Quantitative Measurement of Endogenous Estrogens and Estrogen

Sep 12, 2007 - Epidemiology and Biostatistics Program, Division of Cancer Epidemiology and Genetics, National Cancer Institute,. Bethesda, Maryland 20...
0 downloads 0 Views 205KB Size
Anal. Chem. 2007, 79, 7813-7821

Quantitative Measurement of Endogenous Estrogens and Estrogen Metabolites in Human Serum by Liquid Chromatography-Tandem Mass Spectrometry Xia Xu,† John M. Roman,† Haleem J. Issaq,† Larry K. Keefer,‡ Timothy D. Veenstra,*,† and Regina G. Ziegler§

Laboratory of Proteomics and Analytical Technologies, SAIC-Frederick, Inc., NCI-Frederick, Frederick, Maryland 21702, Laboratory of Comparative Carcinogenesis, Center for Cancer Research, NCI-Frederick, Frederick, Maryland 21702, and Epidemiology and Biostatistics Program, Division of Cancer Epidemiology and Genetics, National Cancer Institute, Bethesda, Maryland 20892

Endogenous estrogen plays a key role in the development of human breast cancer, yet the contribution of specific estrogen metabolites and patterns of estrogen metabolism remains unclear. To determine their individual and joint roles in breast carcinogenesis, it is necessary to be able to measure quantitatively each estrogen metabolite in epidemiologic and clinical biospecimens. In this report, we detail a sensitive, specific, accurate, and precise highperformance liquid chromatography-tandem mass spectrometry method utilizing selected reaction monitoring for measuring the absolute quantities of free (unconjugated) and total (conjugated + unconjugated) endogenous estrogens and estrogen metabolites in human serum from premenopausal and postmenopausal women. The method requires a simple sample preparation and only 0.5 mL of serum, yet is capable of quantifying simultaneously 15 estrogens and estrogen metabolites (EM): estrone and its 2-, 4-, and 16r-hydroxy and 2- and 4-methoxy derivatives; 2-hydroxyestrone-3-methyl ether; 17β-estradiol and its 2-hydroxy and 2- and 4-methoxy derivatives; and estriol, 17-epiestriol, 16-ketoestradiol, and 16-epiestriol. The lower limit of quantitation for each EM was 0.4 pg oncolumn, equivalent to 8 pg/mL (26.5-29.6 fmol/mL) in the original serum sample. Calibration curves were linear over a 103-fold concentration range. For a stripped serum sample containing 8 pg/mL of each EM, accuracy (percent recovery of a known added amount) ranged from 91 to 113%. Intrabatch precision (including hydrolysis, extraction, and derivatization steps) ranged from 7 to 30% relative standard deviation (RSD), and interbatch precision ranged from 8 to 29% RSD. Since distinct roles have been proposed for many of these estrogen metabolites, an accurate, precise, sensitive, and specific method for * Corresponding author. Dr. Timothy D. Veenstra, Laboratory of Proteomics and Analytical Technologies, SAIC-Frederick, Inc., NCI-Frederick, P.O. Box B, Frederick, MD 21702. Phone: 301-846-7286. Fax: 301-846-6037. E-mail: [email protected]. † Laboratory of Proteomics and Analytical Technologies. ‡ Laboratory of Comparative Carcinogenesis. § Epidemiology and Biostatistics Program. 10.1021/ac070494j CCC: $37.00 Published on Web 09/12/2007

© 2007 American Chemical Society

measuring their levels in circulation should suggest new approaches to breast cancer prevention, screening, and treatment. In recent years, the evidence that endogenous estrogen plays a key role in the development of human breast cancer has become overwhelming,1-4 yet the contribution of specific estrogen metabolites and individual patterns of estrogen metabolism remains unclear. A carcinogenic effect can occur through at least two mechanisms. Catechol estrogens, derivatives of estrone and estradiol hydroxylated at position C-2 or C-4 adjacent to the C-3 phenolic hydroxyl, are known to form quinones that react with DNA and form both stable and depurinating DNA adducts. These genotoxic modifications, along with other estrogen-induced DNA damage, can lead to mutation, cell transformation, and tumor initiation in model systems.5-7 Another mechanism, related to tumor promotion rather than initiation and mediated by the estrogen receptor, involves the mitogenic and antiaptotic effects of estrone, estradiol, and some of their hydroxylated metabolites.1,8 Well-designed epidemiologic studies are necessary to further evaluate these hypotheses and provide insight into the estrogenrelated mechanisms of human breast carcinogenesis. Human serum contains both biologically active estrogens, which includes unconjugated parent estrogens, their phase I metabolites, and O-methylated catechol estrogens, as well as biologically inactive estrogens, which includes sulfate or/and glucuronide conjugates of biologically active estrogens. Therefore, quantitatively measuring circulating biologically active endogenous estrogens and estrogen metabolites as well as their inactive sulfate and glucuronide conjugates and comparing them among humans (1) Yager, J. D.; Davidson, N. E. N. Engl. J. Med. 2006, 354, 270-282. (2) Travis, R. C.; Key, T. J. Breast Cancer Res. 2003, 5, 239-247. (3) Clemons, M.; Goss, P. N. Engl. J. Med. 2001, 344, 276-285. (4) Kaaks, R. IARC Sci. Publ. 2001, 154, 149-162. (5) Cavalieri, E. L.; Rogan, E. G. Ann. N. Y. Acad. Sci. 2004, 1028, 247-257. (6) Liehr, J. G. Hum. Reprod. Update. 2001, 7, 273-281. (7) Dowers, T. S.; Qin, Z. H.; Thatcher, G. R.; Bolton, J. Chem. Res. Toxicol. 2006, 19, 1125-1137. (8) Bradlow, H. L.; Sepkovic, D. W. Ann. N. Y. Acad. Sci. 2004, 1028, 216232.

