Determination of Estrogens in Drinking and Waste Waters in the Salt Lake Valley via Solid-Phase Extraction Liquid Chromatography/Mass Spectrometry Method Laura Cotter and Heather McGirk CHEM 3000 Section 003 August 1, 2008
In recent years there has been an increase in the amount of research done on the presence of pharmaceuticals in water. Hormones have especially drawn attention given their potential impact on marine life and human health. The increase in the number of women and men on hormone replacement therapy has also increased the amount of human hormones passing into water. Estrogens are of particular concern. Studies have shown that at relatively low levels, approaching 1-10 ng/L of 17-β-estradiol and 0.1 ng/L of 17α-ethinylestradiol, some species of male fish show feminizing characteristics. Due to this effect, many states have developed methods to detect and remove estrogens from their water, but so far Utah has not adopted any such method. This study examined the amount of estrogen present in water in the Salt Lake Valley. Samples of both drinking water and final effluent wastewater were taken from nine cities/areas in the valley and tested for levels of estrogen using solid-phase extraction and liquid chromatography/mass spectrometry methods. Estrogen levels in drinking water were below the 10 ng/L detection limit of the LC/MS. Estrone was found in all of the wastewater samples, but estriol (E3), 17-β-estradiol (E2) and 17-α-ethinylestradiol (EE2) were each found in only two of the samples. However, those levels that were detected of E2 and EE2 appear to be high enough to have an affect on some species of fish.
2 Introduction This study determined the levels of estrogen found in final effluent wastewater and drinking water. As of March 2008, the state of Utah does not “specifically test for [estrogens]… It is more a problem for the aquatic habitat and the aquatic life. Their threshold…is much, much lower than it would be for humans 1.” Even at low levels estrogens have a major impact on wildlife, and they are more dangerous than many industrial chemicals2. The danger in estrogen comes from their ability to act as endocrine disrupters3. This means that they are able to interfere with the reproductive functions of vertebrates. In certain species of fish, even levels of 1-10 ng/L of 17-β-estradiol and 0.1 ng/L 17-α-ethinylestradiol3 have had effects of feminization, to the point where male fish have begun producing female yolk proteins4. Previous research has shown that when male fathead minnows are exposed to 5 ng/L over the course of a year, they begin producing eggs and the population essentially stops reproducing5. Although the effects on humans by endocrine disruptors found in water have not been thoroughly studied, it has been speculated that they may be linked with early onset puberty, low sperm count and breast and testicular cancers6. In order to determine whether estrogen levels are high enough to be of concern to Utahns, samples were taken from various places across the Salt Lake Valley. These sites included Murray, Cottonwood Heights, Magna, Holladay, Sugarhouse and the University of Utah. Samples were also taken of final effluent wastewater from South Valley, Central Valley and Salt Lake City Water Reclamation Facilities. Two wastewater samples were taken from each plant; one on July 15 and the second on July 18, 2008. The estrogens measured were estrone (E1), 17-β-estradiol (E2), estriol (E3), and 17-αethinylestradiol (EE2). Estrone, estradiol, and estriol are naturally excreted by cycling women (10-100 μg per day), and higher levels are excreted by pregnant women (up to 30 mg per day, mainly consisting of estriol)3. Men excrete estrone at around 5 μg per day7. Another source of these estrogens (E1, E2, and E3) is the use of pharmaceuticals for (1) (2) (3) (4) (5) (6) (7)
Choate-Nielsen, A. (March 11, 2008). Water worries: Utah officials aren’t testing for chemicals. Deseret Morning News, p. A01 Rodriguez-Mozaz, S.; Lopez de Alda, M. J.; Barceló, D. Anal. Chem. 2004, 76, 6998-7006. Baronti, C.; Curini, R.; D’Ascenzo, G.; Di Corcia, A.; Gentili, A.; Samperi, R. Environ. Sci. Technol. 2000, 34, 5059-5066. Kuch, H.; Ballschimter, K. Environ. Sci. Technol. 2001, 35, 3201-3206 Halford, B. C&EN 2008, 86, 13-17. Xiao, X.; McCalley, D.; McEvoy, J. J.Chromatogr., A 2001, 923, 195-204 Belfroid, A.C.; Van der Horst, A.; Vethaak, A.D.; Schafer, A.J.; Rijs, G.; Wegener, J.; Cofino, W. Sci. Total Environ. 1999, 225, 101-108.
