Environmental Risk Assessment of Active Human Pharmaceutical Ingredients in Urban Rivers in Japan

Abstract

Active pharmaceutical ingredients (APIs) have become a public concern owing to their possible adverse effects on aquatic organisms. Ministry of Health, Labor and Welfare in Japan (MHLW) issued “Guidance on the Environmental Risk Assessment (ERA) in new pharmaceutical development” in 2016. To evaluate the validity of phase 1 in the MHLW’s ERA guidance, we monitored the measured environmental concentrations (MECs) of approved APIs in urban rivers and sewage treatment plants (STPs) in Japan and compared these MECs with the predicted environmental concentration (PEC). We collected water samples from urban seven rivers and three STPs during each season. Fifty-one APIs for human and veterinary use and the artificial sweetener sucralose were analyzed by liquid chromatography-tandem mass spectrometry (LC-MS/MS). Forty-four APIs were observed in the rivers and 42 were found in the influent and effluent of STPs, with levels ranging from nanograms to micrograms per liter. The action limit in phase I of the MHLW’s guidance was set to 10 ng/L, and there was no API except for ketoprofen, for which PEC of the MHLW’s guidance (PECjapan) was lower than 10 ng/L and the maximum MEC (MECmax) was 10 ng/L or greater. Almost all APIs also had median MECs that were lower than those of the respective PECjapan. These results indicate that the PECjapan values in phase I of the MHLW’s guidance were appropriate. However, some APIs had MECmax values that were greater than those of the respective PECjapan due to overestimation of the dilution factor of river water and/or underestimation of API production.

Introduction

Since the second half of the 1990s, active pharmaceutical ingredients (APIs) have been a topic of concern.^1–3) Many researchers have reported on the occurrence and behavior of APIs in aquatic environments, and the levels of contamination by APIs, including analgesics, antiphlogistics, lipid regulators, and antidepressants in the aquatic environment, with levels ranging from less than a nanogram to micrograms per liter.^1–5) Their presence in drinking water has been reported by WHO,⁶⁾ and reported in countries such as the U.S.A.,^7–9) China,¹⁰⁾ and Japan.¹¹⁾ Furthermore, the risk they pose to the environment is an established issue. The effects of APIs on human health are unclear.⁶⁾ The effects of APIs on aquatic ecosystems are also of public concern. There is little evidence available on the effects of specific APIs on aquatic wildlife based on actual in situ concentrations, except for endocrine disruptors, such as synthetic estrogens.¹²⁾ In general, the concentrations of APIs that are found to affect algae, daphnia, and fish are 1000-fold greater¹³⁾ and are considerably higher than the measured environmental concentrations (MECs). However, under experimental conditions, selective serotonin reuptake inhibitors, such as fluoxetine and paroxetine, were found to inhibit feeding behavior.¹⁴⁾ Moreover, clarithromycin, an antibiotic, inhibits the growth of algae.¹⁵⁾ High MECs of oseltamivir phosphate, an antiviral medicine used in the treatment of the flu, was also observed to facilitate the development of oseltamivir-resistant influenza viruses in urban rivers in Japan.¹⁶⁾ In addition to the toxicity of specific APIs, the co-occurrence of active substances having a similar mode of action is an important issue in the evaluation of the risks of APIs to ecosystems.¹⁷⁾ Under these circumstances, several nations have developed formalized regulations and frameworks to assess the transport, fate, and effects (toxicity) of APIs in the environment.¹⁸⁾ The U.S. Food and Drug Administration (USFDA), the European Medicines Agency (EMA), and Health Canada are committed to conducting environmental risk assessments (ERAs). The USFDA issued guidelines for the environmental assessment of human drugs and biological agents in 1996.¹⁹⁾ The EMA also provided guidelines for the ERAs of human APIs in 2006.²⁰⁾ In these two ERA systems, the predicted environmental concentration (PEC) of the API is compared to a threshold value in the first step. According to the USFDA guidelines, the PEC is based on the annual production of API and annual sewage volume in the U.S.A. The EMA guidelines state that the PEC is based on the dosage and sewage volume per patient. The PEC of the API in both ERAs was compared to a threshold value in the first-tier assessment. The reliability of the PEC is a key factor in the assessment of environmental risks to aquatic systems. If the PEC of the API is lower than the threshold value, the assessment ends at the first tier. However, if it exceeds the threshold value, API is evaluated at a higher tier of assessment, and the PEC is available comprising the toxicity data for algae, Daphnia, and fish. Therefore, if in the first-tier assessment, the PEC is necessarily conservative, then the PEC of the API should be equal to or higher than the actual concentration in the environment. The global position on including human medicines in ERAs has gradually changed, and the topic of environmentally persistent pharmaceutical pollutants (EPPPs), which are distinguished from persistent organic pollutants, such as polychlorinated biphenyls (PCB), dichlorodiphenyltrichloroethane (DDT), and dioxins, was finally adopted as an emerging policy issue (EPI) in the Strategic Approach to International Chemicals Management at the Fourth International Conference on Chemicals Management in 2015.²¹⁾ Regarding EPPPs, it was agreed that international cooperation is important for promoting awareness, understanding, and countermeasures, such as adoption of EPIs. In Japan, guidance for the ERA of human APIs were established by Ministry of Health, Labour and Welfare (MHLW) of Japan in 2016.²²⁾ No study has evaluated the validity of the MHLW’s ERA guidance.

This study evaluated 51 APIs that were approved and present in the market in Japan with the following objectives: (1) to measure the MECs in seven urban rivers in Japan, (2) to investigate the behavior of APIs in the influent and effluent of three sewage treatment plants (STPs), and (3) to evaluate the validity of PECs calculated following the MHLW guidance, EMA and USFDA guidelines by comparing them to the respective MECs found in urban rivers.

Results

Accuracy of the Solid-Phase Extraction (SPE)-Liquid Chromatography Tandem Mass Spectrometry (LC-MS/MS) Method

The method detection limits (MDLs) of the APIs in river and effluent samples of STP under this analytical method were between 1.0 and 10 ng/L, except for ceftriaxone (445 ng/L), piperacillin (368 ng/L), ibuprofen (45.6 ng/L), sucralose (27.4 ng/L), cefdinir (24.5 ng/L), cefmetazole (20.0 ng/L), cefaclor (11.4 ng/L), and sulbactam (10.4 ng/L) (Supplementary Table S1). For influent samples of STP, the MDLs of APIs were determined to be 5-fold those found in river and effluent samples. API recoveries were examined by the addition of APIs to river water of Riv. No, 3 in Table 1, with final concentrations of 1 µg/L. The recoveries of most APIs were between 32 and 228%, and their relative standard deviations were lower than 17%. Among the APIs, the recoveries were lower than 70% for amantadine (65%), acetaminophen (66%), levofloxacin (53%), sulfamethoxazole (61%), ampicillin (65%), amoxicillin (32%), cefaclor (39%), sulfamonomethoxine (66%), terbinafine (59%), doxycycline (68%), aripiprazole (57%), and simvastatin (52%), and greater than 130% for cefmetazole (228%) and nateglinide (163%) (Supplementary Table S1).

