A review of bone grafting techniques for glenoid reconstruction

Abstract

Background

Traumatic anterior shoulder dislocations can cause bony defects of the anterior glenoid rim and are often associated with recurrent shoulder instability. For large glenoid defects of 20–30% without a mobile bony fragment, glenoid reconstruction with bone grafts is often recommended. This review describes two broad categories of glenoid reconstruction procedures found in literature: coracoid transfers involving the Bristow and Latarjet procedures, and free bone grafting techniques.

Methods

An electronic search of MEDLINE and PubMed was conducted to find original articles that described glenoid reconstruction techniques or modifications to existing techniques.

Results

Coracoid transfers involve the Bristow and Latarjet procedures. Modifications to these procedures such as arthroscopic execution, method of graft attachment and orientation have been described. Free bone grafts have been obtained from the iliac crest, distal tibia, acromion, distal clavicle and femoral condyle.

Conclusion

Both coracoid transfers and free bone grafting procedures are options for reconstructing large bony defects of the anterior glenoid rim and have had similar clinical outcomes. Free bone grafts may offer greater flexibility in graft shaping and choice of graft size depending on the bone stock chosen. Novel developments tend towards minimising invasiveness using arthroscopic approaches and examining alternative non-rigid graft fixation techniques.

Introduction

Anterior glenohumeral instability is common with an incidence of approximately 23.9 per 100,000 person-years. ¹ Traumatic shoulder instability often results in soft-tissue injury of the glenoid labrum, known as a Bankart lesion. However, the traumatic impact forces during a dislocation can cause osseous defects of the glenoid (bony Bankart lesions) and humeral head (Hill-Sachs lesion). ² Glenoid defects have been observed in up to 41% ³ of patients with initial dislocations and 50% ⁴ to 86% ³ in cases of recurrent instability. Hill-Sachs lesions are found in 65% ⁵ of first-time dislocations and approximately 90%^3,5 of recurrent dislocations.

Bony defects of the glenoid can vary from small mobile bony fragments to larger glenoid fractures having undergone long-term resorption of the bone fragment. Substantial damage to the osseous architecture (20–30% of the glenoid) is more likely with recurrent shoulder dislocations or higher impact forces. Failed shoulder stabilisation surgery has been attributed to glenoid defects without an existing bony fragment that are too large to adequately be managed by Bankart repair alone.^6,7 Management of bone loss should consider the size and type of bone defect as well as patient characteristics and functional requirements. Previous authors have discussed the management of recurrent shoulder instability and the glenoid reconstruction techniques available.^8,9 This article expands on existing knowledge and reviews current concepts and novel modifications of existing techniques. We intend to describe the two broad categories of glenoid reconstruction procedures recommended for large bony defects of the glenoid rim (20–30%) without a remaining bony fragment: coracoid transfers and free bone grafting techniques. We explore the distinct characteristics and novel modifications of these techniques and highlight the rationale behind these procedures.

Biomechanics of a glenoid defect

The native glenohumeral joint is highly mobile due to the size disparity between the humeral head and the glenoid fossa. Shoulder stability is mainly provided by a concavity-compression mechanism where the rotator cuff and capsuloligamentous complex pull the humeral head towards the centre of the shallow glenoid fossa. ¹⁰ Glenoid defects impair this mechanism by shortening the effective glenoid arc and reducing articular surface area. The ball and socket morphology of the joint is impaired, and this substantially lowers translational forces required for dislocation.^6,11

Several studies have established a strong inverse relationship between glenoid defect size and shoulder stability.^6,12,13 A biomechanical study by Itoi et al. ⁶ tested the stability of Bankart repaired cadaveric shoulders under different glenoid defect sizes. For glenoid defects with an average width greater or equal to 6.8 mm (21% of glenoid length), there was a significant decrease in the translational force required to displace the humeral head. A defect width 21% of the glenoid length was concluded to cause instability despite soft-tissue stabilisation. In a clinical study of 194 consecutive arthroscopic Bankart repairs, Burkhart and De Beer ⁷ described critical size glenoid defects appearing as an “inverted pear” shape where the inferior glenoid had a smaller diameter than the superior glenoid. At mean follow-up of 27 months, they reported a 67% failure rate for glenoids with critical size defects compared to 4% for glenoids without significant bone loss. Burkhart and De Beer ⁷ concluded that patients with an inverted pear shaped glenoid should undergo a Latarjet procedure rather than arthroscopic Bankart repair alone. Defects of 20% to 30% (more than 6 mm to 8 mm bone loss) have been considered the critical threshold that require glenoid reconstructive procedures.^6,11,12,14

