Measurement of bisphenol A levels in human urine

Abstract

We report a new approach for assessing human exposure to bisphenol A (BPA) by measuring BPA in urine after enzymatic deglucuronidation. This method involves addition of 13C 12-labeled BPA, enzymatic deconjugation, solid-phase extraction, and derivatization with pentafluorobenzyl bromide. The product of the derivatization is separated by gas chromatography followed by mass spectrometric detection using negative chemical ionization and selected ion monitoring. Using this analysis method, urine samples fortified with both a constant level of labeled BPA and a range of unlabeled BPA levels (0.27–10.6 ng/ml) demonstrated constant percentage recovery. In addition, a range of urine sample volumes (0.25–10.0 ml) with constant amounts of added internal standard produced a linear response (r2=0.99). The method limit of detection was 0.12 ng/ml. This method was validated by duplicate analyses using gas chromatography coupled to a high-resolution mass spectrometer.

Introduction

Bisphenol A (BPA), 2,2-bis(4-hydroxyphenyl)propane, is an industrial chemical used in the production of polycarbonate plastics, epoxy resins, and other products. Products potentially containing BPA include dental sealants and fillings, polycarbonate bottles, powder paints, food and drink containers, as well as beverage and food can linings. Prior to 1994, BPA could also be used as an “inert” ingredient in pesticide formulations. BPA has also been shown to be present in the natural environment. In natural water systems, BPA levels vary from <0.01 to 1.9 μg/l and have a half-life of 3–5 days (Matsumoto et al., 1977; Staples et al., 1998).

The relative importance of various sources of human exposure is poorly understood and vigorously debated. BPA appears to leach from lined cans and polycarbonate bottles. The rate of this leaching is under debate (Brotons et al., 1995; Hoyle and Budway, 1997). In addition, the leaching of BPA from dental sealants and fillings has been reported, but these results are also the subject of debate (Olea et al., 1996; Ashby, 1997; Imai, 1999).

When information is lacking about sources of exposure, internal dose measurements (such as monitoring markers in a biological tissue such as serum, urine or feces) must form the basis of human exposure assessment (Pirkle et al., 1995). BPA exposure in humans has not been studied except for one report of serum BPA levels in people. In this study, measurable levels of BPA ranging from <0.1 to 1.6 ng/ml were found in human serum samples (Sajika et al., 1999).

The usefulness of internal dose data can be limited by a lack of information on compartmentation, metabolism, and excretion. The compartmentation, metabolism, and excretion of BPA in humans have not been studied. In rodents, BPA is predominately excreted unmetabolized in feces with a serum half-life of less than 1 h (Pottenger et al., 2000). The other major metabolite excreted, glucuronidated BPA in urine, represents 15–34% of the original dose. BPA is also excreted as the urinary sulfate but at very low levels (<3%). Total urinary excretion may require as much as 2 days (Knaak and Sullivan, 1966; Pottenger et al., 2000). Differing excretion rates by gender and by route of administration were also found in these studies.

In preparation for a population-based human exposure study, we reviewed the literature for current methods for analyzing BPA. We found the methods possessed limitations or were not applicable to human biological samples (Gandara et al., 1993; Del Olmo et al., 1999) except for two techniques described in the previously mentioned study (Sajika et al., 1999). These techniques were analyses of BPA in serum. Assuming the rodent findings apply to humans, measurement of BPA in serum provides information only about very recent exposure (1–3 h), whereas urinary BPA glucuronide levels are a longer lived biomarker (12–48 h).

Due to the difficulty of obtaining and extracting human fecal samples, the ease of urine collection, and the short half-life of BPA in serum, we developed a method to quantify BPA metabolites in human urine. This method extends a similar method to analyze BPA in serum(Yoshimura et al., 2000, unpublished). This new method involves enzymatic deconjugation, solid-phase extraction (SPE), derivatization with pentafluorobenzyl bromide and gas chromatography with mass spectrometric detection (GC/MS). This method is based on previous methods involving pentafluorobenzyl bromide derivatization (Daneshvar and Brooks, 1988; Jacobsson et al., 1988; Hofmann et al., 1990). Confirmation of the GC/MS method and identification of the final derivatized product were performed using gas chromatography coupled to a high-resolution mass spectrometer (GC/HRMS). Preliminary data suggest that the sensitivity of this method will allow measurement of background levels in urine from nonoccupationally exposed humans.

Methods

All reagents and solvents were analytical grade unless specified otherwise (Burdick and Jackson, Muskegon, MI). All glassware was rinsed with hexane followed by acetone.

