Environmental Risk Assessment Reports - Chapter 29 doc

527 CHAPTER 29 International Health Risk Assessment Approaches for Pesticides Colleen J. Dragula Johnson and Gary J. Burin CONTENTS I. Introduction 527 II. Sources of Information 528 III. Differences in Policy and Objectives 528 IV. Performing Assessments 531 V. Issues Affecting Risk Assessment Decisions 534 VI. Status of International Harmonization Efforts 533 VII. Conclusion 534 References 534 I. INTRODUCTION The procedure for conducting risk assessments was originally developed by the National Academy of Sciences in the early 1980s and contained four steps: hazard identification, dose-response assessment, exposure assessment, and risk character- ization. This same basic method continues to be used today as the basis for assessing chemicals. Despite the general scientific consensus on the individual components of a risk assessment, substantial differences exist in the numerical values calculated. This chapter compares and contrasts the techniques used by various countries for assessing the risks associated with pesticide substances. Particular emphasis is placed on quantitating cancer risk, a contentious area within the scientific and regulatory communities for many years. In addition, the impact of mechanism of action (i.e., genotoxic and nongenotoxic) is emphasized, since this often drives the assessment procedures used by countries. LA4111 ch29 new Page 527 Wednesday, December 27, 2000 3:31 PM © 2001 by CRC Press LLC 528 A PRACTICAL GUIDE TO ENVIRONMENTAL RISK ASSESSMENT REPORTS In order to provide more consistency between countries on the methods used to assess risk, a number of scientific meetings and conferences have discussed inter- national harmonization efforts. To determine the status of this issue within the international community, the International Programme on Chemical Safety (IPCS) sponsored a project utilizing a survey questionnaire to obtain information on assess- ment techniques utilized by 21 Organisation for Economic Cooperation and Devel- opment (OECD) and select nonOECD countries. Several review articles and docu- ments have also been prepared on this topic, especially in relation to carcinogenicity (GAO, 1993; OTA, 1993; WHO, 1993). Finally, this chapter addresses a number of policy and technical issues relating to risk assessment. To the extent possible, updates on changes in policy have been included. The primary goal is to highlight differences that currently exist between the U.S. and other countries. II. SOURCES OF INFORMATION It is sometimes difficult to identify the appropriate governing body or group within a regulatory agency to contact with questions concerning risk assessment procedures. GAO (1993) provides a schematic breakdown of the regulatory agency structures in select OECD countries. In addition, the Pesticide Regulation Compendium (PRC, 1993) comprehensively describes the pesticide regulations in over 100 countries, including a description of each country’s regulatory system, data requirements for residues, toxicology, ecotoxicology, and labeling. Table 1 lists the addresses for select regulatory agencies involved in conducting risk assessments. III. DIFFERENCES IN POLICY AND OBJECTIVES To investigate the intercountry differences in approaches to risk assessment, two components of the process have been examined: hazard identification and dose- response. Hazard identification examines all available data in humans and laboratory animals relating to a chemical’s potential to induce toxicity. For carcinogenicity, the International Agency for Research on Cancer (IARC) uses an alphanumeric classi- fication scheme to characterize whether the chemical is a known, probable, or possible human carcinogen. Although the U.S. EPA formerly used an alphanumeric system, in April 1996 it published new cancer risk assessment guidelines (U.S. EPA, 1996). These new guidelines proposed the use of three descriptive categories, similar to those used in the European Union (EU)* to characterize cancer risk (Directive 67/548/EEC).