Analytical Chemistry, Vol. 79, No. 20, October 15, 2007 7813

Figure 1. Method for measuring free (unconjugated) and total (conjugated + unconjugated) endogenous estrogens and estrogen metabolites (EM).

who ultimately develop breast cancer to matched, healthy controls could help gain insight into possible mechanisms of estrogen related breast carcinogenesis. Given the potential importance of individual profiles of estrogen metabolism, we need a direct, quantitative analytical method to measure accurately, reliably, and simultaneously as many estrogen metabolites as possible. Although mass spectrometry (MS)-based methods for quantitatively measuring circulating levels of estrone and 17β-estradiol have been reported,9-13 to the best of our knowledge, only a few published reports have used MS-based methods for the analysis of endogenous estrogen metabolites. Adlercreutz et al. reported a gas chromatography (GC)/MS method for quantitatively measuring 11 endogenous estrogens and estrogen metabolites in late pregnancy human plasma.14 A negative chemical ionization GC/MS method after pentafluoropropionic anhydride derivatization was described for measuring 17β-estradiol, 2-hydroxyestradiol, 4-hydroxyestradiol, 2-methoxyestradiol, and 4-methoxyestradiol in rat plasma spiked with these compounds.15 Lakhani et al. published a liquid chromatography (LC)atmospheric pressure chemical ionization-tandem MS method for measuring 2-methoxyestradiol in human plasma samples from a cancer patient who had received a single oral dose of 2200 mg of 2-methoxyestradiol.16 However, in most cases radioimmunoassay or enzyme immunoassay was the technique used for measuring circulating levels of endogenous estrogen metabolites.17-19 Both these immunoassay procedures can suffer from poor specificity, (9) Siekmann, L. J. Clin. Chem. Clin. Biochem. 1984, 22, 551-557. (10) Thienpont, L. M.; Verhaeghe, P. G.; Van, Brussel, K. A.; De Leenheer, A. P. Clin. Chem. 1988, 34, 2066-2069. (11) Wu, H.; Ramsay, C.; Ozaeta, P.; Liu, L.; Aboleneen, H. Clin. Chem. 2002, 48, 364-366. (12) Nelson, R. E.; Grebe, S. K.; O’Kane, D. J.; Singh, R. J. Clin. Chem. 2004, 50, 373-384. (13) Tai, S. S.; Welch, M. J. Anal. Chem. 2005, 77, 6359-6363. (14) Adlercreutz, H.; Tikkanen, M. J.; Hunneman, D. H. J. Steroid Biochem. 1974, 5, 211-217. (15) Zacharia, L. C.; Dubey, R. K.; Jackson, E. K. Steroids 2004, 69, 255-261. (16) Lakhani, N. J.; Lepper, E. R.; Sparreboom, A.; Dahut, W. L.; Venitz, J.; Figg, W. D. Rapid Commun. Mass Spectrom. 2005, 19, 1176-1182. (17) Emons, G.; Ball, P.; von Postel, G.; Knuppen, R. Acta Endocrinol. 1979, 91, 158-166. (18) Berg, D.; Sonsalla, R.; Kuss, E. Acta Endocrinol. 1983, 103, 282-288. (19) Bradlow, H. L.; Sepkovic, D. W.; Klug, T.; Osborne, M. P. Steroids 1998, 63, 406-413.

7814

Analytical Chemistry, Vol. 79, No. 20, October 15, 2007

accuracy, and reproducibility due to the cross-reactivity and lot-to-lot variation of antibodies.20,21 The need for valid methods to measure endogenous estrogen metabolites in circulation has never been greater. In this report, we present a high performance liquid chromatography-electrospray ionization-tandem mass spectrometry (HPLC-ESI-MS2) method that is capable of measuring quantitatively both free (unconjugated) and total (conjugated + unconjugated) endogenous estrogens and estrogen metabolites in serum from premenopausal and postmenopausal women. EXPERIMENTAL SECTION Reagents and Materials. The 15 estrogens and estrogen metabolites (EM) analyzed in this study, estrone (E1), estradiol (E2), estriol (E3), 2-hydroxyestrone (2-OHE1), 2-methoxyestrone (2-MeOE1), 2-hydroxyestradiol (2-OHE2), 2-methoxyestradiol (2MeOE2), 2-hydroxyestrone-3-methyl ether (3-MeOE1), 4-hydroxyestrone (4-OHE1), 4-methoxyestrone (4-MeOE1), 4-methoxyestradiol (4-MeOE2), 16R-hydroxyestrone (16R-OHE1), 17-epiestriol (17epiE3), 16-ketoestradiol (16-ketoE2), and 16-epiestriol (16-epiE3), were obtained from Steraloids, Inc. (Newport, RI). Four deuteriumlabeled estrogens and estrogen metabolites (d-EM), estradiol-2,4,16,16-d4 (d4-E2), estriol-2,4,17-d3 (d3-E3), 2-hydroxyestradiol-1,4,16,16,17-d5 (d5-2-OHE2), and 2-methoxyestradiol-1,4,16,16,17-d5 (d52-MeOE2), were purchased from C/D/N Isotopes, Inc. (PointeClaire, Quebec, Canada). A fifth d-EM, 16-epiestriol-2,4,16-d3 (d316-epiE3), was obtained from Medical Isotopes, Inc. (Pelham, NH). All EM and d-EM have reported chemical and isotopic purity g98% and were used without further purification. Dichloromethane, methanol, and formic acid were obtained from EM Science (Gibbstown, NJ). Glacial acetic acid, sodium bicarbonate, and L-ascorbic acid were purchased from J. T. Baker (Phillipsburg, NJ), and sodium hydroxide and sodium acetate were purchased from Fisher Scientific (Fair Lawn, NJ). β-Glucuronidase/sulfatase (Helix pomatia, Type HP-2, g500 Sigma units β-glucuronidase and e37.5 units sulfatase activity) was obtained from Sigma Chemical Co. (St. Louis, MO). Dansyl chloride and acetone were purchased (20) Ziegler, R. G.; Rossi, S. C.; Fears, T. R; Bradlow, H. L.; Adlercreutz, H.; Sepkovic, D.; Kiuru, P.; Wahala, K.; Vaught, J. B.; Donaldson, J. L.; Falk, R. T.; Fillmore, C. M.; Siiteri, P. K.; Hoover, R. N.; Gail, M. H. Environ. Health Perspect. 1997, 105 (Suppl. 3), 607-614. (21) Spierto, F. W.; Gardner, F.; Smith, S. J. Steroids 2001, 66, 59-62.

