Fracture and Vertebral Augmentation Utilization Following Androgen Deprivation Therapy in Men With Prostate Cancer: A Propensity Score–matched Cohort Study Authors: Taha Taka, Carson Cummings, Sarah Meng, Andrew Miner, Graydon Dietrich, David Shin, Trupti Trivedi, Rory Goodwin, William Richardson, Olumide Danisa Published: 2026 ABSTRACT AbstractStudy DesignRetrospective cohort study.ObjectiveTo investigate the association between androgen deprivation therapy (ADT), vertebral compression fracture (VCF) incidence, and subsequent vertebral augmentation (vertebroplasty/kyphoplasty) utilization in adult men with prostate cancer.MethodsAdult men with prostate cancer were stratified by ADT exposure and matched 1:1 to non-ADT controls on demographics, comorbidities, and bone health. Primary outcomes were VCF incidence at 3 months, 1 year, 3 years, and 5 years and vertebral augmentation utilization. Secondary outcomes included posterior spinal instrumentation, dependency, fracture burden, and bisphosphonate/denosumab use. Risks were compared using Kaplan-Meier and Cox proportional hazards models, with chi-square tests for categorical outcomes.ResultsAfter matching, 122,511 patients were included per cohort. ADT was associated with higher 5-year VCF incidence (1.95% vs 1.29%; HR, 1.49; p < 0.001), representing a 54% relative increase (NNH = 143), with consistent findings at earlier time points (all p < 0.05). Among 2,926 matched pairs with VCFs, vertebral augmentation utilization was higher with ADT (7.4% vs 5.1%; p < 0.001), representing a 45% relative increase. Posterior instrumentation (2.1% vs 2.0%) and dependency (7.5% vs 7.7%) were similar, while fracture burden did not differ (p = 0.07).ConclusionsADT is associated with increased VCF incidence and vertebral augmentation utilization without differences in posterior instrumentation. This pattern is consistent with osteoporotic insufficiency fractures and supports bone health screening and preventive therapy at ADT initiation. Introduction Androgen deprivation therapy (ADT) is a mainstay of treatment for prostate cancer, the second leading cause of cancer death among men in the United States, with approximately 1.5 million new cases diagnosed each year 1 While ADT is vital for managing adenocarcinoma, the most common subtype of prostate cancer, its use is associated with accelerated bone mineral density loss, with reported annual declines of approximately 2–8% in the lumbar spine, exceeding age-related bone loss 2 This bone loss is a result of androgen deficiency and reduced estrogen conversion, which leads to increased bone resorption.3,4 Consequently, men receiving ADT experience a substantially elevated fracture risk, with prior studies demonstrating a 2- to 5-fold increase compared to non-ADT populations.1,5 Despite these well-established risks, the downstream clinical and procedural burden of ADT-induced spinal fragility remains poorly quantified at the population level. Vertebral compression fractures (VCFs) are the most common osteoporotic fracture and are associated with substantial morbidity, including debilitating pain, spinal deformity, and significant functional decline.6-9 When conservative management fails, vertebral augmentation procedures, including vertebroplasty and kyphoplasty, can be performed to relieve pain and improve function by stabilizing the fracture and restoring vertebral height9,10 While the association between ADT and increased fracture risk is established, whether ADT exposure is associated with increased vertebral augmentation utilization remains unknown. The procedural footprint of ADT-related skeletal fragility has not been quantified. Higher procedural rates may reflect more severe fractures or failure of conservative treatment. In this study, we utilized a large, multi-institutional database to evaluate the association between ADT use and both VCF incidence and downstream vertebral augmentation utilization in men with prostate cancer. Using propensity score matching and longitudinal follow-up, we evaluated both fracture risk and utilization of subsequent vertebral augmentation. We hypothesized that ADT exposure would be associated with both increased VCF incidence and greater utilization of vertebral augmentation, indicating a measurable and potentially preventable procedural burden attributable to treatment-induced bone fragility. Methods Study Design and Data Source This retrospective propensity score-matched cohort study used the TriNetX Global Collaborative Network (TriNetX LLC, Cambridge, MA), a federated health research platform aggregating de-identified