Quantitation of Total Vanadium in Rodent Plasma and Urine by Inductively Coupled Plasma – Mass Spectrometry (ICP-MS)

Abstract

Human exposure to vanadium (V) is anticipated because it is a drinking water contaminant. Due to limited data on soluble V salts, the National Toxicology Program is investigating the toxicity in rodents following drinking water exposure. Measurement of internal V dose allows for interpretation of toxicology data. The objective of this study was to develop and validate an inductively coupled plasma-mass spectrometric method to quantitate total V in rat plasma. The method was linear (r ≥ 0.99) from 5.00 – 1,000 ng V/mL. Intra- and inter-day relative error (% RE) and relative standard deviation (% RSD) of spiked plasma samples were 8.5% – 15.6% RE and ≤ 1.8% RSD and 7.3% – 11.7% RE and ≤ 3.1% RSD, respectively. The limit of detection was 0.268 ng V/mL plasma and absolute percent recovery was 113%. Standards up to 7,500 ng V/mL plasma were diluted into the validated range (5.6% RE, 0.9% RSD). V in extracted plasma samples over 15 days at ambient and refrigerated conditions was from 97.7 – 126% of day 0. Determined plasma V concentrations after three freeze-thaw cycles and after frozen storage for up to 63 days ranged from 100 – 106% and 100 – 122% of day 0, respectively. The method was extended to rat urine (accuracy and precision −2.0 – 0.3% RE and <0.6% RSD, respectively for same linear range). These data demonstrate that the method is suitable to quantitate V in rat plasma and urine.

Introduction

Vanadium (V) has been proposed to play a role in essential biological processes but can be toxic depending on its concentration and chemical form (Frausto da Silva and Williams 2001, Llobet and Domingo 1984). V from naturally occurring and anthropogenic sources is found in the environment in several chemical forms, bound to a range of metal binding agents, and over broad concentration ranges in fresh water, ocean water, and soil (Agency for Toxic Substances and Disease Registry 2012). The element is released into the environment from metallurgical operations, from carbon-based fuel consumption, and from disturbed mineralogical formations. There is evidence that V is also released from fly ash waste disposal activities and accidental releases from holding ponds (Hesterberg et al. 2015; Ramsey et al. 2019; Ruhl et al. 2009). Concentrations of V in the fly ash released in two prominent fly ash spills that occurred in Eden, NC in 2014 and Kingston, TN in 2008 were 40 ± 12 mg/kg (Eden) and 76.7 ± 30.6 mg/kg (Kingston), and surface water concentrations were measured from 5 to 67 ng/mL in the Dan River (in Eden). A recent study found that a significant quantity of water soluble V wasin sediment samples from the Emory River in Kingston, suggesting that environmental impacts may be observed even many years after the events as a result of disturbing the materials in sediment.

Vanadium exhibits complex speciation and can undergo conversions through several chemical forms and at least two oxidation states: vanadyl (V⁴⁺) and vanadate (V⁵⁺) (Al-Kharafi and Badawy 1997; Langmuir et al. 2004). Previous reports have indicated that V⁴⁺ salts have lower toxicity than V⁵⁺ salts (Bevan et al. 1995). However, potential toxicities from the ingestion of V⁴⁺ and V⁵⁺ salts have not been thoroughly investigated, and current reference dose values are based on limited data in rodents (United States Environmental Protection Agency 2002; Llobet and Domingo 1984). Because limited data exist regarding V salts, the National Toxicology Program is investigating the toxicity following drinking water exposure in rodents. Higher toxicity was observed following exposure of rodents to sodium metavanadate (V⁵⁺) compared with vanadyl sulfate (V⁴⁺) for 14 days via drinking water (Roberts et al. 2016).

An assessment of internal dose of vanadium species following exposure to V⁴⁺ and V⁵⁺ salts can be used to interpret toxicology data. We investigated the oxidation state of V salts in rodent plasma after dosing with drinking water and found that all V was present in the V⁴⁺ oxidation state (Harrington et al. 2021). Hence, the analysis of total V is a practical, cost-effective alternative to estimate the internal dose of V following exposure to V⁴⁺ and V⁵⁺ salts.