Table 1. Tandem Mass Spectrometry Selected Reaction Monitoring Conditions EM

transition monitored (m/z)

collision energy (V)

tube lens voltage (V)

E1 E2 E3 2-OHE1 2-MeOE1 2-OHE2 2-MeOE2 3-MeOE1 4-OHE1 4-MeOE1 4-MeOE2 16R-OHE1 17-epiE3 16-ketoE2 16-epiE3 d4-E2 d3-E3 d3-16-epiE3 d5-2-MeOE2 d5-2-OHE2

504 f 171 506 f 171 522 f 171 753 f 170 534 f 171 755 f 170 536 f 171 534 f 171 753 f 170 534 f 171 536 f 171 520 f 171 522 f 171 520 f 171 522 f 171 510 f 171 525 f 171 525 f 171 541 f 171 760 f 170

27 28 28 42 26 43 31 26 42 26 31 27 28 27 28 28 28 28 31 43

180 175 192 224 155 224 173 155 224 155 173 163 192 163 192 175 192 192 173 224

from Aldrich Chemical Co. (Milwaukee, WI). All chemicals and solvents used in this study were HPLC or reagent grade unless otherwise noted. Serum Sample Collection. Six serum samples were collected for this study: two from premenopausal women during the follicular phase of the menstrual cycle, two from premenopausal women during the luteal phase of the menstrual cycle, and two from postmenopausal women. Menstrual and menopausal status was based on self-report. All subjects were healthy, not pregnant, and not taking exogenous hormones. After the blood was allowed to clot for 1 h at room temperature, serum was prepared and aliquots were stored at -80 °C. The blood collection was approved by the NCI/NIH Institutional Review Board. Preparation of Stock and Working Standard Solutions. Stock solutions of EM and d-EM were each prepared at 80 µg/ mL by dissolving 2 mg of each estrogen powder in methanol containing 0.1% (w/v) L-ascorbic acid to a final volume of 25 mL in a volumetric flask. Stock solutions were monitored by measuring the absolute peak height of each EM using liquid chromatography-tandem mass spectrometry (LC-MS2) to verify that no time-dependent degradation of the EM and d-EM standards had occurred. The stock solutions were stable for at least 2 months while stored at -20 °C. Working standard solutions of each EM or each d-EM at 8 ng/mL were prepared by dilutions of the stock solutions with methanol containing 0.1% (w/v) L-ascorbic acid. Calibration Standards and Quality Control Samples. Charcoal-stripped human serum (Golden West Biologicals, Temecula, CA) with no detectable levels of any EM, and containing 0.1% (w/v) L-ascorbic acid, was employed for preparation of calibration standards and quality control (QC) samples. Calibration standards were prepared in charcoal-stripped human serum by adding 20 µL of the d-EM working internal standard solution (0.16 ng of each d-EM) to various volumes of the EM working standard solution, which contained 0.002-2 ng of each EM and were assayed in duplicate. The QC samples were prepared at three levels: 8, 40, and 160 pg/mL (26.5-29.6, 132.4-148.0, 529.5592.2 fmol/mL) of each EM.

Figure 2. High-performance liquid chromatography-electrospray ionization-tandem mass spectrometry selected reaction monitoring (SRM) chromatographic profiles of 15 estrogens and estrogen metabolites (EM) in a charcoal-stripped human serum sample that has been spiked with each EM to a final concentration of 40 pg/mL. Approximately 2 pg of each EM was placed on the column. The X-axis is retention time and the Y-axis is relative intensity.

