Address correspondence to C.G. Daughton, Environmental Chemistry Branch, Environmental Sciences Division, U.S. EPA, ORD/NERL, 944 East Harmon Ave., Las Vegas, Nevada 89119. Telephone: (702) 798-2207. Fax: (702) 798-2142. E-mail: daughton.christian@epa.gov

The authors thank the following people for taking their valuable time to helpfully review both technical and policy aspects of this manuscript: O. Conerly (for the U.S. EPA Office of Water), M. Firestone (for the U.S. EPA Office of Pollution, Prevention, and Toxics), and EHP's anonymous reviewers, all of whom contributed to improving the quality of the manuscript. We also thank G. Wayne Sovocool for assistance in verifying chemical structures and associated chemical data.

The U.S. Environmental Protection Agency (U.S. EPA), through its Office of Research and Development, partially funded and collaborated in the research described here. This manuscript has been reviewed by U.S. EPA and approved for publication. Mention of trade names or commercial products does not constitute endorsement or recommendation by U.S. EPA for use.

Received 23 June 1999; accepted 9 September 1999.

Summary

Risks associated with previously unknown, unrecognized, unanticipated, or unsuspected chemical pollutants in the environment have long been a major concern of environmental scientists. The importance of identifying such emerging risks is reflected in one of the top five goals of the Strategic Plan 2000 for the U.S. Environmental Protection Agency's (U.S. EPA) Office of Research and Development. Early identification and investigation of potential environmental pollution issues before they worsen are critical for protecting ecologic and human health. It is also important to rule out issues that could be of concern but prove otherwise, so that limited resources can be redirected. Ecosystem change is effected by human activities primarily via three routes: habitat fragmentation, alteration of community structure (e.g., via nonindigenous species), and chemical pollution. The scope of the former two is highly delineated and obvious compared with the latter. During the last three decades, the impact of chemical pollution has focused almost exclusively on the conventional "priority" pollutants. This group of chemicals, however, is only one piece of the larger puzzle.

One large class of chemicals receiving comparatively little attention comprises the pharmaceuticals and active ingredients in personal care products (PPCPs), which are used in large amounts throughout the world; quantities of many are on par with agrochemicals. Escalating introduction to the marketplace of new pharmaceuticals is adding exponentially to the already large array of chemical classes, each with distinct modes of biochemical action, many of which are poorly understood. In contrast to agrochemicals, most of these products are disposed or discharged into the environment on a continual basis via domestic/industrial sewage systems and wet-weather runoff. The bioactive ingredients are first subjected to metabolism by the dosed user; the excreted metabolites and unaltered parent compounds can then be subjected to further transformations in sewage treatment facilities. The literature shows, however, that many of these compounds survive biodegradation, eventually being discharged into receiving waters; metabolic conjugates can even be converted back to their free parent forms. Many of these PPCPs and their metabolites are ubiquitous and display persistence in, and bioconcentration from, surface waters on par with those of the widely recognized organochlorine pollutants. Additionally, by way of continual infusion into the aquatic environment, those PPCPs that might have low persistence can display the same exposure potential as truly persistent pollutants since their transformation/removal rates can be compensated by their replacement rates.

Although certain biochemical actions of many drugs in humans have been elucidated, these actions are not necessarily always the ones responsible for the purported physiologic target effects. Sometimes the known pathways of action may have nothing to do with the actual desired effect, as the actual mechanism remains totally unknown. Understanding of the complex biochemical signaling pathways is currently too limited to design drugs that act only via targeted routes, and even then, if their activity can be limited to a single type of receptor, the tissue distribution of the receptor may not be fully known. Unpredicted and unknown side effects are often the norm. The possible actions and biochemical ramifications on nontarget aquatic biota are even less understood; many are totally unknown. The few that are known to elicit subtle but dramatic effects on aquatic life at very low concentrations, however, may point to an ill-defined vulnerability in aquatic ecosystems. A major concern is not necessarily acute effects to nontarget species (effects amenable to monitoring once they are understood), but rather the manifestation of perhaps imperceptible effects that can accumulate over time to ultimately yield truly profound changes--those whose causes would be obscured by time and that would not be distinguishable from natural events. The specter of subtle, cumulative effects could reduce the usefulness of current toxicity-directed screening methods in testing waste effluents for toxicologic end points due to PPCPs. Subtle effects, from low concentrations of bioactive PPCPs, whose continual expression over long periods of time in certain nontarget populations, could lead to cumulative, insidious, adverse impacts that would otherwise be attributed to natural change/adaptation or ecologic succession--any "signal" would be lost among the noise. Current comprehensive environmental risk assessments and epidemiologic studies do not factor in exposures/body burdens from PPCPs and therefore may be flawed by over simplicity.

It is useful to note that the data reported and evaluated in this review reflect the diverse and uneven nature of the PPCP literature published for source/origin, occurrence, distribution, transport, transformation, ecologic exposure and effects, risk assessment, and test strategies. The comprehensiveness of the published literature in each of these areas and across the broad spectrum of PPCP classes is very unequal. This review therefore does not present an exhaustive and rounded view of this emerging topic but rather summarizes most of the significant papers in an integrated, comprehensive manner, and thereby elucidates many of the questions that still need to be addressed by the environmental science community. This review aims to catalyze a discussion on the potential importance of PPCPs in the environment and presents recommendations for focusing further research (Table 1).

Introduction

For the purposes of this discussion, pharmaceutical (and veterinary and illicit) drugs (and the ingredients in cosmetics, food supplements, and other personal care products), together with their respective metabolites and transformation products, will collectively be referred to as pharmaceuticals and personal care products. PPCPs are continually infused into the environment via sewage treatment facilities and wet weather runoff. In many instances, untreated sewage is discharged into receiving waters (e.g., flood overload events, domestic "straight-piping," or sewage waters lacking municipal treatment). In the United States alone, possibly more than a million homes do not have sewage systems but instead rely on direct discharge of raw sewage into streams by straight-piping or by outhouses not connected to leach fields (1). A number of Canadian cities are reported to discharge 3.25 billion liters per day (over 1 trillion liters per year) of essentially untreated sewage into surface waters and the ocean (2). Raw/treated sewage is also disposed of from some locales in the deep ocean where it may possibly remix with upper waters.

We hope that this overview of PPCPs in the environment will a) catalyze a concerted effort among environmental chemists and ecotoxicologists to survey sewage treatment effluents, surface waters/groundwaters, and potable water for the presence of PPCPs and their bioactive transformation products and to determine their origins; b) elucidate the spectrum of possible physiologic effects of PPCPs on nontarget species, especially those that are aquatic; and c) promote discussion of whether this is an environmental issue deserving further investigation. We believe that a scientific debate on this topic is warranted given the evidence that has been accumulating over the last two decades on the occurrence of various pharmaceuticals in sewage effluent and in both surface waters and groundwaters. The big unknown is whether the combined low concentrations from each of the numerous PPCPs and their transformation products have any significance with respect to ecologic function, while recognizing that immediate effects could escape detection if they are subtle and that long-term cumulative consequences could be insidious. Another question is whether the pharmaceuticals remaining in water used for domestic purposes poses long-term risks for human health after lifetime ingestion via potable waters multiple times a day of very low, subtherapeutic doses of numerous pharmaceuticals; this issue, however, is not addressed in this review.

The hypothesis is further complicated by the fact that while the concentration of individual drugs in the aquatic environment could be low (sub-parts per billion or sub-nanomolar, often referred to as micropollutants), the presence of numerous drugs sharing a specific mode of action could lead to significant effects through additive exposures. It is also significant that drugs, unlike pesticides, have not been subjected to the same scrutiny regarding possible adverse environmental effects. They have therefore enjoyed several decades of unrestricted discharge to the environment, mainly via sewage treatment works. This is surprising especially since certain pharmaceuticals are designed to modulate endocrine and immune systems and cellular signal transduction and as such (as opposed to pesticides and other industrial chemicals already undergoing scrutiny as endocrine disruptors) have obvious potential as endocrine disruptors in the environment. Exposure to PPCPs in the environment, especially for aquatic organisms, may differ from that of pesticides and other industrial chemicals in one significant respect--exposures may be of a more chronic nature because PPCPs are constantly infused into the environment wherever humans live or visit, whereas pesticide fluxes are more sporadic and have greater spatial heterogeneity. It is quite apparent that little information exists from which to construct comprehensive risk assessments for the vast majority of PPCPs having the potential to enter the environment.

Although little is known of the occurrence and effects of pharmaceuticals in the environment, more data exist for antibiotics than for any other therapeutic class. This is a result of their extensive use in both human therapy and animal husbandry, their more easily detected effects end points (e.g., via microbial and immunoassays), and their greater chances of introduction into the environment, not just by sewage treatment plants, but also by run-off and groundwater contamination, especially from confined animal feeding operations (CAFOs). The literature on antibiotics is much more developed because of the obvious issues of direct effects on native microbiota (and consequent alteration of microbial community structure) and development of resistance in potential human pathogens. Because of the considerably larger literature on antibiotics, this review only touches on the issue; for the same reason, this discussion only touches on steroidal drugs (those purposefully designed to modulate endocrine systems).

