Anal. Chem. 2000, 72, 4127-4134
Quantitative Detection of Eight Phthalate Metabolites in Human Urine Using HPLC-APCI-MS/ MS Benjamin C. Blount, K. Eric Milgram,† Manori J. Silva, Nicole A. Malek, John A. Reidy, Larry L. Needham, and John W. Brock*
Anal. Chem. 2000.72:4127-4134. Downloaded from pubs.acs.org by AUCKLAND UNIV OF TECHNOLOGY on 01/28/19. For personal use only.
Division of Laboratory Sciences, National Center for Environmental Health, Centers for Disease Control and Prevention, Atlanta, Georgia 30341
Because of the ubiquity of phthalates and their potential role in increasing risk for cancer and reproductive dysfunction, the need for human exposure assessment studies is urgent. In response to this need, we developed a high-throughput, robust, sensitive, accurate, and precise assay for simultaneous measurement of trace levels of eight phthalate metabolites in human urine by HPLCMS/MS. Human urine samples were processed using enzymatic deconjugation of the glucuronides followed by solid-phase extraction. The eluate was concentrated, and the phthalate metabolites were chromatographically resolved by reversed-phase HPLC, detected by APCItandem mass spectrometry, and quantified by isotope dilution. This selective analytical method permits rapid detection (7.7 min total run time) of eight urinary metabolites of the most commonly used phthalates with detection limits in the low nanagram per milliliter range. Assay precision was improved by incorporating 13C4labeled internal standards for each of the eight analytes, as well as a conjugated internal standard to monitor deconjugation efficiency. This selective, sensitive, and rapid method will help elucidate potential associations (if any) between human exposure to phthalates and adverse health effects. The dialkyl or alkyl aryl esters of 1,2-benzenedicarboxylic acid, commonly called phthalates, have a myriad of commercial uses and are considered ubiquitous environmental contaminants. Globally, over 18 billion pounds of phthalates are used each year, primarily as additives to poly(vinyl chloride) (PVC) plastics, as industrial solvents, and as components of many consumer products. Humans are potentially exposed to many products containing phthalates, and specific subpopulations, such as medical patients undergoing transfusions, dialysis, or apheresis are potentially more heavily exposed.1-3 Measurement of an internal dose, or bio* Corresponding author: National Center for Environmental Health, Centers for Disease Control and Prevention, 4770 Buford Hwy. NE, Mailstop F-17, Atlanta, GA 30341; (e-mail)
[email protected]; (fax) 770-488-4609. † Current address: 3550 General Atomics Court, Alanex Division of Agouron Pharmaceuticals, San Diego, CA 92024. (1) Pollack, G. M.; Buchanan, J. F.; Slaughter, R. L.; Kohli, R. K.; Shen, D. D. Toxicol. Appl. Pharmacol. 1985, 79, 257-67. 10.1021/ac000422r Not subject to U.S. Copyright. Publ. 2000 Am. Chem. Soc.
Published on Web 07/27/2000
marker of exposure, is a key aspect of assessing exposure.4,5 Phthalates are lipophilic compounds but are rapidly metabolized in humans6 and therefore do not appear to bioaccumulate. In humans and animals, phthalates are metabolized to their respective monoesters (alkyl or aryl esters of 1,2-benzenedicarboxylic acid, commonly called phthalate monoesters) and further oxidative products, which are excreted through the urine and feces.6,7 In animal studies, several phthalates and/or their monoester metabolites act as potent reproductive8,9 and developmental10,11 toxicants. Despite demonstrated toxicity in animals, general human exposure to phthalate diesters and the monoester metabolites has not been well studied. Previous methods have attempted to directly measure the phthalate diesters in serum but were fraught with contamination problems when low baseline levels were measured.12,13 A total phthalate method in urine14 is limited by the potential for phthalate diester contamination, cumbersome derivatization, and no specific exposure data on individual phthalates or monoester metabolites. A more recent method for measuring urinary metabolites of di-2-ethylhexyl phthalate (DEHP) (2) Sjoberg, P. O.; Bondesson, U. G.; Sedin, E. G.; Gustafsson, J. P. Transfusion 1985, 25, 424-8. (3) Faouzi, M. A.; Dine, T.; Gressier, B.; Kambia, K.; Luyckx, M.; Pagniez, D.; Brunet, C.; Cazin, M.; Belabed, A.; Cazin, J. C. Int. J. Pharmacol. 