Journal of Antimicrobial Chemotherapy (2000) 46, Suppl. S1, 53–58

Introduction

By their nature, medicinal products, such as antibiotics, arebiologically active molecules. Antibiotics are specificallydesigned to control pathogenic bacteria in animals andhumans, but very little is known about their ecotoxicology.After administration of antibiotics to humans, a significantamount is excreted into waste water. Surplus antibiotics arealso sometimes disposed of into household drains. Anti-biotics therefore enter municipal sewage and sewage-treatment plants (STPs). If the drugs are not completelymineralized in the STP, they are released into surface wateror sludge and, if the sludge is used to fertilize arable land,they may enter the topsoil of fields.1,2

Medicinal products are licensed for use by regulatoryauthorities if they comply with scientific criteria on quality,efficacy and safety. The authorities consider safety to theconsumer and to the individuals handling the product dur-ing treatment. In addition to these criteria, the environ-mental risk has recently become a matter of increasingpublic scrutiny and legal requirements. The risk of effectson the environment has led to regulation of new pharma-ceuticals in the USA,3 and a draft on the environmental riskassessment (ERA) of new pharmaceuticals is also pro-

posed in the EU.4 There are, however, no regulations orrequirements concerning the environmental properties or potential effects of existing pharmaceuticals. Conse-quently such data are scarce.

An important step in ERA, particularly for drugs, is acost–benefit analysis. It is necessary to know the risk to theenvironment and the indirect effects on the human body asa result of contamination of the environment to be able todecide whether the benefits exceed the costs. If two ormore drugs have the same therapeutic effect, ERA can alsobe used to determine which has the lower environmentalimpact.

The aim of this paper is to provide data on consumption,estimated environmental concentration and the effects on the environment of three antibiotics widely used for urinary tract infections (mecillinam, trimethoprim andciprofloxacin), enabling a preliminary comparison of theirenvironmental hazard. The structures of these three com-pounds are shown in the Figure, and relevant physico-chemical characteristics in Table I. The spread of bacterialantibiotic resistance genes is another pertinent effect ofantibiotics on the environment. Since this is difficult toquantify and not included in the ERA assessment of anti-biotics, it will not be discussed further in this paper.

53

Environmental risk assessment of antibiotics: comparison ofmecillinam, trimethoprim and ciprofloxacin

B. Halling-Sørensen*, H.-C. Holten Lützhøft, H. R. Andersen and F. Ingerslev

Section of Environmental Chemistry, Department of Analytical and Pharmaceutical Chemistry, Royal Danish School of Pharmacy, Universitetsparken 2, DK-2100 Copenhagen 0, Denmark

The effects of mecillinam, trimethoprim and ciprofloxacin, antibiotics used in the treatment ofurinary tract infections, on the aquatic environment were assessed. Mecillinam and cipro-floxacin were both readily biodegradable (primary degradation) in activated sludge, whereastrimethoprim persisted. The toxicity of these antibiotics towards sludge bacteria, a green alga,a cyanobacterium, a crustacean and a fish were investigated; both mecillinam and cipro-floxacin were highly toxic to the cyanobacterium Microcystis aeruginosa (EC50 in the range5–60 µg/L). Risk characterization for the aquatic environment was performed for the three com-pounds by calculating the predicted environmental concentration (PEC) and the predicted no-effects concentration (PNEC). A PEC/PNEC ratio of <1 indicates that, with the present patternof use, no environmental risk is expected. PEC/PNEC ratios of <1 for present usage in Europewere found for mecillinam and trimethoprim whereas a PEC/PNEC ratio >1 was found forciprofloxacin.

