Pre-operative glenoid bone density varies by indication for reverse shoulder arthroplasty: a comparison of glenoid bone density in proximal humerus fractures, glenohumeral osteoarthritis, and rotator cuff arthropathy Authors: Alec M Reihl, Blake E Bohlig, Nicholas Wiley, Mark Cote, Evan A O'Donnell Published: 2026 ABSTRACT AbstractBackgroundGlenoid bone mineral density (BMD) has important implications for the fixation of glenoid components and may vary based on indications for reverse total shoulder arthroplasty (rTSA).17 Careful consideration of pre-operative bone quality can enable surgeons to improve pre-operative and intraoperative decision-making for glenoid component fixation and reduce associated complications, such as aseptic loosening. There remains a paucity of literature exploring such implications in rTSA. We hypothesize that BMD varies by indication diagnosis for rTSA.MethodsA retrospective chart review was conducted on patients receiving an rTSA from a single surgeon from March 2021-May 2024 with one of the 3 following diagnoses: proximal humerus fracture (FX), glenohumeral osteoarthritis (GHOA), and rotator cuff arthropathy (RCA). Pre-operative computerized tomography scans were analyzed for glenoid BMD. Differences in BMDs across different shoulder pathologies were assessed with general linear regression models adjusted for age and body mass index.ResultsIn sum, 150 patients were identified who underwent an rTSA for one of the 3 following diagnoses: FX (n = 24, 16%), GHOA (n = 64, 43%), and RCA (n = 62, 41%). Fifty-one (34%) of patients were male and 99 (66%) of patients were female. Average subchondral BMD was 150 Hounsfield Units (HUs) (95% confidence interval of 92-207 HUs) for FX, 304 HUs (268-339 HUs) for GHOA, and 208 HUs (172-244 HUs) for RCA. Subchondral bone in patients with GHOA was significantly denser than FX and RCA (P < .01). Average anterior cortex bone density was 1,206 HUs (113-1,299 HUs), 1,262 HUs (1,205-1,319 HUs), and 1,198 HUs (1,140-1,256 HUs) for FX, GHOA, and RCA patients, respectively. No statistically significant differences in anterior cortical HU were observed across the 3 pathologies in this study. Average posterior cortex bone density was 876 HUs (763-990 HUs), 969 HUs (889-1,038 HUs), and 959 (888-1,030 HUs) for FX, GHOA, and RCA patients, respectively. No statistically significant differences in posterior cortical HUs were observed across the 3 pathologies in this study.ConclusionThis study demonstrates that subchondral glenoid BMD varies among patients undergoing rTSA depending on the underlying pathology. Fracture rTSAs have the lowest subchondral BMD when compared to GHOA and RCA patients. BMD of the anterior cortex appears equivalent despite indication for rTSA. Incorporating glenoid BMD assessments into surgical planning may enhance implant fixation and possibly avoid complications such as implant loosening. Further prospective studies are warranted to validate and refine clinical decision-making based on glenoid bone quality.Keywords: Reverse total shoulder arthroplasty, Bone mineral density, Revision reverse total shoulder arthroplasty, Glenoid loosening, Reverse total shoulder arthroplasty complications, Pre-operative evaluation Utilization of reverse total shoulder arthroplasty (rTSA) has been increasing rapidly in prevalence in the United States.2 The incidence of shoulder arthroplasty is predicted to increase from 3-fold to 7-fold by 2030.19 This is in part due to the aging population and expanding indications for rTSA. Due to advancements in implant design, techniques, and experience, the indications for rTSA have expanded from its original indication of rotator cuff arthropathy (RCA) to include primary glenohumeral osteoarthritis (GHOA), massive rotator cuff tears without arthritis, arthroplasty in the setting of poor glenoid bone stock, proximal humerus fractures, inflammatory arthritis, and revision arthroplasty.24 Aseptic glenoid loosening is a rare but catastrophic complication of rTSA requiring revision surgery. In a study of 202 patients, Bizter et al published a rate of aseptic loosening of 1.2% after rTSA.3 Schell et al reported a rate of aseptic loosening at 0.74% in their study of more than 7,000 patients.23 In a recent systematic review, Su et al published that glenoid radiolucencies were reported at rates of 2.4% at a minimum 2-year follow-up.22 Revision rTSA is a morbid procedure with a high complication rate and decreased function compared to primary rTSA.4,5,18 Currently, there is limited published literature discussing the pre-operative glenoid bone mineral density (BMD) in patients prior to an rTSA surgery. As the indications expand and the utilization of rTSA increases, it is vital to understand the differences in patient populations, as this may help to decrease future complications. Significant differences of BMD across shoulder pathologies may translate into different clinical decision-making and management of rTSA