Liquid chromatography as candidate reference method for the determination of vitamins A and E in human serum

Abstract Background Owing to the increasing interest in public health research of antioxidant micronutrients and the inaccuracy of routine serum concentrations of the fat‐soluble vitamins A (retinol) and E (DL‐α‐tocopherol) measurements, we developed a reliable, highly sensitive, robust and rapid method for the quantification of two clinically important lipophilic antioxidants in serum using a reverse‐phase HPLC/DAD method. Method Sample preparation and analytical conditions that would affect extraction efficiency and quantitative results of vitamins A and E were investigated and optimized. Vitamins A and E were extracted from serum via liquid‐liquid extraction (LLE). After adequate sample preparation, the samples were injected directly into the HPLC system with diode‐array detector (DAD). Chromatographic separation was completed in 7 minutes for vitamins A and E. With vitamin A acetate and vitamin E acetate as internal standards, the method was applied to the measurement of vitamins A and E in human serum. Results We evaluated method linearity, accuracy (recovery rate and trueness), precision, carryover, limit of quantitation and limit of detection, and measurement uncertainty. The method was evaluated for trueness using NIST Standard Reference Material SRM 968f. The serum concentration of the studied compounds had a good linear relationship in the range of 0.05 ~ 3.0 μg/mL concentration (r ＝ 0.9998), with 0.0077 μg/mL detection limit and 0.025 μg/mL quantitative limit for vitamin A, respectively, and 1.0 ~ 60.0 μg/mL concentration (r ＝ 0.9999), with 0.40 μg/mL detection limit and 0.50 μg/mL quantitative limit for vitamin E, respectively. The intra‐ and inter‐assay coefficients of variation were calculated by using three concentrations (1, 2, and 3) of the studied compounds in human serum samples. Intra‐assay and inter‐assay precision were 1.23%‐4.97% and 0.97%‐3.79% for vitamin A, respectively, and 0.64%‐4.07% and 0.81%‐5.96% for vitamin E, respectively. The average recovery rates were 100.98% for vitamin A, and 99.21% for vitamin E, respectively. The carryover rate of vitamins A and E was below 1%. As for the evaluation of accuracy, the biases were <± 5% by comparing with NIST standard reference material SRM 968f. Conclusion The method is a simple sample treatment procedure for the determination of fat‐soluble vitamins A and E in human serum with high sensitivity and specificity. The proposed method could be recommended as a candidate reference method for the determination of serum concentrations of the fat‐soluble vitamins A and E in human serum.


Results:
We evaluated method linearity, accuracy (recovery rate and trueness), precision, carryover, limit of quantitation and limit of detection, and measurement uncertainty. The method was evaluated for trueness using NIST Standard Reference Material SRM 968f. The serum concentration of the studied compounds had a good linear relationship in the range of 0.05 ~ 3.0 μg/mL concentration (r ＝ 0.9998), with 0.0077 μg/mL detection limit and 0.025 μg/mL quantitative limit for vitamin A, respectively, and 1.0 ~ 60.0 μg/mL concentration (r ＝ 0.9999), with 0.40 μg/mL detection limit and 0.50 μg/mL quantitative limit for vitamin E, respectively. The intra-and inter-assay coefficients of variation were calculated by using three concentrations (1, 2, and 3) of the studied compounds in human serum samples. Intra-assay and interassay precision were 1.23%-4.97% and 0.97%-3.79% for vitamin A, respectively, and 0.64%-4.07% and 0.81%-5.96% for vitamin E, respectively. The average recovery rates were 100.98% for vitamin A, and 99.21% for vitamin E, respectively. The carryover rate of vitamins A and E was below 1%. As for the evaluation of accuracy, the biases were <± 5% by comparing with NIST standard reference material SRM 968f.

