Simultaneous Determination of Six Mercapturic Acid Metabolites of

1018

Chem. Res. Toxicol. 2009, 22, 1018–1025

Simultaneous Determination of Six Mercapturic Acid Metabolites of Volatile Organic Compounds in Human Urine Yan S. Ding, Benjamin C. Blount,* Liza Valentin-Blasini, Heather S. Applewhite, Yang Xia, Clifford H. Watson, and David L. Ashley Emergency Response and Air Toxicants Branch, DiVision of Laboratory Sciences, National Center for EnVironmental Health, Centers for Disease Control and PreVention, 4770 Buford Highway, NE, Mailstop F-47, Atlanta, Georgia 30341-3724 ReceiVed December 10, 2008

The widespread exposure to potentially harmful volatile organic compounds (VOCs) merits the development of practical and accurate exposure assessment methods. Measuring the urinary concentrations of VOC mercapturic acid (MA) metabolites provides noninvasive and selective information about recent exposure to certain VOCs. We developed a liquid chromatography-tandem mass spectrometry method for quantifying urinary levels of six MAs: N-acetyl-S-(2-carboxyethyl)-L-cysteine (CEMA), N-acetyl-S(3-hydroxypropyl)-L-cysteine (HPMA), N-acetyl-S-(2-hydroxy-3-butenyl)-L-cysteine (MHBMA), N-acetylS-(3,4-dihydroxybutyl)-L-cysteine (DHBMA), N-acetyl-S-(2-hydroxyethyl)-L-cysteine (HEMA), and N-acetyl-S-(phenyl)-L-cysteine (PMA). The method provides good accuracy (102% mean accuracy) and high precision (3.5% mean precision). The sensitivity (limits of detection of 0.01-0.20 µg/L) and wide dynamic detection range (0.025-500 µg/L) make this method suitable for assessing VOC exposure of minimally exposed populations and those with significant exposures, such as cigarette smokers. We used this method to quantify MA levels in urine collected from smokers and nonsmokers. Median levels of creatinine-corrected CEMA, HPMA, MHBMA, DHBMA, HEMA, and PMA among nonsmokers (n ) 59) were 38.1, 24.3, 21.3, 104.7, 0.9, and 0.5 µg/g creatinine, respectively. Among smokers (n ) 61), median levels of CEMA, HPMA, MHBMA, DHBMA, HEMA, and PMA were 214.4, 839.7, 10.2, 509.7, 2.2, and 0.9 µg/g creatinine, respectively. All VOC MAs measured were higher among smokers than among nonsmokers, with the exception of MHBMA. Introduction Volatile organic compounds (VOCs) are chemicals with sufficient vapor pressures to readily vaporize at ambient temperature and pressure. VOCs are ubiquitous, originating from many different natural and anthropogenic sources. They are present in virtually all homes and workplaces. Humans may be exposed to VOCs through inhalation, ingestion, and dermal contact (1). Environmental exposure to VOCs is a public health concern (2) because chronic exposure to some VOCs may increase the risk of leukemia (3), bladder cancer (4), birth defects (5), and neurocognitive impairment (6). One avoidable source of substantial VOC exposure is tobacco smokesthe primary nonoccupational source of exposure to some VOCs (e.g., benzene) in the United States (7). Tobacco smoke contains more than 4000 chemicals, including carcinogenic and toxic VOCs (e.g., benzene, vinyl chloride, ethylene oxide, 1,3-butadiene, acrolein, methanol, and pyridine) (8, 9). The amount of many individual VOCs in mainstream tobacco smoke is typically in the µg per cigarette range. A risk assessment by Fowles and Dybing on chemical constituents in cigarette smoke suggested that mainstream smoke gas phase constituents contribute heavily toward the cancer risk indices (10). The tobacco industry has taken several actions (e.g., charcoal filtration, filter ventilation, and high porosity wrapping paper) to reduce the levels of VOC emission in cigarette mainstream smoke (11). Biomonitoring methods are needed to confirm that reduced levels of VOCs in smoke lead to reduced levels of VOCs in smokers. * To whom correspondence should be addressed. Tel: 770-488-7894. Fax: 770-488-0181. E-mail: [email protected].