patients. Due to the difference in pathophysiology, the authors hypothesize that there is a difference in the BMD when comparing patients undergoing rTSA for proximal humerus fractures (FX), RCA, and GHOA. Understanding this difference in glenoid BMD may allow for indication-specific implants or surgical techniques and thus may impact clinical practice. Methods Patients receiving an rTSA by a single surgeon (E.A.O.) at a single institution from March 2021-May 2024 were retrospectively queried through a research patient data registry (a centralized clinical data registry which gathers data from various hospital systems) search and chart review. Patients were included with a diagnosis of proximal humerus fractures, GHOA, and RCA. Patients were excluded for revision surgery, and any other diagnosis, including avascular necrosis, pyogenic or septic arthritis, and malignancy. rTSAs were indicated in patients with GHOA and concomitant rotator cuff tears, severe glenoid retroversion (>30°), posterior subluxation >80%, or other bony compromise on tomographic imaging. rTSA was indicated for patients after proximal humerus fracture in patients with concomitant shoulder pathology or in patients with fractures not amenable to open reduction internal fixation or nonoperative management. This study was assessed by the institutional review board (No. 2024P002405) and found exempt. A computerized tomography (CT) scan was completed on all patients prior to surgery and was used to assess bone density of the anterior cortex, posterior cortex, and the subchondral bone. Density was measured in Hounsfield Units (HUs). Alhough there was variability in the model of CT scanner used, the CT protocol for all scans completed was the same and included submillimeter cuts and imaging that included the entire scapula. Using a standard Picture Archiving and Communication System, on the axial cut just below the tip of the coracoid, a circle was drawn medial to the subchondral cortical bone inside the anterior and posterior cortex, ensuring that no cortical bone was included. The average HU in the area of the circle was calculated and recorded (Fig. 1).22 To measure the anterior cortex, a line was drawn through the center of the glenoid subchondral cortical bone matching the version of the glenoid. A second line was then placed 10° anterior to that line. A circle was created, including only cortical bone at the point in which the line intersected with the anterior cortex. The average HU in the area of the circle was calculated and recorded (Fig. 2). To measure the posterior cortex, the same line through the center of the glenoid subchondral cortical bone, matching the version of the glenoid, was used. A third line was then placed 10° posterior to that line. A circle was created including only cortical bone at the point in which the line intersected with the anterior cortex. The average HU in the area of the circle was calculated and recorded (Fig. 3). HU measurement of subchondral glenoid. HU, Hounsfield Unit. HU measurement of anterior cortex. HU, Hounsfield Unit. HU measurement of posterior cortex. HU, Hounsfield Unit. Statistical analysis Descriptive statistics including mean and standard deviation for continuous variables and frequency and proportion for categorical variables were calculated to characterize the study groups. General linear regression was used to model differences in density between the diagnostic groups. Age and body mass index (BMI), both of which were scaled to 10-unit increments, were included in the model to obtain adjusted estimates of bone density. The results of inferential analyses are reported with corresponding 95% confidence intervals. All analyses were carried out in R (R Core Team, Vienna, Austria http://www.r-project.org). Results A total of 164 patients were identified as eligible for inclusion, having undergone primary rTSA between March 2021 and May 2024. After review, 150 patients were found to have a pre-operative diagnosis of proximal humerus fractures from traumatic injury, GHOA from chronic degenerative usage, and RCA and were included in the analysis. Patients within the fracture cohort had varied presentations of pre-existing osteoarthritis. The cohort comprised 51 males (34%) and 99 females (66%), with a mean age of 71.4 ± 9.0 years. The surgical indications were distributed as follows: proximal humeral fracture in 24 patients (16%), GHOA in 64 patients (43%), and RCA in 62 patients (41%) (Table I). Mean and standard deviation of the bone density by group are presented in Figure 4. Table IPatient demographicsDemographicsMean ± SDMedian (range)Age (yr)71 ± 9.072 (35 to 93)BMI (kg/m2)30 ± 6.429 (18.6 to 53)Sex Male34% (n = 51) Female66% (n = 99)Diagnosis Proximal humerus fracture16% (n = 24) Glenohumeral osteoarthritis43% (n = 64) Rotator cuff arthropathy41% (n = 62)SD, standard deviation; BMI, body mass index. Patient demographics SD, standard deviation; BMI, body mass index. Mean bone