For the purposes of this document, pharmaceuticals will refer to nonbiologic drugs (i.e., those that do not comprise proteinaceous or nucleotide material). The number of biologics approved by the U.S. Food and Drug Administration (FDA) is growing, and their fate in the environment is unknown. This overview covers only a subset of the commercially available classes of pharmaceuticals and active ingredients in personal care products. The subset of classes discussed in this review comprises the primary classes for which the limited data on environmental occurrence and effects on nontarget species can be found, in a highly fragmented, disjointed, and disparate literature.

Pharmaceutical drugs are chemicals used for diagnosis, treatment (cure/mitigation), alteration, or prevention of disease, health condition, or structure/function of the human body. The definition is extended to veterinary pharmaceuticals and can also be applied to illicit (recreational) drugs. It also must be noted that the active ingredient in a drug may or may not be the actual formulated parent compound. For example, prodrugs such as the esters of clofibric acid, a metabolite of certain lipid regulators, are converted from pharmacologically inactive parent compounds to the physiologically active form. With the exception of antibiotics and antineoplastics, the objective for most drug classes is simply to control symptoms and not to actually cure conditions. As such, many drugs are taken for very long periods, sometimes a good portion of the user's lifetime.

Although drugs are usually designed with a specific mode of action in mind (e.g., methotrexate universally affects all organisms in the same manner--by inhibiting nucleic acid synthesis), they can also have numerous effects on nontarget, or as yet unknown, receptors and possibly cause side effects in the target organism. Furthermore, and of equal importance, nontarget organisms can have receptors, or receptor tissue distributions, that do not exist in the target organisms, and therefore unexpected effects can result from unintentional exposure. This is a primary basis for the hypothesis of this paper.

Pharmaceuticals in the Environment

Sources and Origins

The possibility that pharmaceuticals can enter the environment from a number of different routes and possibly cause untoward effects in biota has been noted in the scientific literature for several decades, but its significance has gone largely unnoticed. This probably results in large part from the international regulation of drugs by human health agencies, which usually have limited expertise in environmental issues. Traditionally, drugs were rarely viewed as potential environmental pollutants; there was seldom serious consideration as to their fates once they were excreted from the user. Then again, until the 1990s, any concerted efforts to look for drugs in the environment would have met with limited success because the requisite chemical analysis tools with sufficiently high separatory efficiencies, to resolve the drugs from the plethora of other substances--native and anthropogenic alike, and low detection limits (i.e., nanograms per liter or parts per trillion), were not commonly available. Other obstacles, which still exist to a large degree, are that many pharmaceuticals and cosmetic ingredients and their metabolites are not available in the widely used environmentally oriented mass spectral libraries. These are available in specialty libraries such as Pfleger (e.g., 3,4), which are not frequently used by environmental chemists. Analytical reference standards, when available, are often difficult to acquire, and are quite costly. The majority of drugs are also highly water soluble. This precludes the application of straightforward, conventional sample clean-up/preconcentration methods, coupled with direct gas chromatographic separation, that have been used for years for "conventional" pollutants, which tend to be less polar and more volatile.

Drugs in the environment did not capture the attention of the scientific or popular press until the last couple of years, with some significant overviews/reviews presented by Halling-Sørenson et al. (5), Montague (6), Raloff (7), Roembke et al. (8), Ternes et al. (9), and Velagaleti (10), among others. The evidence supports the case that PPCPs refractory to degradation and transformation [see Halling-Sørenson et al. (5) for summary of published transformation studies] do indeed have the potential to reach the environment. What is not known, however, is whether these chemicals and their transformation products can elicit physiologic effects on biota at the low concentrations (ng-µg/L) at which they are observed to occur. Another unknown is the actual quantity of each of the numerous commercial drugs that is ingested/disposed. With respect to determining the potential extent of the problem, this contrasts sharply with pesticides in which usage is much better documented and controlled.

A list of the PPCPs covered in this review, together with their chemical names, structures, and some representative environmental occurrence/effects data, is presented in Table 2. These chemicals, together with their synthetic precursors and transformation products, are continually released into the environment in enormous quantities as a result of their manufacture, use (via excretion, mainly in urine and feces), and disposal of unused/unwanted drugs and those that have expired, both directly into the domestic sewage system and via burial in landfills. Although largely unknown, there is evidence that large quantities of prescription and nonprescription, "over-the-counter" (OTC) drugs are never consumed (for any number of reasons) (11), and many of these are undoubtedly eventually disposed down toilets or via domestic refuse.

A striking difference between pharmaceuticals and pesticides with respect to environmental release is that pharmaceuticals have the potential for ubiquitous direct release into the environment worldwide--anywhere that humans live or visit. Even areas considered relatively pristine (e.g., national parks) are subject to pharmaceutical exposures, especially given that some parks have very large, aging sewage treatment systems, some of which discharge into park surface waters and some of which overflow during wet weather events and infrastructure failures (e.g., Yellowstone National Park) (12,13). Other possible sources include disposal of unwanted illicit drugs and synthesis byproducts into domestic sewage systems by clandestine drug operations; disposal of raw products and intermediates (e.g., ephedrine) via toilets is not uncommon in illegal laboratories. Also, in contrast to pesticides, pharmaceuticals in any stage of clinical testing (not yet approved for dispensing by the FDA) are subject to release into the environment, although their overall concentrations would be very low.

Some drugs are excreted essentially unaltered in their free form (e.g., methotrexate and platinum antineoplastics), often with the help of active cellular "multidrug transporters" for moderately lipophilic drugs. Others are metabolized to various extents, which is partly a function of the individual patient and the circadian timing of the dose (the P450 microsomal oxidase system is a major route of formation of more polar, more easily excreted metabolites). Still others are converted to more soluble forms by formation of conjugates (with sugars or peptides). The subsequent transformation products--metabolites and conjugates from eukaryotic and prokaryotic metabolism, and from physicochemical alteration--add to the already complex picture of thousands of highly bioactive chemicals. The FDA refers to all metabolites and physicochemical transformation products, for example, those that range from the dissociated parent compound to photolysis products, for a given drug as structurally related substances (SRSs), which can have greater or lesser physiologic activity than the parent drug.

As in mammals, the metabolic disposition of lipophilic xenobiotics, such as numerous drugs, in vertebrate aquatic species is largely governed by what is referred to as Phase I and Phase II reactions (14); less is known about invertebrate metabolism. Phase I makes use of monooxygenases (e.g., cytochrome P450), reductases, and hydrolases (for esters and epoxides) to add reactive functional groups to the molecule. Phase II uses covalent conjugation (glucuronidation) to make the molecule hydrophilic and more excretable. These reactions are catalyzed by glycosyltransferases and sulfotransferases (for hydroxyaromatics and carboxy groups), glutathione S-transferases (for electrophilic functional groups such as halogens, nitro groups, or unsaturated/conjugated sites), acetyltransferases (for primary amines or hydrazines), and aminoacyltransferases (for forming peptides from carboxy groups using free amino acids). This metabolic strategy creates metabolites successively more polar than the parent compound, thereby enhancing excretion (Figure 1). Considerable interspecies and intraspecies diversity, however, can be observed in actual metabolic potentials. Many drugs and metabolic products, especially those over 400 Da, are concentrated in the bile of fish (vs blood or fat) (15). Although the total amount excreted via the urine may be higher, Guarino and Lech (15) recommend bile analysis to maximize the chance of detecting drugs, especially their conjugates, in fish in order to confirm exposure. They also report that the ratio of drug concentrations in bile to that in the surrounding water can increase many orders of magnitude as exposure duration increases (15). Detection of exposure of fish to many drugs can thereby be facilitated through the analysis of bile.

The introduction of drugs into the environment is partly a function of the quantity of drugs manufactured, the dosage frequency and amount [the 200 most frequently prescribed drugs, representing about two-thirds of all prescriptions filled in the United States for the most recently documented year, are listed in RxList (16)], the excretion efficiency of the parent compound and metabolites, propensity of the drug to sorb to solids, and the metabolic transformation capability of subsequent sewage treatment (or landfill) microorganisms. Publicly owned wastewater treatment plants (POTWs) receive influent from both domestic, municipal, and industrial (including pharmaceutical manufacture) sewage systems. The processed liquid effluents from primary and secondary treatments are then discharged to surface waters and the residual solids (sludge) to landfills/farms; land disposal, including manure from treated animals at CAFOs, creates the potential for introduction into groundwaters or surface waters (via wet weather run-off). Theoretically, PPCPs in sewage sludge applied to crop lands could be taken up by plants.

Figure 1. Metabolic approach to increasing the polarity (and excretability) of drugs.

Compounds surviving the various phases of metabolism and other degradative or sequestering actions (i.e., display environmental persistence) can then pose an exposure risk for organisms in the environment. Even the less/nontoxic conjugates (glucuronides) can later be converted back to the original bioactive compounds via enzymatic (ß-glucuronidases) or chemical hydrolysis (e.g., acetylsalicylic acid can be hydrolyzed to the free salicylic acid). Some degradation products can even be more bioactive than the parent compound. Therefore, conjugates can essentially act as storage reservoirs from which the free drugs can later be released into the environment. Up to 90% of certain drugs can become conjugated (17,18), conjugation varying as a function of chemical class. These pathways of introduction into the environment have been summarized by Velagaleti (10).