| INTRODUC TI ON
Oxidative stress, an imbalance between oxidants and antioxidants in favor of the impaired antioxidants potentially leading to damage, is believed to be implicated in the etiopathology of a number of diseases including cholestatic liver disease, exocrine pancreatic insufficiency, cancer, atherosclerosis, neurodegenerative disorders, cardiovascular disease, and other conditions. [1][2][3] However, interest has grown on the possibility that intracellular and extracellular antioxidants may reduce the risk of oxidative modification. Among fat-soluble vitamins, vitamins A and E are major components of the antioxidant system in humans, protecting cell membranes against peroxidation, 4-6 maintaining organism metabolism and physiological functions. In general, retinol and DL-α-tocopherol are considered to be the most effective biologically active forms of vitamins A and E. In clinical, determination of vitamins A and E is determination of retinol and DL-α-tocopherol in human serum concentration. Vitamin A plays a significant role in a wide range of physiological processes in the human body, including maintaining normal growth and reproduction, embryonic development, vision, immunity response, and neurogenesis. [7][8][9] Vitamin E has the protective effect on other biological antioxidants that are associated with certain cancers, 10 chronic diseases, 11,12 diabetic complications, 13 neurological, and reproductive processes. 9 23,24 or mass spectrometer (MS) 25 are more traditional approaches and are widely used in routine clinical measurements. However, the results obtained with different methods are often not comparable because of inter-method and inter-laboratory variability. In addition, extremely low level of vitamin A in biological samples needs a large quantity of serum and do not allow for accurate quantification and high sensitivity in a single chromatographic run. 26,27 The inadequate accuracy of vitamins A and E measurements hampers the interpretation of data in public health research. In order to improve the accuracy of the routine methods in laboratory medicine, establishing a reference measurement system for value-assigning NIST standard reference material (SRM) and external quality assessment of these vitamins is the best way to help verifying stan- Based on previous reported work, our developed method is simple, rapid, accurate, reliable, sensitive, selective, and cost-effective HPLC method with diode-array detector (DAD), for the simultaneous determination of vitamins A and E in human serum after LLE.
Vitamin A acetate and vitamin E acetate were used as internal standards (IS) to improve the accuracy and precision of the method. The method was also validated strictly and comprehensively, and applied to the determination of vitamins A and E in human serum.

| Materials and reagents
Analytical standards of vitamin A, vitamin E and the internal standards of vitamin A acetate, vitamin E acetate were obtained from Sigma (Sigma-Aldrich Co. LLC.). The Standard Reference Material (SRM 968f) was purchased from National Institute of Standards and Technology (NIST). 2,6-bis (1,1-dimethylethyl)-4-methylphenol (butylated hydroxyl toluene, BHT) was obtained from Sigma (Sigma-Aldrich Co. LLC.). All used solvents including methanol, ethanol, and hexane (chromatography grade) were purchased from Thermo (Thermo Fisher Scientific Inc). A water purification system (Millipore) was used to provide Milli-Q water of ultra-pure quality (18.2 MΩ).
Human serum samples were collected from different healthy volunteers.

Conclusion:
The method is a simple sample treatment procedure for the determination of fat-soluble vitamins A and E in human serum with high sensitivity and specificity. The proposed method could be recommended as a candidate reference method for the determination of serum concentrations of the fat-soluble vitamins A and E in human serum.

K E Y W O R D S
candidate reference method, HPLC, vitamins A and E

| Chromatographic conditions
High-performance liquid chromatography analysis was performed on a SHIMADZU LC-30A system equipped with diode-array detector.
Samples were analyzed using an ODS HYPERSIL C 18 analytical column (3.0 µm, 100 mm × 2.1 mm) connected with a 2.1 mm × 20 mm guard column at a temperature of 30°C, and the flow rate was kept constantly at 0.5 mL/min. The mobile phase employed a gradient elution using the following constituents: Mobile phase A was water; mobile phase B was methanol. The gradient began with 93% methanol/water for 1 minutes, followed ramping to 100% methanol in 0.01 minutes.
Then, 100% methanol was maintained for 2.5 minutes, then returned to initial conditions and equilibrated for 3.5 minutes. The autosampler temperature was 4°C, and the injection volume was 20 μL.

| Calibration preparation and control materials
Volumetric steps were gravimetrically controlled, where indicated, We prepared all internal quality control (IQC) materials using pooled human serum from healthy volunteers. All units were screened and spiked with either vitamin A or vitamin E, or both, as needed, to achieve desired concentrations, and all specimens were stored at −70°C when not in use.

| Sample preparation procedure
Volumetric steps were gravimetrically controlled where indicated.
100 μL serum/working standard solutions was pipetted into a wellcapped 2 mL amber polypropylene tube and accurately weighted.
200 μL of the internal standards was gravimetrically added, and the solution was vortex-mixed for 2 minutes. After An aliquot (800 μL) of the n-hexane containing BHT (1 mg/mL) was added, the content mixed vigorously on a vortex mixer for 5 minutes and centrifuged (8000× g, 5 minutes, 4°C). The upper hexane layer was transferred into 5 mL amber glass tubes. To maximize extraction efficiency, the LLE step was repeated and the combined organic layers were dried under nitrogen at room temperature. The dried residue was dissolved in 100 μL of 95% methanol/water followed by 2 minutes vortex and centrifugation at 13 000× g for 5 minutes at 4°C. Then, 20 μL of this extracted serum was injected into HPLC for analysis.
The calibration solution was validated prior to use by testing with Standard Reference Material (SRM 968f) (NIST), and we treated SRM 968f using the same procedures to complete the method validation. A standard curve was prepared for each run by extracting and analyzing two aliquots of the calibration solution.