Smaller glenoid defects less than 15% to 20% (approximately 5 mm to 7 mm bone loss) can be successfully treated with soft tissue Bankart repair alone or managed conservatively through immobilisation and physiotherapy. This may be suitable for individuals with low risk for recurrent dislocation such as elderly. However, individuals with higher demands may have poorer functional outcomes if small glenoid defects are managed with soft tissue repair alone.^7,14 Shaha et al. ¹⁵ assessed 73 consecutive arthroscopic labral repairs with smaller glenoid bone defects. The mean WOSI (Western Ontario Shoulder Instability Index) score was significantly lower (p = 0.04) for shoulders with ≤ 13.5% bone loss compared to shoulders with >13.5% bone loss (489 and 1014 respectively). Shin et al. ¹⁶ also found significantly worse postoperative functional scores (ASES and SANE) and recurrence rates for defect sizes ≥17.3% of the longest anteroposterior glenoid width. The authors concluded that additional procedures, for example glenoid bone augmentation or Remplissage, should be performed with Bankart repairs for high-risk populations such as athletes and young males.

The existence of Hill Sachs lesions and the concept of glenoid track is another component that influences the critical size threshold and biomechanics of glenoid defects. When the arm is extended and externally rotated, the path traced by the humeral head as it moves along the glenoid creates an area of contact defined as the glenoid track. ¹⁷ Yamamoto et al. ¹⁷ established that a Hill Sachs lesion which overlaps the medial edge of the glenoid track (“off-track” lesions) may engage with the anterior glenoid rim, increasing risk of dislocation. In the setting of anterior glenoid defects, this glenoid track width is decreased due to the shortened glenoid arc. As a result, even small glenoid defects may convert an “on-track lesion” to an “off-track lesion” and further impair glenohumeral stability.^17,18 Arciero et al. ¹⁹ compared the effect of a combined bony Bankart and Hill Sachs lesion biomechanically on 21 cadaveric shoulders. After a Bankart repair of glenoids with a 2 mm osseous defect, the mean force to translate the humeral head 10 mm was 54 ± 13 N. With the addition of a Hill-Sachs lesion, this force decreased significantly to 45 ± 17 N, showing glenoid defects as little as 2 mm may compromise Bankart repairs when bipolar bone loss is present.

Methods

An electronic search of Medline and PubMed databases was conducted using MeSH search tags and key words to find articles describing glenoid reconstruction techniques using coracoid transfers and free bone grafts. The following MeSH search tags were used: “Shoulder Joint/surgery”, Joint instability/surgery”, “Bone Transplantation/methods”, “Glenoid Cavity/surgery”. In addition, the following key words were used: “glenoid defect”, “bony bankart lesion”, “shoulder instability”, “free bone graft”.

Publications were selected based on primary original descriptions of glenoid reconstruction techniques or primary substantial novel modifications of existing techniques. All titles and abstracts were reviewed for relevancy. Non-English articles were excluded. Review articles were not included but references of reviews and references of pertinent articles were examined to target other additional publications not identified in the primary search.

Results

Publications describing glenoid reconstruction techniques found during the search are shown in Table 1.

Table 1.

Original papers and substantial novel modifications of existing glenoid reconstruction techniques found during search of Medline and PubMed databases.

