tandem mass spectrometry

Journal of Chromatography A, 1135 (2006) 203–211

Quantitative determination of folic acid in multivitamin/multielement tablets using liquid chromatography/tandem mass spectrometry Bryant C. Nelson ∗ , Katherine E. Sharpless, Lane C. Sander National Institute of Standards and Technology, Analytical Chemistry Division, 100 Bureau Drive, Stop 8392, Gaithersburg, MD 20899-0001, USA Received 20 July 2006; received in revised form 7 September 2006; accepted 18 September 2006 Available online 2 October 2006

Abstract Two different isotope-dilution liquid chromatography/tandem mass spectrometry (LC/MS/MS) methods for the quantitative determination of folic acid (FA) in multivitamin/multielement tablets are reported. These methods represent distinct improvements in terms of speed and specificity over most existing microbiological and chromatographic methods for the determination of FA in dietary supplements. The first method utilizes an aqueous/organic-based extraction solvent combined with positive-ion mode LC/MS/MS detection of protonated [M + H]+ FA molecules and the second method utilizes a pure aqueous-based extraction solvent combined with negative-ion mode LC/MS/MS detection of deprotonated [M − H]− FA molecules. The LC/MS/MS methods exhibit comparable linear dynamic ranges (≥3 orders of magnitude), limits of detection (0.02 ng on-column) and limits of quantification (0.06 ng on-column) for FA. Two methods employing different extraction and different MS detection modes were developed to allow method cross-validation. Successful validation of each measurement procedure supports the use of either method for the certification of FA levels in dietary supplements. The accuracy and precision of each measurement procedure were evaluated by applying each method to the quantitative determination of FA in a NIST standard reference material (NIST SRM 3280 multivitamin/multielement tablets). The FA measurement accuracy for both methods was ≥95% (based on the manufacturer’s assessment of the FA level in SRM 3280) with corresponding measurement precision values (% RSD) of approximately 1%. Published by Elsevier B.V. Keywords: Dietary supplements; Folic acid; Fortification; Liquid chromatography; Multivitamin/multielement; Standard reference material; Tandem mass spectrometry

1. Introduction Folates are an important class of water-soluble B-vitamins that are essential for normal human cell division and cell growth. A deficiency of folate in the diet is closely linked to the presence of neural tube defects in newborns and to an increased risk of megaloblastic anemia, cancer, Alzheimer’s disease and cardiovascular disease in adults [1,2]. The prevalence of folate deficiency in the United States has been effectively reduced due to a combination of the cereal/grain folate fortification program (initiated in 1998) [3] and the widespread use of folate-fortified dietary supplements (nutraceuticals, multivitamin supplements, multivitamin/multielement supplements, etc.) [4–6]. However, folate deficiency remains one of the most common vitamin defi-

∗

Corresponding author. Tel.: +1 301 975 2517; fax: +1 301 977 0685. E-mail address: [email protected] (B.C. Nelson).

0021-9673/$ – see front matter. Published by Elsevier B.V. doi:10.1016/j.chroma.2006.09.040

ciencies worldwide and is a special concern for the elderly [7]. Folic acid (FA), otherwise known as vitamin M or pteroylglutamic acid (Fig. 1), is the folate vitamer that is traditionally utilized to fortify cereals/grains and dietary supplements. FA is a synthetic vitamer, and it is the fortificant of choice because of its relative stability and increased bioavailability compared to natural folate forms [8,9]. The use of FA-fortified dietary supplements has been dramatically increasing in recent years because of FA’s established role in reducing the prevalence of neural tube defects and its purported role in reducing the risk of cardiovascular disease [4–6]. However, there exists no established reference method for the quantitative determination of FA in dietary supplements in general, nor in multivitamin/multielement supplements in particular. There exists some concern that excess FA in the diet might lead to increased health risks from masking symptoms of vitamin B12 deficiency or to the increased prevalence of life-threatening diseases due to adverse gene selection [4–6].

204

B.C. Nelson et al. / J. Chromatogr. A 1135 (2006) 203–211

Fig. 1. Chemical structure for FA showing the two major ESI fragmentation pathways (A, B). Fragmentation at A results in the characteristic loss of the glutamic acid moiety from the intact molecule. The orientation of the ␣ and ␥ carboxyl groups on the glutamic acid moiety are labeled. Fragmentation at B produces the characteristic folate degradation product, p-aminobenzoylglutamic acid (p-ABG, relative mass 266.3 g/mol). The relative mass of FA is 441.4 g/mol.

Because of these concerns, analytical methods that accurately and specifically measure FA in dietary supplements are needed to provide quality assurance and quality control for commercial products. The traditional approach to the quantification of FA in multivitamin or multivitamin/multielement supplements was based on a microbiological assay [10]. However, microbiological assays are time-consuming, requiring a minimum of 24 h to obtain assay results. Because of the complex matrix of most multivitamin/multielement supplements, current analytical methods are generally based on LC separation of the vitamin/element/excipient components followed by ultraviolet absorbance (UV) detection of FA at 280 or 290 nm [11–20]. All of the reported LC/UV methods provide satisfactory quantification of FA, yet none of the methods have been crossvalidated against other high-specificity analytical methods. Two negative-ion mode LC/mass spectrometry (MS) methods have been developed for the quantification of FA in breakfast cereals and/or multivitamin tablets [18,21]. However, one method shows poor chromatographic retention for FA [18], while the other method [21] utilizes a non-ideal internal standard compound (hippuric acid) for FA quantification. It is not clear why the authors of the last report utilized hippuric acid instead of a stable isotope-labeled version of FA as an internal standard for their determinations; stable isotope-labeled folate vitamers have been commercially available since 2001. Recently, a positive-ion mode LC/MS/MS method for the specific quantification of FA-fortified foods was reported [22]. The method was applied to the successful quantification of FA in multivitamin pharmaceuticals; however, multivitamin pharmaceuticals typically contain FA levels 10 to 100 times higher than over-the-counter multivitamin dietary supplements. Additionally, the quantitative performance of the LC/MS/MS method was not evaluated against another high-specificity method. The development of two potential reference methods based on the use of either positive- or negative-ion mode LC/MS/MS for the quantitative determination of FA in multivitamin/multielement tablets is now reported. The methods were developed using NIST standard reference material (SRM) 3280 multivitamin/multielement tablets as the test sample. The

