Document wD7GzVr3aey2Eg344MpazxnRB

Toxicology and Industrial Health, Vol I, No, 4, 1985 213 PERMISSIBLE CONCENTRATIONS OF CHEMICALS IN AIR AND WATER DERIVED FROM RTECS ENTRIES: A "RASH* CHEMICAL SCORING SYSTEM TROYCE D. JONES, PHIL J. WALSH and ELAINE A. ZEIGHAMI Health Effects and Epidemiology Group, Health and Safety Research Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831 Many chemicab are ofconcern to human health, but only afew have epidemtologically derived risk estimates. About 45,000 chemi cab are listed in RTECS, most of which have had some testing in subhuman modeb. R TECS entries rangefrom cellular effects through organoleptic damage to lethality, with manypathological endpoints Ibied, including mutagenic changes, irritation, teratogenesb, cancer, mortality, etc. However, it b difficult to extend any bio logical test results to human rbk assessments. Ifthe results are extended, the degree of validity b highly uncertain. Thb paper describes a logical basbfor using the entire complex spectrum oftest results to evaluate the overall toxicologicalpotency ofa chemical to be assayed (i.e., an interviewing chemical) and de scribes how to derive tentative, permusible concentrations in air and waterfor any particular chemicalfor which no regulatory guidance exists. Thb approach has been testedfor lb reference chemicab dis cussed in NIOSH Criteria Documents, EPA-CAG reports, etc. The evaluations are uncomplicated, but occasionally it b difficult to match RTECS entriesfor two (Efferent chemicab. Difficult comparIsons may require some familiarity with experimental design and the toxicological literature. One important product ofthb novel approach b that a dbtribution or array ofpotency values b obtainedfor any chemical evaluated. Thb dbtribution reflects many uncertainties stemmingfrom low sta- 1. Send correspondence to: T.D. Jones, Health Effects and Epidemiology Group, Health and Safety Research Division, Oak Ridge National Laboratory. P.O. Box X, Bldg. 4500S, MS F2S6, Oak Ridge, TN 37331, (613} 574-6257. 2. key words: air, chemical scoring, permissible concentrations, relative potency, uncertainty analysis, water. 3. Abbreviations: ACGIH, American Conference of Governmental Industrial Hygienists; ADI, acceptable daily intake; AMTL, ambient monitoring trigger levels; B(a)P, benzo(a)pyrene; CAG, Carcinogen Assessment Group; DMTL, discharge monitoring trigger levels; EPA, V.S. Environmental Protection Agency; LDLo, low acute mortality; LD50, lethal dose to 50% of test organisms; MATE, minimum acute toxicity effluents; MEG, multimedia environmental goals; MTL, monitoring trigger levels; NCI, National Cancer Institute; DMNA, nitrosodiraethylamine; NIOSH, National Institute of Occupational Safety and Health; NRC, National Research Council; PNA, polynuclear aromatics; OECD, Organization for Economic Cooperation and Development; RASH, rapid screening hazard; RTECS, Registry of Toxic Effects of Chemical Substances; TLV, threshold limit values. AP00019996 214 Jones, Walsh and Zeighami tisiicalpower, experimental design, pharmacologicalprocesses, interspecies variability, dose rate, biological effect monitored route of treatment, etc. The array ofrelative values for a particular chemi cal reflects many different biological andphysical conditions. The distribution ofthe array helps to index a composite toxicological profilefor many different biological effects resultingfrom numerous treatment protocols. To minimize the effect ofextreme sensitivity ofcertain (perhaps novel) biological test models, possible errors in the RTECS database, andpossible human pharmacological insensitivity to a particu lar chemical and!or a particular route ofadministration, we consider the interquartile range (i.e., the central 50%) ofthe array of relative potency values between two chemicals being compared as a practical measure of uncertainty. Thus, the range in response derivedfrom variability in relative potency should be useful in addressing the range ofresponse in man as estimatedfrom extrapo lations of test data. INTRODUCTION AND OBJECTIVE Analyses that estimate permissible concentrations ofpollutants in water and air (based on potential health effects) are used to obtain, quickly and cheaply, an approximate idea of the relative hazard of an untested chemical (or technology) or a well-studied chemical under untested conditions. This relative potency or hazard assessment approach is being developed explicitly for setting priorities and research needs related to synfueIs-derived contaminants. Because the mutagenic and carcinogenic potencies of different chemicals have been demonstrated to vary by about seven orders of magnitude, this paper will discuss a method for hazard assessment that is designed (1) for ease of evaluation, (2) for accuracy in the range of twofold to tenfold and (3) to estimate the range of uncertainty associated with the best estimate, which should serve to indicate how relevant to human response the best estimate may be. The best estimate and its range can be used to dispense with many pollution-related health-effects considerations. However, apparent problems may be treated inmore detail before they can be considered insignificant or judged to require pollution abatement technology. The