3 hormone replacement therapy. Ethinylestradiol, a synthetic estrogen and the most potent, is a main component of many birth control pills. It was hypothesized that these combined sources of estrogens would likely result in levels that were in the parts-perbillion, or μg/L, range. There have been a few standard methods of detecting estrogens in water, but the most common is solid phase extraction (SPE) followed by either gas chromatography/mass spectrometry (GC/MS) or liquid chromatography/mass spectrometry (LC/MS). A standard F-test was performed to compare GC (σ = 9-15%, n = 6)2 and LC (σ = <4%, n = 4)4. The test showed that LC does not have a significant improvement in precision over GC. However, GC is limited by factors including high molecular weights and nonvolatility. Procedures using GC/MS also require derivatization of the estrogens before measurements can be taken. Liquid chromatography does not require derivatization, which saves on both time and materials. Solid-phase extraction is usually employed instead of liquid-liquid extraction for several reasons, including the ability to do on-line coupling with chromatography and a higher degree of selectivity8. Therefore, SPE followed by LC/MS was the method used in this study.
Experimental Procedure Collection The samples were collected in 1 liter amber glass bottles from tap water in homes in the areas of the Salt Lake Valley mentioned in the introduction. In order to prevent degradation and clean up any residual chlorine, 50 mg/L of ascorbic acid and 200 mg/L of sodium azide were added to the samples. They were then kept at 4°C and in the dark until filtering and extraction were begun. Wastewater was also collected in amber glass bottles, treated with ascorbic acid and sodium azide and filtered with 0.45 µm Whatman filter paper prior to experimentation. All sample volumes were measured before proceeding. One hundred ng (10 µL of a 10 mg/L methanolic stock solution) of the surrogate standard, estrone-d4, was then added to each filtrate to account for any loss of estrogens throughout the procedure. (8) Ramsey, E. D.; Minty, B.; Rees, A. T. Anal. Commun. 1997, 34, 261-264.
4
Solid-Phase Extraction The solid-phase extraction protocol was adapted from a 9
procedure by Vanderford et al.
Sample Collection
An Isolute C18 (EC) (500
mg/10 mL) was used for each sample and placed on a vacuum manifold. Each cartridge was then preconditioned by passing
Filtration Solid-Phase Extraction
through 5 mL methyl tert-butyl ether (MTBE), 5 mL methanol, and 5 mL of 40% methanol in deionized water. The samples were then passed through the SPE cartridges at a rate of
Dry Down LC/MS
approximately 15 mL/min. The cartridges were then rinsed with 5 mL of reagent grade water. The estrogens and other organics were eluted by flushing with 5 mL of a 10% methanol in MTBE solution and 5 mL of methanol. The samples were
Analysis Fig. 1. Diagram summarizing procedure for determination of estrogens.
collected in test tubes which were then evaporated to dryness on a Büchi Rotavapor R-124, rotating the test tubes at 83 rpm, in a Büchi Waterbath set to 38ºC. The dried samples were reconstituted in 990 μL of MeOH and 10 μL (100 ng) of 17β-estradiol-d3, the internal standard. The test tubes were then vortexed for approximately 20 sec, and each reconstituted sample was transferred to a LC/MS vial using separate, clean Pasteur pipettes. The samples were stored in the freezer until LC/MS was performed. Liquid Chromatography/Mass Spectrometry %A
%B
The final solution was analyzed using Acquity
Time
Flow
(min)
Rate
UPLC HSST3 1.8μm LC/MS with a column size
(mL/min)
of 2.1 x 50 mm and an electrospray interface
Initial
0.600
75.0
25.0
operated in the negative ion mode2. The mass
0.50
0.600
75.0
25.0
spectrometer used was a single quad, as a triple
3.00
0.600
30.0
70.0
3.50
0.600
10.0
90.0
5.00
0.600
75.0
25.0
Table 1. LC Gradient Elution Conditions Used for the Separation of the Estrogens.
quad was not available at the time of experimentation. The solvents were water (A) and acetonitrile (B). Each run had a retention window of 5.00 min and a dwell time of 0.060 sec.
(9) Vanderford, B.J., Pearson, R.A., Rexing, D.J., and Snyder, S.A. Anal. Chem. 75, 6265-6274.