Sample Preparation Procedure. Our sample preparation procedures (Figure 1) were designed to distinguish (1) free (unconjugated) EM and total ((conjugated + unconjugated) EM. For the measurement of total serum EM, 20 µL of the d-EM working standard solution (0.16 ng of each d-EM) was added to a 0.5 mL aliquot of serum, followed by the addition of 0.5 mL of freshly prepared enzymatic hydrolysis buffer containing 2 mg of L-ascorbic acid, 5 µL of β-glucuronidase/sulfatase, and 0.5 mL of 0.15 M sodium acetate buffer (pH 4.1) as previously described.22,23 Samples were incubated for 20 h at 37 °C. After hydrolysis with glucuronidase and sulfatase, each sample underwent slow inverse extraction at 8 rpm (RKVSD, ATR, Inc., Laurel, MD) with 8 mL of dichloromethane for 30 min. After extraction, the aqueous layer was discarded and the organic solvent portion was transferred into a clean glass tube and evaporated to dryness at 60 °C under nitrogen gas (Reacti-Vap III, Pierce, Rockford, IL). To each dried sample, 100 µL of 0.1 M sodium bicarbonate buffer (pH at 9.0) and 100 µL of dansyl chloride solution (1 mg/ mL in acetone) were added. After vortexing, samples were heated at 60 °C (Reacti-Therm III Heating Module, Pierce, Rockford, IL) (22) Xu, X.; Veenstra, T. D.; Fox, S. D.; Roman, J. M.; Issaq, H. J.; Falk, R.; Saavedra, J. E.; Keefer, L. K.; Ziegler, R. G. Anal. Chem. 2005, 77, 66466654. (23) Taylor, J. I.; Grace, P. B.; Bingham, S. A. Anal. Biochem. 2005, 341, 220229.

Analytical Chemistry, Vol. 79, No. 20, October 15, 2007

7815

Table 2. Accuracy and Intrabatch Precision of HPLC-ESI-MS2 Measurement of Estrogens and Estrogen Metabolites (EM) in Serum 8 pg EM/mL serum (26.5-29.6 fmol EM/mL)

40 pg EM/mL serum (132.4-148.0 fmol EM/mL)

160 pg EM/mL serum (529.5-592.2 fmol EM/mL)

EM

accuracy (%)a

precision (%)b

accuracy (%)a

precision (%)b

accuracy (%)a

precision (%)b

E1 E2 E3 2-OHE1 2-MeOE1 2-OHE2 2-MeOE2 3-MeOE1 4-OHE1 4-MeOE1 4-MeOE2 16R-OHE1 17-epiE3 16-ketoE2 16-epiE3

92.5 94.5 95.1 91.9 112.5 93.8 97.5 91.3 97.5 102.0 91.9 96.0 93.8 95.0 92.5

15.9 10.4 11.2 26.4 22.7 19.2 28.9 6.7 21.2 20.8 29.8 10.7 24.9 11.6 20.7

121.4 119.5 106.0 98.3 117.4 90.8 91.3 112.0 96.3 118.5 107.5 114.9 112.2 116.8 97.3

4.2 6.3 12.2 13.6 6.5 5.2 5.9 8.9 16.4 7.7 4.9 7.2 10.9 5.3 9.3

99.8 112.7 111.9 107.2 114.6 97.7 102.2 117.5 111.4 110.7 101.4 100.1 102.3 116.3 98.2

2.5 3.8 13.0 5.1 10.2 3.5 6.4 5.8 3.1 9.8 4.2 3.8 10.1 6.4 4.3

a Accuracy was measured as the percent of the amount of EM known to be added to a serum sample that was recovered. It was based on the mean of four samples hydrolyzed, extracted, derivatized, and analyzed together in one batch. b Intrabatch precision (coefficient of variation) was measured by the percent relative standard deviation (RSD). It was based on four samples hydrolyzed, extracted, derivatized, and analyzed together in one batch.

for 5 min to form the EM and d-EM dansyl derivatives (EM-dansyl and d-EM-dansyl, respectively).22,24-26 Calibration standards and quality control samples were hydrolyzed, extracted, and derivatized following the same procedure as that used for the unknown serum samples. After derivatization, all samples were analyzed by HPLC-ESI-MS.2 For the measurement of free serum EM, the identical sample preparation procedures were used with the exclusion of the β-glucuronidase/sulfatase hydrolysis step. HPLC-ESI-MS2. HPLC-ESI-MS2 analysis was performed using a Finnigan TSQ Quantum-AM triple quadrupole mass spectrometer coupled with a Surveyor HPLC system (ThermoFinnigan, San Jose, CA). Both the HPLC and mass spectrometer were controlled by Xcalibur software (ThermoFinnigan). High performance liquid chromatography was carried out on a 150 mm long × 2.0 mm i.d. column packed with 4 µm Synergi Hydro-RP particles (Phenomenex, Torrance, CA) and maintained at 40 °C. A total of 20 µL of each sample was injected onto the column. The mobile phase, operating at a flow rate of 200 µL/min, consisted of methanol as solvent A and 0.1% (v/v) formic acid in water as solvent B. For the analysis of EM-dansyl and d-EMdansyl, a linear gradient changing the A/B solvent ratio from 72: 28 to 85:15 in 75 min was employed. After washing with 100% A for 12 min, the column was re-equilibrated with a mobile phase composition of 72:28 A/B for 13 min prior to the next injection. The general MS conditions were as follows: source, ESI; ion polarity, positive; spray voltage, 4600 V; sheath and auxiliary gas, nitrogen; sheath gas pressure, 49 arbitrary units; auxiliary gas pressure, 23 arbitrary units; ion transfer capillary temperature, 350 °C; scan type, selected reaction monitoring (SRM); collision gas, argon; collision gas pressure, 1.5 mTorr. The SRM conditions (24) Frei-Hausler, M.; Frei, R. W. J. Chromatogr. 1973, 84, 214-217. (25) Quirke, J. M. E.; Adams, C. L.; Van Berkel, G. J. Anal. Chem. 1994, 66, 1302-1315. (26) Anari, M. R.; Bakhtiar, R.; Zhu, B.; Huskey, S.; Franklin, R. B.; Evans, D. C. Anal. Chem. 2002, 74, 4136-4144.

