Accounting for 5% of all fractures, proximal humerus fracture is one of the most common fractures for osteoporotic group with an incidence of 90.9/100 000 per year. Its incidence rate increases with age, and approximately 70% of proximal humeral fractures occur in patients over 60 years old. Female accounts for the majority of the affected elderly over 80 years old, as the ratio between men and women is 3:7. With obvious displacement and osteoporosis, bone compression and loss are easily formed after fracture, which result in the loss of the unique anatomical mark of humeral neck. The therapeutic effect of conservative treatment for displaced fracture is unsatisfactory. The displacement even causes great difficulty in plate fixation during surgery, which easily triggers complications such as loss of reduction, ischemic necrosis, screw piercing, etc. For osteoporotic proximal humeral fracture, locking plate fixation has the advantages of multi-angle and stable fixation, which can significantly improve postoperative function and reduce complications. However, the complications caused by locking plate still happen. According to the literature, the rate of humeral head bone loss and screw piercing is still as high as 29% with the bone loss of about 6 mm. Allograft bone transplantation can provide good support for humeral head and promote fracture healing, and the clinical effect is satisfactory. Although locking plate fixation with allogeneic bone graft has been applied for the treatment of senile osteoporotic proximal humeral fracture, the clinical efficacy evaluation and imaging data analysis of different types have been barely reported. Therefore, this article aims to explore the application value of allograft bone transplantation in different types of osteoporotic proximal humerus fracture.

(1) General information. A total of 36 patients (17 males and 19 females) were included in this study, and the age ranged from 65 to 82 years with an average of (71.5±5.5) years. The causes of injury include 25 cases of fall damage, 7 cases of traffic accident and 4 cases of sports injury. According to Neer classification, there were 14 cases of surgical neck or anatomic neck part-2 fracture, 13 cases of part-3 fracture and 9 cases of part-4 fracture. Furthermore, there were 4 cases combined with shoulder joint dislocation, 3 cases combined with rib fracture and 1 case combined with clavicular fracture. All patients were treated with open reduction and locking plate internal fixation. (2) Inclusive and exclusive criteria. Inclusive criteria: ① osteoporosis with dual energy x-ray absorptiometry (DXA) T value <-2.5 SD (according to the diagnostic criteria of WHO, DXA T value <-2.5 SD is regarded as osteoporosis) ;② unilateral proximal fracture with displacement > 1 cm and angulation> 45o; ③ Neer part-2, part-3 and part-4 impacted fracture, s with humeral head bone loss; ④ age over 60 years; ⑤ follow-up time ≥1 year. Exclusive criteria: ① pathological fracture or open fracture; ② combined with neurovascular injury ; ③history of diseases that affects limb function, such as previous operation, chronic arthritis, etc. (3) Surgical procedures. The patients were operated under brachial plexus block or general anesthesia. Deltoid-pectoralis approach was used in 26 cases, while deltoid-splitting approach was used in 10 cases. The removal of periosteum and surrounding soft tissue was minimized during the operation. For simple fracture, the intraoperative traction of affected limb should be conducted appropriately by assistants based on the specific conditions. Meanwhile, the affected limb was operated with internal or external rotation if necessary. The humeral head was raised and applied with 2 Kirschner wires of 2.5 mm above the greater tuberosity. The rotation and displacement of humeral head were controlled by Kirschner manipulation lever to assist with reduction. For complex fracture (especially the comminuted fracture with split greater tuberosity and displaced lesser tuberosity) , the insertions of subscapularis, supraspinatus and infraspinatus were penetrated respectively with 2 non-absorbable sutures, and the fragments of greater or lesser tuberosities were indirectly reduced by the traction of these sutures. According to the displacement direction of humeral head, the reduction was achieved through proper pulling, internal or external rotation of humeral shaft and simultaneous poking of humeral head. When the fracture was reduced with satisfactory shape of humeral head, the sutures could be intersected and knotted temporarily to fix the fracture fragments, and the humeral head was percutaneously placed with 2 Kirschner wires above greater tuberosity. With satisfactory reduction of the inferomedial area of bone cortex, 2 Kirschner wires were inserted percutaneously and obliquely upward from the posterior humeral shaft to the inferomedial area of humeral head for temporary fixation. With satisfactory reduction of fracture under fluoroscopy, the metaphysis defect was evaluated under direct vision. The defect over 5 mm was regarded as the standard for bone graft. A moderate amount of allograft bone was transplanted into the gap between fracture end and humeral head. As satisfactory reduction was confirmed under fluoroscopy once again, the fracture was ultimately fixed with locking plate that was placed 5 mm lateral to the intertubercular sulcus and 5-8 mm inferior to the apex of greater tuberosity. When locking screws were pierced proximally, the depth of drilling was controlled carefully. The screws of proper length were selected based on fluoroscopy, that was, the screws reached to the subchondral bone of humeral head (approximately 6 mm from the articular surface) . The inferomedial area of humeral head was inserted with another oblique screw to protect the humeral head from collapse. The suture holes on plate were penetrated with 3 No.2 non-absorbable sutures before placement. After the plate and screws were fixed, the insertions of subscapularis, supraspinatus, infraspinatus and teres minor were penetrated with non-absorbable suture with the knot tightened. For combined lesser tuberosity fracture, one hollow screw of 3.5 mm was used for fixation. After internal fixation, radiographic examination was performed in multiple perspectives including anteroposterior view and auxiliary view to confirm that the screw was not in joint cavity. Afterward, the incision was irrigated and closed. In this study, proximal humeral internal locking system (PHILOS plate, Synthes, Switzerland) was used in 16 cases; locking proximal humeral plate (LPHP, Zimmer, USA) was used in 14 cases; proximal humeral locking plate (Double medical technology Inc., Xiamen) was used in 6 cases. The drainage tube was routinely placed and removed within 24 hours after operation. (4) Postoperative rehabilitation.The active movements of interphalangeal joints and wrist were encouraged on the 1st postoperative day. The flexion and extension of elbow joint and the shoulder abduction of < 90o were conducted 1 week later. The climbing exercises of upper limb was carried out 2 weeks after operation. The shoulder joint was suspended for immobilization at the position of internal rotation and slight anteflexion and abduction within 4 weeks. The isometric exercises of the rotator cuff started at the 6th week, and gradually transferred to the active exercises. (5) Follow-up evaluation. The postoperative follow ups were conducted at the 2nd week, 1st, 2nd, and 3rd months. The radiographic examination was performed every 2 months in the outpatient visit till the fracture was healed. According to the imaging assessment of fracture healing, the neck-shaft angle and the height of humeral head (the distance between the two parallel lines that were respectively over the superior edge of plate and the top of humeral head and vertical to the humeral shaft) were recorded. One year after operation, the curative effect was evaluated by Constant score, UCLA scoring system and visual ana1ogue scale (VAS) . In the meanwhile, the shoulder joint function and postoperative complications in the last follow up were recorded as well, including internal fixation failure, screw cutting-out, humeral head necrosis, infection, etc.