Primary author	Year	Bone graft / procedure	Modifications
Latarjet 20	1954	Coracoid transfer	None – original description
Helfet 21	1958	Coracoid transfer	None – original description
Lafosse et al. 22	2007	Coracoid transfer	Yes – arthroscopic
de Beer 23	2009	Coracoid transfer	Yes – graft orientation
Boileau et al. 24	2016	Coracoid transfer	Yes – arthroscopic, screw free fixation
Xu et al. 25	2020	Coracoid transfer	Yes – arthroscopic, screw free fixation
Warner et al. 26	2006	Iliac crest bone graft – free bone graft	None – original description
Auffarth et al. 27	2008	Iliac crest bone graft – free bone graft	Yes – graft shape, screw free fixation
Provencher et al. 28	2009	Distal tibia allograft – free bone graft	None – original description
Anderl et al. 29	2012	Iliac crest bone graft – free bone graft	Yes – arthroscopic, graft shape, screw free fixation
Kraus et al. 30	2014	Iliac crest bone graft – free bone graft	Yes – arthroscopic
Tokish et al. 31	2014	Clavicle – free bone graft	None – original description
Sanchez et al. 32	2017	Acromion – free bone graft	None – original description
Ogimoto et al. 33	2018	Femoral condyle – free bone graft	None – original description
Smith et al. 34	2018	Proxy glenoid graft – free bone graft	Yes – arthroscopic, screw free fixation

Note: Six publications described coracoid transfers and its modifications. Nine publications described free bone graft procedures and its modifications.

Coracoid process transfers

The original technique described by Latarjet ²⁰ used an open deltopectoral approach to expose the coracoid process. The technique involved an osteotomy of coracoid process between the coracobrachialis and pectoralis minor muscle insertions. The coracoid process was transferred to the anterior glenoid rim through a vertical split in the subscapularis tendon and fixed with a single biocortical screw. ³⁵ Lafosse et al. ²² described an arthroscopic modification of the Latarjet procedure. Drilling and osteotomy of the coracoid process were performed in vivo, and fixation was achieved with two screws to the anterior glenoid. De Beer et al. ²³ described the congruent arc Latarjet modification where the coracoid process was rotated about its axis by 90° and fixation achieved with two biocortical screws (Figure 1).

Figure 1.

Illustration of coracoid graft variations in common coracoid transfer procedures before fixation to the anterior glenoid defect. (a) The graft for the Bristow procedure using the tip of the coracoid process. (b) The graft for the classic Latarjet procedure. (c) The graft for the congruent-arc Latarjet showing a rotation of the classic Latarjet coracoid process graft to maximise articular surface area of the glenoid after repair.

The Bristow procedure, described by Helfet ²¹ , is similar to the Latarjet procedure but used only the tip of the coracoid process and graft fixation was made with sutures instead of a screw. Graft fixation using sutures has also been described by Boileau et al. ²⁴ using a double cortical button technique (Figure 2). Xu et al. ²⁵ proposed a screw-free, single cortical button technique that facilitated graft fixation to the glenoid defect with knotless suture anchors. Graft rotation was minimised in this technique using an additional anti-rotation suture passed superiorly through the cancellous portion of the bone graft.

Figure 2.

Illustration of the cortical button fixation technique described by Boileau et al. ²⁴ The coracoid process was osteotomised and transferred to the anterior glenoid. The entire glenoid was drilled through to allow passage of a 4-strand suture (dotted blue lines). The anterior button has a pegged outlet to avoid damaging the graft with sutures. A sliding knot (Nice-Knot) is tied at the posterior button (marked with red cross) and sutures were tensioned to assist bone-to-bone compression.

Free bone graft procedures

Warner et al. ²⁶ described an open technique for glenoid reconstruction using a tricortical wedge-shaped graft from the iliac crest. A 3 cm long and 2 cm wide graft was harvested and contoured to match the glenoid curvature before fixation to the glenoid using two or three cannulated screws. Auffarth et al. ²⁷ modified this technique by contouring a 1.5 cm long and 1.2–1.5 cm wide iliac crest graft into a J-shape. The keel of the graft was impacted into a crevice created by a chisel just medial to the glenoid rim, allowing for a screw-free method of fixation.

Anderl et al. ²⁹ further proposed an arthroscopic variation to the screw-free glenoid reconstruction with a similar sized J-shaped iliac crest bone graft. This modification required two 1.6 mm pins to be inserted on the short limb of the graft to allow attachment of a graft impactor for graft fixation arthroscopically. Kraus et al. ³⁰ described an arthroscopic iliac crest bone grafting technique using four arthroscopic portals. Graft insertion required temporary removal of the anteroinferior portal cannula, and enlargement of the skin incision by approximately 1 cm. Fixation was achieved with two biodegradable screws.

