Does excessive lateralization influence tuberosity healing in reverse total shoulder arthroplasty for proximal humeral fractures?

Abstract

Background

This study evaluated the influence of stem global lateralization on tuberosity consolidation and functional outcomes in patients undergoing reverse total shoulder arthroplasty (rTSA) for proximal humeral fractures (PHFs).

Methods

A retrospective, two-institution study included patients who underwent rTSA for acute PHFs implanted with one of 3 stems: the Arrow (global lateralization 21.4 mm), the Humelock II (11.2 mm), or the Unique (14.7 mm). Stems were classified as very highly lateralized (VHL; >20 mm, Arrow) or lateralized (L; 10-15 mm, Unique and Humelock II). Tuberosity consolidation was assessed radiographically at a minimum 12-month follow-up. Functional outcomes included forward flexion, external rotation, visual analog scale pain, and Constant Murley score (CS). The lateralization shoulder angle (LSA) and distalization shoulder angle were measured and compared across groups. Complication and reoperation rates were recorded.

Results

A total of 113 patients were included: Arrow (N = 53), Humelock II (N = 40), and Unique (N = 20), comprising 53 in the VHL group and 60 in the L group. The overall consolidation rate was 82.1%. Nonunion rates differed significantly across stems (P = .018): 24.5% for the Arrow, 10.0% for the Humelock II, and 0% for the Unique, with a significant dose–response trend between increasing lateralization and increasing nonunion rate (P = .022). Patients achieving consolidation demonstrated significantly better forward flexion, external rotation, CS, and visual analog scale pain than those with nonunion (P < .05). Even among consolidated patients, the VHL group demonstrated significantly lower forward flexion, higher pain scores, and lower CS than the L group (P < .05). LSA values were similar across all consolidated groups despite a nearly 10 mm difference in global lateralization (P = .112).

Conclusion

A significant dose–response relationship between increasing stem lateralization and tuberosity nonunion rate was observed, with functional consequences extending beyond nonunion alone. Post-operative LSA does not reliably reflect the local mechanical environment at the tuberosity-implant interface. These findings identify proximal stem lateralization as a potentially important modifiable determinant of outcomes after fracture rTSA and underscore the need for prospective studies to define the threshold of lateralization compatible with reliable tuberosity healing.

The use of reverse total shoulder arthroplasty (rTSA) for the surgical management of Proximal humeral fractures (PHFs) in older adults has risen significantly over the past decade, primarily because of its predictable functional outcomes and lower complication rates compared to osteosynthesis and hemiarthroplasty.^1,38 Early designs followed Grammont's medialization principles, which improved pain control and deltoid function but introduced complications including scapular notching, instability, and limited range of motion (ROM).^{20,22,29,38,41},⁴² Lateralized prostheses were subsequently developed to address these limitations by restoring a more anatomic center of rotation; however, excessive lateralization has been associated with soft-tissue overtensioning, stress fractures, nerve injuries, and suboptimal functional outcomes.^3,13,17,23, 24, 25^,27,31,35,39,⁴⁰

Regardless of implant design, anatomical healing of the tuberosities remains a crucial objective in the treatment of PHFs, as it is strongly associated with improved function and reduced complications.^15,21,36 Anatomical restoration of the proximal humerus, including its lateral offset, enhances the wrapping angle of the deltoid, thereby improving its length–tension relationship and increasing its mechanical advantage for shoulder elevation and abduction.4, 5, 6 In this context, the degree of prosthetic lateralization is not only a biomechanical consideration for joint stability but a potentially decisive factor in the local mechanical environment at the tuberosity repair site.

Currently, surgeons use implants with varying degrees of lateralization to treat complex PHFs, yet the literature has not clearly defined the optimal degree of lateralization in this setting, and the direct influence of stem lateralization on tuberosity consolidation rates remains incompletely characterized. We hypothesize that prostheses with greater degrees of global lateralization may impair tuberosity healing by generating excessive mechanical tension at the repair site. The objective of this study was to evaluate the influence of stem lateralization on tuberosity consolidation rates and clinical outcomes in patients undergoing rTSA for PHFs.