1999, 180, 113-21. (4) Pirkle, J. L.; Sampson, E. J.; Needham, L. L.; Patterson, D. G.; Ashley, D. L. Environ. Health Perspect. 1995, 103 (Suppl 3), 45-8. (5) Sampson, E. J.; Needham, L. L.; Pirkle, J. L.; Hannon, W. H.; Miller, D. T.; Patterson, D. G.; Bernert, J. T.; Ashley, D. L.; Hill, R. H.; Gunter, E. W.; et al. Clin. Chem. 1994, 40, 1376-84. (6) Schmid, P.; Schlatter, C. Xenobiotica 1985, 15, 251-6. (7) Albro, P. W.; Hass, J. R.; Peck, C. C.; Jordan, S. T.; Corbett, J. T.; Schroeder, J. J Environ. Sci. Health. B 1982, 17, 701-14. (8) Davis, B. J.; Weaver, R.; Gaines, L. J.; Heindel, J. J. Toxicol. Appl. Pharmacol. 1994, 128, 224-8. (9) Albro, P. W.; Chapin, R. E.; Corbett, J. T.; Schroeder, J.; Phelps, J. L. Toxicol. Appl. Pharmacol. 1989, 100, 193-200. (10) Imajima, T.; Shono, T.; Zakaria, O.; Suita, S. J. Pediatr. Surg. 1997, 32, 18-21. (11) Ema, M.; Harazono, A.; Miyawaki, E.; Ogawa, Y. Reprod. Toxicol. 1996, 10, 365-72. (12) Harvan, D. J.; Hass, J. R.; Albro, P. W.; Friesen, M. D. Biomed. Mass Spectrom. 1980, 7, 242-6. (13) Luster, M. I.; Albro, P. W.; Chae, K.; Clark, G.; McKinney, J. D. Clin. Chem. 1978, 24, 429-32. (14) Albro, P. W.; Jordan, S.; Corbett, J. T.; Schroeder, J. L. Anal. Chem. 1984, 56, 247-50.
Analytical Chemistry, Vol. 72, No. 17, September 1, 2000 4127
Figure 1. Phthalate metabolism and compounds analyzed. Phthalates with the R and R′ groups shown are metabolized to form phthalate monoesters, which are found in urine as glucuronide conjugates or free acids. Phthalates can also be metabolized to further ω and ω - 1 oxidation products. The position of carbon-13 atoms in the stable isotope-labeled monoester internal standards is annotated with an asterisk (*).
is useful, but limited to assessing only DEHP exposure.15 We present a novel method for the accurate, precise, and sensitive measurement of eight phthalate monoesters in human urine. Selective use of β-glucuronidase enzyme to hydrolyze glucuronide metabolites allows for quantification of both free and glucuronidated forms of each phthalate metabolite. By analyzing the phthalate monoesters (the primary urinary metabolites of phthalate diesters), we avoid contamination from the ubiquitous parent compound and directly measure the active metabolites responsible for the reproductive and developmental toxicity of certain phthalates. Additionally, urine collection is less invasive than venous puncture for blood and therefore more suitable for exposure assessment of children. This method uses enzymatic deconjugation, solid-phase extraction, liquid chromatography, and tandem mass spectrometry for the sensitive, selective, and robust detection of these phthalate exposure biomarkers. EXPERIMENTAL SECTION Standard Preparation. Native standards (>98% purity, Figure 1) (monoethyl (MEP), monobutyl (MBP), monocyclohexyl (MCHP), monobenzyl (MBzP), mono-2-ethylhexyl (MEHP), and mono-n-octyl (MOP) phthalate were purchased from LE Scientific (St. Paul, MN). Mono-3-methyl-5-dimethylhexyl (isononyl, MNP) and mono-3-methyl-7-methyloctyl phthalate (isodecyl, MDP) were purchased from Cambridge Isotope Laboratories (Andover, MA). We prepared stock solutions of the eight native standards by accurately weighing ∼5 mg of material, dissolving it in acetonitrile (HPLC grade, Tedia, Fairfield, OH), and quantitatively transferring this solution to a 50-mL volumetric flask. This stock solution was stored at -20 °C in a Teflon-capped glass bottle until use. All standard solutions were prepared in methanol-rinsed and dried glassware. Ten unique standard solutions of phthalate monoesters (15) Dirven, H. A.; van den Broek, P. H.; Jongeneelen, F. J. Int. Arch. Occup. Environ. Health 1993, 64, 555-60.
