Posterior Open-wedge Osteotomy and Glenoid Concavity Reconstruction Using an Implant-free, J-shaped Iliac Crest Bone Graft in Atraumatic Posterior Instability with Pathologic Glenoid Retroversion and Dysplasia: A Preliminary Report

Abstract

Background

Atraumatic posterior shoulder instability in patients with pathologic glenoid retroversion and dysplasia is an unsolved problem in shoulder surgery.

Questions/purposes

In a preliminary study of a small group of patients with atraumatic posterior shoulder instability associated with glenoid retroversion ≥ 15° and glenoid dysplasia who underwent posterior open-wedge osteotomy and glenoid concavity reconstruction using an implant-free, J-shaped iliac crest bone graft, we asked: (1) What proportion of the patients had persistent apprehension? (2) What were the improvements in patient-reported shoulder scores? (3) What were the radiographic findings at short-term follow-up?

Methods

Between 2016 and 2018, we treated seven patients for atraumatic posterior shoulder instability. We performed this intervention when posterior shoulder instability symptoms were unresponsive to physiotherapy for at least 6 months and when it was associated with glenoid retroversion ≥ 15° and dysplasia of the posteroinferior glenoid. All seven patients had a follow-up examination at a minimum of 2 years. The median (range) age at surgery was 27 years (16 to 45) and the median follow-up was 2.3 years (2 to 3). Apprehension was assessed by a positive posterior apprehension and/or posterior jerk test. Patient-reported shoulder scores were obtained and included the subjective shoulder value, obtained by chart review (and scored with 100% representing a normal shoulder; minimum clinically important difference [MCID] 12%), and the Constant pain scale score (with 15 points representing no pain; MCID 1.5 points). Radiographic measurements included glenohumeral arthropathy and posterior humeral head subluxation, bone graft union, correction of glenoid retroversion and glenoid concavity depth, as well as augmentation of glenoid surface area. All endpoints were assessed by individuals not involved in patient care.

Results

In four of seven patients, posterior apprehension was positive, but none reported resubluxation. The preoperative subjective shoulder value (median [range] 40% [30% to 80%]) and Constant pain scale score (median 7 points [3 to 13]) were improved at latest follow-up (median subjective shoulder value 90% [70% to 100%]; p = 0.02; median Constant pain scale score 15 points [10 to 15]; p = 0.03). Posterior glenoid cartilage erosion was present in four patients (all four had Walch Type B1 glenoids) preoperatively and showed no progression until the final follow-up examination. The median (range) humeral head subluxation index decreased from 69% (54% to 85%) preoperatively to 55% (46% to 67%) postoperatively (p = 0.02), and in two of four patients with preoperative humeral head subluxation (> 65% subluxation), it was reversed to a centered humeral head. CT images showed union in all implant-free, J-shaped iliac crest bone grafts. The median preoperative retroversion was corrected from 16° (15° to 25°) to 0° postoperatively (-5° to 6°; p = 0.02), the median glenoid concavity depth was reconstructed from 0.3 mm (-0.7 to 1.6) preoperatively to 1.2 mm (1.1 to 3.1) postoperatively (p = 0.02), and the median preoperative glenoid surface area was increased by 20% (p = 0.02). No intraoperative or postoperative complications were recorded, and no reoperation was performed or is planned.

Conclusion

In this small, retrospective series of patients treated by experienced shoulder surgeons, a posterior J-bone graft procedure was able to reconstruct posterior glenoid morphology, correct glenoid retroversion, and improve posterior shoulder instability associated with pathologic glenoid retroversion and dysplasia, although four of seven patients had persistent posterior apprehension. Although no patients in this small series experienced complications, the size and complexity of this procedure make it likely that as more patients have it, some will develop complications; future studies will need to characterize the frequency and severity of those complications, and we recommend that this procedure be done only by experienced shoulder surgeons. The early results in these seven patients justify further study of this procedure for the proposed indication, but longer term follow-up is necessary to continue to assess whether it is advantageous to combine the reconstruction of posterior glenoid concavity with correction of pathological glenoid retroversion and increasing glenoid surface compared with traditional surgical techniques such as the posterior opening wedge osteotomy or simple posterior bone block procedures.