density by diagnosis. Subchondral bone mineral density The multivariate regression model for subchondral bone density is depicted in Table II. Mean subchondral glenoid BMD varied significantly across diagnostic groups. Age and BMI-adjusted mean BMD was greatest in the GHOA (302 HU) followed by the RCA group (208 HU), and the FX group (153 HU). Compared to the GHOA group, subchondral BMD was −150 HU less (P = .00002) in the FX group and −95 HU less in the RCA group (P = .0003, Table I). R-squared value of the model was 0.16. Table IISubchondral bone densityMultivariate regression of subchondral bone density adjusted for BMI and ageEstimate95% CIP valueGHOA150(82-217)<.001RCA55(−13 to 123).111BMI (10-unit increase)28(−11 to 66).160Age (10-yr increase)−7(−36 to 22).624Subchondral bone density (HU) adjusted meansPairwise comparisonsMean95% CIDifference95% CIP valueFX153(95-210)FX-GHOA150(−217 to −82)<.001GHOA302(267-338)FX-RCA−55(−123 to 13).111RCA208(172-244)GHOA-RCA95(44 to 145)<.001BMI, body mass index; CI, confidence interval; GHOA, glenohumeral osteoarthritis; RCA, rotator cuff arthropathy; HU, Hounsfield Unit; FX, proximal humerus fracture. Subchondral bone density BMI, body mass index; CI, confidence interval; GHOA, glenohumeral osteoarthritis; RCA, rotator cuff arthropathy; HU, Hounsfield Unit; FX, proximal humerus fracture. Anterior cortical bone mineral density The multivariate regression model for subchondral bone density is depicted in Table III. Increasing BMI was associated with decreased anterior cortex bone density with BMI (75 HU average decrease per 10-unit increase in BMI, P = .018). BMD varied considerably less than subchondral density, with minimal differences between the diagnostic groups (Table III). Across the diagnostic groups, GHOA had the greatest adjusted mean BMD at 1,265 HU followed by the FX and RCA groups, both of which were 1,198 HU. R-squared value of the model was 0.06. Table IIIAnterior cortex bone densityMultivariate regression of anterior cortex bone density adjusted for BMI and ageEstimate95% CIP valueGHOA67(−41- to 175).222RCA−0.2(−109 to 108).997BMI (10-unit increase)−75(−136 to −13).018Age (10-yr increase)9(−37 to 55).702Anterior cortex bone density (HU) adjusted meansPairwise comparisonsMean95% CIDifference95% CIP valueFX1,198(1,106-1,290)FX-GHOA−67(−175 to 41).222GHOA1,265(1,209-1,321)FX-RCA−0.2(−108 to 109).997RCA1,198(1,141-1,255)GHOA-RCA67(−13 to 148).099BMI, body mass index; CI, confidence interval; GHOA, glenohumeral osteoarthritis; RCA, rotator cuff arthropathy; HU, Hounsfield Unit; FX, proximal humerus fracture. Anterior cortex bone density BMI, body mass index; CI, confidence interval; GHOA, glenohumeral osteoarthritis; RCA, rotator cuff arthropathy; HU, Hounsfield Unit; FX, proximal humerus fracture. Posterior cortical bone mineral density The multivariate regression model for posterior cortex bone density is depicted in Table IV. Similar to the anterior cortex, posterior cortical BMD varied little across the 3 groups. The adjusted mean posterior cortical HU was again greatest in the GHOA group at 971 HU, followed by the RCA group with an adjusted mean of 959, and the FX group with an adjusted mean of 872. Differences between the groups did not reach statistical significance (Table IV). R-squared value of the model was 0.02. Table IVPosterior cortex bone densityMultivariate regression of posterior cortex bone density adjusted for BMI and ageEstimate95% CIP valueGHOA98(−35 to 232).148RCA87(−47 to 221).203BMI (10-unit increase)−39(−116 to 37).311Age (10-yr increase)6(−51 to 62).848Posterior cortex bone density (HU) adjusted meansPairwise comparisonsMean95% CIDifference95% CIP valueFX872(758-986)FX-GHOA−98(−232 to 35).148GHOA971(901-1,040)FX-RCA−87(−221 to 47).203RCA959(888-1,030)GHOA-RCA12(−88 to 111).819BMI, body mass index; CI, confidence interval; GHOA, glenohumeral osteoarthritis; RCA, rotator cuff arthropathy; HU, Hounsfield Unit; FX, proximal humerus fracture. Posterior cortex bone density BMI, body mass index; CI, confidence interval; GHOA, glenohumeral osteoarthritis; RCA, rotator cuff arthropathy; HU, Hounsfield Unit; FX, proximal humerus fracture. Discussion Our findings demonstrate significantly higher bone density in glenoid subchondral bone in the GHOA group when compared to the FX group and the RCA group. After adjusting for age and BMI, patients with GHOA possess the highest glenoid bone density (302 HU), followed by those with RCA (208 HU), while patients with proximal humerus fractures possess the lowest glenoid bone density (153 HU). These findings indicate that disease-specific pathophysiology leads to differing rates of glenoid remodeling and variations in BMD. The indication for rTSA can provide important information about the bone quality and density, helping guide surgical technique and minimize the risk of complications such as fracture or baseplate loosening in rTSA. Furthermore, we found that the anterior glenoid cortex