All patients were followed up for 6 to 36 months with an average of 31.6 months, and 34 patients acquired bone healing. The healing time was (5.5±0.7) months (3-9 months) , and no allograft rejection occurred. Follow up was carried out 1 year after operation. During the last follow up, the neck-shaft angles of part-2, part-3 and part-4 fractures were 129°±5° (121-135°) , 128°±3° (111-140°) and 121°±4° (110-134°) respectively. The neck-shaft angle of part-4 fracture was remarkably lower than those of part-2 and part-3 fractures, and the difference is statistical significant (P <0.01) . The humeral head height losses of part-2, part-3 and part-4 fractures were (1.7±0.4) mm (1.2-2.7 mm) , (1.8±0.3) mm (1.3-3.1 mm) and (1.9±0.5) mm (1.4-3.5 mm) respectively, and no significant difference was observed among three groups (P>0.05) . The VAS scores of part-2, part-3 and part-4 fractures were (1.7±0.8) points (1-3 points) , (1.5±0.5) points (1-3 points) and (1.4±0.3) points (1-3 points) respectively, and the difference between groups did not show statistical significance (P>0.05) . On the other hand , the Constant scores of part-2, part-3 and part-4 fractures were (82.0±4.2) points (52-90 points) , (78.0±3.6) points (57-91 points) and (63.0±3.8) points (55-89 points) respectively, and the UCLA scores were (31.3±2.1) points (27-35 points) , (30.2±1.5) points (26-33 points) and (27.6±1.6) points (23-31 points) respectively. The shoulder joint functions of part-2 and part-3 fractures were remarkably superior to that of part-4 fracture, and the difference had statistical significance (P <0.01) . Seven cases had complications, and the complication rate was 19.4%. The complication rate of part-4 fracture was up to 44%, which was obviously higher than those of part-2 fracture (7.6%) and part-3 fracture (14.2%) . One patient of part-2 fracture had subacromial impingement syndrome, and the function was acceptable. One patient of part-3 fracture had subacromial impingement syndrome as well, and the pain was still tolerable. Another case had the loss of reduction, and the fracture healed 3 months after revision surgery. One case of part-4 fracture was discovered with the penetration of screw into joint cavity, and the screw was removed later. Two cases underwent hemiarthroplasty due to humeral head necrosis. Humeral head collapse occurred in 1 patient, and his family refused reoperation because of the advanced age.