*Corresponding author. Tel: � 45-35-30-64-53; Fax: � 45-35-30-60-13; E-mail: bhs@dfh.dk

© 2000 The British Society for Antimicrobial Chemotherapy

JAC

B. Halling-Sørensen et al.

Materials and methods

Chemical substances

Test substances were purchased from their manufacturersand stored at 2°C for �24 h. Mecillinam [(6R)-6-(per-hydroazepin-1-ylmethyleneamino)-penicillanic acid, batchno. V27 (99.3% pure), CAS 32887-01-7] was provided by LeoPharmaceutical Products, Ballerup, Denmark; trimetho-prim [2,4-diamino-5-(3,4,5-trimethoxybenzyl)-pyrimidine;batch no. 211962, CAS 738-70-5] by Nomeco, Copenhagen,Denmark; ciprofloxacin–HCl (1-cyclopropyl-6-fluoro-1,4-dihydro-4-oxo-7-piperazin-1-ylquinoline-3-carboxylic acidhydrochloride; batch no. 303477A, CAS 85721-33-1) byBayer A/G, Wuppertal, Germany. Potassium dichromate

(�99.8% pure), 2,3-dichlorophenol, sodium acetate andaniline (analytical grade; Merck, Copenhagen, Denmark)were used as reference compounds in the toxicity studiesand biodegradation tests.

Biodegradation tests

Biodegradation tests were performed in two different testsystems:

(i) Oxytop respirometers (WTW, Weilheim, Germany)were used to measure biological oxygen demand (BOD)according to the guidelines published by the Organizationfor Economic Cooperation and Development (OECD).5

The Oxytop system is a screening system that only evalu-ates whether the compound is readily degradable. Sodiumacetate was used as a reference compound. All compoundswere tested in duplicate experiments with added concen-trations corresponding to theoretical oxygen demands(TODs) of 30 mg/L. The TOD of 30 mg/L was calculated tocorrespond to mecillinam 16.5 mg/L, trimethoprim 19.4mg/L and ciprofloxacin 17.5 mg/L.

(ii) Biodegradation experiments with activated sludge inreactors may be used to determine the half-life of the chem-ical during biodegradation in the activated sludge processin a waste water system. Primary biodegradation experi-ments were conducted in aerated batch reactors5 with 1 Lof liquid (sludge). Experiments were conducted at 22°Cand sludge was preconditioned at room temperature.Sludge was collected from the Lundtofte rensningsanlægsewage treatment plant, Lyngby, Denmark and maintainedaerated until use, with no addition of substrate. Mostsludge was used freshly collected (aeration time 1–2 h) orwas preconditioned by aeration for 1 day. Laboratory reac-tors were fed with sludge (1 L) and mecillinam or trimetho-prim 500 �g/L or ciprofloxacin 250 �g/L.

The biodegradation of the chemicals in the reactors wasfollowed over time by HPLC assay. The half-life (t½) indays was calculated for the primary degradation of thethree compounds based on first-order kinetics. These half-lives were applied in the risk assessment described below.Detailed description and validation of the HPLC methodsused are available from the authors. All three antibioticswere tested in duplicate reactors. Aniline was used as a reference compound.

54

Figure. Chemical structures of (a) mecillinam, (b) trimethoprimand (c) ciprofloxacin.

Table I. Physicochemical data

Solubility log KOW

Antibiotic Compound group (g/L) Syracusea log KOWb

Mecillinam �-lactam antibiotic �2 not found 0.79 � 0.38Trimethoprim dihydrofolate reductase inhibitor �0.3–0.4 0.91 1.33 � 0.75Ciprofloxacin quinolone �2 0.28 1.24 � 0.86

aFrom http://esc.syrres.com/interkow/estsoft.htmbEstimated with ACD Log P software (Advanced Chemical Development, Toronto, Canada).

Environmental risk assessment of antibiotics

Ecotoxicology

Toxicity towards the following organisms was tested: activ-ated sludge bacteria; a green alga, Selenastrum capri-cornutum (eukaroytic); a cyanobacterium, Microcystisaeruginosa (prokaryotic); the zooplankton Daphnia magna;and the zebrafish, Brachydanio rerio. All toxicity tests wereperformed in accordance with OECD or InternationalOrganization of Standardization (ISO) guidelines.5–9

Risk assessment

Environmental risk assessment (ERA) was based on theEU draft guideline document for medicinal products forhuman use4 and was conducted in two phases. For mostmedicinal products, it is anticipated that only the first phaseof the evaluation will be necessary. Phase II of the ERA,which involves a comprehensive assessment of ecotoxi-cology, is considered when the predicted environmentalconcentration (PEC) exceeds a threshold value of 0.001 �g/L in surface water and 10 �g/kg soil. The phase IIassessment involves calculation of a risk quotient betweenthe predicted (PEC) or measured environmental concentra-tion and the predicted no-effect concentration (PNEC),using worst-case assumptions.

