Challenges in describing vitamin D status and activity / Herausforderungen bei der Bestimmung des Vitamin D-Status

Introduction

The formation of vitamin D₃ (cholecalciferol) from its endogenously formed precursor 7-dehydrocholesterol in the skin under irradiation of UV light is a pure photochemical and non-enzymatic process; in many regions of the world – also including the whole of Europe – the intensity of UV light from the sun is insufficient to form vitamin D during several months of the year. Because the contribution of food to the vitamin D pool of the body is low for most individuals, cholecalciferol becomes a conditional vitamin under these conditions [1–6].

The obvious link between exposure to sunlight and calcium handling of the body probably reaches far back to the evolutionary development of fauna outside of the sea with limited availability of calcium; a primarily regulatory role seems likely. However, no conclusive theory explains this link at present. It is surprising that during evolution no enzymatic system emerged to achieve this rather simple reaction step in sterol metabolism, the conversion of 7-dehydrocholesterol to cholecalciferol.

In the past decade, “the vitamin D story” has gained enormous attention worldwide, both among physicians and the general population. Interestingly, this seems to be mainly triggered by potential non-calcemic effects of relative vitamin D deficiency instead of the well-recognized musculoskeletal effects of hypovitaminosis D. Efficiency of vitamin D supplementation is established for the prevention of rickets during the first year of life and for the prevention of falls and fractures in the elderly. However, the effects in the elderly are small to moderate [7–12]. No conclusive and robust data demonstrate protective effects of vitamin D supplementation regarding cardiovascular diseases, diabetes or malignancies [2, 13]. Association between “sub-optimum” vitamin D status and prevalence of many diseases seems to be given; however, causality is uncertain even after these years of intensified research. Inverse causality can also account for these associative observations: for example, patients with major depression typically have little outdoor activity and probably as a consequence a higher risk for poor vitamin D status. Consequently, depression is associated with 25-OHD levels in these patients; however, supplementation of vitamin D in these patients did not have any effect on the course of the disease [14]. Ongoing large randomized controlled studies should bring about relevant new insights concerning the association of vitamin D with health status beyond the bone (e.g., the VITAL study, USA; the VIDAL study, UK; [15]).

Because the predominant and typical clinical manifestation of hypovitaminosis D – impaired mineralization of the bone in rickets and osteomalacia – is not easily described by clinical findings or radiological studies, at present laboratory tests have, without question, a predominant role in characterizing vitamin D status. 25-Hydroxy vitamin D (25-OHD) is by far the most widely used marker in this context, with the number of requests for measurement increasing exponentially in many laboratories. The substantial costs finally associated with this trend at a population level will require prudent utilization management [16, 17]. Consideration of supplementation of vitamin D (and/or intentional sun exposure) is extremely complex – at both a population and individual level – and is not within the scope of this article. Because supplementation of vitamin D in defined populations is inexpensive (in contrast to measuring serum 25-OHD) and probably very safe, diagnostic criteria for hypovitaminosis D cannot be discussed without incorporating therapeutic aspects. Nevertheless, a prerequisite for such discussion – at a population and individual level – is a critical appraisal of the available laboratory tools to describe vitamin D status. Indeed, this is a difficult and complex undertaking for several reasons, which are not at all restricted to issues with serum 25-OHD quantification. In the following sections, we address some important open questions.

Is total serum 25-OHD a valid and reliable marker of individual vitamin D status?

Because cholecalciferol (vitamin D₃) has a short biological half-life, the biochemical characterization of vitamin D status is based on the assessment of surrogate markers (both structurally related and unrelated molecules). Quantification of 1,25-dihydroxyvitamin D in serum, the “D hormone”, as the bio-active end-metabolite of vitamin D seems logical to be quantified in this context; however, it is well recognized that even increased serum concentrations of this analyte can be observed in hypovitaminosis D [18], probably due to secondary hyperparathyroidism.