10.1021/tx800468w CCC: $40.75

The widespread exposure to VOCs merits further research, including developing improved methods for assessing internal dose. Interpreting exposure estimates is complicated because of the variability among humans in the capacity to absorb, distribute, metabolize, and excrete VOCs. Therefore, measuring internal dose is often the best means to assess the impact of human exposure to environmental toxicants, including VOCs that could lead to adverse health outcomes (12). Previous methods for quantifying internal dose levels have focused on measuring VOCs (and their metabolites) in breath, blood, and urine. Immediately after exposure, VOCs can be detected in exhaled breath (13-16). Such measurement techniques are noninvasive, but VOC levels in breath decrease quickly after exposure. Assessing VOCs in blood may yield more complete data from a matrix more similar to target tissue(s) (17, 18); however, these levels can also decrease markedly following the cessation of exposure (19). Measuring VOCs directly in urine may underestimate exposure because evaporative loss may occur during collection or storage. In general, VOC metabolites in urine are less volatile than the parent compounds and can serve as more stable biomarkers, although some of these compounds (e.g., mandelic acid and phenol) lack specificity. Mercapturic acids (MAs) provide a relatively high degree of specificity. The MAs are formed from glutathione (GSH) S-conjugates via the MA pathway. When GSH S-conjugates metabolize to the corresponding MAs, γ-glutamyl and glycinyl residues are removed from the GSH S-conjugate by γ-glutamyltranspeptidase and dipeptidases, with subsequent acetylation to MAs (N-acetyl-L-cysteine S-conjugates) by cysteine S-conjugate

This article not subject to U.S. Copyright. Published 2009 by American Chemical Society. Published on Web 04/23/2009

Quantifying Six Urinary Mercapturic Acids

N-acetyltransferase. Upon exposure, many VOCs form GSH adducts that are further metabolized and excreted as one or more corresponding MAs. Measuring the concentration of these MAs can help identify recent VOC exposure (20). Depending on the parent compound, the amount of an MA can be relatively high, thus allowing researchers to determine its excretion even in cases of low levels of exposure. The main advantages of using MAs to assess human VOC exposure are listed below: noninvasiveness of urine sampling, ability to work with physiological halflives longer than those of the parent VOCs in blood or breath, and sufficiency of nonpersistence that allows for establishing direct relationships between internal dose and specific activities. Therefore, MAs can serve as a quantitative measure of human exposure to certain VOCs. Numerous methods for detecting urinary MAs have been previously developed and applied (20, 21). Analytical techniques range from immunoassay to gas or liquid chromatography coupled with either ultraviolet, fluorescence, or mass spectral detection (20-22). Among these methods, the limits of detection (LOD) vary dramatically (20, 21). Many of these methods focus only on a single MA from a specific parent compound. In reality, most major environmental sources of VOC exposure (e.g., tobacco smoke or automobile exhaust) are comprised of many different VOCs. Only a few methods found in the literature measure multiple MAs; however, these multiple MA measurements typically determine analytes derived from the same parent compound (23, 24). Although not every MA is specific to one VOC, the same MA may be excreted in urine when a person is exposed to different or structurally similar parent compounds. For example, acrylonitrile, vinyl chloride, and ethylene oxide can all be metabolized to N-acetyl-S-(2-hydroxyethyl)-L-cysteine (HEMA) and excreted in urine (20). Although urinary N-acetylS-(3-hydroxypropyl)-L-cysteine (HPMA) is one of the major metabolites of acrolein, it can also be used as a more general biomarker of chemical exposure because acrolein can be an intermediate metabolite of various industrial compounds (20) or endogenous sources (25). This lack of specificity resulting from the multiple sources of exposure can complicate the interpretation of experimental data. Measuring multiple biomarkers of different parent compounds can provide complementary information (26, 27). These studies often require an individual analytical method for each MA, increasing cost and complexity to achieve high throughput. Therefore, quantifying multiple VOC metabolites from multiple parent compounds in a single analysis is efficient for assessing overall internal dose of the mixtures of VOCs to which humans are often exposed. Simultaneously measuring multiple toxicologically related biomarkers can provide a stronger data set for assessing the potential health risk of exposure to a mixture of environmental chemicals. Therefore, we have developed a high-throughput method for quantitative analysis of six MAs from select parent compounds. We can analyze 80 samples per day on our instrument set up. These six MAs include HPMA and N-acetylS-(2-carboxyethyl)-L-cysteine (CEMA) (MA metabolites of acrolein); N-acetyl-S-(3,4-dihydroxybutyl)-L-cysteine (DHBMA) and N-acetyl-S-(2-hydroxy-3-butenyl)-L-cysteine (MHBMA) (MA metabolites of 1,3 butadiene); N-acetyl-S-(phenyl)-Lcysteine (PMA) (MA metabolite of benzene); and HEMA (MA metabolites of acrylonitrile, vinyl chloride, ethylene oxide, etc.). A wide dynamic detection range makes this method suitable for assessing VOC exposure in minimally exposed populations and in those with higher exposure, such as cigarette smokers.