The treatment of osteoporotic proximal humeral fracture with allograft bone transplantation can provide effective support to humeral head, reduce loss of humeral head height and lower related complications. Therefore, the patient can carry out training early and thus acquire the function level closed to preinjury. For part-4 fracture, however, the clinical effect was poor with high rate of complication. The defect of this article is that this research is a simple retrospective study with limited cases. Expansion of sample size, prolonged follow-up and randomized trial are required for further validation of our conclusion.

In 1954, Latarjet firstly reported the open procedure of coracoid transfer with conjoin tendon to anteroinferior glenoid for the treatment of anterior shoulder instability. This technology on the one hand, filled in the glenoid bone defect and restored its size; on the other hand, the conjoin tendon and part of the split subscapularis had "sling" effect and enhanced the obstacle of anteroinferior soft tissue, which significantly reduced the recurrence rate of shoulder dislocation.With the continuous improvement of arthroscopic technology, more and more literatures about arthroscopic Latarjet procedure have been reported recently. Compared with traditional open urgery, arthroscopic procedures have such advantages as small invasion, little blood loss, fast recovery, satisfactory curative effect, etc. However, the arthroscopic technique requires a long learning curve and has some certain difficulties. One is that the arthroscopic operation has multiple approaches, especially the medial approach. The biggest potential risk for the complication of nerve injury is that the approaches are too many and too close to the medial side, especially when the course of brachial plexus have mutation. The second is that it is difficult to fix the coracoid fragment parallel to the glenoid surface with screws. This often lead to the poor position of fragment or the impact of screw head on humeral head cartilage, and even cause complications such as screw loosening or withdraw, fragment fracture, fragment nonunion or absorption, etc. In order to simplify the arthroscopic Latarjet procedure, reduce the complications and improve the clinical efficacy, we adopted the modified arthroscopic Latarjet procedure of double Endobuttons anti-rotation fixation with three approaches for the treatment of patients with recurrent anterior shoulder dislocation of severe bone defect. The fragment position and healing was observed by three-dimensional CT scan. We followed up the clinical efficacy and achieved satisfactory results.

(1) Operation steps.①Coracoid osteotomy and preparation. After general anesthesia, the patient was in beach chair position with routine disinfection and draping. A longitudinal incision of 3 cm was made in the center of coracoid. The coracoid was exposed to cut off the coracoacromial ligament and most of the pectoralis minor muscle. The coracoid was osteotomized for 2 cm to facilitate trimming, and the coracoid pedicle was left 3-4 cm to prevent its impact with axillary nerve. The 3.0 mm and 2.4 mm bone tunnels were drilled respectively at a distance of 6 mm, and the distal pedicle bone tunnel (2.4 mm) was penetrated with PDS suture. The Endobutton (Smiths & Nephew, USA) was placed on the surface of proximal pedicle bone tunnel (3.0 mm) with 3 high tensile-strength sutures penetrated through the central 2 holes to set aside. The other high tensile-strength suture was penetrated through the lateral holes to tie the knot with the suture from anti-rotational anchor. After trimming the coracoid process, the incision was partially closed with 8 mm gap left for A approach which was located right at the anterosuperior side of subscapularis tendon.②Subscapularis split and glenoid bone bed preparation. The lateral approach C was established under arthroscopy through the standard posteior approach (B) . The glenoid bone defect and Hill-Sachs lesion were assessed after joint exploration, and a 2.5 mm hole was drilled on the glenoid for positioning at the direction of 3:00. The C approach was used for observation, and the A approach was used for planer and radiofrequency coblation. After the positioning of switch rod through B approach, the subscapularis was split at the same level of glenoid and the axillary nerve was exposed for protection. The glenoid was observe through the subscapular incision, and then the bone bed of glenoid was refreshed as well.③Transposition and fixation of coracoid fragment. A double cylinder locator of 8 mm was inserted through the switch rod. This locator was 120 mm in length and consisted of 2 hollow titanium alloy tubes at a distance of 5-10 mm which were connected by the titanium alloy of 3 mm thick. One of the hollow tube was 4.5 mm in diameter and used for switch rod insertion, while the other tube was 2.5 mm in diameter and used for the 2.4 mm guide pin insertion. The design and use of this locator were simple and safe. Under the guidance of switch rod, the glenoid was drilled at the position of 5:00 with a guide pin of 2.4 mm after the preoperative CT measurement of its anteroposterior diameter. Then an Endobutton drill of 4.5 mm was used to expand the hole, through which the PDS suture was penetrated to set aside. A Twinfix anchor (Smiths & Nephew, USA) of 2.8 mm was implanted parallel and superior to the bone tunnel of the glenoid to prevent fragment rotation.The small incision of coracoid osteotomy was open to pull out the PDS suture as well as the suture of anchor. After that, the suture of anchor penetrated through the distal bone tunnel. Under the guidance of PDS suture, the 3 high tensile-strength sutures from the preset Endobutton were pulled out through B approach. The coracoid fragment was pulled into the joint and attached to the glenoid. The other Endobutton was penetrated with high tensile-strength sutures, and it was tied and fixed through B approach. The high tensile-strength sutures and the suture of anchor were tied for fixation with Double-Pulley technique to prevent Endobutton rotation.Finally, the B approach was used to observe the relationship between the surfaces of fragment and glenoid. If necessary, the fragment should be shaped to match glenoid. The articular capsule was not sutured. (2) Rehabilitation plan.The shoulder was immobilized with brace in 30° of abduction and 0° of external rotation for 2 weeks, and in the meanwhile, the early active exercises of hand, wrist and elbow were encouraged. The passive pendulum exercise of shoulder joint was allowed 2 weeks later. With the removal of brace, the active rehabilitation training was conducted under the guidance of physicians. The active exercise of biceps brachii muscle with high intensity was forbidden with 6 months after operation. (3) Postoperative evaluation.The position of coracoid fragment was observed by three-dimensional CT scan. From the front view of glenoid, the best position was where the center of fragment was at the direction of 4:30. The range of center was from 2:00 to 5:00, and the center was considered high if it was higher than 3:00 and low if it was lower than 5:00. From the cross-sectional view of CT scan, the best position was where the fragment was even with glenoid or below <5 mm. The fragment was considered to be medial if it was below >5 mm and lateral if it was higher than the glenoid. If the patient complained mainly about the feelings of shoulder instability such as shoulder laxity or dislocation, etc. or both the Apprehension test and the Relocation test were positive, the operation was considered a failure. The shoulder joint function was evaluated by rating scale of the American shoulder and elbow surgeons (ASES) , ROWE score and Walch-Duplay score. The operative time was recorded as well. (4) Statistical analysis.The data were expressed as mean value±standard deviation, and the software of SPSS 18.0 was used for statistical analysis. The difference was statistically significant (P <0.05) .