The idea behind measuring the bio-inactive intermediate metabolite 25-OHD in serum for describing individual vitamin D status is that circulating total (protein-bound) 25-OHD might be in equilibrium with and might reflect whole body pool size of vitamin D and metabolites that are available for further metabolization to the D hormone by 1α-hydroxylase in different tissues. This must be recognized as a hypothesis. If body stores are below certain (and potentially organ-specific) thresholds, there is an insufficient supply assumed for the adequate formation of the bio-active messenger molecule 1,25-dihydroxyvitamin D. The liver is believed to be the most important store of vitamin D, and the main consequence of insufficient formation of 1,25-dihydroxyvitamin D by the kidney is probably reduced calcium absorption by the gut. Notably, this function, however, is not entirely dependent on the regulation of 1,25-dihydroxyvitamin D [2]. Given sufficient supply of precursors for the functionally required and tightly regulated formation of 1,25-dihydroxyvitamin D, biological effects of inflating vitamin stores are not to be expected. This might correspond to vitamin A: deficiency causes xerophthalmia and blindness, representing a huge problem in many populations of the world; however, with sufficient vitamin A supply – as typically given in industrialized countries – supplementation of vitamin A has no effect on visual performance.

There is without doubt an association between very low serum 25-OHD concentrations (<10 ng/mL) and symptomatic hypovitaminosis D, clinically diagnosed as rickets and osteomalacia with bone pain, muscle weakness and typical radiological findings. However, observations in children with rickets show that also seemingly normal 25-OHD serum concentrations can be associated with rickets in the case of low calcium intake [19, 20]. By contrast, obviously not all individuals with serum 25-OHD concentrations <10 ng/mL suffer from symptoms or have abnormal bone mineralization [21]. Thus, it must be assumed that functional vitamin D status is also determined by individual calcium uptake, but also potentially by further variables such as renal calcium excretion, which is dependent on dietary variables such as protein intake [22].

Circulating 25-OHD is tightly bound to vitamin D binding protein (VDBP). Genetic polymorphism is well recognized for this protein, and total 25-OHD serum concentrations have been found to depend on respective genotypes [23]. This suggests that these genotypes determine the bioavailability of vitamin D and may put in question the assumption that total serum 25-OHD equally and conclusively reflects the body stores of vitamin D in a population.

In addition, the downstream processes of the vitamin D endocrine system seem to be subject of genetic variability: different isoforms of 1α-hydroxylase (converting 25-OHD to 1,25-dihydroxyvitamin D) are recognized as well as polymorphism of vitamin D receptor [24–31]. The performance of metabolic functions of the kidneys regarding 1α-hydroxylase is well recognized to be variable with impaired integrity of the organ in chronic diseases.

Taken together, it seems likely that depending on these variables the individual threshold to functional insufficiently of precursor molecules of the D hormone system is variable between individuals. These considerations, in turn, raise concerns if it is reasonable to base recommendations for vitamin D supplementation merely on distinct 25-OHD serum concentration thresholds, as currently advocated [2, 3].

Issues in measuring 25-hydroxyvitamin D

During the past years it has been widely recognized that 25-OHD – the predominantly addressed analyte in the characterization of vitamin D status – is not “an easy analyte”, although circulating in rather high concentrations in the blood [32]. This is, in particular, related to the specificity of tests towards certain compounds, matrix effects, the standardization of commercial tests, and consequently the agreement between tests.

Ligand binding assays for the quantification of 25-OHD are based either on VDBP or antibodies for analyte binding. Owing to the limited availability of epitopes, a competitive test design is used, with labeled 25-OHD competing with the analyte from the sample for a limited number of binding sites in the test system. Those competitive ligand binding assays are known to be, in general, far more prone to matrix effects compared with sandwich immunoassays.

Non-automated ligand binding assays used in the 1990s were rather straightforward: tests of this generation involve a separate protein precipitation step, for example, using acetonitrile which achieves complete release of the analyte from its tight protein bonds [23]. This release seems to be the “Achilles’ heel” in automated assays which became available from approximately 2000 onwards. Obviously, it is difficult to find releasing reagents which are efficient and, by contrast, compatible with test antibodies finally present in a shared test mixture. It must be assumed that in those automated tests a “proportional” rather than a complete release is achieved, which may show considerable between-sample variation.

