Removal of endocrine-disrupting chemical mixtures in water using chlorination and photolysis

Concerns about water quality emerged at the end of the 19th century when the physician John Snow evidenced cholera as a waterborne disease. Thereafter, the major objective of water treatment has been to minimize the risks caused by pathogens in water (Libânio 2010; Jain et al. 2014). Until now, drinking water treatment plants are still designed to remove microorganisms and improve the organoleptic properties (e.g., color and turbidity) of water, whereas chemical substances that may pose risks to human health usually are not the focus of drinking water treatment (Jain et al. 2014; Pai et al. 2020).

Within this scenario, endocrine-disrupting chemicals (EDCs) are commonly not included in drinking water legislations (Aquino et al. 2021). However, these substances are capable to affect the endocrine system of humans and animals (Pereira et al. 2013). The risks associated with EDCs exposure are still not fully understood (Wee & Aris 2019). Despite this, several studies reported associations between exposure to EDCs and alterations in the reproductive, endocrine, and digestive systems (Colborn et al. 1993; Safe 2000; AWWA 2007; Gallo et al. 2016; Sheikh et al. 2016, 2017; Sweeney et al. 2016; USEPA 2016; Vilela et al. 2018).

Currently, diverse chemical substances are classified as EDCs, such as some natural and synthetic hormones, pesticides, surfactants, plasticizers, pharmaceuticals, and personal care products (Bila & Dezzotti 2007; Aquino et al. 2013). Due to their estrogenic activity potency and their growing use, some of these substances have been gaining special attention from the scientific community and public health agencies. Among these substances, estrone (E1), 17β-estradiol (E2), ethinylestradiol (EE2), and nonylphenol (NP) can be cited. E1 and E2 are natural hormones produced and excreted by humans and animals, which can also be used to prevent and control disorders in the female reproductive system (NIH 2022). EE2 is a synthetic estrogen that is widely used in contraceptives and hormone replacement therapies (NIH 2022). In addition, NP is known as the major degradation product of NP ethoxylates, which are nonionic surfactants (Kim 2014).

These EDCs can be released into water sources through raw and treated wastewaters discharge or even through runoff from animal-feeding operations (Ternes et al. 1999; Aquino et al. 2013; Gomes et al. 2022). As a consequence of their massive use, E1, E2, EE2, and NP have been frequently found in water sources and drinking waters (Ding & Tzing 1998; Snyder et al. 1999; Kolpin et al. 2002; Johnson & Jürgens 2003; Moreira et al. 2009, 2011; Resende et al. 2017; Aquino et al. 2021).

Given the increasing concern about EDCs, many analytical methods to detect and quantify these compounds in environmental matrices have been developed in the last decades (Routledge & Sumpter 1996; Alda & Barceló 2001; Magi et al. 2010; Gunatilake et al. 2014; Lopardo et al. 2019; Martin-Yken 2020). In this context, chromatographic methods and in vitro and in vivo assays have been widely used to analyze EDCs in waters (Robitaille et al. 2022). In vitro assays can evaluate the estrogenic activity of a sample based on mechanisms of action of EDCs, such as the case of cell proliferation assays (Shanle & Xu 2011; Serra et al. 2020). A noticeable example is the in vitro yeast estrogen screen (YES), developed by Routledge & Sumpter (1996). The YES assay is capable to quantify a wide range of estrogenic activity concentrations resulting from the synergic effect of the different EDCs present in a sample, as occurs in the environment (Routledge & Sumpter 1996). Its additional advantages include low cost, operational simplicity, and relatively low detection limits (Martin-Yken 2020).

One factor that contributes to the ubiquitous presence of EDCs in the environment is the inefficiency of some treatment processes allied to their continued environmental release. For instance, water clarification is ineffective for EDCs removal (Chen et al. 2007; Azevedo et al. 2020; Pai et al. 2020). However, depending on the process and the operational conditions used, these compounds can be removed during water disinfection (WHO 2011). In this context, chlorination and photolysis have been increasingly studied for this purpose. Previous researches evidence that both chlorination and photolysis may lead to the removal of EDCs (Alum et al. 2004; Coleman et al. 2004; Westerhoff et al. 2005; Fonseca et al. 2011; Pereira et al. 2011, 2013; Schenck et al. 2012; Sarkar et al. 2014; Li et al. 2017; Huang et al. 2022). However, there is a lack of studies that evaluate the removal of the estrogenic activity caused by mixtures of EDCs during disinfection under technically and economically viable conditions.