goal of this paper is to provide an objective, quantitative basis for temporary human exposure guidance. The guidance is to be used in the absence of Federal and state standards or to preview how standards may change as more analytical studies are completed. For example, the Environmental Monitoring Plan for the Synthetic Fuels Corporation (1983) states that the corporation should have the burden ofjustifying the need to monitor specific unrelated substances and ofproviding threshold values above which these substances must be monitored. u ..Jr ` AP00019997 Toxicology and Industrial Health, Vol. 1, No. 4,1985 215 To provide a comprehensive study for each chemical or mix of interest would be impossible. For example, the cost for testing one chemical or mixture in a simple animal study is $0.5 million--and possibly more unless the initial tests are unambigu ous. Testing may then be followed by intensive analytical studies and extrapolation of test results to man. NIOSH Criteria Documents and reports by the EPA-CAG provide quantitative estimates of risk. Other studies, such as the 1ARC monograph series (1979-1983), provide comprehensive reviews of a given chemical from which one can make some qualitativejudgment of the risk level. Thus, a good shortcut is needed to supplement these effort-intensive, but limited-in-scope, activities and to address envir onmental problems stemming from many different fugitive pollutants around a chemi cal dump or an industrial plant. A heuristic approach to health-effects analysis, based on currently available, and constantly increasing, biological test data is shown in Table 1. The method to be proposed in this paper will use data from classes I through V to demonstrate a new chemical scoring process that can be used to guide supplemental monitoring activities (Synthetic Fuels Corp., 1983), to rank different chemical dumps for remedial action or to estimate permissible concentrations of chemical pollutants in air and water. A relative potency approach will be used. The relative potency approach to assessing chemical toxicity is at the core of most efforts to extrapolate from what is known to some hypothesized exposure situation for which no data are available for quantifying risk in a reasonably direct manner (McMillin et al., 1980; Albert et al., 1983; Jones t al., 1983; Dudney et al., 1983). Several recent activities have resulted in philosophical guidelines and principles as related to toxicological risks. Some of the concepts are presented in a carcinogenesis framework, but many of the ideas are generally suitable to assessments involving mutagenesis, teratogenesis, acute effects, cardiovascular disease, etc. (i.e., general toxicology). Representative efforts include: the NRC Steering Committee on Identifi cation of Toxic and Potentially Toxic Chemicals for Consideration by the National Toxicology Program (NRC, 1984); Office of Science and Technology Policy (1984); Interdisciplinary Panel on Carcinogenicity (1984); OECD Chemical Group and Step Systems Group Working on Exposure Analysis (1980); Quantitative Approaches of the Carcinogen Assessment Group (Anderson and EPA-CAG, 1983), etc. Most such activities do not bridge the wide gap between theoretical principles and the needed product--quantitative guidance at a pollution site. The NIOSH, CAG and similar groups have studied a few industries and/or chemicals, but fossil-fuel-based plants must have a reasonably good estimate of pollution levels and toxicities of hundreds or thousands ofchemical compounds and mixtures. The Oak Ridge National Laboratory has worked extensively in both theory of risk analysis and technologyspecific applications (Walsh et al., 1982; Dudney et al., 1983; Jones et al., 1983; Jones, 1984; Jones et al., 1984). Many different chemical scoring systems are available, and we have reviewed the important features ofmore than 40 such systems. Most ofthe systems are designed for a AP00019998 216 Jones, Walsh and Zeighami TABLE 1 A Heuristic Approach to Human Health Analysis Based on Short-term_______Biological Tests and Any Available Epidemiological Data a** Typ* of Data Texie A((it leasee to be Resolved i Epidemiolegietl Specifie agent of iatarest - Stsdy Design - Confounding faetora (a.g., age, atx. smoking# exposure to otber agaats) - Dosa-raapoata nodallag n Epidemlologlcal Agent eiailar to Claas I - Cites I iaanat - Pharmacological diffaraneea - Toxleologieal diffaraaeaa hi Animal Spaelfio ageat of intareat - Collaetioa or aiaialatiea of toxic ageat - Experimental deaign - Physical aealiag of deae to aaa - Biological aealiag of raapoBM to aaa (e.g., eoapariaon ea basis of genetics, morphology* Immunology, ouymolegy) IV Aniatl Agent limilar to Claas III - All Claaa III iaaaaa - Pharmacological diffaraaeaa - Toxioological diffaraaeaa V Data of Mckaalatie Potentially all raise - Mechanistic uderataadiag of dlaeaae preeaaaea - Relative bilnloz of diffaraat tnbtta&ces - Appropriata asrrogata or iadiaator agaata - Halflad modal of Biological orgaaiaatloa (i.a.. molccalaa to aaa) - Individual variability in biological orgaaiaatloa specific purpose, but some are more general. It may seem foolish to propose still another different chemical scoring system, so in the remainder of this section