Materials and methods

A retrospective, descriptive study was conducted across 2 institutions, including all patients who underwent rTSA for 3- and 4-part proximal humeral fractures. This case series protocol was reviewed and approved by the institutional review board (EMSJBIRB-28-2025).

Patient cohorts

The first cohort included patients treated at the primary institution between January 2015 and December 2023. All procedures were performed by D.G. The second cohort comprised patients treated at a secondary institution between August 2018 and June 2020, with all procedures performed by A.R. Data from both cohorts were collected via retrospective review of patient charts and operative documentation.

Inclusion and exclusion criteria

Inclusion criteria for both cohorts were: age 65 years or older, 3- or 4-part displaced proximal humeral fractures according to Neer classification, treatment with rTSA within 4 weeks of injury, and a minimum 12-month clinical and radiological follow-up.²⁸ Exclusion criteria included prior surgery or infection in the affected shoulder, pre-existing neurovascular compromise with signs of deltoid denervation, pathological fractures, and loss to follow-up or incomplete data.

Implant selection by institution

At the primary institution, the Arrow prosthesis was used exclusively from January 2015 through July 2022. In August 2022, the Unique prosthesis became the standard implant following changes to local hospital policies and contracts. At the secondary institution, all patients received the Humelock II prosthesis.

Lateralization measurement and classification

Global lateralization is calculated as the sum of humeral and glenoid lateral offsets. Humeral lateral offset represented the combined humeral stem and insert offsets, while glenoid lateral offset represented the perceived glenosphere radius and center of rotation offset. Relative global lateralization was determined by subtracting the global lateralization of the reference Grammont design (Delta III) from each implant configuration and expressing the difference as a percentage of a standardized baseline.^26,40,⁴³

To evaluate the impact of prosthetic lateralization on greater tuberosity healing, we employed the classification system proposed by Werthel et al,⁴³ which stratifies rTSA implants based on their relative global lateralization: medialized (<5 mm), minimally lateralized (5-10 mm), lateralized (10-15 mm), highly lateralized (15-20 mm), and very highly lateralized (>20 mm). Based on this classification, patients were grouped according to the prosthesis used:

• 1.L group (lateralized): Patients treated with either the Unique prosthesis (Bioprotece, Villa Ballester, Buenos Aires, Argentina), with a global lateralization of 14.7 mm, or the Humelock II prosthesis (FX Solutions, Viriat, France), with a global lateralization of 11.2 mm.
• 2.VHL group (very highly lateralized): Patients treated with the Arrow prosthesis (FH Orthopedics, Mulhouse, France), which has a global lateralization of 21.4 mm.

Data recollection

Demographic information and operative data were collected. Pre- and post-operative ROM, functional outcome scores, and subjective pain scores were obtained. ROM measures assessed included active forward flexion (FF), external rotation (ER), and internal rotation (IR). IR was evaluated as the highest vertebral column the patient could reach with their thumb behind their back and was scored based on the following scale: no IR, 0; buttocks, 2; sacrum, 4; lumbar, 6; and thoracic, 8. Patient-reported outcome scores collected included the Constant Murley score (CS). The visual analog scale (VAS) was used to assess the patient's subjective pain at follow-up, rated on a scale of 0 to 10. Radiographs were obtained at routine post-operative follow-up visits and were reviewed by the performing surgeon. Complications and reoperations were also recorded. Complications included wound dehiscence, infection, dislocation, hematoma, seroma, and neurovascular injury.

Radiographic analysis

Pre-operative X-rays and computed tomography scans were reviewed to classify fracture patterns according to the Neer classification.²⁶ Post-operative follow-up was conducted using anteroposterior (AP), lateral, and axial shoulder radiographs obtained immediately after surgery and at 3, 6, and 12 months, as well as at the last available follow-up. Radiographic analysis included assessing implant positioning, subsidence, loosening, tuberosity position and healing, scapular notching, periprosthetic fractures, and scapular fractures.