Thus, the objective of this work was to evaluate the efficiency of chlorination and photolysis for the estrogenic activity removal resulting from mixtures of E1, E2, EE2, and NP in water.

E1 (≥99%), E2 (≥98%), EE2 (≥98%), NP (analytical standard), and sodium hypochlorite solution (10–15%) were purchased from Sigma-Aldrich. Acetone, hexane, hydrochloric acid, methanol, nitric acid, absolute ethanol, hydrogen peroxide, and the reagents used in the YES assay were purchased from Synth.

The water used in the experiments was sampled in a preserved area located in Torreões (Minas Gerais, Brazil). The sample was analyzed for pH, electrical conductivity (EC), apparent color, turbidity, total hardness, total carbon (TC), total organic carbon (TOC), and free and total chlorine. These analyses were conducted in triplicate (except for total hardness, TC, and TOC) and were performed according to the Standard Methods for the Examination of Water and Wastewater (APHA et al. 2005). Additionally, the sample was also tested to certify the absence of estrogenic activity, using the in vitro YES assay (Routledge & Sumpter 1996) (as described in item 2.2.2). The basic physicochemical properties of the water used in the experiments were as follows: apparent color = 12.8 ± 4.0 uH; turbidity = 0.1 ± 0.0 NTU; pH = 7.9 ± 0.5; EC = 18.6 ± 2.8 μS cm⁻¹; total hardness = 17 mgCaCO₃ L⁻¹; TC = 3.0 mg L⁻¹; and TOC = 0.18 mg L⁻¹.

Before the YES assays, samples were extracted with BondElut C18 (500 mg, 6 mL) cartridges (Agilent). First, 600 mL of water samples were acidified to pH 3.0 with hydrochloric acid. The acidified samples were passed through the extraction cartridges, which were previously conditioned with 6.0 mL of hexane, 2.0 mL of acetone, 6.0 mL of methanol, and 10 mL of acidified deionized water (pH = 3.0). Subsequently, analytes were eluted with 4.0 mL of acetone and the eluates were evaporated under a gentle nitrogen flow. The SPE procedure followed the method described by Pereira (2011). Before the YES assays, analytes were resuspended in 1.0 mL absolute ethanol (Bila et al. 2007).

The in vitro YES assay was used to quantify the estrogenic activity in water samples. In summary, the in vitro YES assay uses a recombinant yeast strain (Saccharomyces cerevisiae) developed by the Genetics Department, Glaxo. The yeast strain contains the human estrogen receptor (hER) and expression plasmids carrying the reporter gene lac-Z, which is capable to encode the enzyme β-galactosidase in the presence of EDCs. Subsequently, the β-galactosidase degrades the substrate chlorophenol red-b-D-galactopyranoside (CPRG) added in the medium into chlorophenol red, promoting a color change from yellow to red (Routledge & Sumpter 1996). Therefore, the estrogenic activity (in E2 equivalents, E2-eq) can be estimated by comparing the absorbance of the medium to the standard curve generated from the absorbances of the assay using known concentrations of E2 (26.6 ng L⁻¹ to 54.48 μg L⁻¹).

The efficiencies of chlorination and photolysis in estrogenicity removal were tested using freshwater spiked with four different mixtures of EDCs: E1 + E2, E1 + E2 + NP, E1 + E2 + EE2, and E1 + E2 + EE2 + NP. In all conditions tested, initial concentrations of E1 and E2 were 100 ng L⁻¹, whereas concentrations of EE2 and NP were 50 and 1,000 ng L⁻¹, respectively. These concentrations were chosen to represent real concentrations found in surface waters, based on previous values reported in the literature (Ding & Tzing 1998; Snyder et al. 1999; Kolpin et al. 2002; Johnson & Jürgens 2003; Moreira et al. 2009, 2011; Resende et al. 2017; Aquino et al. 2021).

Experiments were carried out in the Environmental Quality Laboratory (LAQUA) of the Federal University of Juiz de Fora (UFJF). Each condition described below (i.e., different EDC mixtures and UV doses) was tested in triplicate.

Batch experiments were carried out in a beaker of 1 L containing 700 mL of EDCs solution. A chlorine dose of 2.75 mg L⁻¹, which is comparable to those usually used in drinking water treatment plants (0.2–5 mg L⁻¹ (Brazil, 2021; WHO 2017)), was assessed in this work. The chlorine dose was obtained from a sodium hypochlorite stock solution with 1,000 mg L⁻¹ Cl₂, which was preserved in an amber bottle at 4 °C. The contact time of chlorination was fixed at 30 min. Considering that the chlorine dose can vary according to the working temperature, the chlorine dose applied in the experiments was calibrated before the experiments. The removal of residual chlorine after the tests was carried out by the addition of 0.1-mL sodium metabisulphite 3% (m/V) for each 100-mL sample.