Around one decade ago, clinical reports [33] and questionable data from an epidemiological survey [34] disclosed substantial problems of standardization and lot-to-lot consistency of 25-OHD assays. These observations stimulated the development of isotope dilution mass spectrometry methods [35]. Indeed, 25-OHD is – along with immunosuppressant drugs – one of the analytes which are nowadays rather widely addressed by the innovative technology of liquid chromatography-tandem mass spectrometry (LC-MS/MS). The analyte is efficiently ionized without derivatization and is present in a concentration range above 5 nmol/L that does not represent a big analytical challenge – in contrast to 1,25-OHD, circulating in the lower ng/L range. A substantial number of different realizations have been described, regarding the mode of ionization (ESI, APCI), mass transitions, internal standard compounds, sample preparation (involving in some assays derivatization and often online solid phase extraction), and chromatography. An overview of this heterogeneous background on 25-OHD quantification by LC-MS/MS is given in some recent reviews [36, 37]. Besides individual laboratory developed assays, commercially available kit solutions are more and more widely used. In addition to using kits (including internal standard compounds, calibrator and QC samples, columns, mobile phases), installations are highly heterogeneous because many different MS/MS and LC configurations are applied. The availability and use of commercialized calibration materials has substantially contributed to the harmonization of quantification of serum 25-OHD by LC-MS/MS [38]. In particular, the availability of reference materials (by the US National Institute of Standardization, NIST) and reference procedures was important in this context [39–41]. Ligand binding tests for 25-OHD have been compared with LC-MS/MS methods in many studies [42]. Indeed, at least in the European market regulated by the IVD directive 98/79/EC, it is a legislative requirement for the diagnostic industry to trace back 25-OHD ligand binding assays to LC-MS/MS as a method of higher metrological order. Two dedicated reference methods based on LC-MS/MS have been published and introduced so far [39, 41]; far higher analytical quality is required for such methods compared with routine tests [43], according to their position within the chain of traceability. Commutability of reference materials regarding ligand binding assays has been realized as a challenge.

Aided by the widespread availability of LC-MS/MS, overall relevant progress in the harmonization of 25-OHD testing has been achieved during the past years [44]. The degree of standardization and harmonization of 25-OHD measurement is displayed very comprehensively and is up-to-date by the UK proficiency testing scheme (DEQAS; www.deqas.org). Differentiation of concentrations below and above 20 ng/mL (50 nmol/L) – a concentration which is commonly supposed to be of particular diagnostic importance – is now achieved by the majority of participants. However, all-method coefficient of variances (CVs) of above 20% can be observed for certain samples. Approximately 12% of participating laboratories use LC-MS/MS at present; method specific CVs for this subgroup are typically above 10%, which demonstrates that further improvement in analytical performance is required.

Despite acceptable standardization of calibration, the agreement between automated ligand binding tests with LC-MS/MS is often found rather poor in controlled studies involving clinical samples, with r values <0.9 and intercepts deviating significantly from 1.0 in most studies [42, 45]. This may be due to individual sample matrix effects, which are reliably compensated for by the isotope dilution technique in LC-MS/MS, in contrast to immunoassays. Systematic as well as individual disagreement between different analytical methods for 25-OHD may be related to differential “co-quantification” or cross-reactivity with several related compounds. This includes 25-OHD₂, which is almost exclusively derived from supplements. LC-MS/MS assays specifically quantify 25-OHD₃ and D₂, whereas most – but not all – ligand binding assays detect both compounds almost equally – representing tests for total 25-OHD. Because vitamin D₂ (ergocalciferol) cannot be assumed [46] to be bio-equivalent to naturally occurring cholecalciferol, the co-quantification of both 25-hydroxy-metabolites is not reliable from a diagnostic point of view. Whereas in Europe the use of vitamin D₂ is rather uncommon, it is the only compound which is approved for the treatment of symptomatic hypovitaminosis D in the USA. From a pragmatic point of view, therefore, total 25-OHD assays can be justified.

More recently, two metabolites have been recognized to potentially cross-react with 25-hydroxyvitamin D assays: 3-epi-25-hydroxyvitamin D₃ and 24,25-hydroxyvitamin D₃. 3-Epi-25-hydroxyvitamin D₃ can make up to approximately 10% of total 25-OHD₃ during the first year of life. In adults, the proportion is typically far smaller [47–50]. Activity and biological relevance of 3-epimer is still unclear. Using LC-MS/MS, particular selective chromatographic conditions are required to achieve resolution of the 3-epimer, whereas – surprisingly – many ligand binding assays seem to be isomer-specific.