Sewage treatment plants. Treatment facilities, primarily POTWs or sewage treatment works (STWs), which include privately owned works as well, play a key role in the introduction of pharmaceuticals into the environment [see Rogers (19) for a review of the fate of synthetic chemicals in sewage treatment plants]. STWs were designed to handle human waste of mainly natural origin, primarily via the acclimated degradative action of microorganisms (the efficiency of metabolism of a given drug can increase with duration of treatment because of enzyme induction and cellular adaptation) and the coagulation/flocculation of suspended solids; sometimes, tertiary treatment (e.g., chemical/ultraviolet [UV] oxidation) is used. Most anthropogenic chemicals introduced along with this normal waste suffer unknown fates. Two primary mechanisms remove substances from the incoming waste stream: a) microbial degradation to lower molecular weight products, leading sometimes to complete mineralization--CO₂ and H₂O; and b) sorption to filterable solids, which are later removed with the sludge.

Although the microbiota of sewage treatment systems may have been exposed to many PPCPs for a number of years, two factors work against the effective microbial removal of these substances from STWs. First, the concentrations of most drugs are probably so low that the lower limits for enzyme affinities may not be met. For example, the daily loadings of PPCPs into STWs are largely a function of the serviced human population, the dosages/duration of medications consumed, and the metabolic/excretory half-lives, which are all large variables. As an example, the daily load of a subset of pharmaceuticals to a particular POTW near Frankfurt/Main, Germany, ranged from tens to hundreds of grams, with approximate individual removal efficiencies varying widely from 10 to 100% but trending to around 60% (18). This particular POTW serviced about a third of a million people at a flow rate of roughly 60,000 m³/day. Despite the number of studies on treatment efficiencies, a widespread investigation is still lacking for the differences in removal efficiencies for distinct types of STWs as well as for individual treatment techniques. The extent to which a particular plant uses primary, secondary, and tertiary technologies will greatly influence removal efficiencies; the technologies employed vary widely among cities. The biodegradative fate of most compounds in STWs is governed by nongrowth-limiting (enzyme-saturating) substrate concentrations (copiotrophic metabolism). In contrast, PPCPs are present in STWs at concentrations at enzyme-subsaturating levels, which necessitates oligotrophic metabolism. These micropollutants might be handled by only a small subset of specialist oligotrophic organisms whose occurrence is probably more prevalent in native environments characterized by low-carbon fluxes (e.g., sediments and associated pore waters, where desorption mass transfer is limiting) than in STWs. This means that degradation of PPCPs may occur more prevalently in the receiving waters/sediments than in STWs.

Second, many new drugs are introduced to the market each year; some of these drugs are from entirely new classes never seen before by the microbiota of an STW. Each of these presents a new challenge to biodegradation. A worst-case scenario may not be unusual--the concentration of a drug leaving an STW in the effluent could essentially be the same as that entering. Only the severalfold to multiple order of magnitude dilution when the effluent is mixed into the receiving water, assuming a sufficiently high natural flow, serves to reduce the concentration; obviously, smaller streams have increased potential for having higher concentrations of any PPCP that has been introduced. In general, most pharmaceuticals resist extensive microbial degradation (e.g., mineralization) (10). Although some parent drugs often show poor solubility in water (10), leading to preferential sorption to suspended particles, they can thereby sorb to colloids and therefore be discharged in the aqueous effluent. Metabolites, including breakdown products and conjugates, will partition mainly to the aqueous effluent. Some published data demonstrate that many parent drugs do make their way into the environment (see references cited in Table 2 under "Environmental Occurrence").

The efficiency of removal of pharmaceuticals by STWs is largely unknown. Currently, the most extensive study of treatment efficiency (18) reports removal from German STWs of 14 drugs representing five broad physiologic categories. Removal of the parent compound (keep in mind that possible subsequent metabolites were not accounted for) ranged from 7% (carbamazepine, an antiepileptic) to 96% (propranolol, a beta-blocker); most removal efficiencies averaged about 60%. Fenofibrate, acetominophen, and salicylic acid, o-hydroxyhippuric acid, and gentisic acid (acetylsalicylic acid metabolites) could not be detected in effluent; salicylic acid was found in the influent at concentrations up to 54 µg/L. It is important to understand that absent the stoichiometric accounting of metabolic products, one cannot distinguish between the three major fates of a substance: a) degradation to lower molecular weight compounds, b) physical sequestration by solids (and subsequent removal as sludge), and c) conjugates that can later be hydrolyzed to yield the parent compound (e.g., clofibric and fenofibric acid conjugates) (18). Therefore, by simply following disappearance (removal) of a substance, one cannot conclude that it was structurally altered or destroyed--it may simply reside in another state or form. Identifying metabolic products is difficult not only because of the number of metabolites (sometimes several per parent compound) but also because standard reference materials are difficult to obtain commercially and can be costly.

Despite high removal rates in STWs for some drugs, upsets in the homeostasis of a treatment plant can result in higher than normal discharges. For example, Ternes (18) found that wet weather runoff dramatically reduced the removal rates for certain drugs (e.g., several nonsteroidal anti-inflammatory drugs [NSAIDs] and lipid regulators) in a facility located close to Frankfurt/Main. During the increased period of influent flow, the removal rate dropped to below 5% from over 60% previously; several days were required for the removal rates to recover. Clearly, even for drugs efficiently removed, the operational state of the STW can have a dramatic effect on the removal efficiencies. Other transients that could affect removal include transitions between seasons and sporadic plug-flow influx of toxicants from various sources. Overflows from STW failure or overcapacity events (e.g., floods, excessive water use) lead to direct, untreated introduction of sewage into the environment. In efforts to improve tributary conditions (by increasing stream flow), some cities have considered increasing the percentage of annual overflow events (e.g., see the Portland, Oregon, proposal (20). The highest concentration in an STW effluent reported by Ternes (18) was for bezafibrate (4.6 µg/L); the highest concentration in surface water also was for bezafibrate (3.1 µg/L ppb).

Landfills. PPCPs can be introduced to landfills both directly via domestic and industrial routes and indirectly via sewage sludge. Holm et al. (21) first reported leachates carrying pharmaceuticals from a landfill. Large amounts of numerous sulfonamides (antibiotics) and barbiturates from domestic waste and from a pharmaceutical manufacturer were disposed of at a Danish landfill over a 45-year period. High concentrations (ppm) of many of these drugs were found in leachates close to the landfill; these compounds even accounted for 5% of the total nonvolatile organic carbon found in the leachate. It was also found that the concentrations dropped off dramatically tens of meters down gradient, presumably a result of microbial attenuation.

Drinking water. Few pharmaceuticals have been identified in domestic drinking water, probably because of the dearth of monitoring efforts and because the required detection limits are too low for current routine analytical technology. In Germany, however, clofibric acid concentrations up to 165 ng/L (22) and 270 ng/L (23) have been measured in tap water; the presumed source was from recharged groundwaters that had been contaminated by sewage. Stumpf et al. (24) and Ternes et al. (9) found several pharmaceuticals in German drinking water in the lower nanograms-per-liter range, with a maximum of 70 ng/L for clofibric acid. Additionally, these investigators found that diclofenac, bezafibrate, phenazone, and carbamazepine were sometimes present. In the majority of the samples analyzed, however, no drugs were observed. The investigations performed to date therefore indicate that contamination of drinking water does not appear to be a general problem. Depending on the water source for drinking water production, however, certain facilities can experience contamination, especially if the source is polluted groundwater and if polishing technology does not remove the PPCP [e.g., see Heberer et al. (23) and Stumpf et al. (24)]. A major unaddressed issue regarding human health is the long-term effects of ingesting via potable waters very low, subtherapeutic doses of numerous pharmaceuticals multiple times a day for many decades. This concern especially relates to infants, fetuses, and people suffering from certain enzyme deficiencies (which can even be food-induced, e.g., microsomal oxidase inhibition by grapefruit juice).

Drinking water regulations. Regulations designed to safeguard receiving waters (from sewage treatment) and drinking water were historically designed to protect the consumer from the obvious threats of pathogens, widely used industrial chemicals, and certain radionuclides. The treatment processes used by state-of-the-art POTWs evolved from the need to remove these limited sets of contaminants. In areas of water scarcity, the future will see more and more reuse of treated sewage to meet drinking water needs. This will impose a severe burden on water providers to ensure that all chemical contaminants have been removed to the greatest extent possible. It will also require the ability to identify as many of the plethora of potential chemicals in the upgraded water as possible.

According to the National Research Council (NRC) (25), more than two dozen major U.S. utilities release so much effluent to receiving waters that when the natural flows are low, the discharged waste composes 50% of the eventual flow. Any residual, unidentified contaminants therefore are diluted 2-fold at best. In more densely populated countries (e.g., United Kingdom), this figure can rise as high as 90% of flow during times of low rainfall (26).

Domestic animals. Whereas the concentration of many drugs is greatly attenuated through sewage treatment plants, larger quantities of many pharmaceuticals are used in various animal husbandry operations, especially CAFOs. With aquaculture in particular, which uses many anti-infectives and anesthetics, the chance for introduction into the immediate environment is greatly enhanced, and the possibility of direct human consumption of therapeutic quantities is correspondingly heightened. Even in the United States the extremely large populations of pet dogs and cats are recipients of numerous drugs (e.g., tranquilizers and antidepressants)--some prescribed by veterinarians and others intended for their owners' use as pet owners sometimes administer medications to their pets to test off-label uses for themselves. PPCPs (both veterinary drugs and OTC products) used with terrestrial domestic animals can be dispersed into the environment through the same routes as those PPCPs used for humans, with the added major route of run-off/leaching of on-ground fecal material.