| Optimization of high-performance liquid chromatographic conditions
Analytes were separated on a reversed-phase column using a gradient system of methanol and water. The mobile phase was optimized in order to obtain the best separation of the analytes in the shortest time. Vitamin A, vitamin E, and internal standards as well as pooled serum samples were used for studying the mobile phase composition. Several elutions (mixtures of organic solvents such as acetonitrile, methanol, ethanol, and formic acid) and several gradients were assessed. And other experimental parameters such as analytical column, detector wavelength, flow rate, column oven temperature, injection volume, and internal standard selection were also optimized. The best results were obtained for the conditions that described in "Chromatographic conditions." The criteria were resolution, stability of the absorbance, and analysis duration. According to our results, we can conclude that the presented method is highly robust. Representative chromatograms of the HPLC analysis of mixed standard solution and human serum are shown in Figures 1 and 2.

| Optimization of sample extraction
Sample preparation is essential for accurate analysis. Both pretreatments (protein precipitation and liquid-liquid extraction) were studied. Several protein precipitants (methanol, acetonitrile, ethanol, 2-propanol, zinc sulfate solution, and strong acid) and different extractants (dichloromethane, diethyl ether, n-hexane, and ethyl acetate) were tested. Eventually, LLE was selected to extract the target vitamins in this study. And the results showed that the highest extraction efficiency and fewer impurities were obtained by use of n-hexane. In addition, it well dissolves the fat-soluble vitamins and does not mix with the water fraction of the blood serum matrix. Therefore, n-hexane was selected as the extractant for the study.

| Method validation
The developed method was validated strictly and comprehensively by evaluating the linearity, accuracy (recovery rate and trueness), precision, carryover, limit of quantitation and limit of detection, and measurement uncertainty. The laboratory results showed that the method was accurate and fully validated for the simultaneous deter-

mination of vitamins A and E in serum.
The calibration curves of standard solutions showed good linearity with high correlation coefficients in the range of 0.05 ~ 3.0 μg/ mL and 1.0 ~ 60.0 μg/mL for vitamins A and E, respectively.

Regression equation and correlation coefficient (r) calculated from
the calibration curves of standard solutions for vitamin A were y = 0.512439x + 0.002158, r = 0.9998 and for vitamin E were y = 1.71181x + 0.003990, r = 0.9999 (Figure 3). The target for lower limit of quantification of each analyte was based on clinical need and defined as the lowest tested concentration at which the precision was within <20% and analytical recovery was within 100% ± 20%.
The confirmed lower limit of quantification varied depending on the analytes but was between 0.025 and 0.5 μg/mL for vitamins A and E, respectively.  Table 1, the mean recovery of vitamin A was 100.65% and that of vitamin E was 99.67%. Furthermore, the trueness of the proposed method was demonstrated by analyzing the NIST-certified reference materials.
Each level of the material was measured 3 times, and the measurement results were shown in Table 2. The obtained results were all in good agreement with the certified values.
To test the precision of this method, internal quality control (IQC) materials of serum samples including low-, medium-, and high-concentration serum samples were chosen. Intra-assay and inter-assay coefficient of variations (CVs) and total coefficient of variation were determined in terms of repeatability and quantified by the CV of the replicate measurements. As presented in Table 3, the intra-assay coefficients of variation (CV) of low, middle, and high levels for vitamin A were 4.13%-4.97%, 1.23%-2.04%, and 1.24%-2.16%, and for vitamin E were 1.62%-4.07%, 1.11%-2.32%, and 0.64%-3.14%, respectively (n = 5), whereas the inter-assay coefficients of variation were 0.97%, 3.79%, and 2.57% for vitamin A and 5.96%, 2.62%, and 0.81% for vitamin E, respectively. The data indicated that the assay method showed good repeatability.
To evaluate carryover, samples with high and low concentration were prepared at the upper and lower limit of analytical measurement range and the injection sequence of low-low-low-hi-hi-low-hihi-low-low-low-low-hi-hi-low-hi-hi-low was performed in triplicate with 3 independent extractions. The mean difference of <20% and <3 SD between low-low and hi-low indicated no significant carryover. 32 SD was calculated based on low control-1 values. The data indicated that no significant carryover was observed for vitamins A and E.

| Measurement uncertainty
The measurement uncertainty of the proposed method is strictly evaluated according to the evaluation procedure of measurement uncertainty which were established on Guide to the Expression of Uncertainty in Measurement (GUM, 1999

| CON CLUS ION
In the present study, we developed a reliable, robust, and highly accurate method for the quantification of vitamins A (retinol) and E (DL-α-tocopherol) in serum using a reverse-phase HPLC/UV isocratic method that featured good sensitivity despite using a relatively small amount of serum. The good quality of accuracy and precision val-