Extraction of bisphenol-A and 17β-estradiol from water samples via solid-phase extraction (SPE)

Selection of solid phase (cartridge type)

Selection of a suitable SPE sorbent depends on the interactions between the sorbent and the analyte(s) of interest. The most frequently used sorbents are silica gel, polymer sorbents, and graphitized or porous carbon (Zwir-Ferenc and Biziuk 2006). Sorbents rely on different adsorption mechanisms, such as van der Waals forces (non-polar interactions), hydrogen bonding, dipole-dipole forces (polar interactions), and/or cation-anion interactions (ionic interactions). The chemistry of some common SPE sorbents has been summarized by Zwir-Ferenc and Biziuk (2006). The most common sorption mechanism is based on hydrophobic interactions, and thus reverse-phase alkyl-bonded silica is the most common SPE medium (Pacakova et al. 2009). Reverse-phase SPE separates compounds based on their polarity.

As can be seen from Table 1, several different types of SPE cartridges have been employed to extract BPA and/or E2 from water samples. Some studies have compared the relative effectiveness of different SPE media (Liu et al. 2004, D’Archivio et al. 2007). Overall, the two most common SPE media used for the analysis of BPA and E2 are C₁₈ (typically octadecylsilane) and poly(divinylbenzene-co-N-vinylpyrrolidone), the latter of which is better known by the brand name Oasis HLB (Waters Corp., Milford, MA, USA). C₁₈ is a hydrophobic silica-bonded phase, typically used to adsorb hydrophobic analytes (including weakly hydrophobic analytes) from aqueous solutions (Zwir-Ferenc and Biziuk 2006). Oasis HLB is made from two monomers, hydrophilic N-vinylpyrrolidone and lipophilic divinylbenzene; this chemistry is intended to allow the retention of both polar and non-polar analytes (Tachon et al. 2008). Oasis HLB cartridges have sometimes been reported to have a higher recovery than other SPE sorbents like C₁₈ (OASIS HLB Sample Extraction Products 1998, Liu et al. 2004, D’Archivio et al. 2007).

The efficiency of extraction from water during SPE depends on the sample characteristics, the type of SPE sorbent or cartridge, the concentration of target material, and the eluent used to remove the target analytes from the solid phase (Liu et al. 2004, D’Archivio et al. 2007, Arditsoglou and Voutsa 2008). SPE cartridges have a finite sorption capacity. If the mass of EDC loaded onto the cartridge during the loading step exceeds the sorption capacity, then some fraction of the target analyte will not be retained on the cartridge, and hence will not be recovered during elution. Therefore, to ensure analyte recoveries close to 100%, it is important to consider the capacity of the SPE cartridge and the concentration of target analytes in the water sample.

Selection of eluent

After target analytes are extracted from an aqueous sample onto an SPE medium, the analytes need to be extracted from the SPE medium by an eluent (see Figure 1). The eluent interrupts the interactions between the sorbent and the analytes of interest, allowing the analytes to desorb (Yang et al. 1998). Desirable characteristics of an eluent include high extraction efficiency of the target analytes, low volume required, weak (or no) toxicity, and compatibility with the chromatographic system used (Liu et al. 2004, Arditsoglou and Voutsa 2008). Often, water-miscible organic solvents (e.g., acetone, acetonitrile, methanol, isopropanol) are chosen as eluents; however, as can be seen from Table 1, other organic solvents (dichloromethane, diethyl ether, ethyl acetate, hexane) have also been used in conjunction with SPE for analysis of BPA and E2. The eluents used most commonly for this application are methanol and ethyl acetate.

To concentrate the target analytes, the elution step needs to extract the analytes with a small volume of eluent. Usually, the elution volume is recommended to be 2–10 times the bed volume of the SPE cartridge (Care and Use Manual 2008). Smaller elution volumes result in more concentrated extract, as long as the extraction efficiency is close to 100%.