Previous reports have described analytical methods and approaches designed to determine endogenous V concentrations in biofluids, most commonly plasma, blood, and serum, and environmental samples. For example, one report described a study with a similar analytical approach and a less rigorous digestion involving simple dilution of serum (Nixon et al. 2002). Other studies have described analyses of serum for V, sometimes in a panel of elements, by inductively coupled plasma-optical emission spectroscopy, which has lower sensitivity than inductively coupled plasma-mass spectrometry (ICP-MS), or by neutron activation analysis (NAA), which requires the use of specialized reactors to produce neutron sources with sufficient sensitivity to analyze V (Byrne and Versieck 1990; Pyy et al. 1984; Simonoff et al. 1984; Uchida et al. 1981). To the best of our knowledge, there has not been any previous report of a full validation of an analytical method for determination of total V in rodent plasma and urine.

The purpose of this investigation was to validate an ICP-MS method for measuring total V in rat and mouse plasma and urine to support toxicology studies. The results demonstrate that the analytical method is accurate, precise, and robust in both validated matrices.

Materials and Methods

Reagents

Stock solutions of National Institute of Standards and Technology (NIST)–traceable V (10 and 1,000 μg/mL) were purchased from two sources (High-Purity Standards, Charleston, SC and Inorganic Ventures, Christiansburg, VA). NIST-traceable 1,000 μg/mL solutions of praseodymium (Pr) and indium (In) solutions were purchased from High-Purity Standards. Pooled male Sprague Dawley rat plasma isolated from blood collected using potassium ethylenediaminetetraacetic acid (K₃EDTA) anticoagulant and urine were purchased from BioIVT (Westbury, NY). High-purity, 30% (v/v) non-stabilized hydrogen peroxide (H₂O₂, Suprapur grade) was purchased from EMD Millipore (Billerica, MA) and concentrated nitric acid (HNO₃, Optima grade) was obtained from Thermo Fisher Scientific (Fair Lawn, NJ). Deionized water (DI H₂O, 18 MΩ cm⁻¹) was produced using a purification system from Pure Water Solutions (Castle Rock, CO). Optima-grade hydrochloric acid (HCl) was purchased from Fisher Scientific (Waltham, MA).

Method validation—experimental design

The method was validated in male Sprague Dawley rat plasma for linearity, precision, accuracy, percent recovery, selectivity, analytical limits, and dilution verification using matrix calibration standards and quality control (QC) samples. To the best of our knowledge, there are only a few commercially available certified reference materials for V in plasma, and none in Sprague Dawley rat plasma, so to provide the most accurate measurement of method performance in the primary sample matrix, control Sprague Dawley rat plasma fortified with a stock V standard were used in the current validation as QC samples. The validation range was 5.00 to 1,000 ng V/mL plasma, but additional standards were prepared from 2.50 and 2,500 ng V/mL plasma to challenge the lower and upper limits of quantitation, respectively. Solvent calibration standards, which contained all reagents, but no matrix, were prepared at the same concentrations as the plasma matrix standards to assess the percent recovery and matrix effect.

Linearity was demonstrated with an eight-point matrix calibration curve. The accuracy was expressed as percent relative error (%RE) relative to nominal concentrations. Sensitivity and selectivity were evaluated by preparing replicate matrix standards (n = 6) at two concentrations (i.e., 2.50 and 5.00 ng V/mL plasma) and six matrix blank replicates, respectively. The limit of quantitation (LOQ) was defined to be the lowest matrix standard which was quantitated accurately within 20% of the nominal value with a relative standard deviation (RSD) within 20%. The limit of detection (LOD) was defined to be three times the standard deviation of replicate measurements of the measured LOQ matrix standard expressed as a concentration (Lappas and Lappas 2016). Selectivity was evaluated by comparing instrument response from matrix blanks against response from the matrix LOQ samples. Intra-day precision and accuracy were evaluated with QC samples prepared on the same day at three concentrations. Inter-day precision and accuracy were evaluated at three levels using multiple sets of QC samples analyzed over three days. To evaluate instrument drift during analysis, plasma matrix calibration standards were analyzed at the beginning and end of the first validation experiment. The recovery was also evaluated by comparing the response of matrix standards against responses for corresponding solvent standards prepared at the same level. Dilution verification matrix samples (n = 3) were prepared to demonstrate that rat plasma containing a V level (7,500 ng V/mL plasma) exceeding the validated range (5.00–1,000 ng V/mL of plasma) were successfully analyzed. These verification matrix samples were diluted tenfold with digested matrix blank and analyzed.

