there were no differences in subject shoulder values, glenohumeral arthritis grades at the final follow-up, and revision rates; but the rates of re-dislocation or subluxation were higher in the Bankart repair group. Rossi et al.'s study [21] of rugby players with less than 20% glenoid bone defects also found that the Latarjet procedure had lower recurrence and reoperation rates than Bankart repair despite no differences in the range of motion, Rowe scores, Athletic Shoulder Outcome Scoring System scores, and time of return to sports. Rodkey et al. [22] reported that the recurrence of instability was lower in the primary Latarjet group than with the Latarjet revision patients who failed after Bankart repair, emphasizing the importance of the Latarjet procedure as a primary surgery.

The biggest difference between open and arthroscopic Latarjet procedures is that in open surgery, the anterior joint capsule is reconstructed using the coracoacromial ligament. This is not true of the arthroscopic procedure. To compensate for this, one study reported good results by performing anterior capsular reconstruction using anchors in arthroscopic surgery [23]. In the results of biomechanical studies using cadavers to compare these differences, Schulze-Borges et al. [24] showed that the translation was significantly decreased in the abduction position in open surgery compared to the arthroscopic procedure. However, there was no difference between the two groups in the abduction-external rotation position. Kleiner et al. [25] reported that there was no difference in the translation of the shoulder joint regardless of reconstruction of the anterior capsule and that the range of external rotation movement decreased in the group that underwent anterior capsular reconstruction.

Recently, many clinical trials have been reported to show satisfactory outcomes. Zhu et al. [23] reported that graft union was achieved in all 52 patients after an arthroscopic Latarjet procedure with anterior shoulder instability. Dumont et al. [26] also reported satisfactory clinical outcomes in a study with at least 5 years of follow-up. In addition, several prospective studies have reported equivalent clinical outcomes between arthroscopic and open Latarjet surgery, drawing more attention to its effectiveness [27-30]. Other studies comparing the clinical outcomes of arthroscopic and open Latarjet surgery have been reported recently (Table 1) [27-34]. Hurley et al. [31] compared clinical outcomes of 102 patients with an average of 51.3 months of follow-up and reported that there was no significant difference between open and arthroscopic surgery. Ali et al. [32] observed 48 patients with open and arthroscopic Latarjet surgery for an average of 30.5 months and compared the range of motion, strength, visual analog scale (VAS) scores, Rowe scores, and Western Ontario Shoulder Instability (WOSI) scores. Postoperative computed tomography was used to evaluate the Gerber index, Sugaya index, and superoinferior position of the graft bone, as well as screw angle with glenoid surface in axial images (a-angle). Graft resorption and glenohumeral arthritis grade at final follow-up were also evaluated. These researchers reported that internal rotation was significantly reduced, the WOSI score was higher at final follow-up, and the screw tended to be angulated more medially in the arthroscopic Latarjet group. In a prospective study that analyzed 184 patients, Nourissat et al. [29] compared postoperative pain and WOSI scores and reported less pain on postoperative days 3 and 7 and better WOSI scores at 3 months postoperatively in the arthroscopic group. Marion et al. [27] evaluated VAS scores in the first week after surgery, the position of the graft using computed tomography in the third month after surgery, and WOSI scores in 58 patients. This group observed that patients had less pain in the first week after the surgery and a better equatorial position in arthroscopic surgery; there were no significant difference in the WOSI scores between the two groups. Zhu et al. [30] examined 44 patients with open surgery and 46 patients with arthroscopic surgery for at least 2 years of follow-up and evaluated the shoulder function and 1-year postoperative graft resorption. In that study, the coracoid graft was in a more proper superoinferior position in open surgery, but there were no significant differences in other measures. Graft resorption occurred to a lesser extent in arthroscopic surgery, which may be accounted for by the maintenance of soft tissue and blood supply around the coracoid process.

A systematic review of 89 clinical papers, including 7,175 patients, demonstrated that the short-term complication rate of the arthroscopic Latarjet procedure was 6.8%, which was not significantly different from the 6.1% complication rate of open surgery [35]. As for the types of complications, 3.2% of graft-related complications were coracoid fractures or non-unions, 1.9% were implant-related problems, 0.7% were neurological complications, 0.5% were infections, and 0.5% were other complications. However, while the graft-related complication was mainly non-union after the open surgery, arthroscopic surgery was performed mainly due to the fracture of the coracoid graft. This may be due to the technical difficulties of arthroscopic surgery. In a study on the learning curve of the arthroscopic method reported by Kordasiewicz et al. [34], complications such as graft fracture appeared at the beginning of the learning curve. In a systematic review of 35 clinical papers, Cho et al. [36] analyzed the complications of the Latarjet procedure by categorizing them as intraoperative, postoperative, and instability-related. Intraoperative complications such as graft fractures were higher for the arthroscopic procedure, while instability-related complications were higher for open surgery.