Significant Aspects of Ecotoxicology

Shortcomings of effluent toxicologic screening: comprehensive chemical characterization cannot be replaced--chemical characterization and toxicity screening must be better integrated. There are two debates in the realm of ecotoxicology, both of which have ramifications with respect to performing ecologic risk assessments (ERAs) for PPCPs. The first is the relevance of purposefully simplified, defined-species toxicity tests to predicting/extrapolating pollutant impacts on the more highly organized and complex structural/functional levels of communities or ecosystems (processes) [see Boudou and Ribeyre (27)]; this is truer for PPCPs than for pesticides, as the former were generally never designed to have any intended effects on wildlife and therefore any knowledge as to what types of effects to look for is clearly more limited. Can changes in a complex system be predicted from knowledge of a small subset of the underlying components? The second is the question of whether it is necessary to know the spectrum of possible physiologic effects, given a multitude of organisms, or possible mechanisms (modes) of action before looking for and ascribing causation to changes at the population level and higher. Considering this, one can only pose at this time the rhetorical question as to whether the risk posed by the presence of pollutants in complex waste streams (e.g., PPCPs in STW effluents) can be detected/quantified by the use of current toxicity screening tests never designed to embrace the spectrum of end points (some exquisitely subtle) that may be involved. The most conservative approach would be one that captures the coordinated use of toxicity-directed screening and chemistry-directed characterization, feeding the results of each to the other, to better reveal the nature of any stressors.

Although most pharmaceuticals are designed to target specific metabolic pathways in humans and domestic animals, they can have numerous often unknown effects on metabolic systems of nontarget organisms, especially invertebrates. Although many nontarget organisms share certain receptors with humans, effects on nontarget organisms are usually unknown. It is important to recognize that for many drugs, their specific modes of action even in the target species are also unknown. For these drugs, it is impossible to predict what effects they might have on nontarget organisms. Without knowing the mode of action, coupled with not knowing the possible receptors, it is impossible to design rational toxicity testing procedures at the organism level. In the final analysis, given the vast array of mechanisms of drug action and side effects, the total number of different toxicity tests possibly required to screen the effluent from a typical STW could be impractically large. The current batteries of acute/chronic toxicity tests used for ecotoxicity screening merely supply gross indications of directly measurable acute effects. Even if the known mode of action is considered when selecting ecotoxicity tests [as recommended by Henschel et al. (28)], this falsely presupposes that other modes of action are nonexistent or nominal.

Regulatory agencies only in the last few years have recognized that pharmaceuticals should be screened to determine possible effects on nontarget species. The world's first requirement for ecotoxicity testing as a prerequisite for registration of a pharmaceutical was established in 1995 and first implemented in Germany according to European Union (EU) guideline 92/18 EWG for veterinary pharmaceuticals. For a more in-depth discussion, see Henschel et al. (28), and for a general discussion of the issues in aquatic ecotoxicology, see Boudou and Ribeyre (27).

Screening waste effluent and receiving waters for toxicologic effects can at best be only partially effective because the range of physiologic effects is too broad and relevant to a vast array of aquatic and terrestrial organisms, spanning everything from acute toxicity to very subtle behavioral or genetic changes, of which the consequences are not immediately manifested and can be detected only over long periods of time. There are too many scenarios to discuss in an efficient, comprehensive manner. The complexity of accounting for a wide range of mechanisms of action was made clear in the National Research Council's recent report on endocrine disruptors (29). Although for this class of pollutants the number of modes of action is very large, they represent only a subset of those for PPCPs in general. Quite clearly, any successful toxicity-directed methodology for risk assessment of complex effluents or environmental samples should also make use of a well-developed knowledge of the chemical constituents and their modes of action; current approaches are not yet sufficiently comprehensive. The complexity of this task is further magnified when the effect and necessarily its mode of action have not even been elucidated.

A popular means of attempting to identify the toxic constituents, using toxicity identification and evaluation, in complex waste such as sewage effluent is that of bioassay-directed fractionation screening (30), in which chemical separation techniques yield distinct chemical-class fractions that are then subjected to toxicity testing. Those fractions showing activity against the selected end point are then subjected to chemical identification protocols. Even if one accepts the limitations of selecting appropriate end points (the number with environmental relevance would be enormous), this extremely time-consuming approach would miss any combined effects, whether antagonistic or synergistic, of multiple chemicals. Direct, rigorous chemical characterization of problematic samples clearly must play a role in the identification of toxicants that might elicit previously unrealized toxic effects in nontarget organisms.

The trend toward optically pure pharmaceuticals: fewer side effects and lower concentrations. Most pharmaceuticals are racemic mixtures. For a specific optically active drug, it is theorized that only one of its optical isomers is responsible for the desired physiologic, therapeutic effects; the other isomers are at best inactive, or even worse, responsible for many of the untoward side-effects that most drugs display. A recent trend in the pharmaceutical industry, and now supported by the FDA, is to produce only the optically pure therapeutic isomer (31). This has the potential to not only lessen side effects, but for some drugs, the total dosage can be lessened by at least 50%. This could help in reducing the burden on sewage treatment plants. The significance of the industry's switch to optically pure isomers is that the number of metabolites and other SRSs entering the environment will be reduced at least by half, and the use of the active ingredient will also be reduced by at least 50% because the potency will effectively increase. At the same time, however, the trend of pharmaceuticals toward higher potency will increase the difficulty of environmental monitoring because the required detection levels will be lowered.

Synergistic effects and potentiation: the potentially critical role of "multixenobiotic resistance." The biochemical interactions of drugs, often leading to adverse effects, is well known in humans. Little is known, however, of this interplay in aquatic organisms. The following is provided as an example of the complex potential for adverse drug interactions (one actually leading to increased exposure), as it also illuminates the interwoven pathways that ultimately determine exposure. Mostly during this decade, a new mechanism for elimination of xenobiotics from organisms (first observed in tumor cells) has been elucidated--multidrug transporters. This excretory system, also called multixenobiotic transporters, comprises proteins that facilitate the active export of potentially toxic substances, primarily those of moderate lipophilicity, from inside cells. The best-known transporters are the P-glycoproteinlike (Pgp) transporters (P is for permeability altering), or P170 (because of their 170-kDa mass), which have been well characterized in mammals, especially tumor cells, and bacteria.

The toxicologic significance of these nonspecific transporters in maintaining a first line of defense against exposure to multiple xenobiotics in aquatic species has been largely pioneered and reviewed by Epel (32) and by Kurelec and co-workers (33-35); this system confers what has become known as multidrug or multixenobiotic resistance (MDR or MXR). Although these protective proteins have not been found in all aquatic organisms, they have been found in many, especially filter feeders and bottom dwellers (those having potentially high exposures to xenobiotics). This extrusion pump protein system, and possibly others as yet identified, facilitate the removal and prevent the entrance of those compounds not metabolized or conjugated. They seem to have nonspecific recognition, working for many pesticides, drugs, and natural toxins alike. The action of this transporter system can be inhibited by certain substances such as verapamil (-[3-[[2-(3,4-dimethoxyphenyl)ethyl]methylamino]propyl]-3,4-dimethoxy--(1-methylethyl)benzene acetonitrile), a cardiac drug--calcium ion influx inhibitor--that directly binds to the active site of Pgp. Exposure to verapamil at micromolar concentrations and lower greatly increases the toxicity of a number of drugs or other xenobiotics for many aquatic organisms (32), as the toxicant cannot be readily removed from the exposed organism; exposure time is thereby lengthened by its intracellular accumulation. This elimination system does not function for highly hydrophobic substances (e.g., DDT, polychlorinated biphenyls [PCBs]) and as such might play a more critical role in eliciting effects from exposure to the less hydrophobic PPCPs. Xenobiotics may irreversibly inhibit (cyclosporine A inhibits ATPase), competitively inhibit (verapamil, quinidine, reserpine at low concentrations or high concentrations of general lipophilic compounds such as petroleum oil), or indirectly modulate (e.g., via phosphorylation) MXR regulation or expression (staurosporine inhibits protein kinase C Pgp regulator), resulting in its reversal.

The slow escalation, by induction or genetic enrichment, of MXR occurrence and activity among aquatic organisms can give the illusion that the toxicity potential in the aquatic environment is stable or even decreasing when in reality it may be increasing. The introduction of a new substance, at what would normally be a no-effect level, that disrupts the activity of MXR could thereby lead to a profound cascade of unanticipated and unaccounted-for toxic events--a phenomenon akin to what is being termed toxicant-induced loss of tolerance in humans. Organisms in an aquatic environment that have adapted via MXR to certain levels of a suite of toxicants could experience widespread interspecies toxic events should their MXR be inhibited by the addition of a single agent capable of inhibiting MXR, even one that ordinarily would elicit no effect on its own. The resulting effects would be inexplicable if considered solely on the basis of exposure to the new toxicant.