An aliquot (usually 1 ml) of urine was spiked with 20 μl of 13C 12-ring-labeled BPA (75.7 ng/ml, Cambridge Isotope Laboratories, Andover, MA). To samples requiring deglucuronidation, β-glucuronidase (E. coli, 5 μl, 200 U/ml, Roche Biochemical, Mannheim, Germany) was added followed by incubation at 37°C for 90 min. This treatment is sufficient to deconjugate the glucuronides of both 5-methylumbelliferone and phthalate monoesters (Blount et al., 2000). After enzymatic treatment, 1 ml of formic acid (32%; EM Industries, Gibbsville, NJ) was added to the sample and then buffered with ammonium acetate (250 μl, 1 M, pH 6.5, Sigma, Milwaukee, WI).

An octyldecyl SPE column (500 mg, J.T. Baker, Phillipsburg, NJ) was prepared by washing under partial vacuum (5 mm Hg) with 15 ml of methanol followed by 5 ml of water. Water was left just above the head of the column. The spiked sample was added and pulled with partial vacuum (5 mm Hg) onto the column until it was even with the top of column. The column and sample stood for 10 min without vacuum. Then 5 ml of water/methanol (90:10 v/v) was added to the column and pulled through with partial vacuum (5 mm Hg). The column was allowed to go dry with partial vacuum (15 mm Hg). Clean tubes were placed under the manifold and 8 ml of methanol was added to the column and pulled through with partial vacuum (5 mm Hg). Samples were evaporated just to dryness under a stream of nitrogen using a Turbovap (55°C, 5 psi, 20 min; Zymark, Framingham, MA).

Methylene chloride (1 ml; Calendon Laboratories, Georgetown, Ontario, Canada) was added to the sample tubes and mixed. To the sample tubes, we added 0.5 ml of 0.1 M tetrabutylammonium hydrogen sulfate (Eastman Kodak, Rochester, NY), 0.5 ml of 0.2 M sodium hydroxide (Mallinkrodt Baker, Paris, KY), and 20 μl of α-bromo-2,3,4,5,6-pentafluorotoluene commonly called pentafluorobenzyl bromide (Supelco, Bellefonte, PA). Tubes were mixed for 25 min at room temperature using a mechanical mixer inverting the tubes about twice per minute.

After derivatization, the tubes were centrifuged for 1 min at 1600 rpm. The methylene chloride layer was removed by pipet and placed in a clean tube and concentrated just to dryness under a steam of nitrogen (20 min, 55°C; Turbovap; Zymark). To the tube, 2,2,4-trimethylpentane (0.5 ml, Mallinkrodt Baker) was added and mixed. The 2,2,4-trimethylpentane solution was transferred into clean GC vials.

The derivatized samples were analyzed using a single quadrupole GC/MS (HP 6890 and 5973, Agilent, Wilmington, DE). The injection volume was 2 μl in the splitless mode with the injector temperature of 245°C. The injector purge time was 1 min. The column was a DB-5, 30 m in length with a film thickness of 0.25 μm and internal diameter of 0.25 mm (J&W Scientific, Folsom, CA). The carrier gas was helium with a constant flow of 1 ml/min. The oven temperature program was 60°C initially and a single step to 270°C at 15°C/min. The total run time was 31 min.

The MS used negative chemical ionization (NCI) with methane as the reagent gas with a pressure of 1.8×10 −4 Torr. The source temperature was 260°C. Later experiments suggest that 270°C improves the ion abundance. The transfer line was heated to 280°C. The solvent delay was 5.6 min. The native BPA signal was monitored in selected ion mode at mass to charge ratio (m/z) 407 and the 13C12-labeled BPA signal was monitored at m/z 419. These masses represent the fragment ions after the loss of C 7F5H2 from the disubstituted ether derivatives (Figure 1). The retention time of the derivatized BPA peak was 26.4 min.

Figure 1
figure1

Positive electron ionization (EI+) and negative chemical ionization (NCI) mass spectra of GC peak eluting at 26.4 min.

Validation of this method involved analyzing duplicate samples on both the GC/MS and on the more selective GC/HRMS (Finnigan Mat 900, San Jose, CA). The conditions for the GC/HRMS were as follows: source temperature was 230°C, the reagent gas was isobutane, the transfer line temperature was 270°C, the electron energy was 130 eV, the emission current was 0.18 mA. The GC column was a DB-5, injector temperature was 250°C, the injection volume was 2 μl and the purge time was 1 min. The initial GC oven temperature was 75°C for 1 min then ramped to 200°C at 15°C/min then to 220°C at 10°C/min and then to 270°C at 15°C/min and held for 12 min. The total run time was 27 min.

Results

The product of the derivatization and the compound separated on the GC was the pentafluorobenzyl diether of BPA (Figures 1 and 2). The presence of a peak at 588 amu in the positive electron ionization (EI+) mass spectrum confirms the identity of the separated compound (Figure 1). In negative ionization mode, the predominant ion is formed by the loss of one of the pentafluorobenzyl moieties (Figure 1). The native and labeled BPA peaks elute at 26.4 min and are well resolved from surrounding peaks (Figure 2).