Photolysis experiments were performed in an 800 mL bench-scale reactor equipped with a low-pressure mercury lamp of 80 W placed in a quartz tube and submerged in an axial position in the samples. During the experiments, the system was maintained by continuous stirring with a magnetic stirrer. To maintain steady state conditions during the tests, the lamp was kept on for 15 min before the experiments (Nicole et al. 1990; Huang et al. 2022).

The UV doses used in this work (1.5, 16, 40, and 186 mJ · cm⁻²) were chosen according to values previously established for disinfection processes (Brazil, 2021; LeChevallier & Au, 2004; NSF/ANSI 2021). The dose of 1.5 mJ · cm⁻² is determined by the Brazilian drinking water standard as the minimum UV dose required for drinking water disinfection (Brazil, 2021). The UV dose of 16 mJ · cm⁻² is established by the American National Standards Institute (ANSI) and it is recommended for supplemental treatment of waters that are already of good quality (Class B systems) (NSF/ANSI 2021). The dose of 40 mJ · cm⁻² is recommended by the same institute (ANSI) for the removal of cryptosporidium, giardia, bacteria, and viruses from contaminated water (NSF/ANSI 2021). Finally, the dose of 186 mJ · cm⁻² is recommended for the removal of 99.99% of bacteria and viruses from water according to a guide for water treatment and pathogen control by the World Health Organization (WHO) and The International Water Association (IWA) (LeChevallier & Au, 2004).

Analysis of variance (ANOVA) and the Tukey test were used to compare estrogenicity removal efficiencies of chlorination and photolysis under different conditions at a 95% confidence level. For the case of chlorination, these analyses were used to compare estrogenic activity removal efficiencies from aqueous solutions with different mixtures of EDCs (E1 + E2, E1 + E2 + NP, E1 + E2 + EE2, and E1 + E2 + EE2 + NP). In the case of photolysis, ANOVA and Tukey test were carried out to assess statistically significant differences between the removal of estrogenic activity from both solutions with different mixtures of EDCs and during photolysis at different UV doses. These analyses were performed using SISVAR software.

The physicochemical properties of the water before and after chlorination were summarized in Table S1 (Supplementary Material). As can be seen, in terms of the parameters analyzed, the water used in the experiments had good quality, and its general physicochemical properties were comparable to waters sampled after post-filtration in drinking water treatment plants (Brazil, 2021).

In all conditions tested, pH, apparent color, and turbidity remained practically unaltered after chlorination (Table S1, Supplementary Material). In contrast, EC at the end of the tests was about two-fold higher than in the freshwater, probably due to the addition of chlorine ions in the samples.

After chlorination, the concentration of free chlorine in the E1 + E2 solution was higher than in the other tests. This fact is expected, since a minor concentration of EDCs (100 ng · L⁻¹ E1 and 100 ng · L⁻¹ E2) were available to react with chlorine in this test. The solution prepared with a mixture of E1 + E2 + EE2 had the second highest concentration of free chlorine after chlorination, due to the less concentration of EDCs in the test (100 ng · L⁻¹ E1, 100 ng · L⁻¹ E2, and 50 ng · L⁻¹ EE2) in comparison with E1 + E2 + NP (100 ng · L⁻¹ E1, 100 ng · L⁻¹ E2, and 50 ng · L⁻¹ EE2) and E1 + E2 + EE2 + NP (100 ng · L⁻¹ E1, 100 ng · L⁻¹ E2, 50 ng · L⁻¹ EE2, and 1,000 ng · L⁻¹ NP). Accordingly, the concentration of free chlorine reached 1.2 mg · L⁻¹ in the E1 + E2 + EE2 + NP test, as a result of the high reactivity of this solution in comparison with the other conditions tested. Furthermore, concentrations of total chlorine after chlorination varied from 2.2 to 2.5 mg · L⁻¹, whereas the combined chlorine varied from 0.1 to 0.9 mg · L⁻¹. This indicates a relatively low chlorine demand in the conditions studied (Smith et al. 2021).