Little is known about which xenobiotics have activity within this relatively newly identified class of chemicals, referred to as chemosensitizers, or their frequency of occurrence in the environment. Smital and Kurelec (35), however, showed that unidentified agents in samples from polluted waters enhance the accumulation of aromatic amines in clams, mussels, snails, and sponges. Some examples of known MXR inhibitors (34,35), other than verapamil include trifluoroperazine (Stelazine, which is a calmodulin antagonist and an antipsychotic tranquilizer), reserpine (antihypertensive), quinidine and amiodarone (anti-arrythmics), cyclosporins (immunosuppressants), anthracyclines (noncytotoxic cytotoxin analogs), and progesterone (steroid); some natural substances such as agent(s) in grapefruit juice are also known to inhibit the P-glycoprotein system (36).

Environmental Studies on Pharmaceuticals

Given the numbers and quantities of pharmaceuticals manufactured and used throughout the world and that many of these chemicals are designed to have profound physiologic effects, comparatively little research has been published on their occurrence in the environment, effects on nontarget organisms, or assessment of environmental impact. Literally thousands of distinct drugs are approved for use throughout the world. Many of these are manufactured and used in very large quantities. The world's combined literature (the vast majority of these studies have originated in Europe, but the issue applies equally worldwide) has addressed only a very small percentage of these compounds, and the huge array of associated metabolites and other transformation products, many of which undoubtedly have strong physiologic activity, simply compounds the magnitude of the problem.

When drugs are detected in the environment (e.g., surface waters), their concentrations are generally in the ng/L-µg/L (ppt-ppb) range. Although parts-per-billion concentrations may not pose much acute risk, it is completely unknown whether other receptors in nontarget organisms are sensitive. It must also be recognized that even though individual concentrations of any drug might be low, the combined concentrations from drugs sharing a common mechanism of action could be substantial. Exposures in the aquatic environment are of particular concern, since aquatic organisms (as opposed to those spending at least some time in terrestrial settings) are subject to continual, unabated life-cycle exposures. This is a highly significant consideration for pharmaceuticals (or bioactive metabolites) that are refractory to structural transformations and are continually introduced into surface waters from sewage treatment plants. Moreover, the polar, nonvolatile nature of most drugs prevents their escape from the aquatic realm. Effectively, even PPCPs with relatively short environmental half-lives assume the qualities of highly persistent pollutants because they are continually replenished by infusion to the aquatic environment from STWs.

Environmental Occurrences

(Note: The names, structures, Chemical Abstracts Service Registry Numbers, and some of the data for environmental occurrences cited in this paper are summarized in Table 2.) Probably the first report of a prescription drug in the environment (sewage treatment effluent) was made over 20 years ago by Garrison et al. (37), who reported clofibric acid (the active metabolite from the lipid regulators clofibrate, etofibrate, and theofibrate) concentrations of 0.8-2.0 µg/L in raw sewage and activated sludge effluent. They also found the ubiquitous caffeine and nicotine to be the two most prevalent compounds in influent and effluent from activated sludge, but they did not find the parent clofibrate in any sample. In parallel, Hignite and Azarnoff (38) reported salicylic acid and clofibric acid in the influent and effluent from a Kansas City, Missouri, municipal sewage treatment plant [the history of clofibric acid identified in the environment has been summarized by Stan and Heberer (39)]. Clofibric acid was routinely detected in the effluent of this Missouri STW at an average effluent rate of 2.1 kg/day; over a 10-month period its loading remained in the tight range of 0.76-2.92 kg/day. Similarly, salicylic acid, a hydrolytic metabolite of aspirin, averaged 8.6 kg/day but ranged more widely from 0.55 to 28.7 kg/day. Stan and Heberer also observed that the influent concentrations of clofibric acid were only 20% higher than the effluent concentrations, showing that this chemical resisted removal by the STW. In contrast, for salicylic acid, the influent concentration was about an order of magnitude higher than the effluent, showing more efficient removal.

It therefore was clearly recognized over 20 years ago that the continual, daily introduction of kilogram quantities of drugs from a given STW into receiving waters could result in sustained concentrations with the potential to lead to exposures in aquatic organisms. Little more transpired in the literature, however, during the next 15 years, although clofibric acid continued to appear in a number of monitoring efforts that did not target PPCPs. The most complete investigation to date of the occurrence of pharmaceuticals in both the influent and effluent of POTWs (and also in various surface waters) has been published by Ternes (18).

The distribution of pharmaceuticals is a large function of their production volumes, which can rival those for many pesticides. There are thousands of registered drugs that are dispensed both as prescriptions and OTC; this makes it difficult to estimate usage rates for those pharmaceuticals sold via both routes (e.g., many analgesics). In Germany, roughly 2,900 drugs are permitted in human medicine alone (18). Many countries dispense drugs in the absence of prescriptions. The two primary sources for release into the environment are from human and veterinary applications. Ternes (18) states that at least for lipid regulators and NSAIDs the source is almost entirely from human usage, as these drugs are infrequently (or never) used in veterinary medicine. In general, the literature shows that most pharmaceuticals, when detected, are present in surface waters in a concentration range of 1 ng/L-1 µg/L. To put this in perspective, Richardson and Bowron (40) state that 1,000 kg of a chemical distributed evenly among the rivers in England and Wales would yield a concentration of about 0.1 µg/L. Many pharmaceuticals are consumed in amounts far exceeding this; in fact, Richardson and Bowron report 170 pharmaceuticals used annually in excess of this amount.

Terrestrial and Atmospheric Exposure

Minor route for PPCPs in contrast to pesticides. The majority of PPCPs introduced into the environment is undoubtedly into aquatic systems; the terrestrial environment receives only a secondary input. Although the primary source for terrestrial exposure is probably from disposal of biosolids from STWs and from animal wastes both applied to land and stored in open-air pits (waste lagoons), other possible sources for veterinary pharmaceuticals result from animal dips and direct deposition of dung from medicated animals. To date, most attention has been focused on the application of animal wastes to land, primarily because of the suspected introduction of antibiotics and nutrients, not because of PPCPs other than veterinary antibiotics, which are used in comparatively smaller amounts. It should be noted, however, that even though the introduction of veterinary antibiotics into the environment, both terrestrial and aquatic, via animal wastes is widely discussed, the topic has experienced little attention in the peer-reviewed literature (41,42). This topic also relates directly to the human health concern of introducing/promoting antibiotic resistance in bacteria, both native to and introduced into the environment [see section on "Antibiotics" and Williams and Heymann (43)].

The polar nature of the majority of drugs/metabolites leads to facile leaching from land disposal areas into groundwater or wet weather runoff into surface waters. The remainder (largely those designed to pass the blood-brain barrier) have lipophilic character, rendering them prone to bioconcentration from consumption of water or bioaccumulation from consumption of tissue.

Dung-feeding fauna such as birds, beetles, worms, flies, and microorganisms could experience immediate exposure to excreted terrestrial veterinary pharmaceuticals and metabolites. These organisms in turn could suffer effects themselves from exposure or, alternatively, pass on accumulated residues further up the food chain. All other routes of dispersal to other environmental compartments also play roles, with the distinct exception of direct volatilization, because nearly all PPCPs, with the exception of medical gases and fragrances in contrast with many other anthropogenic compounds are polar or otherwise nonvolatile. The major volatile pharmaceuticals are the inhalable anesthetics (e.g., halothane); these hydrofluoroalkanes are known to oxidize in the atmosphere, like the conventional hydro[chloro]fluorocarbon refrigerants, to yield the highly persistent, toxic, and ubiquitous product trifluoroacetic acid (TFA). This source of TFA is believed to be minor (44).

Drug Classes and Environmental Occurrences

Hormones/Mimics

Potential for receptor interaction may not be rare. An excellent overview of hormone systems is given by the Endocrine Disruptor Screening and Testing Advisory Committee (EDSTAC) (45). Steroids were the first physiologic compounds to be reported in sewage effluent (46-49) and as such were the first pharmaceuticals to capture the attention of environmental scientists. Estrogenic drugs, primarily synthetic xenoestrogens, are used extensively in estrogen-replacement therapy and in oral contraceptives, in veterinary medicine for growth enhancement, and in athletic performance enhancement. A special issue of The Science of the Total Environment (50) is devoted to drugs (especially hormones) as pollutants in the aquatic environment.

Although the synthetic oral contraceptive (17-ethynylestradiol) occurs generally at low concentration (< 7 ng/L) in POTW effluent, it is still suspected, in combination with the steroidal estrogens 17ß-estradiol and estrone [30], of causing vitellogenin production (feminization) in male fish. Feminization is a phenomenon first observed for fish in sewage treatment lagoons in the mid-1980s (26). An overview of pharmaceutical hormones in the environment is presented by Arcand-Hoy et al. (51). The estrogenic activity of various waters (from sewage to drinking water) has been shown to vary dramatically, spanning six orders of magnitude. Some other widely used synthetic hormone modulators include Proscar/Propecia (finasteride: an androgen hormone inhibitor) and various thyroxine analogs (thyroid hormones); nothing is known of the environmental fates of these compounds. In general, the lipophilicity of these hormones is sufficiently great that at least a large portion are removed via sorptive processes in sewage treatment (52,53) and therefore partition to the sludge; but even the low concentrations that remain in the effluents may be capable of exerting physiologic effects in aquatic biota.