methods were developed based on different tablet extraction techniques and different mass spectrometric detection principles to allow method cross-validation and to ensure independence of the analytical results. The analytical development and performance characteristics of both LC/MS/MS methods are described herein. 2. Experimental1 2.1. Safety considerations The handling of organic solvents and organic/inorganic acids should be regarded as potentially hazardous. Safe working conditions (use of safety goggles and disposable protective gloves) and solvent disposal procedures should be established before initiating work. 2.2. Reagents and materials FA (lot 089H1371, purity = 99.7%), dithiothreitol (DTT), potassium phosphate (K2 HPO4 ), ascorbic acid (AA), ammonium hydroxide (5 mol/L), acetic acid and formic acid were obtained from Sigma Chemical Company (St. Louis, MO). [13 C5 ]-FA (lot TE-2135A, purity = 98.9%) was obtained from Merck Eprova AG (Schafhausen, Switzerland). The purity values for FA and [13 C5 ]-FA were determined using a combination of LC/UV, LC coupled with charged aerosol detection and the manufacturer’s purity assessment (thin layer chromatography). The identity of FA and [13 C5 ]-FA were confirmed by direct-infusion MS and MS/MS analyses at NIST. HPLC-grade methanol was obtained from J.T. Baker (Phillipsburg, NJ). SRM 3280 was obtained from the standard reference materials group at NIST. Each tablet was formulated as single-unit daily dosage with a nominal tablet weight of 1500 mg. Each tablet contained 1 Certain commercial equipment, instruments and materials are identified in order to specify experimental procedures as completely as possible. In no case does such identification imply a recommendation or endorsement by the National Institute of Standards and Technology (NIST) nor does it imply that any of the materials, instruments or equipment identified are necessarily the best available for the purpose.

B.C. Nelson et al. / J. Chromatogr. A 1135 (2006) 203–211

13 vitamins, including FA, and 18 elements. SRM 3280 has been prepared as part of a collaborative effort between NIST and the National Institutes of Health’s Office of Dietary Supplements (NIH/ODS) in support of the dietary supplement ingredient database that is being produced by the U.S. Department of Agriculture [23,24]. SRM 3280 was prepared, using normal manufacturing procedures, as a non-commercial batch of multivitamin/multielement tablets. Because some of the individual vitamins are coated or encapsulated to provide stability, and grinding would compromise this coating, the SRM will be provided in bottles containing 30 whole tablets. Because between-tablet homogeneity is not expected, it is necessary to grind (homogenize) a number of tablets and then remove a test portion for analysis. There are no stability data on the ground tablet, although FA levels in ground samples were observed to be stable for at least 24 h when the samples were protected from light and oxygen. Purified water (18 M), prepared using a Millipore Milli-Q purification system, was used to prepare all samples and standards. All other chemical reagents and solvents were ACS reagent grade unless stated otherwise. Reagent concentrations given in terms of percent (%) are to be considered as mass fractions (g/g) in all listed procedures. Preparation of analyte stocks/standards, samples and calibrants were performed gravimetrically in all listed procedures, except where noted otherwise. Additionally, all procedures were conducted using helium-degassed buffers/solvents under subdued lighting conditions. 2.2.1. Preparation of FA stock solutions Five FA and one [13 C5 ]-FA stock solutions (200 ng/␮L) were prepared by weighing the appropriate folate powder into an amber glass vial containing a weighed amount of degassed 20 mmol/L K2 HPO4 buffer (pH 7.4). The concentrations for the folate stock solutions were assigned using previously published procedures [25]. Briefly, the concentrations for the folate stock solutions were determined by measuring the UV absorbance of 40X dilutions of each stock and using a literature absorptivity value of 27.6 × 103 L mol−1 cm−1 at 282 nm for FA and [13 C5 ]FA [26]. The concentrations were then corrected for impurities that contributed to the absorbance readings by performing an LC analysis on each stock solution using absorbance detection at the same wavelength used for the absorptivity determination. The folate concentrations were calculated by multiplying the spectrophotometric concentrations by the peak ratio (folate peak area/total peak area) determined from the LC analysis. An appropriate amount of solid DTT was added to each stock solution so that the concentration of DTT in solution was 0.1%. Diluted standards of FA and [13 C5 ]-FA were prepared, as needed, by diluting weighed portions of the appropriate folate stock with weighed portions of degassed 20 mmol/L K2 HPO4 /0.1% DTT buffer (pH 7.4). The FA and [13 C5 ]-FA stock and diluted-standard solutions were stored at 4 ◦ C until needed. 2.2.2. Preparation of calibrants Five independent calibrants were prepared gravimetrically by adding sequentially increasing amounts of the appropriate FA stock solution and constant amounts of the [13 C5 ]-FA diluted-

