Plasma periostin associates significantly with non-vertebral but not vertebral fractures in postmenopausal women: Clinical evidence for the different effects of periostin depending on the skeletal site

Background: Periostin is preferentially expressed by the periosteum, which mainly covers the long bones. Therefore, the role of periostin in osteoporotic fracture (OF) may differ depending on bone type. We performed a case-control study to investigate whether periostin can serve as a predictor of OF risk, particularly after dividing OFs into non-vertebral and vertebral fractures. Methods: Among 532 consecutive postmenopausal women not taking any drug or without any disease that could affect bone metabolism, 133 cases with OF (i.e., non-vertebral and/or vertebral fractures) and 133 age- and body mass index-matched controls were enrolled. Non-vertebral (i.e., forearm, humerus, hip, and pelvis; n = 81) and morphological vertebral (n = 62) fractures were identified by an interviewer-assisted questionnaire and lateral thoracolumbar radiographs, respectively. Bone mineral density (BMD) and plasma periostin levels were also measured. Results: Plasma periostin was markedly higher in subjects with non-vertebral fracture than their controls even after adjustment for BMD and potential confounders (P = 0.006). Each standard deviation increment of plasma periostin was associated with a multivariable-adjusted odds ratio of 1.59 for non-vertebral fracture. The odds for non-vertebral fracture were 2.48-fold higher in subjects in the highest periostin tertile compared with those in the lowest periostin tertile (95% confidence interval = 1.10-5.61). However, associations between plasma periostin and vertebral fracture were not observed, regardless of the adjustment model used. Consistently, plasma periostin levels were inversely associated with proximal femur BMD (P = 0.007 to 0.030) but not lumbar spine BMD. In subgroup analyses, plasma periostin had no correlation with the levels of classical bone turnover markers. Conclusions: Plasma periostin may be a potential biomarker of the risk of OF, especially in non-spinal skeletal sites, such as the limbs, rather than spine.