In addition to these synthetic steroids and xenoestrogens is a suite of naturally occurring estrogen hormones, for example, phytoestrogens such as the complex series of leguminous isoflavonoids, including genistein, daidzein, and glycitein in soy. Further complicating the picture are a host of newly suspected endocrine-disrupting compounds (EDCs), more recently referred to as hormonally active agents (HAAs) by the NRC (29), which have gained attention in the last few years, primarily as a result of the 1996 publication Our Stolen Future by Colburn et al. (54). These inadvertent EDCs include such commonly recognized industrial pollutants and products as halogenated dioxins/furans, PCBs, organohalogen pesticides, phthalates, and bisphenol A.

The issue of screening many of the major commercial chemicals (over 87,000 total) for endocrine disruption potential has been formalized with the creation of the EDSTAC, which had been charged by the U.S. EPA with the task of implementing a screening and testing program by August 1999 (45). The Chemical Manufacturers Association (CMA) also has launched an intensive health effects investigation for over 3,000 high-volume chemicals (called the Health and Environmental Research Initiative) (55). It is significant, however, that pharmaceuticals are not specifically targeted by the EDSTAC (or the CMA) in its tiered screening program that focuses on pesticides, commodity chemicals, naturally occurring nonsteroidal estrogens (phytoestrogens and mycotoxins), food additives, cosmetics, nutritional supplements, and representative mixtures (for possible synergistic effects). Even though the strategy gives top priority to "chemicals with widespread exposure at the national level" (55), PPCPs are not specially targeted. It is also significant that the screening strategy will initially focus on only the three primary hormone systems--estrogen, androgen, and thyroid--hormone systems of relatively unknown importance to invertebrates (45).

A controversial hypothesis regarding multiple toxicants (sharing a common mode of action), when each is present at a low level, is that of synergism. Evidence of synergism among estrogenic mimics (where the effect can be elicited at orders-of-magnitude lower concentration than predicted by additive action) was reported by Arnold et al. (56). This study created much controversy by purporting synergistic action of low-level chemical mixtures. Subsequent studies by Gaido et al. (57) and others rebutted this hypothesis. They did not find any evidence of synergism in mixtures of mild estrogenic pollutants. McLachlan (58) later withdrew the article by Arnold et al. (56), but the issue has not been put to rest, especially given Arnold's other publications on this subject including Arnold et al. (59) and references cited therein. Another controversial issue is that of inverted (U-shaped) dose-response in which toxicity diminution tracks lower concentrations down to a certain level, at which point toxicity again increases. Consequently, higher dose effects might not be useful in predicting the type or magnitude of effects from lower doses (29). This unresolved issue, coupled with the controversy of whether toxicity thresholds necessarily exist, could severely impede EDSTAC's ability to reach its objective because the concentration ranges that must be investigated would be greatly expanded.

Low molecular weight nonpeptidyl molecules can mimic hormones. Another subclass of hormonelike substances includes those that are being purposefully designed to mimic the activity of therapeutically significant hormones. A long-sought objective has been to obviate the need for hormone-replacement therapy (e.g., insulin) by designing small synthetic (nonpeptidyl) molecules that mimic the hormone's effect yet can be ingested orally, taken up by the gut, and remain stable for a sufficiently long period of time in the blood. The first report of a "designer" hormone mimic (60,61), a polybenzimidazole that activates the receptor for a cytokine that regulates white blood cell production, perhaps portends the advent of many synthetic hormone mimics in therapeutic medicine. If the finding can be generalized, it could mean that the possible routes of hormone disruption by simple molecules could extend beyond that of the estrogen/androgen system.

With the exception of estrogenic mimics, the possibility of disrupting the activity of proteinaceous hormones by lower molecular weight anthropogenic chemicals has been held in low regard. This view has been based on the fact that a relatively large, complex proteinaceous molecule (the hormone) neatly "fits" within the complex three-dimensional domain of its target receptor, whereas in contrast a much smaller nonproteinaceous molecule would have little to offer in terms of recognition specificity. It has been believed that the complexity of larger proteins such as insulin was required to enable recognition by the corresponding receptors; smaller compounds simply did not convey enough three-dimensional information to have high-binding constants for one or multiple receptors.

The report by Tian et al. (60) demonstrates for the first time that a relatively small nonpeptide molecule can bind to a receptor normally dedicated to a proteinaceous hormone. While this has high therapeutic significance (this research might catalyze concerted attempts to develop the first protein-mimicking and therefore perhaps hormone-mimicking low molecular weight drugs), it also alludes to the possibility that existing anthropogenic compounds might have a greater chance of interacting with hormone receptors than was previously believed. Although the synthetic substance was three to six orders of magnitude less potent, its ability to bind to the receptor was undisputed (in the mouse in vitro and, more importantly, in vivo).

Antibiotics

In addition to pathogen resistance, genotoxicity may be a concern. A large body of literature exists on antibiotics in the environment. Veterinary and animal husbandry, especially aquaculture, usage plays a major role in their introduction into the environment. In one study of hospital effluent, fluoroquinolones was the chemical class contributing the major portion to overall DNA toxicity (62); ciprofloxacin, for example, was identified at 3-87 µg/L. Hirsch et al. (41) analyzed German STW effluents and groundwaters/surface waters for 18 antibiotics representing macrolides, sulfonamides, penicillins, and tetracyclines. Although the penicillins (susceptible to hydrolysis) and the tetracyclines (can precipitate with calcium and similar cations) were not found, the others were detected in the microgram per liter range. Indeed, the rampant, widespread (and sometimes indiscriminate) use of antibiotics, coupled with their subsequent release into the environment, is the leading proposed cause of accelerated/spreading resistance among bacterial pathogens, which is exacerbated by the fact that resistance is maintained even in the absence of continued selective pressure (an irreversible occurrence). Sufficiently high concentrations could also have acute effects on bacteria. Such exposures could easily lead to altered microbial community structures in nature and thereby affect the higher food chain. Their use in aquaculture results in eventual human consumption. For a discussion of promotion of antibiotic resistance, see the policy article by Witte (63). Hartmann et al. (62) propose that genotoxicity in hospital effluent may result more from antibiotics than from antineoplastics.

Recently, a number of stream surveys documented the significant prevalence of native bacteria that display resistance to a wide array of antibiotics including vancomycin (64). Isolates from wild geese near Chicago, Illinois, are reported to be resistant to ampicillin, tetracycline, penicillin, and erythromycin (65). All these reports could simply indicate that the natural occurrence of antibiotic resistence in native bacterial populations is much higher than expected or that these bacteria are being selected for by the uncontrolled release of antibiotics into the environment. If the latter is true then, excluding the significance of antibiotics themselves in the environment, their occurrence can be viewed as marking or indicating the possible presence of other PPCPs.

Blood Lipid Regulators

Fibrates--high usage. Fibric acid metabolites--ubiquitous, persistent pollutants. Clofibric acid was the first prescription drug (actually an SRS) reported in a sewage effluent (37,39), and it continues to be one of the most frequently reported PPCPs in monitoring studies. Clofibric acid (2-[4]-chlorophenoxy-2-methyl propanoic acid), the active metabolite from a series of widely used blood lipid regulators, and which also happens to be structurally related to the phenylalkanoic acid herbicide mecoprop (the methylphenoxy structural analog), has captured much attention from investigators in Europe. Stan et al. (22) first reported clofibric acid in Berlin tap water at concentrations between 10 and 165 ng/L. Heberer and Stan (66) found clofibric acid at levels up to 4 µg/L in groundwater under a sewage treatment farm; they also found clofibric acid concentrations up to 270 ng/L in drinking water samples. They concluded that it is not removed by sewage/water treatment processes.

Buser et al. (67) report finding clofibric acid in various Swiss waters ranging from rural to urban lakes. Concentrations ranged from 1-9 ng/L (ppt), whereas the parallel concentrations for mecoprop were higher at 8-45 ng/L; little of either compound was found in a relatively remote mountain lake, indicating no atmospheric deposition. Because this drug is not manufactured in Switzerland, its route of introduction into the environment had to be through medical use and subsequent excretion/disposal. Although these concentrations are very low, they are significant in that they are similar to the concentrations found for any of the conventional ubiquitous and persistent pollutants, sometimes referred to as persistent organic pollutants (POPs) or persistent bioaccumulative toxicants (PBTs) such as lindane [see Jones and de Voogt (68) for an overview]. In one of the lakes studied, Buser et al. (67) calculated steady-state amounts of clofibric acid to be roughly 19 kg (with export and import amounts balancing each other). Perhaps more significantly, they also found amounts of clofibric acid up to 7.8 ng/L in the North Sea; the parallel concentrations of mecoprop in the same North Sea samples were lower, up to only 2.7 ng/L, indicating that mecoprop was less persistent than clofibric acid.

Stumpf et al. (24) and Ternes (18) reported bezafibrate, gemfibrozil, and clofibric/fenofibric acids in river waters at the nanogram per liter level. Stumpf et al. (69) reported that the removal efficiencies from Brazilian STWs for clofibric/fenofibric acids, bezafibrate, and gemfibrozil ranged from only 6-50%, verifying extremely limited degradation for these compounds. This chemical class is ubiquitous because the daily human dosages are generally high (grams per day). Buser et al. (67) concluded that the concentrations seen in urban Swiss and German rivers, coupled with essentially the same concentrations in the North Sea, lead to an annual input of 50-100 tons of clofibric acid into the North Sea. The concentration of clofibric acid in the environment is more a function of dilution than of degradation. Clofibric acid is the most widely and routinely reported drug found in open waters. It would be expected that its occurrence in other parts of the world would parallel these studies.