7816 Analytical Chemistry, Vol. 79, No. 20, October 15, 2007

Table 3. Interbatch Precision of HPLC-ESI-MS2 Measurement of Estrogens and Estrogen Metabolites (EM) in Seruma

EM E1 E2 E3 2-OHE1 2-MeOE1 2-OHE2 2-MeOE2 3-MeOE1 4-OHE1 4-MeOE1 4-MeOE2 16R-OHE1 17-epiE3 16-ketoE2 16-epiE3

8 pg EM/mL 40 pg EM/mL 160 pg EM/mL serum serum serum (26.5-29.6 fmol (132.4-148.0 fmol (529.5-592.2 fmol EM/mL) EM/mL) EM/mL) 16.0 13.9 16.7 13.6 23.6 14.5 24.5 7.8 10.1 21.9 24.2 10.2 28.8 19.3 14.3

11.7 7.0 11.3 7.3 14.0 6.9 10.2 8.1 8.8 14.6 10.2 11.9 8.7 11.0 9.1

3.8 6.9 5.3 8.0 7.3 3.8 5.1 8.2 4.6 6.2 7.9 9.3 9.6 7.2 3.7

a Interbatch precision (coefficients of variation) was measured by the percent relative standard deviation (RSD). It was based on the means for four samples hydrolyzed, extracted, derivatized, and analyzed in each of four batches.

for the protonated molecules [MH]+ of EM-dansyl and d-EMdansyl are described in Table 1. The following MS parameters were used for all experiments: scan width, 0.7 u; scan time, 0.50 s; Q1 peak width, 0.70 u full-width half-maximum (fwhm); Q3 peak width, 0.70 u fwhm. Ion Suppression Study. To screen for potential ion suppression problems, a dilute solution of the dansylated EM and d-EM mixture was infused at a constant rate into the effluent flowing from the LC system into the mass spectrometer to create an elevated, but constant, baseline signal using the ESI-MS2 SRM conditions as described. After a steady baseline was obtained, a 20 µL dansylated blank serum sample extract was injected into the LC system. Any eluted material from the injected serum

Figure 3. High-performance liquid chromatography-electrospray ionization-tandem mass spectrometry selected reaction monitoring (SRM) chromatographic profiles of (A) free (unconjugated) and (B) total (conjugated + unconjugated) estrogens and estrogen metabolites in a serum sample from a premenopausal woman in the luteal phase of her menstrual cycle. The X-axis is retention time and the Y-axis is relative intensity.

sample extract that suppresses ionization in the mass spectrometer would cause a drop in the baseline intensity.27 In addition, the same protocol was used to monitor any SRM intensity drop during a blank injection. Quantitation of EM. Quantitation of serum EM was carried out using Xcalibur Quan Browser (ThermoFinnigan). Calibration curves for each EM were constructed by plotting EM-dansyl/dEM-dansyl peak area ratios obtained from calibration standards versus amounts of EM and fitting these data using linear regression with 1/X weighting. The amount of each EM in a serum sample was then interpolated using this linear function. Deuteriums positioned R to the carbonyl group of labeled ketolic estrogens were especially susceptible to exchange loss during sample preparation and analysis. To ensure the accuracy of the quantitative analyses, only deuterium-labeled EM that did not undergo exchange loss under the conditions used were employed as internal standards in this study. On the basis of the similarity of structures and retention times, d4-E2 was used as the internal standard for E2 and E1; d3-E3 for E3, 16-ketoE2, and 16R-OHE1; d3-16-epiE3 for 16-epiE3 and 17-epiE3; d5-2-MeOE2 for 2-MeOE2, 4-MeOE2, 2-MeOE1, 4-MeOE1, and 3-MeOE1; and d5-2-OHE2 for 2-OHE2, 2-OHE1, and 4-OHE1. Absolute Recovery of EM after Hydrolysis and Extraction. To one set of six 0.5 mL aliquots of the charcoal-stripped human serum, 20 µL of the EM working standard solution (0.16 ng of each EM) was added, followed by the hydrolysis and extraction (27) King, R.; Bonfiglio, R.; Fernandez-Metzler, C.; Miller-Stein, C.; Olah, T. J. Am. Soc. Mass Spectrom. 2000, 11, 942-950.

procedure described above. A second set of six 0.5 mL aliquots of the charcoal-stripped human serum was treated identically, except that the EM solution was added after the hydrolysis and extraction procedure. Both sets of samples were derivatized and analyzed in consecutive LC-MS2 analyses. The absolute recovery of each EM after hydrolysis and extraction was calculated by dividing the mean EM-dansyl peak area from the second set of samples into the mean EM-dansyl peak area measured for the first set. Accuracy and Precision of the Serum EM Analysis. To assess the accuracy and precision of this method, four replicate 0.5 mL aliquots from each of the three QC samples [8, 40, and 160 pg/mL (26.5-29.6, 132.4-148.0, and 529.5-592.2 fmol/mL) of each EM in serum] were hydrolyzed, extracted, derivatized, and analyzed in each of four different batches. Accuracy was measured as the percent of the amount of EM known to be added that was recovered and was based on the mean of four samples analyzed in one batch.28,29 The intrabatch and interbatch precision (coefficients of variation) were measured by the percent relative standard deviation (RSD).28,29 Intrabatch precision was based on four samples analyzed in one batch, and interbatch precision was based on the means for four samples in each of four batches.