24,25-Dihydroxyvitamin D₃ is a product of inactivation and, in particular, is present in highly replete vitamin D states. Cross-reactivity of ligand binding assays reported by the assay manufacturers is up to >100%, a fact still hardly addressed by the DEQAS proficiency testing scheme. Taking into consideration that serum concentrations of 24,25-dihydroxyvitamin D₃ are approximately 10% of the concentrations of 25-OHD₃, interference of immunoassays by this metabolite are of concern and may explain some of the scatter seen in method comparison between the immunoassay and LC-MS/MS [51].

If the systematic (calibration-related) bias of 25-OHD results in the range of 10% is superimposed with random imprecision and bias caused by differential patterns of co-detected metabolites and differential degree of analyte release from its protein bonds, a substantial uncertainty of routine tests has to be expected when two different analytical methods are applied to individual samples. Consequently, discrepant classification will be observed rather frequently if dichotomous diagnostic cut-off values are applied. In most cases, laboratory results represent one piece in a diagnostic mosaic together with a number of anamnestic, clinical and technical findings. However, when dichotomous cut-off values are applied to determine specific therapeutic measures, the performance of analytical methods is extremely challenged, regarding standardization, reproducibility and accuracy. This is excellently exemplified for HbA_1c in diagnosing and monitoring of diabetes mellitus [52]. Similarly, decision-making regarding vitamin D supplementation based merely on 25-OHD results above or below 20 ng/mL (50 nmol/L) – as widely suggested – is hardly paralleled by any decision in endocrinology and a huge challenge for analytical quality. It must be recognized that the level of analytical reliability of 25-OHD testing is still far from that realized for HbA_1c measurement, with long-term CVs of <2%.

The choice of analytical method for 25-OHD measurement in individual laboratories – different automated and non-automated tests, HPLC-UV, or LC-MS/MS – has to include many aspects and is often a matter of substantial debate. LC-MS/MS is far more demanding for laboratory technicians to implement and to maintain compared with uniform automated ligand binding tests. Investments for LC-MS/MS instruments are rather substantial (>€200,000) and not covered by in vitro diagnostic companies, as applies for convenient and easy-to-handle automated ligand binding tests. By contrast, the running costs per sample are moderate in LC-MS/MS because only simple substances are required. No reagent lot-to-lot shifts are expected in contrast to immunoassays where the production, for example, of new antibody lots can be a problem. LC-MS/MS is not interfered by anti-reagent antibodies, which represent a pitfall in all ligand binding assays. Because the analyte detection of LC-MS/MS is based on the molecular mass of the analyte and its molecular disintegration behavior and because individual matrix effects are corrected for by the isotope dilution technique, this technique is indeed likely to be of superior reliability compared with ligand binding assays for 25-OHD measurement. However, also pitfalls of LC-MS/MS tests have to be recognized [53]. Besides fundamental analytical issues such as deuterium-hydrogen scrambling of internal standards, this also includes incorrect calibration [54, 55] and gross errors (such as sample identification mistakes) because LC-MS/MS assays are only partially automated so far. Regarding patients’ outcome there are no data available which would indeed favor the use of one certain analytical platform for 25-OHD measurement.

Are several markers necessary to characterize the vitamin D status?

Given the substantial concerns about analytical quality, diagnostic reliability and biological validity of the mainstream vitamin D marker 25-OHD, it should be thoroughly discussed and studied if other available laboratory parameters – alone or in synopsis – are clinically superior in characterizing vitamin D status. In this context, quantification of plasma parathyroid hormone (PTH) is of particular interest. The application of this parameter is limited in a physician’s office setting by the requirement to keep the sample material chilled until analysis. Different target epitopes in different generations of PTH assays and variable cross-reactions with truncated molecules are analytical issues, also leading, in part, to assay-specific reference ranges.

PTH can be looked upon as a functional marker of vitamin D supply, because an increase indicates reduced availability of calcium in the circulation. Secondary hyperparathyroidism is both an indicator of hypovitaminosis D and an effector of this condition by mobilizing calcium from the bone. PTH is assessed to define hypovitaminosis D: responsiveness of PTH (by decreasing) in response to vitamin D application is an indicator of functional vitamin D deficiency. It is recognized, however, that the threshold of serum 25-OHD regarding PTH-responsiveness is rather variable [56], underpinning the concerns about the diagnostic utility of 25-OHD.

