Abstract
Background
Liquid chromatography-tandem mass spectrometry (LC-MS/MS) is becoming increasingly popular for measuring 25-hydroxyvitamin-D (25-OH-D). Results submitted to the International Quality Assessment Scheme (DEQAS) have shown poor interlaboratory agreement. We investigated whether the use of a common standard would reduce interlaboratory imprecision.
Methods
A commercial standard and two controls were distributed with the DEQAS samples in January 2008. Participants were asked to calculate the results of samples and controls using their usual standard and the commercial standard. A method questionnaire was also distributed.
Results
Use of a common standard reduced the mean interlaboratory imprecision (coefficient of variation [CV]) for total 25-OH-D from 16.4% (in-house standards) to 10.4% (common standard). For 25-OH-D3 and 25-OH-D2, the mean CVs were reduced from 16.7% and 21.1% to 8.5% and 12.6%, respectively. Mean values obtained for total 25-OH-D using the common standard were higher by 6.1%.
Conclusions
Use of a common standard improved agreement among laboratories using LC-MS/MS methods for 25-OH-D. This suggests that problems with assay standardization contribute to interlaboratory imprecision. This may be related to the nature of the matrix used for working standards or errors in the calibration of stock standard solutions of 25-OH-D. Some participants used a gravimetric method, others UV spectrophotometry, to establish the concentration of stock solutions. Among the latter group there was uncertainty over the molar absorption coefficient of 25-OH-D solutions. We conclude that LC-MS/MS is not yet sufficiently robust to become the reference method for 25-OH-D and that gas chromatography-mass spectrometry might be a more suitable candidate.
Introduction
The International Quality Assessment Scheme (DEQAS) for 25-hydroxyvitamin-D (25-OH-D) was established in 1989 and now has over 450 participants in 32 countries (April 2008).
The majority of participants use manual or automated immunoassays. However, methods based on liquid chromatography-tandem mass spectrometry (LC-MS/MS) are becoming increasingly popular but interlaboratory coefficients of variation (CV) of this group are relatively high. Unlike immunoassay kits, which include standards, liquid chromatography methods are often calibrated with ‘in-house’ standards. We wished to investigate the possibility that differences in assay standardization might be responsible for poor interlaboratory agreement among users of liquid chromatography methods for 25-OH-D.
Methods
DEQAS distributes five liquid human serum samples quarterly at ambient temperature. Results are analysed by the method of Healy.
1
Detailed operation of the scheme has been described previously
2
and information is available at the DEQAS website (
In the January 2008 distribution, participants known to use liquid chromatography methods were sent a lyophilized commercial standard and two controls (Chromsystems GmbH, Munich, Germany), in addition to the five DEQAS samples. One DEQAS sample (330) included the serum of a subject taking a vitamin D2 supplement.
Participants were asked to analyse the Chromsystems controls and DEQAS samples and to calculate the results against the common (commercial) standard as well as their usual standard. Participants were also sent a questionnaire requesting details of their assay, with particular reference to standard preparation. Here, we report the questionnaire responses and results from the users of LC-MS/MS methods.
Results
Questionnaires were completed by 22 LC-MS/MS users of whom six routinely used a standard purchased from Chromsystems and 16 used in-house standards prepared in various matrices (animal/human serum, bovine/human albumin solutions, ethanol/methanol:water) which were extracted in the same way as samples. Participants were asked how they calibrated their stock standard solutions of 25-OH-D3 and 25-OH-D2. Of the 16 LC-MS/MS users preparing in-house standards, two used a gravimetric method and 14 calculated the concentration from the measured UV absorbance at Amax. The molar absorption coefficients (MACs) used in calculating 25-OH-D3 concentrations (participant numbers in parentheses) were 18200 (two), 18300 (ten), 18584 (one) and 19400 (one). The MACs used for 25-OH-D2 were 18200 (two), 18300 (seven), 18900 (one) and 19400 (four). Hexa-deuterated 25-OH-D3 was used as an internal standard by 21 of the group. The other participant used deuterated delta-9-tetrahydrocannabinol.