Table 1. The Situation in Upstream Areas of the Sampling Sites in Selected Urban Rivers in Japan

River					Sewage treatment plants (STPs) in river basinb)					River flow/STP effluentd) (DF)
No.	Name	Sampling site bridge name	Monitoring stationa)		Number of STPs	Treatment process of STPsc)	Inhabitants (person)	Effluent (m3/d)	Effluent/inhabitants (L/person/d)
			Flow (m3/d)	Name
1	Shinkawa	Shinkawachou	429000*	—	1	AS	341000	185000	543	2.3
2	Hirose	Miyazawa	1011000	Hirose Bridge	1	A2O	46000	15000	326	67.4
3	Tama	Tamagawara	2057000	Ishihara	8	AS, A2O, O3	2118000	708000	334	2.9
4	Tsurumi	Ochiai	916000	Kamenoko Bridge	3	AS, A2O	1016000	304000	299	3.0
5	Shonai	Shonaigawa	2307000	Biwajima	10	AS, A2O	727000	356000	490	6.5
6	Katsura	Miyamae	4562000	Hazukashi	2	A2O, AO, O3	876000	680000	776	6.7
7	Chikugo	Mametsu	10852000	Senoshita	18	AS, OD	342000	99000	289	109.6

a) Average river flow in 2016, Water Information System, Ministry of Land, Infrastructure, Transport and Tourism in Japan. *: data from Hokkaido Pref. —: no monitoring station. b) Data from each Bureau of Sewerage of the following Prefectures: River No. 1: Hokkaido, No. 2: Iwate, No. 3: Tokyo, No. 4: Kanagawa, No. 5; Aich and Gifui, No. 6: Kyoto, No. 7: Fukuoka, Saga, and Oita. c) AS: activated sludge, AO: anaerobic-oxic, A2O: anaerobic-anoxic-oxic, O3: ozone, OD: oxidation ditch. d) DF: dilution factor.

Occurrence of APIs in the Seven Urban Rivers in Japan

Information about the sampling sites on the seven urban rivers is provided in Table 1. The flow rates at the urban river sampling sites were between 429000 and 10852000 m³/d. Effluent from STPs flowed into each river upstream of the sampling site. The drainage of each river basin ranged between 299 and 776 L/person/d. The DF of the summed STP effluent in each of the seven rivers was calculated to be between 2.3 and 109.6.

The detection frequencies of the APIs, which were calculated as the number of detections above the MDL, and the maximum MECs of the APIs in the seven rivers are shown in Fig. 1 and Supplementary Table S2. The river water was sampled once per season (four seasons) in each river. Forty-four of the 51 APIs were detected in the samples. The APIs detected at relatively high frequencies were as follows: crotamiton, olmesartan, valsartan, carbamazepine, irbesartan, sulpiride, candesartan, bezafibrate, and amantadine were detected over 80% and 100% frequency. DEET, clarithromycin, losartan, diclofenac, memantine, diphenhydramine, trimethoprim, and erythromycin, and the artificial sweetener sucralose occurred at frequencies ranging between 70 and 80%. Epinastine, phenytoin, acetaminophen, levofloxacin, and roxithromycin were detected at frequencies between 60 and 70%.

Fig. 1. Detection Frequency and Maximum Measured Environmental Concentration (MECmax) of Active Pharmaceutical Ingredients (APIs) in Seven Urban Rivers in Japan

† Veterinary medicine.

In this study, the maximum MECs of the API were lower than 1000 ng/L. The maximum MECs of sulpiride, crotamiton, acetaminophen, piperacillin, olmesartan, and sucralose were between 500 and 1000 ng/L. Furthermore, the maximum MECs of valsartan, DEET, clarithromycin, bezafibrate, epinastine, cefaclor, cefmetazole, erythromycin, irbesartan, ketoprofen, oxytetracycline, losartan, and candesartan ranged from 100 to 500 ng/L. Amantadine, diphenhydramine, levofloxacin, ibuprofen, ampicillin, diclofenac, rosuvastatin, carbamazepine, memantine, sulfamethoxazole, trimethoprim, cefdinir, roxithromycin, lincomycin, sulbactam, phenytoin, diltiazem, olmesartan medoxomil, benzylpenicillin, and mefenamic acid had maximum MECs of 10–100 ng/L. Although macrolide antibiotics (clarithromycin, erythromycin, and roxithromycin), sulfa drugs (sulfamethoxazole and trimethoprim), and new quinolone antibiotics (levofloxacin) were observed at relatively high frequencies among the antibiotics measured, the other antibiotics investigated were not detected and/or had both a low frequency and a low MEC. The median MECs of the APIs detected in the seven sampled rivers were between 3.5 to 81.9 ng/L, except for crotamiton which was 172.7 ng/L, (Supplementary Table S2).

The sum of the API MECs in the urban rivers was in the following order: River No. 1 > No. 3, No. 4 > No. 5, No. 6 > No. 7 > No. 2. The API MECs in rivers with low DFs were higher than those in rivers with high DFs. The relationship between the MEC and the sum of the MECs in the seven rivers is shown in Fig. 2. The MECs of olmesartan, carbamazepine, crotamiton, irbesartan, sulpiride, valsartan, and memantine were correlated with the sum of the API MECs. Their correlation factors (r) were 0.855, 0.897, 0.865, 0.936, 0.866, 0.706, and 0.861 (p < 0.01), respectively. The artificial sweetener sucralose, which is an indicator of sewage contamination in rivers, was detected at a high frequency (78%), and was correlated with the sum of API MECs (r = 0.853, p < 0.01).

Fig. 2. Relationship between the Measured Environmental Concentrations (MECs) of APIs and the Sum of the API MECs in Sampled Urban Rivers in Japan

**: p < 0.01.

API MECs in winter and spring tended to be higher than those in summer and autumn. However, the river flows in winter and spring were lower than those in summer and autumn (Fig. 3). The MECs of crotamiton, epinastine, and DEET appeared to be higher in winter, spring, and summer, respectively, in some of the rivers (Fig. 4). The other APIs examined exhibited no obvious seasonal changes.

Fig. 3. Relationship between River Flow Rate and the Sum of the Measured Environmental Concentrations of APIs in Urban Rivers in Japan

Closed squares (■), open squares (□), open circles (○), and closed circles (●) present spring, summer, autumn, and winter, respectively. *: p < 0.05.

Fig. 4. Seasonal Changes in Active Pharmaceutical Ingredient Loads in Seven Urban Rivers in Japan

Occurrence of APIs in the Influent and Effluent of Urban STPs in Japan

The API MECs in the three STPs located near urban rivers were examined (Table 2). STP B was the largest of those examined. The water discharge per inhabitant in the STPs ranged from 237 to 421 L/person/d. The treatment processes in the STPs were similar, and involved screening, activated sludge, and partially anaerobic-anoxic-oxic (A2O) treatment, followed by chlorine treatment for disinfection.

Table 2. Characteristics of the Sewage Treatment Plants (STPs) in Urban Areas in Japan

STP	Inhabitants (person)	Influenta)		Influent/Inhabitants (L/person/d)	Effluenta)		Treatment processb)
		Average (m3/d)	S.D. (m3/d)		Average (m3/d)	S.D. (m3/d)
A	2670000	633000	125000	237	621000	125000	AS, A2O, chlorination
B	2127000	895000	56000	421	935000	56000	AS, A2O, chlorination
C	489000	169000	14000	346	175000	14000	AS, A2O, chlorination

a) Sampling time: Feb., Aug., and Oct. in 2017, Jan. and, Apr. in 2018; S.D.: standard deviation. b) Mainly AS, partially A2O; AS: activated sludge, A2O: anearoboc-anoxic-oxic.

