Glenoid bone grafting in primary anatomic total shoulder arthroplasty: a systematic review

Abstract

Background

Primary anatomic total shoulder arthroplasty can be challenging in patients with complex glenoid wear patterns and bone loss. Severe retroversion (>15°) or significant bone loss may require bone grafting. This review summarizes the rate of revision and long-term outcomes of anatomic total shoulder arthroplasty with bone graft.

Methods

A systematic search of MEDLINE, Embase, PubMed, and CENTRAL databases was conducted from the date of inception to 23 October 2018. Two reviewers independently screened articles for eligibility and extracted data for analysis. The primary outcome was rate of revision. The secondary outcomes were rate of component loosening, functional outcome, and range of motion.

Results

Of the 1056 articles identified in the search, 26 underwent full-text screening and 7 articles were included in the analysis. All procedures were one-stage anatomic total shoulder arthroplasties. The rate of revision was 5.4% with component loosening and infection listed as indications over a weighted mean follow-up period of 6.3 years. Complications occurred in 12.6% of patients.

Conclusion

Glenoid bone grafting in anatomic total shoulder arthroplasty results in comparable revision rates and improvement in pain compared to augmented glenoid components and reverse shoulder arthroplasty. Due to the low quality of evidence, further prospective studies should be conducted.

Level of evidence

IV

Introduction

Anatomic total shoulder arthroplasty (aTSA) results in pain relief and improvement in function among patients with degenerative conditions of the shoulder. Approximately 10% of component failures in aTSA can be attributed to glenoid component loosening.^1–8 One challenge in aTSA is adequately correcting deformity and seating the glenoid component in the native glenoid without violating the subchondral bone, which can be particularly difficult in severe asymmetric glenoid wear. Without correction, excessive residual retroversion can lead to poor positioning of the glenoid component causing radiolucent lines and humeral head dislocation after aTSA. Clinically, uncorrected glenoid component results in a 2.5-fold increase in aTSA failure compared to corrected counterparts. ⁹ Several methods to address this glenoid pathology have included eccentric reaming, augmented glenoid components, and bone grafting to correct the glenoid defect followed by insertion of a standard polyethylene glenoid component.

Bone grafting has been presented in the literature as an attractive option in the setting of severe bone loss. Cortical and cancellous bone from the humeral head and iliac crest is used to correct retroversion, restore bone stock, provide a biological basis for healing, prevent subluxation of the humeral head component, and prevent component penetration of the glenoid vault.^9–11 However, some studies have highlighted concerns about failure of graft incorporation or loss of glenoid fixation causing inconsistent clinical benefit. ¹² Cement fixation of an all polyethylene glenoid component may be challenging due to the theoretical potential for non-union and graft resorption. Additionally, there is a question about how much of the glenoid component will achieve fixation into native bone, potentially resulting in premature implant loosening and failure.

In the present study, we systematically reviewed the available literature to determine (a) the short- and middle-term revision rate of aTSA with glenoid bone grafting, (b) complications following aTSA with glenoid bone grafting, (c) the bone graft union rate following aTSA with glenoid bone grafting, and (d) clinical outcomes following aTSA with bone grafting, including improvements in pain and function.

Despite the increasing utilization of glenoid bone grafting, our understanding of the outcomes and complications is limited. The primary objective of this systematic review is to quantify the rate of revision and to explore common causes for revision after aTSA with bone grafting. Additionally, we aim to explore the incidence of component loosening, and improvement in functional outcome scores and range of motion.

Material and methods

This study was conducted following the methods of the Cochrane Handbook for Systematic Review of Interventions ¹³ and is reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement. ¹⁴

Search strategy

A systematic search was conducted using MEDLINE, CENTRAL, Embase, and PubMed from the database inception date to 23 October 2018. Search strategy was developed by investigators with methodological and content expertise and a list of search terms is provided in supplementary Appendix 1. To enhance search sensitivity, we incorporated Medical Subject Headings and Emtree headings and subheadings (supplementary Appendix 1). A hand search of related references and cited articles was also completed.

Screening and data abstraction

The title, abstract, and full-text screening was completed independently by two reviewers (BZ and GN) using a piloted electronic data collection form (Excel; Microsoft). Inter-rater agreement was described by the kappa coefficient. Data abstracted included the rate of revision after aTSA with bone graft, indications for revision, rate of component loosening, union rate, functional outcomes, and pain relief after surgery.