Chem. Res. Toxicol., Vol. 22, No. 6, 2009 1019

Experimental Procedures Source of Materials. CEMA, CEMA-13C3, HPMA, HPMA-d6, N-acetyl-S-(1-hydroxymethyl-2-propenyl)-L-cysteine (MII), N-acetylS-(1-hydroxymethyl-2-propenyl-d6)-L-cysteine (MII-d6), DHBMA, HEMA, HEMA-d4, PMA, and PMA-13C6 were custom-synthesized by Cambridge Isotope (Andover, MA) with purity greater than 95%. DHBMA-d7 (98% purity) was purchased from Toronto Research Chemicals (Toronto, Ontario, Canada). All solvents were highperformance liquid chromatography (HPLC)-grade. Standard and Internal Standard Solution. All neat and isotopically labeled materials were dissolved in HPLC-grade water to obtain individual stock solutions. A standard solution containing six MAs was prepared in water and aliquoted. The concentrations of the six MAs varied from 50 to 1000 µg/L. Standard solutions were prepared daily by dilution with a mobile phase solution to final concentrations covering the linear range of the assay. An internal standard solution containing six isotopically labeled MA analogues was also prepared in water, with the concentrations of six labeled MAs ranging from 100 to 480 µg/L. Both standard and internal standard solutions were stored at 4 °C. Sample Collection. Spot urine samples were collected from local, anonymous donors (59 nonsmokers and 61 smokers) and stored at -70 °C until analysis. The study protocol was reviewed and approved by the CDC institutional review board. All subjects gave informed written consent before participating in the study. The smoking status was determined from self-reporting when urine was anonymously donated. An enzyme-based colorimetric method on a Roche Hitachi 912 Chemistry Analyzer was used to assess the creatinine concentration in urine (28), and these data were used to generate creatininecorrected urinary MA data. We analyzed total urinary 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanol (NNAL) among smokers according to a previously published method (29). Sample Preparation. Frozen urine samples were thawed to room temperature and mixed to suspend any settled precipitate. A 100 µL urine aliquot was transferred to an autosampler vial and spiked with 50 µL of an internal standard solution. The sample was diluted with a 0.5% acetic acid water solution to a final volume of 1 mL. A 10 µL aliquot was injected on-column for HPLC. Instrumental Analysis. Samples were analyzed by use of an Agilent 1100 liquid chromatograph (Agilent Technologies, Wilmington, DE) coupled with an API 4000 triple quadruple mass spectrometer (Applied Biosystems, Foster City, CA). Samples were injected on a Waters Xterra C18 MS column (4.6 mm × 50 mm i.d., 5 µm particle size, Waters Corp., Milford, MA). Solvent A was 0.5% acetic acid in water, and solvent B was acetonitrile. The solvent elution program used the following time and rate setting {[A] (%), rate (µL/min)}: 0-3.5 min (90, 300), 3.6-6 min (60, 500), and 6.1-12 min (90, 500). The mass spectrometer was operated in negative ion electrospray mode. Mass spectral data on precursor and product ions were collected in multiple reaction monitoring mode. The quantification/confirmation ion pairs, declustering potential, entrance potential, collision energy, and cell exit potential were optimized for each analyte. Data Analysis. The Analyst software version 1.4.1 (Applied Biosystems) was used to process peak area determinations for all samples, blanks, standards, and quality control (QC) materials. Each ion of interest in the reconstructed ion chromatogram was automatically selected and integrated. The peak integrations were manually inspected for errors and, if necessary, manually reintegrated. For each analyte, two pairs of transition ionssone for quantification and one for confirmationswere collected to verify analyte identity. The acceptable ratio of peak areas for quantifying and confirming transition ions of unknown samples was within 30% of that for QC materials. Method Validation. To assess method validity, we divided the urine pool sample into three portions. The first two portions were spiked with two different concentrations of standard solution. The third portion was used as the sample blank. Accuracy and precision were calculated from five replicate analyses of these three pools.

