Environmental protection: Theory and practice - ACS Publications

---

Environmental protection: Theory and practice

Renate D. Kimbrough U.S. Environmental Protection Aaencv Washington, DC 20460 I

_

In the past 20 years, the fields of toxicology, environmental health, and ecology have expanded exponentially. Sophisticated methods of assessing risks have been developed. Elaborate tests can be conducted to determine the toxic properties of chemicals. Increasingly sensitive analytical methods have been developed to detect smaller and smaller concenaations of chemicals in a variety of media. However, this newly gained scientific knowledge has not been properly incorporated into environmental laws and regulations.

The US. environmental laws (1) that have been enacted do not have a consistent approach toward improving public health and environmental quality. In spite of our best intentions, public pressure and lack of adequate coordination can lead to attention and resources being devoted to minor problems while important issues are left unattended. Less air pollution, clean drinking water, efficient management of waste, and protection and preservation of OUT natural environment are laudable goals, but when the various laws were developed not much thought was given to how this can best be accomplished. The coordination between environmental programs and the manner in which science is incorporated into poli-

1442 Environ. Sci. Technol., Vol. 24, No. 10. lea0

cy decisions need to be improved. In this paper I discuss these issues as they relate to the evaluation of risk. When the approaches to risk assessment and risk management are examined, it becomes apparent that many factors influence these processes. Most important are the different mandates of the various statutory authorities under which EPA operates. The agency is divided into a number of p r o g r a m offices that implement these statutes. Their risk analyses, resulting cleanup levels, discharge limits, and predicted ecological and health risks may differ. The 10 EF'A regions also operate independently of each other. Because many of their functions are divided along the same programs that exist in

This slticle not subject to US. copyright. Published 1990 American Chemical Sodety

the headquarters office, a given region must respond independently to different program offices. Thus, differences in policy or lack of coordination in headquarter operations is most acutely felt at the regional level. Determination of risk The driving force to reduce exposure of humans and the environment to manmade chemicals is the assumption that at certain concentrations such chemicals are harmful. This is not disputed. However, it is difficult to characterize this concept further. If such chemicals are already in the environment, should they be removed and to what extent? What concentration of a given chemical should be permitted in water, air, and other media? The information necessary to address these questions consists of two parts: the assessment of the levels of the pollutants and the resulting degree of exposure, and the assessment of the health and environmental effects of the pollutants. Based on this information the potential risk can be assessed. Such an approach seems straightforward; however, it is often not possible to quantitate or define a particular risk with certainty. Risk assessment Using standard assumptions in risk assessment means that an estimate of risk can be stated and reviewed by others, and comparisons can then be made between different risk assessments. However, there is a fundamental prob lem with the use of standard assumptions. Because biological mechanisms of action vary, wrong comparisons may be made. EPA’s cancer risk assessment guidelines (2) recognize the need to incorporate biological data. To some extent the current effort to update these guidelines also reflects a growing scientific consensus that more attention should be paid to the biological signifcance of such data. Risk assessments are based on factors such as dosage, buman health effects, effects in animals, the bioavailability of chemicals, and the chemical form (3,4). The lowest dose tbat results in adverse health effects in humans is usually unknown for industrial chemicals and pesticides. For many indusnial chemicals no or very little information is available on any adverse human health effects. This has resulted in the development of extrapolation techniques from animal studies to predict human health effects. For some but not all chemicals, information is available about acute and chronic toxic effects in laboratory animals. Although it is customary to extrapolate from animals to humans, it is not always clear which animal species is

the best surrogate for humans. For instance, the neurotoxicity of organophosphorus pesticides is best tested in chickens or cats. The guinea pig is the best surrogate to test allergic reactions. Some lesions that develop in animals are not relevant for humans. For instance, in rats some organic tin compounds cause an inflammation of the wall of the bile duct where it enters the duodenum. This lesion only occurs in species where the bile duct and the pancreatic duct have a common course, which is not the case in humans (5). Similarly, a number of chemicals cause tumors of the kidneys in male Wistar rats because these animals form large amounts of alpha-2u-globulin. This globulin may combine with chemicals, resulting in an irritating effect on the epithelium of the kidneys and in the formation of kidney cancer in the male rat. Female rats and mice form little alpha2u-globulin and do not develop cancer

