Overreporting of vitamin D deficiency by the Roche Elecsys vitamin D3 (25-OH) method

<b>Background</b>: Vitamin D deficiency is common. Recently Roche Diagnostics removed their Elecsys Vitamin D3 (25OH) electrochemiluminescence immunoassay (ECLIA) from use, citing deteriorating traceability to the reference method (liquid chromatography tandem mass spectrometry; LCMSMS). We investigated the performance of the Roche assay (2 assay formulations) against an LCMSMS method and the widely used DiaSorin radioimmunoassay (RIA) method.<br /><br /><b>Methods</b>: Two sets of samples from separate populations were assayed for vitamin D. The first set was assayed using three different methods: RIA (DiaSorin) in 2004, polyclonal ECLIA (Roche) in early 2009 and LCMSMS in early 2010. The second set was assayed using polyclonal and monoclonal ECLIA (Roche) and LCMSMS in mid-2010.<br /><br /><b>Results</b>: The correlation of the polyclonal ECLIA with the RIA was poor (ECLIA = 0.45 x RIA + 19, r² = 0.59, n = 773). LCMSMS results correlated with RIA (RIA = 0.86 x LCMSMS + 4, r² = 0.69, n = 49) better than with polyclonal ECLIA (polyclonal ECLIA = 0.55 x LCMSMS + 6, r² = 0.62, n = 55) despite a storage interval of 6 years.<br /><br />In recently collected samples monoclonal and polyclonal immunoassays gave similar results (monoclonal ECLIA = 0.93 polyclonal ECLIA -3, r² = 0.60, n = 153). The correlation between monoclonal Roche ECLIA and LCMSMS in these samples was very poor (monoclonal ECLIA = 0.31 x LCMSMS + 23, r² = 0.27).<br /><br /><b>Conclusions</b>: At the time of its removal from the market, the Roche Elecsys Vitamin D3 (25OH) assay showed unacceptable performance, underestimating vitamin D levels. It seems that this bias preceded the introduction of the monoclonal assay. The worldwide distribution of the assay and the duration of this bias likely led to a significant number of patients starting supplementation unnecessarily.