Some previous researchers have compared different eluents, to determine which provides the best performance when extracting EDCs from SPE cartridges (Liu et al. 2004, Arditsoglou and Voutsa 2008, Suri et al. 2012). Liu et al. (2004) found that, when extracting BPA and E2 from Oasis HLB cartridges, acetone, methanol, and ethyl acetate performed well, but methanol gave slightly higher recoveries. Suri et al. (2012) found that, for eluting target analytes from C₁₈ cartridges, methanol and toluene showed higher extraction efficiency than acetone, hexanol, or acetonitrile, with methanol showing the highest recoveries. Arditsoglou and Voutsa (2008) found that when Oasis HLB cartridges were used to extract BPA and E2 from artificial seawater, acetone provided higher recoveries than methanol or ethyl acetate. Overall, there does not appear to be a clear answer as to which eluent is “best” for extraction of EDCs from SPE cartridges; which eluent performs best is likely to depend on the target analyte(s), the type of SPE cartridge, and other experimental conditions. Methanol appears to be a suitable “default” eluent for extracting EDCs from C₁₈ or poly(divinylbenzene-co-N-vinylpyrrolidone) cartridges, based on its relatively low cost and relatively good performance in previous research.

Selection of derivatization agent

Chemical derivatization alters the chemical structure of the target analyte(s) to a form that provides better GC (Drozd 1981). Derivatization is often carried out to convert polar N–H, O–H or S–H groups into non-polar groups (Jeannot et al. 2002, Basheer and Lee 2004, Carpinteiro et al. 2004, Chang et al. 2005, Li et al. 2006, Yang et al. 2006, Zhang et al. 2006, Möder et al. 2007, Pan and Tsai 2008, Orata 2012). Many EDCs (including both BPA and E2) contain polar hydroxyl, phenolic, or carboxylic groups, so that derivatization is often necessary when EDCs are analyzed by GC/MS. The derivatized analytes typically are more volatile and produce sharper peaks than the underivatized forms, thereby improving the chromatography, and may also provide better thermal stability and/or more favorable fragmentation patterns during MS.

There are three common groups of derivatization methods: silylation, acylation, and alkylation (which includes esterification). For each type of derivatization, different derivatization agents are available. Many of these have been employed in the studies listed in Table 1. For instance, silylation agents include N,O-bis(trimethylsilyl)-acetamide (BSA), N,O-bis(trimethylsilyl)-trifluoroacetamide (BSTFA), N-methyl-N-(trimethylsilyl)-trifluoroacetamide (MSTFA), N-(tert-butyl-dimethylsilyl)-N-methyltrifluoroacetamide (MTBSTFA), and N-trimethylsilyl imidazole (TMSI). Many silylation agents contain a small amount of trimethylchlorosilane (TMCS; also known as trimethylsilyl chloride or chlorotrimethylsilane) as a catalyst. The most common alkylation agent for derivatization of BPA and E2 is pentafluorobenzyl bromide (PFBBr), but phenyltrimethylammonium hydroxide has also been used (Bolz et al. 2000, Körner et al. 2000, Stuart et al. 2005). Acylation is not common for E2 or BPA, but some studies have used heptafluorobutyric anhydride or trifluoroacetic anhydride to acylate E2.

Silylation is the most commonly used technique among the three candidate derivatization methods. The advantages of silylation are: (1) the ability to silylate a wide variety of compounds; (2) the large number of silylating reagents available; and (3) easy preparation. Silylation reagents can derivatize almost all functional groups (hydroxyl, carboxylic acid, amine, thiol, phosphate) that might cause problems during gas chromatographic separation. The trimethylsilyl (TMS) group is the typical silyl group to enhance chromatographic peak or detectability (Knapp 1979, Evershed 1993, Orata 2012).

An interesting question is whether one silylation agent is preferable to others for derivatization of EDCs. Gehrke (1968) compared BSTFA and BSA for silylation of amino acids and found that BSTFA reacts more quickly and more completely than BSA. Quintana et al. (2004) compared the reactivity of MTBSTFA, BSTFA, and MSTFA with the aromatic and aliphatic hydroxyl groups contained in six estrogenic chemicals. They found that MTBSTFA and BSTFA can react with hydroxyl groups only at certain positions within the chemical structure of the target analytes, but that MSTFA was able to silylate hydroxyl groups at any position, perhaps because its smaller molecular size reduced steric hindrance. However, for four of the six estrogens (including E2), BSTFA was observed to perform as well as MSTFA. As can be seen from Table 1, BSTFA and MSTFA have both been used successfully by many research teams for derivatization of BPA and E2, although BSTFA appears to be employed more frequently.