For the aquatic compartment, it was assumed that allantibiotics are used in the year they are purchased; that theantibiotics are released into the environment withoutdegradation in man or the sewage system; and that the usepattern is evenly distributed temporally and spatially. Theconcentration in water of the active substance (PECw) wascalculated from the following equation:4

PECw �A � (100 R)

(1)365 � P � V � D � 100

where A is the amount used (kg/year), R is the percentageremoval (set to zero when information on biodegradationin the STP is missing or when worst-case conditions areassessed), P is the population of Europe (383 million in1999), V is the volume of waste water per capita per day(0.2 m3) and D is the dilution factor in the environment (adefault factor of 10 was used). A hydraulic retention time(HRT) of 10 h was used in the PECw calculations, incor-porating the degradation.

For compounds sorbing to sludge, the concentration insludge can be calculated if the partition coefficient betweensludge and water, Kd, is measured or estimated based onthe octanol–water partition coefficient, Kow, and the frac-tion of organic carbon in sludge (foc � 0.35), and calculatedas follows:

Kd � foc � 0.41 � Kow (2)

Experimental Kds were obtained in sorption experimentswith sludge (H. C. Holten-Lützhøft and B. Halling-Sørensen, unpublished results). Kows were from the litera-ture or calculated with the ACD log P software (Advanced

Chemical Development, Toronto, Canada). A Kow cor-rected for the effect of ionized groups on Kow-dependentdistributions (Dow) can be applied for the Kd calculation(Equation 3).

Dow �Kow (3)

1 � 10pH–pKa

The calculation of the predicted concentration in sludge(Csludge) is based on no degradation or other loss of thedrugs. The concentration of the antibiotics in sludge can becalculated as:

Csludge �Mmedical compound (4)

Vw/Kd � Msludge

where Kd based on experimental Kd, Dow or Kow, respec-tively, is inserted for Kd. The total annual volume of wastewater in Denmark (Vw) is 3.8 � 108 m3. For Europe Vw canbe calculated as P � V � 365, i.e. 383 million � 0.2 m3 � 365� 2.8 � 1010 m3. For the mass of sludge, Msludge, a value of151 159 tonnes in 1997 was used for Denmark. Mmedicinal

compound, the annual loading of antibiotic to the environmen-tal compartment, is often assumed to equal consumptionbecause of lack of information.

Results and discussion

Physicochemical properties of the chemicals

n-Octanol–water partition coefficients were obtained fromliterature using the Syracuse database at http://esc.syrres.com/ and calculated using ACD log P software. No pub-lished value was found for mecillinam, but agreementbetween calculated and literature values for the two othersubstances suggests that the range of the calculated value for mecillinam is appropriate. Based on the available n-octanol–water partition coefficients, it is consideredappropriate to use static exposure systems in the toxicitytests.

No values for the vapour pressure of any of the threesubstances were found in the literature. However, themolar weights (�200 g/mol) and the presence of severalpolar groups in each of the three molecules suggests thatvapour pressure as well as air–water partition coefficientswill be small. Therefore no special precautions wereemployed in the test procedure to handle loss of test sub-stance by evaporation.

Degradation test

None of the compounds was found to sorb strongly tosludge. The partition coefficients, Kd, between water andsludge for mecillinam, trimethoprim and ciprofloxacinwere found to be 55, 76 and 417, respectively.

55

B. Halling-Sørensen et al.

Table II shows the results of the biodegradation studiesperformed with the Oxytop system and the activatedsludge reactors. Biodegradation in the Oxytop system wasvery poor: all three compounds persisted for the entire 28 days of the test period. Using this system, which is con-sidered an initial screening system, all three compoundswere therefore assessed as being not readily degradable.

The next step was assessment of the primary degrada-tion of the compounds in sludge reactors. Results per-formed with an initial concentration of 500 �g/L ofmecillinam showed that biodegradation occurred rapidly inthis test system. The estimated t½ (assuming first-orderdegradation) was 0.5 days for one sludge column and 0.7days for the other. �-Lactams such as mecillinam are alsosubject to abiotic degradation such as hydrolysis. This mayspeed up degradation in the STP.