Figure 2
figure2

GC chromatograms. The human urine extracts are identical except for undergoing or omitting enzyme treatment. (a) BPA standard in water (0.53 ng/ml); (b) unspiked human urine extract; (c) unspiked human urine extract — deconjugated.

Calibration curves for BPA were linear from 0.26 to 10.6 ng/ml for the GC/MS and the GC/HRMS based on subtracting out a low background level (0.06 ng/ml urine; Figure 3). The detection limit for the GC/MS was 0.12 ng/ml (N=11) and for the GC/HRMS was 0.11 ng/ml (N=7). These detection limits were calculated by three standard deviations of the mean blank level of: 0.110±0.3 ng/ml (GC/MS) and 0.077±0.04 ng/ml (GC/HRMS). The method reproducibility at a sample concentration of 1 ng/ml was ±6% at 95% confidence(2 σ). Constant percentage recovery (101.6±8.1%) and linear correlation were found upon analysis of samples spiked with 0.25 to 10 ng BPA/ml (Figure 4). This relation shows that percentage recovery does not vary with concentration. As desired, the amount of BPA recovered varies linearly with the amount added.

Figure 3
figure3

Calibration curves for the GC/MS. The amount injected onto the column can be calculated by multiplying the amount on this graph by the injection volume of 0.02 ml.

Figure 4
figure4

Standard addition and recovery in human urine.

This method was validated by analysis of duplicate samples by both GC/MS and GC/HRMS. The results obtained on duplicate samples analyzed by both instruments were not significantly different (Table 1). This agreement suggests that the GC/MS does not suffer from any major interferences that could inflate the values.

Table 1 Comparison of BPA levels in spiked human urine (ng/ml) .

To improve the sensitivity of this method, different volumes of spiked urine samples were analyzed (Figure 5A and B). Sample volumes between 0.25 and 10 ml can be used with comparable results but larger sample volume may allow lower detection limits.

Figure 5
figure5

Effect of changing urine volume. (a) High volume and (b) low volume.

Pooled human urine samples analyzed by this method demonstrated no detectable BPA before deglucuronidation. After deglucuronidation, pooled urinary BPA levels varied from 0.11 to 0.51 ng/ml urine (Table 2).

Table 2 Free and total BPA levels in unspiked human urine (ng/ml)

Discussion

We have developed a highly sensitive and highly selective technique that should allow determination of background BPA levels in nonoccupationally exposed humans. The method involved SPE, derivatization with pentafluorobenzyl bromide and GC/MS. An alternative approach, the derivatization of BPA to trimethylsilyl (TMS) derivatives followed by GC/MS with EI+ was also attempted (data not shown). Although the TMS technique produced about 10-fold greater sensitivity when analyzing standards in water, the PFB technique described in this paper produced fewer interfering peaks when analyzing human urine or serum (data not shown). Other studies involving less complex matrices may be well served by using the TMS approach.

This method was developed for use in studies of human exposure to BPA and correlation with possible health effects. In rats, BPA produces a variety of hormone-like effects (Laws et al., 2000; Long et al., 2000). The health effects of BPA exposure on humans have not been studied.

This method's use of urine permits the inclusion of children in studies without requiring venous puncture. The method also offers flexibility by allowing use of different urine volumes. Studies with stored urine specimens often have different volumes available for analysis.

By omitting the enzymatic deglucuronidation step, we were able to measure free BPA in urine and serum. Preliminary data suggests that BPA in human serum is not present as the glucuronide but is present as the free phenol (Yoshimura et al., 2000). However, in our pilot work developing this method, BPA glucuronide was found in urine with very little free phenol (Table 2).

We will use this method to assess the BPA levels in human urine samples to provide a reference range. In addition, variations in human metabolism will also be examined by monitoring both free and glucuronidated BPA in urine and serum. Further, this method should allow easy expansion to related compounds like octyl and nonyl phenol as well as tetrabromobisphenol A and bisphenol F.

Abbreviations

BPA:

bisphenol A

EI+:

positive electron ionization

GC/HRMS:

gas chromatography coupled to high-resolution mass spectrometric detection

GC/MS:

gas chromatography coupled to mass spectrometric detection

Hg:

mercury

m/z:

mass to charge ratio

NCI:

negative chemical ionization

SPE:

solid-phase extraction

TMS:

trimethylsilyl

References

  1. 1

    Ashby J, Bisphenol A and dental sealants: the inappropriateness of continued reference to a single female patient, Environ Health Perspect (1997) 105: 362

  2. 2

    Blount BC Milgram KE Silva M Malek N Reidy JA Needham LL and Brock JW, Quantitative detection of eight phthalate metabolites in human urine using HPLC-APCIMS/MS, Anal Chem (2000) 72: 4127–4134