4128
Analytical Chemistry, Vol. 72, No. 17, September 1, 2000
were prepared in water (HPLC grade, Tedia) from stock solutions. These solutions spanned the desired analytical range for each analyte (∼1-2500 ng/mL) without exceeding the linear dynamic range of the method. Standard solutions were also stored at -20 °C in Teflon-capped bottles except for the working solution (4 °C). Stock and working solutions of the 13C4-labeled standards (monoethyl-13C4, monobutyl-13C4, monocyclohexyl-13C4, monobenzyl-13C4, mono-2-ethylhexyl-13C4, mono-n-octyl-13C4, mono-3-methyl5-dimethylhexyl-13C4, and mono-3-methyl-7-methyloctyl-13C4 phthalate) were prepared similarly to the native standards and stored sealed at -20 °C until use (>98% chemical purity, Cambridge Isotope Laboratories). The working internal standard solution was stored at 4 °C. The isotopic purity of each internal standard was confirmed empirically by tandem mass spectral analysis. The amounts of native compounds were less than 0.2% for all internal standards except MEHP-13C4 (4.6%). The minimal contributions of the isotope to the native ion and native to the isotope ion were corrected for all reported data. Analyte and internal standard structures are shown in Figure 1. An additional internal standard solution of 4-methylumbelliferone glucuronide (>98%, Sigma Chemical, St. Louis, MO) was prepared in water as described for native standard solutions. This deconjugation internal standard was added to all samples to monitor β-glucuronidase (Escherichia coli K12, Roche Biomedical, Mannheim, Germany) enzyme activity16 as assessed by the levels of 4-methylumbelliferone released from the conjugated internal standard. Phthalate monoesters (pure standards) were stable for at least 6 months when stored in amber bottles at 4 °C. Sample Preparation. Human urine (1.00 mL) was thawed, vortex mixed, sonicated (5 min), and dispensed into a borosilicate glass test tube (16 × 125 mm, Corning, Corning, NY). Samples were subsequently buffered with ammonium acetate (250 µL, 1 M, pH 6.5, Sigma Chemical Co.) and spiked with isotopically labeled and deconjugated internal standards (50 µL, 0.24-1.00 ng/µL). After E. coli β-glucuronidase (5 µL, 200 units/mL) was added, the samples were sealed with Teflon-lined screw caps and gently mixed. Subsequent incubation at 37 °C for 90 min resulted in quantitative glucuronide hydrolysis to release the free phthalate monoesters. After enzymatic deconjugation, the samples were loaded onto solid-phase extraction cartridges (Oasis HLB, Waters, Milford, MA) and treated with various solvents and buffered aqueous solutions (see below and Figure 2). We prepared basic buffer by adding concentrated ammonium hydroxide (1 mL, 30% NH3 solution, J.T. Baker, Phillipsburg, NJ) to a 50:50 acetonitrile/water (200 mL) and stored it sealed at room temperature. Basic buffer was discarded after one week. We prepared acidic buffer (pH 2.0 ( 0.1) by making a solution of 0.14 M NaH2PO4 (ultrapure bioreagent, Fisher, Pittsburgh, PA) and 1.0% of concentrated H3PO4 (85%, J.T. Baker) and stored it sealed at room temperature. Acidic buffer was discarded after one month. Cartridges were processed on a vacuum manifold equipped with single-use Teflon flow lines (Supelco, Bellefonte, PA), minimizing the potential for carryover. (16) Valentı´n-Blasini, L.; Blount, B. C.; Rogers, H. S.; Needham, L. L. J. Exposure Anal. Environ. Epidemiol., in press.
Figure 2. Method for the analysis of eight phthalate monoesters in human urine.
First, an SPE cartridge was equilibrated with 1.0 mL of acetonitrile followed by 2.0 mL of basic buffer. This SPE cartridge was used to retain compounds that are hydrophobic at basic pH while allowing the more acidic analytes to elute. Next, the deconjugated urine samples were diluted with 1.0 mL of basic buffer solution and vortex mixed for 5 s. These samples were then added to an equilibrated SPE cartridge (3 mL/60 mg of Oasis HLB, Waters). The eluate was collected in borosilicate glass tubes. Residual analyte on the first cartridge was eluted by adding a second 1.0 mL of basic buffer to the cartridge. The combined basic buffer eluates were acidified by adding 3.0 mL of acidic buffer and mixed thoroughly. SPE solvent flows were monitored while samples were on the cartridges to maximize sorbent-analyte interaction (UCL ) 7.9 ng/mL, 0.99). The calibration curve typically spanned 3 orders of magnitude and was linear across this entire range, including the lowest levels as shown in the figure inset.
