Ecological Risk Assessment of Pharmaceuticals and Personal Care Products in the Water Environment of 15 Cities in Japan

To assess the ecological risk of pharmaceuticals and personal care products (PPCPs) to the water environment of several cities in Japan, major local environmental research institutes, a private company, and an academic institution launched a joint research project in 2019. Under this initiative, local environmental research institutes surveyed the concentrations of 46 types of PPCPs at 59 points distributed across 15 cities in Japan. IDEA Consultants, Inc. calculated the unknown values of predicted no-effect concentration (PNEC) of six chemicals (telmisartan, candesartan, fexofenadine, diphenhydramine, diphenyl sulfone, and ketotifen) through bioassay experiments on aquatic organisms. Among the researched chemicals, the concentrations of clarithromycin, 14-hydroxyclarithromycin, erythromycin, diclofenac, carbamazepine, and telmisartan exceeded the PNEC in at least one sampling point. However, ozone treatment removed most of these chemicals, except for certain phosphate ester ame retardants. The mass balance of chemicals in the Tamagawa River owing through Tokyo Prefecture was calculated by multiplying the concentration of each chemical with the ow rate at each sampling point in the river. The measured load of most chemicals at each sampling point of the Tamagawa River coincided to a certain extent with the cumulative load accumulated from the tributaries and sewage treatment plants to the uppermost point (Nagata Bridge). However, the measured load of diclofenac was signicantly smaller than the estimated values at each sampling point, suggesting that diclofenac photodegrades while owing down the river. of the ecological risks posed by PPCPs in Japan. The study demonstrated that environmental surveys could be eciently conducted over a wide area through joint research. However, the study was conducted in winter, and it is known that the discharge of PPCPs into the environment varies with the season (Golovko et al. 2014; Marques dos Santos et al. 2019). Therefore, we plan to conduct a survey in summer to better understand the concentrations, and therefore risks, of PPCPs. In addition, we plan to conduct a survey of wastewater discharged from sites other than STPs, especially commercial sites, to better understand the various sources of PPCP emissions.


Introduction
Chemicals such as drugs, cosmetics, pesticides, and fertilizers are essential to maintain the current lifestyle of most societies worldwide. However, several of these chemicals are potential environmental pollutants (such as dioxins, polychlorinated biphenyls, and dichlorodiphenyltrichloroethane) and adversely affect resources and living beings (Carson 1962 institutes investigated the PPCP contamination values, and the private company performed the toxicity assessment to provide information for the ecological risk assessment. Additionally, 11 regional environmental research institutes also cooperated in water sampling. This was the rst time in Japan that a wideranging survey and risk assessment of PPCPs in the water environment across 15 areas were conducted.  Table 1 shows the chemicals analyzed in this study, which include pharmaceuticals such as antibiotics, antihypertensives, and antihistamines, and phosphate ester ame retardants (PFRs); these were selected based on the comprehensive analysis of former research (Environmental Restoration and Conservation

Reagents
Agency of Japan 2020) while considering their high consumption and frequent detection in aquatic environments. Information on the primary use or origin of each chemical was collected from several sources (DrugBank, ChemIDplus Advanced, and PubChem). Deuterium or 13 C labeled analytes were used in surrogates to maintain the highest quantitative accuracy for the analysis. For the analysis of pharmaceuticals, standard chemicals were individually dissolved with methanol at 100-1000 ng mL − 1 as the rst solution, which was then mixed and diluted to 1.0 ng mL − 1 with methanol. The surrogate mixture solution was prepared by a similar method. To prepare the standard PFRs and their surrogates, mixture solutions (Hayashi Pure Chemicals) were purchased and prepared using the same method as that employed for the pharmaceutical samples. The chemicals were eluted from the dried cartridge using 3 mL of methanol, followed by the addition of 3 mL of acetone and 2 mL of dichloromethane; these three solvents were mixed in a 10-mL glass tube. The mixed solvent was concentrated with nitrogen gas to approximately 0.1 mL and adjusted to 1 mL with 80% methanol aqueous solution. The solution was ltered with a syringe lter (Millex®-LG, pore size: 0.2 µm, diameter: 4 mm; Merck Millipore Corporation, Burlington, MA, USA) and analyzed with LC-MS/MS (Xevo-TQS: Waters Associates). Fig. S1 (Online Resource 1) shows the analysis procedure for the pharmaceuticals.