electronic health record data from over 170 healthcare organizations. The platform includes diagnoses, procedures, medications, and demographic information derived from participating institutions. Data were accessed on April 12–13, 2026. The time period analyzed in this study was 2008 to 2025. All analyses were performed using the TriNetX platform’s built-in analytics tools, which utilize R statistical software (R Foundation for Statistical Computing). The use of de-identified data qualified this study for institutional review board exemption, and the study was conducted in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines. 11 Study Population Adult male patients (≥18 years) with a diagnosis of prostate cancer (ICD-10-CM: C61) were identified. Patients were stratified into two cohorts based on androgen deprivation therapy (ADT) exposure. The ADT cohort included patients with documented ADT exposure, while the non-ADT cohort included patients without any recorded ADT exposure during the study period. Patients with long-term systemic corticosteroid use were excluded from both cohorts due to the known effects of steroids on bone metabolism. 3 For the primary analysis of vertebral compression fracture (VCF) incidence, patients with a prior history of VCF before the index event were excluded. For the secondary analysis evaluating post-fracture outcomes, patients with a history of vertebral augmentation before the index VCF were excluded. Cohort selection and matching are summarized in Figure 1. International Classification of Diseases (ICD - 10) and Current Procedural Terminology (CPT) codes are utilized for defining post-fracture surgical and medical outcomes Exposure Definition ADT exposure was defined by the presence of medications consistent with androgen suppression, including gonadotropin-releasing hormone (GnRH) agonists (goserelin, triptorelin, histrelin), GnRH antagonists (degarelix, relugolix), and antiandrogen therapies (bicalutamide, flutamide, nilutamide, apalutamide). Abiraterone was included given its use in combination with androgen deprivation therapy in standard clinical practice.1,12 Index Event and Follow-Up Index events were defined according to cohort-specific criteria derived from the TriNetX cohort definitions. For the ADT cohort, the index date was defined as the earliest recorded ADT exposure following prostate cancer diagnosis. For the non-ADT cohort, the index date corresponded to the earliest qualifying diagnosis of prostate cancer. Outcomes were assessed within a 5-year time window beginning 1 day after the index event and extending to 1825 days, consistent with the TriNetX analytic framework. Patients were censored at the time of the last recorded clinical activity. Outcome Definitions The primary outcomes were vertebral compression fracture (VCF) incidence rates at different time points and vertebral augmentation utilization within 3 and 6 months time-points among the patients who developed VCFs. Notably, vertebral augmentation utilization was specifically assessed at these time intervals to ensure the procedure was directly linked to the initial compression fracture. As more time passes between the diagnosis and the intervention, the likelihood of a direct clinical association decreases. Within these VCF cohorts, additional outcomes were posterior spinal instrumentation and dependency. Surgical and medical outcomes and complications were also assessed at the first-, third-, and fifth-year time points after VCF diagnosis. Vertebral augmentation procedures were identified using standardized procedural coding in the TriNetX platform, and dependency was defined using ICD-10-CM codes for care dependence. A complete list of diagnostic codes is provided in the Supplementary Table. Propensity Score Matching 1:1 propensity score matching was performed between ADT and non-ADT cohorts using nearest-neighbor matching without replacement and a caliper width of 0.1 standard deviations. Covariates included demographic characteristics (age, race), comorbidities (diabetes mellitus, chronic kidney disease, cardiovascular disease, and chronic pulmonary disease), and baseline bone health indicators (osteoporosis, body mass index, tobacco use, and metastatic disease) (Tables 1 and 2). Covariate balance after matching was assessed using standardized mean differences, with values less than 0.1 indicating adequate balance. 