Stock solutions

Prior to use, all labware were soaked in 1% (v/v) HNO₃ for a minimum of 12 hours, rinsed repeatedly with deionized H₂O, and allowed to dry in a clean, high-efficiency particulate air (HEPA)–filtered hood to mitigate the risk of environmental metal contamination. Intermediate stock solutions of internal standard (1,000 ng Pr/mL) and V (100 and 1,000 ng/mL) were prepared in 5% (v/v) aqueous HNO₃ from commercial sources.

Matrix standard and blank preparation

Matrix calibration standards and QC samples were prepared with different final volumes: 50 mL for calibration standards and 10 mL for QC samples. Plasma matrix standards were prepared by adding control plasma (0.500 mL) to 50-mL, graduated plastic tubes and fortifying over the concentration range using intermediate stocks. Matrix and reagent blanks were prepared by transferring 0.100 mL of pooled control plasma or 0.100 mL of deionized H₂O, respectively, to 15-mL plastic tubes. Plasma QC samples were prepared by aliquoting 0.100 mL of pooled control plasma to 15-mL tubes and fortifying with V to contain low, medium, high, and dilution verification concentrations (7.50, 250, 800, and 7,500 ng V/mL plasma, respectively). To assess matrix effects, 0.100 mL plasma were digested as described below prior to addition of V at the same concentration as matrix calibration standards and comparing the calculated concentrations with those of the matrix calibration standards.

To each sample, HNO₃ (2.50 mL for calibration standards and 0.500 mL for matrix QC samples) was added, the digestion tubes were lightly capped and 30 minutes of pre-digestion time were performed at ambient temperature in a clean plastic hood. All samples were digested in a DigiPREP graphite heating block from SCP Science (Champlain, NY) at 65°C for 60 minutes and allowed to cool. An aliquot of 30% H₂O₂ was added to each sample (1.25 mL for calibration standards and 0.250 mL for QC samples) and returned to the heating block at 65°C. The temperature was ramped to 95°C in 60 minutes and held for an additional 120 minutes. The samples were cooled to room temperature in a clean plastic hood, intermediate internal standard stock solution (0.500 mL for calibration standards and 0.100 mL for QC samples) was added, and all were diluted to final volume with deionized H₂O, capped, and vortex mixed. A diagram of the sample preparation is shown in Figure S1 (Supplemental Information).

Stability of vanadium in plasma

Stability of vanadium in matrix was evaluated under a range of conditions intended to mimic short- and long-term storage of V in plasma and sample extracts during analysis. Analysis period stability (APS) was evaluated at 10.0, 100, and 500 ng V/mL plasma (n = 4 replicates each) after storage on the autosampler at ambient temperature for 6 days. Long-term storage stability (up to 60 days) and freeze-thaw cycle stability (three freeze–thaw cycles of at least 24 hours) of V in plasma were evaluated at 15.0 and 750 ng V/mL plasma (n = 3 replicates each).

Partial method validation in Sprague Dawley male rat urine

A partial validation was performed for the analysis of V in male SD rat urine because initial efforts to measure V in urine matrix against a plasma matrix calibration curve were not successful. It was hypothesized that the differing chloride content of the two matrices made them incompatible for quantification against each other, and we concluded that a partial method validation was required to demonstrate acceptable quantification against a urine matrix calibration curve. A urine matrix calibration curve was prepared over the validated calibration range (5.00 to 1,000 ng V/mL urine) and urine matrix QC samples were prepared in triplicate at 10.0, 100, and 500 ng V/mL urine. Precision and accuracy of the measured V concentrations were assessed to verify performance in urine matrix.

ICP-MS analysis

Samples were analyzed for total V by ICP-MS using an X-Series II quadrupole ICP-MS (Thermo Fisher Scientific, Bridgewater, NJ) equipped with a Peltier-cooled spray chamber and an ASX-500 autosampler (Teledyne CETAC Technologies, Omaha, NE). ICP-MS instrumental parameters including lens voltages, gas flow rates, and plasma strength were optimized daily by aspiration of a tuning solution containing 10 ng In/mL and 1% (v/v) of HCl to minimize background V signal due to polyatomic interferences (< 200 counts per second) and maximize In signal intensity. For each calibration standard, the ⁵¹V signal intensity was corrected for internal standard intensity by monitoring the signal of a constant concentration of Pr in all standards and samples. The ratio of Pr signal in each standard and sample to the Pr signal in the initial blank is calculated to adjust the intensity of the V signal in each standard and sample which accounts for drift over time. The adjusted analyte signal was plotted against the standard concentration and an unweighted linear least-squares regression fit was calculated to produce the calibration curve for the experiment. Total V concentration for all samples was calculated from the linear regression equation and the measured V signal intensity in each sample and adjusted to correct for background V signal. Indium was included as a secondary internal standard during method development to allow for comparison of results between different IS elements, and Pr was selected as the internal standard element during the full validation.