Level of Evidence

Level IV, therapeutic study.

Introduction

Posterior shoulder instability has been thought to affect only a small proportion of patients with glenohumeral instability [19], but this may be because symptoms and clinical examination results are less clear than in anterior shoulder instability [42]. Patients frequently report subluxation rather than true dislocations, and some even report only an unstable, painful shoulder [5, 16, 48]. With increasing knowledge and better clinical examination and imaging techniques, posterior shoulder instability is being diagnosed more frequently [4, 47]. There are many risk factors for posterior shoulder instability; these include connective tissue disorders, hyperlaxity, pathologic muscle patterns, a horizontally oriented acromion in the sagittal plane [38], increased retroversion, glenoid dysplasia [22, 29, 31, 32], and less frequently, trauma [19].

Glenoid dysplasia is thought to be a result of scapular developmental anomalies [9, 36]. It is associated with a deficiency of the osseous posterior rim, malformation of the mechanical alignment or orientation of the socket (increased retroversion), and hyperplasia of the posterior labrum and cartilage [9, 23, 52]. The alteration in the articular alignment resulting in a posteriorly oriented glenoid surface relative to the scapula is a known risk for posterior instability as well as persistent or recurrent instability following posterior stabilization procedures [22, 26, 29, 32]. Glenoid dysplasia is seen with static posterior subluxation of the humeral head, and as such, it is still an unrecognized entity responsible for glenohumeral osteoarthritis in young adults [7, 49, 50]. And, importantly, this problem remains largely unsolved.

Nonoperative management in patients with atraumatic posterior shoulder instability associated with glenoid dysplasia may provide initial but inconsistent symptom relief [46, 55], but it cannot solve its underlying causes [26]. Surgical options in these patients, who often are younger, focus on joint-preserving treatment. Posterior capsulolabral repair or capsulorrhaphy have had variable results with regard to stability restoration [17, 18], with one study reporting recurrence in 23% of shoulders treated [17]. Biomechanical studies also showed that posterior capsulolabral repair for shoulders with glenoid retroversion of more than 10° may not be sufficient to restore stability [27]. Osseous procedures such as posterior iliac crest bone grafting (ICBG) can increase the posteroinferior glenoid joint’s surface [53], and certain bone block procedures can reconstruct a glenoid concavity [41]. However, pathologic retroversion cannot be addressed; therefore, some authors have focused on correcting glenoid retroversion with an opening-wedge osteotomy, as described in detail by Scott [45]. The reported results of such osteotomies in patients with glenoid dysplasia vary [28, 35], with studies reporting substantial complications [24, 30]. Although there are no long-term results on this procedure, there are some concerns about iatrogenic arthropathy after correction of version resulting in increased glenohumeral contact pressure [25], which may to some extent be caused by the procedure’s inability to address the pathology of the posteroinferior glenoid rim. Therefore, the senior author (CG) introduced a novel surgical technique using an implant-free, J-shaped ICBG, which is established for treating recurrent anterior shoulder instability [1, 43]. This novel procedure can reconstruct a posterior glenoid concavity and correct pathologic glenoid retroversion.

In a preliminary study of a small group of patients with atraumatic posterior shoulder instability associated with glenoid retroversion ≥ 15° and glenoid dysplasia who underwent posterior open-wedge osteotomy and glenoid concavity reconstruction using an implant-free, J-shaped iliac crest bone graft, we asked: (1) What proportion of the patients had persistent apprehension? (2) What were the improvements in patient-reported shoulder scores? (3) What were the radiographic findings at short-term follow-up?

Patients and Methods

Study Design and Setting

This is a retrospective, single-center study that evaluated patients treated by three experienced, high-volume subspecialty shoulder surgeons (CG performed four procedures, KW performed one, and the third surgeon, who was not a study author, performed the remaining two procedures). All three surgeons were experienced in posterior shoulder stabilization procedures, including posterior open-wedge osteotomies and simple posterior bone block procedures.

Patients

Between April 2016 and February 2018, we treated seven patients for atraumatic recurrent posterior shoulder instability with the J graft procedure. We performed this intervention when posterior shoulder instability symptoms were unresponsive to physiotherapy for at least 6 months and when it was associated with glenoid retroversion of at least 15° and dysplasia of the posteroinferior glenoid. All seven patients had a follow-up of a minimum 2 years. The median (range) age at surgery was 27 years (16 to 45) and the median follow-up was 2.3 years (2 to 3).