As can be seen, the highest removal efficiency (99.7%) occurred in the experiment with E1 + E2, which resulted in a residual estrogenic activity of 0.4 ng E2-eq. L⁻¹. Noticeably, the removal efficiency decreased to 98.8% when adding NP to this mixture (E1 + E2 + NP). As the relative estrogenic activity potency of NP is about 2.57 × 10⁵ times lower than E2 (Beck et al. 2006), the addition of NP did not increase substantially the initial estrogenic activity of the solution. Nevertheless, the presence of NP available to react with chlorine resulted in a reduction in estrogenic activity removal efficiency and, consequently, in a higher effluent concentration (1.4 ng E2-eq. L⁻¹), which can indicate the formation of chlorinated by-products that can cause estrogenic activity in the final effluent. A comparable outcome occurred in the experiment with E1 + E2 + EE2, which had a removal efficiency of 98.7%. As EE2 has a relative estrogenic activity potency of about 0.75–2.46 (Beck et al. 2006; Combalbert & Hernandez-Raquet, 2010), the initial estrogenic activity increased due to the addition of EE2, leading to an estrogenic activity of 2.3 ng E2-eq. L⁻¹ after chlorination. Surprisingly, the addition of NP and EE2 (E1 + E2 + EE2 + NP) did not affect the average removal efficiency of chlorination which was maintained equal to 99.5%, with a residual estrogenic activity concentration of 1.0 ± 0.4 ng E2-eq. L⁻¹. This is interesting, as EDCs tend to occur simultaneously in the environment. Therefore, these results evidence that chlorination can be an effective process to remove the estrogenic activity from complex matrices such as freshwater.

The remaining estrogenic activity after chlorination obtained in this study agreed with values reported in the literature. Fan et al. (2013) assessed the presence of EDCs in finished waters from 62 drinking water treatment plants in China. In their work, E1 and E2 were detected in 53 plants, with maximum concentrations of 0.1 and 1.7 ng · L⁻¹, respectively, whereas NP was found in 55 plants with concentrations of up to 558 ng · L⁻¹. A similar study was conducted in Brazil by Gerolin (2008), in which E1, E2, EE2, and NP were identified at concentrations of 0.1, 1.48, 472, and 87 ng · L⁻¹, respectively. In Spain, Rodriguez-Mozaz et al. (2004) detected EE2 at a concentration of 2.5 ng · L⁻¹, which is lower than the value reported by Gerolin (2008) and more concordant with the concentrations obtained in the present study.

Despite the satisfactory results, the risks associated with human exposure to estrogens are not fully understood. Therefore, it is not possible to affirm that the remaining estrogenic activity determined in this work is not like to affect human health. Different values for the acceptable daily intake (ADI) of EDCs and their possible guideline values in drinking water have been proposed. ADIs from 13 to 50 ng · kg⁻¹ body weight d⁻¹ were established for E1 by previous toxicological studies, which evidenced estrogenic responses and effects on the endocrine system and liver of humans with no-observed-adverse-effect levels (NOAEL) varying from 0.004 to 0.005 ng · kg⁻¹ body weight d⁻¹ (AWWA 2007; USEPA 2016). For the case of E2, an ADI of 50 ng · kg⁻¹ body weight d⁻¹ was established based on the promotion of serum levels of follicle-stimulating hormone, angiotensinogen, sex hormone binding globulin, and corticosteroid-binding globulin in women (AWWA 2007; EPHC/NRMMC/AHMC, 2008; FAO/WHO, 2000). Additionally, dose–response studies evidenced the development of breast, uterus, and mesentery cancers in rats after exposure to E2, with a slope factor of 1/39,000 (USEPA 2016). More restrictive ADIs, from 0.1 to 5 ng · kg⁻¹ body weight d⁻¹, were established for EE2 based on its effects on the endocrine system and liver of humans (AWWA 2007; USEPA 2016). In contrast, ADIs from 15,000 to 50,000 ng · kg⁻¹ body weight d⁻¹ were estimated for NP, based on effects on the reproductive system and decreases in body and organ weight of rats (EU, 2002; AWWA 2007; EMEA, 2015; USEPA 2016). Considering these ADIs, possible guideline values for maximum acceptable concentrations of E1, E2, EE2, and NP in drinking waters may vary from 78 to 300 ng · L⁻¹ (E1), 8.0 to 300 ng · L⁻¹ (E2), 3.0 to 150 ng · L⁻¹ (EE2), and 90,000 to 300,000 ng · L⁻¹ (NP) (Aquino et al. 2021).