Consolidation (union) of the tuberosity was defined on the AP shoulder radiograph in neutral rotation as visualization of the greater tuberosity positioned laterally to the humeral stem and in contact with the diaphysis. (Fig. 1). Failure of consolidation (nonunion) was diagnosed based on radiographic signs of nonunion or bone resorption (lysis) visible on AP shoulder and axial scapular radiographs (Fig. 2).

Figure 1.

Intra-operative photograph of tuberosity repair technique.

Figure 2.

Anteroposterior shoulder radiograph demonstrating anatomic consolidation of the greater tuberosity.

The lateralization shoulder angle (LSA) and distalization shoulder angle (DSA) were measured on the last available AP radiograph to characterize the angular orientation of the reconstructed joint in vivo. LSA was defined as the angle formed between 2 lines: one connecting the superior glenoid tubercle to the most lateral border of the acromion, and another connecting the most lateral border of the acromion to the most lateral border of the greater tuberosity.⁷ DSA was defined as the angle formed between a line connecting the most lateral border of the acromion to the superior glenoid tubercle and a line connecting the superior glenoid tubercle to the most superior border of the greater tuberosity.⁷

Statistical analysis

Numerical variables were described using the mean, median, and standard deviation, as appropriate, based on their distributions. Qualitative variables were reported as frequencies and percentages. The Student t-test (or Mann–Whitney U test, when applicable) was used to compare numerical variables between 2 groups, and the chi-square test (or Fisher exact test) was used for categorical variables. One-way analysis of variance (ANOVA) was used to simultaneously compare numerical variables across 3 stem groups. Where ANOVA reached significance, pairwise post hoc comparisons were performed using Fisher exact test. A Cochran-Armitage test for trend was applied to evaluate the dose–response relationship between increasing global lateralization and nonunion rate across the 3 stem designs. Nonunion rates were reported with 95% confidence intervals (CIs) calculated using the Clopper-Pearson exact method. A Bonferroni correction was applied for multiple pairwise comparisons, with a corrected significance threshold of α = 0.017 for 3 simultaneous comparisons. For significant between-group differences in functional outcomes, effect sizes were calculated using Cohen d to quantify the magnitude of differences. The number needed to treat was calculated to estimate the clinical impact of stem selection on the prevention of nonunion. A P value of less than .05 was considered statistically significant for all analyses. Data were entered into an Excel spreadsheet (Microsoft, Redmond, USA). All statistical calculations were performed using Python (version 3.x; Python Software Foundation) with the SciPy and NumPy libraries.

Surgical technique

All patients were positioned in the beach-chair position, and a standard deltopectoral approach was used in each case. For glenoid component implantation, a 10° inferior tilt was established, corresponding to a reverse shoulder angle of 0°.⁵ Glenoid preparation involved reaming followed by central drilling to accommodate either the keel (VHL group) or the post (L group). The metaglene was then impacted and secured with peripheral screws. In the VHL group, 2 nonlocking screws were used. In the L group, which employed a circular metaglene, 4 screws were inserted: 2 locking and 2 nonlocking.

Afterward, the glenosphere was attached and secured with a central screw. The humeral components for the Arrow and Unique prostheses were cemented in 20° of retroversion, guided by an alignment rod parallel to the flexed forearm. A cement restrictor was inserted into the medullary canal before cementation. Following the placement of the polyethylene insert, the final reduction was performed to ensure proper stability and ROM. In contrast, the Humelock II stem was press-fit (noncemented) and implanted with 20° of retroversion. It was secured with 2 locking screws to stabilize the metadiaphysis.

Tuberosity fixation was performed using the technique described by Boileau, aiming for anatomical reduction (Fig. 3). Tuberosities were secured laterally to the humeral stem using a cerclage technique, with 4 horizontal and 2 vertical sutures for robust stabilization.^5,15

Figure 3.

