Vitamin D plays a vital role in skeletal metabolism, calcium homeostasis, 123 and also in the functioning of the immune, cardiovascular, and reproductive systems 45. Vitamin D deficiency leads to rickets and osteomalacia and is also associated with breast and colorectal cancers, multiple sclerosis, dementia, rheumatoid arthritis, diabetes, Parkinson's and Alzheimer's diseases 67. Despite numerous reports, the associations of Vitamin D deficiency with health and diseases are subject to debate, partly owing to inadequacies in current approaches to measurement of serum levels. Depending on the source, Vitamin D is produced in two forms: Vitamin D2 and Vitamin D3, which differ by the presence of a double bond and methyl group on the aliphatic side chain. The issues involved in assessing Vitamin D status arise from the complexities of the metabolic pathways leading to a number of active forms. The complex metabolic pathway for Vitamin D3 is summarized in Figure 1.

Vitamin D3 is formed from its precursor 7-dehydrocholesterol in the skin by ultraviolet B light (medium wavelength, 290-315 nm) and Vitamin D2 originates from dietary sources together with some fraction of D3. In the liver, Vitamins D3 and D2 undergo hydroxylation reactions catalyzed by 25-hydroxylase, which leads to the formation of pharmacologically active metabolites 25OHD3 and 25OHD2 respectively (collectively termed as 25OHD). Further metabolism (in the presence of 1α,hydroxylase) in the kidney produces the pharmacologically active metabolites 1-alpha,25-dihydroxyvitamin-D3 (1α,25(OH)₂D2) and 1-alpha,25-dihydroxyvitamin-D2 (1α,25(OH)₂D3) along with the minor metabolite 24,25(OH)₂D3 8910.

Since 25OHD has significant effects on health and wellbeing, there has been a substantial interest in improving the relevant analytical techniques 1112131415161718192021222324252627282930. Owing to a long serum half-life, measurement of total 25OHD (25OHD2 and 25OHD3) is the routinely used approach for assessing the total circulating Vitamin D status 1011121314. In immunoassay techniques, a measure of total metabolite concentration and equivalent detection of both 25OHD2 and 25OHD3 is challenging, as binding proteins show a higher affinity for 25OHD3 than 25OHD2 15161718. Reports have shown inter-laboratory and inter-method variations in results for Vitamin D determinations 192021.

LC-MS/MS is currently the best technique available for the correct quantification of 25OHD3 and 25OHD2 2223 and it also has the capability to overcome most of the problems associated with protein binding assays. LC-MS/MS is a more favourable technique because sample derivatisation is not required, run time is very short and an internal standard is used which usually compensates for any matrix related and instrumental effects 242526272829303132.

However, the LC-MS/MS approach is also subject to interferences 3334353637. Along with matrix related, instrumental and analytical interferences, endogenous 25OHD determinations have also been shown to suffer from epimeric and isobaric interferences 38394041. Epimers are non-super imposable (or non mirror images) that only differ in the configuration at one carbon atom (Figure 2). Epimers and isobars are compounds with the same molecular weight as Vitamin D metabolites and form the same mass to charge parent and product ion pairs upon ionisation. Moreover, the separation of interfering epimers and isobars is also essential, because they can overlap chromatographically with Vitamin D metabolites or internal standard peaks and give false estimates of true Vitamin D levels. 25OHD3 is the most abundant Vitamin D metabolite in circulation and 3-epi-25OHD3 is the most prevalent epimer of 25OHD3. There are two compounds known to cause isobaric interferences in 25OHD analysis; 1α-hydroxyvitamin-D3 (1αOHD3), which is an exogenous pharmaceutical compound and 7α-hydroxy-4-cholestene-3-one (7αC4), which is an endogenous bile acid precursor 30383940414243.

The epimerization of 25OHD3 and 1α,25-(OH)₂D3 results in the formation of 3-epi-25OHD3 and 3-epi-1α,25(OH)₂D3 epimers respectively as shown in Figure 2. The epimers of 25OHD differ in configuration at third carbon atom (C-3) (shown by dashed highlights in Figure 2) that is attached to a hydroxyl group. Hydroxylation of 3-epi-25OHD3 forms 3-epi-1α,25(OH)₂D3 41.

The aim of this study was to develop a novel LC-MS/MS method that can accurately identify and quantitate 25OHD3 and 25OHD2 and chromatographically separate epimers and isobars.

Calibration curves and quality controls at low, medium and high concentration levels were prepared from the multi-level reference standards with known amounts of 25OHD. The assays were validated for specificity, recovery, linearity and intraday/interday precision and accuracy. The lower limit of detection (LLOD) for all analytes was found to be 0.25 ng/mL. The LLOD was determined by decreasing the analyte concentrations until a response equivalent to 3 times the background level was observed. The relative extraction recoveries at a final concentration of 50 ng/mL were determined by comparing the representative peak areas of extracted analytes (N = 6) to un-extracted analytes. The un-extracted analyte solutions were prepared in methanol. Validation results are summarised in Table 1.

Under- or over-estimation of actual 25OHD3 concentrations may occur owing to co-eluting epimers (e.g.3-epi-25OHD3) and isobars (e.g.7αC4) 39404142. The use of high resolution microbore ZORBAX column in tandem with chiral ULTRON column not only facilitated accurate determination of the analyte ions but also chromatographically separated all isobars and epimers from 25OHD co-eluting peaks.