API MDLs in the influent and effluent samples are listed in Supplementary Table S1. The MDLs of ceftriaxone and piperacillin were higher than those of the other APIs. For the influent, the APIs that were detected at relatively high frequencies (>60%) in the seven rivers were observed in all the water samples. The APIs with maximum MECs (>1 µg/L) in the influent were crotamiton, DEET, valsartan, bezafibrate, clarithromycin, acetaminophen, sucralose, levofloxacin, cefmetazole, ibuprofen, and terbinafine (Fig. 5 and Table 3). The APIs that were detected in the influent were also observed in the effluent, and their MECs, except for crotamiton, sucralose, and levofloxacin, were lower than 1 µg/L (Supplementary Table S3). The average removal rates of crotamiton, olmesartan, carbamazepine, irbesartan, memantine, trimethoprim, erythromycin, epinastine, roxithromycin, and lorazepam were lower than 20% in the STPs. The average removal rates of carbamazepine, irbesartan, memantine, erythromycin, epinastine, roxithromycin, and lorazepam were negative values (Fig. 5).

Table 3. MEC/Predicted Environmental Concentration (PEC) of APIs Using the Environmental Risk Assessment Systems of MHLW, EMA, and the USFDA

No.	APIs	DOSEaia) (mg/d)	Productionb) (kg/year)	MECmax in 7 rivers (ng/L)	MHLWc)				EMAd)				USFDAe)
					PECjapan	MECmax/PEC			PECema	MECmax/PEC			EIC-aquatic (ng/L)	MECmax influent of 3 STPs (ng/L)	MECmax/EIC
					DF = 10 (ng/L)	DF = 10	DF = 5	DF = 2	DF = 10 (ng/L)	DF = 10	DF = 5	DF = 2
1	Crotamiton	—	58800#	845.0	384	2.203	1.102	0.441	NC	NC	NC	NC	3836	2290	0.597
2	Olmesartan	40	18648#	570.6	122	4.691	2.346	0.938	200	2.853	1.427	0.571	1216	562	0.462
3	Valsartan	160	27800#	405.1	181	2.234	1.117	0.447	800	0.506	0.253	0.101	1813	2080	1.147
4	Carbamazepine	1200	53752	60.5	351	0.172	0.086	0.034	6000	0.010	0.005	0.002	3506	98	0.028
5	Irbesartan	200	17500#	161.9	114	1.418	0.709	0.284	1000	0.162	0.081	0.032	1142	177	0.155
6	Sulpiride	600	26516	916.5	173	5.298	2.649	1.060	3000	0.305	0.153	0.061	1730	684	0.395
7	Candesartan	10	4891#	113.0	32	3.541	1.771	0.708	50	2.260	1.130	0.452	319	135	0.422
8	Bezafibrate	400	84504	200.2	551	0.363	0.182	0.073	2000	0.100	0.050	0.020	5512	4380	0.795
9	Amantadine	300	81906	95.3	534	0.178	0.089	0.036	1500	0.064	0.032	0.013	5343	241	0.045
10	DEET	—	—	260.8	NC	NC	NC	NC	NC	NC	NC	NC	NC	4350	NC
11	Clarithromycin	800	130632	225.6	852	0.265	0.132	0.053	4000	0.056	0.028	0.011	8521	1830	0.215
12	Sucralose	—	—	992.4	NC	NC	NC	NC	NC	NC	NC	NC	NC	8430	NC
13	Losartan	100	16242	116.8	106	1.102	0.551	0.220	500	0.234	0.117	0.047	1059	282	0.266
14	Diclofenac	100	48966	66.3	319	0.208	0.104	0.042	500	0.133	0.066	0.027	3194	194	0.061
15	Memantine	20	1896#	47.4	12	3.834	1.917	0.767	100	0.474	0.237	0.095	124	100	0.810
16	Diphenhydramine	160	13400#	82.0	87	0.939	0.469	0.188	800	0.103	0.051	0.021	874	635	0.726
17	Trimethoprim	320	3776	44.1	25	1.789	0.894	0.358	1600	0.028	0.014	0.006	246	155	0.631
18	Erythromycin	1200	253043	162.3	1651	0.098	0.049	0.020	6000	0.027	0.014	0.005	16506	670	0.041
19	Epinastine	20	3942	197.1	26	7.664	3.832	1.533	100	1.971	0.985	0.394	257	330	1.281
20	Phenytoin	300	4139	19.5	27	0.723	0.361	0.145	1500	0.013	0.007	0.003	270	27	0.099
21	Acetaminophen	1500	155999	745.8	1018	0.733	0.366	0.147	7500	0.099	0.050	0.020	10176	58600	5.759
22	Levofloxacin	400	15570	72.0	102	0.709	0.355	0.142	2000	0.036	0.018	0.007	1016	2210	2.176
23	Roxithromycin	300	6348	30.0	41	0.725	0.363	0.145	1500	0.020	0.010	0.004	414	173	0.419
24	Sulfamethoxazole	2000	18878	46.0	123	0.373	0.187	0.075	10000	0.005	0.002	0.001	1231	420	0.341
25	Diltiazem	200	25515	18.6	166	0.112	0.056	0.022	1000	0.019	0.009	0.004	1664	46	0.027
26	Ketoprofen	150	1024#	150.1	7	22.470	11.235	4.494	750	0.200	0.100	0.040	67	789	11.813
27	Lorazepam	3	135#	7.3	1	8.290	4.145	1.658	15	0.487	0.243	0.097	9	26	2.949
28	Mefenamic acid	1500	5649	11.8	37	0.321	0.160	0.064	7500	0.002	0.001	0.000	368	123	0.333
29	Cefmetazole	2000	19023	178.8	124	1.441	0.720	0.288	10000	0.018	0.009	0.004	1241	1340	1.080
30	Rosuvastatin	5	3764#	64.9	25	2.644	1.322	0.529	25	2.596	1.298	0.519	246	420	1.712
31	Lincomycin	2000	38904	29.6	254	0.117	0.058	0.023	10000	0.003	0.001	0.001	2538	13	0.005
32	Ampicillin	3000	20385	68.2	133	0.513	0.256	0.103	15000	0.005	0.002	0.001	1330	30	0.023
33	Oxytetracycline†	—	186476	142.4	1216	0.117	0.059	0.023	NC	NC	NC	NC	12164	ND	NC
34	Sulbactam	1000	14782	25.3	96	0.263	0.131	0.053	5000	0.005	0.003	0.001	964	200	0.207
35	Olmesartan medoxomil	40	13300#	15.3	87	0.176	0.088	0.035	200	0.076	0.038	0.015	868	ND	NC
36	Amoxicillin	1000	142918	4.8	932	0.005	0.003	0.001	5000	0.001	0.000	0.000	9323	ND	NC
37	Benzylpenicillin†	—	803265	12.0	5240	0.002	0.001	0.000	NC	NC	NC	NC	52398	17	0.000
38	Clofibric acid	1500	—	5.2	NC	NC	NC	NC	7500	0.001	0.000	0.000	NC	12	NC
39	Amlodipine	5	17347	6.2	113	0.055	0.027	0.011	25	0.247	0.124	0.049	1132	73	0.065
40	Ibuprofen	600	12567	68.7	82	0.838	0.419	0.168	3000	0.023	0.011	0.005	820	1450	1.769
41	Cefdinir	300	12429	41.6	81	0.513	0.256	0.103	1500	0.028	0.014	0.006	811	ND	NC
42	Piperacillin	4000	11018	623.8	72	8.680	4.340	1.736	20000	0.031	0.016	0.006	719	ND	NC
43	Cefaclor	750	10886	188.4	71	2.653	1.327	0.531	3750	0.050	0.025	0.010	710	280	0.394
44	Chlortetracycline†	—	374438	7.2	2443	0.003	0.001	0.001	NC	NC	NC	NC	24425	ND	NC
45	Sulfamonomethoxine	1000	33521	6.6	219	0.030	0.015	0.006	5000	0.001	0.001	0.000	2187	49	0.023
46	Terbinafine	125	9009	ND	59	NC	NC	NC	625	NC	NC	NC	588	1170	1.991
47	Doxycycline	200	531820	ND	3469	NC	NC	NC	1000	NC	NC	NC	34691	22	0.001
48	Nateglinide	360	5805	ND	38	NC	NC	NC	1800	NC	NC	NC	379	41	0.107
49	Ceftriaxone	2000	12466	ND	81	NC	NC	NC	10000	NC	NC	NC	813	ND	NC
50	Tylosin A†	—	448513	ND	2926	NC	NC	NC	NC	NC	NC	NC	29257	35	0.001
51	Aripiprazole	24	1024#	ND	7	NC	NC	NC	120	NC	NC	NC	67	ND	NC
52	Simvastatin	20	26394	ND	172	NC	NC	NC	100	NC	NC	NC	1722	98	0.057