Methodological quality assessment

Two reviewers (BZ and GN) performed an independent assessment of the methodological quality using the Methodological Index for Non-Randomized Studies (MINORS) ¹⁵ tool for all nonrandomized studies. Kappa values were calculated to quantify agreement between reviewers. Level of evidence was graded according to the criteria of Wright et al. ¹⁶

Eligibility criteria

Studies were included if they enrolled patients over the age of 18 who underwent aTSA with bone grafting. Studies were excluded if the indication for surgery was revision of a previously failed TSA procedure or if an augmented implant was used. Study designs including editorials, reviews, expert opinions, and basic science articles were excluded. There were no restrictions regarding the type of autologous graft, length of follow-up, publication date, or language of publication.

Statistical analysis

Graft union, revision, and complication rates were pooled using Microsoft Excel. For categorical variables, including radiological lucency lines, patients within each 0.50 mm incremental category were compared. For satisfaction ratings, patients were also pooled together into three groups: unsatisfactory, satisfactory, and excellent.²⁰ Functional outcome and range of motion were presented as an average at pre-op and at last follow-up with confidence intervals when possible.

Results

Search results and study characteristics

The systematic literature search identified 1056 relevant citations. After duplicate removal, 805 citations underwent title and abstract screening with excellent agreement between reviewers (kappa = 0.83). Of these, 26 met the eligibility criteria for full-text screening (kappa = 0.86). At the end of screening, seven articles^{9–11,17–20} met all inclusion criteria. Figure 1 demonstrates the PRISMA diagram showing the article screening process and reasons for exclusion.

Figure 1.

PRISMA diagram. TSA: total shoulder arthroplasty.

All seven of the included articles were retrospective cases series (level IV evidence). Six were conducted in the United States^{9–11,18–20} and one study was conducted in Switzerland. ¹⁷ All patients were followed-up for at least two years. The weighted mean length of follow-up was 75.5 months, standard deviation 25.6 months. A total of 161 patients received 167 aTSAs, of which 58 (34.7%) were completed with polyethylene components and 109 (64.1%) were with metal-backed components. The mean age was 59.4 years (range 26–81). Four studies reported a rate of 1.97% (83 bone graft aTSA/4218 of all aTSAs) as the frequency of bone grafting used across their sampling population of all aTSAs performed at that facility.^9,18–20

The most common indication for aTSA with bone graft is shoulder osteoarthritis (n = 124) followed by chronic dislocation (n = 16), rheumatic arthritis (n = 9), traumatic arthritis (n = 6), capsulorrhaphy arthroplasty (n = 3), cuff-tear arthropathy (n = 3), inflammatory arthritis (n = 2), and others (n = 4). All but three patients received an autograft from the humoral head. Two of these patients had an iliac crest allograft, and the third patient had an unspecified allograft. Table 1 describes the study characteristics of the seven included studies.

Table 1.

Study characteristics.

Study	Study timeframe	Level of evidence	No. of included patients	No. of TSAs	No. of patients screened a	Mean age at surgery (range)	Source of graft	MINORS score 21
Sabesan et al. 9	2002–2009	IV	12	12	213	55.8 (46–74)	Autograft from humeral head	13
Gazielly et al. 17	1997–2005	IV	33	39	–	68.1 (51–83)	Autograft from humeral head	12
Nicholson et al. 10	2004–2014	IV	28	28	–	61.4 (33–76)	Autograft from humeral head	12
Klika et al. 19	1976–2008	IV	24	25	2607 a	65.5 (39–87)	Autograft from humeral head	13
Hill and Norris 11	1980–1989		17 b	17	–	56.4 (30–84)	15 shoulders used autograft from humeral head; 2 used allograft from iliac crest	13
Steinmann and Cofield 18	1976–1992		28 b	28	935	56 (39–78)	27 autograft from humeral head; 1 unspecified allograft	13
Neer and Morrison 20	1973–1985		19	19	463	52.7 (26–68)	Autograft from humeral head	13

MINORS: Methodological Index for Non-Randomized Studies; TSA: total shoulder arthroplasty.

^aThe number of patients with TSA screened to identify patients who received bone grafting.

^bOne patient was excluded from the analysis after receiving bone grafting as part of revision procedures after a previous TSA.

Risk of bias

Most studies were similar in design and execution and thus were scored similarly on the MINORS scale. ¹⁵ The mean minor score was 12.7 with a range from 12 to 13. Lack of prospective data collection, exclusion of patients lost to follow-up, and lack of prospective sample size calculation were commonly cited reasons for reduced MINORS scores.