PFRs
Plastic materials were avoided in the extraction of PFRs since certain PFRs exhibit plasticizer properties. InertSep Glass PLS-3 (200 mg/6 mL) was used for PFR extraction after being preconditioned with 8 mL of methanol followed by 10 mL of ultrapure water. To analyze the water samples, 10 µL of surrogate mixture solution (1.0 mg L − 1 for each chemical) was added to 200 mL of river water samples (50 mL for STP in uents, and 100 mL for STP e uents); the subsequent mixture was poured into a polypropylene container connected to a concentrator. Although plastic materials were avoided in the PFRs analysis, glass containers were excessively heavy and unstable to be connected to a concentrator.
Before the extraction, the containers were thoroughly washed with ultrapure water and methanol. The samples were loaded to the solid-phase cartridges at 10 mL min − 1 . The cartridge was then washed with 10 mL of ultrapure water and dried with a ush of nitrogen gas. The chemicals were eluted from the dried cartridge with 8 mL of acetone. The solvent was concentrated with nitrogen gas to approximately 0.1 mL and adjusted with 1 mL of methanol. The solution was ltered with a syringe lter (Millex®-LG, pore size: 0.2 µm, diameter: 4 mm; Merck Millipore Corporation) and analyzed with LC-MS/MS (Xevo-TQS: Waters Associates). Fig. S2 (Online Resource 1) shows the analysis procedure for PFRs, and Table S1 (Online Resource 1) shows the LC-MS/MS analysis conditions for pharmaceuticals and PFRs.

Quality control
Recovery tests were conducted as quality tests using STP e uent and ultrapure water. The method detection limits (MDLs) and method quanti cation limits (MQLs) were estimated based on data from eight different analyses of the standard PPCP mixture added to ultrapure water at 0.5 and 5.0 ng L − 1 .

PNECs of PPCPs
For the risk assessment, the measured value of each analyte was compared with the respective predicted no-effect concentration (   Among the predicted no-effect concentrations (PNECs) of TBOEP, 21000 ng L − 1 was calculated by dividing the acute toxicity data for sh (Oryzias latipes) of factor of 1000, following the guidelines of the Initial Environment Risk Assessment of Chemicals.
The PNEC of 14-hydroxyclarithromycin, 270 ng L − 1 was calculated by dividing the chronic toxicity data for algae (Anabena os-aquae) of 2.7 µg L − 1 (NOEC) the guidelines of the Initial Environmental Risk Assessment of Chemicals in Japan.
For each chemical, the PNEC used for the risk assessment in this study is highlighted and written in bold.

River water samples
The total number of sampling points was 59. The water samples were collected by hanging a stainless-steel bucket from either the bridge or shore for each sampling point from July to September 2019. If the sampling point was in uenced by the ebb and ow of the tide, sampling was conducted during the low tide. The collected samples were sent to the Tokyo Metropolitan Research Institute for Environmental Protection under refrigeration. The received samples were immediately processed by a solid-phase extraction method and then analyzed using LC-MS/MS. The sampling points of this study (considering the four institutes participating in this joint research) are shown in Table S2 (Online Resource 1) and Fig. S3 (Online Resource 1).

PPCPs in STPs
As

Mass balance of chemicals in the Tamagawa River in Tokyo
The mass balance of six chemicals, whose concentrations exceeded PNEC in at least one sampling point, was calculated for the Tamagawa River owing through Tokyo.
The Tamagawa River has a simple channel because its ow is not in uenced by the tide from the Chofu water intake, and it does not branch out until its estuary (Fig. 1). Furthermore, its shallowness allows the ow velocity entering the river channel to be measured more conveniently. The width of the Tamagawa River was rst measured, then divided into 10-12 sections, and nally, the ow velocity of each section was measured using a current meter. When the depth of each sampling point was ≥ 40 cm, the ow velocity was measured at two depth points: 20% and 80% of the depth from the surface to the bottom; when the depth was < 40 cm, it was measured at 60% of the depth. The ow rates from the STPs were obtained from the Tokyo Bureau of Sewerage.