a) Attached ethical document information of the drug, Pharmaceuticals and Medical Devices Agency in Japan; —: no data. b) Current Survey of Production Concerning Pharmaceutical Industry (Fundamental Statistics), 2015, Ministry of Health, Labour and Welfare (MHLW) in Japan; —: no data. c) PECjapan = (Production (kg/year) × Fseason (1))/WEST (350 L)/Population (120000000 persons)/DILUTION (10)/Day (365 d). d) PECema = (DOSEai × Fpen (0.01))/WESTinhab (350 L) × DILUTION (10). e) EIC-Aquatic (ng/L) = A(kg/year)/B (4.2 × 10¹⁰ L/d)/C (365 d) × D (10⁹ µg/kg) × 10³ (µg/ng), MECmax is the maximum concentration in the effluent of three STPs in urban areas in Japan. † Veterinary medicine; #: NDB data (the second, 2015); NC: not calculable; ND: less than method detection limit (MDL) in supplementary Table S1.

Fig. 5. MECs of APIs in Influent and Effluent and Their Removal Rates at Urban Sewage Treatment Plants in Japan

† Veterinary medicine.

The loads of crotamiton, epinastine, DEET, and acetaminophen in the influent samples from the STPs were the highest in January, April, August, and January, respectively (Fig. 6). The maximum loads of crotamiton, epinastine, DEET, and acetaminophen in the influent of the three STPs were 1013, 269, 1683, and 46169 mg/d/1000 people, respectively. Changes in the valsartan load were relatively small. The maximum loads of DEET and acetaminophen in the effluent decreased to 168 and 13.5 mg/d/1000 people, respectively.

Fig. 6. Seasonal Changes in Active Pharmaceutical Ingredient Loads in Influent and Effluent at Urban Sewage Treatment Plants (STPs) in Japan

Closed squares (■), open circles (〇), and open triangles (△) represent STP A, B, and C, respectively.

Estimation of PEC Using the Three ERA Methods

The API MECs in the seven urban rivers were compared with their PECs calculated based on the MHLW guidance and EMA guidelines. Moreover, the API MECs in the inflow samples of the three STPs were compared with their EIC-Aquatic, which was calculated according to the USFDA guidelines (Table 3). For the MHLW guidance, the annual production of APIs in Japan was calculated based on data from the Annual Report of Pharmaceutical Industry Production Dynamics Statistics 2015. However, OTC production was not included in these statistics. When the production date was not listed in the report, the National Database of Health Insurance Claims and Specific Health Checkups of Japan (NDB, the second, 2015) were used.²³⁾

The relationships between the median MECs (MECmedian) of APIs in seven urban rivers in Japan and the PECjapan and PECema (the PEC calculated via the EMA method) are presented in Fig. 7. The ratios of MECmedian to PECjapan (MECmedian/PECjapan) for each API were mostly less than 1, with a minimum value of 0.0014, a maximum value of 3.97, and a median value of 0.207; however, the ratios for lorazepam and ketoprofen were slightly greater than 1, at 3.97 and 1.26, respectively. On the other hand, the MECmedian/PECema ratio for each API was less than 1, with a minimum value of 0.0005, maximum value of 0.474, and a median value of 0.010. The PECjapan of each API was one or two orders of magnitude lower than the PECema, and each PECjapan was closer to the MECmedian than the respective PECema.

Fig. 7. Relationship between Predicted Environmental Concentrations of APIs, Calculated According to Japan’s Guidance (PECjapan) and the European Medicines Agency’s Guidelines (PECema), and the Median of the Measured Environmental Concentrations (MECmedian) of APIs in Urban Rivers in Japan

PECjapan (DF = 10) and PECema (DF = 10) are listed in Table 3. APIs with the detection frequency of less than 3 in seven urban rivers are excluded. Closed circles (●) and closed triangles (▲) represent ketoprofen and lorazepam, respectively.

In the case of the maximum API MECs in the seven urban rivers, when the default DF value of 10 in the MHLW guidance was used, 16 APIs were underestimated. When using a DF value of 2, sulpiride, epinastine, ketoprofen, lorazepam, and piperacillin were underestimated. In River No. 1, sulpiride, ketoprofen, lorazepam, and piperacillin were underestimated. In River No. 4, epinastine, lorazepam, and ketoprofen were underestimated. Except for River No. 1 and 4, no APIs were underestimated in any of the other rivers that were investigated using a DF value of 2, as shown in Table 4 and Supplementary Table S4.

Table 4. Frequency of Underestimation of Active Pharmaceutical Ingredients (APIs) Following MHLW Guidance

Rivera）		Number of APIs, MEC > PECb)
No.	DF	MHLW guidance
		DF = 10	DF = 5	DF = 2	DF = 1
1	2.3	35	13	7	1
2	67.4	0	0	0	0
3	2.9	19	8	0	0
4	3.0	30	15	3	0
5	6.5	3	0	0	0
6	6.7	1	0	0	0
7	109.6	1	0	0	0

a) No. and DF refer to Table 1; DF = dilution factor. b) Total number of APIs in each river is 204 (51 APIs × 4 seasons).

For the EMA guidelines, DOSEai was obtained from the information in the attached ethical document of each drug, supplied by the Pharmaceuticals and Medical Devices Agency in Japan. When the DF value was set to 10, olmesartan, candesartan, epinastine, and rosuvastatin were underestimated. The rivers in which these APIs were underestimated were River No. 1, 3, and 4. When the DF value was 2, no APIs were underestimated.

Using USFDA guidelines, the APIs were estimated at the end of the pipeline. In this study, API MECs at the three STPs were used to calculate the EIC-Aquatic of the APIs. Valsartan, epinastine, acetaminophen, levofloxacin, ketoprofen, lorazepam, cefmetazole, rosuvastatin, ibuprofen, and terbinafine were underestimated.

