Commercial assays for serum osteocalcin give clinically discordant results.

Serum samples from 9 healthy controls and from subjects with primary hyperparathyroidism (n = 5), Paget disease (n = 3), pregnancy (n = 5), glucocorticoid therapy (n = 5), postmenopausal osteoporosis (n = 10), and renal failure (n = 10) were used to assess the clinical agreement among eight commercially available assay kits for osteocalcin (OC). These kits differ in their assay configurations (six radioimmunoassays, two immunoradiometric assays), standards (five bovine, three human), and antibodies (six polyclonal, two monoclonal). Individual results were divided by the mean OC of the control subjects for each assay and expressed as percentage deviations. The expected wide variation in absolute OC concentrations between kits was only partially reduced by this transformation. Agreement was equally poor when absolute OC concentrations were compared with the reference ranges quoted by the manufacturers. The discordance was particularly marked in renal failure, presumably because of immunoreactive fragments, and in osteoporosis. Systematic differences could not be attributed to assay format, species source of standard, or antibody specificity. We conclude that results cannot be compared between assays even when normalized against healthy subjects, and that standardization is needed.