From October 2014 to October 2016, a total of 50 patients with recurrent shoulder dislocations of severe bone defects received arthroscopic Latarjet procedure of double Endobuttons anti-rotation fixation with three approaches. There were 39 males and 11 females (45 left shoulders and 5 right shoulders) with an average age of (27.2±2.1) years (15-45 years old) . The average time from primary dislocation to surgery was (24.3±12.7) months (6-120 months) . The average percentage glenoid bone defect was (20.1%±2.9%) (18%-39%) . The average episodes of shoulder dislocations was (16.1±2.4) times (3-40 times) . There were 4 revision cases, including 3 recurrent cases after Bankart repair and 1 recurrent case of multiple directions instability (MDI) after capsule constriction. The operation time ranged from 57 min to 150 min with an average of (105±38) min. (1) Postoperative posititon of coracoid bone fragment. Glenoid surface observation: according to the standard of fragment center at the position of 4:30, there were 45 cases of good position, 2 cases of high position and 3 cases of low position. Cross-sectional observation: there were 43 cases of good position, 1 case of medial position and 6 cases of lateral position. It was worth noting that the coracoid fragments which were superolateral to the glenoid in 6 patients had been moulded and become even with the glenoid during CT examination of 6 months after operation. (2) Postoperative absorption and healing of coracoid fragment.There were different levels of coracoid fragment absorption, and the stabilization time was 2-6 months with an average of (3.3±0.6) months after operation. The absorption ratio was 15%-60% with an average of (27.5±3.8) %. Thereinto, 8 cases had a ratio of <20%, 25 cases had a ratio of 20%-30%, 35 cases had a ratio of 30%-40% and 5 cases had a ratio of >40%. All the patients achieved fragment union 6 months after operation or revision without soft tissue healing or partial healing. (3) Postoperative clinical efficacy evaluation.The patients were followed up for 6-24 months with an average of (13.2±3.6) months. In the preoperative and final follow-ups, the mean ASES score was (80.2±16.2) points vs. (95.2±5.6) (P <0.05) , the mean ROWE score was (40.2±9.8) vs. (94.5±2.7) (P <0.05) , and the mean Walch-Duplay score was (67.5±10.2) vs. (95.6±3.2) (P <0.05) . All the patients returned to normal life, and compared with pre-operation, 45 (90%) patients had more intense exercises. No recurrence of shoulder dislocation or instability occurred, and the Apprehension test and the Relocation test were all negative.

The modified arthroscopic Latarjet procedure of double Endobuttons anti-rotation fixation with three approaches was adopted for the treatment of recurrent anterior shoulder dislocation with severe glenoid bone defect. With this technique, not only were the operation process and surgical instrument simple, but also the coracoid fragment was in accurate position with high healing rate. As this technique reduces the majority of arthroscopic approaches and reduces the risk of potential nerve injury, it can be a reliable and effective alternative to traditional Latarjet procedure.