Serum levels of a cathepsin‐K generated periostin fragment predict incident low‐trauma fractures in postmenopausal women independently of BMD and FRAX

N Bonnet, E Biver, T Chevalley, R Rizzoli… - Journal of Bone and …, 2017 - academic.oup.com
N Bonnet, E Biver, T Chevalley, R Rizzoli, P Garnero, SL Ferrari
Journal of Bone and Mineral Research, 2017academic.oup.com
Periostin is a matricellular protein involved in bone formation and bone matrix organization,
but it is also produced by other tissues. Its circulating levels have been weakly associated
with bone microstructure and prevalent fractures, possibly because periostin measured by
the current commercial assays does not specifically reflect bone metabolism. In this context,
we developed a new ELISA for a periostin fragment resulting from cathepsin K digestion (K‐
Postn). We hypothesized that circulating K‐Postn levels could be associated with bone …
Abstract
Periostin is a matricellular protein involved in bone formation and bone matrix organization, but it is also produced by other tissues. Its circulating levels have been weakly associated with bone microstructure and prevalent fractures, possibly because periostin measured by the current commercial assays does not specifically reflect bone metabolism. In this context, we developed a new ELISA for a periostin fragment resulting from cathepsin K digestion (K‐Postn). We hypothesized that circulating K‐Postn levels could be associated with bone fragility. A total of 695 women (age 65.0 ± 1.5 years), enrolled in the Geneva Retirees Cohort (GERICO), were prospectively evaluated over 4.7 ± 1.9 years for the occurrence of low‐trauma fractures. At baseline, we measured serum periostin, K‐Postn, and bone turnover markers (BTMs), distal radius and tibia microstructure by HR‐pQCT, hip and lumbar spine aBMD by DXA, and estimated fracture probability using the Fracture Risk Assessment Tool (FRAX). Sixty‐six women sustained a low‐trauma clinical fracture during the follow‐up. Total periostin was not associated with fractures (HR [95% CI] per SD: 1.19 [0.89 to 1.59], p = 0.24). In contrast, K‐Postn was significantly higher in the fracture versus nonfracture group (57.5 ± 36.6 ng/mL versus 42.5 ± 23.4 ng/mL, p < 0.001) and associated with fracture risk (HR [95%CI] per SD: 2.14 [1.54 to 2.97], p < 0.001). After adjustment for aBMD, FRAX, bone microstructure, or BTMs, K‐Postn remained significantly associated with fracture risk. The performance of the fracture prediction models was improved by adding K‐Postn to aBMD or FRAX (Harrell C index for fracture: 0.70 for aBMD + K‐Post versus 0.58 for aBMD alone, p = 0.001; 0.73 for FRAX + K‐Postn versus 0.65 for FRAX alone, p = 0.005). Circulating K‐Postn predicts incident fractures independently of BMD, BTMs, and FRAX in postmenopausal women. Hence measurement of a periostin fragment resulting from in vivo cathepsin K digestion may help to identify subjects at high risk of fracture. © 2017 American Society for Bone and Mineral Research
Oxford University Press