Provencher et al. ²⁸ described an open glenoid reconstruction technique using fresh osteochondral distal tibial allografts. The lateral third of the distal tibial plafond was used as the graft and contoured to match the glenoid shape. The graft was fixed to the glenoid using two 3.5 mm cortical screws.

Sanchez et al. ³² discussed the feasibility of a J-shaped graft harvest from the acromion. The study utilised 40 cadaveric shoulders with 20 acromial harvests done via a lateral approach and 20 via a posterior approach. In 100% of the cases, a graft was successfully harvested, measuring at least 1.5 cm long, 1.5 cm wide and 0.6 cm high.

A pilot study by Tokish et al. ³¹ first described the distal clavicle as a potential graft. Eight patients were selected for a pilot study with an initial diagnostic arthroscopy performed to confirm a >20% osseous defect size. A 3 cm horizontal incision over the acromioclavicular joint allowed the harvest of the distal 6–8 mm clavicle which was fixed to the glenoid with either suture anchors or a 3.75 mm screw.

A case report by Ogimoto et al. ³³ described an arthroscopic glenoid reconstruction technique using an osteochondral autograft from the femur. A 20 mm long, 12 mm wide and 7 mm deep bone graft with attached cartilage surface was harvested from the contralateral femoral condyle and contoured to fit the native glenoid shape. The graft was attached to the anterior defect with two Herbert screws.

Smith et al. ³⁴ developed the buckle-down technique as a novel screw-free graft fixation procedure in their biomechanical study. This technique used a total of three suture anchors with two positioned at the glenoid defect site for graft fixation and the third “off-site anchor” placed 10 mm superiorly at the glenoid neck. A four-hole cortical button was used to facilitate the organisation of the suture threads for graft attachment (Figure 3).

Figure 3.

Illustration of the buckle-down fixation technique described by Smith et al. ³⁴ Three suture anchors were used for graft fixation. Two anchors were placed at the defect site with an off-site anchor placed superiorly to augment and maintain suture tension in the repair construct. Organisation of suture strands was achieved through using a four hole endobutton.

Discussion

The non-anatomic nature of coracoid transfers

Coracoid process transfer procedures were hypothesised to restore glenohumeral stability via three factors: (1) the bone block itself, (2) a sling effect from the conjoint tendon and (3) ligamentous reinforcement by the coracoacromial ligament after repairs. The bone block effect physically increases the glenoid concavity and articular surface area at time zero with the shape and positioning of a bone graft contributing most to the increased stability.^36,37 The sling effect and the role of the coracoacromial ligament were investigated by Yamamoto et al. ³⁸ in a biomechanical study of eight shoulders. The authors found that the conjoint tendon formed a buttress to support the humeral head and prevent anterior displacement at mid-range and end-range arm positions. With the arm in maximum external rotation and 60° abduction, approximately 76% to 77% of the restored stability was due to the sling effect with the remaining 23% to 24% attributed to the coracoacromial ligament from the capsular repair. These findings note a commonly discussed rationale for coracoid transfer procedures. Hypothetically, the Latarjet procedure may have greater stabilisation potential due to the conjoint tendon effect, particularly at apprehensive positions of the arm.

Biomechanical evidence also describes the importance of the bone block itself in restoring stability. Montgomery et al. ³⁶ investigated the effects of the bone block effect on four cadaveric shoulders in a biomechanical study. The study measured balance stability angles, the maximum angle between a transducing force and glenoid centreline before dislocation occurs, with a larger angle corresponding to greater joint stability. The average balance stability angle for a 21% defect glenoid was 14° ± 2°. After repair with a bone graft, this increased significantly to 31° ± 9° (p = 0.033) for an 8 mm uncontoured graft and 46° ± 4° (p < 0.0001) for an 8 mm contoured graft. This study concluded that stability can be restored with free bone grafting procedures alone and the role of additional tendinous attachments in non-anatomic coracoid transfers may be supplementary and not a compulsory factor in treatment selection.