205

standard solution to 10 mL portions of degassed 20 mmol/L K2 HPO4 /0.1% DTT buffer (pH 7.4). The nominal FA concentrations in the calibrants ranged from 8 ng/␮L to 12 ng/␮L. The nominal [13 C5 ]-FA concentration in the calibrants was 12 ng/␮L. The calibrants were subdivided into 1 mL aliquots and stored at −20 ◦ C until needed. 3. Methods 3.1. Positive-ion mode LC/MS/MS The total contents (30 tablets) of one bottle of the SRM were homogenized using an automatic mortar grinder. A 300 mg sample was weighed into a 15 mL centrifuge tube and spiked with 500 ␮L of [13 C5 ]-FA standard (200 ng/␮L). The sample was diluted with 10 mL of 0.1% DTT in 50/50 methanol/water (apparent pH 7.5), vortex mixed and subjected to rotational mixing on an orbital rotator for 30 min. The sample (pH 6.2) was then centrifuged at 3000 × g for 10 min and the supernatant was removed and transferred to a clean 15 mL centrifuge tube. One milliliter of the supernatant was filtered through a regenerated cellulose filter (0.45 ␮m pore) directly into a sample vial. Sample extracts and calibrants were injected (10 ␮L) onto the LC/MS/MS system. Analyte/internal standard peak area ratios (area/area) and mass ratios (mg/mg) from the calibrants were subjected to linear least squares regression analysis to produce calibration curves (y-intercept model) and calibration equations. FA in the sample extracts was quantified on the basis of the relevant calibration equation and the FA/[13 C5 ]-FA peak area response ratio detected in the sample extract. Samples and calibrants were injected (one time) and analyzed using the following analysis sequence: calibrant 1–5, sample, sample, sample, etc. end. Experiments were conducted on a HP1100 LC Series system coupled to an Applied Biosystems 4000 Q-Trap MS/MS system. The Q-Trap MS/MS system was operated as a triplequadrupole MS/MS system in positive electrospray-ionization (ESI) mode. The LC system was outfitted with a binary pump, a variable-wavelength UV absorbance detector, a temperature controlled (10 ◦ C) autosampler and an in-line mobile phase vacuum degasser. Samples were analyzed using a Supelco Discovery HS-F5 (pentafluorylphenyl) analytical column (4.6 mm × 150 mm, 5 ␮m particle diameter) with an attached HS-F5 guard column (3 mm × 20 mm, 5 ␮m particle diameter) held at 30 ◦ C ± 1 ◦ C. The LC elution conditions were as follows (all solvent percentages are volume fractions): mobile phase A = 1% formic acid in water; mobile phase B = 1% formic acid in methanol; time program = 0 min, 60% A/40% B; 10.0 min, 60% A/40% B; 10.1 min, 0% A/100% B; 12.0 min, 0% A/100% B; 12.1 min, 60% A/40% B; 15.0 min, 60% A/40% B; flow rate = 500 ␮L/min. FA and [13 C5 ]-FA were detected and quantified using multiple-reaction-monitoring (MRM) of the protonated [M + H]+ analyte molecules. MS/MS operating parameters are summarized in Table 1. For the collisioninduced-dissociation (CID) LC/MS/MS experiments, spectra were collected from m/z 50 to m/z 500 in 1 s.

206

B.C. Nelson et al. / J. Chromatogr. A 1135 (2006) 203–211

3.2. Negative-ion mode LC/MS/MS Sample preparation, extraction and analysis were identical to the procedures previously described for the positive-ion mode LC/MS/MS method except that the sample was extracted using a strongly alkaline extraction solution containing no organic solvent component. Briefly, the sample was weighed and spiked with [13 C5 ]-FA standard and then diluted with 10 mL of 0.1% DTT/0.3% ammonium hydroxide solution (pH 11.1), vortex mixed and subjected to rotational mixing on an orbital rotator for 15 min. The sample (pH 10.1) was then centrifuged at 3000 × g for 10 min, and the supernatant was removed and transferred to a clean 15 mL centrifuge tube. One milliliter of the supernatant was filtered through a regenerated cellulose filter (0.45 ␮m pore) directly into a sample vial for analysis. The instrumental setup and LC column were identical to the setup and column utilized for the positive-ion mode LC/MS/MS method; however, the LC elution conditions and MS/MS parameters were optimized for negative-ion mode ESI detection. The LC elution conditions were as follows: mobile phase A = 0.1% acetic acid in water; mobile phase B = 0.1% acetic acid in methanol; time program = 0 min, 55% A/45% B; 10.0 min, 55% A/45% B; 10.1 min, 0% A/100% B; 12.0 min, 0% A/100% B; 12.1 min, 55% A/45% B; 15.0 min, 55% A/45% B; flow rate = 500 ␮L/min. FA and [13 C5 ]-FA were detected and quantified using MRM of the deprotonated [M − H]− analyte molecules. For the CID LC/MS/MS experiments, spectra were collected from m/z 50 to m/z 500 in 1 s. MS/MS operating parameters for negative-ion mode detection are summarized in Table 1. 3.3. Preparation of linearity standards An internal standard stock solution (30 mL) containing approximately 12 ng/␮L [13 C5 ]-FA in 20 mmol/L K2 HPO4 /0.1% DTT buffer (pH 7.4) was prepared in an amber glass vial. An analyte stock solution (10 mL) containing a FA concentration of 234 ng/␮L was prepared using the Table 1 LC/MS/MS instrument parameters for the quantitative determination of FA in multivitamin/multielement tabletsa Parameter

MRM transition (m/z) DP (V) CE (V) CXP (V) EP (V) CAD gas (kPa) Curtain gas flow (kPa) Gas 1 flow (kPa) Gas 2 flow (kPa) Dwell time (ms/ion) Ion spray voltage (V) Source temperature (◦ C)

Positive-ion mode

Negative-ion mode

FA

[13 C5 ]-FA

FA

[13 C5 ]-FA

442 → 295 61 23 10 10 34 138 345 345 500 5000 500

447 → 295 61 23 10 10 34 138 345 345 500 5000 500

440 → 311 −95 −32 −23 −10 68 69 138 276 500 −4000 700

445 → 311 −95 −32 −23 −10 68 69 138 276 500 −4000 700

DP, declustering potential; CE, collision energy; CXP, collision exit potential; EP, entrance potential; CAD, collision activated dissociation. a Testing was conducted with SRM 3280.