Because hypovitaminosis D is not the only cause of secondary hyperparathyroidism, for example, malabsorption has to be considered in increased results. Measurement of PTH may have the strength to give information on both vitamin D supply and nutritional calcium intake. Indeed, the characterization of dietary calcium supply by laboratory tests is of great interest: according to epidemiological data, calcium intake is critically low in many populations [3], and assessing individual calcium intake using food balances, food questionnaires and tables is cumbersome. Moreover, factors such as phytate content of the diet impacts calcium absorption, whereas high protein consumption increases renal elimination [57]. There is, by contrast, growing evidence that over-supplementation with calcium may be a relevant problem regarding cardiovascular disease, with calcium having a far smaller relative range of tolerable daily intake compared with vitamin D [3]. Measurement of plasma PTH in an osteological setting further incorporates screening for primary hyperparathyroidism, which has a considerable prevalence in elderly individuals. In turn, interpretation of PTH results regarding primary hyperparathyroidism requires excluding vitamin D deficiency. Typically, of course, secondary hyperparathyroidism due to hypovitaminosis D and primary hyperparathyroidism are differentiated by serum calcium concentrations, which are high in the latter case.

It should not be overlooked that the concept of functional surrogate markers for micronutrients is very powerful – that is, in diagnosing cobalamin deficiency, where homocysteine and methyl malonic acid are recognized as offering far higher diagnostic reliability compared with serum cobalamin or holotranscobalamin measurements.

Low serum calcium and inorganic phosphate is typically reported for cases of severe vitamin D deficiency, but these analytes are not widely proposed for a screening for hypovitaminosis D. Serum calcium can be looked upon as the terminal functional marker and may potentially allow a synoptic assessment of vitamin D and calcium status. The costs for calcium measurement are substantially lower compared with 25-OHD and/or PTH measurement. Calcium measurement is rather precise and well standardized, and intra-individual biological variation is low. Measurement of bio-active free ionized calcium is looked upon as the gold standard, but sample pH is critical and shipment from a doctor’s office is not possible. Correction of serum calcium for total protein or albumin is easy to perform and seems to increase the usefulness of the test. It seems warranted to thoroughly study the potential value of serum calcium measurement in the assessment of vitamin D and nutritional calcium status – also in response to vitamin D supplementation.

Alkaline phosphatase is a further classical and low cost marker of functional hypovitaminosis D (as long as hepatobiliary issues can be excluded). A potentially useful contribution of more recent bone turnover markers is not conclusively addressed by studies so far. In general, however, use of these novel markers (such as CTx and osteocalcin) is hampered by substantial intra-individual variation, also requiring very exact standardization of sampling schedules.

A “multi-marker approach” to the characterization of vitamin D status could potentially involve – besides 25-OHD – other metabolites of vitamin D (such as 24,25-dihydroxyvitamin D which seems to show a highly dynamic increase in the case of fully replete vitamin D status), PTH, serum calcium and phosphate, serum-calcium-phosphate product, urine calcium and selected bone markers in the future. Innovative biomathematical methods of pattern recognition may potentially be useful in this context – particularly to predict the likeliness of significant clinical benefit from vitamin D supplementation. Also, more complex metabolomic platforms might contribute to the description of vitamin D status in a further perspective. The goal of innovative approaches should be the characterization of both vitamin D and calcium status, aiming to increase the performance of laboratory testing in a clinical decision-making context. Initiation of long-term vitamin D supplementation is the most relevant therapeutic intervention, and the incidence of falls and fractures is the relevant clinical endpoint to be assessed for hypovitaminosis D – because effects beyond this are still hypothetical. Bone histology is also considered to be the most reliable indicator of bone health related to vitamin D status, but rarely available. Priemel et al. correlated bone histology of individuals who died suddenly (predominantly from accidents) with serum 25-OHD of the dead bodies [21]. In the majority of individuals with low 25-OHD, normal histology was observed; in these cases, supplementation with vitamin D would not have been useful from an osteological point of view. Unfortunately, PTH and calcium could not be studied in this investigation. Thus, the predictive power of 25-OHD regarding bone health seems to be poor, and innovative approaches should aim to substantially improve the performance of laboratory testing.

