An evaluation of the Fracture Risk Assessment Tool (FRAX®) as an indicator of treatment efficacy: the effects of bazedoxifene and raloxifene on vertebral, nonvertebral, and all clinical fractures as a function of baseline fracture risk assessed by FRAX®.

SUMMARY: The relationship between baseline Fracture Risk Assessment Tool (FRAX®) and treatment efficacy was evaluated using data from a pivotal phase 3 study. Relative risk of vertebral, nonvertebral, and all clinical fractures decreased with increasing probability of fracture for bazedoxifene (BZA) versus placebo but remained generally constant for raloxifene (RLX).

INTRODUCTION: To determine whether the FRAX® predicts osteoporosis treatment efficacy, we evaluated reductions in fracture incidence associated with BZA and RLX according to baseline fracture risk determined by FRAX® using data from a phase 3 osteoporosis treatment study.

METHODS: Hazard ratios (HRs) for effects of BZA and RLX versus placebo on incidence of vertebral, nonvertebral, and all clinical fractures were calculated using a Cox regression model. Cox regression analyses were performed in subgroups at or above 10-year fracture probability thresholds determined by FRAX®.

RESULTS: HRs for the risk of vertebral, nonvertebral, and all clinical fractures versus placebo decreased with increasing 10-year fracture probability for BZA, while those for RLX remained stable. In all 10-year fracture probability subgroups, all BZA doses significantly reduced vertebral fracture risk versus placebo (HR = 0.22-0.66). BZA at 20, 40, and 20/40 mg significantly reduced risk of nonvertebral fractures (HR = 0.45, 0.44, and 0.45, respectively) and all clinical fractures (HR = 0.38, 0.41, and 0.40, respectively) for ≥20.0 % fracture probability. Vertebral fracture risk reductions for RLX 60 mg versus placebo were significant in subgroups at lower fracture probabilities (≥2.5- ≥ 10.0 %), but not higher (≥12.5 %), and in no subgroups for nonvertebral or all clinical fractures.

CONCLUSION: The antifracture efficacy of BZA increased with increasing baseline FRAX® score, but there was no clear relationship between RLX and baseline FRAX®. These findings provide independent confirmation of current literature, suggesting that the relationship between FRAX® and treatment efficacy varies for different agents.