Previous studies evidenced that the presence of chlorine can promote the removal of EDCs from water by aromatic halogenation and removal efficiencies of up to 98% had already been reported (Hu et al. 2003; Alum et al. 2004; Westerhoff et al. 2005; Choi et al. 2006; Pereira et al. 2011; Schenck et al. 2012; Li et al. 2017; Shao et al. 2018). However, the initial concentrations of EDCs used in these studies were at least 10 times higher than those used in the present work. This fact suggests that chlorination can be effective to remove EDCs even at low concentrations which are more likely to be found in the environment. The Europe Union (EU) established a watch list of substances and compounds of concern for water intended for human consumption in the revised Drinking Water Directive (EU Directive 2020/2184) (EU, 2020). In this watch list E2 and NP have guidance values of 1 and 300 ng · L⁻¹, respectively (EU, 2020). Values considering the worst case, for E2, that was achieved only for the E1 + E2 mixture.

It is worth mentioning that chlorination of EDCs may lead to the formation of different by-products and the toxicity of many of these products is still not well described in the literature (Bila et al. 2007; Pereira et al. 2011; He et al. 2016; Li et al. 2017; Leusch et al. 2019). According to Leusch et al. (2019), by-products formed during chlorination are unlikely to increase estrogenic activity. However, their findings suggest that these by-products may be more reactive and have increased non-specific toxicity (Leusch et al. 2019). It is also important to highlight that the potential by-products from the reaction between chlorine and EDCs may include trihalomethanes and halogenic acetic acids (He et al. 2016; Li et al. 2017).

In this work, the UV light intensity was estimated as 0.297 mW · cm⁻², which is near to the lower value used by Zhang et al. (2010) (0.246 mW·cm⁻²) and lower than other light intensities reported in the literature (1.5–500 mW · cm⁻²) (Caupos et al. 2011; Chowdhury et al. 2011; Kovacic et al. 2018; Li et al. 2013; Liu & Liu 2004; Liu et al. 2017; Martínez-Zapata et al. 2013; Mboula et al. 2015; Zhang et al. 2010). After determining the light intensity, the times required to provide the tested UV doses were estimated as follows: 5 s for the dose of 1.5 mJ · cm⁻², 54 s for the dose of 16 mJ · cm⁻², 75 s for 40 mJ · cm⁻², and 626 s for 186 mJ · cm⁻². As expected, photolysis did not lead to major changes in the basic physicochemical properties of water in all combinations and doses tested (Table S2).

Previous researches evidence that pH may affect photolysis efficiency (Neamţu & Frimmel 2006). Some studies suggest that photolysis in alkaline media may result in slightly increased EDCs removal efficiencies (Liu & Liu 2004; Neamţu & Frimmel 2006), but this is not a consensus (Chowdhury et al., 2011). Thus, as a preliminary analysis on the effectiveness of photolysis in the proposed scenario, no adjustments in pH before photolysis were performed in this work. Given this fact and considering possible costs related to the adjustment of pH prior to photolysis in a real scale operation, further analyses on the effects of changes in pH on photolysis at the proposed optimal conditions and the potential technical and economic aspects are recommended. Water temperature was higher for the dose of 186 mJ cm⁻² than for the other doses, as a response of the heat generated during UV light use. In fact, the use of cooling systems can be required for large exposure times.

The lowest dose tested, which is recommended for disinfection by the Brazilian drinking water standard (Brazil, 2021) was capable to remove 6–23% of estrogenic activity, with effluent estrogenic activity of up to 138 ng E2-eq. · L⁻¹. The UV doses established by the American National Standards Institute (NSF/ANSI 2021) (16–40 mJ · cm⁻²) led to removals from 57.8 to 94.4%. As aforementioned, the dose of 40 mJ · cm⁻² is recommended to remove viruses and pathogenic bacteria from water. Regarding estrogenic activity, concentrations of up to 24 ng E2-eq. · L⁻¹ remained after photolysis. This value is higher than possible health-based guideline values for E2 and EE2 (8 and 3 ng · L⁻¹) (Aquino et al. 2021), which suggests that the refereed UV dose (40 mJ · cm⁻²) may not be sufficient to minimize potential risks of the exposure to EDCs through drinking water intake. On the other hand, the dose recommended by ANSI was sufficient to remove the estrogenic activity from the mixtures with a remaining concentration of less than the method detection limit (approximately 1 ng · L⁻¹). For this, an exposure time of 10 min and 26 s was required. However, it is worth to mention that this time can be reduced by providing a high UV light intensity.