‘‘

I

99

I

If [manmade] chemicals are already in the environment, should they be removed and to what extent? I

from a virtually safe dose in animals is used for such an extrapolation. However, for many chemicals, only results from short-term testing are available. To compensate, larger uncertainty factors are then employed. However, using short-term studies for extrapolation and large uncertainty factors of 1,000 or 10,ooO leads to inconsistency in the prediction of risk for similar classes of chemicals simply because more is known about some chemicals than about others. Finally, some effects are not clearly adverse, and the interval between a dose that produces observed effects and a dose whose effects are not noticeable will vary, depending on the dose intervals used by the experimenter. Thus, the available data will introduce inconsistencies in the evaluation of chemicals. Use of the nonthreshold concept. Mathematical models for extrapolating from high to low doses have been developed for cancer as an endpoint. In this case, the acceptable dose (typically the dose that leads to risks in the lod to lF7range) is lower than the dose that would be acceptable for any other chronic toxic effect. However, it may not always be appropriate to use one of these mathematical models (7-9).Most models are based on the assumptions that the chemicals initiate cancer by affecting deoxyribonucleic acid (DNA) either directly or indirectly and that they represent complete carcinogens. But chemicals may also affect the development of cancer by an epigenetic mechanism for which the presently used models of extrapolation are not appropriate. If the proximate carcinogen is a metabolite, then the rate at which this metabolite is formed may differ at high and low doses and may vary among species. A particular proximate carcinogen may not be formed at all if, for instance, the primary pathway must be overwhelmed for a secondary pathway to come into play. Because of the small number of animals used in animal studies, high doses are usually given to make bioassays more sensitive. At such high doses, cells of the target organs may undergo cell necrosis (i.e., tissue injury or cell death). Cell necrosis stimulates cell growth, proliferation, and division, producing an additional promotional effect on the development of cancer in target organs. Indirect DNA damage through replication error would then be more likely. This would not happen at doses at which no cell necrosis occurs. At low doses, genotoxins and epigenetic agents are more likely to be bound to substances in the cytoplasm and are thus more efficiently metabolized and excreted. Therefore,

1

of the hdney. simi~ar~y, humans ah0 form little of this globulin, and therefore this kidney lesion has no relevance for human health (6). Customarily two approaches are used to extrapolate from data obtained in laboratory animal studies to humans. For all effects other than cancer a threshold is assumed and uncertainty factors are applied. For chemicals that produce cancer in laboratory animals it is assumed in the United States that no threshold exists, and mathematical models are applied that assume that the dose response is more or less linear. If lifetime Studies are available in animals, the results are customarily used to predict chronic toxicity in humans. For cbronic toxic effects it is assumed that an adequate margin of safety is developed if an uncertainty factor of 100

Environ. Sci. Technol., VoI. 24, No. 10, 1990 1443

the response to a carcinogen would no longer be linear. At low doses, the lack of a catabolic effect coupled with the defense mechanisms of the cell, such as metabolism and excretion of the chemicals and repair of DNA, may for all practical pqoses result in athreshold I;lther than a continuum of linearity (10). The need to fully utilize such biological data is in part outlined in the EPA guidelines for cancer risk assessments (2). However, as currently written the guidelines do not adequately describe the weight that should be given to such factors. They are therefore usually not followed. A default position is assumed instead, and the risk assessment process using the linearized multistage model is applied if the chemical has been shown to produce tumors in animals. There may be instances where this assumption is correct. However, the generalization of these assumptions to all instances where cancer has been prOauced in l a b oratory animals at maximal tolerated doses is inappropriate. Expressing uncertainty in risk assessment. When using the multistage linearized model to estimate the risk for cancer, risk assessors usually express this risk as an “upper bound” or as a “plausible upper bound” of a virtually safe dose. This number is meant to be a dose below which the risk is probably less than one cancer in a million, in 1oO,OOO, or in lO,oOO, depending on the theoretical level of protmtion desired. In reality it is likely that the risk is much less; it may even be zero. The upper bound is simply a prediction based on statistical calculations for which no xientific p m f exists. This aspect of the risk assessment process is lost when risk managers and the public at large use the upper bound numbers. These numbers are interpreted as being factual and predicting m e , “actuarial” risks. Thus the public as well as the risk managers misuse the results of risk assessments. In addition to this problem of misuse and misunderstanding, other methodological problems exist as well. These problems result from the type of data used for risk assessments. The amount, quality, and appropriateness of the toxicity data that form the basis for the risk assessment should be critically evaluated. The uncertainties introduced hy hiological factors such as pharmacokinetics and species variation should be considered and discussed. Exposure assessment methodologies used in risk assessments (e.&, air models, pathways of chemicals, bioavailability, and assumed total dose from point sources) should be verified hy real data before they are generically applied to risk assessments. Once a risk assessment is completed, it should be determined whether the re-

sults make “hiological sense.’. Cost, feasibility, accessibility, general background levels of similar chemicals in the environment, and the form of the chemical should also be considered. Whether a particular source makes a major contribution to overall exposure is i m p tant. Site-sperific exposures Chemicals are now measured in a variety of seaings, and their risks are estimated on a laal or r e g i d Mi.For meSe situations exposw assessments may be a ma@ factorin risk assessments. For exposure assessments, hypothetical scenarios are developed that may or may not resemble reality. Often the mode of transpolt or the bioavailahility of the chemical is unknown Ill I. How-

l

44

Once a risk assessment completed, i be determined whether the results make ‘biological sens

I

I

I

exp&ure may be dveres&ated and giving a range, risk assessors may present the worst case estimate. The risk assessors may point out what these estimates really mean, but these qualifiers are often not considered when risk management decisions are made by regulators or by members of the public.