Clinically, Moroder et al. ³⁹ conducted a prospective randomized trial of 60 patients and found comparable clinical WOSI scores and radiological outcomes between an open Latarjet and iliac crest graft transfer at 6 -, 12 - and 24-months follow-up. There were no significant differences in strength and range of motion except for diminished internal rotation in the Latarjet cohort. These findings indicate that the role of the conjoint tendon and the sling effect offered by the Latarjet procedure may not be as important as previously thought. Notably, iliac crest grafts have also been used as a revision procedure after a failed Latarjet. ⁴⁰ The non-anatomic nature of coracoid transfers has made revision procedures challenging due to scarring and distortion of anatomy around the subscapularis muscle and brachial plexus. For patients at high risk of recurrent dislocations, such as younger populations or people with greater functional demands, free bone graft procedures are an option to consider as they provide greater anatomical resemblance to the native glenohumeral joint.

Complications of bone grafting procedures

A systematic review by Griesser et al. ⁴¹ of 1904 shoulders found a complication rate of 30% after mean follow-up of 6.8 years. Poor positioning of the coracoid graft has been reported in 15% ⁴² to 36% ⁴³ of surgeries and has been identified as a key factor that leads to these unsatisfactory outcomes.^40,42 Lateral overhang of the graft has been identified as a risk factor for development of arthritis. In a long-term retrospective study of 68 shoulders averaging 20 years postoperatively, Mizuno et al. ⁴⁴ observed that 16 cases had new arthritis or progression of preoperative arthritis. Of those 16 cases, 44% had a lateral overhang of the coracoid graft. For shoulders without new or progression of arthritis, only 4% had a lateralised graft (p < 0.001). A graft placed too medially has been found by Hovelius et al. ⁴⁵ to increase the risk of recurrent instability by diminishing the bone block effect. Out of seven shoulders with the graft placed too medially (at least 10 mm medial to the rim), two had subluxated and two had re-dislocated at minimum two years follow-up. This prevalence was significantly greater than shoulders where the graft was placed less than 10 mm medial to the rim (p < 0.01). A biomechanical study by Ghodadra et al. ⁴⁶ on 12 cadaveric specimens found that glenohumeral contact pressures were restored closest to normal when grafts were placed flush. At 2 mm lateral placement, peak contact pressure in the joint increased to 250% of normal particularly at the posterosuperior quadrant of the humerus with the authors noting this posterior concentration of force potentially increasing risk of arthritis. At 2 mm medial placement, the authors noted no significant difference in contact pressure compared to a defect glenoid, suggesting inadequate restoration of the glenoid arc. It is thus recommended that the optimal position for a coracoid bone graft was one placed flush against the glenoid surface.

Complications arising from screw usage for graft fixation have also been commonly reported. Griesser et al. ⁴¹ found that out of 132 patients that required an unplanned revision surgery, 35% of these cases were due to hardware-related complications such as loose, broken migrated or painful screws. Inappropriately sized screws can cause shoulder pain, with the screw head rubbing against the subscapularis muscle or impinging with the humeral head. ⁴⁷ Poor screw orientation can also increase risk of neurovascular injury. In an anatomic study, Lädermann et al. ⁴⁸ noted that a graft placed too superiorly would risk iatrogenic injury to the suprascapular nerve due to close proximity (around 4 mm) with the superior screw. Screws that are positioned too inferior reduces the space available to securely grip into the glenoid and may result in bone block non-union due to rotational instability. ⁴⁷

The use of a single screw for graft fixation was historically thought to reduce rates of impingement and neurovascular complications. ⁴⁹ Single screw fixation techniques have also been used when the osteotomised graft size was too small to accommodate two screws. A study of single screw fixation by Schroder et al. ⁵⁰ in 52 shoulders found two patients required revision surgery to remove hardware due to impingement. A larger study by Hovelius et al. ⁵¹ in 118 patients found a single case that required removal of the screw due to close proximity to the glenohumeral joint. A systematic review of 24 studies by Cowling et al. ⁵² found that rates of revision procedures for single screw fixation techniques were lower than two screws fixation techniques (1.3% and 5.2% respectively). However, single screws were more prone to loosening compared to two screws (2.0% and 1.5% respectively). Although neurovascular injury is less commonly reported with single screw fixation techniques, graft fixation appears more compromised. Recurrent instability rates as high as 15.4% have been observed. ⁵⁰ Cowling et al. ⁵² noted studies using two screws had overall mean rate of loosening rate of 1.5% compared with 2.1% for single screws. Hardware breakage was also higher with single screws (0.4%) compared to fixation with two screws (0.2%). Graft non-union in the Latarjet procedure occurs in up to 9.1% ⁴¹ of cases and have been seen more commonly in single screw fixation techniques or when poor decortication and flattening of the graft and glenoid surface occurred intraoperatively. ⁴⁷