Nonopioid Analgesics/Nonsteroidal Anti-Inflammatory Drugs

Stumpf et al. (24) were the first to identify diclofenac, ibuprofen, acetylsalicylic acid, and ketoprofen in sewage and river water. Ternes (18) reported levels of diclofenac, indometacine, ibuprofen, naproxen, ketoprofen, and phenazone in POTW effluent exceeding 1 µg/L; all these except ketoprofen were also found in surface waters at concentrations severalfold lower. In another study, Ternes et al. (70) reported average concentrations of acetylsalicylic acid generally less than 1 µg/L in most POTW effluents as well as less than 0.14 µg/L in rivers. They also reported salicylic acid concentrations of 54 µg/L in POTW influents, with two other acetylsalicylic metabolites, gentisic acid (4.6 µg/L) and o-hydroxyhippuric acid (6.8 µg/L). While low levels (0.5 µg/L) of salicylic acid appeared in the effluents, no detectable amounts of the metabolites could be found. Ternes et al. (70) also found naproxen in all POTW effluents examined and in river waters (~0.05-0.4 µg/L); two veterinary NSAIDs, meclofenamic and tofenamic acids, were not detectable in any river sample. In their screening of waters in Berlin, Heberer et al. (23) found that the most prevalent drugs, other than clofibric acid, were the NSAIDs diclofenac, ibuprofen, and propyphenazone. In groundwater from a drinking water plant, they found diclofenac, ibuprofen, and N-methylphenacetin (from phenacetin) (23). In the influent to Swiss STWs, Buser et al. (71) found diclofenac at concentrations of 0.5-1.8 µg/L, whereas the concentrations in the respective effluents were only moderately reduced (at most 50%). In the receiving water (Swiss lakes/rivers), they found 11-310 ng/L but only 1-12 ng/L in exiting waters. They concluded that photolysis was the major cause of the diminished concentrations of diclofenac in surface waters (71). Buser et al. (72) showed that ibuprofen, while present in influents at 1-3.3 µg/L, was easily degraded to yield low effluent concentrations (nanograms/liter) in contrast to other NSAIDs, which were more refractory. This study is also one of the few that examined the enantiomeric selectivity in the degradation of the parent optical isomers as well as the production of metabolites.

Beta-Blockers/ß₂-Sympathomimetics

Hirsch et al. (73) and Ternes (18) identified the beta-blockers metoprolol and propranolol, with lesser amounts of betaxolol, bisoprolol, and nadolol, in POTW effluent. Only metoprolol and propranolol were found in surface waters at concentrations just above the limit of detection. The ß₂-sympathomimetics (bronchodilators) terbutalin and salbutamol (albuterol in the United States), but rarely clenbuterol and fenoterol, were detected in POTW effluent and only at low concentrations, less than 0.2 µg/L. They were rarely seen in surface waters. It may be significant to note that medications delivered by inhalers could result in portions of the dose being deposited externally because of improper dosing technique.

Fenfluramine (N-ethyl--methyl-3-[trifluoromethyl] benzene ethanamine hydrochloride), known as Pondimin in addition to other brand names, is a sympathomimetic amine, which was used as a popular diet (anorectic) drug and was removed from the U.S. market in 1998 by the FDA because of heart valve damage. Although no one has looked for fenfluramine in sewage, it is known to enhance the release of serotonin (3-(2-aminoethyl)indol-5-ol or 5-hydroxytryptamine creatinine sulfate [5-HT]); in the crayfish, 5-HT in turn triggers release of ovary-stimulating hormone, resulting in larger oocytes with enhanced amounts of vitellin (consequences unknown) (74). Similarly, in fiddler crabs, fenfluramine at a dose of 125 nmol stimulates (through 5-HT) the production of gonad-stimulating hormone, which accelerates testicular maturation (75).

Antidepressants/Obsessive-Compulsive Regulators

Subtle but possibly profound effects on nontarget [aquatic] species. Selective serotonin reuptake inhibitors (SSRIs) are a major class of widely prescribed antidepressants that includes Prozac, Zoloft, Luvox, and Paxil. These drugs enjoy widespread and heavy use. One of the few series of studies reported in the literature that addresses the effects of drugs on nontarget organisms (albeit not the intent of the studies) was performed in a quest for more effective spawning inducers for economically important bivalves (76). Fong's studies and those of other physiologists studying the function of serotonin in a wide array of aquatic creatures could prove highly significant in any discussion of the importance of low levels of pharmaceuticals in the environment. Fong's work is perhaps the most significant to date for showing the potential for dramatic physiologic effects on nontarget species (in this case invertebrates) by low (ppb) concentrations of pharmaceuticals.

Serotonin is a biogenic amine common in both vertebrate and invertebrate nervous systems. SSRIs increase serotonin neurotransmission by inhibiting its reuptake at the synapses by inhibiting the transporter enzymes. In addition to playing a key role in mammalian neurotransmission, serotonin is involved in a wide array of physiologic regulatory roles in molluscs, among most other creatures. For bivalves, reproductive functions including spawning, oocyte maturation, and parturition are regulated by serotonin, (76). Serotonin controls a wide spectrum of additional behaviors and reflexes in molluscs, including heartbeat rhythm, feeding/biting, swimming motor patterns, beating of cilia, and induction of larval metamorphosis (77). It also stimulates release of various neurohormones in crustaceans (hyperglycemic hormone, red pigment-dispersing hormone, neurodepressing hormone, and molt-inhibiting hormone) and ovarian maturation (78).

It has long been known that serotonin at concentrations of 10^-4 to 10^-3 M (~0.18-1.8 g/L) induces spawning in bivalves. Some commercial farmers make use of this by adding serotonin to induce spawning. Fong (76) found that Prozac (fluoxetine) and Luvox (fluvoxamine) are the most potent inducers ever found, eliciting spawning behavior in zebra mussels at aqueous concentrations many orders of magnitude lower than serotonin. Fluoxetine elicited significant spawning in male mussels at concentrations of 10^-7 M (~150 µg/L); females were an order of magnitude less sensitive at 10^-6 M. Fluvoxamine was the most potent of the SSRIs, eliciting significant spawning in male mussels, at 10^-9 M (~0.318 µg/L); females were two orders of magnitude less sensitive, at 10^-7 M. In males, spawning was complete in the first hour, while females were slower (within 2 hr). Paxil (paroxetine) was the least potent of these three SSRIs, eliciting male spawning, but to a lesser degree, at 10^-6 M, and having no inducing effect on females at any concentration. It should be noted that Fong states that the evidence is not clear whether these compounds are indeed acting as SSRIs, or via some other mechanism. It is also unknown how these compounds are taken up by molluscs (76).

In another study, Fong et al. (79) showed that fluvoxamine induces significant parturition in fingernail clams at 1 nM; 1 nM fluvoxamine also potentiated the effect of 10 µM 5-HT by almost 5-fold. Paroxetine was less potent, requiring a concentration of 10 µM to effect significant parturition. In contrast, even at concentrations of 100 µM, fluoxetine displayed no effect, although it was capable at 5 µM of potentiating 5-HT at concentrations that were otherwise subthreshold. It is interesting that the order of potency for inducing parturition in clams differs from the order for induction of spawning in mussels (above). This points to the complexity of considering any approach involving extrapolations from one species to another or from one drug to another within a given class.

In crustaceans, Kulkarni et al. (74) found that fluoxetine significantly potentiates the effect of 5-HT in crayfish, enhancing the release of ovary-stimulating hormone, which results in larger oocytes with enhanced amounts of vitellin; any ecologic consequences of higher vitellin protein levels are unknown. Similarly, in fiddler crabs, fluoxetine at a dose of 125 nmol stimulates (through 5-HT) the production of gonad-stimulating hormone, which accelerates testicular maturation (75).

It is clear that aquatic life can be exquisitely sensitive to at least some of this class of compounds. Although some SSRIs are extremely potent, others have almost no effect, which possibly makes the approach of assessing ecologic risk on a class-by-class basis infeasible.

Concentration of SSRIs plays a complicated role with respect to effects. For example, Couper and Leise (77) found that while injected fluoxetine induced significant metamorphosis in a gastropod, 10^-4 M induced less metamorphosis than 10^-6 M. Simple extrapolations of effects from higher concentrations do not necessarily have any relevance to effects at lower concentrations.

The potential for SSRIs to elicit subtle effects on aquatic life is further extended by serotonin reuptake mechanisms that also are a factor in snails and squids (76), particularly in the regulation of aggression (80). Yet another example of a subtle effect that would go unnoticed is the fighting behavior of lobsters, in which serotonin causes behavior reversal by stimulating subordinates to engage in fighting against dominants by reducing their propensity to retreat (80).

Antiepileptics

Antiepileptics are ubiquitous and prevalent due to poor STW removal. Carbamazepine was the drug detected most frequently and in highest concentrations during a study by Ternes (18). This drug was detected in all POTWs and receiving waters, with a maximum concentration of 6.3 µg/L. Ternes hypothesized that the ubiquitous occurrence resulted from the very low removal efficiency from POTWs, which was calculated to be only 7%. Sacher et al. (81) found carbamazepine levels in the river Rhine in Germany up to 0.90 µg/L and always above 0.1 µg/L.