12 ng/␮L [13 C5 ]-FA solution as diluent. A set of 19 volumetric serial dilutions was prepared from the FA stock solution covering a FA concentration range from 0.001 to 234 ng/␮L using the 12 ng/␮L [13 C5 ]-FA stock solution as diluent. The linearity standards were analyzed using both LC/MS/MS methods to estimate each method’s linear dynamic range, limit of detection (LOD) and limit of quantification (LOQ). 4. Results and discussion 4.1. Development of LC/MS/MS methods Building upon our recent research on the positive-ion mode LC/MS [27] and LC/MS/MS [25] analysis of folates in human serum, three reversed-phase LC columns (a phenylpropyl phase, a C18 phase, a pentafluorylphenyl phase) were screened for the analysis of FA. The columns were screened utilizing a mobile phase consisting of 1% formic acid in 50/50 water/methanol (500 ␮L/min) with positive-ion mode MRM detection (FA = m/z 442 → m/z 295). The three columns produced excellent chromatographic peak shapes and adequate retention for FA in phosphate-buffered standards; however, the pentafluorylphenyl phase column (Discovery HS-F5) allowed for the most flexible control of analyte retention based upon the composition of the mobile phase. Final LC parameters (column temperature, flow rate, etc.) were optimized with the HS-F5 column (see Section 3). Positive-ion mode MRM detection (Table 1) of protonated FA and [13 C5 ]-FA molecules was optimized using a combination of direct-infusion and flow-injection MS/MS studies. Researchers have suggested that FA has better ionization characteristics (formation of more intense molecular ions) with negative-ion mode MS/MS compared to positive-ion mode MS/MS [28,29]. This is supported by the observation that deprotonated FA molecules show lower amounts of cationization and suffer less signal suppression than protonated FA molecules during MS/MS analysis [29]. However, this observation has not been demonstrated under traditional reversed-phase LC/MS/MS conditions. Reversed-phase LC column and mobile phase conditions that are nominally compatible (provide both good chromatographic retention and formation of abundant molecular ions) for FA in negative-ion mode have not been reported. Under reversed-phase conditions, FA must be protonated and non-ionized (Fig. 1) in order for it to be effectively retained oncolumn. The pKa’s of the ␣-COOH and ␥-COOH on FA are 3.5 and 4.8, respectively [30]. Protonation is characteristically achieved via addition of a discrete level of a volatile organic acid (acetic acid, formic acid, propionic acid, etc.) to the LC mobile phase; the protons from the acid ensure that FA exists as predominantly uncharged molecules in solution (during chromatographic separation) and as positively charged ions in the gas phase (during positive-ion mode ESI analysis). The presence of free protons in solution is intuitively advantageous for positive-ion mode ESI. It has recently been documented that gas phase analyte deprotonation during negative-ion mode ESI analysis can also be achieved via addition of low (0.1 ␮mol/L

B.C. Nelson et al. / J. Chromatogr. A 1135 (2006) 203–211

to 100 mmol/L) levels of volatile organic acids to the mobile phase [31,32]. In negative-ion mode ESI, protons from water and/or the analytes are electrochemically reduced to hydrogen gas at the ESI spray tip [32]. This process generates an excess of negative charge on the resulting droplet surface. The additional source of protons (free protons) coming from the organic acid modifier further enhances the electrochemical reduction of H+ at the ESI spray tip, leading to an even greater level of negative charge on the droplet surface (the droplet surface has increased basicity). The presence of the excess negative charge on the droplet surface promotes efficient gas phase transfer of protons from the analyte (deprotonation) to the acid anion. Further, the acetate anion has the highest gas phase proton affinity (r G◦ = 1427.0 kJ/mol ± 8.4 kJ/mol) [32] of the common acid modifiers which suggested that acetic acid might be ideal for the negative-ion mode analysis of FA. Direct-infusion of FA dissolved in 50/50 water/methanol containing 0.1% acetic acid confirmed the presence of intense deprotonated FA molecules (m/z 440) in full-scan mode. Control experiments based on directinfusion of FA in 50/50 water/methanol produced no detectable deprotonated FA molecules. Preliminary MRM transitions for FA (m/z 440 → m/z 311) and [13 C5 ]-FA (m/z 445 → m/z 311) were obtained via direct-infusion of the analytes in the acetic acid solvent. On the basis of these initial data, development of a negative-ion mode LC/MS/MS method was evaluated using acetic acid and the HS-F5 LC column. LC mobile phases consisting of 0.01% (1.8 mmol/L) to 0.1% (18 mmol/L) acetic acid in 50/50 water/methanol (500 ␮L/min) were tested for optimal retention and sensitivity in negative-ion mode. FA was sufficiently retained (capacity factor ≥1.0) at all concentrations of acetic acid; however, peak tailing was noted with acetic acid concentrations less than 0.1%. Strong MRM transitions for FA were observed using acetic acid from 0.01 to 0.1%; however, because FA peak tailing was not significant using 0.1% acetic acid, this concentration was selected as the modifier concentration for the analytical method. Using 0.1% acetic acid in 50/50 water/methanol, development of the negative-ion mode LC/MS/MS method was completed by fine-tuning the LC conditions and MS/MS instrumental parameters as described in Section 3 and in Table 1. 4.2. Development of FA extraction procedures Successful quantification of FA in multivitamin formulations requires careful selection of extraction solvents and extraction procedures. Extensive review of the literature revealed no consensus regarding optimal extraction protocols [11–20,22,33,34]. Protocols utilizing aqueous or organic solvent extraction solutions (acidic, alkaline or neutral pH) combined with room temperature shaking, stirring, sonication and/or high-temperature (50 or 100 ◦ C) procedures have all been reported. Folates are generally unstable compounds that decompose or undergo form interconversion depending on their physical state. FA, in crystalline form, is known to decompose ≤1%/year under normal storage conditions (20 ◦ C, 65% humidity, protected from light and oxygen) [35]. However, FA in solution (water solubility = 1600 ng/mL at 25 ◦ C) [36] rapidly decomposes