All posterior instability events were dynamic, involuntary, and atraumatic, and posterior shoulder instability was clinically confirmed by a positive jerk [33] and/or posterior apprehension test result [54]. Although there is no consensus on the definition of glenoid dysplasia, the presence of a rounded “lazy J form” or a triangular osseous “delta form” deficiency of the posteroinferior glenoid rim on CT have been identified as characteristics for glenoid dysplasia and to be associated with atraumatic posterior shoulder instability [52]. In addition, several authors have described increased glenoid retroversion as a risk factor for posterior instability and persistent or recurrent instability after posterior stabilization procedures, and the generally accepted value of a retroversion of no less than 15° was used for the indication of the posterior J-bone graft procedure [6, 21, 26, 40]. Glenoid dysplasia was assessed using preoperative CT arthrography according to the classification of Weishaupt et al. [52], and preoperative glenoid retroversion was assessed according to the technique of Friedman et al [15]. Patients with B2 glenoid morphology, previous posterior bone block procedures, multidirectional instability, voluntary posterior instability, or connective tissue disorders were excluded from this procedure.

The dominant shoulder was affected in three patients, one patient smoked cigarettes, and four were laborers. The median age at which posterior instability symptoms were first recognized, defined as posterior subluxation or difficulties or pain with activities at 90° of active forward elevation in the sagittal plane, was 15 years (IQR 6) in five patients and unknown in the remaining two patients. Four patients reported recurrent posterior shoulder subluxations, and the other three patients could not recall clear posterior subluxations or dislocations but reported apprehension. The posterior J-bone graft procedure was the primary surgical intervention in three patients, and the remaining four had undergone arthroscopic soft tissue stabilizations; three underwent posterior Bankart repair and one underwent posteroinferior capsulorrhaphy.

The median preoperative glenoid retroversion was 16° (IQR 6°), the median glenoid surface area (which was assessed on an en face view according to the Pico method [37]) was 594 mm² (IQR 324), and the median glenoid concavity depth (which was assessed on axial CT images [13]) was 0.3 mm (IQR 1.1). In addition to the dynamic posterior instability element in all seven patients, preoperative CT images showed static posterior humeral head subluxation, which was defined as more than 65% of humeral head subluxation, according to the modified subluxation index [51], in four of seven shoulders. Preoperative glenoid morphologic findings according to the criteria from Walch et al. [50] showed posterior glenoid cartilage erosion (that is, a Type B1 glenoid) in four of seven patients. Preoperative dislocation arthropathy according to the classification of Samilson and Prieto [44] was mild in two patients and moderate in one; the remaining four patients had no preoperative signs of dislocation arthropathy.

Surgical Technique

The patient was placed in a lateral decubitus position on the contralateral side. Through a posterior vertical skin incision over the glenohumeral joint, the deltoid was split along its fibers, and the interval between the infraspinatus and teres minor muscles was established without detachment of either of these muscles. The posterior scapular neck was exposed and a vertical capsulotomy was performed. The capsule was secured at each site with one Vicryl suture. A Fukuda ring-retractor (Innomed Inc) was inserted to expose the glenohumeral joint, and posteroinferior glenoid rim deficiency was confirmed in all seven patients. The plane of the glenoid was marked with a K-wire introduced posteriorly, lying flat on the glenoid cartilage. Glenoid osteotomy was performed between 15 and 20 mm medial and parallel to the posterior glenoid rim parallel to the intraarticular K-wire. A 20-mm osteotome was used, and the osteotomy was performed in a unicortical fashion. The suprascapular nerve lies medial to the osteotomy and was not dissected. With the chisel in situ, the osteotomy was opened very carefully to prepare for graft insertion. The width of the long leg of the J-shaped ICBG determines the degree to which glenoid retroversion is corrected and was assessed based on preoperative CT planning. We aimed to correct to neutral version. The ultimate extent of the opening osteotomy was determined intraoperatively using a measurement angle, and we determined the width of the long leg of the J-shaped ICBG, which was usually 5 mm.