  3. 3

    Brotons JA Olea-Serrano MF Villalobos M Pedraza V and Olea N, Xenoestrogens released from laquer coating in food cans, Environ Health Perspect (1995) 103: 6008–6012

  4. 4

    Daneshvar M, and Brooks JB, Improved procedure for preparation of pentafluorobenzyl derivatives of carboxylic acids for analysis by gas chromatography with electron capture detection, J Chromatgr (1988) 433: 248–256

  5. 5

    Del Olmo M Zafra A Navas NA and Vilchez JL, Trace determination of phenol, bisphenol A, and bisphenol A digycidyl ether in mixtures by excitation fluorescence following micro liquid–liquid extraction using partial least squares regression, Analyst (1999) 124: 385–390

  6. 6

    Gandara JS Abuin SP Losado PP and Lozano JS, Determination of bisphenol A and F in noncured epoxy resins by RP–HPLC–fluorescence techniques, J Chromatogr Sci (1993) 31: 450–454

  7. 7

    Hofmann U, Holzer S, and Meese CO, Pentafluorophenyldiazoalkanes as novel derivatization reagents for the determination of sensitive carboxylic acids by gas chromatography–negative ion mass spectrometry, J Chromatogr (1990) 508: 349–356

  8. 8

    Hoyle WC and Budway R, Bisphenol A in food cans, Environ Health Perspect (1997) 105: 570–571

  9. 9

    Imai Y, Comments on estrogenicity of resin-based composites and sealants used in dentistry, Environ Health Perspect (1999) 107: A290

  10. 10

    Jacobsson S, Larsson A, and Arbin A, Extractive pentafluorobenzylation of formic, acetic and levulinic, benzoic and phthalic acids studied by liquid chromatography and dual-oven capillary gas chromatography, J Chromatogr (1988) 447: 329–340

  11. 11

    Knaak J, and Sullivan L, Metabolism of bisphenol A in the rat, Toxicol Appl Pharmacol (1966) 8: 175–184

  12. 12

    Laws SC Carey SA Ferrell JM Bodman GJ and Cooper RL, Estrogenic activity of octylphenol, nonylphenol, bisphenol A, and methoxychlor in rats, Toxicol Sci (2000) 254: 154–167

  13. 13

    Long X Steinmetz R Nira B-J Caperall-Grant A Young PCM Nephew KP and Bigsby RM, Strain differences in vaginal responses to the xenoestrogen bisphenol A, Environ Health Perspect (2000) 108: 243–247

  14. 14

    Matsumoto G, Ishwatari R, and Hanya T, Gas chromatographic–mass spectrometric identification of phenols and aromatic acids in river waters, Water Res (1977) 11: 693–698

  15. 15

    Olea N Pulgar R Perez P Olea-Serrano F Rivas A Novillo-Fetrell A Pedraza V Soto AM and Shonnenshein C, Estrogenicity of resin-based composites and sealants used in dentistry, Environ Health Perspect (1996) 104: 298–305

  16. 16

    Pirkle J Sampson EJ Needham LL Patterson DG and Ashley DL, Using biological monitoring to assess human exposure to toxicants, Environ Health Perspect (1995) 103: 45–48

  17. 17

    Pottenger LH Domoradzki JY Markham DA Hansen SC Cagen SZ and Wacchter JM Jr., The relative bioavailability and metabolism of bisphenol A in rats is dependent upon the route of administration, Toxicol Sci (2000) 54: 3–18

  18. 18

    Sajika J, Takahashi K, and Yonekubo J, Sensitive method for the determination of bisphenol-A in serum using two systems of high performance chromatography, J Chromatogr B Biomed Sci Appl (1999) 736: 255–261

  19. 19

    Staples CA Dorn PB Klecka GM O'block ST and Harris LR, A review of the environmental fate, effects, and exposures of bisphenol A, Chemosphere (1998) 36: 2149–2173

  20. 20

    Yoshimura Y Brock JW Makino, T and Nakazawa H 2000 Determination of bisphenol A in human serum by GC/MS, Anal Chem Acta In press

Download references

Acknowledgements

This work was supported in part by a grant-in-aid from the Ministry of Health and Welfare of Japan. Brand names are given for reference and do not constitute endorsement.

Author information

Correspondence to JOHN W BROCK.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

BROCK, J., YOSHIMURA, Y., BARR, J. et al. Measurement of bisphenol A levels in human urine. J Expo Sci Environ Epidemiol 11, 323–328 (2001). https://doi.org/10.1038/sj.jea.7500174

Download citation

Keywords

  • biomonitoring
  • bisphenol A
  • GC/MS
  • human
  • pentafluorobenzyl bromide
  • solid-phase extraction
  • urine

Further reading