Quality control
Recovery rates, calculated by subtracting the concentration of the sample without the chemical from that of the sample with the chemicals added, was within the ranges of 71-143% (average: 98%) and 61-147% (average: 99%) for ultrapure water and the STP e uent, respectively.

Calculation of MDLs and MQLs
The MDLs and MQLs of the analyzed chemicals were calculated based on the results of the recovery tests conducted by adding the chemicals to ultrapure water and adjusting the concentration at the lowest calibration curve level. Eight extracted samples were analyzed with LC-MS/MS, and the respective t-values were calculated based on the standard deviation for each chemical. The MDLs and MQLs were calculated using the following formulas (Currie 1997): MDL = 2 × t (n-1, 0.05) × σ n-1 × 2, and MQL = 10 × σ n-1 ; where t (n-1, 0.05) indicates the t-value (one-sided) for the risk factor of 5% and the degree of freedom of n-1.

PNECs of PPCPs
The assessment factors were obtained from the "Guidelines for the Initial Environmental Risk Assessment of Chemicals," published by the Ministry of Environment, Japan (2019). Following these guidelines, when chronic toxicity data were obtained for the three species ( sh, algae, and crustacean), the assessment factor was 10; however, when chronic data were obtained for one or two species, the assessment factor was 100. Among the three methods used to calculate PNECs, TG-212 is considered an acute toxicity test, while the others are chronic toxicity tests. Therefore, the assessment factor for the six analyzed chemicals was set at 100.
Algae are highly susceptible to 14-hydroxyclarithromycin, compared to other species, with half (50%) maximal effective concentrations (EC50) > 2000 µg L -1 for sh and EC50 > 2000 µg L -1 for crustaceans; the concentration that results in 50% inhibition of growth rate (ErC50) of algae is 27.2 µg L -1 (Baumann et al. 2015). In this case, according to the guideline published by the Ministry of Environment, Japan (2019), the assessment factor can be assumed to be 10.
Therefore, the PNEC of 14-hydroxyclarithromycin was set at 270 ng L -1 , calculated by dividing the lowest NOEC (2.7 µg L -1 ) by the assessment factor of 10.
Similarly, for tris(2-butoxyethyl) phosphate (TBOEP), the minimum PNEC was calculated by dividing 21 mg L -1 (96 h-lethal concentration for 50% of animals (LC50)) (obtained from the toxicity tests conducted by the Ministry of the Environment, Japan) by the assessment factor of 1000.

River water samples
Analyzed data for PPCP detected at a concentration more than the PNEC of each chemical or 1000 ng L -1 , in at least one sampling point, are shown in Table 3, and the details of the data are shown in Tables S3 and S4. The names of the municipalities were indicated with letters (A to K), and the sampling points in each municipality were assigned identi cation numbers (e.g., "A-1" or "C-5") based on the data obtained from the 11 institutes collaborating in this joint research. The blank test value was subtracted from the raw measured values for N,N-diethyl-m-toluamide (DEET) and diphenyl sulfone because the blank test values were above the MDL for both chemicals. The gray cells shown in Table 3 indicate that the measured values exceeded the respective PNECs. The concentrations of clarithromycin, 14-hydroxyclarithromycin, erythromycin, carbamazepine, diclofenac, and telmisartan exceeded their respective PNECs at several sampling points, and almost all these points were in the downstream areas.  Table 4 shows the PPCP concentrations at each process step in the six investigated STPs. The removal rates of the macrolide antibiotics, i.e., clarithromycin, 14-hydroxyclarithromycin, and erythromycin, were approximately 24-53% (35% on average), 18-57% (36% on average), and − 59-84% (− 10% on average), respectively. The concentration of erythromycin in the e uent of F STP was below the MQL. As mentioned above, all e uent and in uent STP samples analyzed in this study were composite samples used to determine the average variability of pollution load throughout the day. However, when preparing composite samples, sampling of both in uent and e uent fractions starts simultaneously. In typical STPs, there is a time lag of 6-8 h even only for a reactor in which conventional activated-sludge treatment is conducted (Japan Sewage Works Association 2013). Therefore, even for composite samples, the gap due to time lag cannot be compensated totally. hydroxycarbamazepine, and carbamazepine-10,11-epoxide) were scarcely removed, and this trend was also reported by Jelić et al. (2011). The overall concentrations of chemicals were greatly reduced by ozone treatment in STP A. In terms of PFRs, triphenyl phosphate (TPhP), TBOEP, and tricresyl phosphate (TCP) were relatively well removed (average removal rates of 61%, 73%, and 75%, respectively). In addition, the ratio of PFRs subjected to biodegradation removal by sewage treatment was almost negligible. For example, the removal rate of triethyl phosphate (TEP) calculated by the EPI suite and attributed to biodegradation was 0.09%, while the total removal rate was 1.87%. Similarly, a TPhP removal rate of 0.56% was attributed to biodegradation, while the total removal rate was 60.71%. The correlation factor between removal rates and octanol-water partition coe cient (log K ow ) of PFRs was 0.6989. In contrast, the removal rates obtained by ozonation ranged from − 23% for tributyl phosphate (TBP) to 38% for TBOEP, with an overall low value for PFRs. The removal of hydrophobic chemicals, such as TBOEP and TBP, was nearly negligible, including that by ozonation. Among PFRs, tris(2-chloroethyl) phosphate (TCEP), tris(1,3-dichloro-2-propyl) phosphate (TDCPP), and tris(2-chloroisopropyl) phosphate (TCPP) were particularly unsusceptible to ozonation (removal rates < 5%); however, their concentrations were found to be reduced by advanced oxidation processes (AOPs; UV/H 2 O 2 treatment) (Yuan et al. 2015

