Undercarboxylated osteocalcin measured with a specific immunoassay predicts hip fracture in elderly women: the EPIDOS Study.

Increased levels of circulating undercarboxylated osteocalcin (ucOC), measured indirectly with the hydroxyapatite (HAP) binding assay, have been shown to predict hip fracture risk in a small group of elderly institutionalized women. The aim of this study was to confirm these findings in a prospective cohort study (EPIDOS prospective study) of 7598 healthy, independently living women over 75 yr of age. One hundred and four women who sustained a hip fracture during a 22-month follow-up period were age matched with 255 controls who did not fracture. Baseline samples were collected before hip fracture for measurement of total OC and ucOC, assessed either with the HAP binding assay or directly with a new enzyme-linked immunosorbent assay (ELISA). This direct ELISA uses human recombinant noncarboxylated OC as a standard and two monoclonal antibodies, one of which was raised against the 14-30 Glu synthetic peptide. We found that the intra- and interassay variations are less than 11%, and this assay exhibits a 5% cross-reactivity with purified human bone OC, used as a source of carboxylated OC. ucOC levels measured with this ELISA correlated well with the HAP binding assay in the population of 359 elderly women (r = 0.82; P < 0.0001). We estimated the risk of hip fracture for women with levels of ucOC in the highest quartile of values for the 255 controls. We found that increased levels of ucOC measured by ELISA were associated with increased hip fracture risk with an odds ratio (OR) of 1.9 (95% confidence interval, 1.2-3.0), and the ELISA had a greater sensitivity than the HAP assay. In contrast, total OC was not associated with hip fracture risk. After adjustment for femoral neck bone mineral density (BMD) and mobility status assessed by gait speed, ucOC still predicted hip fracture with an OR of 1.8 (1.0-3.0). Women with both femoral neck BMD in the lowest quartile and ucOC in the highest quartile were at higher risk of hip fracture, with an OR of 5.5 (2.7-11.2), than those with only low BMD or high ucOC levels. In conclusion, we have developed a new specific ELISA for serum ucOC, with low cross-reactivity with carboxylated OC and increased specificity and sensitivity over the HAP assay. Using this new ELISA, we found that ucOC, but not total OC, predicts hip fracture risk independently of femoral neck BMD in elderly women drawn from the general population. Thus, ucOC measurement could be combined with bone mass determination to improve the assessment of hip fracture risk in elderly women.