207

Table 2 Extraction solvents tested on SRM 3280a Solvent ID

Extraction solvent compositionb

FA/[13 C5 ]-FA area ratio

A B C D E F G H I J

1% Formic acid/0.1% AA 0.1% Phosphoric acid/0.1% AA 0.1% Formic acid/0.1% AA 0.1% Acetic acid/0.1% AA 0.1% Hydrochloric acid/0.1% AA 0.1% Formic acid/0.1% AA (5 min boiling) 0.1% DTT in methanol 0.1% DTT in water 0.1% DTT in 50/50 water/methanol 0.3% Ammonium hydroxide/0.1% DTT

0.25 0.46 0.47 0.64 0.81 0.89 0.90 0.97 1.03 1.04

a b

All extraction solvents were degassed with helium for 15 min prior to use. AA, ascorbic acid; DTT, dithiothreitol.

when exposed to ultraviolet light, oxygen, trace metals and/or increased temperature [26,37]. If experimental conditions are not designed to prevent or reduce exposure to the previously listed physical factors, FA in solution will irreversibly degrade to p-aminobenzoyl-l-glutamic acid (pABG) and 6-formylpterin [38]. Extraction of FA from the SRM was evaluated using a limited selection of aqueous and aqueous/methanol extraction solvents (Table 2). FA is slightly soluble in methanol [36], and methanol is fully compatible with ESI MS. Other potential organic solvents, such as dimethylformamide, pyridine and phenol provide greater solubility for FA [36]; however, these solvents pose greater health risks and are not compatible with ESI MS. The extraction efficiency of the test solvents was determined by adding a known amount of [13 C5 ]-FA to homogenized tablet samples and monitoring the resulting FA/[13 C5 ]-FA area ratio in fully processed samples. Successful execution of the extraction study is based on having prior knowledge of the approximate amount of analyte in a given test sample (for SRM 3280, the approximate FA level was given as 400 mg/kg). With this knowledge, a known amount of isotope-labeled internal standard can be spiked into the test sample to produce an expected analyte/internal standard (FA/[13 C5 ]-FA) response ratio close to unity. For a given test solvent, the closer the analyte/internal standard response ratio is to unity, the better the extraction efficiency. For these experiments, the samples were prepared, extracted and analyzed according to the procedures given in Section 3 for the positive-ion mode LC/MS/MS method, except that each sample was subjected to vibrational shaking instead of rotary mixing. AA or DTT was added to each extraction solvent to prevent oxidative decomposition of FA in solution; AA is effective for acidic pH solutions, and DTT is effective for neutral pH or alkaline pH solutions. The area ratio results (Table 2) indicated that the best extraction solvents for FA were either neutral or alkaline pH solvents (solvents H, I, J). Acidic pH solvents (solvents A–F) resulted in area ratios (0.25 to 0.89) lower than the area ratios (0.97 to 1.04) observed for neutral or alkaline pH solvents. Solvent G was also a neutral pH solvent, however, the MRM peaks for FA and [13 C5 ]-FA in solvent G were more than 1 min in width, and the absolute area counts were more than six times lower than the absolute area counts observed for the analytes in the other neu-

208

B.C. Nelson et al. / J. Chromatogr. A 1135 (2006) 203–211

tral or alkaline pH solvents. Thus, solvent G was not considered an appropriate solvent for FA extraction. In general, as the acidity of the extraction solvent increased, the extraction ratio decreased (see Table 2, solvents H, C, A). This indicates that the solubility of FA tends to decreases with decreasing solvent pH as reported previously [37]. However, the current data suggests that some acids solubilize FA better than other acids (hydrochloric acid > acetic acid > formic acid/phosphoric acid). Interestingly, addition of thermal energy (boiling at 100 ◦ C) to solvent C (0.1% formic acid) significantly increased the observed extraction ratio (solvent F). Additionally, the absolute area counts for FA and [13 C5 ]-FA increased by a factor of two with the use of solvent F. Boiling the sample improved the extraction of FA, and the addition of thermal energy to the extraction process did not lead, as expected, to complete decomposition of FA and [13 C5 ]-FA; this observation is likely due to the presence of AA and to the absence of oxygen in the extraction solvent. Among all of the tested solvents, solvents I and J produced the highest area ratios (≈1.0), indicating that these two solvents were the best solvents for the extraction of FA from the SRM. Solvent I (weakly alkaline pH solvent) produced higher absolute area counts than solvent J (strongly alkaline pH solvent) for FA and [13 C5 ]-FA using positive-ion mode MRM detection, while this trend was reversed using negative-ion mode MRM detection. In actuality, the final sample pH of the tablet extract (pH ≈ 6) using solvent I was acidic, while the final sam-