A bicortical ICBG including the crest and inner table was harvested using osteotomes and an oscillating saw. The graft was then molded into a J-shaped graft using the oscillating saw. The graft typically measures 25 mm wide (superoinferior dimension of a glenoid in the en face view), 30 mm long (long leg; AP dimension), and 10 mm high (graft surface; AP dimension). To reconstruct a posteroinferior glenoid concavity, we molded the graft surface to a ramp-like structure. The dimension of the short leg of the J-bone graft usually measures 20 to 25 mm (mediolateral dimension) (Fig. 1).

Fig. 1.

The J-shaped iliac crest bone graft typically measures 25 mm wide (superoinferior dimension of the glenoid in the en face view), 30 mm long (long leg; AP dimension), and 10 mm high (graft surface; AP dimension). The graft surface is molded to a ramp-like structure to reconstruct the concavity of the posterior glenoid. The dimension of the short leg of the J-bone graft usually measures 20 to 25 mm (mediolateral dimension). The long leg and short leg comprise cortical bone on the outside and cancellous bone on the inside to facilitate bony ingrowth.

The short leg of the J-bone graft should comprise the distance between the medial boarder of the osteotomy and the most lateral point of the osseous glenoid plane after opening the osteotomy to neutral retroversion, and its definitive length therefore depends on the amount to which glenoid retroversion is corrected. The J-shaped ICBG was gently inserted into the osteotomy site and then impacted into place with press fit (Fig. 2). Because of press fit of the graft and compression forces of the rotator cuff, no further stabilization of the ICBG was necessary. The capsule was then closed with resorbable sutures, leaving an extraarticular graft (Fig. 3). No specific labral repair was performed.

Fig. 2.

A-D Intraoperative photographs show the posterior J-bone graft procedure. (A) A bicortical iliac crest bone graft including the crest and inner table was harvested. (B) The graft was then molded into a J-shaped graft with a ramp-like surface (arrow) to reconstruct the concavity of the posterior glenoid. (C) Posterior glenoid osteotomy parallel and medial to the glenoid plane was performed. (D) A J-bone graft was then gently inserted into the osteotomy site and impacted into place with press fit; G = glenoid; GN = glenoid neck; HH = humeral head; I = inferior; JBG = J-bone graft; M = medial; OT = osteotomy.

Fig. 3.

A-D These drawings show correction of glenoid retroversion and posterior glenoid rim deficiency with the posterior J-bone graft procedure. (A) Glenoid dysplasia was associated with pathologic glenoid retroversion and deficiency of the posterior glenoid concavity. (B) A unicortical posterior glenoid osteotomy was performed between 15 and 20 mm medial and parallel to the posterior glenoid rim according to preoperative CT planning. (C) The osteotomy was then opened very carefully to prepare it for graft insertion. The width of the long leg of the J-bone graft determined the degree to which the retroversion of the glenoid was corrected, which was aimed to be neutral. (D) The ramp-shaped graft surface reconstructed the posterior glenoid concavity and compensated for the posterior glenoid rim’s deficiency.

For 6 weeks, the shoulder was immobilized in a brace in neutral rotation with the arm at the side, with permitted passive and active-assisted anterior elevation and abduction in neutral rotation until 60°. No combined adduction and internal rotation in the horizontal plane was allowed for 6 weeks. After 4 months, patients could return to sports.

Primary and Secondary Study Outcomes

Our primary goal was to assess the proportion of the patients who had persistent apprehension. To achieve this, one orthopaedic surgeon in training (MW) who was not involved in patient care examined the patients for posterior apprehension and/or subluxation. Posterior apprehension was defined as a positive posterior apprehension [54] and/or positive posterior jerk test result [33]. The test results were considered positive if there was subluxation with pain or an uncomfortable sensation reproducing the patient’s symptoms [42]. Subluxation was defined as the subjective sensation of posterior translation of the humeral head over the glenoid rim followed by spontaneous reduction [3].

Our secondary study goals were the improvements in patient-reported shoulder scores and shoulder ROM, and the radiographic findings at short-term follow-up. One orthopaedic surgeon in training (MW) who was not involved in patient care summarized all scores and shoulder ROM by chart review. Patient-reported shoulder scores were the subjective shoulder value [20], with 100% representing a normal shoulder and a minimum clinically important difference (MCID) of 12% [8], and the Constant pain scale score, with 15 points representing no pain and an MCID of 1.5 points [34].