Selection of sample volume

In general, it is expected that an inverse relationship would exist between the concentration of target analytes in the aqueous sample and the required volume of sample that must be loaded on the SPE cartridge. If concentrations of target analytes are relatively high (μg/l range), then not much sample volume is required to load an acceptable mass of analyte onto the SPE cartridge. If concentrations of target analytes are low (ng/l range), then significantly more sample volume would be required to load the same analyte mass onto the cartridge. Many of the entries in Table 1 corroborate this general idea. For BPA or E2 concentrations in the μg/l range, typically no more than 1 l of sample volume is required for SPE (Jeannot et al. 2002, Ding and Chiang 2003, Latorre et al. 2003, Suzuki et al. 2004, Wang et al. 2005, Yang et al. 2006, Gatidou et al. 2007). However, for BPA or E2 concentrations in the low ng/l range, typically at least 2 l of sample volume are required (Desbrow et al. 1998, Routledge et al. 1998, Huang and Sedlak 2001).

Likewise, detection limits for BPA and E2 can be affected by sample volume. Suri et al. (2012) reported method detection limits for E2 of 30 ng/l when 200 ml of sample volume was used, 0.10 ng/l when 1000 ml of sample volume was used, and 0.03 ng/l when 3000 ml of sample volume was used. This is consistent with the pattern that higher sample volumes are required for lower concentrations of analytes.

This suggests that, when preparing calibration curves with standards of known concentrations, the volume of standard employed should vary according to the concentration of the standard. By varying the volume such, it may be possible to extend the range of linearity of the calibration curve and/or to reduce the method detection limits.

Remaining questions/room for improvement

To optimize the SPE method and GC/MS analysis for detection and quantification of EDCs, each step in the SPE process (Figure 1) can be optimized. For instance, key factors that can be considered include, but are not limited to, the following. (1) What is the best type of SPE cartridge for the analytes and aqueous matrix under consideration? (2) How should the SPE cartridge be conditioned? (3) What sample volume is appropriate for the expected concentrations of target analytes? (4) At what flow rate, or under what degree of vacuum, should the sample be loaded onto the SPE cartridge? (5) What solvents, and at what volumes, should be used to wash the cartridge after sample loading? (6) Which eluent or combination of eluents is best to remove the target analytes from the SPE cartridge? (7) What volume of eluent(s) should be used? (8) Should the eluent be evaporated to further concentrate the target analytes, and if so, what process should be used for solvent evaporation? (9) What derivatizing agent (or combination of agents) is best? (10) What conditions (temperature, time, etc.) should be used for the derivatization process? (11) How do the answers to the preceding questions depend upon the chemistry of the aqueous matrix and/or the target analytes?

Some of these questions can be answered in part by examining the existing literature, as in Table 1 and the preceding discussion in this paper. To fully answer every one of these questions would likely take years of effort from multiple research groups working in parallel. In the remainder of this paper, we contribute to this effort by considering the following with respect to analysis of BPA and E2. (1) Which type of SPE cartridge provides better analyte recovery, C₁₈ or poly(divinylbenzene-co-N-vinylpyrrolidone)? (2) Which derivatization agent is preferable, MSTFA or BSTFA? (3) To what extent can the range of linearity of GC/MS calibration curves be extended by varying the sample volume applied during SPE?

Chemicals

Methanol (HPLC grade), BPA (purity grade >99%), E2 (purity grade >99%), BSTFA with 1% TMCS, and MSTFA with 1% TMCS were all purchased from Sigma-Aldrich, St. Louis, MO, USA.

Aqueous samples

Primary stock standard solutions (1000 mg/l) of both BPA and E2 were prepared in methanol by dissolving 0.100 g of analyte into 100 ml methanol. Stock solutions were stored at 4°C in a refrigerator. Then, aqueous samples were prepared daily by dilution of the stock solutions with deionized water. The concentrations of aqueous samples ranged from 1 ng/l to 100 μg/l for SPE. The aqueous samples were prepared from the primary stock solutions by diluting with deionized water, using sequential dilutions when necessary to obtain low concentrations. Methanol content in the aqueous samples was 0.1% or lower (by volume, before mixing) in all aqueous samples, and was therefore considered negligible.