Shoulder X-ray imaging demonstrating radiographic signs of pseudoarthrosis.

Rehabilitation protocol

All patients followed the same rehabilitation protocol. Patients used an arm sling in IR for 4 to six weeks. Starting on the first post-operative day, patients began self-directed rehabilitation, including pendulum exercises for the shoulder and active ROM exercises for the elbow, wrist, and hand.

Starting in the third week, patients initiated guided passive ROM exercises under the supervision of a physical therapist. Once the sling was discontinued, active shoulder mobility was permitted. Muscle strengthening exercises began 90 days post-operatively, followed by a gradual return to all strength-dependent activities.

Results

Final cohorts

At the primary institution, 82 patients were initially identified between January 2015 and December 2023. One patient (1.2%) with pre-operative axillary nerve neurapraxia was excluded, and 8 patients (9.9%) were lost to follow-up, yielding a final cohort of 73 patients: 53 received the Arrow prosthesis (VHL group) and 20 received the Unique prosthesis (L group). At the secondary institution, 40 patients treated with the Humelock II prosthesis met the inclusion criteria and were assigned to the L group. The final study population comprised 60 patients in the L group and 53 patients in the VHL group.

Patient demographics

A total of 113 patients underwent rTSA for acute PHFs and were included in this study: 53 implanted with the Arrow stem, 20 with the Unique stem, and 40 with the Humelock II. The Unique and Humelock II cohorts were combined into a single L group (L; N = 60) for comparative analyses. Patient characteristics are summarized in Tables I and II.

Table I.

Patient characteristics of Arrow, Unique, and Humelock II cohorts.

Demographic variables	Arrow	Unique	Humelock II
Age (yr)	77.1 ± 6.3	74.1 ± 6.0	78.3 ± 4.9
Gender
Female	46 (87%)	16 (80%)	40 (100%)
Male	7 (13%)	4 (20%)	0 (0%)
Operated side
Right	32 (60%)	14 (70%)	30 (75%)
Left	21 (40%)	6 (30%)	10 (25%)
NEER classification
3-part	28 (53%)	15 (75%)	31 (78%)
4-part	25 (47%)	5 (5%)	9 (22%)
Days to surgery	13.3 ± 8.0	8.5 ± 4.5	4.1 ± 3.4

Table II.

Patient characteristics of the VHL and the LL group.

Demographic variables	VHL	L
Age (yr)	77.1 ± 6.3	76.4 ± 4.3
Gender
Female	46 (87%)	56 (93%)
Male	7 (13%)	4 (7%)
Operated side
Right	32 (60%)	44 (73%)
Left	21 (40%)	16 (26%)
NEER classification
3-part	28 (52.8%)	46 (76.7%)
4-part	25 (47.2%)	14 (23.3%)
Days to surgery	12.8 ± 6.7	5.6 ± 4.3

VHL, very highly lateralized; L, lateralized.

Mean age was 77.1 ± 6.3 years in the Arrow group, 74.1 ± 6.0 years in the Unique group, and 78.3 ± 4.9 years in the Humelock II group (range: 61-94 years), with no significant difference across the 3 stems (ANOVA F = 2.96, P = .06) or between the VHL and L groups (P = .517). Mean follow-up was 27.3 ± 7.2 months across the entire cohort.

The cohort was predominantly female, consisting of 102 women (90.2%) and 11 men (9.8%). All patients in the Humelock II cohort were female (N = 40), compared to 87% in the Arrow group (N = 46) and 80% in the Unique group (N = 16). The L group consisted of 56 females, and the VHL group consisted of 46 females. Gender distribution did not differ significantly between the VHL and L groups (P = .343).

The average time from fracture occurrence to surgical intervention was 8.7 ± 6.6 days (range: 1-28 days).

Regarding fracture classification, 74 patients (65.5%) had Neer 3-part proximal humeral fractures, and 39 patients (34.5%) had Neer 4-part fractures. No statistically significant differences were found between the VHL and L groups in terms of patient demographics or fracture characteristics.

