The Stryker Perform and Univers Apex short stems have different patterns of stress shielding but comparably low rates of loosening after reverse total shoulder arthroplasty

Abstract

Background

The purpose of this study was to compare short-term radiographic outcomes between the Apex Univers Revers (Arthrex, Inc.) and the Stryker Perform Reversed (Stryker Inc.) short-stem reverse total shoulder arthroplasty (rTSA) implants.

Methods

A retrospective review was conducted of patients who underwent rTSA between 2021 and 2023 by 2 fellowship-trained surgeons at separate institutions. Consecutive primary cases with ≥2 years of follow-up were included. Immediate and final post-operative radiographs were reviewed for signs of loosening and stress shielding, which were graded in 5 zones and classified according to the Schnetzke bone adaptation grades. Partial or complete calcar osteolysis was also recorded. Multivariable regression was performed, and multiplicity was controlled within outcome families using the Holm method.

Results

A total of 40 Univers Revers and 46 Stryker Perform implants met the study criteria. Groups were similar in age, sex, diagnosis, and follow-up duration. By zone, remodeling was more frequent with Univers Revers in Zone 2 (32.5% vs. 7.5%; adjusted P = .05) and Zone 5 (35.0% vs. 10.0%; adjusted P = .070). All other zones showed no significant differences between groups (all adjusted P > .05). Partial calcar osteolysis was observed in 1 Univers Revers (2.5%) and 1 Stryker Perform (2.2%) (P = 1.000). The proportion of implants at risk for loosening was similar between Univers Revers (2.5%) and Stryker Perform (6.5%) (P = .620). Subsidence occurred in 1 Univers Revers (2.5%) and 2 Stryker Perform (4.3%) (P = 1.000); all stabilized without progression or need for revision. One of the Stryker Perform cases demonstrating subsidence also exhibited a minor stem shift, corresponding to 3 unique implants (6.5%) considered at risk for loosening.

Conclusion

There is no apparent difference in short-term risk for loosening between the Stryker Perform and Arthrex Apex Univers Revers short stem designs in rTSA. Remodeling patterns differed by zone, which may reflect the differing stem designs.

This study was conducted under approval of the Southern Oregon Institutional Review Board (SOIRB) (Asante Rogue Regional Medical Center, Asante Three Rivers Medical Center, and Asante Ashland Community Hospital; IRB approval number: SOIRB #15-001; IRB Protocol: Arthrex, Inc. Total Shoulder Arthroplasty Multi-Center Prospective Registry).The supplied IRB documentation confirms SOIRB #15-001 and identifies this registry protocol; the July 31, 2025, date pertains specifically to approval of Protocol Amendment 8, so I would not present it as the study's original approval date.

Reverse total shoulder arthroplasty (rTSA) indications continue to increase.² Historically a salvage procedure, it now has become routine in many instances to be implanted in young active patients.² The preservation of host bone in this patient population remains a focus of discussion, especially in light of data showing a 153% increase in revision shoulder arthroplasty over a recent 6-year span and projections of further growth.² Several manufacturers have developed short stems in an effort to preserve proximal humeral bone and minimize stress shielding.^5,27

Historically stress shielding was first discussed with the use of total hip implant designs with complete diaphyseal fit, reducing load in the proximal femur and resulting in osteolysis of the proximal femur via Wolff law.^5,28 While not weight bearing, humeral stems in early rTSA designs exhibited similar issues, with osteolysis of the proximal humerus in long-stem prostheses being more common than in their modern short-stemmed counterparts.²¹ Computational and finite element studies support these findings, but also show that malalignment and oversizing can negate the benefit.^17,21 Clinical series similarly demonstrate that high filling ratios and varus placement are associated with increased resorption, even in short-stem designs.^9,23,25 Despite these insights, few studies have directly compared outcomes between contemporary stem designs.

Radiographic measurements have highlighted the multifactorial nature of this phenomenon, with implant geometry, canal fill, and alignment serving as key determinants of remodeling. Stress shielding has been reported in 20% to 50% of arthroplasty cases, most often in the lateral metaphysis and calcar, though usually without early loosening or failure.^5,8 Meta-analyses confirm a high incidence, particularly in long stems and in patients with high filling ratios, but the clinical impact remains debated.^6,8,10 As a result, implant design has become a focus of recent investigation. Short-stem constructs promote metaphyseal load transfer and reduce proximal stress shielding compared with standard-length stems.⁴ Computational and finite-element analyses show that metaphyseal fixation lowers cortical stress, although malalignment or oversizing can negate this benefit, and clinical studies link high canal-fill ratios or varus alignment to increased remodeling even in short-stem designs.^9,20