Rate of revision/complications

Of the 167 aTSAs performed, nine required revision within the follow-up period resulting in a revision rate of 5.4%.^9,11,17,19 A subgroup of two studies utilizing only polyethylene implants reported a revision rate of 3.95% (3 revisions out of 76 TSA performed). The most common indication for revision was aseptic glenoid loosening (n = 8) and infection (n = 1). Sabesan et al. and Hill and Norris described that cases of infection and aseptic loosening^9,11 were treated with removal of the glenoid components. Sabesan et al. ⁹ performed two revision hemiarthroplasties and Klika et al. ¹⁹ performed two revision aTSAs. Two other studies requiring revisions did not describe the specific revision procedure.^11,17 Revision procedures occurred at a mean of 32.2 months after the index of operation and had a bimodal distribution. Five revisions occurred in the first year and involved graft collapse, non-union, and dislodgement of glenoid and bone graft.^9,11,19 Four revisions occurred after one year and was often indicated for glenoid loosening, instability, and pain.^9,11,17,19

Secondary surgery to improve function was required for three patients experiencing broken screws, anterior dislocation, posterior subluxation, and capsular stiffness. Arthroscopic surgery was performed to remove the broken screws. ⁹ A soft tissue reconstruction was chosen for the anterior dislocation and surgical tightening of the posterior shoulder capsule was elected to correct for posterior subluxation. ⁹ Stiffness in one patient was corrected through capsular release. ¹⁸

Two other complications were reported that did not require revision, which were two patients who experienced strength deficiency following “massive rotator cuff tears” during the follow-up period. ¹¹ In addition to the complications that resulted in revision, six patients had shifting of the glenoid component between the early and last follow-up radiographs, and two had strength deficiency secondary to massive rotator cuff tears. Thus, the pooled complication rate was 12.6%.

Radiographic findings

Graft incorporation was assessed radiographically in each study. Union was defined as no complete radiolucent lines at the baseplate/bone graft/glenoid interfaces and no evidence of component loosening at final follow-up. The union rate is 82%. There are 30 shoulders with evidence of non-union.

Five studies reported on radiolucent lines of the glenoid component on plain radiographs, examining both the width of the radiolucent lines as well as whether they were incomplete or complete. In four studies with standardized reporting, 53 shoulders (61.6%) had no radiolucent lines, 23 (26.7%) shoulders had 1 mm lines (including complete and incomplete), and 10 (11.6%) shoulders had greater than 1.5 mm lines out of 86 shoulders sampled,^10,11,19,20 Gazielly et al. ¹⁷ developed a radiolucency scoring system based on the width of the radiolucent lines, as well as the number of locations where the lines were seen to extrapolate risk of radiographic loosening. The group reported 31 cases with no radiographical loosening, two cases had possible loosening, and six cases showed definitive loosening. ¹⁷

Functional outcomes and range of motion

Across the three studies reporting functional outcomes, all patients saw improvement in their clinical scores.^9,10,17 Scoring system included Constant score, Simple Shoulder Test, American Shoulder and Elbow Surgeons Shoulder Score, and Penn Shoulder Score (Table 2). Heterogeneity of data with respect to follow-up times and reporting methods was not conducive to meta-analysis of the data. Improvement in range of motion is shown in Figure 2, which demonstrates a statistically significant improvement in range of motion on the scales for forward flexion and external rotation.

Table 2.

Comparison of preoperative and postoperative functional outcome scores.

Study ID	Scoring system	Mean preoperative	Mean postoperative	Percent change in functional score
Sabesan et al. 9	Penn Shoulder Score31	38.7	79.4	+105.2
Gazielly et al. 17	Constant Score33	33.5	73.0	+117.9
Nicholson et al. 10	SST Score34	4	10	+150
	ASES Score29	39	90	+130.8
	Functional Score30	11	25	+127.3

ASES: American Shoulder and Elbow Surgeon Shoulder Score; SST: Simple Shoulder Test.

Figure 2.

Improvement in range of motion. Error bars represent the 95% confidence interval (CI) and is graphically depicted¹¹ where studies reported sufficient data for a CI calculation.

Three studies examined postoperative pain and identified that all patients had severe, debilitating pain preoperatively.^11,18,19 Of the 65 patients who were assessed for postoperative pain on a categorical scale, 55 (84.6%) had no to slight pain, 9 (13.8%) had moderate pain, and 1 (1.5%) had severe pain (Table 3). Nicholson et al. ¹⁰ and Gazielly et al. ¹⁷ both showed a statistically significant improvement in patient-reported pain after aTSA with bone grafting.

Table 3.

Comparison of preoperative and postoperative pain scores.^20,32

Study ID	No pain	Slight pain	Moderate pain	Severe pain
Klika et al. 19	12	11	2	0
Hill and Norris11,a	10		2	0
Steinmann and Cofield 18	16	6	5	1

^aOnly completed for patients who did not have prosthetic failure and revision surgery.

