Treatment-Related Changes in Bone Turnover and Fracture Risk Reduction in Clinical Trials of Anti-Resorptive Drugs: A Meta-Regression.

Few pooled analyses of antiresorptive (AR) treatment trials relate short-term changes in bone turnover markers (BTMs) to subsequent fracture reduction. Such information would be useful to assess new ARs or novel dosing regimens. In the Foundation for the National Institutes of Health (FNIH) Bone Quality project, we analyzed individual-level data from 28,000 participants enrolled in 11 bisphosphonate (BP) and three selective estrogen receptor modulator (SERM) placebo-controlled fracture endpoint trials. Using BTM results for two bone formation markers (bone-specific alkaline phosphatase [bone ALP] and pro-collagen I N-propeptide [PINP]) and two bone resorption markers (N-terminal and C-terminal telopeptide of type I collagen) and incident fracture outcome data, we performed a meta-regression relating the mean net effect of treatment on change in bone turnover (active minus placebo % difference after 3 to 12 months) to the log of study-wide fracture risk reduction, and used linear regression to plot the best fitting line. Separate analyses were performed for incident morphometric vertebral, nonvertebral, and hip fractures over 1 to 4 years of follow-up. Change in bone ALP and PINP were available for over 16,000 and 10,000 participants, respectively. For vertebral fracture, the results showed a strong relationship between treatment-related bone ALP or PINP changes and vertebral fracture risk reduction (r2  = 0.82 [p < 0.001] and r2  = 0.75 [p = 0.011], respectively) Relationships were weaker and no longer statistically significant for nonvertebral (r2  = 0.33 [p = 0.053] and r2  = 0.53 [p = 0.065], respectively) and hip fracture (r2  = 0.17 [p = 0.24] and r2  = 0.43 [p = 0.11], respectively) outcomes. Analyses limited to BP trials gave similar results. For all fracture types, relationships were weaker and nonsignificant for bone resorption markers. We conclude that short-term AR treatment-related changes in bone ALP and PINP strongly predict vertebral fracture treatment efficacy, but not nonvertebral or hip fracture treatment efficacy. Change in bone formation markers might be useful to predict the anti-vertebral fracture efficacy of new AR compounds or novel dosing regiments with approved AR drugs.