Factors afleeting risk assessment Chemical analysis. The concentrations at which most chemicals are deemed to represent an acceptable risk are minuscule. It usually is not appreciated that higher levels will o h not result in any harm. Furhermore, the ability to measure chemicals accurately decreases with increasingly lower concentrations. For instance, there may not be any true difference between 5 and 10 pph (pgkg) of a given chemical in a medium such as soil or air. Finally, the analytical methods used for many chemicals have not been standardized nor are the same analytical

M44 Envimn. Sci. Technol., Vol. 24, No. IO, 1990

methds used even for the same medium. For instance, different analytical methds for the same analyte may have to be used for drinkiig water and for groundwater simply because the requirements for analyses were developed in different programs. It is often argued that precision and accuracy in chemical analysis are of minor importance because the uncertainties in a risk assessment are usually several orders of magnitude. Thus, imprecise and inaccurate chemical analytical data do not have much of an impact. However, this is only partially correct. In some situations a doubling of the risk resulting in a lower concentration as an acceptable level may have a major impact on compliance cost. Moreover, because of the a c t i d n o action significance attached to numerical standards (e& maximum concentration limits), “minor” inaccuracies are very important and very costly in practice. Furthermore, how results of chemical analyses that are below the limit of detection are entered into the risk assessment process may have a significant impact on the results of the risk assessment. Dosage. For chemicals present in food or water a dose is not easily calculated even if chemical concentration and consumption are known. Absorption of the chemical may vary depending on other food constituents present, the age of the person exposed, and the form of the chemical. Examples are the higher absorption of lead by the gastrointestinal tract of infants, the differences in absorption of lead acetate and lead sulfide, and the effect of particle size and of calcium on the absorption of lead. Estimating uptake of chemicals from air and soil is even more prohlematic. Air models presently in use appear to determine the upper bound of the risk without giving any infamation on what the m a e likely exposure would be. If the chemical is present as an aemwl ormachedto particles in air, the size of the droplets or pamcles determines whether the chemical reaches the lungs. It is presently not known to what extent chemicals attached to such particles areabsabed.

Soil also binds chemicals. If organic chemicals are tightly bound to soil, they will be poorly absorbed. Thus, uptake of chemicals present in the environment may vary widely. Metals. Metals may be present in a variety of forms and species. Some forms of metals are practically insoluble and thus are not well absorbed. Poorly soluble forms of metals do not pose the same threat to humans or animals as do the more soluble forms. The adverse impact of metals also depends on the composition of the mixture of trace elements in which the metals are found. In the

presence of calcium, phosphates, and zinc, for instance, some metals are much less well absorbed. Sampling and analysis Representative environmental sampling is complex, and chemical analysis of very low concentrations of chemicals in environmental samples is expensive and difficult. If the sample collection is not representative, then exposure may be over- or underestimated. Limited budgets often lead to adeemphasis of quality control and quality assurance. Thus, the quality of the data available to the risk assessor may vary widely. Usually the uncertainties for a given analytical result are not stated. Because of these variations in methodology, it is difficult to conduct trend analysis to determine whether levels of pollutants are increasing or decreasing over time. Furthermore, results obtained under different programs cannot be c m lated because of the inconsistencies in the methods used (12,131. Impact on the environment Presently, little information is available to guide us in assessing the risk of industrial chemicals in ecosystems. We know little about fate and transport of many chemicals. Furthermore, actions intended to protect groundwater may in some areas change land usage with pollution of groundwater by other chemicals. Insufficient baseline information exists to intelligently correlate effects in biota with specific chemical exposure. Whether malformations observed in birds or tumors in fish are the result of local pollution usually cannot be properly evaluated. Few case histories have been successfully completed in which an observed lesion or population decline has been reliably related to an inferred cause (14). The prediction of the effects of exposure on environmental systems is also limited. Most exposure assessment models are directed at predicting human exposure; implications for ecosystems are iess well understood.

