Comparison of a Reformulated Automated Vitamin D Assay, Its Predecessor, and 2 Contemporary Formulations

Historically, 25-hydroxyvitamin D (25OHD) assays have under- or over-recovered 25-hydroxyvitamin D2 (25OHD2), but assay manufacturers have modified their reagents to address this problem. In this study, we compared the second- and third-generation Roche assays as well as two contemporary offerings from Diasorin and Beckman against liquid chromatography-tandem mass spectrometry (LC-MS/MS).

We identified 50 remnant serum samples with 25OHD concentrations from across the analytical range of the second-generation Roche assay. To increase 25OHD2 representation, we identified 25 additional samples from individuals prescribed high-dose vitamin D2 supplements. We tested samples on Roche assays and circulated to laboratories performing Beckman and Diasorin 25OHD assays. We tested samples by LC-MS/MS to obtain concentrations for 25OHD2 and 25-hydroxyvitamin D3.

Mean overall bias for each assay was 5.1 ng/mL or less against the LC-MS/MS measurement; mean proportional bias was 8.7% to 12.1%. Some individual specimens had much larger bias. 25OHD2 was under-recovered on average, but the bias for the third-generation Roche assay represents a significant improvement over the previous assay, and mean bias for current generation assays was no worse than -3.2 ng/mL. In most cases, clinical classification by automated assay values agreed with clinical classification by LC-MS/MS; where present, disagreements occurred near classification thresholds.

Automated 25OHD assays continue to improve, and 25OHD2 recovery no longer appears to be a significant concern for the assays evaluated here. All assays evaluated were adequate for clinical classification of vitamin D nutritional status and are suitable for routine use, including in patients prescribed high-dose vitamin D2.