Accuracy and clinical implications of seven 25-hydroxyvitamin D methods compared with liquid chromatography-tandem mass spectrometry as a reference

Background: The most reliable assessment of vitamin ID status is measurement of plasma 25-hydroxyvitamin D (25[OH]D) concentration. High variability in 25(OH)D measurements due to utilized test and assay technologies and the lack of standardization against reference materials and reference method often confounds proper assessment of vitamin ID status. Methods: We evaluated the accuracy of six routinely available methodologies: high-performance liquid chromatography (HPLC), the IDS-radioimmunoassay (IDS-RIA) and enzyme immunoassay (IDS-EIA), the Nichols Advantage automated protein-binding assay (Advantage), two versions of the DiaSorin automated immunoassay (Liaison 1 and Liaison 2) - and one prototype automated immunoassay (Elecsys) for assessment of the 25(OH)D-3 status in a cohort of 300 randomly selected patients' samples compared with the reference method liquid chromatography-tandem mass spectrometry (LC-MS/MS). Results: Passing-Bablok regression analysis demonstrated a slope for each method compared with LC-MS/MS that varied from 0.62 (IDS-EIA) to 1.0 (HPLC). The Advantage and the Liaison 1 showed significant deviation from linearity with highly variable individual results vs. the LC-MS/MS. Difference plots revealed a considerable persistent proportional bias for the IDS-RIA and IDS-EIA. All evaluated methods except HPLC demonstrated a more or less considerable deviation of individual 25(OH)D-3 values compared with LC-MS/MS defined target concentrations. Conclusions: Standardization of methods for the quantification of 25(OH)D on a human-based sample panel by means of LC-MS/MS would help to reduce the inter-method variability with respect to accuracy existing in 25(OH)D measurement considerably. However, there will still remain differences in the accuracy of methods utilizing sample purification before final quantification or immunological reaction when compared with those methods without separate sample purification.