Discussion

Occurrence of APIs in the Seven Sampled Rivers in Japan

In this study, prescription and OTC drugs, such as antipyretic, analgesic, anti-allergic, anti-diabetic, antihypertensive, anti-anxiety, anti-dementia, antiepileptic, and antibiotic drugs for human use, as well as some veterinary antibiotics were selected. These APIs, except for the antibiotics, were produced at a relatively high rate in Japan,²⁴⁾ and were frequently detected around the world.^1,2) Antibiotics for human use were selected from those discussed in previously published literature,^25,26) and/or those with an annual production greater than 10000 kg/year in Japan.²⁴⁾ For the selected antibiotics used for veterinary purposes, annual production was greater than 10000 kg/year in 2013 in Japan.²⁷⁾ Previous research has clarified that the discharge sources of APIs in rivers are mainly STPs in urban areas. The sewage system in Japan is well established, with a penetration rate of 91.4%, as established at the end of the 2018 fiscal year. It is well known that the API MECs in rivers in urban areas are generally higher than those in rural rivers.²⁸⁾ Therefore, seven urban rivers in Japan were selected. The DFs at the sampling sites in each river were different. The sum of the API MECs in River No. 1, 3, and 4, which had DFs lower than 3, were higher than those of River No. 2 and 7, which had DFs higher than 67. However, no correlation was observed between the DFs of the drainage of the STPs and the sum of their API MECs. The artificial sweetener sucralose is an indicator compound for STP discharge,²⁹⁾ but its concentration in the rivers was also correlated with the DFs of the drainage of the STPs. Nakada et al.³⁰⁾ suggested that crotamiton and carbamazepine in river water are also indicators of STP discharge. In addition to these two APIs, the MECs of olmesartan, irbesartan, valsartan, sulpiride, clarithromycin, amantadine, and memantine in the river water were correlated with the sum of the API MECs in this study. These seven APIs can be used as indicators to predict the degree of API MECs in Japan. The antihypertensive drugs olmesartan, irbesartan, valsartan, candesartan, and losartan, which are angiotensin II receptor blockers (ARBs), were detected in urban rivers in Germany.^31,32) The detection frequencies in the aforementioned studies and the MECs in this study were similar. The detection rates of ARBs are high, possibly due to their daily consumption, high excretion rate in their unchanged form, and low removal rate in STPs. Olmesartan is used as a prodrug with a medoxomil ester. However, the detection frequency and MEC of the medoxomil ester were lower than those of olmesartan in the seven rivers sampled in this study. This is due to the high excretion rate of hydrolysates in urine.³³⁾ Candesartan is also used as a cilexetil ester, and most of the ester is hydrolyzed in the intestine.³⁴⁾ Therefore, the detection frequency and MEC of the ester may be lower than those of candesartan. This suggests that it is necessary to measure the active metabolite of the prodrug in river water.

Hyperlipidemia drugs, such as rosuvastatin and simvastatin, which are hydroxymethylglutaryl-CoA (HMG-CoA) reductase inhibitors, showed lower detection rates and MECs than ARBs. Their detection rates were also low in Fracne.³⁵⁾ Bezafibrate^36–39) and clofibric acid^38,40) were observed at MEC levels ranging from ng/L to µg/L in Canada, U.S.A., Brazil, and Italy. In Japan, bezafibrate was detected at high frequencies, but clofibric acid was not detected. This might be to the low production of clofibrate (less than 1 t per year) in recent years. For calcium channel blockers, the detection frequency and MECs of diltiazem were higher than those of amlodipine, the daily dose of diltiazem was higher than that of amlodipine, and the removal rate of diltiazem was lower than that of amlodipine. The antiepileptic drug carbamazepine has not only been observed at high frequencies in rivers worldwide but has also been observed in seawater that receives urban river water.^{35,37,39,41–46)} Carbamazepine has also been widely detected in Japan. Memantine, a therapeutic drug for Alzheimer’s dementia, was detected at relatively high frequencies in rivers. In Japan, Alzheimer’s dementia has been increasing annually in recent years. Therefore, other drugs used to treat the disease may also be observed in urban rivers. Among the antibiotics, macrolides, such as clarithromycin, erythromycin, and roxithromycin, the sulfa drug sulfamethoxazole, and the new quinolone antibiotic levofloxacin were detected at high frequencies. Other antibiotics tested were not observed or had relatively low MECs. This detection pattern of antibiotics was similar to that in rivers in the U.S.A., the EU, and Asia.^15,37,41,47)

In this study, veterinary antibiotics, such as oxytetracycline, benzylpenicillin, and chlortetracycline, were investigated and compared with antibiotics for human consumption. Oxytetracycline and tetracycline have been detected in rivers in the U.K.⁴⁸ and Norway.⁴⁹⁾ However, their detection rates and MECs are extremely low in urban rivers in Japan.

Seasonal changes in API MECs have been previously reported. API MECs in rivers were highest in winter in Finland.⁵⁰⁾ The MECs of OTC drugs in rivers increased during the hay fever season in Australia.⁵¹⁾ In Japan, epinastine, DEET, and crotamiton increased during spring, summer, and winter, respectively.

Occurrence and Fate of APIs in STPs in Japan

Almost all the APIs detected in the urban rivers were observed in the influent and effluent of the STPs. Moreover, their MECs were between ng/L and µg/L. No seasonal changes were observed in the API loads of the STP influent in Greece.⁵²⁾ In Italy, the values in summer were approximately half those in winter.³⁶⁾ In this study, the loads of most APIs changed little, however, crotamiton, epinastine, DEET, and acetaminophen exhibited seasonal changes. Epinastine is an antihistamine and has been approved as an OTC drug in recent years. Its usage increases from February to April, when hay fever becomes prevalent.⁵³⁾ DEET usage increases from June to September when mosquitoes are rampant.⁵⁴⁾ The use of crotamiton and diphenhydramine increases in winter when itching develops owing to low humidity. Acetaminophen is co-administered as an antipyretic when the flu develops in winter.⁵⁵⁾ The API loads of the effluent did not change significantly, except for the APIs with low removal rates. Furthermore, the amount of effluent at the STPs did not change significantly, except after heavy rainfall. API MECs in rivers may depend on the volume of surface water provided by rainfall, as mentioned in a previous study.⁵⁶⁾ Assessment of the removal rates of APIs in STPs has revealed that carbamazepine, crotamiton, clarithromycin, erythromycin, roxithromycin, diclofenac, olmesartan, candesartan, and irbesartan^31,57) are removed marginally by microbial degradation and/or absorption in sludge. In this study, the removal rates of these APIs were also found to be relatively low, and adamantane derivatives, such as amantadine and memantine, are also difficult to remove at STPs. The MECs of some APIs were found to be higher after treatment processing at STPs. Marx et al. reported that the macrolides azithromycin, clarithromycin, and roxithromycin, as well as trimethoprim and clindamycin, show low (<20%) or no removal at STPs in Germany.⁵⁸⁾ Giebultowicz et al. also showed that azithromycin, clindamycin, lincomycin, roxithromycin, and thiabendazole were associated with negative removal rates at STPs in Poland.⁵⁹⁾ Negative removal rates of several sulfonamides, macrolides, and fluoroquinolones were observed in more than half of the STPs in China.⁶⁰⁾ Erythromycin and clarithromycin were not removed at STPs in Taiwan.⁶¹⁾ In these previous studies on the behavior of antibiotics in STPs, the removal rates of macrolide antibiotics erythromycin, clarithromycin, and roxithromycin were lower than those of other antibiotics at STPs employing activated sludge. In the present study, the mean removal rates of macrolide antibiotics were relatively low: clarithromycin (20%) > roxithromycin (−22%) > erythromycin (−204%). The causes of negative or low removal efficacy are considered to be: (1) metabolites of antibiotics, such as glucuronide conjugates excreted from humans, that undergo hydrolysis during the treatment process and revert to the parent compound, and (2) macrolide antibiotics, which were adsorbed on suspended solids, biofilms, and bottom sediments in sewage and would be desorbed and introduced into the effluent.^59,60) For example, 80% of the administered dose of lorazepam is excreted as glucuronide conjugate,⁶²⁾ which might be hydrolyzed to lorazepam during the treatment process at STPs.