Graft resorption and osteolysis are a complication of bone grafting techniques that are usually not associated with a clinical recurrence of instability or poorer functional outcomes. ⁴⁷ Willemot et al. ⁴² noted graft resorption was the main failure mechanism in 6 out of 21 patients who underwent a Latarjet procedure and was only diagnosed after recurrent instability prompted radiologic investigations. Di Giacomo et al. ⁵³ noted a higher mean osteolysis rate of 59.5% in a study of 26 Latarjet procedures but found no patients had post-operative recurrent dislocations, pain, stiffness, or subtle instability. Bone resorption may be explained as a radiological finding post-operatively with little clinical impact and not requiring specific management. Coracoid graft resorption has been reported to occur significantly less in patients with anterior glenoid bone loss than compared to those without glenoid bone loss.^40,53 This suggests that the sling and capsule effect of coracoid transfers may assist in restoring stability for patients with smaller glenoid defects.

In a study of 10 patients, Boehm et al. ⁵⁴ found substantial resorption of all iliac crest bone grafts postoperatively. The authors found an increase of glenoid defect area from 0.6% after surgery to 14.0% at one year. There was a corresponding decrease of glenoid surface area from 118.4% to 86.6%, representing total resorption of the bone graft. Rate of graft resorption with distal tibial allografts varies. Provencher et al. ⁵⁵ noted a 3% distal tibia allograft lysis rate at average of 45 months follow-up in 27 patients. Amar et al. ⁵⁶ conducted follow-up CT at 6.3 months of 42 patients and found 13 patients had <50% graft resorption and 5 patients had >50% resorption. Graft osteolysis and resorption rates for clavicle, acromion and femoral condyle bone grafts have not been reported. Substantial graft resorption may result in implant complications such as screw heads becoming prominent, resulting in impingement and pain. Avoiding complete devascularisation of bone grafts during the procedure may lower this risk.^47,53

Novel modifications

Arthroscopic approaches to glenoid reconstruction have been described.^{22, 24, 25, 29, 30} The proposed advantages of an arthroscopic approach compared to an open approach include less pain, quicker rehabilitation and improved cosmetic outcome. Lafosse ⁵⁷ argued that bone graft positioning was more accurate due to the multiple views offered by the arthroscope. In his arthroscopic Latarjet technique, Lafosse ⁵⁷ reported no cases of recurrent dislocation in the first 100 shoulders at 26 months follow-up with patients returning to work within one month and returning to sport in 10 weeks (Table 2). A five-year minimum follow-up study of 62 patients by Dumont et al. ⁵⁸ described the long-term outcomes of the arthroscopic Latarjet procedure. The authors found a recurrent instability rate of 1.6%. Eight patients returned due to hardware complications to have screws removed and one patient required total shoulder arthroplasty for arthritis. With an arthroscopic technique, concurrent pathologies, such as labral tears, can also be more easily treated compared to an open approach. However, despite these theoretical advantages, systematic reviews have found no differences in rates of recurrence or instability between an arthroscopic and open approach.^{59, 60} There exists a steep learning curve with Bonnevialle et al. ⁶¹ estimating it requires up to 30 cases for surgeons to become proficient with arthroscopic Latarjet procedures. Arthroscopic techniques may still be warranted with patients who expect a minimally invasive surgery and may continue to increase in demand.

Table 2.

Summary of complications and outcomes of clinical studies for selected original descriptions of glenoid reconstruction techniques.