13 Table 1.Propensity-matched comparison of demographic characteristics of prostate cancer patients with vertebral compression fractures with and without androgen deprivation therapyCharacteristicBefore MatchingAfter MatchingADT patients (N, %) (N =131,883)Control (N = 1,090,739)P ValueADT patients (N, %) (N =122,511)Control (N = 122,511)P Value Age at Index 72.6 ± 9.269.1 ± 10.1<0.0172.5 ± 9.272.5 ± 9.20.9White68,433 (52.1)580,613 (56.4)<0.0163,410 (51.8)63,742 (52.0)0.2Black or African American17,479 (13.3)118,723 (11.5)<0.0116,094(13.1)15,668 (12.8)0.3Asian7,762 (5.9)31,528 (3.1)<0.017,032 (5.5)6,955 (5.4)0.5Unknown Race26,234 (20.0)241,919 (23.5)<0.0125,550 (19.9)25,554 (20.0)1Native Hawaiian/Other PI489 (0.4)2,545 (0.2)<0.01412 (0.3)355 (0.3)1American Indian/Alaska Native306 (0.2)2,337 (0.2)1271 (0.2)241 (0.2)1 Propensity-matched comparison of demographic characteristics of prostate cancer patients with vertebral compression fractures with and without androgen deprivation therapy Table 2.Propensity-matched comparison of clinical characteristics of prostate cancer patients with vertebral compression fractures with and without androgen deprivation therapy. (BMI = Body Mass Index)DescriptionBefore MatchingAfter MatchingADT patients (N, %) (N =131,883)Control (N = 1,090,739)P ValueADT patients (N, %) (N =122,511)Control (N = 122,511)P ValueSecondary Neoplasm (C79)18,581 (13.5)9,502 (0.9)<0.019,424 (7.4)9,358 (7.3)0.6Ischemic Heart Disease (I20-I25)24,327 (17.7)118,904 (11.3)<0.0121,614 (16.9)22,143 (17.2)0.4Chronic Lower Resp. Dis. (J40-J44)15,624 (11.4)78,875 (7.5)<0.0113,857 (10.9)14,520 (11.3)0.3Diabetes Mellitus (E08 - E13)22,631 (16.5)113,202 (10.7)<0.0120,249 (15.8)20,732 (16.1%)0.4Chronic Kidney Dis. (N18)14,410 (10.5)58,787 (5.6)<0.0112,489 (9.7)12,865 (10.0)0.4BMI (Z68)10,959 (8.0)64,088 (6.1)<0.019,274 (7.2)9,490 (7.4)0.2Osteoporosis (M81)3,485 (2.5)9,381 (0.9)<0.012,846 (2.2)2,711 (2.1)0.8Tobacco Use (Z72.0)2,773 (2.0)14,929 (1.4)<0.012,399 (1.9)2,429 (1.9)0.7 Propensity-matched comparison of clinical characteristics of prostate cancer patients with vertebral compression fractures with and without androgen deprivation therapy. (BMI = Body Mass Index) Statistical Analysis Baseline characteristics were summarized using descriptive statistics. Categorical variables were compared using chi-square tests, and continuous variables were compared using t-tests. Kaplan–Meier survival analysis was performed to estimate VCF-free survival, with differences between cohorts assessed using the log-rank test. Hazard ratios were calculated using Cox proportional hazards models. Hazard ratios were used in preference to odds ratios because VCF incidence was analyzed as a time-to-event outcome with censoring at the last recorded clinical activity. Patient counts for all outcomes are reported alongside percentages in Table 3. For secondary outcomes, risk differences and relative risks were used to compare rates of vertebral augmentation, posterior instrumentation, and dependency between cohorts. Fracture burden was evaluated by comparing the mean number of VCF events per patient. All statistical tests were two-sided, with a significance threshold of p < 0.05. Table 3.Matched cohort size and outcome proportions by androgen deprivation therapy (ADT) exposureOutcomePatients per matched cohort, nADT, %Non-ADT, %Absolute difference, percentage pointsp value Primary analysis: VCF incidence VCF at 1 year122,5110.90.60.3< 0.01VCF at 3 years122,5111.61.00.6< 0.01VCF at 5 years122,5111.951.290.66< 0.01 Post-VCF matched analysis Vertebral augmentation2,9267.45.12.3< 0.001Posterior spinal instrumentation2,9262.12.00.10.926Dependency2,9267.57.7−0.20.73n is the number of patients in each propensity score–matched cohort (the denominator for each analysis), not the number of patients experiencing the outcome. Percentages are the proportion of each cohort with the outcome, as reported in the Results. Absolute differences are ADT minus non-ADT. ADT, androgen deprivation therapy; VCF, vertebral compression fracture. Matched cohort size and outcome proportions by androgen deprivation therapy (ADT) exposure n is the number of patients in each propensity score–matched cohort (the denominator for each analysis), not the number of patients experiencing the outcome. Percentages are the proportion of each cohort with the outcome, as reported in the Results. Absolute differences are ADT minus non-ADT. ADT, androgen deprivation therapy; VCF, vertebral compression fracture. Results Cohort Formation and Characteristics A total of 1,222,622 adult male patients with prostate cancer were identified within the TriNetX Global Collaborative Network. After application of inclusion and exclusion criteria, patients were stratified by androgen deprivation therapy (ADT) exposure. Following 