Background/Purpose: Denosumab (DMAb) treatment has been shown to decrease fracture (Fx) risk. Unlike bisphosphonates, DMAb is a monoclonal antibody against RANKL. Discontinuation is characterized by reversal of effect, including transient increases in bone turnover markers to above-baseline (BL) levels returning to BL within 2 yr and loss of bone mineral density (BMD) to BL levels. An analysis from FREEDOM showed similar subject incidence of Fx in DMAb and placebo (PBO) groups upon discontinuation. Recent reports describe isolated cases of multiple (≥2) vertebral Fx (VFx) after DMAb cessation (Aubrey-Rozier OI 2015; Anastasilakis OI 2015; Popp OI 2016). To better understand VFx incidence—particularly multiple VFx—upon DMAb cessation, we describe VFx risk and possible determinants in subjects who discontinued DMAb during the 3-yr FREEDOM or 7-yr Extension.

Methods: Subjects who received ≥2 doses of investigational product (IP; DMAb 60 mg Q6M or PBO) and discontinued IP but stayed in study ≥7 M after the last dose were included in the analysis. Subjects who discontinued DMAb from FREEDOM or Extension were analyzed as one group. A logistic regression model explored covariates related to off-treatment new VFx or multiple new VFx.

Results: Of 1001 subjects who discontinued DMAb during FREEDOM or Extension, 56 (5.6%) sustained new VFx. Upon DMAb discontinuation, new VFx incidence increased relative to the on-treatment period but stayed within the range of those who discontinued PBO, ie, had never been treated (Table). The same pattern was seen in subjects with prior VFx, who had higher on- and off-treatment VFx rates than the overall population. Among subjects with off-treatment new VFx, a greater percentage of those who discontinued DMAb (34/56 [60.7%]) than PBO (10/29 [34.5%]) sustained multiple new VFx. Logistic regression models found prior VFx before or during treatment to be the strongest predictor of off-treatment new VFx, including multiple VFx (odds ratio 2.1-3.4). Off-treatment femoral neck BMD loss was a weak covariate.

Conclusion: Discontinuation of DMAb is associated with an increase in VFx rate to levels comparable to PBO. Among subjects who sustained new VFx after DMAb cessation, there was a greater incidence of multiple new VFx than in PBO. Subjects with prior VFx are at high risk for off-treatment new VFx and should continue osteoporosis therapy. Consequently those who discontinue DMAb should transition to another therapy after the 6-M dosing interval. Table. Off-treatment vertebral fracture incidence in subjects who discontinued IP in FREEDOM or its Extension