SPE

Here we describe the SPE method used to prepare a sample for analysis by GC/MS. Oasis HLB glass cartridges (5 ml, 200 mg sorbent) or Fisher C₁₈ cartridges were placed on a vacuum manifold (SPE 24-port Vacuum manifold, purchased from Fisher Scientific, Pittsburgh, PA, USA). The cartridges were conditioned with 40 ml of deionized water and 25 ml of methanol, both of which were drawn through the cartridges under low vacuum to remove residual bonding agents. Then, a known volume of aqueous sample was loaded onto the cartridge and flowed through under slight vacuum (flow rate=60 ml/min). We tested different volumes of samples ranging from 10 ml to 4 l, and different EDC concentrations ranging from 1 ng/l to 100 μg/l. During the sample loading step, the target compounds were extracted from the aqueous samples onto the SPE cartridges. After loading, the cartridges were washed with 20 ml of deionized water, and then dried for 5 min under vacuum in order to remove excess water remaining on the cartridge. Then, the adsorbed analytes were eluted to 10 ml vials from the cartridges with 5 ml methanol at a flow rate of 5 ml/min.

Derivatization

Due to the presence of polar functional groups in BPA and E2, which can give rise to poor chromatographic peaks, derivatization was necessary. We based our derivatization method on that of Zhang et al. (2006). The methanol eluent collected from SPE was evaporated in a Rotavapor R-210 rotary evaporator (Buchi, Flawil, Switzerland). The dry residues were derivatized either by BSTFA with 1% TMCS, or by MSTFA with 1% TMCS. For either agent, 100 μl of derivatization reagent was added to each reaction vial. Then, the vials were closed and placed in an oven at 65°C for 25 min. Because the reaction time and temperature can affect the derivatization process, these parameters were investigated to optimize the derivatization conditions. Derivatization was carried out in triplicate on test samples (100 μg/l) at reaction times of 10, 20, 25, 30, 35, and 40 min. When reaction times of 25 min or longer were applied, there were no underivatized compounds in the chromatograms, so 25 min was selected as the reaction time for the remainder of the study. Similarly, derivatization was tested in triplicate at ambient temperature and at temperatures of 30, 40, 50, 60, 65, 70, and 75°C. The extent of derivatization increased with temperature up to a temperature of 65°C, so 65°C was selected as the reaction temperature for the remainder of the study.

Once the derivatization was completed, 1 μl of the reaction mixture was injected into the GC/MS system within 30 min, to avoid reaction inversion.

GC/MS instrumentation and operating conditions

Analyses were carried out on a Varian CP-3800 gas chromatograph directly connected to a Saturn 2000 ion-trap mass spectrometer (Varian, Palo Alto, CA, USA). An HP-5MS capillary column (30 m×0.25 mm inner diameter, 0.25 μm film, 5% phenyl-dimethylsiloxane phase, Agilent, Santa Clara, CA, USA) was used for chromatography. Helium (99.9995% purity) was used as carrier gas at a constant flow rate of 1.0 ml/min. The injection port temperature was 280°C and injections were performed using a splitless mode. The GC oven temperature program was as follows: hold for 1 min at 80°C, increase at 15°C/min to 240°C, hold for 1 min, increase at 10°C/min to 280°C, and hold for 5 min. Data acquisition was performed in full scan mode measuring m/z from 69 to 614. The transfer line temperature of the GC/MS was set at 170°C, and the manifold temperature was set at 160°C. The electron emission current of GC/MS was 10 μA (70 eV), multiplier voltage was 1500 V, and automatic gain control target was 20,000.

BPA and E2 were quantified by the area of the peak corresponding to a particular fragment on the MS. We refer to these fragments as the diagnostic ions for each compound. The m/z ratios for the diagnostic quantitative ions are 357 for BPA and 416 for E2. These m/z ratios correspond to major peaks in the mass spectra of the derivatized (silylated) compounds.

Comparison of cartridge types

When comparing C₁₈ to poly(divinylbenzene-co-N-vinylpyrrolidone) cartridges, no major differences were observed. With poly(divinylbenzene-co-N-vinylpyrrolidone) cartridges, the average recoveries observed were 94.4% for BPA and 97.6% for E2. With C₁₈ cartridges, the average recoveries were 93.2% for BPA and 96.8% for E2. The differences between the cartridge types were not statistically significant. For subsequent experiments, we chose poly(divinylbenzene-co-N-vinylpyrrolidone) to extract the EDCs from the water samples because in recent work by other research groups, it appears to be more common than C₁₈ (see Table 1).