Results for the DEQAS samples, Chromsystems controls and standard are given in Table 1. Four participants reported 25-OH-D2 in one or more of the samples 326–328; results ranged from 2.0 to 24.8 (median 4.7) nmol/L. These results were used in the calculation of total 25-OH-D but were not included in the 25-OH-D2 comparison data.
Inter-laboratory imprecision (CV%) and mean concentrations (nmol/L) of Total 25-OH-D, 25-OH-D3 and 25-OH-D2 for DEQAS samples (326-330), Chromsystems controls (C1, C2) and standard (CS) calculated with the participants’ usual standards (US) and the Chromsystems’ standard (CS). Figures in parenthesis are the concentrations given in the Chromsystems package insert
Key: *Results calculated after the removal of 1 outlier (>3 SD from Mean)
After calculating the means and standard deviation (SD), outliers (results >3 SD from the mean) were removed and the means and SDs were recalculated.
Use of a common standard reduced the mean interlaboratory imprecision (CV) for total 25-OH-D from 16.4% (in-house standards) to 10.4% (common standard). For the individual metabolites, 25-OH-D3 and 25-OH-D2 (sample 330 and Chromsystems controls only), the mean CVs were reduced from 16.7% and 21.1% to 8.5% and 12.6%, respectively. Mean values obtained for total 25-OH-D using the Chromsystems standard were higher by 6.1%.
With the exception of sample 330, inter-laboratory CVs for 25-OH-D3 fell to below 10% when a common standard was used. The more modest fall for total 25-OH-D probably reflects inconsistencies in the detection of 25-OH-D2.
Discussion
A potential source of error in LC-MS/MS assays is the medium in which working standards are prepared. Signal intensities in LC-MS/MS assays are affected by the standard matrix 3 and the accuracy of results relies on the internal standard behaving identically to the analyte. Used appropriately, the internal standard should compensate for differences in extraction efficiency and procedural losses, rendering the need for standard extraction an unnecessary complication. Bovine serum albumin solutions, used as a standard matrix by four participants, may contain vitamin D-binding protein and 25-OH-D as contaminants, another possible source of error.
Inaccurate calibration of standards is a potential source of interlaboratory imprecision in all 25-OH-D methods. The concentration and purity of stock standard solutions should be established using a double-beam scanning spectrophotometer. Vitamin D metabolites have a well-defined absorbance spectrum with an absorbance maximum at 265 nm and a minimum at 228 nm. A fall in the ratio A max:A min to below 1.5 indicates deterioration of the standard, 4 which should be replaced or purified by liquid chromatography. Unfortunately, the MACs used to calculate the concentration of 25-OH-D3 and 25-OH-D2 solutions are still a subject of uncertainty. We have been unable to find the definitive reference to this but, in a personal communication, Dr Hector F DeLuca, Steenbock Research Professor, Department of Biochemistry, University of Wisconsin-Madison, a leading authority on vitamin D, indicated that 18200 is used in his laboratory for both 25-OH-D3 and 25-OH-D2. Professor DeLuca pointed out that the chromophore in both metabolites is the same and therefore the MACs should be identical. In our experience, using the weight given on manufacturers' data sheets or weighing solid 25-OH-D can be unreliable, as the starting material may contain impurities. Furthermore, different preparations may vary in their crystalline composition, some being monohydrates, others not.
Our findings have highlighted the important influence of assay standardization on interlaboratory precision of LC-MS/MS methods for serum 25-OH-D. Experience with immunoassays suggests that the widespread adoption of LC-MS/MS by routine clinical laboratories will depend on the support offered by instrument manufacturers and the availability of reagent ‘kits’. However, all manufacturers, including Chromsystems, will need to show greater transparency about their products, particularly with respect to the provenance of standards. No doubt, LC-MS/MS methods will continue to evolve and proposals that this technique might be adopted as a reference procedure for 25-OH-D 5 may eventually be accepted. Meanwhile, gas chromatography-mass spectrometry, historically used as the reference measurement procedure for other steroids, 6 might be a more suitable candidate.
Footnotes
Acknowledgements
We would like to thank Rob Jones of Chromsystems for provision of the lyophilized standard and controls, the International Quality Assessment Scheme participants who took part in this study and to Professor Hugh Makin, St Bartholomews and the Royal London School of Medicine and Dentistry, London, UK for his helpful comments during the preparation of this manuscript.