The detection of human and veterinary pharmaceutical compounds in surface water has received considerable attention in recent years. The improved precision and accuracy of analytical methods for trace organic chemicals, which include pharmaceutical products and many other types of consumer products, has led to concerns about potential exposure to humans, animals and aquatic life. Humans and companion animals, for example, could potentially be exposed to these chemicals via drinking water and aquatic biota could be exposed in surface waters. Most reporting in the popular press and some scientific papers has focused on the detection of trace concentrations of pharmaceuticals.^1-6

It should be noted that like any chemical product following disposal and excretion, low concentrations of pharmaceutical compounds are expected to be detectable in water. Equilibrium chemistry dictates that all chemicals used will be distributed in the environment. This means that even compounds that are highly degradable may be detected in the environment in very small concentrations given adequate analytical methods. Therefore, it is important to evaluate the risks associated with these concentrations.  Studies have shown that for medicines found in surface waters, the detected concentrations of pharmaceutical products are far below levels that may affect human health.^7,8 The same could be said for impacts to aquatic life, with the possible exceptions of hormonally active chemicals and antibiotics.⁹

While the major route for active pharmaceutical ingredients (APIs) to reach water is via excretion from patients following prescribed usage¹⁰, there is widespread concern for contamination from manufacturing effluents. Therefore, it is important to reduce wastewater discharges of APIs to levels below thresholds of concern. Unfortunately, for many older, previously approved pharmaceuticals on the market, little or no environmental fate and effects data are available. Additionally, during the research and development of APIs, chronic aquatic data are usually not developed until the drug reaches Phase II or III human clinical trials. Chronic studies are generally performed on fish, daphnia and algae, per the EU and U.S. guidelines for environmental risk assessments of pharmaceuticals.^11,12 However, these studies may take up to two years to complete. Therefore, a process must be in place to ensure wastewater concentrations do not exceed potentially toxic thresholds of APIs prior to a pharmaceutical going to market, i.e., during research and process scale-up activities. Once production of larger quantities is possible, chronic toxicity studies are initiated and chronic aquatic data are derived (Figure 1).

Previous research has shown that acute toxicity data may be leveraged as screening tools when chronic data are unavailable.¹³ Therefore, preliminary predicted-no-effect concentrations (PNECs) can be calculated using appropriate assessment factors (AFs)¹³ and acute aquatic toxicity endpoints. AFs are intended to account for uncertainty due to inter-species, intra-species and lab-to-field variation. Once established, chronic data can be used, which require less stringent AFs.¹⁴ Special classes of compounds based on their intended mode of action and/or potency may trigger the need for more stringent AFs (e.g., female sex hormones).¹⁵

Another important consideration is potential contamination of downstream potable water intakes. Acceptable daily exposure (ADE) values are derived using data from preclinical and clinical studies conducted to evaluate the safety and efficacy of new drugs for patients. These limits are also used to support cleaning validation programs to prevent cross-contamination in multi-product manufacturing facilities and to establish safe levels of exposure for workers handling APIs.¹⁶ The ADE can also be applied to establish safe levels for humans following potential exposure via the environment.

The objective of this paper is to describe one method for setting limits for APIs in research and manufacturing effluent discharge in order to reduce potential downstream contamination, and subsequent adverse effects, to both human and aquatic life. One common API, sitagliptin phosphate—hereby referred to as sitagliptin—will be used as an example to illustrate this methodology. Sitagliptin is indicated for the treatment of Type 2 diabetes and is the active ingredient in Januvia. It has been on the market since 2006, and is produced in large quantities. Additionally, sitagliptin is manufactured in North America, Europe and Asia, which results in the potential for global impacts.

Derivation of environmental quality criteria to protect aquatic life (EQC_E)
Medicines are subjected to a battery of environmental fate and effects tests, which are appropriate for evaluating potential environmental impacts from pharmaceutical residues in the environment. Tests are typically performed in independent research laboratories accredited to perform regulatory testing. Timing for test initiation is triggered by the phase of the clinical program, and the types of tests are commensurate with the quantity of material produced at the site (Figure 1). Studies are carried out following protocols issued by the Organization for Economic Co-operation and Development (OECD) and in full compliance with Good Laboratory Practice (GLP) principles and regulations, when possible. Prior to environmental testing, a 1 ppb de minimis value is applied as the default environmental quality criterion to protect aquatic life (EQC_E), unless the API is expected to be more hazardous based on data from early clinical studies or mode of action effects (e.g., some hormones). The 1 ppb default value is consistent with regulatory guidance in the U.S., and also leverages recent scientific work¹³ where 102 pharmaceuticals were reviewed and found that for most classes of APIs, the 1 ppb de minimis value as a PNEC is appropriately conservative except for sex hormones and antineoplastic compounds with estrogenic activity.