Calibration curves

The calibration curves of EDCs that were extracted by SPE and derivatized are presented in Figures 2 and 3 as measured peak area vs. injected EDC mass. The injected EDC mass is calculated as the volume, V, of sample loaded onto the SPE cartridge (ranging from 20 ml to 4000 ml) times the concentration, C, of target EDC in the sample (ranging from 0.001 μg/l to 100 μg/l). This calculation assumes that 100% of the EDC mass loaded onto the SPE cartridge was recovered and injected into the GC; as noted above, observed recoveries were in the range of 93%–97%, so the assumption of 100% recovery does not introduce much error. Tables 2–5 show the relation between loaded sample volume, concentration of EDC sample, and measured area of derivatized EDC fragment.

Figure 2

Calibration curve of bisphenol-A analyzed by solid-phase extraction (SPE) followed by gas chromatography and mass spectrometry (GC/MS). Top panel: derivatized with N,O-bis(trimethylsilyl)-trifluoroacetamide (BSTFA); bottom panel: derivatized with N-methyl-N-(trimethylsilyl)-trifluoroacetamide (MSTFA).

Figure 3

Calibration curve of 17β-estradiol analyzed by solid-phase extraction (SPE) followed by gas chromatography and mass spectrometry (GC/MS). Top panel: derivatized with N,O-bis(trimethylsilyl)-trifluoroacetamide (BSTFA); bottom panel: derivatized with N-methyl-N-(trimethylsilyl)-trifluoroacetamide (MSTFA).

It can be seen from Figures 2 and 3 that the calibration curves, presented as measured peak area vs. injected EDC mass, are linear over multiple orders of magnitude. This wide range of linearity is useful, because it means that the analytical method is likely to be applicable even to samples where the approximate concentration range is unknown. Also, it means that a single analytical method is likely to be applicable to studies in which samples have widely varying concentrations, e.g., if the concentration of a target analyte is attenuated by 99% or even 99.9%. In cases where the research team knows that the sample concentrations vary over a smaller range, a calibration curve can be developed for that particular concentration range, which may result in an R² value higher than those observed in Figures 2 and 3 (R²=0.986 or 0.993 for BPA, R²=0.994 or 0.999 for E2).

For BPA (Figure 2), the calibration curve was generated from samples that meet three criteria: the concentration C_BPA was between 0.010 μg/l and 100 μg/l; the sample volume V was between 20 ml and 4000 ml; and the BPA mass loaded (M=V*C_BPA) was between 30 ng and 5000 ng. The third criterion implies, for instance, that for samples where we used a volume V=100 ml, the calibration curve includes all results for which 0.30 μg/l≤C_BPA≤50 μg/l, but not for samples outside this concentration range. As can be seen from Figure 2, the measured peak area is linear with respect to the injected BPA mass for samples meeting the three necessary criteria.

For E2 (Figure 3), similar behavior was observed, but the range of linearity is even greater for E2 than it is for BPA. For E2, the calibration curves were generated from samples which met the following three criteria: the concentration C_E2 was 0.010–100 μg/l; the sample volume V was 20–4000 ml; and the E2 mass loaded (M=V*C_E2) was 20–20,000 ng. The third criterion implies, for instance, that a sample volume of V=500 ml could be used to quantify concentrations in the range 0.040 μg/l≤C_E2≤40 μg/l. As can be seen from Figure 3, the measured peak areas are linear with respect to the E2 mass injected for samples meeting these criteria.

For both BPA and E2, we did test several samples that did not meet one of the requisite criteria (e.g., samples of concentration C<10 ng/l, or for which V*C was not in the specified range). These samples generally did not follow the same linear behavior, and are not included in Figures 2 and 3. Hence, there is some limitation on the range of linearity for the SPE method; if the concentration is too low or too high, the measured peak area is not likely to fall on the calibration curves provided. However, this limitation is not severe; simply by choosing the sample volume appropriately, the SPE method may be applied to samples of BPA or E2 in the concentration range 10 ng/l to 100 μg/l, i.e., four orders of magnitude of concentration. We found that C=10 ng/l is a practical lower limit of quantification for the SPE method for both BPA and E2.

Derivatization agent

MSTFA (with 1% TMCS) and BSTFA (with 1% TMCS) were selected and compared as derivatization agents in this work because of their common usage by previous research groups (Table 1). We observed that silylating the target analytes sharpened the chromatographic peaks. Retention times increased by only about 10 s for the silylated compounds compared to the non-derivatized compounds. Results from MS verified that the target analytes were silylated (m/z=357 for BPA and 416 for E2).