** Although assignment of these qualitative cancer ratings tends to be * The European Community (now called the European Union) was established in 1958 by the Treaty of Rome and is responsible for developing governmental policy primarily through legislation known as regulations and directives. Much of the legislation affecting environmental issues involves directives. These directives are binding on the member nations with respect to the end result, but allow each state to individually decide the means by which they will implement the directive (OTA, 1993). This allows for considerable flexibility in fine-tuning the specifics of the directives. ** EU Directive 67/548/EEC addresses the classification, packaging, and labeling of dangerous sub- stances and was last appended by the Seventh Amendment. Annex VI of this directive provides guidance for characterizing the carcinogenicity of chemicals in the EU. LA4111 ch29 new Page 528 Wednesday, December 27, 2000 3:31 PM © 2001 by CRC Press LLC INTERNATIONAL HEALTH RISK ASSESSMENT APPROACHES FOR PESTICIDES 529 fairly consistent between countries, the situation for the dose-response evaluation is significantly different. In the U.S., dose-response data are frequently assumed to be linear in the non- experimental low-dose region and are assessed through the derivation of cancer potency factors. Using mathematical models (i.e., the linearized multistage model, see below), upper limits of risk are calculated that yield a cancer potency factor. This factor is multiplied by the estimated exposure to yield a single risk value. In actuality, the “true” risk actually lies between this calculated upper limit (i.e., 95th percentile) and zero (EPA,1989). The 15 EU member nations (Austria, Belgium, Denmark, Finland, France, Ger- many, Greece, Ireland, Italy, Luxembourg, The Netherlands, Portugal, Spain, Swe- den, and the U.K.) use a significantly different approach than the U.S. for assessing the dose-response of carcinogenic pesticides. In these countries, cancer is thought to be a threshold response yielding a “safe” dose below which there is no risk. The EU also focuses on mechanism of action (e.g., genotoxic vs. nongenotoxic) in dealing with carcinogenic or potentially carcinogenic pesticides; registration of pesticides which are genotoxic is not permitted. Acceptable daily intakes (ADIs) are derived for nongenotoxic pesticides using NOELs or NOAELs* from animal carci- Table 1 Regulatory Agencies Responsible for Conducting Pesticide Risk Assessments Australia The Scientific Director Chemicals Safety Unit DHHLGCS, PO Box 9848 Canberra ACT 2601 Bulgaria National Center of Hygiene Ecology and Nutrition D. Nestorov Str. 15 1431 Sofia, Bulgaria Canada Director General Food Directorate Health Protection Branch Health and Welfare Canada Health Protection Branch Bldg.,Tunney’s Pasture Ottawa, Ontario Canada K1A OL2 China Institute for the Control of Agrochemicals, Ministry of Agriculture (ICAMA) Liang Maqiao, Chaoyang Qu, Beijing 100026, China Czechoslovakia National Institute of Public Health, National Reference Centre for Pesticides Srobarova 48, 100 42 Prana 10 Czech Republic Egypt Central Agricultural Pesticides Lab Ministry of Agriculture Dekki, Giza, Egypt France DGCCRF Commission of Toxicity 59 Boulevard Vincent Auriol, 75703 Paris cedex 13 Germany Bundesgesundheitsamt 1. Eachgebeit C I 4 2. Postfach 33 00 13 3. B-1000 Berlin 33 4 India Secretary, Central Insecticides Board & Registration Committee Directorate of P.P.Q & S NH-IV, Faridabad 121001 Korea Agricultural Chemicals Research Institute 249 SeudoonDong SuweonSi KyunggiDo Republic of Korea Thailand Director of Agricultural Regulatory Division Department of Agriculture Bangkok 10900 United States Office of Pesticide Programs, Health Effects Division Environmental Protection Agency Washington, DC * The NOAEL is the highest dose administered that produces “no statistically or biologically significant increases in frequency or severity of adverse effects.’’ The NOEL is defined in the same manner, with the exception that there is no increase in the frequency or severity of effects (Hallenbeck and Cunningham, 1985). LA4111 ch29 new Page 529 Wednesday, December 27, 2000 3:31 PM © 2001 by CRC Press LLC 530 A PRACTICAL GUIDE TO ENVIRONMENTAL RISK ASSESSMENT REPORTS nogenicity studies. This approach has occasionally been applied in the U.S. to assess pesticides for which the weight-of-evidence for carcinogenicity is not convincing. Thus, the use of a threshold-based approach in the U.S. is a function of the weight- of-evidence rather than whether or not genotoxicity is presumed to be responsible for tumor induction. Given the release of the new cancer risk assessment guidelines, however, the procedures used by the U.S. may change. Health Canada reported employing different risk assessment techniques depend- ing on the mechanism of tumor induction. For genotoxic pesticides, a quantitative risk assessment is performed whereas for nongenotoxic pesticides an ADI is derived in conjunction with a weight-of-evidence assessment (Dragula and Burin, 1994). Specific information on the type of quantitative risk assessment model used was not supplied. Typically, the federal government is responsible for these assessments, and not the individual provinces