Tuberosity consolidation and nonunion rates

The overall nonunion rate was 17.0% (N = 17/98). Nonunion rates differed significantly across the 3 stem designs (χ² = 8.07, P = .018). The Arrow stem demonstrated the highest nonunion rate at 24.5% (13/53; 95% CI: 13.8-38.3%), compared to 0% for the Unique stem (0/20; 95% CI: 0-16.8%) and 16.0% for the Humelock II stem (4/25; 95% CI: 2.8-23.7%). Pairwise Fisher exact tests demonstrated a significant difference between the VHL and L groups (P = .016, odds ratio [OR] = 0.22) (Tables III and IV).

Table III.

Comparison of consolidation rate between the Arrow, Unique, and Humelock II cohorts.

Prosthesis	Union	Nonunion	Total
Arrow	40 (75.5%)	13 (24.5%)	53 (100%)
Unique	20 (100%)	0 (0%)	20 (100%)
Humelock	36 (90.0%)	4 (10.0%)	40 (100%)
Total	96 (85.0%)	17 (15.0%)	113 (100%)

Table IV.

Comparison of consolidation rate between the VHL group and the L group.

Group	Union	Nonunion	Total
VHL	40 (75.5%)	13 (24.5%)	53 (100%)
L	56 (93.3%)	4 (6.7%)	60 (100%)
Total	96 (85.0%)	17 (15.0%)	113 (100%)

VHL, very highly lateralized; L, lateralized.

A statistically significant trend of increasing nonunion rate with greater global lateralization was identified (Cochran-Armitage test, P = .022). For every 4 patients implanted with the Unique stem instead of the Arrow, one additional nonunion would be prevented (number needed to treat = 4.1); this figure was 5.6 for the combined Unique and Humelock II group.

Lateralization shoulder angle

LSA values across consolidated cases were similar across the 3 stem designs: 89.3 ± 9.0° for Arrow, 86.0 ± 7.8° for Unique, and 91.0 ± 6.9° for Humelock II. One-way ANOVA across the 3 consolidated groups did not reach statistical significance (F = 2.24, P = .112), indicating that LSA does not reflect the degree of stem lateralization. Nonunion cases demonstrated consistently lower LSA values (mean 84.7°) compared to all consolidated groups (mean 91.6° ± 9.6°, P = .009).

Distalization shoulder angle

The DSA did not differ significantly across the 3 stem designs (Kruskal–Wallis H = 2.15, P = .342), with mean values of 43.9 ± 9.9° for the Arrow group, 46.5 ± 10.8° for the Unique group, and 44.6 ± 8.7° for the Humelock II group. No pairwise comparison reached statistical significance (Arrow vs. Unique P = .168, Arrow vs. Humelock II P = .854, Unique vs. Humelock II P = .221). Similarly, no significant difference in DSA was observed between the VHL and L groups (43.9 ± 9.9° vs. 45.4 ± 9.6°, P = .375). The mean difference between any 2 groups was less than 3°, with negligible effect sizes across all comparisons (d < 0.27).

Comparison of clinical outcomes of each stem

Among patients with tuberosity consolidation, functional outcomes were comparable between Arrow and Unique consolidated groups across all domains except VAS pain, where Unique patients reported significantly lower pain scores (0.9 vs. 1.6, P = .024). The Humelock II consolidated group demonstrated significantly better FF (150.0 ± 20.1° vs. 133.6 ± 14.0°, P = .001), ER (39.0 ± 13.7° vs. 30.4 ± 10.2°, P = .024), lower VAS pain (0.9 vs. 1.6, P = .014), and higher CS (79.2 ± 17.6 vs. 68.3 ± 6.8, P = .004) compared to the Arrow consolidated group (Table V).

Table V.

Range of motion and patient-reported outcome measures of Arrow, Unique, and Humelock II cohorts.