Biodegradation studies performed with trimethoprim500 �g/L in the reactors showed that degradation had notreached 50% at day 25. Trimethoprim persisted in the tworeactors with a t½ of 22 days and 41 days, respectively.Ciprofloxacin 250 �g/L was degraded quickly until 50%had been degraded. Its t½ was 2.5 days in one reactor and1.6 days in the other. This result is consistent with a similarstudy investigating the primary degradation of cipro-floxacin,12 in which it was found that the t½ for ciprofloxacinin activated sludge was between 1 and 5 days, depending

on the sludge concentration. Compared with a compoundlike pentachlorophenol, which is recognized as barelydegradable under such test circumstances [t½ (first ordermodel) about 5 days], trimethoprim is extremely persistent,whereas mecillinam and ciprofloxacin are readily degrad-able (primary degradation). Because results show thatciprofloxacin probably was not mineralized under the testconditions, more studies need to be performed on metabo-lites. This study and that of Hartmann and co-workers12

indicate that compounds resembling ciprofloxacin persistin the sludge. The reference compound aniline was degradedrapidly (t½ � 0.7 days) in the reference reactor, showing theability of the activated sludge to degrade chemical com-pounds.

Toxicity tests

Table III shows the water concentration giving 50% effect(EC50) or no-observed-effect concentration (NOEC)obtained for the reference chemicals and three tested com-pounds.

Sludge bacteria. Only ciprofloxacin showed high potencyfor the sludge bacteria. Mecillinam and trimethoprim wereharmless to the activated sludge bacteria.

56

Table II. Biodegradation data obtained with the screening test11 (data are expressed as readily degradable orpersistent) and the activated sludge reactors (data are expressed as half-lives (days))

Test Mecillinam Trimethoprim Ciprofloxacin Sodium acetate Aniline

Screening test persistenta persistenta persistenta readily degradable NAActivated sludge reactor 0.5–0.7 22–41 1.6–2.5 NA 0.7

NA, not assessed.aPersisted for 28 day test period.

Table III. Toxicity data: EC50 with corresponding 95% confidence limits or no-observed-effect concentration (NOEC) (mg/L) for the compounds tested

EC50 (mg/L)

Species mecillinam trimethoprim ciprofloxacin

Bacteria (activated sludge)a 62.1 17.8 0.61 (0.31–1.22)S. capricornutumb NOEC � 300 110 (64–192) 2.97 (2.41–3.66)M. aeruginosa 0.06 (0.05–0.06) 112 (100–126) 0.005 (0.004–0.006)D. magna, 48 h test NOEC � 300 123 NOEC � 60B. rerio, 72 h test NOEC � 100 NOEC � 100 NOEC � 100

a3,5-Dichlorophenol was used as a reference compound in the test. EC50 was in accordance with that given in theguidance protocol, 5.8 (95% CI 3.8–8.8) mg/L.bPotassium dichromate was used as reference compound in the test. EC50 was in accordance with that given in theguidance protocol, 0.59 (95% CI 0.46–0.75) mg/L.

Environmental risk assessment of antibiotics

Green algae and cyanobacteria. Only ciprofloxacin wastoxic to S. capricornutum (EC50 � 2.97 mg/L), whereasmecillinam (NOEC � 300 mg/L) and trimethoprim (EC50

� 110 mg/L) were toxic only at environmentally irrelevantconcentrations. The highest mecillinam concentrationapplied was 300 mg/L, so this concentration is thereforedetected as the lower limit of NOEC. Both mecillinam andciprofloxacin were highly toxic to M. aeruginosa (see TableIII), with EC50s in the range 5–60 �g/L. M. aeruginosa wasgenerally much more sensitive than S. capricornutum toboth mecillinam and ciprofloxacin. This is not surprising, asM. aeruginosa is a prokaryote whereas S. capricornutumis a eukaryote. Similar results have been reported for other antibiotics.13 The EC50 of potassium dichromate was0.59 mg/L (95% CI 0.46–0.75 mg/L), within the limits indicated in the guidance protocol5 [0.53 mg/L (95% CI0.20–0.75 mg/L)].