within Canada. Publicly available information on pes- ticide regulatory practices can be obtained through “Backgrounders” published by the Pesticides Directorate. For occupational exposures, Denmark, The Netherlands, and the U.K. use quan- titative risk assessment techniques (i.e., mathematical models) to generate a risk value which represents the probability of human cancer risk. This probability reflects the expected or the best estimate of the human cancer risk likely to occur in a population. Finally, a significant portion of countries simply adopt the carcinogenicity assess- ment policies or evaluations developed by other countries or scientific groups. For example, Bulgaria and Thailand rely on the evaluations derived by MARC (which provides statements regarding the weight-of-evidence for carcinogenicity based on animal and epidemiological data); Korea relies on the risk assessment techniques and decisions developed by the EPA; Egypt utilizes the evaluations provided by the Joint FAO/WHO Meeting on Pesticide Residues; Czechoslovakia uses the risk assessment methods developed by WHO in their Environmental Health Criteria document (WHO, 1990); and Spain relies on criteria described in Annex VI of the EU Directive 67/548/EEC for dangerous substances (Dragula and Burin, 1994). The most outstanding difference between the risk assessment goals or objectives in the U.S. vs. the EU is the default assumption in the U.S. that carcinogenicity is a nonthreshold, or linear process where every increase in dose is associated with an increased risk. The U.S. has utilized the linearized, multistage model for most potential human carcinogens, including pesticides, whereas the EU has generally used the ADI approach. The U.S. also publishes extensive guidelines outlining the procedures for per- forming carcinogenic and other types of risk assessments. GAO (1993) referred to this policy as “transparent” and noted that such detailed procedures were not yet readily available in the EU and other OECD countries surveyed. However, the specific guidance provided by the EPA may have the unintended effect of forcing risk assessment decisions (i.e., through the application of default assumptions) in ways not consistent with expert scientific judgement. The interviewees considered the U.S. process to be less flexible than the procedures used in other OECD countries, a fact they viewed as a weakness when compared to their system’s ability to flexibly address specific issues on a case-by-case basis (GAO, 1993). LA4111 ch29 new Page 530 Wednesday, December 27, 2000 3:31 PM © 2001 by CRC Press LLC INTERNATIONAL HEALTH RISK ASSESSMENT APPROACHES FOR PESTICIDES 531 IV. PERFORMING ASSESSMENTS Two primary procedures exist for estimating the carcinogenic risks associated with exposure to pesticides (and other substances): mathematical models and ADIs. In general, quantitative risk assessment refers to the use of models since an actual risk value is calculated. This method is much more complex than the ADI approach in which the actual “risk” is presumed to be nonexistent. The ADI is derived from the highest dose in an animal study at which no adverse effects occur and is compared to the estimated human exposure. The U.S. typically employs models using sophisticated computer software to estimate cancer risk. The Netherlands is the only other country that regularly per- forms quantitative risk assessment utilizing such models, and it only uses these models for occupational exposures, not for pesticide assessments. Quantitative assessments using models are performed on a limited basis in Canada, Denmark, Germany, and the U.K., but again not for pesticides. In the U.S., the risk associated with carcinogenic substances is assessed using the linearized, multistage model. This model has two main constraints. First, it assumes no threshold for effects. In other words a single molecule of the pesti- cide/substance can induce the molecular events necessary to produce cancer (Barnes and Dourson, 1988). As a result, zero risk is only achieved when zero exposure occurs. Second, it assumes that the dose-response curve is linear in the low-dose region. This means that an increase in dose results in a proportional incremental increase in cancer risk. The decision to use this model was made in the 1970s and was based upon uncertainty regarding the shape of the dose-response curve in the nonexperimental, low-dose region. EPA believed that it was prudent to be conser- vative where the public was concerned and, thus, chose the linearized multistage model in order to provide the greatest protection. This model remains the default method for estimating the cancer risk