Two blinded readers (LE, MW) evaluated all radiographs and CT images independently. Preoperatively and at the final follow-up examination, patients had standardized true AP, axillary lateral, and scapular lateral radiographs and CT of the involved shoulder. AP radiographs were evaluated for the grade and progression of dislocation arthropathy [44]. On true lateral radiographs, the acromion orientation in the sagittal plane was measured as previously described [2, 11, 12, 38] and included the posterior acromial tilt, anterior and posterior acromial coverage, and posterior acromial height. CT images were assessed for glenoid erosion according to the criteria from Walch et al. [50] and posterior humeral head subluxation (> 65% of humeral-head subluxation) according to the modified subluxation index [51]. We also evaluated ICBG union, correction of glenoid retroversion according to the technique of Friedman et al. [15], augmentation of the glenoid surface area on an en face view according to the Pico method [37], and preoperative and postoperative glenoid concavity depth on axial CT images [13].

Further parameters assessed by chart review were intraoperative or postoperative complications and reoperations.

Ethical Approval

Ethical approval for this study was obtained from the Cantonal Ethics Commission of Zurich (project number 2017-01059); all included patients provided written consent for the study.

Statistical Analysis

Distribution of the data was assessed with the Shapiro-Wilk test. Preoperative and postoperative clinical and radiographic outcome measures were compared using the Wilcoxon signed-rank test. For categorical variables, the chi-square test and the Fisher exact test (if n < 5) were used. Significance was set as p < 0.05, and all p values were two-tailed.

The interobserver reliability for dislocation arthropathy (r = 0.97 [95% CI 0.84 to 0.99]), glenoid erosion (r = 1), posterior humeral head subluxation (r = 0.93 [95% CI 0.67 to 0.99]), glenoid retroversion (r = 0.83 [95% CI 0.34 to 0.97]), glenoid surface area (r = 0.97 [95% CI 0.87 to 0.99]) and concavity depth (r = 0.86 [95% CI 0.43 to 0.97]), and orientation of the acromion in the sagittal plane (r = 0.86 [95% CI 0.44 to 0.97]) was either excellent or very good.

Results

Persistent Instability

In four of seven patients, posterior apprehension was positive, but none reported resubluxation.

Shoulder Scores

The preoperative subjective shoulder value (median [range] 40% [30% to 80%]) and Constant pain scale score (median 7 points [3 to 13]) were substantially improved at latest follow-up (median subjective shoulder value 90% [70% to 100%]; p = 0.02; median Constant pain scale score 15 points [10 to 15]; p = 0.03) (Table 1).

Table 1.

Clinical findings preoperatively and at the most recent follow-up examination

Variable	Preoperative	Postoperative	Change	p valuea
Number of shoulders	7	7
SSV, %	40 (30-80)	90 (70-100)	+ 50	0.02
Constant pain scale score, points	7 (3-13)	15 (10-15)	+ 8	0.03
ROM
AAE, °	150 (90-170)	170 (130-180)	+ 20	0.15
Abduction, °	150 (90-170)	160 (140-180)	+ 10	0.25
External rotation, °	60 (40-80)	50 (40-80)	−10	0.72
Internal rotation, points	8 (0-10)	8 (8-10)	0	0.16

Data are presented as the median (range).

^ap values are based on the Wilcoxon signed-rank test; an SSV of 100% represents a normal shoulder with an MCID of 12% [8]; a Constant pain scale score of 15 represents no pain with an MCID of 1.5 points [34]; SSV = subjective shoulder value; AAE = active anterior elevation; MCID = minimum clinically important difference.