The treatment processes of the three STPs in this study were AS and partially A2O, which is typical in Japan. The removal rate of each API varied depending on the STP, therefore, refining the PEC would be difficult.

Estimation of PEC from the Three ERA Methods

To date, several ERA systems for APIs have been applied to protect the environment from the potential adverse effects of APIs. In the present study, the suitability of three systems from Japan, the EU, and the U.S.A. were investigated for the early stage of each system, that is, phase I or the EIC, with previously approved APIs in Japan. To avoid unacceptable environmental risks, API PECs in ERA systems should not underestimate MECs.

According to phase I of the MHLW ERA guidance, if PECjapan is 10 ng/L or greater, or logPow is 3.5 or greater, an environmental assessment should be conducted using a ready biodegradability test (OECD301), algae (OECD201), Daphnia (OECD202), and fish (OECD203).²²⁾ In this study, there was no API, except for ketoprofen, for which PECjapan was lower than 10 ng/L and MECmax was 10 ng/L or greater. Almost all APIs, except for ketoprofen and lorazepam, also had median MECs that were lower than those of the respective PECjapan. Therefore, PECjapan values in phase I are considered to be reasonable.

When utilizing the default values of the ERA systems, some APIs had the maximum MECs that were greater than PECs. The numbers were the highest when the guidelines were in the following order: MHLW > USFDA > EMA. The PECs of sulpiride, epinastine, lorazepam, ketoprofen, and piperacillin were lower than those of the MECs, even if the DF was set to 2 in MHLW’s ERA system. The causes of underestimation might be as follows: epinastine and ketoprofen were detected in all the river water samples, except for the latter in River No. 2, these drugs were used all over Japan as prescription and OTC medications, although their OTC production was not included in the calculation. Lorazepam was detected only in River No. 1, 3, and 4. Thus, its usage might not have occurred throughout Japan in proportion to the population. In small and mid-sized rivers sampled in Germany, the overestimation of carbamazepine could be due to the local situation.⁴³⁾ In contrast, there were no underestimated APIs with DF = 2 using the EMA method. The PEC and MEC of APIs calculated with the EMA system in other countries showed that the default DF value of 10 could be lower than the real DF = 3 in the fields in France.⁶³⁾ The PECs of some APIs were lower than their MECs in Italy,⁶⁴⁾ and the API PECs of large rivers were adequate but those of small rivers were underestimated in the U.K.⁶⁵⁾ There was no correlation between DF and MEC in a river in Germany.⁴³⁾ API PECs were not accurate due to incorrect API production values in Spain,⁶⁶⁾ some APIs showed PEC > MEC of more than ±10-fold in Portugal,⁶⁷⁾ and among the 16 selected APIs, 3 APIs, carbamazepine, atenolol, and pravastatin, provided adequate PECs in France.³⁵⁾ These results show that the situation in and around river basins affects the API PECs. For the development of new APIs, Japan’s ERA system considers seasonal changes in API usage. The use of epinastine, crotamiton, and DEET in this study and oseltamivir in a previous study⁵⁵⁾ were found to vary with the seasons. Kasprzyk-Hordern et al.⁵⁶⁾ reported that API MECs in river water in the U.K. were affected by rainfall. In Japan, the DF was found to decrease to approximately 2 from winter to spring in certain urban rivers owing to reduced rainfall. Therefore, the DF should therefore be set at 2 as the worst-case scenario.

Based on the above, the following points should be considered when calculating the PECjapan of new APIs: (1) MECmedian and MECmax may exceed 10 ng/L even if PECjapan is less than 10 ng/L if production volume is underestimated, as in the case of ketoprofen; (2) in the case of APIs with regional bias, such as lorazepam, MECmax may be close to 10 ng/L even if PECjapan is less than 1 ng/L; and (3) some APIs may have an MEC to PEC ratio of more than 1 if the DF of river water is set to a value larger than the actual DF. Further, for the ERA of previously approved APIs, the development of a database of accurate production values for APIs is desired. Azuma et al.⁶⁸⁾ reported that the calculation of exact PECs could be obtained using the shipping amount, rather than the sales amount. Sato et al.²³⁾ suggested that prescription values from NDB were suitable for reflecting the usage of APIs. If a database of OTC medication use is established, it could help calculate PECs more accurately.

Conclusion

We examined the validity of the ERA of APIs for human use in Japan. PECjapan in phase I is considered to be adequate. The conflicting results of PEC versus MECmax values indicated that the shortcomings in the models that PECs were the right DF and revealed the consumption patterns of APIs at a local scale. The MECs of most APIs that have already been approved in urban rivers were lower than their PECs when the DF was set to the actual DF in the field. In Japan, the DF decreases to approximately 2 from winter to spring in some urban rivers owing to reduced rainfall. Therefore, the DF shoul be set at 2 as the worst-case scenario. Several issues were observed in this study. Although each of the four seasons was investigated, the results did not support the seasonal variation factor introduced in the MHLW guidance. Future research should focus on this aspect. Accurate production (consumption) data, including OTCs, are necessary to calculate the PEC of drugs already on the market.

Experimental

Reagents and Chemicals

Fifty-one APIs (included in Fig. 1 and Supplementary Table S1), the artificial sweetener sucralose, and other solvents were purchased from FUJIFILM Wako Pure Chemical Corporation (Osaka, Japan), except for the deuterium-labeled carbamazepine (carbamazepine-d₁₀), which was purchased from CDN isotopes (Pointe-Claire, Quebec, Canada).

Sampling Sites and Collection of Water Samples

Twenty-eight river water samples were collected from seven urban rivers in Japan during each of the four seasons: autumn (September to November 2015), winter (December 2015 to February 2016), spring (March to May 2016), and summer (June to August 2016). The rivers are labeled from No. 1 to 7 in Table 1. Sampling was performed at sites downstream from STPs. To investigate the seasonal changes in the loading and removal rates of APIs in STPs, influent and effluent samples from three STPs in urban areas in Japan were collected as 24-h flow-proportional composite samples, once per season from February 2017 to April 2018 (Table 2). The sewage samples were collected and stored at 4 °C using refrigerated automatic samplers equipped with multiple containers (Model 6700FR, ISCO, Lincoln, NE, U.S.A., or LYSAM-R, NKS, Osaka, Japan), while maintaining a constant time interval and then manually compositing the individual sample portions proportional to the flow.

The river and sewage samples were collected in 1 L amber glass bottles, which were pre-cleaned with 50 mL acetone and transported at 4 °C to our laboratory. The water samples were stored at 4 °C in a refrigerator, and extraction of the samples by SPE was performed within 48 h after collection. The final solutions from the SPE were stored at −20 °C in a freezer, and the APIs were analyzed by LC-MS/MS within 2 weeks after SPE.

Sample Preparation

Each 500-mL sample of river water was acidified to pH 3–4 with formic acid and passed through tandem SPE cartridges. The first cartridge was a Sep-Pak PS-2 Plus (300 mg, 80 µm, Waters, Palo Alto, CA, U.S.A.), and the second cartridge was an OASIS HLB Plus (225 mg, 60 µm, Waters). These were pre-washed with 5 mL of acetonitrile (CH₃CN), followed by 5 mL of water at a flow rate of 20 mL/min. In the case of the STP influent (100 mL) and effluent (500 mL) samples, solids were separated using a glass filter (45 mm I.D., pore size 0.45 µm. Millipore, Billerica, MA, U.S.A.) prior to SPE. The filter was sonicated for 5 min in 5 mL of methanol. The methanol solution was added to the filtrate, and then the sample was subjected to SPE. The two cartridges were then air-dried for 30 min. The analytes were eluted from the tandem SPE cartridges by back-flushing using 5 mL of CH₃CN. The CH₃CN solution eluted from the SPE cartridges was concentrated to approx. 100 µL under a stream of nitrogen at 40 °C, and then the final solution was increased to a volume of 500 µL using purified water. A 5-µL portion of a 10 mg/L carbamazepine-d₁₀ solution was added to the final solution as an internal standard.