Despite these shared goals, short-stem systems differ in geometry and fixation philosophy. The Stryker Perform Reversed (Stryker Inc.; Kalamazoo, MI, USA) employs a proximally porous-coated design intended for purely metaphyseal fixation, concentrating load within cancellous bone, whereas the Apex Univers Revers (Arthrex, Inc.; Naples, FL, USA) uses a proximally tapered configuration that engages both the metaphysis and proximal diaphysis along the calcar. The purpose of this study was to compare radiographic remodeling, signs of loosening, and clinical outcomes between the Apex Univers Revers and the Stryker Perform Reversed short-stem rTSA implants. We hypothesized that there would be no difference in risk for loosening or stress shielding between the 2 designs.

Materials and methods

This was a retrospective comparative study of patients who underwent rTSA between 2021 and 2023 at 2 independent institutions. Patients were included if they underwent a primary rTSA with either the Stryker Perform Reversed or the Arthrex Univers Revers and had minimum 2-year clinical and radiographic follow-up. Exclusion criteria were revision surgery, inadequate imaging, or incomplete clinical outcome data.

Surgical technique

All rTSAs were performed with a deltopectoral approach. One surgeon routinely used the Stryker Perform for all rTSAs (K.K.). This stem is available in short (65 mm) and standard (115 mm) lengths with a proximal porous coating and is designed for metaphyseal fixation (Fig. 1, A). The other surgeon used the Apex Univers for all rTSAs (P.J.D.) This stem is available in short (60–65 mm) and standard (95–147 mm) lengths with proximal grit blasting and a medial-lateral taper designed for both metaphyseal as well as proximal diaphyseal fixation at the level of the calcar (Fig. 1, B). Both surgeons broached the humeral canal with a ‘love at first bite’ technique, stopping once initial rotational control was achieved rather than continuing to cortical chatter. Autograft impaction grafting was not used in any cases. The goal was to achieve rotational stability and a flush fit of the humeral cut surface. All stems were placed with an intended humeral inclination of 135°. The subscapularis was routinely repaired by both surgeons. Post-operative rehabilitation was not standardized.

Figure 1.

Representative short-stem humeral component designs. (A) Perform Reversed (Stryker, Inc.) humeral component. (B) Univers Revers Apex (Arthrex, Inc.) humeral component.

Baseline demographic variables included age, sex, side, body mass index, smoking status, diagnosis, and length of follow-up. Functional outcomes at final follow-up included visual analog scale pain, American Shoulder and Elbow Surgeons (ASES), Single Assessment Numeric Evaluation scores. This information was used for post-operative evaluation only of implants considered to be at risk for loosening. Comparison between implants could not be performed as 1 surgeon did not obtain pre-operative functional outcome scores for pairwise comparison.

Immediate post-operative and 2-year follow-up anteroposterior radiographs obtained as part of routine post-operative follow-up were evaluated for signs of loosening or stress shielding using a five-zone system (Fig. 2).³ Radiographs were independently reviewed by 2 fellowship-trained shoulder surgeons who were not the treating surgeons. Signs of loosening were defined as radiolucency across 3 or more zones, subsidence on serial radiographs, or varus/valgus stem shift. Partial calcar osteolysis was recorded separately, and advanced adaptive changes were defined as Schnetzke Grade ≥2, representing moderate to severe cortical thinning, osteopenia, or metaphyseal resorption.^25,26 Inter-rater reliability was assessed using intraclass correlation coefficients (two-way random effects, absolute agreement), which demonstrated good to excellent agreement (intraclass correlation coefficient 0.70–0.94).¹⁵

Figure 2.

Five-zone classification of the proximal humerus used for radiographic assessment of stress shielding and bone adaptation. Zone 1 represents the proximal lateral cortex; Zone 2, the distal lateral cortex; Zone 3, the proximal medial cortex (calcar); Zone 4, the distal medial cortex; and Zone 5, the stem tip.

Statistical analysis was performed on the full cohort. Continuous variables were compared using Welch t-test or Mann–Whitney U test, as appropriate. Categorical variables were analyzed with chi-square or Fisher exact test. Multiplicity was controlled within outcome families using the Holm method. Multivariable linear and logistic regression models were constructed to identify independent predictors of clinical and radiographic outcomes. Statistical significance was defined as P < .05. Given the retrospective, exploratory nature of this study, no a priori power calculation was performed. A post hoc power analysis was conducted after data collection, which indicated that approximately 421 shoulders per group would be required to detect a 20% difference in loosening rates with 80% power at α = 0.05.