the entire process along with unknown urine samples to monitor for contamination. Before daily instrumental analyses, a known standard was injected to confirm acceptable chromatographic resolution and mass spectral sensitivity. After instrumental verification, a full set of 10 standards was analyzed, followed by the unknowns, QC samples and the blank. The analysis was completed by a repeated sequence of the 10 standards. All standards injected on the same day were then used to generate a daily calibration curve for each analyte with correlation coefficients typically greater than 0.99 (known concentration vs analyte/internal standard ratio, Figure 6). We required both the reagent blank and QC materials to meet clear specifications before approving a batch. If a reagent blank exceeded the limit of dection (LOD), we rejected the batch and reextracted the samples (with the exception of MBP). Consistent background MBP contamination (1.5-2 ng/mL in blanks) was subtracted from all data; contamination above this relatively low level resulted in batch reanalysis. Only 1 of 37 batches was reanalyzed for this reason. A batch for each individual analyte
RESULTS AND DISCUSSION Solid-phase extraction recoveries of the eight phthalate monoesters from human urine were excellent (>90%). The lowest recoveries were found for MEHP, MOP, MNP, and MDP, probably because of incomplete elution. Elution with acetonitrile followed by ethyl acetate improved the recoveries of longer side chain phthalate monoesters to acceptable levels. Extraction of water-based standards produced lower analyte recoveries than did urine-based standards (78-91%). The salt content of urine may have raised extraction recoveries because synthetic urine also produced improved recoveries. The variability in urinary sample composition will probably produce variations in recoveries and mass spectral responses. These variations are corrected by use of stable isotope-labeled internal standards for each of the analytes. To improve limits of detection, many instrumental parameters were optimized. The two most important instrumental parameters affecting APCI ion transmission in Finnigan TSQ mass spectrometers are the capillary voltage and the tube lens offset. Most analysts optimize these two parameters in a serial fashion; however, these two parameters are highly interdependent. Therefore, unless a more sophisticated approach to optimization is used, the instrument’s full analytical potential will not be realized. To gain maximum analytical sensitivity and instrument precision, we wrote an instrument control language (ICL) procedure to optimize the capillary and tube lens voltages in a factorial fashion. A solution of MDP (1 ng/µL) was infused (10 µL/min) with a constant flow of LC mobile phase (50% acetonitrile, 6 mM ammonium acetate, pH 6.5, 1.2 mL/min) into the APCI interface while the capillary voltage was held constant and the tube lens voltage was varied. This process was repeated for a series of capillary voltages (Figure 7) to identify the optimal settings of these two interdependent parameters. Similar analyses were performed on the other phthalate monoesters to ensure optimal ion transmission for all analytes. After optimization, atmospheric pressure chemical ionization of all phthalate monoesters produced negatively charged molecular ions ([M - H]-) in suitable abundance. Collision-induced dissociation produced informative daughter ion spectra. Optimal parent ion fragmentation was produced by altering the collision offset (Coff, Table 1). All phthalate monoesters produced phthalatespecific negative ions at m/z 77, 105, 121, and 147, corresponding to putative benzyl, benzaldehyde, benzoate, and phthalic anhydride fragment anions (Figure 4). Stable isotope-labeled internal standards fragmented similarly with ions at m/z 79, 108, 124, and 151, consistent with the presence of four carbon-13 atoms. Chromatographic separation was required to fully distinguish two structural isomers, MEHP and MOP, given similarities in daughter ion spectra (Figure 4). The daughter ions used for quantification were chosen as the most abundant ions that did not compromise the specificity of the measurement (Table 1). Analytical Chemistry, Vol. 72, No. 17, September 1, 2000
4131
Figure 7. Co-optimization of related APCI parameters for MDP. An instrument control language program was written to allow for cooptimization of tube lens offset and heated capillary voltage. Table 2. Quality Control, Precision, and Detection Statistics for Measuring Phthalate Monoesters in Human Urine QC low
QC high
detection (ng/mL)
analyte
ppb
% RSD
ppb
% RSD
LOD (3S0)
% detectn
MEP MBP MCHP MBzP MEHP MOP MNP MDP
8.5 7.1 7.1 7.9 10.9 16.2 29.8 20.9
12.5 17.2 11.9 14.1 13.9 13.2 14.1 14.7
489 101 105 150 116 151 145 157
4.9 5.7 5.8 10.6 9.6 8.6 9.3 10.7
1.0 0.6 0.7 0.8 1.2 0.9 0.8 1.5
98 100 22 100 81 19 18 51
This approach yielded excellent limits of detection (Table 2). The analytical LOD for each of the eight analytes was calculated as 3S0, where S0 is the value of the standard deviation as the concentration approaches zero.18 S0 was determined by analyzing quintuplicate sets of the lowest five standards and plotting the standard deviation versus the known standard concentration. The y-intercept of the best-fit line of this plot was used as S0. The LOD for analysis of phthalate monoesters in 1 mL of urine ranged from 0.5 to 2 ng/mL (Table 2). The SPE protocol used in this study resulted in improved sensitivity and reproducibility compared with nontreated urine; however, some matrix-associated components remained after cleanup. Most of these potential interferences eluted early in the chromatogram, with variable effects. Most urine samples showed a signal enhancement (2-4-fold) effect that was strongest for the analytes eluting earliest in the chromatogram (MEP, MBP). (17) Peck, C. C.; Albro, P. W. Environ. Health Perspect. 1982, 45, 11-7. (18) Taylor, J. K. Quality Assurance of Chemical Measurements; CRC Press: Boca Raton, FL, 1987.