1:1 propensity score matching, 122,511 patients were included in each cohort (Figure 1). Baseline demographic and clinical characteristics were well balanced between cohorts after propensity score matching, with all standardized mean differences less than 0.1 (Tables 1 and 2). Vertebral Compression Fracture Incidence and Fracture Burden ADT was associated with a higher 5-year incidence of vertebral compression fractures (VCFs) (1.95% vs 1.29%, p < 0.01) and increased utilization of vertebral augmentation among patients who developed fractures (7.4% vs 5.1%, p < 0.001). Similar findings were noted at the first (0.9% vs 0.6%, p < 0.01) and third-year (1.6% vs 1.0%, p < 0.01) time points (Tables 3 and 4). The ADT cohort had a higher hazard of VCF compared to the non-ADT cohort (HR 1.49, p < 0.001). Among patients who developed VCFs, there was no statistically significant difference in fracture burden between cohorts. The mean number of VCF events per patient was 3.48 in the ADT cohort compared to 3.17 in the non-ADT cohort (p = 0.07). Table 4.Comparative analysis of vertebral compression fracture incidence and subsequent vertebral augmentation rate in prostate cancer patients following androgen deprivation therapy. (VCF = Vertebral compression fracture; ADT = Androgen deprivation therapy)Clinical OutcomeProstate cancer without ADTProstate cancer with ADTRelative Difference/Ratiop-valueVCF Incidence: 3 Months0.34%0.47%0.73< 0.01VCF Incidence: 1 Year0.60%0.93%0.64< 0.01VCF Incidence: 3 Years1.03%1.63%0.63< 0.01VCF Incidence: 5 Years1.29%1.95%0.66< 0.01Vertebral Augmentation:3 months post-VCF4.24%5.64%0.750.0134Vertebral Augmentation:6 months post-VCF4.44%6.53%0.68<0.01 Comparative analysis of vertebral compression fracture incidence and subsequent vertebral augmentation rate in prostate cancer patients following androgen deprivation therapy. (VCF = Vertebral compression fracture; ADT = Androgen deprivation therapy) Vertebral Compression Fracture Incidence and Fracture Burden with Bisphosphonate or Denosumab Use Analysis of antiresorptive therapy within the propensity-matched cohort showed no statistically significant difference in the incidence of VCF between patients receiving IV bisphosphonates or denosumab and those not receiving them. At the 5-year follow-up, the primary outcome rate was 2.09% in the IV bisphosphonate group compared to 2.13% in the non-exposure group (p = 0.81). This lack of statistical significance was consistent across earlier time points, with 1-year rates of 1.12% vs 1.0% (p = 0.374) and 3-year rates of 1.79% vs 1.87% (p = 0.637). Post-VCF Matched Cohort for Vertebral Augmentation Utilization and Subsequent Outcomes Among patients who developed VCFs, a secondary propensity score-matched analysis was performed, resulting in 2,926 matched pairs between the ADT and non-ADT cohorts. Within the matched VCF cohort, the ADT group had a higher rate of vertebral augmentation than the non-ADT group (7.4% vs 5.1%, p < 0.001), representing a 45% relative increase in vertebral augmentation utilization (Tables 3 and 4). There were no statistically significant differences between cohorts in rates of posterior spinal instrumentation (2.1% vs 2.0%, p = 0.926) or dependency outcomes (7.5% vs 7.7%, p = 0.73). Discussion In this large propensity-matched cohort, ADT exposure was associated with both increased vertebral compression fracture (VCF) incidence and higher vertebral augmentation utilization in men with prostate cancer. While it is established that ADT is associated with elevated fracture risk, this study extends prior work by demonstrating a corresponding increase in utilization of downstream vertebral augmentation. Rates of posterior spinal instrumentation and dependency were similar between cohorts, suggesting that ADT-related fractures are not more structurally severe but result in increased procedural intervention. Even in the absence of differences in functional outcomes, increased procedural utilization represents additional healthcare utilization and costs. To our knowledge, this is the first population-level study to quantify the downstream procedural burden of ADT-related skeletal fragility. The observed increase in VCF risk is consistent with prior literature demonstrating an elevated fracture risk in men receiving ADT.14-16 While previous studies have reported increased rates of osteoporosis and fragility fractures, this study extends those findings through a large propensity score-matched analysis with longitudinal follow-up, demonstrating that approximately 1 in 143 patients treated with ADT will develop