measurements were significantly more dense than the posterior cortex across all pathologies.14,15 These findings have important implications for surgical technique and may help optimize implant positioning, especially positioning of baseplate placement. Large registry studies often report revision of rTSAs at 1% per year.20 One of the more common modes of failure in rTSAs is aseptic baseplate loosening. This complication, while less frequent with modern implant designs, remains a significant cause of failure, revision surgery, and patient morbidity. Baseplate loosening is multifactorial, with key contributors including poor bone quality, suboptimal fixation (eg, use of short or noncortically engaging central screws or pegs), baseplate seating, and baseplate malorientation of version or inclination.3,10,16 Observed variations in glenoid bone density between GHOA, RCA, and proximal humerus fractures may have important implications for presurgical planning and patient outcomes as BMD has been shown to affect screw fixation strength in the setting of rTSA.4,15,16,21 Patient-specific risk factors such as rheumatoid arthritis, diabetes, and glenoid morphology (notably Walch B2 and B3 glenoid types) further increase the risk of failure.20 Revision cases and those requiring bone grafting for severe glenoid defects are particularly prone to failure.1,3,6 Additionally, the use of allograft for bone grafting may present particular technical challenges contributing to increased micromotion and failure. Biomechanical studies confirm that insufficient initial fixation due to short central pegs or poor bone density may lead to micromotion and a higher risk of baseplate loosening.16,23 Clinically, baseplate loosening results in pain and loss of function, and often necessitates complex revision surgery, which is associated with higher complication rates and less predictable outcomes.7,11 Our study found anterior glenoid cortex yields the most dense bone across all 3 primary indications for rTSA. While traditional approaches prioritize anatomic restoration of glenoid version, our findings may suggest that strategic engagement of the denser anterior cortical bone may provide a benefit of fixation in high-risk patients where baseplate loosening risk may be higher. Although more research is required, this is a possible clinical option in patients with concerns for lower BMD and departure from current practice patterns. This study has several limitations, many of which are inherent to its retrospective design. Data collection was restricted to existing imaging and medical records, introducing potential selection bias and limiting control over confounding variables. Although 2-dimensional CT-based measurements provide a reliable estimate of bone density, they do not directly quantify BMD as obtained through quantitative CT or dual-energy x-ray absorptiometry.13 Furthermore, HU values may vary across scanners and routine CT is not calibrated against a reference phantom, potentially affecting measurement accuracy. Degenerative changes and other imaging artifacts may also influence HU readings. Additionally, bone density is affected by comorbidities such as osteoporosis, diabetes, rheumatoid arthritis, smoking, renal disease, and variations in glenoid morphology, which were not accounted for in this study. Finally, as this analysis was conducted at a single institution, the findings may not be generalizable to broader populations or differing imaging protocols. Our findings align with established arthroplasty principles across other specialties. In hip and knee arthroplasty, lower pre-operative BMD consistently correlates with increased aseptic loosening, periprosthetic fractures, and revision surgery rates.8,9,12,17 Shoulder arthroplasty follows a similar pattern, with osteoporosis increasing periprosthetic fracture odds by 2.24 times and prosthetic joint infection risk by 1.68 times.19 This study demonstrates that anterior cortical bone density is higher than posterior density across primary indications for rTSA, offering surgeons practical guidance to optimize fixation in patients with poor bone quality and potentially reduce complications. Conclusion Subchondral glenoid BMD varies among patients undergoing rTSA depending on the underlying pathology. Patients with GHOA have significantly higher bone density of subchondral bone when compared to patients with a pre-operative diagnosis of RCA and proximal humerus fracture. There is consistently superior bone density of the anterior glenoid cortex across all 3 studied pre-operative diagnoses: GHOA, RCA, and proximal humerus fracture. This study may assist with pre-operative planning, although further studies are needed. Disclaimers: Funding: No funding was disclosed by the authors. Conflicts of interest: The 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. Footnotes This study was assessed by the institutional review board (No. 2024P002405) and found exempt. References