Studies conducted to evaluate the environmental fate and effects are presented in Table 1. Acute—48 to 96 hours—studies along with the measurement of octanol-water partition coefficient (Kow) are initiated early in product development once sufficient quantities are produced to allow for testing. These tests are not complicated, provide fast results and offer an indication of relative toxicity. The octanol-water partition coefficient provides information about the potential for bioaccumulation, difficulty of analysis in different media and whether terrestrial testing may be warranted.  Chronic toxicity studies are performed on at least three trophic levels, generally fish, daphnids and algae, and are provided as part of the regulatory filing. Additionally, tests on biodegradation, transformation in sediments, and bioconcentration are also conducted. The EMA guideline mandates that a ready biodegradation study be conducted per OECD 301. As very few pharmaceutical compounds meet the criteria for ready biodegradability per OECD 301, an alternate study per OECD 314 (sludge die-away) may be initiated instead, as this study is more informative about the potential biodegradation in wastewater treatment plants.

Results of the toxicity studies are used to develop Environmental Quality Criteria (EQC_E), which are concentrations considered to be safe to aquatic life. These EQCs are derived from the most sensitive endpoint determined during studies of sentinel species—typically fish, daphnia and algae. If only acute data are available, a preliminary EQC (pEQC_E) is developed using the lowest concentration that affects 50% of the population (EC₅₀) and an appropriate applied AF. Typically, an AF of 100 is considered appropriate for intermittent releases, which might be possible during pilot plant production. If chronic data are available for at least three trophic levels, then a final EQCE is developed using the no-observed-effect concentration (NOEC) and the applied AF is reduced to 10. However, it should be noted that if information suggests that the above assessment factors are not protective enough (e.g., based on mode of action), higher AFs will be used. Likewise, if the mode of action is known and the most sensitive species has been identified (e.g., blue-green alga for a large number of antibiotics), then the requirement for chronic tests in fish and daphnia may be waived or given lower priority.

Based on data obtained from environmental fate studies (e.g., organic carbon partitioning; sediment transformation), there may be a need to minimize exposure to terrestrial ecosystems. Results indicating that APIs have high binding to sludge and the ability to unduly impact terrestrial receptors are highlighted so that sites may consider this concern in their evaluations. Likewise, if bioconcentration tests indicate the potential for bioconcentration in species, additional assessments on this exposure route may be conducted on a site-specific basis.

Derivation of environmental quality criteria to protect human health (EQC_H)
Acceptable daily exposures (ADEs) are typically developed using the lowest clinical dose of the API and application of various uncertainty factors to account for inter-individual variability and extrapolation to doses that are not expected to produce any adverse or clinically significant effects.¹⁷ Early in development, an ADE can be derived using the threshold of toxicological concern (TTC) concept.¹⁶ The default standard water ingestion rate of a 70 kg average adult is 2 L/day.¹⁸ If the ADE is set based on a route of exposure other than ingestion, oral bioavailability is taken into account since the potential exposure pathway is through drinking water. Since the ADE assumes 100% bioavailability, the bioavailability correction factor (α) is 100 divided by the percent oral bioavailability (e.g., α=100/50=2). A criterion to protect human health (EQCH) can therefore be calculated as follows:

EQC_H = ADE*α
2L/day

All data that support the development of EQC_E and EQC_H values are summarized for distribution to site EHS professionals, as well as to external partners. These summaries assist sites with determining how to limit environmental impacts from API production. Also, as data on physical-chemical properties and environmental fate are included in the summary, such data may be leveraged to inform on treatment technologies, if needed.

Derivation of environmental quality limits (EQlS)
EQC values are based on the underlying toxicity data associated with the API. These values are independent of manufacturing or research location. The potential for environmental impact is site-specific and draws on some key attributes of the area including geographic location, proximity to receiving water and treatment options.  Effluent from manufacturing processes can be discharged directly into surface water, to on-site wastewater treatment plants (WWTPs) or to off-site WWTPs.