Radiographic Findings at Short-term Follow-up

Dislocation arthropathy at final follow-up was advanced (Grade 3) in two patients, moderate (Grade 2) in one, and mild (Grade 1) in one. One patient with a preoperative Grade 1 showed progression by two grades, and the other three progressed by one grade (Table 2). Posterior glenoid cartilage erosion showed no progression until the final follow-up examination. The median (range) humeral-head subluxation index decreased from 69% (54% to 85%) preoperatively to 55% (46% to 67%) postoperatively (p = 0.02), and in two of four patients with preoperative humeral head subluxation (> 65% subluxation), it was reversed to a centered humeral head. The two patients with uncorrected humeral head subluxation had advanced dislocation arthropathy at final follow-up, and these two patients were the only ones with persistent mild pain (that is, Constant pain scale score of 10 points). Notably, both patients were older than 40 years of age at the time of surgery (42 and 45 years).

Table 2.

Characteristics of the seven patients undergoing the posterior J-bone graft procedure

Number	Gender (man/woman)	Age at surgery in years	Prior posterior soft tissue stabilization	Preoperative glenoid retroversiona	Postoperative glenoid retroversiona	Preoperative posterior humeral head subluxationb	Postoperative posterior humeral head subluxationb	Preoperative posterior glenoid erosionc	Postoperative posterior glenoid erosionc	Preoperative dislocation arthropathyd	Postoperative dislocation arthropathyd
1	M	42	0	15	0	1	1	1	1	2	3
2	M	27	1	25	-1	1	0	1	1	0	1
3	W	20	0	16	4	0	0	0	0	0	0
4	M	32	1	21	-5	1	0	1	1	1	2
5	W	18	1	16	6	0	0	0	0	0	0
6	W	16	1	15	4	0	0	0	0	0	0
7	M	45	0	17	-3	1	1	1	1	1	3

^aGlenoid retroversion was assessed according to the technique by Friedman et al. [15], and a negative value indicates anteversion.

^bPosterior humeral head subluxation was defined as an index > 65% according to the subluxation index described by Walch et al. [51].

^cPosterior glenoid cartilage erosion (such as the B1 glenoid) was defined according to the Walch classification [50].

^dDislocation arthropathy was graded from 1 (mild) to 3 (advanced) according to the classification of Samilson and Prieto [44].

CT images showed union in all implant-free, J-shaped iliac crest bone grafts. The median (range) preoperative retroversion was corrected from 16° (15° to 25°) to 0° postoperatively ( [-5° to 6°]; p = 0.02). The median preoperative glenoid surface area was increased from 594 mm² (403 to 892) preoperatively to 747 mm² (557 to 980) postoperatively (p = 0.02), which corresponded to a relative gain of 20% of the glenoid surface area. The median glenoid concavity depth was reconstructed from 0.3 mm (range -0.7 to 1.6 mm) preoperatively to 1.2 mm (1.1 to 3.1) postoperatively (p = 0.02) (Fig. 4).

Fig. 4.

A-B The posterior J-bone graft procedure was assessed preoperatively and postoperatively with CT. (A) Preoperative CT shows pathologic glenoid retroversion of 16° and a flat glenoid concavity. (B) Postoperative CT shows correction of glenoid retroversion to 1°. The reconstructed glenoid concavity is seen through the posterior J-bone graft.

Measurement of sagittal acromial morphology revealed anormal values of median (range) posterior acromial tilt of 65° (55° to 73°), anterior acromial coverage of -9° (-17° to 3°), posterior acromial coverage of 51° (range 33° to 63°), and posterior acromial height of 31 mm (18 to 45) (Table 3).

Table 3.

Radiographic findings preoperatively and at final follow-up

Variable	Preoperative	Postoperative	p valuea
Number of shoulders	7	7
Dislocation arthropathy, pointsb	0 (0 to 2)	1 (0 to 3)	0.06
None, n	4	3
Grade 1, n	2	1
Grade 2, n	1	1
Grade 3, n	0	2
Type B1 glenoid, nc	4	4
Posterior humeral head subluxation index, %d	69 (54 to 85)	55 (46 to 67)	0.02
Posterior humeral head subluxation, n	4	2	0.30
Graft union, n		7
Glenoid retroversion, °	16 (15 to 25)	0 (-5 to 6)	0.02
Glenoid surface area, mm2	594 (403 to 892)	747 (557 to 980)	0.02
Glenoid concavity depth, mm	0.3 (-0.7 to 1.6)	1.2 (1.1 to 3.1)	0.02
Posterior acromial tilt, °		65 (55 to 73)
Anterior acromial coverage, °		-9 (-17 to 3)
Posterior acromial coverage, °		51 (33 to 63)
Posterior acromial height, mm		31 (18 to 45)

Data are presented as the median (range) or n.

