Background
Glenoid component loosening is still one of the major problems in shoulder arthroplasty. According to a systematic review of the current literature, radiolucent lines have been reported to occur at a rate of 7.3% per year with over 70% prevalence at 10 years follow up of total shoulder arthroplasties. Revisions due to glenoid loosening were performed at close to 1% per year following implantation [
1]. Multiple factors including the method of glenoid preparation, cementing technique, implant-material etc. are considered potential reasons for loosening within the cement-bone interface. In the systematic review, the significant factors included Walch classification, gender, and diagnosis [
1]. The design of the glenoid component (implant) has been suggested as another critical factor and therefore has led to concerns regarding optimal prosthetic design.
Adequate initial fixation strength is thought to be crucial for long-term stability of the glenoid component and ultimately the clinical success of total shoulder arthroplasty. Previous authors have identified eccentric loading and the resulting rocking of the glenoid component as an important biomechanical factor for implant loosening [
2‐
4]. As a result, many designs have been developed with the intent to improve fixation of the glenoid. To date, keeled and pegged constructions have emerged as the most widely utilized designs. Results of recent radiographic studies have favored the pegged over the keeled glenoid designs at early follow-up such as 26 months [
5]. However, others could not demonstrate significant differences in radiographic follow up studies with an intermediate follow up (<45 months) [
6].
Our objective was to evaluate the effects of implant design (keel vs. peg) on initial stability and postoperative radiolucency. The purpose of this study was to determine whether the morphology of the glenoid that yielded the strongest primary stability with the least amount of micro-motion under eccentric loading would also exhibit less radiolucency in patients at a minimum of 2 years following total shoulder arthroplasty.
Discussion
The goal of this study was to determine if the biomechanical performance of two glenoid components would correspond to the degree of radiolucency in patients at a minimum of 2 years following total shoulder arthroplasty. While the pegged component demonstrated more micro-motion during eccentric axial loading both with and without transverse force loads, the keeled component presented with a greater degree of radiolucency on postoperative x-rays. These findings provide support for both glenoid components from different perspectives and when taken together, highlight the need for well-constructed clinical studies to determine whether glenoid design influences outcome and patient satisfaction.
From a biomechanical perspective, the keeled component demonstrated less micro-motion in all orientations except for superior displacement. These findings are consistent or lower than those of Collins et al. [
4] whose paper served as the basis of our testing protocol. The lack of a significant difference in superior displacement could be explained by the bony anatomy of the glenoid. Anglin et al. [
11] reported a more stable bony socket in the superior glenoid as compared to the anterior, posterior, and inferior glenoid rim. Checroun et al. [
12] reported that 71% of the 412 glenoids examined in their study displayed a pear-shaped form. These pear-shaped glenoids are described as having decreased width in the superior portion as compared to the inferior aspect [
13]. Based on the differences in the dimension of the bony anatomy, complete coverage of the superior glenoid may not always be possible with an implant [
14]. This may give the superior aspect of the glenoid component an advantage in terms of stability with eccentric loading. The anterior, posterior, and inferior aspects of the glenoid are well covered by the implant leaving less area to distribute eccentric loads in comparison to the superior aspect of the component where the lack of complete coverage results in bone above the implant thereby providing a larger surface area for eccentric force distribution.
The physical dimensions of the anchor on each component may have impacted stability. The single-anchor keeled glenoid is more compact and uniform in its design compared to the three-anchor pegged glenoid, which may influence the stability of the articulating surface of the component attached to the anchor (Fig.
4). This may explain the deformation of the pegged component during transverse loading. When loading the pegged glenoid with an inferior eccentric force, the component exhibited reciprocal displacements with motion in the anterior to posterior plane occurring in one direction and motion in superior to inferior plane occurring in opposite directions. For example, under an inferior eccentric force the anterior and posterior aspects of component both displaced anteriorly while the superior aspect displaced inferiorly and the inferior aspect displayed superiorly. This pattern of movement is considered a deformation phenomenon, indicating that the component itself had become deformed and may be explained by the connection between the articulating component and the anchoring component, which seems to be larger in the keeled glenoid. Additionally, the cement bone interface may have a role, but the impact cannot be answered with this biomechanical protocol [
15].
The amount of bone removed when preparing the glenoid for implantation of the keeled or pegged components may have influenced the stability of the components. When investigating the different anchor sizes, we found a difference in volumes of the keeled (1022.6 mm3) versus the pegged (662.8 mm3) components indicating a greater amount of bone removal is required for the keeled glenoid. Additionally, the volume and weight of cement used during implantation showed no statistically significant differences between the two implants. These results may appear odd based on the assumption that a higher volume of implant would require more cement. A possible explanation of this finding may be that the preparation of the keeled component leads to the removal of more bone, which is typically cancellous bone. The cement for the keeled component is pressurized into the glenoid vault, which has less cancellous bone by virtue of the bone preparation, and therefore a better apposition to the cortical bone of the glenoid. With the pegged system, the bone removal is less, and therefore more of the cement is fixed within cancellous bone, which is less rigid and may deform under the testing conditions.
