Recurrent dislocation of shoulder joint is very common in young sport populations. For the patients with low exercise requirement and no obvious shoulder defect, the simple repair of joint capsule and labrum complex (Bankart) can achieve good clinical results. However, there are still high risks of recurrence and dislocation in patients who love physical activity or (and) suffer from glenoid bone defect. This is a serious challenge for orthopedic and sports medicine physicians. Latarjet surgery is considered as coracoid transposition, which is the coracoid osteotomy performed by passing coracoid bone through subscapularis muscle and screwing it to the anteroposterior margin of scapula. This technique can lead to significant decline of the postoperative recurrence of should joint dislocation. The rate of returning to preoperative exercise and patient satisfaction increase significantly. With the development of arthroscopic technique, Lafosse et al. first performed the Latarjet surgery under arthroscopy in 2007 and achieved good clinical result. Arthroscopic Latarjet has the advantages of less trauma, less bleeding, faster recovery and the ability to observe other diseases in joint. This technique has been promoted and improved by most scholars. Boileau et al. changed the traditional way of coracoid bone fixation with screw and first reported the arthroscopic Latarjet surgery of coracoid fixation using double-cortical buttons. They believe that the use of double buttons fixation can effectively avoid the complications such as poor position of bone due to poor screw implantation angle, bone joint degeneration caused by the loosening of bone due to bone absorption and the injuries of plexus and its branches caused by broken nail and over-internal approach .Whether it is traditional open or arthroscopic screw-fixed Latarjet surgery, a large number of reports has shown that the absorption and non-union of coracoid bone after screw-fixed Latarjet lead to postoperative recurrence or dislocation or shoulder pain, degeneration and other symptoms. Most scholars believe that ideally the coracoid should be fixed below the center of joint plane and flush with the plane of articular surface, or be located slightly inward of glenoid or at least not above the plane of glenoid. The posterior humeral head collides with a bone or screw to cause joint degeneration. However, with the fixation by double-buttons method which is different from the strong fixation of screw, the internal fixation between coracoid bone and joint tendon is flexible. Our previous study found that the arthroscopic double-buttons method Latarjet fixed coracoid bone at a level higher than articular surface. After a year or so of shaping, the bone finally merged with articular surface in parallel. The "pear-shaped" form was restored, which formed an arcuate articular surface consistent with the trajectory of humeral head, and the shoulder joint did not exhibit impact degeneration. Therefore, the purposes of this study were to observe the shape change process of coracoid bone after arthroscopic double-buttons Latarjet surgery and to report a new shape that is different from conventional screw fixation Latarjet.

1. Research methods: A retrospective analysis of 70 patients who underwent arthroscopic double-buttons Latarjet surgery between October 2014 and October 2016.Inclusion criteria: (1) patient who was under 40 and was diagnosed with recurrent shoulder anterior dislocation; (2) preoperative CT showed obvious bone defect in joint. (3) The surgical procedure was arthroscopic double buttons Latarjet; (4) postoperative CT showed that the center of coracoid bone was located at the joint 4:00-5:00; (5) the patient agreed to receive CT examinations preoperatively, immediate postoperatively, 1 month postoperatively, 3 months postoperatively, 6 months postoperatively and 12 months postoperatively; (6) Patient data were complete, and follow-up was at least 12 months. Exclusion criteria: (1) shoulder joint osteoarthritis; (2) history of previous shoulder joint surgery; (3) the center of coracoid bone block with a position above 4:00 or below 5:00 or the bone mass 5mm above joint surface. 2. Preoperative evaluation: The age of initial dislocation, the number of dislocations, the severity of shoulder index (ISIS) and the degree of joint relaxation (Beighton score) were recorded in detail. Shoulder joint function assessment was performed using American shoulder and elbow surgeons (ASES) and Rowe scoring system. Surgical indications: (1) ISIS >3 points; (2) dislocation time >5 times; (3) joint humeral defect >15% or bone defect >10% with Hill-sachs injury. 3. Surgical methods: The specific surgical method of arthroscopic double-buttons Latarjet has been reported in the previous study. 4. Rehabilitation program: After the operation, the shoulder joint was fixed with external rotation 0° neck sling for 3 weeks. Under the guidance of rehabilitation instructor, passive movement exercise of the affected limb was started. After 6 weeks, the sling can be removed, and the daily activities can be gradually restored. However, vigorous resistance and elbow flexion of great muscle strength should be avoided. Full range of shoulder joint activity was gradually restored 3 months after surgery, and physical exercise was gradually resumed 6 months after surgery, and normal exercise level was completely restored 12 months after surgery. 5. Postoperative follow-up plan: The patients returned to hospital for follow-up 2 weeks, 1 month, 3 months, 6 months, 12 months and 18 months after the operation. Afterward, they were followed up once a year. Recurrence, dislocation, subluxation or instability during follow-up was considered as surgical failure. The shoulder function at the end of follow-up was recorded. 6. CT evaluation: Three-dimensional CT examination of bilateral shoulder joints in the same window was performed before the operation to understand the shape of the "pear" joints of shoulder joint and to evaluate the defect of shoulder joint (Figure 1) . According to the relative position of coracoid graft on the transverse section of CT, the population was divided into two groups: Group A, the coracoid graft was higher than the joint surface; Group B, the corcoid graft was level with or lower than the articular surface within 5 mm. Three-dimensional CT examination was performed 3 months, 6 months and 12 months after the surgery, and the shape of coracoid bone was evaluated in the transverse and sagittalsctions . 7. Statistical methods: Clinical data were analyzed using SPSS11.5 statistical software. Values are expressed as mean±standard deviation. Comparison of count data was performed using a multivariate analysis Fisher's exact test (Fisher test) . The count comparison was performed using the student-t test (student-t test) . P<0.05 was statistically significant.

