Introduction: Denosumab is a potent antiresorptive treatment often prescribed in treatment of severe osteoporosis. The observed resurgence of bone resorption markers after treatment discontinuation as well as loss of bone mineral density suggests a so-called rebound effect around 6- 30 months after the last injection consequently increasing the risk of fractures. This study aims to evaluate through a longitudinal analysis the probability and occurrence of fractures after denosumab discontinuation. Moreover, the predictability at the individual level based on bone mineral density values, bone resorption markers and treatment characteristics is evaluated in order to assess if clinical or biological monitoring could be used to prevent this serious consequence of denosumab discontinuation
Methods: The 784 women from the cohort of the Swiss Denosumab Group, affected either by postmenopausal osteoporosis or by breast cancer treated by aromatase inhibitors were included. Only patients who had gotten at least two doses of Denosumab and subsequently discontinued treatment at least one year were included. An online questionnaire recorded baseline characteristics and covered separately periods before, during and after denosumab administration in order to assess treatment patterns and clinical outcomes. Beta-Crosslaps (CTX) were used as bone resorption markers (BRM) and Dual X-ray imaging served to measure bone mineral density (BMD). Only osteoporotic fractures, occurring after minor trauma or everyday activities, were included in this study. The main outcome was the occurrence of fractures after denosumab discontinuation, in order to describe the evolution of the probability over time after the last denosumab injection as well as by integrating BRM, BMD and treatment pattern. Fracture occurrence and their probability was assessed with linear and nonlinear regression models by taking time-related information into consideration. Hierarchical mixed-effect longitudinal models as well as models inspired from pharmacokinetics, especially Bateman model for drug monitoring, were used.
Results: A 27% decrease of BRM levels has been observed when bisphosphonates (BP) were administered at any time in the patient’s history compared to cases where no BPs were administered, but no correlation between BRM follow-up and fracture occurrence has been found. The fracture incidence evolved from 5.5% during denosumab treatment to 12.5% after denosumab discontinuation with a mean occurrence at 13.1 months after treatment cessation. The modelled peak fracture probability at an individual level was at 9.8 months following the last denosumab injection with a baseline expectation of 1.8/10’000 fractures per month increasing to 33/10’000 at the abovementioned modelled peak. BPs given at any time significantly reduced the overall monthly fracture risk by 83%. The lowest fracture incidence has been observed when BPs were administered before and after denosumab. Breast cancer did not increase basal fracture probability but showed a two-fold increase of the fracture risk at the peak. No significant impact of BRMs on fracture occurrence has been found. BMD had an effect on basal risk of fractures, with 1 T-score unit at lumbar spine decreasing the expected number of fractures at all three sites.
Conclusion: Our study does not support the suggestion that closely monitoring BRM or BMD after densoumab discontinuation could predict and be used to prevent rebound fractures. A BP treatment given at any time significantly reduced fracture incidence after treatment discontinuation, with the most protective effect when given before and after denosumab administration.