Functional outcomes	Arrow	Unique	Humelock II
Forward flexion (°)	128.8 ± 15.3	137 ± 14.0	141.2 ± 27.6
External rotation (°)	26.7 ± 11.6	32 ± 10.8	39.2 ± 12.9
Internal rotation	Lumbar	Lumbar	Lumbar
Constant score	64.9 ± 9.0	70.4 ± 7.0	75.5 ± 18.6
Visual analog scale	1.7 ± 1.1	0.9 ± 1.1	1.1 ± 1.3

VHL vs. L group comparison

When all patients were grouped as either VHL or L group, the L group demonstrated significantly better outcomes across FF (139.4 ± 22.5° vs. 128.8 ± 15.3°, P = .003, d = 0.56), ER (36.0 ± 12.4° vs. 26.7 ± 11.6°, P = .001, d = 0.78), VAS pain (1.0 ± 1.2 vs. 1.7 ± 1.1, P = .003, d = 0.59), and CS (73.2 ± 14.7 vs. 64.9 ± 9.0, P = .003, d = 0.70). IR did not differ between groups (P = .186). When restricted to consolidated cases only, the L group maintained significantly better FF (143.7° vs. 133.6°, P = .006), VAS pain (0.9 vs. 1.6, P = .004), and CS (74.9 vs. 68.3, P = .014), with a trend toward better ER (35.6° vs. 30.4°, P = .076) (Table VI).

Table VI.

Range of motion and patient-reported outcome measures the VHL and the L group.

Functional outcomes	VHL	L
Forward flexion (°)	128.8 ± 15.3	139.4 ± 22.5
External rotation (°)	26.7 ± 11.6	36.0 ± 12.4
Internal rotation	Lumbar	Lumbar
Constant score	64.9 ± 9.0	73.2 ± 14.7
Visual analog scale	1.7 ± 1.1	1.0 ± 1.2

VHL, very highly lateralized; L, lateralized.

Nonunion vs. consolidated outcomes

Pooled across all stems, patients with tuberosity nonunion (N = 17) demonstrated significantly worse FF (109.4 ± vs. 138.7°, P < .001, d = 1.82), ER (21.4° vs. 33.0°, P = .001, d = 0.97), and CS (54.7 vs. 71.6, P < .001, d = 1 .56) compared to consolidated patients (N = 96). VAS pain trended toward significance (P = .050). Within the Arrow stem, nonunion was associated with significantly worse FF (113.8° vs. 133.6°, P < .001), ER (15.6° vs. 30.4°, P < .001), IR (P = .016), and CS (54.3 vs. 68.3, P < .001), while VAS pain did not differ (P = .889). Within the Humelock II stem, nonunion was associated with significantly worse FF (95.0° vs. 150.0°, P = .002), higher VAS pain (2.2 vs. 0.9, P = .036), and lower CS (55.8 vs. 79.2, P = .026), but ER and IR were preserved (40.0° vs. 39.0°, P = .818, and 5.5 vs. 5.0, P = .967, respectively). When comparing the Arrow nonunion directly against the Humelock II nonunion, ER was significantly better in the Humelock II nonunion group (40.0° vs. 15.6°, P = .004), while the CS did not differ (55.8 vs. 54.3, P = .732).

Complications

The overall complication rate across all 113 patients was 5.3% (N = 6/113). Individual stem rates were 5.7% for Arrow (N = 3/53, 95% CI: 1.2-15.7%), 10.0% for Unique (N = 2/20, 95% CI: 1.2-31.7%), and 2.5% for Fx (N = 1/40, 95% CI: 0.1-13.2%). Complications included a periprosthetic fracture, a chronic periprosthetic infection, and implant subluxation in the Arrow cohort. Two complications occurred in the Unique subgroup (1 periprosthetic fracture, 1 acromial fracture), and one in the Humelock II subgroup (acromial fracture) (Table VII). When grouped as VHL vs. L, complication rates were 5.7% for VHL (N = 3/53) and 5.0% for the L group (N = 3/60) (OR = 0.88, P = 1.000).

Table VII.