The majority of patients were either exceptionally satisfied with the outcome (59.5%, 50 out of 84 sampled patients) or satisfied with the outcome (26.2%, 22 out of 84). Only 14.3% (12 out of 84) found the clinical results of the procedure to be unsatisfactory (Table 4).

Table 4.

Postoperative patient satisfaction rating.^20,32

Study ID	Unsatisfactory	Satisfactory	Excellent
Klika et al. 19	2	5	18
Hill and Norris11,a	3	6	3
Steinmann and Cofield 18	5	10	13
Neer and Morrison 20	2	1	16

^aOnly completed for patients who did not have prosthetic failure and revision surgery.

Discussion

Although bone grafting is not commonly indicated in the setting of aTSA, for a subset of patients suffering from substantial glenoid bone deficiency it is an attractive option to prevent poor positioning of the glenoid component and humeral head subluxation. In this study, the rate of revision for aTSA with bone graft was found to be 5.4%, with aseptic loosening as the most common indication for surgery. The total complication rate was 12.6%, which is similar to the complication rate for aTSA without bone loss utilizing third generation implants, metal-backed prostheses, and modern stemless implant (10.6–12.8%).^21,22 On radiographs, graft union rates were 82.0%. More than 85% of patients were satisfied or very satisfied with the improvement in functional outcome and reduction in pain. This evidence suggests that bone grafting has a relatively low rate of revision and high levels of patient satisfaction.

The primary concern in the literature around aTSA with bone graft is early glenoid component loosening. We found the rate of revision secondary to aseptic glenoid loosening to be 4.79%, accounting for 89% of the revision procedures. Furthermore, 10.8% of all shoulders had complete radiolucency lines >1.5 mm suggesting an increased risk of glenoid loosening. In a recent literature review on bone grafting in the setting of reverse shoulder arthroplasty, the authors found rates of glenoid loosening to be around 3.1%. ²³ Our ability to analyze the long-term data is limited by the short follow-up time for existing studies.

In the setting of significant bone loss and deformity, there is also the option to use augmented polyethylene glenoid components to compensate for glenoid retroversion above 10–15°. There are limited studies on the outcomes of augmented polyethylene glenoid components. Case series by Lenart et al. ²⁴ and Favorito et al. ²⁵ identified low complication and revision rates. Limitations with respect to study quality and data heterogeneity make direct comparisons difficult. ²⁶ The two approaches are not necessarily comparable, as a review highlighted that after 20° of preoperative retroversion, bone grafting is helpful for adequate correction of version. ²⁷ While augmented glenoid components are promising, further research with respect to long-term outcomes with larger cohort sizes is needed. The more modern augments that have been recently introduced may contribute to a lower complication and failure rate.

Another option for glenohumeral arthritis is a reverse total shoulder arthroplasty (RTSA) that can be performed with or without bone grafting. Some have highlighted concerns with increased complication rates and limitations on activity after RTSA. ¹² A recent systematic review pooled 254 patients across seven studies and found a pooled mean complication rate of 17.6%, ²⁸ which is higher than our complication rate of 12.6%. Similar to our study, both techniques show a statistically significant improvement in functional outcomes and range of motion. ²⁸

Our study has numerous strengths in its rigorous methodology with a systematic review of several databases and duplicate data screening and extraction. This is the first study to quantify the rates of revision, complication, and patient satisfaction across all TSA with bone graft studies.

However, the quality of our study was limited by low level of evidence from the included retrospective case series. Each study had small sample sizes because of the low incidence of aTSA with bone graft and the need for patients with long-term follow-up. Specific to the functional outcomes and pain score, there was inconsistent reporting, making the data difficult to pool and thus meta-analysis could not be performed. Some of the studies include older metal backed glenoid implants, which have fallen out of clinical use.

We identified variability in surgical technique to be an additional element of bias. Indications for bone grafting varied from osteoarthritis to chronic dislocation and often the degree of retroversion and bone loss necessitating bone graft was not specified, making it difficult to identify when it can be ideally used.

Bone graft resorption is a gradual process, with rates of failure increasing significantly overtime. A future area of exploration would be to compare the long-term failure and revision rates of bone graft aTSA with conventional aTSA.

Conclusion

Glenoid bone grafting in the setting of aTSA results in comparable rates of revision (5.4%) and improvement in pain compared to glenoid augment and RSA. The most common cause of failure is glenoid loosening, which was defined by clinical complications and radiographical evidence of radiolucency lines. Due to the low quality of evidence, further prospective studies should be conducted to support clinical decision making.