ple pH of the tablet extract (pH ≈ 10) using solvent J remained strongly alkaline. Undoubtedly, the solution characteristics (pH) of the tablet extracts moderate the ionization of the analytes. Consequently, solvent I was deemed the appropriate extraction solvent for extracting and quantifying FA using positive-ion mode LC/MS/MS and solvent J was better suited for negativeion mode LC/MS/MS analysis. 4.3. Confirmation of extraction completeness The type of extraction procedure, vibrational shaking versus rotational mixing, was investigated for each LC/MS/MS method. No significant differences between the extraction ratios were observed, indicating that both extraction procedures were equivalent when extracting single samples. The repeatability of the vibrational shaking procedure was not consistent (due to sample capacity limitations) when extracting batches of five or more samples, whereas the repeatability of the rotary mixing procedure was consistent within batches. Therefore, rotational mixing was adopted as the standard extraction procedure for both LC/MS/MS methods. The appropriate length of time for extracting samples was determined using timed ([0,15,30,45,60] min) extraction studies. The FA/[13 C5 ]-FA extraction ratio for the positive-ion method was maximal and stable from the 30 min time point onward. The extraction ratio for the negative-ion method was maximized from the 15 min time point onward. Additional testing was conducted to verify complete extraction

Fig. 2. A representative positive-ion mode CID spectrum (product ion spectrum) of a FA standard (200 ng/␮L). The ions located at m/z 424 (a), m/z 313 (b) and at m/z 295 (c) can be tentatively identified as [M + H − H2 O]+ , [M + H − glutamic acid]+ and [M + H − glutamic acid − H2 O]+ , respectively. All profiles were collected using the LC/MS/MS conditions described in Section 3.

B.C. Nelson et al. / J. Chromatogr. A 1135 (2006) 203–211

209

Fig. 3. A representative negative-ion mode CID spectrum (product ion spectrum) of a FA standard (200 ng/␮L). The ions located at m/z 422 (a), m/z 396 (b), m/z 378 (c), m/z 311 (d) and at m/z 293 (e) can be tentatively identified as [M − H − H2 O]− , [M − H − CO2 ]− , [M − H2 O − CO2 ]− , [M − H − glutamic acid]− and [M − H − glutamic acid − H2 O]− , respectively. All profiles were collected using the LC/MS/MS conditions described in Section 3.

of FA from the SRM. For both the positive-ion and negativeion methods, samples were continuously extracted for 24 h using solvent I and solvent J, respectively. For the positiveion method, there were no significant differences between the 30 min or 24 h FA/[13 C5 ]-FA area ratios indicating that complete (100%) extraction of FA was achieved within the optimized 30 min extraction period. For the negative-ion method, there were no significant differences between the 15 min or 24 h FA/[13 C5 ]-FA area ratios indicating that complete extraction of FA was achieved within the optimized 15 min extraction period. 4.4. Confirmation of FA identity The identity of FA in SRM 3280 samples was confirmed on the basis of CID experiments using both LC/MS/MS methods and comparing the CID spectrum for a FA standard solution against the CID spectrum for the analysis of a sample extract. For the positive-ion mode method, the protonated FA molecule (m/z 442) was fragmented to produce characteristic positiveions in the standard and sample at m/z 424, m/z 313 and m/z 295. A representative positive-ion mode spectrum of FA from the analysis of a standard is shown in Fig. 2. For the negativeion mode method, the deprotonated FA molecule (m/z 440) was fragmented to produce characteristic negative-ions in the standard and sample at m/z 422, m/z 396, m/z 378, m/z 311 and m/z

293. A representative negative-ion mode spectrum of FA from the analysis of a standard is shown in Fig. 3. 4.5. Method performance characteristics The detection and quantification sensitivity for each LC/MS/MS method was characterized using serially prepared FA linearity standards (see Section 3). The instrument response for FA was linear over more than three orders of magnitude using either positive-ion or negative-ion mode detection. Complete data regarding the analytical linear dynamic range, LOD and LOQ sensitivity for each LC/MS/MS method are provided in Table 3. 4.6. Quantification of FA in NIST SRM 3280 Each method was independently applied to the quantification of FA in SRM samples. Twelve bottles of the SRM were opened and the tablets were homogenized and extracted as described in Section 3 (six bottles per method). Characteristic chromatograms from a tablet analysis using positive-ion mode LC/MS/MS are shown in Fig. 4. The figure shows the analyte and internal standard MRM ion channels and the UV absorbance chromatogram. No interferences were detected on either MRM ion channel (Fig. 4A and B). The UV channel (Fig. 4C) shows a peak eluting at the retention time of FA; however, the leading

210

B.C. Nelson et al. / J. Chromatogr. A 1135 (2006) 203–211

Table 3 Method performance characteristics Analytical parameter

FA (ng) by (+) LC/MS/MS

FA (ng) by (−) LC/MS/MS

Linear dynamic rangea LODa,d LOQa,e

0.02–73b 0.02 0.06

0.02–293c 0.02 0.06

a Determined via analysis of serially diluted calibrants as described in Section 3. All FA values are masses of analyte injected on-column. b Calculated r2 = 1.0, slope = 0.1082 (0.0003), y-intercept = 0.0050 (0.0008), SE estimate for regression line = 0.0025. Values in parentheses signify the standard error (SE). c Calculated r2 = 1.0, slope = 0.1056 (0.0004), y-intercept = 0.0077 (0.0031), SE estimate for regression line = 0.0107. Values in parentheses signify the standard error (SE). d The LOD is the minimum detectable analyte signal that is at least three times the noise signal. e The LOQ is calculated by multiplying the LOD by a factor of three. The S:N ratio for FA at the listed LOQ is ≥10.

