Relationship between bone mineral density changes with denosumab treatment and risk reduction for vertebral and nonvertebral fractures

Abstract

Dual-energy X-ray absorptiometric bone mineral density (DXA BMD) is a strong predictor of fracture risk in untreated patients. However, previous patient-level studies suggest that BMD changes explain little of the fracture risk reduction observed with osteoporosis treatment. We investigated the relevance of DXA BMD changes as a predictor for fracture risk reduction using data from the FREEDOM trial, which randomly assigned placebo or denosumab 60 mg every 6 months to 7808 women aged 60 to 90 years with a spine or total hip BMD T-score < -2.5 and not < -4.0. We took a standard approach to estimate the percent of treatment effect explained using percent changes in BMD at a single visit (months 12, 24, or 36). We also applied a novel approach using estimated percent changes in BMD from baseline at the time of fracture occurrence (time-dependent models). Denosumab significantly increased total hip BMD by 3.2%, 4.4%, and 5.0% at 12, 24, and 36 months, respectively. Denosumab decreased the risk of new vertebral fractures by 68% (p < 0.0001) and nonvertebral fracture by 20% (p = 0.01) over 36 months. Regardless of the method used, the change in total hip BMD explained a considerable proportion of the effect of denosumab in reducing new or worsening vertebral fracture risk (35% [95% confidence interval (CI): 20%-61%] and 51% [95% CI: 39%-66%] accounted for by percent change at month 36 and change in time-dependent BMD, respectively) and explained a considerable amount of the reduction in nonvertebral fracture risk (87% [95% CI: 35% - >100%] and 72% [95% CI: 24% - >100%], respectively). Previous patient-level studies may have underestimated the strength of the relationship between BMD change and the effect of treatment on fracture risk or this relationship may be unique to denosumab.

Fig. 1

Relationship between new or worsening vertebral fracture incidence at 36 months and percent change from baseline in total hip BMD at 36 months. Adjusted estimates for a baseline lumbar spine BMD T-score of −2.5. Data represent the 5th through the 95th percentiles of total hip BMD percent change. The density curves at the bottom represent the distributions of total hip BMD change at 36 months for each treatment group. For both denosumab and placebo, the risk of new or worsening vertebral fracture decreased with increasing percent change in total hip BMD but the slope of the curves differed between treatment groups (interaction p value = 0.0003).

Fig. 2

Relationship between nonvertebral fracture incidence at 36 months and percent change from baseline in total hip BMD at 36 months. Adjusted estimates for a baseline total hip BMD T-score of −2.5. Data represent the 5th through the 95th percentiles of total hip BMD percent change. The density curves at the bottom represent the distributions of total hip BMD change at 36 months for each treatment group. For both denosumab and placebo, the risk of nonvertebral fracture decreased with increasing percent change in total hip BMD. The data suggest similar relationships (slopes) for both treatment groups (interaction p value = 0.38).