an additional VCF associated with ADT exposure. The similar fracture burden per patient between cohorts suggests no significant difference in fracture frequency among affected individuals, indicating that the observed increase in VCF incidence is driven by more patients experiencing fractures rather than increased recurrence per patient. A key finding is the increased rate of vertebral augmentation procedures among ADT patients who develop VCFs. Vertebral augmentation procedures are widely utilized for symptomatic VCFs when conservative management fails, with randomized controlled trial evidence, including the CAFE trial, demonstrating significant pain reduction and functional improvement in cancer patients.17,18 In this study, ADT patients underwent vertebral augmentation at significantly higher rates despite having a similar number of fractures per patient. This suggests that fractures in the ADT population are not more structurally severe, but are more likely to require procedural intervention. Alternatively, increased oncologic surveillance in ADT patients may lead to higher rates of fracture detection and subsequent intervention. Importantly, the absence of increased posterior instrumentation indicates that these fractures do not represent more complex or unstable injury patterns requiring surgical stabilization. Our study did not differentiate between vertebroplasty and kyphoplasty, though prior literature suggests similar clinical outcomes with potential differences in complication profiles.19,20 The pattern of results supports characterizing ADT-related VCFs as osteoporotic insufficiency fractures rather than as structurally catastrophic events. Equivalent rates of posterior instrumentation and dependency outcomes indicate similar functional trajectories following fractures. This pattern is consistent with ADT-induced bone loss driven by increased resorption, resulting in fragility rather than metastatic fractures.3,21 These results show an important and potentially modifiable clinical issue. Current guidelines recommend baseline bone health assessment and antiresorptive therapy for men initiating ADT, yet adherence remains low.12,22-25 The increased utilization of vertebral augmentation observed in this study represents a measurable downstream consequence of treatment-induced bone fragility and may reflect missed opportunities for prevention. Notably, our sub-analysis of patients receiving intravenous (IV) bisphosphonates did not demonstrate a significant reduction in the cumulative incidence of VCF compared to patients not receiving IV bisphosphonates. At 5 years, the rates of primary outcome were functionally equivalent between the two groups (2.09% vs 2.13%, p = 0.81). Similar non-significant results were observed at earlier time points, including 3 years (1.80% vs 1.87%, p = 0.64) and 1 year (1.1% vs 1.0%, p = 0.37). This lack of significance may reflect poor adherence to IV bisphosphonates or denosumab, given efficacy demonstrated in prior literature. 26 Specifically, a prior prospective study has shown a significant reduction in incident vertebral fractures (7.1% vs 29.0%) when bone-active drugs are utilized compared to unmanaged controls. 26 Additionally, a prior systematic review and meta-analysis of randomized controlled trials determined that bisphosphonate therapy significantly reduced the risk of vertebral fractures (RR 0.80) and osteoporosis (RR 0.39). 27 The findings of our study could also be attributed to confounding by indication, in which patients with lower baseline bone mineral density or declining bone mineral density were more likely to be prescribed bisphosphonates. Ultimately, the use of anti-resorptive medications remains a cornerstone of bone health management. The increased utilization of vertebral augmentation observed in this study represents a measurable economic burden. In this cohort, the higher augmentation rate corresponds to approximately 67 additional procedures per 2,926 patients with VCFs. With each procedure costing approximately $9,000–$13,000, this amounts to several hundred thousand dollars in excess procedural costs. 28 More recent administrative data suggest substantially higher total hospitalization costs, indicating that this estimate may be conservative. 