Complications.

Complications	Arrow	Unique	Humelock II
Periprosthetic fracture	1	1	0
Acromion fracture	0	1	1
Implant subluxation	1	0	0
Chronic prosthetic infection	1	0	0
Total	3	2	1

Discussion

The present study investigated the relationship between stem global lateralization and tuberosity consolidation in patients undergoing rTSA for Neer 3- and 4-part PHFs across 3 stem designs. The most clinically significant finding was the marked difference in nonunion rates (24.5% vs. 10.0% and 0%), with a significant dose–response trend between increasing global lateralization and increasing nonunion rate (P = .022). Analysis of post-operative LSA revealed similar values across all 3 consolidated groups despite a nearly 10-mm difference in global lateralization (P = .112); however, nonunion cases showed lower LSA values than consolidated cases, likely reflecting a measurement artifact in which failed tuberosity healing causes medial shift of the lateral reference landmark, producing a smaller angle as a consequence rather than a cause of nonunion. More broadly, LSA reduces the complex three-dimensional geometry of the implant to a single scalar value and does not characterize the mechanical environment at the tuberosity-implant interface. Importantly, the consequences of excessive lateralization extended beyond nonunion alone: even among consolidated patients, the VHL group demonstrated significantly lower FF and CS and higher pain scores than the L group. While these findings identify stem lateralization as a potentially important modifiable factor, tuberosity consolidation remains a multifactorial process governed by complex interactions among biological and biomechanical variables, and the present study does not establish lateralization as an independent cause of nonunion.

Reported rates of tuberosity consolidation after rTSA for fractures vary widely in the literature, ranging from 37% to 90%.^2,8, 9, 10, 11, 12^{,16,18,21,34,35,39} Tuberosity healing is now widely accepted as being associated with improved functional outcomes and lower complication rates.³⁷ In a multicenter study, Sabah et al³⁶ reported a significant association between anatomic healing and higher CS, Subjective Shoulder Value, FF, and ER, as well as a reduced incidence of prosthetic dislocation and humeral loosening. Similarly, Gallinet et al¹⁵ found that a 57% consolidation rate was associated with a lower instability rate (P < .01). In a meta-analysis of 376 cases, Buchanan et al¹⁶ reported a 78.5% tuberosity healing rate, correlated with improved ROM and CS (P < .001). Our findings are consistent with this body of evidence: with an overall consolidation rate of 82.1%, patients who achieved tuberosity healing demonstrated significantly better FF, ER, and CS, as well as lower VAS pain scores (P < .05).

However, a limitation of all these studies is the indiscriminate grouping of implants with different lateralization profiles, without analyzing how progressive increases in lateralization may alter the biomechanical environment for healing. This is a critical issue, especially in elderly patients, where bone quality is often compromised by osteoporosis, making the tuberosities fragile structures prone to migration, resorption, or osteointegration failure if subjected to excessive tension.⁶ While lateralization of the center of rotation has demonstrated biomechanical advantages in rotator cuff arthropathy, including improved ROM, extrapolating these benefits to fracture cases introduces unique challenges.^9,14,32 The tension that benefits cuff performance in cuff tear arthropathy may be detrimental in the early fracture setting, where tuberosities must heal without excessive traction. Several biomechanical studies have shown that excessive lateralization increases soft-tissue tension, creating a suboptimal environment for osteointegration in fragile bone. Experimental evidence suggests that excessive stress at the repair site may promote fibrous rather than bony healing by inducing micromovements and deformations.^11,33 In this context, exaggerated lateralization may lead to repetitive microdisplacements of the fixed tuberosities, alter contact pressures, and compromise local neovascularization.