Risk management and reduction Different environmental laws were developed for air and water, and for waste and landfills. Decisions are often made in isolation in different programs responsible for these areas, resulting in apparent inconsistencies. Only recently have attempts been made to determine whether reducing a chemical or banning it in a particular medium such as water will make a significant contribution to overall reduction of exposure of humans and the environment. For instance, in regulating levels ~~

of chemicals in water, it is usually not considered whether humans would get much greater exposure from air or from food or by using these substances in their daily lives. Which measure of exposure reduction to certain chemicals is the most economic or the most effective usually is not examined either. Because decisions in different programs are made in isolation without a review of their overall present and future impact, other environmental and public health problems may be created. At times the technology needed to reduce chemicals in water or air to very low concentrations or to a nondetectable level may be either very expensive or not available. Under such circumstances, goals to reduce the concentrations of these chemicals may be set that are unachievable, unrealistic, and even unnecessary. By removing chemicals from water, their presence may be increased in sludge, creating sludge disposal problems. When reducing the impact of chemicals and solid waste on the environment we must examine what the benefits m.If for less cost we can sipiicantly reduce wntamination (e.& 95% vs. 99% at higher cost) with almost equal protection of the environment, we should settle for a less costly approah and use the money saved in other a m s hat need attention. We should reexamine whether vast sums of money should be spent to reduce the levels of chemicals in the environment to levels that cannot be adequately measured and that professional judgment would regard as presenting an acceptable risk. Such money might be more effectively spent on an integrated national waste management and waste reduction program.

to respond to these demands never gives EPA quite sufficient time or resources to develop a long-range plan in which certain goals can be perused and the effectiveness of control efforts can be determined. The opinions expressed in this article are those of the author and do not necessarily reflect a policy position of the Environmental Protection Agency.

The courts, the public, and Congress Because of lack of resources, lack of available technology, and the inherent difficulty in dealing with the long-range problems of environmental pollution, EPA is perceived by some critics as being ineffective, sympathetic with industry, and inefficient in implementing envkonmental laws. This perception results in directives from the courts and demands from environmental groups, Congress, and state legislators to develop certain controls and produce certain data and results in a specified period of time. These various specific problems are addressed piecemeal and haphazardly. Thus controls are developed without evaluating whether their application makes any major contribution to environmental improvement and whether the technology is sufficiently simple that these demands can actually be implemented by the infrastructure that presently exists in the United States. Having

Terbhenyli, Nophihole& Dibembdio& and Related Produeis; Elsevier: Amster-

References (I)

(2) (3)

(4)

(5) (6)

(7) (8)

(9)

(IO) (11)

D o h , G. D. “An Infloductory Guide to he Statutory Authorities of the U.S. Environmental h t e c t i o n Agency”: U.S. Environmental h t e c t i o n Agency: Washington, DC, 1987; EPA-905/9-87M)3. Fed. Regist. 1986, Sept. 24,33992-34003. ”Risk Assessment in the Federal Government: Managing the Process”; National Academy of Sciences: Washington, DC. 1983. Nichols, A. L.; Zeckhauser, R. 1. Regularim 1986, Novemkr/December. 1%23. Barnes, 1. M.; Magee, P. N. J . Pothol. Bocreriol. 1958, 75, 267-79. Galdsworthy, T. L. et al. Toxicol. Appl. Phormocol. 1988,%, 367-79. Crump, K. S. J . Environ. Pnfhol. Toxicol. 1981,5, 6 7 5 4 4 . Gaylor, D. W.; Kodel. R. 1. J. Environ. Porhol. Toxicol. 1980.4, 305-12. Hogan,M. D.; Hoel, D. G. In P-ipla and methods of roxicology: Hayes, A.W., Ed.; Raven h s : New Y& 1982: pp. 711-31. Aldridge, W. N. Ann. Rev. Phormcol. Toxicol. 1986, 26, 39-58. Kimbrough. R. D. et al. J . Toxicol. Environ. Health 1984,14, 47-93.

(12) Availability, Adequacy ond Comparobiliry of Tesring Procedures for the AMlysis of Pollwnnrs Estobished Under Section 3 W h ) of the Federal Wofer Pollution Control Acf. Repon to Congress; US. Environmental h t e c t i o n Agency. o f f i c e of Research and Develoment. Environmental Monitoring System’s Labaratory: Cincinnati, OH, Sept. 1988: EPA/€03D-87/030. lensen, A. A. In Kimbmugh, R. D.: lensen. A. A,. Eds.. Holoeenared Biohenvls.

,

1989; pp. 345-380. , dam. Gilbertson, M. In Kimbmugh, R. D.; Jensen. A. A,. Eds.. Halozenored Biohenvls.

dam,1989; pp. 103-127.

Rena4e D. Kimbrough is an advisor for Medical Toxicology and Risk Evaluation in the Office of the EPA administrator. She holds an M.D. dearee with fraininc in Dathology. She is a Diplomate of theAmerican Board of Toxicology and is licensed to Practice medicine in and Illinois. She has worked in toxicology and environmental health since 1962, has than 100 publications in refereed journals, and h a edited or written chapters for several books.

Envimn. Sci. Technol., Vol. 24. No. 10, 1990 1445