29 In contrast, while our results did not show significance in VCF rate with antiresorptive therapies such as denosumab and bisphosphonates, their use has been shown to reduce fracture risk significantly in other studies and is less costly over a comparable time horizon.30,31 Ultimately, this topic of the effectiveness of anti-resorptive therapy would benefit from both a cost-benefit analysis and an adherence analysis to provide further insights into the overall clinical utility. Furthermore, this area would benefit from further research regarding patient use and longitudinal outcomes to better characterize the real-world impact of these treatments. Our results have several clinical implications. ADT exposure should be recognized as a risk factor in men presenting with VCFs, and treatment history should be incorporated into clinical evaluation. These encounters represent an opportunity to address underlying bone health, including coordination with oncology or primary care providers to ensure appropriate screening and initiation of preventive therapies. Limitations Several limitations should be considered. The retrospective design introduces the potential for residual confounding despite propensity score matching and precludes causal inference. Reliance on administrative coding (ICD-10 and CPT) may result in misclassification of diagnoses and procedures, and the TriNetX platform lacks granular clinical detail, including imaging findings, fracture severity, symptom burden, bone mineral density, and specific indications for vertebral augmentation. Medication exposure was defined based on recorded prescriptions and does not confirm adherence, dosage, or duration of therapy. Although equivalent posterior instrumentation rates support a predominantly osteoporotic fracture pattern, the contribution of occult metastatic disease to VCF incidence cannot be fully excluded. Additionally, patients receiving ADT may undergo more frequent imaging through oncologic surveillance, which may increase fracture detection and intervention rates. Finally, the TriNetX network predominantly includes large academic and integrated healthcare systems, which may limit generalizability to smaller practice settings. Conclusion In men with prostate cancer, ADT was associated with a modest absolute increase in 5-year vertebral compression fracture incidence (1.95% vs 1.29%; absolute risk difference, 0.66 percentage points) and in vertebral augmentation utilization among those who fractured (7.4% vs 5.1%; absolute difference, 2.3 percentage points), without differences in posterior instrumentation or dependency outcomes. The results show a pattern consistent with osteoporotic insufficiency fractures. Although the absolute risk to any individual patient is small, the number of men receiving ADT is large, and these findings represent a measurable and potentially preventable procedural burden associated with treatment-related bone fragility. Systematic bone health screening and preventive therapy at ADT initiation may reduce the need for downstream fracture-related interventions and their associated costs. Ethical Considerations This study used de-identified data from the TriNetX Global Collaborative Network and was therefore exempt from institutional review board approval. The study was conducted in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines. Acknowledgements The authors have no acknowledgments to declare. Footnotes Author Contributions: Taha Taka: Conceptualization, Methodology, Formal analysis, Data curation, Investigation, Writing – original draft, Writing – review and editing, Project administration. Carson Cummings: Methodology, Data curation, Investigation, Writing – original draft, Writing – review and editing. Sarah Meng: Methodology, Data curation, Investigation, Writing – original draft, Writing – review and editing. Andrew Miner: Data curation, Investigation, Writing – review and editing. Graydon Dietrich: Project administration, Investigation, Writing – review and editing. David Shin: Writing – original draft, Writing – review and editing. Trupti Trivedi: Methodology, Validation, Writing – review and editing, Supervision. Rory Goodwin: Conceptualization, Validation, Writing – review and editing, Supervision. William Richardson: Conceptualization, Validation, Writing – review and editing, Supervision. Olumide Danisa: Conceptualization, Methodology, Validation, Writing – review and editing, Supervision,s Project administration. Funding: The authors received no financial support for the research, authorship, and/or publication of this article. The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. ORCID iDs Taha Taka https://orcid.org/0009-0002-2553-3017 Carson Cummings https://orcid.org/0009-0002-1539-0964 Sarah Meng https://orcid.org/0009-0000-2587-4706 Graydon Dietrich https://orcid.org/0009-0002-4820-7927 Rory Goodwin https://orcid.org/0000-0002-6540-2751 References