Primary author	Technique summary	Follow-up	Patients, n	Recurrent instability rate and complications	Return to work/sport	Patient reported outcomes/scores
Lafosse et al. 22	Coracoid transfer – arthroscopic	At 18 months and 26 months	100	No recurrent dislocations Early: 2 haematomas, 1 graft fracture, 1 transient nerve palsy Late: 4 non-union, 3 graft lysis around screws, 4 screw removal	Mean return to work: 1 months Mean return to sport: 10 weeks	At 18 months: 80% – excellent 18% – good 2% – disappointed At 24 months: 91% – excellent 9% – good
de Beer 23	Coracoid transfer – graft orientation (congruent arc Latarjet)	Mean: 59 months (range: 32–108 months)	102	4 recurrent dislocations, 1 recurrent subluxation	N/A	Mean constant: 94 Mean Walch-Duplay: 92
Boileau et al. 24	Coracoid transfer – arthroscopic, screw free fixation	Mean: 14 months (range: 6–24 months)	76	1 traumatic subluxation 7 graft non-union No hardware failures/graft migration	93% returned to preinjury level of sport at follow-up	Mean Rowe: 95 (range: 84–100) Mean Walch-Duplay: 96 (range: 86–100)
Xu et al. 25	Coracoid transfer – arthroscopic, screw free fixation	Mean: 40.3 ± 5.8 months	102	No redislocation, 1 traumatic subluxation 1 case of shoulder joint stiffness No postoperative infection, axillary nerve injury, vascular injury	85% returned to preoperative level of exercise at follow-up	Mean ASES: 95 ± 6 Mean Rowe: 95 ± 3 Mean Walch-Duplay: 96 ± 3
Warner et al. 26	Iliac crest bone graft	Mean: 33 months	11	No recurrent instability 2 patients report mild pain with overhead activity	All patients returned to preinjury level of sport at follow-up	Mean ASES: 94 Mean Rowe: 94
Auffarth et al. 27	Iliac crest bone graft – J shaped graft, screw free fixation	Mean: 106.2 months	47	No cases of instability, 1 traumatic graft fracture 2 cases haematomas, 2 cases subcutaneous wound infection 5 cases nerve palsy 19 cases arthropathy	94% return to sport at follow-up 98% return to work at follow-up	Mean Rowe: 94 Mean constant: 94
Provencher et al. 55	Distal tibia allograft	Mean: 45 months (range: 30–66 months)	27	No cases of recurrent instability 1 allograft lysis 1 case of infection with P. acnes	N/A	Mean ASES: 91 WOSI: 11% of normal Mean SANE: 90
Kraus et al. 30	Iliac crest bone graft – arthroscopic	Mean: 20.6 months (range: 12–65 months)	15	No recurrent subluxations or dislocations 1 case hypoesthesia of iliac crest	N/A	Mean constant: 85 (range: 73–98) Mean Rowe: 88 (range: 65–100) Mean SSV: 85% (range 50%–100%) Mean WOSI: 77% (range: 46%–93%)

Non-rigid fixation techniques have been developed aiming to reduce complications seen with using screws. In their double cortical button fixation technique, Boileau et al. ²⁴ noted no neurological complications or revision surgeries at 14 months (Table 2). Gendre et al. ⁶² also found no neurological or hardware complications in a study of 70 patients using the same double cortical button fixation technique. Short-term recurrent instability rates for screw-free graft fixation range from 1.3% ²⁴ to 8.3%. ⁶³ Biomechanically, a study on eight shoulders by Provencher et al. ⁶⁴ compared the strength of traditional screw fixation with this novel technique and found comparable biomechanical strength for coracoid bone fixation. The mean load to failure for screw fixation was not significantly different from cortical button fixation at 226 ± 114 N and 266 ± 73 N, respectively. Smith et al. ³⁴ found that for an ovine shoulder, repair using the “buckle-down” fixation technique increased the force required for dislocation by 101% (p < 0.05) to 170 ± 17 N which was 72% of control values (234 ± 26 N). Smith et al. ³⁴ proposed that the buckle-down technique may be implementable as an all-arthroscopic procedure due to the standard arthroscopic instruments used in their study. The authors noted this procedure to be biomechanically sound and simple to execute.