Multiple reaction monitoring (MRM) mode was used for the analysis of 25OHD2 and 25OHD3. The MRM parameters were optimised using direct infusion of 0.1 μg/mL solutions of standard compounds. The MRM sequence consisted of two periods executed sequentially to monitor different transition pairs using parameters optimised for each period. The most abundant MRM ion transitions for each analyte were acquired using the conditions given in Table 2.

According to a joint position statement issued by British Association of Dermatologists, Diabetes UK, the Multiple Sclerosis Society UK, the National Heart Forum UK and the National Osteoporosis Society UK, the threshold level quoted for Vitamin D deficiency is ≤ 10 ng/mL (≤ 25 nmol/L) 48. Individuals with ≤ 10 ng/mL of 25OHD are regarded as Vitamin D deficient. Some experts define Vitamin D deficiency as 25OHD levels ≤ 20 ng/mL 495051. However, the levels of Vitamin D can vary between individuals and currently there is no standard set for optimal Vitamin D levels 495051. Amongst the five samples analysed by LC-MS/MS, three were found to be below the effective 25OHD levels. Vitamin D serum results are shown in Table 3.

The levels of potential interference owing to epimers and isobars are shown in Table 4. Clearly, if not chromatographically separated, these interferents could make a major contribution (in the range of 14 to 55%) to the total measured level of Vitamin D. The comparison between levels in Tables 3 and 4 reveals that where epimers and isobars are not accounted for, the measurements of Vitamin D levels are inflated to appear as normal. In these cases a diagnosis of deficiency may be missed as illustrated in Figure 5. The percentage of interfering epimer and isobars in serum, are detailed in Table 4. The isobar (1αOHD3) was not detected in any samples.

These results confirm the complexities involved in measuring Vitamin D status and reinforce the need for the uniform adoption of improved accurate assays. Measurements of 25OHD in human serum using competitive immunoassays are difficult owing to lipophilicity/tight binding to the Vitamin D-binding protein (DBP), even at very low serum concentrations 15. In addition, immunoassays for 25OHD have been reported to cross-react with 24,25(OH)₂D3, an intermediate product during the formation of 1,25(OH)₂D3 in the kidney 1617 (Figure 1). Many commercial immunoassays can only measure 25OHD3 and are not suitable to monitor supplementation with Vitamin D2, which is derived from plant sources and widely used in many countries for fortification of foods 18. Hence, LC-MS/MS analysis is preferred but it has been shown that it is also subjected to interference issues 19202122232425262728. In our new method, the use of an electrospray ionisation technique has overcome the problems associated with in-source transformation of 25OHD3-sulfate and other metabolites to 25OHD3 which may occur when using atmospheric pressure chemical ionisation mode 29.

The use of tandem column technology, an optimised mobile phase composition and a modified extraction method not only separate the epimers and isobars but also eliminates the interferences caused by early eluting salts (e.g. sodium) and the late eluting phospholipids, which may interfere with analyte ionisation 293031323334. Moreover, the installation of a low dispersion inlet filter together with a chiral guard column also minimises the interferences resulting from early eluting amino acids and late eluting xenobiotics. The column also separates interfering compounds present at higher concentration which have been shown to overlap or share exact mass transitions with 25OHD e.g. detergents and phthalates etc 293031.

The separation of epimers and isobars from the target analyte is critical because they can overlap 25OHD peaks and form the same masses upon ionisation, thus compromising the true status of 25OHD in circulation 383940414243. The epimer of 25OHD3 is known to have the same effects on suppressing parathyroid hormone (PTH) secretion, but it has negligible calcium-producing effects 4041. Recently, it was concluded that the absence of external standardisation for the 25OHD assay might lead to greater variations and false results 33. To minimise the inter-laboratory and inter-method variations in our LC-MS/MS analysis, we have used reference materials for preparation of our calibrants and quality controls, along with the introduction of a new internal standard. The latter was required as, when using the internal standard hexadeuterated-25OHD3, certain fragment transitions of the parent ions could lead to greater isobaric interferences such as the transition 407.7 > 389.7 37. Stanozolol-D3 as a new internal standard prevents isobaric interferences by using the transition 332.2 > 81.2.

To date, only one transition of the parent ion to product ion has been investigated, instead of our three-precursor ions to fragment ions transitions (according to the qualifier-qualifier principle) 29. To our knowledge, this is first time that three transitions of precursor ions to product ions have been used for quantification of 25OHD3, 25OHD2, 3-epi-25OHD3, 1αOHD3 and 7αC4 as shown in Table 2. The sum of the three transitions not only increases the sensitivity of the assay but also minimises isobaric interferences.

To summarise, this is the first LC-MS/MS method for the determination of 25OHD which includes separation and quantification of epimers and isobars. It is a fast liquid-liquid extraction method, which does not require complicated derivatisation procedures, hence reducing assay time and variability. This method can quantify the different 25OHD forms accurately, which is difficult to achieve using immunoassay methods. This methodology can be used for accurate blood testing to prevent falsely elevated Vitamin D levels being reported.

The LC-MS/MS method was free from all types of interferences arising due to epimeric, isobaric, instrumental and matrix components, which may interfere with analyte ionisation. Thus the LC-MS/MS method provides a robust, specific, reliable and reproducible technique as a solution to the problems identified in relation to current assays for Vitamin D. Removal of uncertainties in Vitamin D measurement, is required in order to progress the current understanding of the roles of Vitamin D in health and disease via more rigorous clinical trials 7.

The authors declare that they have no competing interests.

AP and DPN initiated the study. The method development was conducted by IS with contributions from DPN and JB. RJ provided samples for analysis. All authors contributed to the study design, preparation of the manuscript and have read and approved the final version.