The size and nature of the receiving stream and WWTP will dictate, in part, the allowable discharge of a specific API from the site.  The assessments assume worst case, low-flow conditions of the receiving water body.  These are obtained from published data (ex. 7Q10 in U.S.) or historic information available for the water body. The environmental quality limit (EQL) is calculated by multiplying the EQC value (mg/L) by the stream low-flow value (L/day) to derive a site-specific limit (mg/day).

Mass balance assessment
Small quantities of APIs may be released in wastewater effluents as part of the manufacturing processes. A mass balance approach should be used to determine the potential mass of API discharged (POD) into the receiving stream on a daily basis. Discharge quantities may be estimated based on manufacturing yield projections or sampling of rinse water or discharge from a building location. The volume of water discharged from the manufacturing plant and the volume of water discharged from the WWTP are then used to estimate the mass discharge of API at the point of discharge into the receiving stream. This mass balance assessment provides a predicted daily mass discharge of API into the receiving stream.

Application of the EQL
The predicted daily mass of API in the receiving stream is then compared to the EQL. If the discharge mass exceeds half of the EQL, then a refined mass balance approach can be used to develop a more accurate estimation. For example, more detailed modeling considering fate and transport in wastewater systems (e.g., degradation, sorption) will be used to more accurately predict the removal of the API in the system. If the refined calculations still reveal a potential exceedance of half of the allowable limit, additional action is required to reduce the API discharge to meet the EQL. Manufacturing sites have the responsibility to implement any and all mechanisms to reduce discharges to acceptable levels. This may include segregation of waste, additional treatment, etc. A process flow diagram for EQC implementation is presented in Figure 2.

Example
An example EQC evaluation for sitagliptin is summarized below.

A full dataset for both acute and chronic endpoints are available for sitagliptin (Table 1). Therefore, a final EQC_E of 0.9 mg/L is derived based on the fathead minnow chronic study and an AF of 10. Prior to generation of chronic data, the preliminary EQC_E of 0.4 mg/L was applied to pilot plant activities based on the green algae EC₅₀ of 39 mg/L.
An acceptable daily exposure (ADE) value of 1 mg/day was determined for sitagliptin using the lowest clinical dose of 25 mg as a lowest-observed-adverse-effect level (LOAEL) and a composite uncertainty factor of 25 to account for interindividual variability and extrapolation to a no-observed-adverse-effect level (NOAEL).The EQCH is calculated to be 0.5 mg/L. No correction for oral bioavailability was required (α=1).

Selection of derived EQC values
Each site that manufactures an API derives EQL values that are specific to their operations and the surrounding environment. If no potable water intakes exist downstream of manufacturing discharges, then only the EQC_E should be applied. However, if it is not known whether potable water intakes exist, then both EQC values should be used.
For one site manufacturing sitagliptin, no potable drinking water intake exists downstream of the point of discharge of the effluent, so the EQL value was calculated using the EQC_E of 0.9 mg/L and appropriate assumptions regarding on site dilution beyond the sampling point and the boundary of the mixing zone at the point of discharge. In this case, the EQL is calculated and, based on the effluent volume, the maximum amount that could be discharged in the effluent would be 6.8 kg/day.

If one were to assume that sitagliptin production was 300 kg/day with a 0.5% loss estimate and was sent to an offsite WWTP with a total flow of 4 x 10⁵ L/day, the estimated quantity that could enter the mixing zone of 3.8 x 10-⁶ kg/day is well below the EQL-derived maximum allowable discharge amount of 06.8 kg/day and no further refinements would be necessary.

Conclusion
The above methodology outlines steps that can be taken to determine what, if any, impacts to human health and the environment will occur, and at what concentrations, following the discharge of APIs from manufacturing effluent. EQC values are compound-specific but can be applied globally because site-specific effects are taken into account when setting EQLs. It should be noted that EQLs are dependent upon the characteristics of the receiving stream, and thus discharges to a small stream have the potential to have much higher impact than to a larger body of surface water. This approach provides a consistent methodology applicable to all sites but has the ability to address site-specific or API-specific concerns under special circumstances.

Pharmaceutical companies play a vital role in human health, but are also responsible stewards for the health of the environment. While the largest source of surface water pharmaceutical contamination comes from excretion following patient use, mass discharges in manufacturing and research waste water effluents should be controlled to levels that are protective of human health and aquatic life. 

Acknowledgements
The authors would like to thank the many Merck scientists and engineers that have contributed to the development of this methodology.