Daphnia magna. Mecillinam was tested in a limit test7 attwo concentrations, 300 and 1000 mg/L. At these concen-trations it did not kill D. magna. Ciprofloxacin was tested ina similar limit test at 30 and 60 mg/L (at concentrations of�60 mg/L, ciprofloxacin crystallized and was, therefore, nolonger bioavailable, so it was not tested at higher concen-trations). At these concentrations, ciprofloxacin did not killthe organisms studied. Trimethoprim was tested at fiveconcentrations between 30 and 300 mg/L. Table III showsthe toxicity of the three compounds towards D. magna:trimethoprim had minor effects on D. magna.

Zebrafish. A limit test was performed in accordance withOECD guidelines8 at two concentrations (30 and 100mg/L) for all three compounds. For all three compounds,an NOEC (72 h) of 100 mg/L was found. Assuming thattoxicity fits a probit-type effect curve, this indicates a 99.9%confidence level for the true EC50 being �100 mg/L.

Environmental loading

Because data on environmental loading were not availablefor the antibiotics tested, the environmental loading wasassumed to be the greatest possible, i.e. equal to the consumption of the compounds. Data for Europe (Spain,Portugal, Italy, Greece, Germany, Switzerland, France,

Austria, Denmark, Belgium, The Netherlands, UK, Ire-land, Norway, Sweden and Finland; total population 383million) was obtained from IMS14 for the first 3 months of1999 and extrapolated to give a value for the whole of 1999.These IMS data exclude consumption at hospital level inSpain, Portugal, Greece, Switzerland and Ireland. The total environmental loading in 1999 for mecillinam, cipro-floxacin and trimethoprim was estimated as 2.0, 186.2 and47.8 tonnes/year, respectively.

Preliminary risk characterization

Risk characterization is the estimation of the incidence ofthe adverse effect occurring in an environmental compart-ment as a result of actual or predicted exposure to a sub-stance. It generally involves the integration of an effectassessment (calculating the PNEC) and an exposure assess-ment (calculating the PEC). We calculated PEC in twoways: (i) not including degradation data (worst-case scenario) (we call this PEC1) and (ii) including degradationdata (PEC2) (Table IV).

Many international regulatory frameworks oftenexpress risk as a PEC/PNEC ratio. It should be noted thatthese ratios are not absolute measures of risk.

The worst-case PEC estimations, PEC1 (R � 0 in Equation 1), for the three compounds show that (based onEuropean consumption data) none of the three compoundsexceeds the threshold values in the EU (1994) draft guide-line4 of 0.001 �g/L for surface water. With today’s patternof use, the compounds would not, therefore, require phaseII assessment. However, changes in the consumption pattern of the three compounds throughout Europe or incertain member countries could enhance the PEC andtherefore trigger a demand for more information.

The aim of this study was to compare the environmentalfate and effects of the three compounds, as part of thephase II assessment, on the aquatic environment using bacteria (activated sludge), green algae and cyanobacteria,daphnia and fish as target organisms. The terrestrial com-partment was not assessed because none of the compoundswas found to sorb strongly to activated sludge (Kd 500)and are not used in veterinary therapy. An assessment fac-tor of 100 was used in accordance with OECD guidelines,5

as acute toxicity data on algae, crustaceans and fish were

57

Table IV. Estimated PECs, PNEC (mg/L) and PEC/PNEC ratios based on European consumption in 1999

Compounds PEC1 (mg/L)a PEC2 (mg/L)b PNEC (mg/L) PEC1/PNEC ratio PEC2/PNEC ratio

Mecillinam 7.0 � 10–6 5.9 � 10–8 0.0006 0.01 9.8 � 10–3

Trimethoprim 1.7 � 10–4 1.7 � 10–4 0.18 9.5 � 10–4 9.4 � 10–4

Ciprofloxacin 6.7 � 10–4 6.3 � 10–4 0.00005 13.3 12.7

aCalculated without including biodegradation (R � 0 in Equation 1); simulates a worst-case assessment.bCalculated including biodegradation (R � (percentage degraded in the STP during the 8 h HRT) in Equation 1) and simulates a best-case assessment.