Healing rates of 83% ⁶² to 91% ²⁴ have been achieved using cortical buttons for fixation. Xu et al. ²⁷ reported no postoperative neurovascular complications or infections out of the 102 patients included in their single cortical button study with subjective pain and functional status (measured with ASES, Rowe and Walch-Duplay scores) all improving significantly after at least three years follow-up (Table 2). Proper vertical graft positioning was achieved in 98 grafts with 100 grafts achieving bone union and all grafts underwent remodelling towards the intact glenoid shape within 2 years. However, there lacks evidence for long-term for graft resorption rates and remodelling for this technique.

Techniques eliminating implant-related complications entirely have been described by using a J-shaped iliac bone graft.^27,29 Anderl et al. ²⁹ noted significant improvement in all perioperative scores (Rowe, constant, VAS, SSV) and no recurrent instability at two-year follow-up using an arthroscopic J-shaped iliac crest bone grafting technique. A long-term study of the same technique in 46 patients at 18 years follow-up found a single case of traumatic subluxation with no further recurrent instability. ⁶⁵ However, five patients noted hypoesthesia at the iliac cresta and there were two cases of postoperative haematoma.

Bone grafts containing osteochondral surfaces have the proposed advantages of mimicking the glenoid cartilage and providing smoother articular surfaces during joint motion after recovery.^28,31,33 The distal tibia was noted to closely mimic the radius of curvature for the native glenoid. ²⁸ Wong et al. ⁶⁶ found graft union was seen in 94% of patients with the tibial allograft compared to 75% in the Latarjet procedure when using computed tomography assessment at minimum six months follow-up. The authors further found no statistically significant difference between the anteroposterior dimensions and cross-sectional area offered by the coracoid and distal tibial grafts suggesting equal graft resorption rates. At a longer follow-up of 45 months in 27 patients, Provencher et al. ⁵⁵ found a healing rate of 89% with no cases of recurrent instability (Table 2).

An osteochondral autograft of the contralateral femoral condyle has also seen positive results in a case study with the patient’s Rowe scores increasing from 15 to 95 after 7 months and postoperative CT showing glenoid defect size decreasing from 37% to 11%. ³³ Osteochondral bone graft techniques may be considered for patients without osteoarthritis, allowing preservation of the cartilaginous component during graft harvest. A biomechanical study has also noted a lower mean contact pressure for an osteochondral distal clavicle graft compared to a coracoid graft which may translate to reduced long-term joint wear in patients. ⁶⁷ However, there is no definitive evidence that osteochondral grafts are superior to other bone grafts for glenoid reconstruction due to limited studies and long-term data.

Acromion and distal clavicle bone grafts for glenoid reconstruction may allow for a single surgical site and avoid donor site morbidity seen with iliac crest graft transfers.^27,31,32,65 Sanchez et al. ³² observed a J-shaped graft similarly harvested at the iliac crest can be harvested at the posterior acromion angle. Tokish et al. ³¹ described a pilot study of eight patients for glenoid reconstruction with distal clavicle graft. Compared to distal clavicle allografts, the distal clavicle also has greater availability and lower costs while maintaining the osteochondral component. The bone stock of the clavicle and acromion also provide flexibility in graft harvest size, allowing further tailoring of the graft intraoperatively to best mimic the glenoid surface and achieve a smooth repair.

Conclusion

Coracoid transfer procedures and free bone grafting procedures are two broad techniques for glenoid reconstruction that are recommended in settings of large glenoid defects without a mobile bony fragment. Coracoid process transfers have been well established in literature and are attractive for the potentially increased stability due to the sling effect. Free bone grafting procedures using the iliac crest provide a more anatomic reconstruction and have had similar clinical outcomes. Alternative grafts also offer flexibility in graft shape and size depending on the bone stock chosen. Emphasis has been placed on modifications that minimise invasiveness using arthroscopic approaches and reduce implant-related complications by using non-rigid fixation techniques. Techniques involving a single surgical field that use an inexpensive and readily available bone graft are further considerations in future developments targeting glenoid reconstruction.

ORCID iDs

Patrick H Lam https://orcid.org/0000-0001-6196-1794

George AC Murrell https://orcid.org/0000-0002-8251-1327