4132 Analytical Chemistry, Vol. 72, No. 17, September 1, 2000
Absolute internal standard signal levels indicate that this phenomenon is not caused by SPE extraction or eluate concentration. Instead, the effect may involve improved ionization as described previously.19 Quantification was based on stable isotope dilution; therefore, variation in absolute signal did not significantly alter quantitative accuracy unless ionization was dramatically suppressed (phenomenon observed for MEP only in 2% of the samples analyzed). Full-scan mass spectra of urine samples with total suppression of the 13C4-MEP and MEP signals indicated a large interfering peak coeluting with MEP. This ionization suppression phenomenon was minimized by subsequent reanalysis of less sample volume (5 µL) and resulted in quantifiable 13C4-MEP signals. Variation in matrix effects between samples probably resulted from varying urine composition attributable to dietary, pharmaceutic, and genetic factors. Any urinary matrix effect on the calibration curve was evaluated by analyzing the standards prepared in water versus those spiked into human urine. Standards spiked with urine produced calibration curves with slopes not significantly different from the slopes produced by standards prepared in water. Therefore, no interfering matrix effect was observed for the range of analyte concentrations measured, and calibration curves were produced using data collected by analyzing standards and internal standard prepared in water. Interday variability of daily calibration curve slopes was minimal; the relative standard deviation of 59 calibration curves over 3 months ranged from 5 to 9% depending on the analyte. A typical daily calibration curve based on triplicate analysis of 10 standard solutions across the linear range produced excellent linear correlation coefficients (typically >0.99, Figure 6). The calibration curve typically spanned 3 orders of magnitude with a working range from the LOD (0.5-2 ng/mL) to the highest standard (400-2500 ng/mL). A broad linear range was required because of highly variable levels in human urine specimen; some analyte levels varied over 4 orders of magnitude.20 QC material (spiked urine) was analyzed with each batch of samples, and the resulting data were evaluated using a Shewhart QC plot (Figure 5, MOP QC low pool). Each point in this means plot represents the average analyte value found daily upon analyzing multiple aliquots of the same QC material. The means plot assesses the accuracy of the method relative to the originally characterized levels in the QC pool. A variance plot for each analyte and level was used to assess precision. The coefficients of variation (37 sample batches analyzed over 11 weeks) for each analyte in QC-high and QC-low pools are shown in Table 2. The QC-high pool contained ∼100 times the LOD for each analyte; all CVs were 5-11%. The low-QC pool contained lower levels of analyte (∼10 times LOD) and therefore had larger CVs. Nonetheless, all but one of these analytes had CVs less than 15%, indicating that the acceptable reproducibility of this trace analysis method. To verify the absence of significant phthalate monoester contamination, representative specimen cups, tubes, pipet tips, and autosampler vials were prescreened and found to be phthalate monoester free (< LOD). In addition, synthetic urine21 was stored (19) Barr, D. B.; Ashley, D. L. J. Anal. Toxicol. 1998, 22, 96-104. (20) Blount, B. C.; Silva, M. J.; Caudill, S. P.; Needham, L. L.; Pirkle, J. L.; Sampson, E. J.; Lucier, G. W.; Jackson, R. J.; Brock, J. W. Environ. Health Perspect., in press. (21) Gustafsson, J. E.; Uzqueda, H. R. Clin. Chim. Acta 1978, 90, 249-57.
Figure 8. Enzyme deconjugation time optimization for MBP (open diamonds) and 4-MU (solid circles) glucuronides in urine. Time zero represents urine with no enzyme added; note the presence of a low level (3 ng/mL) of unconjugated MBP.
at 4 and -20 °C for three months in polypropylene specimen cups with no sign of phthalate monoester contamination leaching from the cup. Additionally, SPE cartridges screened for phthalate diesters contained insignificant levels of parent compound (