^ap values are based on the Wilcoxon signed-rank test.

^bDislocation arthropathy was graded from 1 (mild) to 3 (advanced) according to the classification of Samilson and Prieto [44].

^cPosterior glenoid cartilage erosion was defined according to the Walch classification [50].

^dPosterior humeral head subluxation was defined as an index > 65% according to the subluxation index described by Walch et al. [51].

Previous arthroscopic soft tissue stabilizations had no significant influence on any of the clinical or radiographic outcome measures (p > 0.05).

Complications and Reoperations

No intraoperative complications were recorded. There were also no postoperative complications including graft-related complications such as graft dislocation or resorption, and none of the included patients underwent a reoperation.

Discussion

Involuntary, atraumatic posterior shoulder instability is rare and difficult to diagnose. It frequently is associated with glenoid dysplasia. Although the degree to which glenoid dysplasia is clinically relevant is not yet known, the combination of deficiency of the posteroinferior glenoid rim and glenoid retroversion of at least 15° are accepted risk factors for atraumatic posterior shoulder instability [6, 21, 26, 40, 52] and for persistent or recurrent instability after posterior stabilization procedures [22, 26, 29, 32]. Treatment of this rare condition is controversial. Soft tissue procedures may not be sufficient to restore stability in cases with pathologic retroversion [27], and osseous procedures such as a simple posterior ICBG can increase the posteroinferior glenoid joint’s surface but not correct retroversion [53]. Correction of glenoid retroversion can be achieved with an opening-wedge osteotomy [45], but there are concerns regarding complications including iatrogenic arthropathy [25]. These unpredictable results and the characteristics of the underlying pathology in patients with atraumatic posterior instability associated with glenoid dysplasia led the senior author (CG) to introduce this novel surgical technique using an implant-free, J-shaped ICBG. The key findings of this small, retrospective study with patients treated by experienced shoulder surgeons were that in at least the short-term, a posterior J-bone graft procedure corrected posterior shoulder instability and pain and substantially improved but did not fully correct posterior apprehension, which persisted in four of seven patients who underwent the procedure. Structurally, the procedure was also able to reconstruct normal posterior glenoid morphology and correct glenoid retroversion, but four of seven patients in this study showed progression of their preoperative osteoarthritic changes, and two patients remained slightly painful.

Limitations

This study should be interpreted in light of a number of important limitations. The two main limitations of this study are the small number of patients and the short-term follow-up. The small number of patients is an important limitation because it precludes any assessment of what the actual complication risk will be as this operation comes into wider use. This operation is challenging, and we do not recommend it to inexperienced shoulder surgeons. Although we did not see complications in this small study, this operation is large enough that if more are done, one would reasonably expect complications to occur. Larger studies are needed to determine the frequency and severity of those complications. Performance bias is worth considering here as well, as three very experienced subspecialty shoulder surgeons performed all the procedures. In our department, the operations remain reserved for the most experienced surgeons and will not be delegated because of its complexity and the fear of potential complications.

With a minimum follow-up of only 2 years, the results of this operation may look better than they likely will over time; the benefits of the procedure are evident, but the natural history of this condition has not yet played itself out. We continue to follow these patients and anticipate reporting on them with longer follow-up. Recurrence of instability, progression of osteoarthritis, and recurrence of static posterior subluxation are potential complications that may only manifest themselves at long-term.