Results and discussion

Method development

Before validation, the V method was developed in Sprague Dawley male rat plasma (Harrington et al. 2021). For V at the ng/mL levels, ICP-MS is an excellent technique for collecting high-throughput measurements. However, particularly in biomatrices, ³⁵Cl¹⁶O is a problematic polyatomic interference for ⁵¹V measurements. Several strategies have been developed to mitigate polyatomic interferences, including use of collision cell technology (CCT) mode, reaction cell mode, or high-resolution mass spectrometry (HR-MS) on a sector field mass spectrometer (Chrastny et al. 2006, Liu and Jiang 2002, Pick et al. 2010). Reaction cell mode involves introducing a gas (e.g., hydrogen, ammonia) into the ion path to react with the interferent and prevent it from passing through the quadrupole (D’Ilio et al. 2011), and is generally viewed as most appropriate for trace V analysis (Chrastny et al. 2006).

A method development experiment was conducted to evaluate the use of reaction cell mode on a quadrupole ICP-MS against use of a HR-MS system for interference correction. Calibration of rat plasma matrix standards covered a comparable linear range with similar correlation coefficients by both instruments (Table S1) and the accuracy of V in spiked plasma matrix QC samples (n = 5 replicates at 2.50; 5.00; 100; 250; 1,000; and 2,500 ng V/mL plasma levels) was acceptable for both interference correction approaches: −4.6–4.0% average RE for both approaches and precision ranging from 0.63% to 9.5% RSD (Tables S2–S3). A comparison of the results suggested a modest improvement through use of the sector field ICP-MS. An interlaboratory performance study (Nixon et al. 2002) indicated comparable performance between reaction cell and sector field instruments, consistent with our observations. Because the overall method performance was comparable on both instruments, the quadrupole instrument was selected for validation experiments to provide the most cost-effective and broadly transferable method.

Method validation in plasma

The method previously developed in Sprague Dawley male rat plasma was optimized and validated here. Method validation data are given in Table 1. The calibration curves were linear from 5.00 to 1,000 ng V/mL plasma. A representative calibration curve for V in rat plasma is provided in the Supplemental Information (Figure S2). All matrix calibration standards were acceptable for all validation days, with %RE ranging from −1.4% to 5.1%. The average absolute percent recovery for matrix standards across the validation range was 113%. Matrix effects ranged from 1.0% to 2.4% across the validation range, indicating that sample processing had minimal impact on V quantification in plasma.

Table 1.

Method validation data for vanadium (V) in plasma

Parameter	Result
Concentration range (ng V/mL plasma)	5.00 – 1,000
Correlation coefficient (r)	> 0.99
Matrix standard accuracy (%RE)	−1.4 – 5.1%
LODa (ng V/mL plasma)	0.268
LOQb (ng V/mL plasma)	5.00
Selectivity (mean matrix blank response relative to LOQ, %)	17.3%
Mean Instrument drift (%)	16.4%
Absolute recovery (%)	110 – 122%
Precision and accuracyc
Intra-day accuracy (mean %RE)	8.5 – 15.6%
Intra-day precision (%RSD)d	≤ 1.8%
Inter-day accuracy (mean %RE)	7.3 – 11.7%
Inter-day precision (%RSD)	≤ 3.1%
Dilution verification (up to 7,500 ng V/mL plasma)
Accuracy (mean %RE)	5.6%
Precision (%RSD)	0.9%

^aLOD = Limit of detection.

^bLOQ = Lower limit of quantitation, or the lowest calibration standard that exhibited −20% to 20% relative error (RE).

^cPrecision and accuracy were determined for triplicate QC samples fortified at three levels, 7.50, 250, and 800 ng V/mL plasma.

^d%RSD = percent relative standard deviation.