edge of the peak has a skewed shape, suggesting a co-eluting interferent. Characteristic chromatograms from a tablet analysis using negative-ion mode LC/MS/MS are shown in Fig. 5. No interferences were detected on either the analyte or internal standard MRM ion channels (Fig. 5A and B). The UV channel (Fig. 5C) shows a peak eluting at the retention time of FA; however, the peak has a closely eluting interferent that would make accurate integration difficult. Analytical results for both methods are given in Table 4. The precision of each method is excellent as shown by overall RSDs ≤ 1%. The mean FA levels, 401 and 413 mg/kg, determined by positive-ion mode and negativeion mode LC/MS/MS, respectively, differ by approximately 3%. The manufacturer’s assessment of the FA level in SRM 3280 was stated as 630 ␮g FA per tablet. This correlates to a relative accuracy of 96 and 98%, for the positive-ion mode and negative-ion mode LC/MS/MS methods, respectively. The small measurement SDs and the strong correlation between both LC/MS/MS determinations and the manufacturer’s FA assessment, combined with the small difference between method means, supports the overall accuracy and reliability of each method. Addition-

Fig. 5. Typical negative-ion mode MRM chromatograms and ultraviolet absorbance chromatogram of FA extracted from SRM 3280. (A) MRM chromatogram of unlabeled FA. (B) MRM chromatogram of [13 C5 ]-FA. (C) UV absorbance chromatogram of tablet extract monitored at 282 nm. All profiles were collected using the LC/MS/MS conditions described in Section 3. Table 4 Quantification of FA in NIST SRM 3280 using LC/MS/MSa

Overall mean SD Overall RSD (%) ␮g FA/tabletb,c

Bottle #

FA (mg/kg) by (+) LC/MS/MS

Bottle #

FA (mg/kg) by (−) LC/MS/MS

1 2 3 4 5 6

392 ± 0.88 404 ± 1.1 406 ± 3.9 401 ± 1.4 404 ± 1.0 400 ± 3.4

7 8 9 10 11 12

411 ± 1.2 412 ± 1.5 412 ± 3.1 417 ± 1.4 415 ± 3.3 412 ± 1.5

401 4.9 1 602

413 2.7 1 620

a All values have been corrected for the consensus purity value (99.7%) for the folic acid primary reference standard. Each reported value represents the determination (mean ± SD) from two independent sample preparations. b Calculated based on a nominal tablet mass of 1500 mg. c The mean level of FA in SRM 3280 as determined by the manufacturer was 630 ␮g/tablet. This level was determined via LC/UV analysis.

ally, microbiologic assay of SRM samples (n = 8) returned an FA level (mean ± SD) of 399 mg/kg ± 38 mg/kg; the mean microbiologic FA determination is highly concordant with both mean LC/MS/MS FA determinations. The data provided in Table 4 will be combined with results reported by collaborating laboratories to generate the final certified value for FA in SRM 3280. 5. Conclusions

Fig. 4. Typical positive-ion mode MRM chromatograms and ultraviolet absorbance chromatogram of FA extracted from SRM 3280. (A) MRM chromatogram of unlabeled FA. (B) MRM chromatogram of [13 C5 ]-FA. (C) UV absorbance chromatogram of tablet extract monitored at 282 nm. All profiles were collected using the LC/MS/MS conditions described in Section 3.

The recommended daily intake (RDI) for FA lies between 400 and 1000 ␮g/day [39,40]. A minimal level of 400 ␮g/day has been shown to prevent or reverse the effects of folate deficiency [40]. The daily dose (one tablet) of FA provided by SRM 3280 (Table 4) ranges from 602 to 620 ␮g, levels concordant with the RDI range. Two potential reference methods using isotopedilution LC/MS/MS for the quantitative determination of FA in multivitamin/multielement tablets have been developed, validated and applied to the measurement of FA in SRM 3280. Each method uses substantially different FA extraction (aque-

B.C. Nelson et al. / J. Chromatogr. A 1135 (2006) 203–211

ous/organic versus aqueous) and detection (positive-ion MRM mode versus negative-ion MRM mode) conditions which allows for measurement independence and method cross-validation. The LC/MS/MS methods are much faster than microbiological assays, and the methods do not require the utilization of chelating agents or the use of heat during the extraction of FA as required by several of the recently reported LC/UV-based methods. The methods have LOQs that are as good as or dramatically better than existing LC methods combined with UV or MS detection. Since both the positive-ion and negative-ion mode LC/MS/MS methods demonstrate comparable performance characteristics, either method could potentially serve as an independent reference method for the quantitative determination of FA in multivitamin/multielement tablet formulations. Acknowledgements The authors gratefully acknowledges Nathan Dodder (NIST, Gaithersburg, Maryland) and Mary Frances Picciano (NIH, Bethesda, Maryland) for their critical reviews of the manuscript during its preparation. The work reported in this manuscript was partially funded through the NIH – Office of Dietary Supplements. References [1] [2] [3] [4] [5] [6] [7]

[8] [9] [10] [11] [12]