How can the long-term vitamin D status be assessed?

Because bone mineralization and remodeling – as the main field of vitamin D action – is a rather slow process, in particular, long-term vitamin D status is of interest [58]. In many – if not most – regions of the world, vitamin D status is highly variable depending on the season at a population level. In Germany, approximately a 1.6-fold higher median of 25-OHD was reported for June compared with March [59]. It is assumed that vitamin D or metabolites have a half-life of approximately 60 days in the body, explaining low levels, particularly, at the end of the winter season. Owing to the shallow position of the sun in winter, no photochemical conversion of cholecalciferol to 25-OHD is maintained.

Mankind had to cope with this phenomenon for several 1000 years in all regions of the world, with a certain distance to the equatorial areas. Thus, an annual sequence of a certain time period of low vitamin D status with a period of fully replete vitamin D status seems to be a quasi-physiological condition. The duration and extent of poor vitamin D status is probably dependent on how efficiently vitamin D stores of the body are filled during the summer months. This again is determined by latitude, skin pigmentation, lifestyle and clothing habits, and therefore is highly variable [60]. Additionally, a high consumption of meat can contribute to vitamin D status due to its 25-OHD content.

It seems evident that a limited period of low 25-OHD serum concentrations with elevated PTH over 2–3 months in a person with relevant outdoor activity from spring to autumn has a different impact on health compared with poor vitamin D status throughout the year in an institutionalized elderly individual with very limited outdoor activity or in an infant. However, a single-point laboratory assessment of vitamin D status, for example, in March may observe similar 25-OHD concentrations and has little diagnostic power to discriminate these two conditions (Figure 1).

Figure 1

Typical long-term courses of serum 25-hydroxyvitamin D.

Line A may represent a person with a continuously high degree of sun exposure throughout the year as found, for example, in equatorial regions. Line B may represent a person in Central Europe with substantial outdoor activity in summer and vitamin D deficiency during winter. Line C represents a person with sustained vitamin D deficiency throughout the year, for example, in a chronically disabled elderly person. Note that a single measurement of 25-hydroxyvitamin D in winter may not be appropriate to display the fundamental differences in vitamin D status of individual B and C, respectively (arrow).

It is evident that – whatever analytes are involved – a comprehensive description of individual long-term vitamin D status has to include several observation points over the year. Respective models might assess an “area-under-the-year-curve” of defined analytes, or assess the duration of deficiency in months per year. Obviously, describing long-term vitamin D status has to make a compromise between practicability and comprehensiveness.

Recently, a mathematical approach has been suggested to predict average 25-OHD serum concentrations over the whole year from a single, random sampling [61]. Just considering the different amplitudes of UV light irradiation at different latitudes, however, makes this approach very questionable.

Conclusions

Empirically, low serum 25-OHD in combination with increased alkaline phosphatase and low calcium seems to be a useful set of markers to verify candid vitamin D deficiency in a standard clinical setting. Regarding questionable milder forms of hypovitaminosis D (“insufficiency”) also addressed in epidemiological research, the diagnostic validity of single serum 25-OHD measurement seems to be very questionable.

The current debate on potential health effects of vitamin D status at a population level and on the policy of supplementation (with or without individual laboratory testing) is still far from obtaining consensus. In this context, it must be recognized that there are many open questions concerning the reliability and biological appropriateness of the currently applied approach to characterize vitamin D status almost exclusively based on serum 25-OHD measurement.

The enormous interest in vitamin D emerging during recent years (including aggressive direct-to-patient marketing for testing of the “sunshine vitamin”) is widely perceived as hype. Nevertheless, poor bone quality, falls and fractures represent a huge and growing burden of disease worldwide, and it is reasonable to assume that both long-term vitamin D and calcium status is of relevance in this context. Consequently, improvement of laboratory diagnostic tools to characterize this system is indeed of utmost importance.

Conflict of interest statement

Authors’ conflict of interest disclosure: The authors stated that there are no conflicts of interest regarding the publication of this article.

Research funding: None declared.

Employment or leadership: None declared.

Honorarium: None declared.