Analytical Variability Among Methods for the Measurement of 25-Hydroxyvitamin D

Abstract

To compare total 25-hydroxyvitamin D [25(OH) D] results measured by 3 direct immunoassays, including the previous version of the DiaSorin Liaison2 assay and the current versions of the Siemens Centaur2 and the Abbott Architect assays, with results measured in serum extracts by liquid chromatography/tandem mass spectrometry (LC/MS) and radioimmunoassay (RIA). Our study sample consisted of 163 consecutive clinical specimens submitted to our laboratory for 25(OH)D testing. Regression and bias analyses of the data revealed that results measured by the 3 direct immunoassay methods had high degrees of random variability and bias relative to the results determined by LC/MS and RIA. The relative biases between results measured by the direct assays and the comparison methods exceeded a recommended criterion for the total allowable error of a 25(OH)D test in as many as 48% of our clinical specimens. Of the subjects in our study sample, 33, 37, 30, 45, and 71 were classified as vitamin D deficient based on results determined by LC/MS, RIA, Liaison2, Architect, and Centaur2, respectively. Intermethod variability in 25(OH)D assays continues to limit our progress toward the establishment of reference values for 25(OH)D in health and our efforts to gain a better understanding of the role of vitamin D insufficiency as a risk factor for disease.