J.G. Donnelly, Crit. Rev. Clin. Lab. Sci. 38 (2001) 183. A.M. Molloy, J.M. Scott, Public Health Nutr. 4 (2001) 601. DHHS, FDA, Fed. Regist. 61 (1996) 8781. G. Andria, I. Scala, G. Sebastio, Nutr. Metab. Cardiovasc. Dis. 15 (2005) 87. M.C. Cornel, D.J. de Smit, L.T.W. de Jong-van den Berg, Reproductive Tox. 20 (2005) 411. M. Lucock, Z. Yates, Nature Rev. 6 (2005) 235. R. Clarke, J.G. Evans, J. Schneede, E. Nexo, C. Bates, A. Fletcher, A. Prentice, C. Johnston, P.M. Ueland, H. Refsum, P. Sherliker, J. Birks, G. Whitlock, E. Breeze, J.M. Scott, Age Ageing 33 (2004) 34. H. McNulty, K. Pentieva, Proc. Nutr. Soc. 63 (2004) 529. A. Melse-Boonstra, P. Verhoef, C. West, Curr. Opin. Clin. Nutr. Metab. Care 7 (2004) 539. W. Horwitz, Official Methods of Analysis of AOAC International, AOAC International, Gaithersburg, 2000. R. Amidzic, J. Brboric, O. Cudina, S. Vladimirov, J. Serb. Chem. Soc. 70 (2005) 1229. I.J. Holcomb, S.A. Fusari, Anal. Chem. 53 (1981) 607.

211

[13] U. Holler, C. Brodhag, A. Knobel, P. Hofmann, V. Spitzer, J. Pharm. Biomed. Anal. 31 (2003) 151. [14] M.A. Kall, P. Norgaard, S.J. Pedersen, T. Leth, J. Pharm. Biomed. Anal. 23 (2000) 437. [15] G. Klaczkow, E. Anuszewska, Acta Pol. Pharm. 57 (2000) 257. [16] B. Kledjus, J. Petrlova, D. Potesil, V. Adam, R. Mikelova, J. Vacek, R. Kizek, V. Kuban, Anal. Chim. Acta 520 (2004) 57. [17] L.A. Kozhanova, G.A. Fedorova, G.I. Baram, J. Anal. Chem. 57 (2002) 40. [18] P. Stokes, K. Webb, J. Chromatogr. A 864 (1999) 59. [19] W.H. Tafolla, A.C. Sarapu, G.R. Dukes, J. Pharm. Sci. 70 (1981) 1273. [20] P.F. Chatzimichalakis, V.F. Samanidou, R. Verpoorte, I.N. Papadoyannis, J. Sep. Sci. 27 (2004) 1181. [21] Z. Chen, B. Chen, S. Yao, Anal. Chim. Acta 569 (2006) 169. [22] M. Rychlik, Anal. Chim. Acta 495 (2003) 133. [23] J.T. Dwyer, M.F. Picciano, J.M. Betz, K. Fisher, L.G. Saldanha, E.A. Yetley, P.M. Coates, K. Radimer, B. Bindewald, K.E. Sharpless, J. Holden, K. Andrews, C. Zhao, J. Harnley, W. Wolf, C. Perry, J. Food Comp. Anal. 19 (2006) S108. [24] J.M. Holden, K.W. Andrews, C. Zhao, A.L. Schweitzer, C.R. Perry, J. Harnly, W. Wolf, J.T. Dwyer, M.F. Picciano, J.M. Betz, L.G. Saldanha, E. Yetley, K. Fisher, K.E. Sharpless, K. Radimer, B. Bindewald, FASEB J. 19 (2005) A86. [25] B.C. Nelson, M.B. Satterfield, L.T. Sniegoski, M.J. Welch, Anal. Chem. 77 (2005) 3586. [26] R.L. Blakley, The Biochemistry of Folic Acid and Related Pteridines, John Wiley & Sons, New York, 1969. [27] B.C. Nelson, J.J. Dalluge, S.A. Margolis, J. Chromatogr. B 765 (2001) 141. [28] C.M. Pfeiffer, Z. Fazili, L. McCoy, M. Zhang, E.W. Gunter, Clin. Chem. 50 (2004) 423. [29] S.D. Garbis, A. Melse-Boonstra, C.E. West, R.B. van Breemen, Anal. Chem. 73 (2001) 5358. [30] J.D.M. Patring, S.A. Lanina, J.A. Jastrebova, J. Sep. Sci. 29 (2006) 889. [31] N.B. Cech, C.G. Enke, Mass Spectrom. Rev. 20 (2001) 362. [32] Z. Wu, W. Gao, M.A. Phelps, D. Wu, D.D. Miller, J.T. Dalton, Anal. Chem. 76 (2004) 839. [33] M.S. Aurora-Prado, C.A. Silva, M.F.M. Tavares, K.D. Altria, J. Chromatogr. A 1051 (2004) 291. [34] C. Paveenbampen, D. Lamontanaro, J. Moody, J. Zarembo, C. Rehm, J. Pharm. Sci. 75 (1986) 1192. [35] A. Vora, A. Riga, D. Dollimore, K. Alexander, J. Therm. Anal. Calorimetry 75 (2004) 709. [36] S. Budavari, The Merck Index, Merck & Co, Whitehouse Station, 1996. [37] J.F. Gregory, Chemical and Nutritional Aspects of Folate Research: Analytical Procedures, Methods of Folate Synthesis, Stability, and Bioavailability of Dietary Folates, Academic Press, San Diego, 1989. [38] M.K. Off, A.E. Steindal, A.C. Porojnicu, A. Juzeniene, A. Vorobey, A. Johnsson, J. Moan, J. Photochem. Photobiol. 80 (2005) 47. [39] S.F. Choumenkovitch, J. Selhub, P.W.F. Wilson, J.I. Rader, I.H. Rosenberg, P.F. Jacques, J. Nutr. 132 (2002) 2792. [40] L.B. Bailey, Nutr. Rev. 62 (2004) S14.