(28) Duncan, M. W.; Gale, P. J.; Yergey, A. L. Principles of Quantitative Mass Spectrometry, 1st ed.; Rockpool Press: Denver, CO, 2002. (29) Swartz, M.; Krull, I. S. Analytical Method Development and Validation, 1st ed.; Marcel Dekker Inc.: New York, 1997.

Analytical Chemistry, Vol. 79, No. 20, October 15, 2007

7817

Figure 4. High-performance liquid chromatography-electrospray ionization-tandem mass spectrometry selected reaction monitoring (SRM) chromatographic profiles of (A) free (unconjugated) and (B) total (conjugated + unconjugated) estrogens and estrogen metabolites in a serum samples from a premenopausal woman in the follicular phase of her menstrual cycle. The X-axis is retention time and the Y-axis is relative intensity.

RESULTS AND DISCUSSION To establish the analytical parameters required for quantitative measurement, 15 unconjugated EM were added to charcoalstripped human serum to a final concentration of 40 pg/mL of each EM. After extraction and derivatization of the sample (as described in the Experimental Section), 20 µL of the extract’s 200 µL final volume (containing 2 pg of each EM) was injected into the column and analyzed by HPLC-ESI-MS2 operating in the SRM mode. The chromatographic profile resulting from this analysis is shown in Figure 2. With the use of a methanol/water linear gradient, all 15 EM were resolved by reversed phase C18 chromatography within 70 min. The only EMs that were not fully baseline resolved were 16R-OHE1 and 16-ketoE2. In almost every case, the peak shapes showed excellent symmetry. No serum matrix-induced ion suppression was observed in EM-dansyl and d-EM-dansyl eluted regions using the HPLC-ESI-MS2 SRM conditions described in this study. In addition no drop in the SRM intensities of EM-dansyl and d-EM-dansyl were observed at the end of run. Even though only a simple serum sample preparation and a methanol/water gradient elution were employed for the current method, it is capable of simultaneously quantifying 15 endogenous EM in human sera. An effective clinical assay will have the ability to measure specific analytes accurately over a wide range of concentrations with adequate sensitivity. With our method, the calibration curves for quantifying each serum EM in this study were linear over a 103-fold concentration range with linear regression correlation coefficients ranging from 0.9977 to 0.9986. The standard error of 7818 Analytical Chemistry, Vol. 79, No. 20, October 15, 2007

the estimate (SEE) and the relative standard error of the estimate (RSEE) for the linear regression correlation coefficients ranged from 0.0315 to 0.2035 and from 3.3% to 4.8%, respectively, which indicated that the confidence intervals of the slopes were tight and the intercepts were close to zero. The signal-to-noise (S/N) ratios obtained from analyzing 0.5 mL of the lowest QC sample [8 pg/mL (26.5-29.6 fmol/mL) of each EM, representing 0.4 pg (1.3-1.5 fmol) of each EM on-column] were consistently greater than 50. The accuracy, intrabatch precision, and interbatch precision at this EM level were 91-113%, 6.7-29.8%, and 7.8-28.8%, respectively, which met our criteria for an acceptable limit of quantitation (LOQ). Therefore, the LOQ for each EM in serum, using the described method, was 8 pg/mL (26.5-29.6 fmol) EM and 0.4 pg (1.3-1.5 fmol) EM oncolumn. Although, in general, atmospheric pressure chemical ionization (APCI) is less influenced by matrix effects, the HPLCAPCI-MS2 SRM and HPLC-APCI/atmospheric pressure photoionization (APPI)-MS2 SRM methods that we tested performed poorly in detecting and quantitating the EM-dansyl species in the low QC samples (8 pg/mL EM) used in this study. However, the ESI-based method described in this study allows the coupling of nanoflow capillary-LC with tandem MS, which resulted in a significant improvement in sensitivity and limit of quantitation. Since there were no previous studies that provided representative concentrations of all individual EM in human serum, physiological levels of serum 17β-estradiol in premenopausal women, postmenopausal women, and men were used as the reference for the three

Figure 5. High-performance liquid chromatography-electrospray ionization-tandem mass spectrometry selected reaction monitoring (SRM) chromatographic profiles of (A) free (unconjugated) and (B) total (conjugated + unconjugated) estrogens and estrogen metabolites in a serum sample from a postmenopausal woman. The X-axis is retention time and the Y-axis is relative intensity.