1020

Chem. Res. Toxicol., Vol. 22, No. 6, 2009

Table 1. Chemical Structures of MAs and Their Parent Compounds

Ding et al. Table 2. Multiple Reaction Monitoring Analysis of MAs and Their Internal Standardsa MA

ion pair b

CEMA HPMA MHBMA DHBMA HEMA PMA

234/162 234/128c 220/91b 220/89c 232/103b 233/104c 250/121b 250/128c 206/77b 206/75c 238/109b 239/110c

CEd -15 -15 -20 -30 -15 -15 -20 -16 -20 -40 -20 -20

internal standard

ion pair

CEd

CEMA-13C3

237/162

-16

226/97

-20

238/109

-20

257/128

-20

210/81

-22

244/115

-22

HPMA-D6 MHBMA-D6 DHBMA-D7 HEMA-D4 PMA-13C6

a Other parameters (V): declustering potential, -35; entrance potential, -9; and collision exit potential, -10. b Quantitation. c Confirmation. d CE (V), collision energy.

* HEMA is also formed from bromoethanol, chloroacetaldehyde, ethylene, chloroethylene, 1,2-dichloroethane, and 1,2-dibromoethane.

All samples were subjected to the standard sample preparation procedure and the subsequent HPLC-tandem mass spectrometry (MS/MS) analysis. Accuracy was calculated as the mean of the experimentally determined concentration from replicate analysis divided by the nominal concentration. Method precision was determined by calculating the relative standard deviations of five replicate measurements (intraday). We used an anonymously collected human urine sample (nonsmoker) to set target ranges for a QC pool at a low level (QCL). Another QC pool at a higher level (QCH) was prepared in a similar manner. These QC pools were aliquoted and stored at -20 °C until use. Interday precision was determined by calculation of the relative standard deviation of QC material analyzed 30 times during 7 months. We prepared proficiency testing samples by diluting standard mix solution to four different concentrations. We verified assay accuracy of these samples by blind analysis. Statistical Analysis. Levels of urinary MA and creatinineadjusted urinary MA were compared between nonsmoker and smoker groups by use of t-tests (Excel), and these levels were visually compared by use of box plots (SigmaPlot). Linear regression analysis (Excel) was used to analyze relationships between levels of urinary MA and NNAL among smokers. An analysis of variance (ANOVA) statistical procedure was used to calculate the relation coefficients and significance (Excel). All calculated p values less than 0.05 were considered statistically significant.

Results Method Development. Because the diverse chemical nature of substituted alkyl groups on the various analytes ranged from nonpolar to polar (Table 1), we evaluated HPLC columns with different properties to optimize chromatographic resolution, peak shape, and analysis time. The Waters Xterra C18 MS column provided optimal chromatographic resolution and peak shape.

We also optimized the column length, particle size, solvent elution conditions, and flow rates for optimal sensitivity, specificity, and throughput. The total run time per sample was 12 min, including equilibration. We monitored a second precursor/product ion pair to confirm analyte identity to maintain high specificity with the simple sample preparation (Table 2). The quantitation to confirmation ion peak area ratios for each of the six MA analyte were measured using reference materials and compared with the corresponding peak area ratios in unknown samples. Agreement of the ratio of responses by quantitation and confirmation transitions provided additional data confirming correct peak identification. Reconstructed ion chromatograms for MAs from human urine showed excellent sensitivity and chromatographic resolution (Figure 1). Urine samples from 59 nonsmokers and 61 smokers were analyzed to determine the initial target ranges of MAs. On the basis of the results, a 10-point calibration curve covering 3 orders of magnitude was constructed to quantify MA levels found in urine from both groups (Table 3). Linearity was tested (least-squares regression) for each analyte. The R2 values of all six MA standards were typically greater than 0.99. Method Validation. Blank samples were analyzed with every batch of unknown samples to exclude possible carryover and to detect systemic contamination. Measurements were also made to determine how much native analyte contributed to the isotopelabeled internal standard and vice versa. Cross-interference was insignificant. The detection limit (LOD) for each MA was calculated as three times the standard deviation at the lowest standard concentration; it was in the sub-µg/L range (Table 3). The method accuracy was assessed by spiking two concentrations of known amounts of MAs in nonsmoker urine. The mean accuracies for all analytes ranged from 95 to 109%. Intraday precision for all analytes was less than 8%, with an average of 3% (Table 3). In 30 measurements made during a 7 month interval, interday precision was less than 12% for QCL pool and less than 15% for QCH pool (Table 3). The validity of our method was also tested by purposely varying five factors that influenced assay accuracy. After testing different dilution factors, we concluded that 1:10 dilution was superior to either 1:5 (too much matrix effect) or 1:50 (sample too diluted). For the dilution solvent, we tested synthetic urine and mobile phase solution and water. We found that the concentrations of the highest standard were underestimated in synthetic urine because of ion suppression and possible salt effects. We found no difference between the mobile phase solution and the water. Therefore, the mobile phase solution