of carcinogens. Alternative approaches may be implemented when the draft cancer risk assessment guidelines are formally adopted. As noted above, most countries rely on ADIs for establishing exposure limits for carcinogenic pesticides. These limits are not the same as risk values because they represent levels at which no risk is predicted to occur. ADIs are calculated by dividing the NOAEL or NOEL by Safety Factors (SF), also called uncertainty factors. These factors reflect the reliability and consistency of the experimental animal data. Generally, the more SFs applied the less confidence is placed in the data. The majority of the nations polled in the IPCS survey (Australia, Belgium, Bulgaria, Canada, China, Czechoslovakia, Denmark, France, Germany, India, Japan, Korea, The Netherlands, Spain, and the U.S.) routinely utilized the NOAEL/SF approach for estimating the risks associated with pesticide exposure (Dragula and Burin, 1994). As previously noted, the U.S. often uses quantitative models for estimating cancer risks. The reliance on the NOAEL/SF approach is easy to understand: it is simple to use and provides a clear, limit value below which exposures are considered accept- able. Although many countries currently rely on this approach for assessing cancer risks there are practical advantages to the risks calculated from quantitative models. LA4111 ch29 new Page 531 Wednesday, December 27, 2000 3:31 PM © 2001 by CRC Press LLC 532 A PRACTICAL GUIDE TO ENVIRONMENTAL RISK ASSESSMENT REPORTS These include the ability to calculate “cost” per tumor in risk benefit comparisons and to develop “bright lines” or clear risk values upon which regulatory decisions can be based (e.g., a cancer risk of 1 in 1 million). V. ISSUES AFFECTING RISK ASSESSMENT DECISIONS The mechanism by which substances induce cancer (e.g., genotoxic vs. nongeno- toxic) is the single most important factor affecting the approach many countries take to assessing human cancer risk.* Terms used to describe the mechanism of non- genotoxic and genotoxic substances are threshold and nonthreshold, respectively. There appears to be general international consensus that most biologic effects occur through a threshold mechanism (i.e., there is a dose, which is unique for each substance and endpoint, below which no adverse effect or response is observed); however, there is some debate about whether this is also true for carcinogenicity. According to U.S. policy, cancer can develop from a single event with no threshold. Yet, several nongenotoxic mechanisms of action have been identified and studied extensively. These include thyroid tumors induced by hormone imbalance and kidney tumors associated with α−2µ globulin. Two other issues affect the conduct of risk assessments and represent substantial differences in international scientific opinion: the significance of tumors arising at the maximum tolerated dose (MTD); and the significance of mouse liver tumors (MLTs) in extrapolating carcinogenic risk to humans. Not only do these issues symbolize a significant divergence of opinion and policy between the U.S. and other countries, they also represent barriers between countries with respect to data inter- pretation in carcinogenicity studies. Most countries believe that tumors only occurring above the MTD result from physiologic changes that cannot be directly associated with the substance adminis- tered. Such changes include, but are not limited to, cell death and concomitant cellular repair processes, metabolic overload, pharmacokinetic alterations, and hor- monal changes. Australia, France, Bulgaria, and Canada specifically concluded that tumor data collected at the MTD were either of limited or no usefulness for assessing carcinogenic risks in humans (Dragula and Burin, 1994). In comparison, the 1986 EPA cancer risk assessment guidelines state that data must be collected at the MTD in order to provide adequate statistical power for assessing the carcinogenicity of a substance (U.S. EPA, l986). An agency position document stated that the purpose of the MTD was to “vigorously” test a substance “for oncogenic potential at levels somewhat below test levels which might compro- mise survivability” (Farber, 1987). Interestingly, EPA did not specifically use the term “MTD” in the proposed new cancer risk assessment guidelines, but it did state that the high dose in a carcinogenicity bioassay should produce some toxicity (not * Genotoxic compounds directly interact with DNA and alter it permanently such that the change is passed on to future generations of cells. This is referred to as a heritable change and the substances that produce this effect are often termed “initiators” to describe the initiation of cancer. In comparison, non- genotoxic substances simply enhance or promote the growth of cells already genetically altered or initiated; they do not interact directly with the DNA to produce any alterations. These substances are referred to as “promoters.” LA4111 ch29 new Page 532 Wednesday, December 27, 2000 3:31 PM © 2001 by CRC Press LLC INTERNATIONAL HEALTH RISK ASSESSMENT APPROACHES FOR PESTICIDES 533 to exceed a 10% reduction in body weight gain during the lifespan of the animal) without unduly affecting the survival or the nutrition and health of the test species (U.S. EPA, 1996). Of the 16 countries responding to related questions in the IPCS survey, eight (Bulgaria, Denmark, Germany, India, Japan, The Netherlands, Spain, and the U.S.) stated that dose selection “frequently” affected carcinogenicity study results and another two (Belgium and China) stated that it “sometimes” affected study results (Dragula and Burin, 1994). Furthermore, seven of the countries concluded that the highest dose used was too high (Belgium, Bulgaria, Czechoslovakia, Denmark, Germany, Japan, and Spain) and five felt that it was inadequately justified (Canada, France, Germany, India, and the U.S.). These survey results point to the need for international consensus on the highest dose to be used in animal carcinogenicity studies and clarification as to whether this dose should be the MTD. Clearly, many countries disagree with the U.S. criteria for determining whether dosing is adequate. The U.S. has also developed a policy/regulatory position on the significance of MLTs for estimating carcinogenic risk to humans. EPA (1986) indicated that MLTs, under particular conditions, provide sufficient evidence of carcinogenicity in ani- mals, even if they are the only type of tumor response observed. This “sufficient” classification may be withdrawn under particular conditions but the policy is in direct conflict with the opinions and policies of other countries (Dragula and Burin, 1994). In India, for example, MLTs typically receive little “weight” as indicators of carci- nogenic potential in humans. Bulgaria also assumes that these tumors are of little relevance in estimating human cancer risks. In Germany, the significance of MLT data is assessed on a case-by-case basis. Because MLTs are a frequent finding in carcinogenicity studies, the different views on their significance represent a barrier to risk assessment harmonization. VI. STATUS OF INTERNATIONAL HARMONIZATION EFFORTS Significant interest in harmonizing risk assessment techniques has culminated in a number of international scientific conferences as well as national and international sponsorship of data gathering activities on the subject. At present, scientists are attempting to determine the status of risk assessment procedures in different nations and have discovered that some countries have had the opportunity to focus efforts on it far more than others. This information represents a first step towards identifying the areas most likely to benefit from harmonization efforts. In the IPCS survey, 17 of the 18 nations clearly indicated that they were interested in harmonizing risk assessment procedures (Dragula and Burin, 1994). However, the area(s) that each country identified as being most in need of harmonization activities varied considerably and included virtually every topic related to toxicology. This diversity of responses truly reflects the varying interests of the countries polled. Involvement in international harmonization endeavors is not limited to pesticide chemicals. In fact, the pharmaceutical industry has recently worked to harmonize various topics of concern. Several consensus documents have been published under the sponsorship of the International Congress on Harmonization. It would appear LA4111 ch29 new Page 533 Wednesday, December 27, 2000 3:31 PM © 2001 by CRC Press LLC 534 A PRACTICAL GUIDE TO ENVIRONMENTAL RISK ASSESSMENT REPORTS that some of the lessons learned by pharmaceutical manufacturers may also be applied to pesticide manufacturers. VII. CONCLUSION The four-step paradigm originally proposed in 1983 by the National Academy of Sciences continues to be the most widely used methodology for presenting the results of risk assessments. However, as this chapter shows, significant differences exist with respect to individual country’s policies for interpreting toxicity data and assess- ing carcinogenic risk. These differences are particularly interesting given the fact that toxicity data requirements for pesticide registration are fairly standardized across most countries (GAO, 1993 and PRC, 1993). They appear to