The LOQ was established as 5.00 ng V/mL plasma (mean %RE = 18.0% and RSD = 1.5%), and LOD was established as 0.268 ng V/mL plasma. Negligible background was observed in matrix blanks (mean measured instrument response was 17.3% of the instrument response for LOQ standards). An interlaboratory study of V in serum (Nixon et al. 2002) reported a slightly lower LOD than reported here but did not indicate the method used to calculate the LOD. Additionally, the difference in matrix (serum versus plasma) makes direct comparison of analytical figures of merit difficult. An early metareview of 28 analytical techniques for vanadium in serum indicated that a detection limit of approximately 0.20 ng/mL is necessary to accurately quantify V in biofluids (Heydorn 1990). Although serum was the focus of the metareview, it is anticipated that plasma analysis would have comparable requirements; it should also be noted that the metareview was performed early in the development of ICP-based analysis techniques, which were not yet capable of attaining such low detection limits. A more recent study of trace metals in human plasma using triple quadrupole ICP-MS indicated a limit of detection for V of 1.0 ng/mL as part of a multielement panel (Tonelli et al. 2017). Another study of vanadium in urine and seminal plasma indicated a quantitation limit of 0.0046 ng/mL in seminal plasma by a quadrupole system, although a comparison to the method validated here would be difficult as the referenced method was developed and validated in urine and the matrices studied there (human seminal plasma) are different from those validated here (rat blood plasma) (Wang et al. 2018; Wang et al. 2017). It is also likely that direct comparison of the analytical limits would not be possible as the reported limit of quantitation is very low, indicating it may have been calculated in the absence of matrix (although this is not explicitly stated by Wang).

Mean intra-day and inter-day accuracy and precision were determined by analyzing plasma matrix QC samples prepared in triplicate on three validation days at three concentrations (7.50, 250, and 800 ng V/mL plasma). The mean data are shown in Table 1. Intra- and inter-day %RE was 8.5 to 15.6% and 7.3 to 11.7%, respectively, demonstrating the accuracy of the method. Intra- and inter-day % RSD was <1.8% and <3.1%, respectively, demonstrating that the method can be used to precisely quantitate V in plasma. Precision and accuracy data from the triplicate plasma samples prepared outside of the linear range showed that the dilution effects did not adversely impact method performance for concentrations up to 7,500 ng V/mL plasma (5.6% average RE and 0.9% RSD).

Instrument drift was evaluated by analyzing the matrix calibration curve at the beginning and end of the longest analytical day (the first experiment consisted of 101 injections during a 5-hour analytical run) and comparing the determined concentrations between the measurements. Across the calibration range, the average %RE was 16.4%, indicating that the instrument drift was minimal.

The current validation demonstrated that the method is fit for use and produces accurate results with high precision at trace concentrations. These characteristics are particularly important due to the anticipated low endogenous levels of V in biofluids and the possibility of contamination from environmental sources, collection media, and even the skin of the animal (Martin et al. 2020). Martin et al. (2020) reviewed analytical methods for vanadium in biological tissues and indicated that average concentrations in serum were 0.83 ± 0.09 ng/mL. Although the detection limits noted for high resolution sector field ICP-MS analysis methods were lower than those obtained with the current quadrupole instrument (0.00001 ppb, likely calculated from the standard deviation of method blank samples and ignoring matrix factors), the limits obtained here are sufficient to detect and quantify V in rat plasma collected from toxicology studies, which have been shown to be elevated well above endogenous levels (Harrington et al. 2021).

Stability of vanadium in plasma

Analyte stability was evaluated under several storage conditions with comparison to freshly prepared (Day 0) samples. Analysis period stability was determined by storing digested and extracted samples at ambient (up to 6 days) and refrigerator (up to 15 days) temperatures. Freeze–thaw stability was determined in plasma stored frozen (nominal −70°C) and subjected to three complete freeze–thaw cycles. All results were within 100 – 126% of Day 0 (Table 2) demonstrating stability of V under these conditions of storage. Long-term storage stability was determined by storing spiked plasma samples at −80°C for up to 60 days. All results were within 100 – 118% of Day 0 (Table 2), demonstrating stability of V after prolonged frozen storage of plasma samples (at least 60 days).

Table 2.

Stability data for V in plasma.