levels of QC samples (8, 40, and 160 pg of each EM per mL of charcoal-stripped human serum, which is equivalent to 26.5-29.6, 132.4-148.0, and 529.5-592.2 fmol EM/mL). Overall, excellent accuracy and precision were obtained for the analysis of all three quality control serum samples (Tables 2 and 3). Each sample was taken through the complete process of hydrolysis, extraction, derivatization, and HPLC-ESI-MS2 analysis. The accuracy for the 8, 40, and 160 pg/mL EM (26.5-29.6, 132.4148.0, and 529.5-592.2 fmol/mL EM) QC samples was 91-113%, 91-121%, and 98-118%, respectively, as shown in Table 2. The intrabatch precisions, based on four replicate samples assayed in one batch, were 6.7-29.8%, 4.2-16.4%, and 2.5-13.0% RSD for the 8, 40, and 160 pg/mL EM (26.5-29.6, 132.4-148.0, and 529.5592.2 fmol EM/mL) QC samples, respectively (Table 2). The interbatch precisions, based on the means for four replicate samples assayed in each of four batches, were 7.8-28.8%, 6.914.6%, and 3.7-9.6% RSD for the 8, 40, and 160 pg/mL EM (26.529.6, 132.4-148.0, and 529.5-592.2 fmol EM/mL) QC samples, respectively (Table 3). The absolute recovery of each EM after hydrolysis and extraction from a serum sample containing 320 pg/mL of each EM ranged from 84.5 to 92.2%. Once the HPLC and MS conditions necessary to resolve and quantify the 15 EM in spiked serum samples had been established and validated, we proceeded to measure the levels of free and total EM in human serum samples: two from premenopausal women in the luteal phase of their menstrual cycle, two from premenopausal women in the follicular phase of their menstrual cycle, and two from postmenopausal women. For each serum sample, four aliquots of the same sample were analyzed, with each

taken through complete sample processing and data acquisition. The HPLC-ESI-MS2 SRM profiles of free EM in one of the serum samples obtained from a premenopausal luteal phase woman is shown in Figure 3A. Five EM, E1, E2, E3, 2-MeOE1, and 2-MeOE2, could be readily quantified. The SRM profiles from the same serum sample, analyzed after a β-glucuronidase/sulfatase hydrolysis step so that the total amount of each EM could be measured, are shown in Figure 3B. Not only were the 5 free EM detected in greater abundance, but the other 10 EM could also be detected and quantified. These results suggest that these 10 EM exist mainly in conjugated form, as β-glucuronides or sulfates, in premenopausal luteal phase serum. The same analyses for serum from a premenopausal follicular phase woman are shown in Figure 4A,B. As observed for the premenopausal luteal phase serum, free forms of E1, E2, E3, 2-MeOE1, and 2-MeOE2 could be detected and quantified (Figure 4A). Incorporation of the β-glucuronidase/sulfatase hydrolysis step enabled all 15 EM to be measured (Figure 4B). As with the premenopausal luteal phase serum sample, these results show that circulating EM exist predominantly as β-glucuronide and sulfate conjugates. Similar analyses to measure free and total EM were performed on two serum samples obtained from postmenopausal women. In the analysis of free EM, 2-MeOE2 could only be detected in serum obtained from one of the two postmenopausal women (Figure 5A). As with the premenopausal serum samples, inclusion of the β-glucuronidase/sulfatase hydrolysis step enabled all 15 EM to be detected and quantified and indicated that in postmenopausal Analytical Chemistry, Vol. 79, No. 20, October 15, 2007

7819

Table 4. Concentrations of Free (Unconjugated) and Total (Conjugated + Unconjugated) Estrogens and Estrogen Metabolites (EM) in Serum from Two Premenopausal Luteal Phase (LP-1 and LP-2), Two Premenopausal Follicular Phase (FP-1 and FP-2), and Two Postmenopausal (PostM-1 and PostM-2) Womena LP-1

LP-2

FP-1

FP-2

PostM-1

PostM-2

E1 E2 16R-OHE1 16-ketoE2 E3 16-epiE3 17-epiE3 2-OHE1 2-OHE2 4-OHE1 2-MeOE1 2-MeOE2 3-MeOE1 4-MeOE1 4-MeOE2

58.1 (215.2) 79.8 (293.2) ND ND 17.2 (59.8) ND ND ND ND ND 19.8 (65.9) 4.5 (14.8)b ND ND ND

49.2 (182.2) 65.5 (240.6) ND ND 15.7 (54.5) ND ND ND ND ND 11.1 (36.8) 3.5 (11.5)b ND ND ND

Free EM 17.2 (63.7) 19.5 (71.8) ND ND 9.3 (32.2) ND ND ND ND ND 7.3 (24.4)b 2.3 (7.5)b ND ND ND

83.3 (308.3) 108.2 (397.4) ND ND 23.9 (82.8) ND ND ND ND ND 27.3 (91.1) 7.6 (25.0)b ND ND ND

24.6 (91.2) 8.7 (31.9) ND ND 7.5 (26.0) b ND ND ND ND ND 2.6 (8.8) 1.1 (3.6)b ND ND ND

40.7 (150.6) 21.3 (78.3) ND ND 8.3 (28.7) ND ND ND ND ND 4.9 (16.3) ND ND ND ND

E1 E2 16R-OHE1 16-ketoE2 E3 16-epiE3 17-epiE3 2-OHE1 2-OHE2 4-OHE1 2-MeOE1 2-MeOE2 3-MeOE1 4-MeOE1 4-MeOE2

777.9 (2878.8) 174.2 (640.1) 13.4 (46.8) 25.4 (88.7) 45.2 (156.8) 6.7 (23.4)b 2.6 (8.9)b 514.6 (1798.0) 32.7 (113.5) 64.4 (224.9) 41.9 (139.7) 10.3 (34.1) 9.4 (31.3) 1.0 (3.2)b 1.4 (4.7)b

671.9 (2486.8) 122.0 (448.1) 16.6 (58.0) 16.0 (56.0) 57.0 (197.8) 8.1 (28.1) 4.5 (15.6)b 201.6 (704.3) 48.0 (166.4) 31.8 (111.1) 29.5 (98.1) 8.4 (27.6) 4.8 (16.1)b 1.0 (3.4)b 0.9 (2.8)b

