Clin Orthop Relat Res. 2026 Jul 07.
Xi-Qian Heng,
Wei-Hsin Lin,
Chun-Chieh Yeh,
Liang-Yen Liu,
Cheng-Han Chung,
Hung-Kuan Yen,
Fang-Lan Chang,
Ho-Min Chen,
Ming-Hsiao Hu,
Shau-Huai Fu,
Chen-Yu Wang,
Olivier Q Groot.
BACKGROUND: Denosumab is widely used for preventing skeletal-related events (SREs) in patients with bone metastases. As advances in systemic cancer therapies have improved survival, the long-term safety and efficacy of denosumab warrant reevaluation. Balancing efficacy in preventing SREs with the cumulative risks associated with prolonged denosumab therapy, particularly osteonecrosis of the jaw (ONJ), a rebound phenomenon of accelerated bone turnover after abrupt discontinuation, and subsequent fragility fractures, has become increasingly important.
QUESTIONS/PURPOSES: (1) Is a lower cumulative number of denosumab doses associated with the risk of surgically treated SREs, ONJ, osteoporotic fractures, or all-cause mortality after dose reduction? (2) Is an extended maximum injection interval associated with the risk of surgically treated SREs, ONJ, osteoporotic fractures, or all-cause mortality after dose reduction? (3) Is a higher frequency of treatment gaps > 60 days associated with the risk of surgically treated SREs, ONJ, osteoporotic fractures, or all-cause mortality after dose reduction?
METHODS: This nationwide retrospective comparative study utilized data from Taiwan's National Health Insurance Research Database, a population-level resource accounting for 99.9% of Taiwan's residents. By capturing comprehensive medical claims with continuous patient enrollment, this database functionally eliminates loss to follow-up, enabling the construction of highly reliable longitudinal patient histories to evaluate long-term treatment patterns and clinical outcomes. We identified 35,325 patients with breast, prostate, or lung cancer with radiologically confirmed skeletal metastases who initiated denosumab therapy between January 1, 2012, and December 31, 2020. We excluded 75.4% (26,635) of patients, primarily because they received < 3 initial injections (36%), died within 1 year (10%), had prior use of a bone-targeting agent (9%), or switched to another bone-targeting agent during follow-up (2%). Among the 24.6% (8690) of eligible patients (median [IQR] age 65 years [58 to 74]; 58% [5039] female), breast cancer was the most common diagnosis (41% [3577]), followed by lung (30% [2577]) and prostate cancer (29% [2536]). Given that all administered doses throughout the study were standard 120-mg injections, dose reduction strictly reflected decreased administration frequency. Three dose reduction strategies were evaluated based on treatment patterns within the first year after receiving a mandatory three-dose induction phase in the initial 4 months: (1) cumulative dosage (3, 4 to 7, 8 to 10, or ≥ 11 injections), (2) maximum injection interval (≤ 60 days, 61 to 90 days, > 90 days, or no further treatment), and (3) frequency of gaps > 60 days (0, 1, ≥ 2, or no further treatment). Following the initial 1-year exposure window, patients were tracked up to a maximum of 3 years (median [IQR] time 19 months [10 to 33]) or until the occurrence of a study outcome or death. The primary outcome was surgically treated SREs. We excluded nonoperatively treated fractures because conservative management lacks pathologic confirmation in claims data, which introduces substantial ascertainment bias by confusing true metastatic events with osteoporotic or traumatic fractures using ICD codes alone. Therefore, to ensure diagnostic accuracy, an SRE was operationally defined as a fracture surgical procedure accompanied by a concurrent bone tumor excision surgical code, which under the national health insurance system requires a confirmatory pathologic report for reimbursement. Secondary outcomes included surgically treated ONJ, osteoporotic fractures, all-cause fractures, and all-cause mortality. To explicitly differentiate true osteoporotic fractures from SREs or high-energy trauma, we utilized a strict surgical code-based exclusionary definition: We isolated surgical procedures utilizing nontraumatic fracture codes while explicitly ruling out any concurrent tumor excision codes or traumatic fracture codes. To evaluate these outcomes, we estimated cause-specific HRs using Cox proportional hazards models. Additionally, to account for the competing risk of mortality, we calculated subdistribution HRs using Fine and Gray models. Without randomization, patients in this retrospective cohort inherently had unequal baseline risks of skeletal complications. To isolate the independent effect of the dosing strategy from inherent baseline imbalances, both models utilized multivariable analyses to calculate adjusted HRs (cause-specific adjusted HR [cs-aHR] for the Cox models and adjusted subdistribution HR [aSHR] for the Fine-Gray models), controlling for observable confounders that independently influence bone fragility, specifically age, sex, comorbidities, baseline medications, and prior fracture history.
RESULTS: Compared with patients receiving ≥ 11 doses, patients receiving only three doses were associated with a higher risk of surgically treated SREs (cs-aHR 3.20 [95% confidence interval (CI) 1.22 to 8.37]; p = 0.03) and lower risk of surgically treated ONJ (cs-aHR 0.21 [95% CI 0.07 to 0.65]; p < 0.001), with no difference in surgically treated osteoporotic fractures. Compared with patients with intervals of ≤ 60 days, patients extending the maximum injection interval beyond 90 days were associated with a higher proportion of surgically treated SREs (cs-aHR 2.50 [95% CI 1.35 to 4.66]; p = 0.003) and less surgically treated ONJ (cs-aHR 0.54 [95% CI 0.35 to 0.81]; p = 0.002), with surgically treated osteoporotic fractures again being no different. Compared with patients with 0 gaps, patients experiencing ≥ 2 treatment gaps for > 60 days were associated with a lower risk of surgically treated ONJ (cs-aHR 0.36 [95% CI 0.18 to 0.73]; p = 0.001) but showed no association with surgically treated SREs and osteoporotic fractures. Lower treatment intensity across all strategies was associated with higher all-cause mortality. In subgroup analyses stratified by primary cancer site, dose reduction was associated with a higher risk of surgically treated SREs in patients with prostate cancer across all three strategies. Specifically, patients who cumulated only three doses (cs-aHR 17.42 [95% CI 1.87 to 162.50]; p < 0.001), patients extending their maximum injection interval to no further treatment (cs-aHR 11.15 [95% CI 1.58 to 78.90]; p = 0.02), and patients experiencing a frequency of gaps leading to no further treatment (cs-aHR 11.22 [95% CI 1.59 to 79.23]; p = 0.02) were associated with higher risk of surgically treated SREs. Conversely, dose reduction strategies showed no association with the risk of surgically treated SREs in patients in either the breast or lung cancer subgroups. A lower risk of surgically treated ONJ was consistently observed across all types of dose reduction strategies. Finally, the subgroup findings for surgically treated osteoporotic fractures and all-cause mortality mirrored the overall cohort, demonstrating no association with surgically treated osteoporotic fractures and a generally higher risk of mortality associated with lower treatment intensity across cancer types.
CONCLUSION: In this nationwide retrospective comparative study, extending the denosumab dosing interval up to 90 days maintained efficacy against SREs and was associated with a low risk of ONJ. However, dosage intervals beyond 90 days or very low cumulative dosing were associated with a high risk of SREs. These findings support the potential feasibility of modest dose reduction in clinical practice, but they highlight the need for larger controlled trials to confirm the efficacy of extending dosing intervals and adjusting cumulative dosage.
LEVEL OF EVIDENCE: Level III, therapeutic study.