Analysis of APIs by LC-MS/MS

The analytes were measured under the following conditions: LC model, AQUITY UPLC (Waters): L-column 2 ODS, 2.0 ×100 mm, particle size 3 µm (Chemicals Evaluation and Research Institute, Tokyo, Japan), and column temperature 40 °C. A gradient elution program was achieved using a mixed-solvent system of 0.1% formic acid (v/v) in water (A) and 0.1% formic acid in CH₃CN (B) at a flow rate of 0.2 mL/min under a program of 0.0–2.0 min (95% B), 2.0–3.0 min (80% B), 3.0–18.0 min (0% B), 18.0–23.0 min (0% B), and 23.01 min (95% B) to control the column. The LC system was coupled with an MS model, Xevo TQD (Waters), equipped with an electrospray ionization source and interface, furthermore, the conditions were as follows: ion source temperature 150 °C; desolvation temperature 550 °C; ionization mode, electron spray ionization (ESI); capillary voltage 3.0 kV; cone gas flow, 85 L/h; desolvation gas flow, 500 L/h; and detector voltage, 650 V. The ESI mode, cone voltage, collision energy, precursor ion, and product ion of each analyte are listed in Supplemental Table S1. Under these LC-MS/MS conditions, the instrumental detection limit and limit of quantification (LOQ) of each analyte were measured by 10 analyses of a 100 µg/L standard solution. The method detection limit (MDL) of APIs was calculated according to USEPA guidance.⁶⁹⁾ The MDL from the USEPA guidance and LOQ were compared, and the larger of the two was used as the MDL in this study. The results are presented in Supplementary Table S1.

Calculation of API PECs

The API PEC in local surface water (PECjapan, µg/L) was calculated according to Japan’s guidance²²⁾ as follows:

  

where PRODUCT is the production per year (kg/year), Fseason is a factor for seasonal change from 1 to 4, with a default of 1; Population is the population of Japan (person); WEST is the amount of wastewater per person per day (350 L/person/d), DILUTION is the dilution factor (DF, 10), Day is the number of days per year (365 d/year), and CF is the conversion factor (10⁹ µg/kg).

The API PEC in local surface water (PECema, µg/L) was calculated according to EMA²⁰⁾ as follows:

  

where DOSEai is the maximum daily dose consumed per patient (mg/person/d), Fpen is the fraction of market penetration (0.01), WESTinhab is the amount of wastewater per inhabitant per day (200 L/person/d), and DILUTION is the DF (10).

The API PEC (EIC-Aquatic, µg/L) released into the aquatic environment was calculated according to the USFDA guidelines¹⁹⁾ as follows:

  

where A is the production for direct use as an active moiety (kg/year), B is the amount entering the STP (1/L/d, in the case of Japan: 4.2 × 10¹⁰/L/d, Water Resource Department, Ministry of Land, Infrastructure, Transport and Tourism, 2014,⁷⁰⁾ C is the number of days per year (year/365 d), and D is the conversion factor (10⁹ µg/kg).

Acknowledgments

The authors are thankful to the staff of the sewage treatment plants for their help in sample collection. The present research was supported in part by a Health and Labor Sciences Research Grant, (No. 17mk0101022h0003), from the Japan Agency for Medical Research and Development (AMED).

Conflict of Interest

The authors declare no conflict of interest.

Supplementary Materials

The online version of this article contains supplementary materials.