5 The internal standard provided an approximation of the recovery percentage of the LC/MS. Standards of estrone (E1), 17-β-estradiol (E2), estriol (E2) and 17-αethinylestradiol (EE2), were taken from a stock solution of 1000 ng/L and diluted to concentrations of 10 ng/L, 50 ng/L, 100 ng/L, 500 ng/L and 1000 ng/L. The standards were run in triplicate. A plot of the ratio of the area under the internal standard curve to the areas of the estrogens studied versus the actual concentrations provided a method of determining the relative concentrations of the estrogens in the drinking and wastewater samples.
Fig. 2. Chromatogram of 1000 ppb standard: Ethinylestradiol (295), estriol (287), estradiol-d3 (274.1), estrone-d4 (273.2), estradiol (271.1), and estrone (269.2).
Data and Results Three chromatograms were obtained for each standard solution. Estrogens were identified by retention time and molecular weight. Retention times were as follows: estriol – 1.04 min; estradiol – 1.75; ethinylestradiol – 1.88; and estrone – 1.94. The deuterated forms of estradiol and estrone had nearly identical retention times.
6 Peak areas of each estrogen from the standards were averaged to give calibration curves of the ratio of the internal standard to the area under the curve of the estrogen versus actual concentration. Standard deviations and 95% confidence intervals were also calculated using the standard peak areas. Concentrations of estrogens in the samples were calculated by using the linear regression slope and intercept from the calibration curves. The error in the intercept was much smaller than the value of the intercept, so the trendline could not be forced to zero.
70 60 50 40 30 20 10 0 ‐10 0
LC‐MS Calibration Curve ‐ Estriol (E3 287.2) w/I.S. Ratio of Response to I.S. Response (Area Under Curve)
Ratio of Response to I.S. Response (Area Under Curve)
LC‐MS Calibration Curve ‐ Estrone (E1 269) w/I.S. y = 0.0533x ‐ 3.0677 2
R = 0.9804
200
400
600
800
1000
1200
80 y = 0.0703x ‐ 4.8376
60
2
R = 0.9579
40 20 0 ‐20
0
200
Actual Concentration (ug/L)
800
1000
1200
LC‐MS Calibration Curve ‐ Estradiol (E2 271) w/I.S. 120 Ratio of Response to I.S. Response (Area Under Curve)
40 Ratio of Response to I.S. Response (Area Under Curve)
600
Actual Concentration (ug/L)
LC‐MS Calibration Curve ‐ Ethinylestradiol (EE2 295.2) w/I.S. y = 0.0323x ‐ 1.7642
30
2
R = 0.9821
20 10 0 ‐10
400
0
200
400
600
800
Actual Concentration (ug/L)
1000
1200
100
y = 0.0934x ‐ 7.0525
80
2
R = 0.9359
60 40 20 0 ‐20 0
200
400
600
800
Actual Concentration (ug/L)
Fig. 3. Calibration curves of estrone, estriol, ethinylestradiol, and estradiol. Error bars show 95% confidence intervals. Linear regression equations are shown on graphs.
1000
1200
7
Collection Site
Initial
Percent
E1
sE1
E2
sE2
E3
sE3
EE2
sEE2
Volume
Recovery
(ppb)
(ppb)
(ppb)
(ppb)
(ppb)
(ppb)
(ppb)
(ppb)
(mL) Holladay
922
76.67
n.d.
N/A
n.d.
N/A
n.d.
N/A
n.d.
N/A
Cottonwood Heights
980
76.73
n.d.
N/A
n.d.
N/A
n.d.
N/A
n.d.
N/A
Magna
900
76.75
n.d.
N/A
n.d.
N/A
n.d.
N/A
n.d.
N/A
South Valley WRF (7/15)
935
111.9
0.090
0.003
n.d.
N/A
0.15
0.01
0.91
0.04
U of U
980
89.55
n.d.
N/A
n.d.
N/A
n.d.
N/A
n.d.
N/A
Murray
934
93.86
n.d.
N/A
n.d.
N/A
n.d.
N/A
n.d.
N/A
SLC WRF (7/18)
938
118.0
0.177
0.007
0.14
0.01
0.29
0.02
0.27
0.01
SLC WRF (7/15)
929
85.50
0.091
0.004
n.d.
N/A
0.131
0.009
n.d.
N/A
South Valley WRF (7/18)
1038
116.8
0.063
0.003
n.d.
N/A
0.129
0.009
n.d.
N/A
Central Valley WRF (7/15)
911
136.2
0.067
0.003
n.d.
N/A
n.d.