B. Halling-Sørensen et al.

available. The lowest EC50 (see Table III) obtained for thethree compounds were therefore recalculated to a PNECof 0.0006, 0.18 and 0.00005 mg/L for mecillinam, trimetho-prim and ciprofloxacin, respectively. Based on thesePNECs, the PEC/PNEC ratio was calculated.

Table IV shows PEC/PNEC calculations for the threeantibiotics at European level, excluding the degradationdata (PEC1/PNEC; worst-case scenario) and includingthese data (PEC2/PNEC). A PEC1/PNEC ratio of �1 wasfound for ciprofloxacin; the ratios for mecillinam andtrimethoprim were both 1. This is a result of the lowerconsumption of trimethoprim, even though it is assessed asthe most persistent of the three compounds. The picturedid not change much after taking the biodegradation of thecompounds into consideration because mecillinam andtrimethoprim still had PEC2/PNEC ratios 1, suggestingthat no environmental hazard would occur with today’spattern of use. The PEC2/PNEC ratio was still ��1 forciprofloxacin, identifying a possible environmental hazardwith today’s pattern of use. This study would suggest thatmore information on the fate and effects of ciprofloxacin inenvironmental matrices should be provided.

Acknowledgement

The authors gratefully acknowledge the technical assis-tance of Susanne Hermansen, The Royal Danish School ofPharmacy.

References

1. Halling-Sørensen, B., Nors Nielsen, S., Lanzky, P. F., Ingerslev,F., Holten Lützhøft, H. C. & Jørgensen, S. E. (1998). Occurence, fateand effects of pharmaceutical substances in the environment—areview. Chemosphere 36, 357–93.

2. Jørgensen, S. E. & Halling-Sørensen, B. (2000). Drugs in theenvironment. Chemosphere 40, 692–9.

3. Food and Drugs Administration. (1995). Guidance for Industryfor the Submission of an Environmental Assessment in Human DrugApplication and Supplements. CDER, Washington.

4. European Union. (1994). Assessment of Potential Risks for theEnvironment Posed by Medical Products for Human Use, ExcludingProducts Containing Live Genetically Modified Organisms. EU AdHoc Working Party, III/5504/94, draft 4. EMEA, London.

5. Organisation for Economic Cooperation and Development.(1992). Report of the OECD Workshop on the Extrapolation of Lab-oratory Aquatic Toxicity Data to the Real Environment. EnvironmentMonograph 59, OECD, Paris.

6. Organisation for Economic Cooperation and Development.(1984). Algal Growth Inhibition Test. Environment Guideline no. 201,OECD, Paris.

7. Organisation for Economic Cooperation and Development.(1984). Daphnia sp., Acute and Immobilisation Test and Reproduc-tion Test. Environment Guideline no. 202, OECD, Paris.

8. Organisation for Economic Cooperation and Development.(1992). Fish, Acute Toxicity Test. Environment Guideline no. 203,OECD, Paris.

9. International Organization of Standardization. (1998). WaterQuality—Determination of the Inhibitory Effect of Water Constituentson the Growth of Activated Sludge Microorganisms. ISO 15522. ISO,Geneva.

10. Nyholm, N., Ingerslev, F., Berg, U. T., Pedersen, J. P. & Frimer-Larsen, H. (1996). Estimation of kinetic rate constant for biodegra-dation of chemicals in activated sludge waste water treatment plantsusing short term batch experiments and microgram/L range spikedconcentrations. Chemosphere 33, 851–64.

11. Organisation for Economic Cooperation and Development.(1998). OECD Guidelines for Testing of Chemicals. OECD, Paris.

12. Hartmann, A., Alder, A. C., Koller,T. & Widmar, R. M. (1998).Bioavailability and elimination of the fluoroquinolone antibioticciprofloxacin in wastewaters using a modified OECD 302B assay. InAbstracts of the Nineteenth ETAC Annual Meeting, Charlotte, NC,Poster PHA101, p. 275. SETAC, Brussels, Belgium.

13. Halling-Sørensen, B. (2000). Algal toxicity of antibacterialagents used in intensive farming. Chemosphere 40, 731–9.

14. Anon. (1999). IMS Health Q3/1999. Maxims Database, London.

58