Quantifying Six Urinary Mercapturic Acids

Chem. Res. Toxicol., Vol. 22, No. 6, 2009 1021

Figure 1. Multiple reaction monitoring chromatograms of urinary MAs from a urine sample. CEMA, m/z 234 f 162; HPMA, m/z 220 f 91; MHBMA, m/z 232 f 103; DHBMA, m/z f 121; HEMA, m/z 206 f 77; PMA, m/z 238 f 109. The urinary CEMA, HPMA, MHBMA, DHBMA, HEMA, and PMA for this particular subject (smoker) are 5.72, 1360, 22.6, 965, 256, and 1.47 µg/L, respectively.

Table 3. Method Validation Parameters for Measuring Urinary MAs MA CEMA HPMA

standard range (µg/L) 0.3-300 0.5-500

MHBMA

0.075-75

DHBMA

0.4-400

HEMA

0.075-75

PMA

0.025-25

a

nominal concentration (µg/L)

mean accuracy (%)

precision (intraday) (%)

precision (interday) (%)

3 30 10 100 1.5 15 4 40 1.5 15 0.5 5

102 99 109 95 106 100 103 103 103 102 101 103

4.1 2.0 3.0 2.6 1.7 2.5 5.3 3.5 3.6 3.1 7.5 2.8

8a 9b 8a 12b 12a 10b 12a 13b 11a 15b 8a 8b

LODc (µg/L)

LOD (µg/L) (literature)

0.15 0.20 0.05 0.14 0.03 0.01

200d 0.9e 0.1f 5f 0.68g 0.04h

QCL. b QCH. c Detection limit. d Onkenhout et al. (40). e Carmella et al. (37). f Van Sittert et al. (30). g Barr et al. (44). h Lin et al. (34).

was used for sample dilution. Thoroughly mixing urine before taking an aliquot for analysis was crucial. Failure to follow this procedure yielded lower measured values. For sample integrity and stability, we tested QC samples after multiple freeze-thaw cycles. Our six MA analytes were stable in urine samples following 10 freeze-thaw cycles. Additionally, we found no changes in MA analyte levels in urine samples (pH range

6.5-7.2) stored at room temperature (18-21 °C) for a day or stored at 4 °C for 1 week. Although these analytes appear to be stable, urine composition and pH can vary quite broadly. Therefore, we recommend freezing samples as soon as possible following sample collection and minimizing subsequent freeze-thaw cycles to preserve analyte stability. We also tested the stability of working solutions of standards and isotopically

1022

Chem. Res. Toxicol., Vol. 22, No. 6, 2009

Ding et al.

Table 4. Levels of Selected MAs in Urine Collected from Nonsmokers (N ) 59) and Smokers (N ) 61) range (µg/L) [frequency of detection (%)] CEMA HPMA MHBMA DHBMA HEMA PMA a

creatinine-corrected range (µg/g) a

nonsmoker

smoker

P value

nonsmoker

smoker

P valuea

ND-94 (97%) ND-128 (75%) ND-73.4 (78%) ND-329 (93%) ND-1.44 (7%) ND-0.26 (2%)

29-1240 (100%) 80.9-4030 (100%) ND-132 (87%) 113-1830 (100%) ND-20.8 (84%) ND-37.7 (89%)