High removal efficiencies of the studied EDCs during photolysis had already been reported in the literature. However, EDC concentrations tested in earlier studies were considerably higher (10–1,950 μg · L⁻¹ for estrogens and up to 10 mg · L⁻¹ for NP) than those tested in the present work (in the 50–100 ng · L⁻¹ for estrogens and 1,000 ng · L⁻¹ for NP) (Zhang et al. 2010; Martínez-Zapata et al. 2013; Carlson et al. 2015 Huang et al. 2022). One of the major reasons for using relatively high initial concentrations during experiments to assess efficiencies of treatment processes can be associated with the dependency on analytical methods capable to detect EDCs at low concentrations. Within this context, in the case of EDCs with relatively high estrogen activity potency (e.g., E2 and EE2) or mixtures of EDCs, methods to detect/quantify estrogenic activity (e.g., YES assay) can be useful tools for assessing processes efficiency at initial concentrations comparable to those which naturally occur in the environment.

Carlson et al. (2015) reported pseudo first-order rate constants (k) in the order of 10⁻⁴ min⁻¹ for the removal of E1, E2, EE2, and NP (initial concentrations: 20–1,500 μg · L⁻¹) from deionized water during photolysis. A rate constant in the order of 10⁻³ was observed for the removal NP (10 mg · L⁻¹) from ultrapure water in the study of Martínez-Zapata et al. (2013), whereas Huang et al. (2022) observed rate constants from 10⁻³ to 10⁻² for E1, E2, E3, and EE2 (10 μg · L⁻¹). Similarly, Zhang et al. (2010) reported that the removal rate constant of EE2 (610–1,950 μg · L⁻¹) photolysis increased from 10⁻³ to 10⁻¹ min⁻¹ with increasing the UV light intensity from 0.2456 to 1.5 mW cm⁻².

This suggest therefore that EDCs degradation rate during photolysis may be strongly influenced by the initial concentration of EDCs in water. Interestingly, as concentrations of EDCs in freshwater usually remain in the order of ng L⁻¹ (Ding & Tzing 1998; Snyder et al. 1999; Kolpin et al. 2002; Johnson & Jürgens 2003; Moreira et al. 2009, 2011; Resende et al. 2017; Aquino et al. 2021), experiments with low concentrations may reflect more closely the EDCs degradation in a real application scenario.

The results therefore evidence that photolysis can be efficient for the removal of estrogenic activity caused by the studied EDCs during drinking water treatment. This can be achieved by providing an UV dose capable to remove estrogenic activity from water. At the operating conditions of this work, this dose can be obtained with an exposure time of about 10 min, but this contact time can be decreased by using reactors that can provide high UV light intensities.

Some studies have been reported the formation of EDC by-products during photolysis, raising concerns about the toxicity of these products (Caupos et al. 2011; Li et al. 2013; Mboula et al. 2015; Souissi et al. 2012). However, little information concerning their toxicity is currently available. In this regard, studies have pointed out that the phenol ring of EDCs may persist after photolysis, suggesting the endocrine disrupting potential of by-products (Mboula et al. 2015; Souissi et al. 2012). Furthermore, as reported by Souissi et al. (2012) photolysis can lead to the formation of hydroxylated by-products and it is important to highlight that some hydroxylated products of EDCs (e.g., 2-hydroxyestradiol, 4-hydroxyestradiol, 16α-hydroxyestradiol, and 4-hydroxyestrone) can be carcinogenic.

At the conditions investigated, both chlorination and photolysis resulted in high removal efficiencies of the estrogenic activity induced by the mixtures of EDCs tested in this work. Chlorination resulted in a removal of up to 99.7% of the estrogenic activity of samples at a chlorine dose of 2.75 mg · L⁻¹ and contact time of 30 min. In all EDC solutions tested, photolysis led to removal efficiencies higher than 99.9% at a UV dose of 186 mJ · cm⁻². It is worth highlighting that the EDC concentrations used in this study were based on values observed in water sources, whereas the conditions tested were determined according to water disinfection operating conditions, which corroborates the applicability of the proposed techniques in drinking water treatment plants. Furthermore, these processes may lead to the formation of by-products and little information concerning the health risks associated with these products in water is currently available. In this context, a combination of in vivo and in vitro techniques can be valuable for the assessment of the occurrence and the potential risks related to these products in treated waters.

All relevant data are included in the paper or its Supplementary Material.

The authors declare there is no conflict.