N/A
n.d.
N/A
Central Valley WRF (7/18)
998
110.7
0.065
0.003
0.091
0.006
n.d.
N/A
n.d.
N/A
Table 2. Estrogen concentrations from drinking and wastewater samples. Sx = the standard deviation. Percent recovery is based off of the internal standard (17-β-estradiol-d3). n.d. = not detected
Concentrations were calculated as follows: For estrone at SLC WRF (7/15): Calibration curve concentration (CCC) = (peak area/IS peak area - y-intercept) / slope CCC = [(1901/2426) – (-3.0577)] / 0.0533 = 72.1 ppb Concentration Factor (CF) = initial volume / final volume CF = 929 mL / 1 mL = 929 mL Percent Recovery (PR) = CCCE1-d4 / (initial concentration * CF / 100) PR = 87.4 / [ (0.11 ppb * 929 mL) / 100 ] = 85.5% Estrogen Concentration = CCC / PR / 100 / CF [E1] = 72.1 / ( 85.5 / 100 ) / 929 = 0.091 ppb No drinking water samples had estrogens above the detection level (10 ng/L). Estrone was present in all wastewater samples and estriol was present in South Valley WRF and SLC WRF wastewater but was not detected in the Central Valley WRF samples. 17-β-estradiol levels in final effluent water from Central Valley and Salt
8 Lake City plants appear to be above levels that may affect male fish (1-10 ppt)3. South Valley and Salt Lake City appear to have levels of 17-Îą-ethinylestradiol that may affect male fish (0.1 ppt)3
Discussion and Conclusions After chromatographs of the standards were obtained, it became apparent that there was an impurity in the estrone-d4 surrogate standard. The impurity was of about the same mass and had a similar retention time as 17-β-estradiol. This made integration of the E2 peaks difficult and inaccurate. Until this error is corrected in a second set of experiments, no accurate determination of wastewater purity can be made. estrogen std 1000ppb_2 1.75 352619
%
100 0.28 1815
0
0.40 3912
0.25 d 1000
0.50 b 2
1.42 4314
1.03 171
0.75
1.00
1.25
1.50
1.80 2.06 192744 221
1.75
2.00
2.35 2.48 2.65 369 1103 791
2.25
2.50
2.76 600
2.75
Fig. 4. E2 peak is at 1.75
Although most recovery rates were feasible, a few from wastewater samples appear to have gained estrogens in the experimentation process. This could be due to matrix effects from the electrospray interference. 10 The stability of estrogens may also be problematic and lead to decreased recovery rates. For example, at higher temperatures nitrifying bacteria that break down EE2 are more active than at lower temperatures. Since these samples were taken during summer months, EE2 levels may be higher during autumn and winter months11. South Valley WRF employs a final UV sterilization step which may lead to further removal of estrogens12. Estradiol is also naturally oxidized to estrone, so E2 in final effluent wastewater may be an imprecise measure of how much E2 entered the facility12. Estriol and estrone were observed in almost all of the wastewater samples. This is most likely due to the fact that both of these estrogens have low degradation constants and do not adsorb or absorb very easily. Estradiol has the highest degradation constant of the estrogens and will sorb very readily to most surfaces. Ethinylestradiol (10) (11) (12)
Richardson, S. and Ternes, T. Anal. Chem 2005, 77, 3807-3838. Vader, J.; Ginkel, C.; Sperling, F.; Jong, J.; de Boer, W.; Graaf, J.; Most, M.; Stokman, P. Chemosphere 2000, 41, 1239-1243 Andersen, H.; Siegrist, H.; Halling-Sorensen, B.; Ternes, T. Environ. Sci. Technol. 2003, 37, 4021-4026.
9 is only found in two samples; however both samples have relatively high concentrations. This may be due to ethinylestradiolâ&#x20AC;&#x2122;s extremely low degradation constant. It also tends to sorb less than estriol or estrone13. There are estrogens in final effluent wastewater in the Salt Lake Valley. However, without further experimentation and improvements, it is difficult to say exactly how much and if they are at levels immediately harmful to the environment. Another set of tests will need to be run with a new surrogate standard that has been tested for impurities. Derivitization may lead to better sensitivity and therefore better recovery rates, which would provide more accurate estimates of estrogen levels. (13) Ren, Y.; Nakano, K.; Nomura, M.; Chiba, N.; Nishimura, O. Water Research 2007, 41, 3089-3096