Repeated Romosozumab With Short-Term Ibandronate Sequencing in High-Risk Osteoporosis: A Retrospective Single-Center Cohort Study

Abstract Long-term real-world evidence on repeated romosozumab treatment remains limited. This retrospective single-center cohort included 68 patients who initiated treatment between April 2020 and April 2026. Nineteen patients discontinued during the 5-year observation period, and 49 completed follow-up and formed the longitudinal analytical cohort. Patients received romosozumab for 12 months followed by oral ibandronate 100 mg once monthly for 3 months before reassessment for another course. All patients received eldecalcitol 0.75 μg daily. Recorded lumbar spine and femoral young adult mean values, bone turnover markers, numerical rating scale scores, persistence, fractures, and prespecified safety events were evaluated. Longitudinal outcomes were analyzed using linear mixed-effects models with a patient-level random intercept. Lumbar spine young adult mean increased from 73.3% at baseline to 92.9% at 5 years, with a model-estimated change of 19.6 percentage points (95% confidence interval, 17.9–21.3; p<0.001). Femoral young adult mean increased from 66.2% to 76.1%, with a change of 9.9 percentage points (95% confidence interval, 9.2–10.5; p<0.001). Numerical rating scale scores decreased from 8.9 to 1.4, with an estimated change of −7.5 (95% confidence interval, −7.9 to −7.1; p<0.001). Interval-based treatment persistence was 72.1% at 5 years (95% confidence interval, 59.8%–81.2%). One recurrent vertebral fracture was recorded, whereas no cardiovascular events, hypocalcemia, osteonecrosis of the jaw, or atypical femoral fractures were recorded. Repeated courses were accompanied by longitudinal improvements in recorded skeletal measures and bone turnover markers. Because this was an uncontrolled retrospective study and patient-level baseline data for the 19 patients who discontinued were unavailable in the analytical dataset, the findings should be considered descriptive and hypothesis-generating.