Antineoplastics

Antineoplastics are highly [geno]toxic compounds, primarily from hospitals, with poor removal from STWs. Antineoplastic agents, antitumor agents primarily used only within hospitals for chemotherapy, are found sporadically and in a range of concentrations, probably because only small amounts are introduced to STWs via domestic sewage because of their long-lived physiologic retention. These compounds act as nonspecific alkylating agents (i.e., specific receptors are not involved) and therefore have the potential to act as either acute or long-felt stressors (mutagens/carcinogens/teratogens/embryotoxins) in any organism. The fact that two oxazaphosphorines, ifosfamide and cyclophosphamide, were found in certain effluents in the low microgram-per-liter range indicates that these highly toxic compounds, which are probably refractory to microbial degradation at POTWs (82), can find their way into the environment. Indeed, Steger-Hartmann et al. (82) found levels of cyclophosphamide in sewage influent from servicing hospitals ranging from undetectable to 143 ng/L; the levels in the effluent reached 17 ng/L.

Additional evidence pointing to the refractory nature of ifosfamide is presented by Kümmerer et al. (83), who found that concentrations of ifosfamide in hospital effluent matched the predicted values of up to 1.91 µg/L; also the concentrations in the influent and effluent of POTWs that serviced chemotherapy hospitals were essentially unchanged (influent/effluent maximum, 43 ng/L; median, 6.5-9.3 ng/L). Kümmerer et al. (83) found ifosfamide to be totally refractory to removal by POTWs and to totally resist alteration during a 2-month bench-scale POTW simulation.

Another class of antineoplastics, the platinates, includes carboplatin and cisplatin. Although the stability of these compounds in sewage systems is unknown, Kümmerer et al. (84) calculated that if they were present in hospital sewage effluents as the intact parent compound, they could be present at daily average concentrations of up to 600 ng/L (on the basis of total platinum). Although the majority of the dose for these compounds is excreted in the urine in the first day, a large amount (~30%) resides in the body and is slowly excreted over a period of years and therefore could be excreted to residential sewage systems. Falter and Wilken (85) showed that while these compounds are difficult to determine analytically, their potential to remain in the aqueous phase after sewage treatment is high.

White and Rasmussen (86), in the most detailed overview to date on the genotoxicity of wastewaters, elaborate that while the genotoxic potency of industrial wastewaters is often the highest, the overall loading of genotoxic compounds to surface waters is far greater, up to several orders of magnitude, from municipal treatment plants. They present a striking correlation between the occurrence of direct-acting mutagens in surface waters and the human population served by the discharging STWs. This correlation points to the activities/metabolism of humans, not industrial activities, as the origin for these mutagens. A number of possible sources for the mutagens are discussed, an obvious one of which is antineoplastic drugs.

These data point to antineoplastics as a class of drugs of potential concern for environmental effects, not just for their acute toxicity but perhaps more for their ability to effect subtle genetic changes, the cumulative impact of which over time can lead to more profound ecologic change. Hospitals are the major source of genotoxic drugs. POTWs that service hospitals, especially multiple hospitals, are likely candidates for releasing these chemicals into surface waters.

Impotence Drugs

This class of drugs displays widespread use, new modes of action, and unknown effects on nontarget organisms. Even though a number of drugs from various chemical classes have been used over the years for treating impotence, the emergence of Viagra (sildenafil citrate) has focused tremendous attention on this market. The significance of this therapeutic class of drugs, with new ones awaiting FDA approval, is that they all tend to have distinct modes of action, most of which differ from those of traditional drugs. While potential effects on wildlife are totally unknown, the fact that Viagra, for example, works by inhibiting a phosphodiesterase responsible for regulating the concentration of cyclic guanosine monophosphate, which indirectly relaxes muscles and increases blood flow (87), gives cause for concern regarding the disruption of this common phosphodiesterase in unintended target species. Impotence drugs will prove to have very high usage rates, especially since they are one of the most common drugs available without prescription over the Internet, yielding high potential for environmental exposure and possibly nontarget effects.

Tranquilizers

Little is known about possible occurrence of tranquilizers. Ternes (18) reported diazepam in almost half of the POTWs but only in low concentrations of less than 0.04 µg/L; it could not be detected in surface waters. Genicola (88) reports diazepam in the groundwater from a monitoring well at a Superfund site near Atlantic City, New Jersey. Concentrations were approximately 10-40 µg/L and probably originated in a landfill in which pharmaceutical manufacturers disposed of chemicals.

Retinoids

High usage rates and profound activity in amphibians lends cause for concern. Retinoids, low molecular weight lipophilic derivatives of vitamin A, can have profound effects upon the development of various embryonic systems (89), especially amphibians in which retinoic acid receptors have been hypothesized to play a role in frog deformities. Although naturally occurring, retinoids have been used for a number of years for a wide array of medical conditions including skin disorders (e.g., Accutane [isotretinoin] for acne), antiaging treatments (e.g., Retin-A [tretinoin] for skin wrinkles), and cancer (e.g., Vesanoid [tretinoin] for leukemia). Isotretinoin (13-cis-retinoic acid) is related to both retinoic acid and retinol (vitamin A). Tretinoin is among the top 200 prescribed drugs in the United States. Methoprene, an insecticidal synthetic retinoic acid mimic, is photolabile and yields numerous photo-products, some of which also elicit strong retinoic acid activity (90). Although retinoic acids would also be expected to be photolabile (and therefore not persistent), their products may also still possess receptor activity.

Diagnostic Contrast Media

Diagnostic contrast media have very high usage rates, display considerable persistence, show no evidence for mineralization, and have low physiologic activity. Detailed X-ray images of soft tissues are routinely captured by the use of contrast media. Some of the more widely used members of contrast media are highly substituted and sterically hindered amidated, iodinated aromatics such as diatrizoate and iopromide (91), which are used worldwide at annual rates exceeding 3,000 tons. Kalsch (91) found these compounds to be quite resistant to transformation in STWs and in river waters. When transformations were effected, they merely terminated with unidentified resistant metabolites. Ternes et al. (92) recently reported significant amounts of iopromide in rivers.

In municipal STW effluents, Ternes et al. (92) found concentrations as high as 15 µg/L (iopamidol) and 11 µg/L (iopromide). In an STW close to Frankfurt/Main, they found two other contrast agents, diatrizoate and iomeprol, at concentrations up to 8.7 µg/L, as well as iothalamic acid and ioxithalamic acid in the nanogram-per-liter range. In rivers and streams, five iodinated diagnostics were repeatedly detected, with median values up to 0.49 µg/L for iopamidol and up to 0.23 µg/L for diatrizoate. Isolated maximum values above 100 µg/L for diatrizoate indicated that relatively high local concentrations can occur, especially in small streams containing a high percentage of STW discharges. Maximum groundwater concentrations for iodinated contrast agents ranged up to 2.4 µg/L and may well represent a worst case with respect to occurrence of pharmaceuticals in native waters. In Germany alone, individual contrast agents can experience annual usage rates of 100 tonnes. Such high usage, coupled with inefficient human metabolism (95% unmetabolized) and ineffective elimination of iodinated contrast agents by STWs, can lead to very high environmental accumulations and persistence. Despite these negative attributes, contrast agents have no bioaccumulation potential and low toxicity (93); Steger-Hartmann et al. (93) also found no acute toxicity for bacteria (Vibrio fisheri), algae (Scenedesmus subspicatus), crustaceaens (Daphnia), and fish (Danio rerio, Leuciscus idus melanotus) exposed to no more than 10 g/L of iohexol, iotrolan, diatrizaote, or iopromide.

Personal Care Products in the Environment

For the purposes of this review, personal care products are defined as chemicals marketed for direct use by the consumer (excluding OTC medication with documented physiologic effects) and having intended end uses primarily on the human body (products not intended for ingestion, with the exception of food supplements). In general, these chemicals are directed at altering odor, appearance, touch, or taste while not displaying significant biochemical activity. Most of these chemicals are used as the active ingredients or preservatives in cosmetics, toiletries, or fragrances. They are not used for treatment of disease, but some may be intended to prevent diseases (e.g., sunscreen agents). In contrast to drugs, almost no attention has been given to the environmental fate or effects of personal care products--the focus has traditionally been on the effects from intended use on human health. Many of these substances are used in very large quantities frequently more than recommended.

Personal care products differ from pharmaceuticals in that large amounts can be directly introduced to the environment. For example, these products can be released directly into recreational waters or volatilized into the air (e.g., musks). Because of this direct release they can bypass possible degradation in POTWs. Also, in contrast to pharmaceuticals, less is known about the effects of this broad and diverse class of chemicals on nontarget organisms, especially aquatic organisms. Data are also limited on the unexpected effects on humans. For example, common sunscreen ingredients, 2-phenylbenzimidazole-5-sulfonic acid and 2-phenylbenzimidazole, can effect DNA breakage when exposed to UV-B (94).

The quantities of personal care products produced commercially can be very large. For example, in Germany alone the combined annual output for eight separate categories has been estimated (95) at 559,000 tons for 1993 (Table 3). A few examples are given below of common personal care products that are ubiquitous pollutants and that may possess substantial bioactivity.