References

• 1) Daughton C., Ternes T. A., Environ. Health Perspect., 107 (Suppl. 6), 907–938 (1999).
• 2) Kolpin D. W., Furlong E. T., Meyer M. T., Thurman E. M., Zaugg S. D., Barber L. B., Buxton H. T., Environ. Sci. Technol., 36, 1202–1211 (2002).
• 3) Fang W., Peng Y., Muir D., Lin J., Zhang X., Environ. Int., 131, 104994 (2019).
• 4) Beek T., Weber F.-A., Bergmann A., Hickmann S., Ebert I., Hein A., Küster A., Environ. Toxicol. Chem., 35, 823–835 (2015).
• 5) Ebele A. J., Abdallah M. A.-E., Harrad S., Emerging Contam., 3, 1–16 (2017).
• 6) WHO, “Pharmaceuticals in drinking-water,” Number of pages: 52, ISBN: 978 9241502085, 2012.
• 7) Benotti M. J., Trenholm R. A., Vanderford B. J., Holady J. C., Stanford B. D., Snyder S. A., Environ. Sci. Technol., 43, 597–603 (2009).
• 8) Furlong E. T., Batt A. L., Glassmeyer S. T., Noriega M. C., Kolpin D. W., Mash H., Schenck K. M., Sci. Total Environ., 579, 169–1642 (2020).
• 9) Kibuya F. A., Gall H. E., Elkin K. R., Swistick B., Veith T. L., Watson J. E., Elliott H. A., J. Environ. Qual., 48, 1057–1066 (2019).
• 10) Ben Y., Hu M., Zhang X., Wu S., Wong M. H., Wang M., Andrew C. B., Zheng C., Water Res., 175, 115699 (2020).
• 11) Simazaki D., Kubota R., Suzuki T., Akiba M., Nishimura T., Kunikane S., Water Res., 76, 187–200 (2015).
• 12) Kidd K. A., Blanchfield P. J., Mills K. H., Palace V. P., Evans R. E., Lazorchak J. M., Flick R. W., Proc. Natl. Acad. Sci. U.S.A., 104, 8897–8901 (2007).
• 13) Watanabe H., Tamura I., Abe R., Takeuchi H., Nakamura A., Suzuki T., Hirose A., Nishimura T., Tatarazako N., Environ. Toxicol. Chem., 35, 996–1006 (2016).
• 14) Cunningham V. L., Constable D. J. C., Hannah R. E., Environ. Sci. Technol., 38, 3351–3359 (2004).
• 15) Bamann M., Weiss K., Maletzki D., Schüssler W., Schudoma D., Kopf W., Kühnen U., Chemosphere, 120, 192–198 (2015).
• 16) Ghosh G. C., Nakada N., Yamashita N., Tanaka H., Environ. Health Perspect., 118, 103–107 (2010).
• 17) Ågerstrand M., Berg C., Björlenius B., Breitholtz M., Brunström B., Fick J., Gunnarsson L., Larsson D. G. J., Sumpter J. P., Tysklind M., Rudén C., Environ. Sci. Technol., 49, 5336–5345 (2015).
• 18) Lee D., Choi K., Sci. Total Environ., 671, 1026–1035 (2019).
• 19) U.S. FDA, CDER/CBERCMC 6, Revision 1, “Guidance for Industry Environmental Assessment of Human Drug and Biologics Applications,” 1998.
• 20) EMEA/CHMP, “Guideline on the Environmental Risk Assessment of Medicinal Products for Human Use,” 2006.
• 21) International Conference on Chemicals Management, ICCM Fourth session, Geneva, 28 September–2 October 2015. Annotations to the Provisional Agenda, 2015.
• 22) Ministry of Health, Labor and Welfare in Japan, “Guidance for environmental risk assessment for developing new drug,” 30 March, Yakuseishinsahatsu 0330-1gou. 2016.
• 23) Sato K., Watanabe H., Ikeda T., Miida H., Ohtaki K., Watanabe K., Regul. Toxicol. Pharmacol., 99, 98–104 (2018).
• 24) Ministry of Health, Labor and Welfare in Japan, “Current Survey of Production Concerning Pharmaceutical Industry (Fundamental Statistics),” 2013.
• 25) Kümmerer K., Henninger A., Clin. Microbiol. Infect., 9, 1203–1214 (2003).
• 26) Brain R. A., Solomon B. K., Brooks B. W., “Targets, Effects and Risks in Aquatic Plants Exposed to Veterinary Antibiotics,” ed. by Henderson K.L., Coats J.R., American Chemical Society, Washington, DC, 2009, pp. 169–189.
• 27) National Veterinary Assay Laboratory, Ministry of Agriculture, Forestry and Fisheries in Japan, “Annual Report of Sales Amount and Sales Volume of Veterinary Drugs, Quasi-drugs and Medical Devices,” 2013.
• 28) Lindim C., van Gils J., Georgieva D., Mekenya O., Cousins I. T., Sci. Total Environ., 572, 508–519 (2016).
• 29) Oppenheimer J., Eaton A., Adruzzaman M., Aghani A., Water Res., 45, 4019–4027 (2011).
• 30) Nakada N., Kiri K., Shinohara H., Harada A., Kuroda K., Takizawa S., Takada H., Environ. Sci. Technol., 42, 6347–6353 (2008).
• 31) Bayer A., Asner R., Schüssler W., Kopf W., Wei K., Sengl M., Letzel M., Environ. Sci. Pollut. Res. Int., 21, 10830–10839 (2014).
• 32) Letzel T., Byer A., Schulz W., Heermann A., Lucke T., Greco G., Grosse S., Schussler W., Sengl M., Letzel M., Chemosphere, 137, 198–206 (2015).
• 33) Laeis P., Püchler K., Kirch W., J. Hypertens., 19 (Suppl.), S21–S32 (2001).
• 34) See S., Stirling A. L., Am. J. Health Syst. Pharm., 57, 739–746 (2000).
• 35) Celle-Jeanton H., Schemberg D., Mohammed N., Huneau F., Bertrand G., Lavastre V., Le Coustumer P., Environ. Int., 73, 10–21 (2014).
• 36) Castiglioni S., Bagnati B., Fenelli R., Pomati F., Calamari D., Zuccato E., Environ. Sci. Technol., 40, 357–363 (2006).
• 37) Nakada N., Komori K., Suzuki Y., Konishi C., Houwa I., Tanaka H., Water Sci. Technol., 56, 133–140 (2007).
• 38) Hughes S. R., Kay P., Brown L. E., Environ. Sci. Technol., 47, 661–677 (2013).
• 39) Yang L., He J. T., Su S. H., Cui Y. F., Huang D. L., Wang G. C., Environ. Sci. Pollut. Res. Int., 24, 15838–15851 (2017).
• 40) Carmona E., Andreu V., Picó Y., Sci. Total Environ., 484, 53–63 (2014).
• 41) Carballa M., Omi F., Lema J. M., Chemosphere, 72, 1118–1123 (2008).
• 42) Yoon Y., Ryu J., Oh J., Choi B.-G., Snyder S. A., Sci. Total Environ., 408, 636–643 (2010).
• 43) Meyer W., Peich M., Beie S., Behrendt J., Gulyas H., Otterpohl R., Environ. Monit. Assess., 188, 487 (2016).
• 44) Dai G., Wang B., Fu C., Dong R., Huang J., Deng S., Wang Y., Yu G., Environ. Sci. Process Impacts, 18, 445–455 (2016).
• 45) Ma R., Wang B., Yin L., Zhang Y., Deng S., Huang J., Wang Y., Yu G., J. Hazard. Mater., 323 (Pt A), 147–155 (2017).
• 46) Björlenius B., Ripaszám M., Haglund P., Lindberg R. H., Tysklind M., Fick J., Sci. Total Environ., 633, 1496–1509 (2018).
• 47) Oĝuz M., Mihçiokur H., Environ. Toxicol. Pharmacol., 38, 79–83 (2014).
• 48) Jones O. A. H., Voulvoulis N., Lester J. N., Water Res., 36, 5013–5022 (2002).
• 49) Grung M., Källvist T., Sakashaug S., Skurtveit S., Thomas K. V., Ecotoxicol. Environ. Saf., 71, 328–340 (2008).
• 50) Vieno N. M., Tuhkanen T., Kronberg L., Environ. Sci. Technol., 39, 8220–8226 (2005).
• 51) Roberts J., Kumar A., Du J., Hepplewhite C., Ellis D., Christy A. G., Beavis S. G., Sci. Total Environ., 541, 1625–1637 (2016).
• 52) Papageorgiou M., Kosma C., Lambropoulou D., Sci. Total Environ., 543 (Pt A), 547–569 (2016).
• 53) Horiguti S., Saito Y., Bull. Tokyo Med. Dent. Univ., 14, 141–155 (1967).
• 54) Tsuda Y., Me. Entomol. Zool., 63, 95–101 (2012).
• 55) Azuma T., Nakada N., Yamashita N., Tanaka H., Environ. Sci. Technol., 46, 12873–12881 (2012).
• 56) Kasprzyk-Hordern B., Dinsdale R. M., Guwy A. J., Water Res., 42, 3498–3518 (2008).
• 57) Onesios K. M., Yu J. T., Bouwer E. J., Biodegradation, 20, 441–466 (2009).
• 58) Marx C., Günther N., Schubert S., Oertel R., Ahnert M., Krebs P., Kuehn V., Sci. Total Environ., 538, 779–788 (2015).
• 59) Giebultowicz J., Nałęcz-Jawecki G., Harnisz M., Kucharski D., Korzeniewska E., Płaza G., Molecules, 25, 1470 (2020).
• 60) Zhang X., Zhao H., Du J., Qu Y., Shen C., Tan F., Chen J., Quan X., Environ. Sci. Pollut. Res. Int., 24, 16478–16487 (2017).
• 61) Lin A. Y.-C., Lin C.-F., Tsai Y.-T., Lin H. H.-H., Chen J., Wang X.-H., Yu T.-H., Water Sci. Technol., 62, 2450–2458 (2010).
• 62) Elliott H. W., Br. J. Anaesth., 48, 1017–1023 (1976).
• 63) Coetsier C. M., Spinelli S., Lin L., Roig B., Touraud E., Environ. Int., 35, 787–792 (2009).
• 64) Castiglioni S., Fanelli R., Calamari D., Bagnati R., Zuccato E., Regul. Toxicol. Pharmacol., 39, 25–32 (2004).
• 65) Burns E. E., Carter L. J., Kolpin D. W., Thomas-Oates J., Boxall A. B. A., Water Res., 137, 72–85 (2018).
• 66) De Garcia S. O., Pinto G. P., Encina P. G., Mata R. I., Sci. Total Environ., 444, 451–465 (2013).
• 67) Pereira A., Silva L. J. G., Lino C. M., Meisel L. M., Pena A., Sci. Total Environ., 603–604, 226–236 (2017).
• 68) Azuma T., Nakada N., Yamashita N., Tanaka H., Chemosphere, 138, 770–776 (2015).
• 69) U.S.EPA, “Definition and Procedure for the Determination of the Method Detection Limit,” Revision 2, EPA 821-R-16-006, 2016.
• 70) “Water in Japan,” Water Resource Department, Ministry of Land, Infrastructure, Transport and Tourism, Chiyoda-ku, Tokyo, 2014, p. 4.