Results

Cohort overview

A total of 86 rTSAs met the study criteria, including 40 Arthrex Apex Univers Revers and 46 Stryker Perform stems (Table I). There was no difference in baseline demographics between groups.

Table I.

Demographic characteristics of the study cohorts

Variable	Arthrex Apex (n = 40)	Stryker Perform (n = 46)	P value
Age (years)	73.1 ± 7.0	70.9 ± 7.1	.144
Male sex	21 (53.0%)	24 (52.0%)	.878
Right side	22 (55.0%)	28 (60.9%)	.740
Osteoarthritis	22 (55.0%)	21 (46%)	.273
Cuff arthropathy	17 (43.0%)	25 (54%)	.273
Follow-up, months	27.0 ± 7.2 (8-50)	27.6 ± 5.4 (17-41)	.653

n, number; SD, standard deviation.

Radiographic findings

The distribution of Schnetzke grades did not differ significantly between groups (Table II). Grade 0 changes were observed in 9 Arthrex cases (22.5%) and 10 Stryker cases (21.7%) (P = 1.000). Grade 1 changes were most common, seen in 24 Arthrex cases (60.0%) and 29 Stryker cases (63.0%) (P = .642). Grade 2 changes occurred in 7 Arthrex cases (17.5%) and 5 Stryker cases (10.9%) (P = .518), while Grade 3 changes were present in 2 Stryker cases (4.3%) but absent in the Arthrex cohort (P = 1.000). Partial calcar osteolysis was observed in 1 Univers Revers (2.5%) (Fig. 3) and 1 Stryker Perform (2.2%) case (P = 1.000) (Fig. 3).

Table II.

Distribution of Schnetzke bone adaptation grades by implant

Grade	Arthrex Apex (n = 40)	Stryker Perform (n = 46)	P value
0	9 (22.5%)	10 (21.7%)	1.000
1	24 (60.0%)	29 (63.0%)	.642
2	7 (17.5%)	5 (10.9%)	.518
3	0 (0.0%)	2 (4.3%)	1.000

n, number.

Schnetzke Grade — 0 = no adaptive changes, 1 = mild adaptive changes, 2 = moderate adaptive changes, 3 = severe adaptive changes.

Figure 3.

Partial calcar osteolysis following short-stem rTSA with the Arthrex Apex Univers Revers humeral component. (A) Immediate post-operative radiograph. (B) Radiograph at approximately 15 months post-operatively demonstrating development of partial calcar osteolysis without subsidence or radiolucent lines in other zones. (C) Final follow-up radiograph at approximately 20 months post-operatively demonstrating persistent partial calcar osteolysis without progression to complete resorption. The implant was not classified as at risk for loosening according to the study criteria. rTSA, reverse total shoulder arthroplasty.

By zone, stress shielding was observed in 14 of 40 Arthrex cases (35%) and 20 of 46 Stryker cases (50%) in Zone 1 (P = .600). Zone 2 demonstrated stress shielding in 13 of 40 Arthrex cases (32.5%) and 3 of 46 Stryker cases (7.5%) (P = .05). Zone 3 changes were absent in the Arthrex group and present in 2 of 46 Stryker cases (5.0%) (P = .780). Zone 4 changes were infrequent, identified in 1 of 40 Arthrex (2.5%) and 1 of 46 Stryker (2.2%) (P = 1.000). In Zone 5, stress shielding occurred in 14 of 40 Arthrex cases (35%) and 4 of 46 Stryker cases (10%) (P = .070) (Table III).

Table III.

Stress shielding and radiolucency findings by zone and implant

Zone	Arthrex Apex	Stryker Perform	P value
	n (%)	n (%)
Zone 1
Stress shielding	14 (35.0%)	20 (50.0%)	.600
Radiolucency	11 (27.5%)	16 (40.0%)	.344
Zone 2
Stress shielding	13 (32.5%)	3 (7.5%)	.050
Radiolucency	0 (0.0%)	6 (15.0%)	.026
Zone 3
Stress shielding	0 (0.0%)	2 (5.0%)	.780
Radiolucency	2 (5.0%)	4 (10.0%)	.675
Zone 4
Stress shielding	1 (2.5%)	1 (2.5%)	1.000
Radiolucency	2 (5.0%)	3 (7.5%)	1.000
Zone 5
Stress shielding	14 (35.0%)	4 (10.0%)	.070
Radiolucency	8 (20.0%)	6 (15.0%)	.770

Radiolucency, presence of a radiolucent line ≥2 mm. Zones: Zone 1 = proximal lateral cortex; Zone 2 = distal lateral cortex; Zone 3 = proximal medial cortex (calcar); Zone 4 = distal medial cortex; Zone 5 = stem tip. Stress shielding is reported as presence or absence in each zone; cortical thinning grades are reported separately. P values are Holm-adjusted for multiple comparisons within outcome families.