Storage condition	Relative standard deviation (%)	Mean Percentage (%) of Day 0
Extract stability a
Refrigerated extracts (15 days, 4°C)	≤ 1.6	105 – 126
Ambient extracts (6 days)	≤ 1.0	97.7 – 113
Plasma stability b
Freeze–thaw (3 cycles over 3 days)	≤ 2.5	100 – 106
Frozen matrix (−80°C, 60 days)	≤ 2.6	100 – 122

^aExtract stability (refrigerated and ambient) was determined for quadruplicate QC samples fortified at 10.0, 100, and 500 ng V/mL plasma.

^bFreeze-thaw and frozen sample stability was determined for triplicate plasma samples fortified at 15.0 and 750 ng V/mL plasma.

Urine matrix partial validation

We attempted to apply the plasma matrix calibration method to analysis of V in male Sprague Dawley rat urine secondary matrix samples, but the results were not acceptable due to elevated relative errors of plasma matrix QC samples. We hypothesized that this result was due to chloride interference considerations discussed earlier. In urine samples, differences in homeostatic factors including hydration status, health conditions, and other factors can result in varying chloride concentrations both between individuals and at different sample collection times (Wang et al. 2013). Potentially large sample-to-sample chloride concentration variability is an important concern for V measurements by ICP-MS because of the ³⁵Cl¹⁶O polyatomic interference for ⁵¹V and the importance of matrix-matched tuning to mitigate it in quadrupole-based reaction cell instruments. Therefore, it was deemed necessary to perform a partial validation for rat urine to verify the performance of the digestion and analysis methods for quantification of V.

Male Sprague Dawley urine QC samples were prepared and analyzed against spiked urine matrix calibration standards and the accuracy and precision of the measured V concentration was calculated at low (10.0 ng V/mL urine), medium (100 ng V/mL urine), and high (500 ng V/mL urine) levels. Resulting unweighted linear regression parameters are shown in Table 3, showing comparable background (y-intercept), linear range, and linear correlation for urine matrix calibration, compared to the plasma matrix calibration from the primary matrix validation. Comparing sensitivity between the plasma calibration curves and the urine calibration curve demonstrates why the partial validation had to be done against a urine calibration curve, as the slopes of the two calibration approaches differed by a factor of more than 2. Although some of the variability of the slope may be due to day-to-day variability of instrument operation, this significant difference may also indicate matrix effects likely due to differing chloride concentrations between urine and plasma. However, it is important to note that the accuracy and precision indicate that the interference correction was sufficient to ensure accurate results.

Table 3.

Partial validation of the determination of V in male SD rat urine calibrated against matrix-matched calibration curve.

Parameter	Value
Concentration range (ng V/mL urine)	5.00–1,000
Slope	37376
y-Intercept	370.8
Correlation coefficient (r)	> 0.999
Accuracy (%RE)a	−2.0 – 0.3%
Precision (%RSD)	≤ 0.6%

^aAccuracy and precision of urine samples was calculated from triplicate measured V content of urine samples fortified 10.0, 100, and 500 ng V/mL urine.

Across all three urine matrix QC levels, −2.0–0.3% average RE was observed with RSD ≤ 0.6% (n = 3), indicating acceptable recovery of urine QC samples measured against a urine matrix calibration curve. These results demonstrate accurate, precise measurement of V in urine against a urine matrix calibration curve processed by the same acid digestion method.

Conclusion

Vanadium is an emerging contaminant of concern because of its presence in carbon-based fuels and its potential for accidental release into the environment. Although it is broadly known that some V forms are more toxic than others, detailed toxicological assessments are needed for a range of V salts for both V⁴⁺ and V⁵⁺ oxidation states. A simple, cost-efficient, and robust analytical method is also needed to support these toxicological studies to determine the biodistribution and uptake of the salts by test animals.

For the current study, a method for determining the total V in rat plasma was subjected to rigorous validation. The results showed that precision and accuracy met acceptance criteria, demonstrating reliable method performance within and between days. Matrix calibration produced acceptable results for analysis of plasma matrix QC samples. The use of matrix calibration curves is generally preferable, especially when endogenous analyte levels are relatively low or are consistent between samples and over time (Levine et al. 2011). Stability analysis at a range of conditions indicated that fortified plasma and extracts were stable for up to 60 days. Finally, the V content of urine samples was accurately determined against a urine matrix calibration curve using the same digestion method as for plasma.