Over the last few decades the pegged and keeled glenoid components have been investigated regarding their ability to restore native glenoid function. Several biomechanical testing protocols and computer assisted finite element models were developed to determine which implant is more favorable. The proposed benefit of the pegged configuration is a more equal force distribution on the subjacent bone stock as demonstrated by finite element analysis [
16,
17]. In contrast to the pegged confirmation, the keeled implant was designed to allow for easier surgical implantation and it may have also been designed as a keel due to manufacturing limitations when the first keeled glenoids were made. Lacroix et al. [
18] compared the pegged versus keeled components and predicted that in 94% of pegged implants and 68% of keeled implants the cement has a greater than 95% probability of survival in normal bone. In bone of patients suffering from rheumatoid arthritis (RA), 86% of the pegged implants and 99% of the keeled implants were reported to have a greater than 95% probability of survival. Further, the results showed that bone stress is not substantially affected by the implant design, leading the authors to conclude that the pegged anchorage would be superior in the normal bone while keeled system would be superior for patients with rheumatoid arthritis or osteoporotic bone.
While the pegged component exhibited a greater amount of micro-motion during biomechanical testing, radiolucency was greater in patients with a keeled component. Edwards et al. [
5] randomized 53 patients undergoing total shoulder arthroplasty (TSA) to either a pegged or keeled glenoid implant. At initial post-op examination, there was no difference in radiographic findings, but after a mean follow-up of 26 months the rate of glenoid radiolucency was significantly higher in patients with keeled glenoids (46%) as compared to patients with pegged glenoids (15%) (
p = 0.003). Furthermore Gartsman et al. [
19] reported an increased rate of radiolucency in keeled implants after 6 weeks with a rate of 39% and a significantly lower rate in pegged implants with a rate of 5% (
p = 0.026). These findings are consistent with our results, showing more radiolucency for keeled compared to pegged glenoid components.
There is considerable debate regarding the relationship between radiographic findings and clinical failure. Long-term results from Torchia et al. [
20] suggest a positive correlation. Walch et al. [
21] reported that glenoid component failure is multifactorial and speculated that the preservation of glenoid bone stock is the most important factor in providing long-term resistance to the stress.
Other authors have not reported radiologic differences in patients with long-term follow up. Gazielly et al. [
22] reported on long-term survival of keeled glenoid components in TSA with a mean follow-up of 8.5 years using a bone compaction and cement pressurization technique. These results were comparable to pegged components with radiological glenoid loosening of 15.5%. Throckmorton et al. [
6] investigated 100 patients undergoing primary TSA with pegged and keeled glenoid components. At mean follow-up of 51.3 months 8% of pegged implants and 4% of keeled implants demonstrated radiographic lucency however there was no differences in clinical outcomes at intermediate-term follow-up (
p = 0.74). Walch et al. [
21] performed a multicenter study evaluating 518 TSA more than five years out from surgery. Radiographic loosening was present in 33% of the keeled components and was associated with three predominant patterns: 1) superior tilting, 2) subsidence and 3) posterior tilting. The authors proposed that the subchondral bone quality beneath the implant component is important to maintain the position of the glenoid over time.
The optimal method of long-term glenoid fixation has yet not been defined. Metal-back glenoids have the disadvantage of requiring more significant initial bone resection, risk of late metal on metal debris, increased overstuffing, and higher revision rates [
23]. Boileau et al. [
24] compared the cemented polyethylene glenoid to an unique uncemented metal-back glenoid component in a prospective, double blind randomized study. The results show more favorable outcomes with cemented polyethylene glenoids based on the significantly higher incidence of loosening with this unique metal-backed glenoid design as compared to polyethylene components. These findings are supported by the results from Fox et al. [
25] who investigated 1337 patients with 1542 TSA using 6 types of glenoids (cemented, not cemented, polyethylene, keeled, pegged and metal-back). They concluded that optimal implant survival was achieved with the cemented all-polyethylene glenoid components with 15-year follow-up. Cemented all-polyethylene pegged or keeled glenoids are widely considered the optimal implants, as their outcomes are believed to be the most reliable [
21,
25‐
27]. An additional advantage is the minimal amount of bone removal required for proper placement. Further research and development needs to be continued to determine the ideal shape of the glenoid and the method of fixation associated with the highest rate of radiographic and clinical stability.
There are limitations to this study. The in vitro nature of biomechanical evaluation can be a limiting factor in the translation of the findings to the in vivo conditions of the shoulder complex. This is particularly true for load distribution in shoulder replacement with its specific three-dimensional forces. Accurate replication of these forces in a cadaveric study is a challenge. Another limitation to this study is the fact that biomechanical testing has been performed with single loads, whereas occurrence of radiolucency is depended on repetitive cycles over the course of time in an actively remodeling system under ever changing loading scenarios. Thus, a correlation is hard and it is not clear if the cement bone interface has a significant contribution to this effect. In addition, the mean age of all cadaveric specimens was 61.9 ± 10.6 and therefore raises concerns about the quality of bone. However, shoulder replacement is commonly used in older patients and the bone mineral density of each cadaveric specimen was measured to ensure comparable results. Additionally, biomechanical testing with cadaveric specimens does not allow the effects of biological healing to be measured and, therefore, we are able to draw conclusions only for the primary stability of the joint at a time point immediately after implantation of the glenoid component. Furthermore, we did not evaluate the humeral head component position (imperfect or non-anatomical head replacement) on the radiographs as this may have affected the glenoid radiolucency and age difference in the compared groups may have influence on the radiographic findings, too (mean difference 6.7 years). Additionally, only two surgeons performed the arthroplasty. One only used keeled, the other one only used pegged components. This has a limitation by the surgeon and may provide a selection bias, but both surgeons are experienced and specialized shoulder surgeons.