There is also an element of assessment bias as the surgeon’s team assessed all outcome measures. We tried to mitigate by blinding; radiographic and clinical assessment were completed without the treating surgeon. Although there is no universal definition of glenoid dysplasia, our study was well-defined in terms of glenoid anomalies: All patients had glenoid retroversion of between 15° and 25° and osseous deficiency of the posteroinferior glenoid, which are known risk factors for atraumatic posterior shoulder instability [6, 21, 26, 40, 52]. We consider these elements as the factors identifying glenoid dysplasia, and they may be associated with posterior static subluxation with no more than B1 changes. On the other hand, there was some heterogeneity in the study group; four of seven patients had previous posterior soft tissue stabilization, which as far as we can tell in this small series did not appear to influence our clinical or radiographic short-term results. It remains challenging to distinguish between static and dynamic posterior shoulder instability in this setting, and whether there is an element of combined dynamic and static posterior instability needs to be further investigated. In our study, all patients had clinical signs of dynamic posterior shoulder instability assessed with the posterior apprehension and/or jerk test, but also four of seven showed posterior humeral head subluxation on preoperative CT scans, which, surprisingly, was corrected at this follow-up time in two patients after the posterior J-bone graft procedure. Whether this procedure can be beneficial for more dysplastic/retroverted glenoids (that is, C glenoids) remains unknown. At our institution, we therefore perform the posterior J-bone graft procedure in patients with atraumatic posterior shoulder instability, preferably without obvious preoperative arthropathy, unresponsive to nonoperative management such as physiotherapy over a period of minimum 6 months, and with associated structural abnormalities such as deficiency of the posteroinferior glenoid rim and glenoid retroversion of between 15° and 25°.

Persistent Instability and Shoulder Scores

Although posterior apprehension persisted in four of seven patients, shoulder scores as patient-reported pain scores improved by an amount larger than the known MCIDs [8, 34] in this small group. Pain and mild disability persisted in two patients who had radiographic signs of arthritis preoperatively, and these radiographic changes progressed. We also observed that both were older than 40 years of age at the time of the procedure, and in accordance with the observations of Smith and Bunker [46], with increasing age the symptoms are generally more disabling and the results of this procedure may be less predictable. Our results suggest that the procedure may be more reliable in patients without degenerative joint changes.

Radiographic Findings at Short-term Follow-up

We observed consistent radiographic union and general improvement of glenoid alignment and surface area. The procedure was effective in correcting pathologic glenoid retroversion in all included patients. Such extensive correction of glenoid version has raised concerns about iatrogenic arthropathy because of increased glenohumeral contact pressure [25]. Although posterior glenoid erosion did not progress, two of seven patients in this study showed advanced arthropathy at the final follow-up interval. These two patients also had uncorrected posterior humeral head subluxation and had radiographic signs of osteoarthritis preoperatively. Static posterior humeral head subluxation is sometimes observed in patients with atraumatic posterior instability. In a series by Ortmaier et al. [39], posterior open-wedge osteotomy of the scapular neck was performed in relatively elderly patients with preexisting glenohumeral arthropathy associated with static posterior humeral head subluxation. Although pathologic glenoid retroversion was corrected, the humeral head could not be recentered with their procedure. In our series, two of four patients with preoperative posterior humeral head subluxation had a centered glenohumeral joint at the final follow-up examination, but this was at a follow-up of 2 years. In the other two patients, the humeral head remained posteriorly subluxated. These two patients showed preoperative osteoarthritic changes that progressed over time. We think that in patients with preoperative humeral head subluxation and radiographic signs of glenohumeral arthropathy, the natural history of this condition is less likely to be altered with the posterior J-bone graft procedure, a phenomenon that has also been observed in the surgical treatment of recurrent anterior instability in patients older than 40 years of age [10, 14]. Another newly discovered risk factor for posterior instability is the morphology of the posterior acromion in the sagittal plane [38], which was also observed in this series.

Conclusion

In this small, retrospective series, patients treated with a posterior J-bone graft procedure by experienced shoulder surgeons for recurrent posterior shoulder instability obtained satisfactory clinical results at short-term follow-up. The procedure restored normal glenoid shape, glenoid version, and improved posterior shoulder instability associated with pathologic glenoid retroversion and dysplasia, although four of seven patients had persistent posterior apprehension. Although there were no complications in this series of patients treated by very experienced, specialized shoulder surgeons, the size and complexity of this procedure make it likely that as more patients have it, some will develop complications; future studies will need to characterize the frequency and severity of those complications, and we recommend that this procedure be done only by experienced shoulder surgeons. The early results in these seven patients justify further study of this procedure for the proposed indication to explore the theoretical advantages of reconstructing posterior glenoid concavity, increasing glenoid surface, and correcting pathologic glenoid retroversion over posterior open-wedge osteotomy or simple posterior bone block procedures.