Conclusions
In this study, we conducted a nationwide survey and ecological risk assessment of PPCPs in water environments in Japan. For risk assessment, PNEC values were collected from the reported literature, such as studies on ecotoxicological aspects. For pharmaceuticals whose PNEC values have not been clari ed, their PNECs were calculated by conducting bioassay experiments on aquatic organisms. In the risk assessment study, 59 water samples were collected from sampling points throughout Japan and analyzed for chemicals using LC-MS/MS. The PNEC concentration was exceeded at one point, at least, for clarithromycin, 14-hydroxyclarithromycin, erythromycin, diclofenac, carbamazepine, and telmisartan. The removal rates for these chemicals at STPs were investigated by comparing their concentrations in the in uent and e uent.
Although many pharmaceuticals and PFRs were not effectively removed by the activated-sludge treatment, almost all of them were decomposed by ozonation, except for certain PFRs. The mass balance of six chemicals, whose concentrations exceeded PNEC at one sampling point at least, was calculated by multiplying the concentration of each chemical and the ow rate through the Tamagawa River. The results indicated that diclofenac is likely photodegraded while owing down the river. Through this joint research project, many institutes and different industries were able to work together to contribute to a deeper understanding of the ecological risks posed by PPCPs in Japan. The study demonstrated that environmental surveys could be e ciently conducted over a wide area through joint research. However, the study was conducted in winter, and it is known that the discharge of PPCPs into the environment varies with the season (Golovko et al. 2014; Marques dos Santos et al. 2019). Therefore, we plan to conduct a survey in summer to better understand the concentrations, and therefore risks, of PPCPs. In addition, we plan to conduct a survey of wastewater discharged from sites other than STPs, especially commercial sites, to better understand the various sources of PPCP emissions.

Declarations
Ethics approval and consent to participate Not Applicable

Consent for publication
Not Applicable Availability of data and materials All data generated or analyses during this study are included in this published article and its supplementary information les.

Competing interests
None Funding This study was supported by the Environment Research and Technology Development Fund (JPMEERF20195054) of the Environmental Restoration and Conservation Agency of Japan.

Authors' contributions
Conceptualization of the research project, writing the original draft, and funding acquisition were performed by TN. Methodology was performed by MK and YM (PPCP analysis) and TO and AS (bioassay experiments with aquatic organisms). Project administration was performed by MK and YM (Tokyo prefecture), TT (Osaka city), CM (Hyogo prefecture), and HH (Nagoya city). Investigation was performed by DA and MO (Osaka city), YH and TK (Hyogo prefecture), HH (Nagoya city). Review and editing of the manuscript were performed by DA, YH, and HH. AS advised on toxicity information. All authors read and approved the nal manuscript.