References

1. Snyder EM, Pleus RC, Snyder SA. 2005. Current Issues -- Pharmaceuticals and EDCs in the U.S. Water Industry - An Update (PDF). American Water Works Association 97:32-35.
2. Kümmerer K. 2009. The presence of pharmaceuticals in the environment due to human use – present knowledge and future challenges. Journal of Environmental Management 90:2354-2366.
3. Murray KE, Thomas SM, Bodour AA. 2010. Prioritizing research for trace pollutants and emerging contaminants in the freshwater environment. Environmental Pollution 158:3462-3471.
4. Focazio M, Kolpin D, Barnes K, Furlong E, Meyer M, Zaugg S, Barber L, Thurman M. 2008. A national reconnaissance for pharmaceuticals and other organic wastewater contaminants in the United States--II) untreated drinking water sources. Sci Total Environ 402:201-216.
5. Benotti M, Trenholm R, Vanderford B, Holady J, Stanford B, Snyder S. 2009. Pharmaceuticals and endocrine disrupting compounds in U.S. drinking water. Environ Sci Technol 43:597-603.
6. Padhye L, Yao H, Kung'u F, Huang C. 2014. Year-long evaluation on the occurrence and fate of pharmaceuticals, personal care products, and endocrine disrupting chemicals in an urban drinking water treatment plant. Water Res 15:266-276.
7. Schwab B, Hayes E, Fiori J, Mastrocco F, Roden N, Cragin D, Meyerhoff R, D'Aco V, Anderson P. 2005. Human pharmaceuticals in US surface waters: a human health risk assessment. Regul Toxicol Pharmacol 42:296-312.
8. Cunningham V, Olson SBM. 2009. Human health risk assessment from the presence of human pharmaceuticals in the aquatic environment. Regul Toxicol Pharmacol 53:39-45.
9. Anderson PD, D'Aco VJ, Shanahan P, Chapra SC, Buzby ME, Cunningham VL, DuPlessie BM, Hayes EP, Mastrocco FJ, Parke NJ, Rader JC, Samuelian JH, Schwab BW. 2004. Screening Analysis of Human Pharmaceutical Compounds in U.S. Surface Waters. Environmental Science & Technology 38:838-849.
10. Aga DS. 2008. Advances in the Analysis of Pharmaceuticals in the Aquatic Environment. In Aga DS, ed, Fate of Pharmaceuticals in the Environment and in Water Treatment Systems. CRC Press, Taylor & Francis Group, Boca Raton, FL.
11. EU EMA. 2006. Guideline on the Environmental Risk Assessment of Medicinal Products for Human Use. In Agency EM, ed, EMEA/CHMP/SWP/4447/00. Committee for Medicinal Products for Human Use (CHMP), London, UK.
12. US FDA. 1998. Environmental Assessment of Human Drugs and Biologics Application. In US Food and Drug Administration CfDEaRC, ed.
13. Vestel J, Caldwell DJ, Constantine L, D'Aco, VJ, Davidson, T, Dolan, DG, Millard SP, Murray-Smith R, Parke NJ, Ryan JJ, Straub JO, Wilson, P. 2015. Use of Acute and Chronic Ecotoxicity Data in Environmental Risk Assessment of Pharmaceuticals.  Environmental Toxicology & Chemistry DOI: 10.1002/etc.3260
14. ECHA. 2008. Guidance on information requirements and chemical safety asssesment. Chapter R.10: Characterisation of dose [concentration]-response for environment. REACH Regulation (EC) No 1907/2006 of the European Parliament and of the Council of 18 December 2006. European Chemicals Agency, Helsinki, Finland.
15. Caldwell DJ, Mastrocco F, Hutchinson TH, Länge R, Heijerick D, Janssen C, Anderson PD, Sumpter JP. 2008. Derivation of an Aquatic Predicted No-Effect Concentration for the Synthetic Hormone, 17α-Ethinyl Estradiol. Environmental Science & Technology 42:7046-7054.
16. Dolan DG, Naumann BD, Sargent EV, Maier A, Dourson ML. 2005. Application of the threshold of toxicological concern concept to pharmaceutical manufacturing operations. Regul Toxicol Pharmacol:1-9.
17. Sargent EV, Faria E, Pfister T, Sussman RG. 2013. Guidance on the establishment of acceptable daily exposure limits (ADE) to support Risk-Based Manufacture of Pharmaceutical Products. Regulatory Toxicology and Pharmacology 65:242-250.
18. EPA. 2004. Estimated Per Capita Water Ingestion and Body Weight in the United States–An Update, Washington, DC.