Implants at risk of loosening

A total of 4 implants were identified as having radiographic risk variables for loosening, including 1 Arthrex (2.5%) and 3 Stryker (6.5%) short-stem rTSAs (P = .620) (Table IV). In the Stryker group, 1 patient showed radiolucency across ≥3 zones, 2 demonstrated subsidence, and one of the subsidence cases also exhibited a minor stem shift. The Arthrex group included 1 case of subsidence. The Arthrex subsidence case occurred in a 70-year-old male and was observed at 76 days post-operatively. The implant remained stable through 2 years of follow-up without the development of radiolucent lines (Fig. 4), and the patient had an ASES score of 98.3 and subjective shoulder value score of 99% at final follow-up. The 2 Stryker subsidence cases occurred in a 61-year-old female and a 68-year-old male and were first identified at 934 and 898 days post-operatively, respectively (≈916 ± 25 days). One of these cases demonstrated progressive calcar osteolysis with subsidence that subsequently stabilized without revision (Fig. 5). The 2 Stryker patients had ASES scores of 100 and subjective shoulder value scores of 100% at final follow-up. No infections, reoperations, or revisions were performed in either cohort during the study period.

Table IV.

Radiographic findings associated with risk of humeral stem loosening

Variable	Arthrex Apex (n = 40)	Stryker Perform (n = 46)	P value
Radiolucency ≥3 zones	0 (0.0%)	1 (2.2%)	1.000
Subsidence	1 (2.5%)	2 (4.3%)	1.000
Stem shift	0 (0.0%)	1 (2.2%)	1.000
Implants at risk	1 (2.5%)	3 (6.5%)	.620

Figure 4.

Arthrex Apex Univers Revers case demonstrating post-operative subsidence. (A) Immediate post-operative radiograph. (B) Radiograph demonstrating post-operative subsidence. The implant remained stable through 2 years of follow-up without development of radiolucent lines.

Figure 5.

Progressive complete calcar osteolysis and subsidence following short-stem rTSA with the Stryker Perform Reversed humeral component. (A) Immediate post-operative radiograph. (B) Early post-operative follow-up radiograph demonstrating initial cortical thinning. (C) Interval radiograph demonstrating progressive calcar osteolysis. (D) Subsequent interval radiograph demonstrating complete calcar osteolysis with continued humeral stem subsidence. (E) Final follow-up radiograph demonstrating persistent complete calcar osteolysis with stabilized subsidence. The implant remained clinically stable without revision at final follow-up. rTSA, reverse total shoulder arthroplasty.

Discussion

This comparative study evaluated radiographic remodeling and early implant stability after rTSA with 2 contemporary short-stem designs. There was no difference in risk for loosening between the Arthrex Univers Revers and Stryker Perform Reversed implants at short-term follow-up. The overall rate of stress shielding was also similar between groups. The most notable difference in stress shielding between groups was a higher frequency of Zone 2 stress shielding with the Arthrex component. Collectively, these findings suggest that implant design may influence regional remodeling patterns without compromising early fixation or function.^1,5,8

Stem geometry and fixation philosophy are major determinants of bone adaptation in rTSA.^9,11,28 Short stems that achieve metaphyseal fixation minimize diaphyseal engagement and reduce cortical stress shielding while maintaining proximal load transfer.^16,25 Technical and patient factors such as undersizing, varus alignment, and low bone mineral density may also contribute to the severity and distribution of remodeling.^9,17,18 Together, these data underscore the complex interaction between implant design, surgical technique, and bone quality in shaping remodeling patterns after reverse shoulder arthroplasty.

Contemporary rTSA studies have demonstrated variable patterns of proximal humeral remodeling influenced by implant geometry and fixation philosophy. Goñalons-Giol and Ventura-Parellada identified remodeling most commonly within the lateral cortex (Zone 2 in the seven-zone system) and highlighted female sex as a risk factor,⁸ findings consistent with alignment-dependent patterns described by Abdic et al¹ and Kramer et al.¹⁴ Whereas Inoue et al¹² observed remodeling in Zones 1 and 3 with pedestal formation in Zone 4. In the present cohort, Zone 2 remodeling (lateral cortex) was more common in the Arthrex design, mirroring the pattern described by Goñalons-Giol and Ventura-Parellada.⁸ This finding suggests that stem geometry, specifically partial diaphyseal engagement and proximal canal fill, may alter regional loading patterns and remodeling distribution. Differences in fixation philosophy between the 2 stems may contribute to this observation. The Apex Univers employs a proximally tapered stem that engages both the metaphysis and proximal diaphysis along the calcar, creating a mixed load-transfer pathway that increases cortical stress in this region. In contrast, the Stryker Perform relies on purely metaphyseal fixation through a porous-coated proximal body, limiting diaphyseal engagement and concentrating load more proximally. These design differences may contribute to the observed variation in regional remodeling; however, the influence of implant geometry cannot be isolated from unmeasured technical factors such as stem alignment and canal fill.