In the study by Dot Pascuet et al,³¹ a 54% rate of tuberosity nonunion was reported in fractures treated with rTSA using a highly lateralized implant with a large-diameter humeral tray (44 mm). The authors attributed this high rate to intraoperative difficulties in achieving proper anatomical repair due to excessive tension generated by the implant's size. This finding is consistent with our study, in which the most lateralized stem had the highest nonunion rate. The convergence of post-operative LSA values across all 3 stem designs in our study suggests that joint-level reconstruction was equivalent. The similar LSA across consolidated groups therefore supports the interpretation that the differences in tuberosity union rates are more likely attributable to the local mechanical conditions imposed by proximal stem lateralization than to glenoid-side factors, though the retrospective nature of this study and the inability to measure tuberosity tension in each patient preclude definitive causal attribution.

The only comparative study analyzing healing between medialized and lateralized implants in fracture settings is that by Verdano et al,³⁹ who reported 87.5% tuberosity consolidation with medialized designs vs. 50% with lateralized ones. These findings are consistent with our results, in which the most lateralized stem demonstrated the highest nonunion rate at 24.5%, with a significant difference between the most and least lateralized groups (P = .016, OR = 0.22). Together, these data support the hypothesis that increasing lateralization of the center of rotation progressively increases tension on soft tissues and tuberosities, hindering their stability and healing. In contrast, Holt et al,¹⁹ in a systematic review, found higher consolidation rates with lateralized implants (88%) than with medialized implants (72%), and the authors acknowledged that this difference likely reflects concurrent advances in surgical technique, tuberosity repair methods, and biomaterials rather than implant design alone. More recently, O'Sullivan et al³⁰ demonstrated that increased humeral lateralization is significantly associated with stress fractures of the acromion and scapular spine, further supporting the notion that the mechanical consequences of excessive lateralization extend well beyond the tuberosity interface.

Limitations

One limitation of this study is its retrospective design, which inherently introduces potential selection bias and limits the ability to establish causality. Although we attempted to minimize this effect by matching patients on baseline characteristics for group comparisons, unmeasured confounders may still exist. The retrospective nature of the analysis also limits control over perioperative variables, which may influence tuberosity healing. In addition, the inclusion of different centers, surgeons, and implant systems introduces potential confounding factors that may affect external validity. The relatively small sample size within individual subgroups limits the precision of consolidation estimates and the statistical power of subgroup comparisons, and results should be interpreted with appropriate caution. Furthermore, the absence of an advanced imaging protocol across centers for assessing tuberosity consolidation introduces potential variability in outcome classification. Interobserver and intraobserver reliability of both the radiographic consolidation assessment and LSA measurement were not formally evaluated. An important limitation also relates to the inherent design differences among implants within the lateralized group: the Humelock II employs a press-fit system with locking screws to stabilize the metadiaphysis, whereas the Unique and Arrow prostheses were cemented, a fixation method historically associated with diminished bone osteointegration that may independently affect healing outcomes. Furthermore, the Humelock II is a fracture-specific stem with a low-profile metaphyseal geometry specifically designed to accommodate tuberosity reattachment compared to the Arrow and Unique prostheses, which have more prominent metaphyseal profiles that may create a step or gap between the stem and the repaired tuberosity. Finally, prospective randomized or registry-based studies comparing defined lateralization profiles with standardized repair techniques and follow-up protocols are necessary to more definitively isolate the independent effect of stem lateralization on tuberosity healing.

Conclusion

This study supports rTSA as an effective treatment for complex PHFs, demonstrating a high overall tuberosity consolidation rate and confirming that tuberosity healing is associated with superior functional outcomes. A significant dose–response relationship was observed between increasing stem lateralization and the nonunion rate. These findings suggest that excessive proximal stem lateralization may create a biomechanically unfavorable environment for osteointegration at the tuberosity, particularly in elderly patients with compromised bone quality. Post-operative LSA, while useful as a global geometric descriptor, does not capture the local mechanical environment at the tuberosity-implant interface. These findings highlight proximal stem lateralization as a potentially important and modifiable determinant of outcomes after fracture rTSA. Prospective randomized studies comparing defined lateralization profiles with standardized repair techniques and post-operative protocols are warranted to establish causality and define the threshold of lateralization compatible with reliable tuberosity healing.