1. Case data: There were 28 patients in the Group A and 42 patients in the Group B. One patient in the Group B group was lost after 6 months. There is no significant difference in age, sex ratio, ISIS score or joint humeral defect between the two groups. 2. The observation of the shape of the coracoid: Immediately after the surgery, CT showed that the coracoid of Group A was higher than the articular surface by (3.4±1.5) mm. On the other hand, the coracoid of Group B was (1.8±1.2) mm below the articular surface. After the operation, the total healing of coracoid graft to the glenoid rim was (4.5±1.5) months for group A and (4.0±1.3) months for group B. There was no case of bone non-union in both groups.According to the CT cross-sectional observation (horizontal direction) , the bone of Group A higher than the articular surface was absorbed and finally formed arc with the concentric surface of articular surface. All achieved bone healing without the formation of shoulder osteoarthritis caused by the impact of umeral head and coracoid. The distance between the outer edge and glenoid of Group B group was slightly absorbed with time with an average of about (0.32±1.1) mm. Finally, the bone was healed without shoulder osteoarthritis.According to the vertical direction of CT three-dimensional reconstruction en-face, the coracoid grafts of Group A and Group B not only healed in the contact area of ??bone but also had osteophytes formed on the upper and lower edges of bone to fill the joints with the extension of time. In the defected area, the bone between final bone and joint was fused. The results of final healing were that the formation of bone tended to fill the center of the "pear" shape and that the excess bone was basically absorbed, forming a morphology similar to the "pear" shape of the healthy side of shoulder joint. 3. Postoperative clinical efficacy evaluation: All patients were followed up for 12-24 months with an average of (14±2.8) months. One patient in Group B had fat liquefaction after the operation, and the wound healed after 2 weeks of dressing change. There was no complication such as wound infection, radial nerve injury or coracoid fracture in both groups. All patients returned to normal life with no dislocation or instability, and the apprehension test and re-dislocation test were negative. 61 patients (90%) were able to carry out vigorous competitive activities. The mean ASES scores of Group A before and at the end of follow-up were (60.8±18.1) and (90.7±15.5) (P<0.01) , and the ROWE scores were (48.4±10.5) and (88.6±17.5) (P<0.01) . The average ASES scores of Group B before and at the end of follow-up were (58.7±13.2) and (85.4±17.8) (P<0.01) , and the ROWE scores were (40.4±9.8) and (87.3±15.4) (P<0.01) .

The form of coracoid graft remodeling in arthroscopic double-buttons Latarjet procedure is different from the traditional screw fixation Latarjet procedure. The absorption of coracoid graft is found when the bone block is placed higher than glenoid rim in a lateral way without impingement and joint osteoarthritis. A lot of bone callus formation around the superior and inferior aspects of coracoid graft makes the bone block to be normal"pear"shaped glenoid rim.