Total EM 192.6 (712.8) 31.2 (114.7) 10.3 (36.1) 11.6 (40.6) 25.1 (87.1) 3.8 (13.1)b 2.2 (7.8)b 75.9 (265.1) 15.6 (54.2) 14.5 (50.6) 10.4 (34.5) 5.0 (16.7)b 2.7 (8.9)b 0.6 (1.8)b 0.8 (2.8)b

1270.2 (4701.1) 218.5 (802.6) 45.3 (158.4) 48.7 (170.0) 75.0 (260.4) 10.2 (35.4) 5.5 (19.0)b 522.8 (1826.6) 39.6 (137.3) 78.5 (274.4) 64.7 (215.4) 16.2 (53.5) 3.9 (13.1)b 1.0 (3.5)b 2.5 (8.4)b

376.7 (1394.2) 13.2 (48.6) 8.4 (29.4) 11.6 (40.4) 34.0 (117.9) 4.3 (14.8)b 2.2 (7.7)b 64.0 (223.7) 10.5 (36.5) 9.2 (32.2) 4.3 (14.3)b 2.1 (7.1)b 1.4 (4.8)b 0.4 (1.2)b 1.0 (3.4)b

507.4 (1877.9) 89.7 (329.4) 9.2 (32.0) 10.4 (36.2) 21.8 (75.8) 3.3 (11.3)b 1.5 (5.1)b 81.0 (282.9) 11.6 (40.3) 13.7 (47.8) 8.5 (28.4) 2.9 (9.6)b 1.2 (4.0)b 0.2 (0.8)b 1.1 (3.7)b

a Concentration is expressed as the mean of four replicate aliquots hydrolyzed, extracted, derivatized, and analyzed together in one batch. It is presented in both pg/mL and (fmol/mL). ND: analyte was not detected. b The estimated concentration is below the lower limit of quantitation but above the limit of detection (signal-to-noise ratio > 3).

women, circulating EM exist primarily in the conjugated form (Figure 5B). The serum concentrations for free and total EM in two premenopausal luteal phase women, two premenopausal follicular phase women, and two postmenopausal women are presented d in Table 4. In all six serum samples, total E1 was at the highest concentration of the 15 EM. In the four premenopausal serum samples, total 2-OHE1 and total E2 were the next two most abundant EM, generally followed by total E3 and total 4-OHE1. In the postmenopausal serum samples also, total 2-OHE1 tended to be relatively abundant. As described above, only five free EM (E1, E2, E3, 2-MeOE1, and 2-MeOE2) could be detected in serum from the premenopausal and postmenopausal women. In the premenopausal women, free E2 was at the highest concentration while in the postmenopausal women free E1 was highest (Table 4). Next most abundant among the free EM in circulation were E3 and 2-MeOE1 with free 2-MeOE2 at the lowest concentration. CONCLUSIONS As time passes, more and more evidence is accumulating that demonstrates a link between endogenous estrogen and the risk of developing human breast cancer. Specifically, epidemiologic studies have shown that women with high levels of urinary and circulating estrogen are at an elevated risk of postmenopausal 7820 Analytical Chemistry, Vol. 79, No. 20, October 15, 2007

breast cancer and possibly premenopausal breast cancer. However, the contributions of individual estrogen metabolites and patterns of estrogen metabolism to the underlying mechanisms of human breast carcinogenesis remain unclear. To evaluate the role of estrogen metabolism, it is critical that we be able to quantitatively measure each individual metabolite in collected biospecimens. This manuscript presents a sensitive, specific, accurate, and precise HPLC-ESI-MS2 method for simultaneously measuring 15 endogenous EM in human serum. The method can differentiate between the total concentration of each EM, including conjugated and unconjugated forms, and the concentration of the free, or unconjugated, form of each EM. Each standard curve was linear over a 103-fold concentration range (0.2-200 pg of EM oncolumn), with the RSEE for the linear regression lines ranging from 3.3 to 4.8%. The lower limit of quantitation for each EM was 8 pg/mL (26.5-29.6 fmol/mL) in serum and 0.4 pg (1.3-1.5 fmol) on-column. Accuracy ranged from 91 to 113%. Intrabatch precision, for samples prepared concurrently, ranged from 7 to 30% RSD; and interbatch precision, for samples prepared in several batches, ranged from 8 to 29% RSD. Furthermore, this method is robust and rapid enough to handle the large numbers of samples collected in epidemiologic and clinical studies. This new analytic tool for quantitating the absolute amounts of total and free EM in circulation should benefit research on the mechanisms underlying

human breast carcinogenesis and, therefore, suggest new approaches to breast cancer prevention, screening, and treatment. ACKNOWLEDGMENT This project has been funded in whole or in part with federal funds from the National Cancer Institute, National Institutes of Health, under Contract N01-CO-12400, and by the Intramural Research Program of the NIH, National Cancer Institute, Center for Cancer Research, and Division of Cancer Epidemiology and

Genetics. The content of this publication does not necessarily reflect the views or policies of the Department of Health and Human Services nor does mention of trade names, commercial products, or organizations imply endorsement by the United States Government. Received for review March 9, 2007. Accepted July 26, 2007. AC070494J

Analytical Chemistry, Vol. 79, No. 20, October 15, 2007

7821