By comparing the calibration curves obtained in Figures 2 and 3, we can assess the relative performance of MSTFA and BSTFA as derivatization agents. Examination of Figures 2 and 3 shows that the calibration curves are nearly identical for the two derivatization agents. The range of linearity was the same for both agents. For analysis of BPA (Figure 2), the slopes of the calibration curves differ by only about 7% for the two derivatization agents. For analysis of E2 (Figure 3), the slopes differ by only about 1%. This result is consistent with those of some previous researchers, who observed similar performance with MSTFA or BSTFA (Quintana et al. 2004, Shareef et al. 2006, Zhou et al. 2007, Sebők et al. 2008).

Calibration curves

Many previous researchers, as summarized in Table 1, used a fixed aqueous sample volume for all analyses, regardless of the expected analyte concentration in the sample. Here, we have developed calibration curves for our target analytes based on a range of sample volumes from 20 ml to 4 l. The advantage to this is that it can extend the concentration range over which a linear calibration curve can be obtained. As seen from Tables 2–5, the lowest sample volume (20 ml) is useful only for the highest concentrations of analytes, and the highest sample volume (4 l) is useful only for the lowest concentrations of analytes. Even if we selected the “best” single sample volume, the calibration range would be limited to two orders of magnitude for BPA (e.g., 0.01–1 μg/l with 3000 or 4000 ml) and three orders of magnitude for E2 (e.g., 0.01–10 μg/l with 2000 ml). By using multiple sample volumes, we extend the range of concentrations to four orders of magnitude, 0.01–100 μg/l. This appears to be a wider range than was achieved by most of the researchers listed in Table 1.

In some applications, it might not be necessary to develop a calibration curve that covers four orders of magnitude in analyte concentration, because the analyte concentrations might be known to vary over a relatively small range. For instance, if a group were monitoring the concentrations of BPA and E2 in the effluent of a wastewater plant over time, we might expect the contaminant concentrations to be relatively steady, and the concentrations might fluctuate over time only by a factor of 10 or so. However, other applications would benefit from the ability to monitor analyte concentrations over multiple orders of magnitude. For instance, if a research group were looking at removal or biodegradation of EDCs to protect human health, it might be desirable to quantify contaminant removals up to 99.9% or even 99.99% efficiency, which would require the ability to measure analyte concentrations over multiple orders of magnitude. In such instances, we recommend using a range of sample volumes, where the sample volume selected varies inversely with the expected concentration in the sample.

Selection of derivatization agent (MSTFA or BSTFA)

For BPA and E2, we observed no difference in the performance of MSTFA (with 1% TMCS) and BSTFA (with 1% TMCS) as derivatizing agents. This might be partly because we selected the derivatization temperature and time (65°C for 25 min) to maximize the degree of silylation, i.e., complete derivatization was achieved with both agents tested. We conclude that for analysis of BPA and E2, either derivatizing agent may be used with equal confidence. However, this conclusion might not be valid for other target analytes besides BPA and E2. For instance, Quintana et al. (2004) observed that MSTFA could fully silylate 17α-ethinylestradiol (EE2), but that BSTFA could silylate only the aromatic hydroxyl groups, not the aliphatic. Zhou et al. (2007) similarly saw differences in the degree of silylation of estrone and EE2 depending on which derivatization agent was selected, and on whether small amounts of TMCS or TMSI were added as silylation promoters. Therefore, we conclude that silylation is likely a viable method of derivatization for many EDCs (including steroid hormones, bisphenols, and alkylphenols), but that research groups should evaluate candidate derivatization agents for their effectiveness on the particular analyte(s) of interest.

Applicability to other EDCs

In this paper we have considered two particular EDCs, BPA and E2. In many cases, the techniques and conditions employed here for BPA and E2 may also apply to other chemically similar EDCs, such as alkylphenols or steroid hormones; indeed, many of the studies listed in Table 1 included other EDCs besides BPA and E2. However, researchers should not necessarily assume that all chemically similar EDCs will behave the same during SPE. It has been observed, for instance, that recovery of nonylphenol by SPE sometimes differs from the recovery of BPA (Carabias-Martínez et al. 2004, Beck et al. 2005). Researchers should exercise care in ensuring that methods employed are appropriate for the particular analyte(s) of interest.