arise largely in the area of data interpretation reflecting each country’s autonomy in regulatory decision- making. This desire for individuality complicates efforts to enter international mar- kets, since a pesticide developed in one country and regarded as noncarcinogenic may be considered carcinogenic in another country. Johnson (1989) captured this reality succinctly when stating that although significant agreement exists with respect to the scientific principles for regulating pesticides, the application of these principles is based on various factors, not the least of which are “national, social, and institu- tional structures and values.” To date, the U.S. has taken the international lead in developing risk assessment methodologies. More countries are becoming involved in developing approaches to risk assessment, such as The Netherlands. International scientific bodies, such as WHO, IARC, IPCS, OECD, and the Pan American Health Organization, are also playing an increasingly important role in advancing this process, especially for developing countries (OTA, 1993). In conclusion, the following two general statements can be made regarding the current state of risk assessment and its future direction. First, the status quo results in an unnecessary duplication of resources and the generation of potential trade barriers. Second, harmonization efforts are underway to establish a uniform approach to data interpretation, especially for carcinogenicity data, in risk assessments. It is clear that international harmonization of risk assessment procedures is an objective which will result in more consistent international risk decisions. REFERENCES Anon., Pesticide Regulation Compendium , 4th ed., Editions Agrochimie, Switzerland, 1993. Barnes, D.G. and Dourson, M., Reference dose (RfD): description and use in health risk assessments, Regul. Toxicol. Pharmacol. , 8, 471, 1988. Dragula, C. and Burin, G., International harmonization for the risk assessment of pesticides: results of an IPCS survey, Regul. Toxicol. Pharmacol ., 20, 337, 1994. Farber, T.M., A Position Document of the U.S. Environmental Protection Agency Office of Pesticide Programs: Selection of a Maximum Tolerated Dose (MTD) in Oncogenicity Studies , Office of Pesticide Programs, Washington, 1987. LA4111 ch29 new Page 534 Wednesday, December 27, 2000 3:31 PM © 2001 by CRC Press LLC INTERNATIONAL HEALTH RISK ASSESSMENT APPROACHES FOR PESTICIDES 535 Government Accounting Office, Pesticides: A Comparative Study of Industrialized Nations’ Regulatory Systems, Report to the Chairman of the Committee on Agriculture, Nutrition, and Forestry , U.S. Senate, Washington, 1993. Hallenbeck, W.H. and Cunningham, K.M., Qualitative evaluation of human and animal stud- ies, in Quantitative Risk Assessment for Environmental and Occupational Health , Lewis Publishers, Ann Arbor, MI, 4041, 1988. Johnson, E.L., Pesticide residues, In International Food Regulation Handbook: Policy, Sci- ence, Law , Middlekauff, R.D., and Shubik, P., Eds., 253, 1989. Office of Technology Assessment, Appendix A: International risk assessment, in Researching Health Risks , U.S. Government Printing Office, Washington, 187, 1993. U.S. Environmental Protection Agency, Guidelines for Carcinogen Risk Assessment , Federal Register, 51, 33992, 1986. U.S. Environmental Protection Agency, Proposed Guidelines for Carcinogen Risk Assessment , Federal Register, 61(79), 17959, 1996. U.S. Environmental Protection Agency, Risk Assessment Guidance for Superfund,Vol. I: Human Health Evaluation Manual (Part A), Interim Final , Office of Emergency and Remedial Response, Washington, 1989. World Health Organization, Classification Systems on Carcinogens in OECD Countries — Similarities and Differences , prepared by Norway and The Netherlands, 1993. World Health Organization, Environmental Health Criteria, Principles for the Toxicological Assessment of Pesticide Residues in Food, Vol. 104 , International Programme on Chem- ical Safety, Geneva, Switzerland, 1990. LA4111 ch29 new Page 535 Wednesday, December 27, 2000 3:31 PM © 2001 by CRC Press LLC . the assessment procedures used by countries. LA4111 ch29 new Page 527 Wednesday, December 27, 2000 3:31 PM © 2001 by CRC Press LLC 528 A PRACTICAL GUIDE TO ENVIRONMENTAL RISK ASSESSMENT REPORTS . Cunningham, 1985). LA4111 ch29 new Page 529 Wednesday, December 27, 2000 3:31 PM © 2001 by CRC Press LLC 530 A PRACTICAL GUIDE TO ENVIRONMENTAL RISK ASSESSMENT REPORTS nogenicity studies PRACTICAL GUIDE TO ENVIRONMENTAL RISK ASSESSMENT REPORTS These include the ability to calculate “cost” per tumor in risk benefit comparisons and to develop “bright lines” or clear risk values upon