Prior systematic reviews comparing short vs. standard-length stems have reported similarly low loosening rates with short stems (Ciuffreda et al,⁴ Harrell et al¹⁰), though those studies evaluated stem length as the primary variable rather than comparing designs within the same stem-length category.²¹ The present study differs by directly comparing 2 short-stem designs with distinct fixation philosophies. Our series demonstrated mild remodeling predominantly in both cohorts, with advanced changes uncommon. Zone 2 remodeling was more frequent with the Arthrex design, although no apparent difference in loosening or instability was observed between groups.

There was no statistical difference in radiographic risk for loosening between the Univers Revers and Stryker Perform short-stem implants (2.5% vs. 6.5%; P = .620). A post hoc power analysis indicated that approximately 421 shoulders per group would be required to detect a 20% difference in loosening rates with 80% power at α = 0.05. Given that the present study included only 40 and 46 shoulders per group, respectively, it was substantially underpowered to detect differences in loosening rates. Accordingly, the observed 4-percentage-point difference that did not reach statistical significance may reflect insufficient power and should not be interpreted as equivalence. A Type II error cannot be excluded.

Notably, the cases that subsided appeared to stabilize without progression, and there was no need for revision. In the present cohort, the isolated cases of subsidence may represent early adaptive settling rather than mechanical failure, as rTSA stems are loaded predominantly in compression. However, this interpretation is based on a limited number of cases in this study, and whether subsidence in short-stem rTSA is consistently benign requires validation in larger, prospectively designed cohorts. That said, this interpretation aligns with prior biomechanical studies showing that reverse shoulder stems experience compressive rather than tensile forces that favor stable fixation.^20,24 Clinical series have likewise demonstrated that minor subsidence and proximal remodeling rarely progress to loosening or compromise fixation.^7,8,26 Consensus recommendations emphasize standardized radiographic surveillance to identify subtle bone adaptations and late instability, underscoring the need for adequately powered studies to further define these risks.²⁸

Limitations

This study has limitations, including its retrospective design, reliance on AP radiographs, and 2-year follow-up, which likely underestimates long-term remodeling. Second, the study was not powered to detect small differences in loosening between implant designs, as indicated by the post hoc analysis. Longitudinal studies have shown that stress shielding continues to progress beyond 2 years, with Khorram et al¹³ reporting increased severity between 2 and 5 years, Navas et al¹⁹ noting new resorption between 4 and 6 years, and Raiss et al²² observing bone loss in nearly 40% of patients at 5 years. Stem alignment and canal fill ratio were not assessed in this study; prior work has specifically evaluated the relationship between these technical factors and stress shielding patterns in short-stem shoulder arthroplasty.^9,21,22 Differences in these parameters between groups may therefore confound the attribution of remodeling differences to implant geometry alone, and future studies should incorporate these measurements. Additionally, 1 or more authors have a financial relationship with Arthrex, Inc., which may represent a potential source of bias in the interpretation of findings related to the Arthrex implant. A performance bias exists inherent to this study design, as each implant was implanted exclusively by a single surgeon at a separate institution. Observed differences in remodeling patterns may therefore reflect variation in surgical technique, patient selection, or institutional practice rather than implant design alone. Finally, longer-term follow-up and larger multicenter cohorts are needed to determine whether early adaptive settling influences fixation durability over time. Nevertheless, these data demonstrate differences in the distribution of remodeling between the 2 implant cohorts while showing no apparent difference in early fixation or function.

Conclusion

There is no apparent difference in short-term risk for loosening between the Stryker Perform and Arthrex Apex Univers Revers short stem designs in rTSA. Remodeling patterns differed by zone, which may reflect the differing stem designs.

Disclaimers:

Funding: No funding was disclosed by the authors.

Conflicts of interest: Patrick J. Denard reports that he has received consulting fees and royalties from Arthrex, Inc. for work outside the submitted work. Kevin Kruse reports that he has worked as a consultant for Arthrex, Inc. Any additional authors, their immediate families, and any research foundations with which they are affiliated have not received any financial payments or other benefits from any commercial entity related to the subject of this article.