we will briefly describe six ofthe best and/ or most prominent From this minireview, it will be clear why we propose a new RASH assessment for chemicals. Permissible concentrations of 65 chemicals in water are estimated in the NIOSH Criteria Documents, and the EPA-CAG activity has estimated carcinogenic risk for 43 chemicals in water and 21 chemicals in air. These efforts usually involve very limited data selection, selection of one particular dose-response model or selection of some arbitrary level of risk (e.g., 10"s)(Anderson and EPA-CAG, 1983). Permissible concen- ~.r APOOOf9999 Toxicology and Industrial Health, Vol l. No. 4, 1985 217 trations given by the Criteria Documents and by CAG reports may have a significant degree ofuncertainty but are intended to avoid underestimating risk. For example, the estimates of risk given for bischloromethylether seem to be extremely high. These" estimates are high probably because the data analysts chose the most sensitive biologi cal tests and the one-hit model (in the Criteria Document [Sittig, 1980]), a linear model or a linearized multistage model (by the CAG [Albert et al., 1977; Anderson and EPA AG, 1983]). CAG and Criteria Document concentrations were generally derived for extreme safety, or for a risk level of 10~s, although the composite biological potency ofbischloromethyl ether may be substantially lower, based on all the available test data (Lewis and Tatken, 1980). Nevertheless, the Criteria Document and CAG values should be taken as primary standards or "benchmarks" because of the large investment of effort and scientific experience. However, uncertainty in animal models cannot be eliminated and should be used to our advantage in extrapolating to humans. CAG estimates are somewhat different from values in the Criteria Documents. A single number (for relative potency) may help to classify a kinetic system (i.e., a specified biological model under fixed conditions), but cannot be considered adequate to profile the kinetic system under different operating conditions. Thus, there may be no best estimate except for a fixed biological test model subjected to constant biological and physical conditions. A third benchmark value (in addition to CAG and Criteria Document values) may be derived by relative comparisons between TLVs as defined by the ACGIH (1980b). TLVs are provided for about 500 chemicals. For most of the chemicals studied, relative potency estimates derived from the three benchmark values are in close agreement; however, for certain chemicals, there may be significant spread. Any scoring method that can serve as a good shortcut to ranking technological priorities or to assessing risk for many unregulated chemicals should be expected to differ somewhat from individual benchmark values. But to be useful, there must be reasonable agreement between estimates of a candidate system and the benchmark values for most regulated chemicals; Le., no consistent bias. Most chemical scoring systems are special-purpose systems not well suited to ranking chemical dumps or guiding industrial plants. The systems are generally characterized by (1) dependence on expert judgment, which leads to much controversy between experts when comparing different chemicals, different exposures or establishing reme dial priorities^(2) arbitrary safety factors that change from chemical to chemical, (3) very limited selection of test data--even limited selection of data from one doseresponse experiment (i.e., not all data from a given study are used [Anderson, and EPA-CAG 1983]), (4) being subject to false positive and false negative conclusions (false conclusions are greatly reduced by using all or much test data and by considering the complex progression oftoxicological responses, in contrast with assessment activi ties that select one or two biological tests and on this basis consider potential risks to humans), (S) arbitrary combination of numerical subscores, (6) inaccuracy when i i i I AP00020000 218 Jones, Walsh and Zeighami compared with the benchmarks (which are subject to errors, but because of the extensive analytical effort of a great many experts, the benchmark values are assumed to be accurate in this analysis) and (7) lacking a reasonable estimate of possible uncertainty. For general problems, two of the more prominent scoring systems are the EPA-MATE/MEG approach (Kingsbury et al., 1979) and the method of McMillin ct ah (1980), commonly referred to as the Monsanto system. The MATE/MEG approach is based mostly on the rat oral LDso for some chemicals and the NIOSH ordering number for other chemicals. The ordering number is a four-digit number that simply reveals the history of toxicity testing on a specific chemical--nothing more. For example, the first digit corresponds to the "highest priority species" that has been tested and found to have a positive response. Species assignments include: human=7; monkey = 6; domestic animal = 5; rat = 4; mouse = 3; and hamster = 2. This first digit dominates the resulting hierarchy. The EPA is currently reviewing a report on MTLs for process characterization studies (Kingsbury and Chessin, 1985). The MTLs are offered to guide data acquisition but are not intended for risk assessments or to identify concerns. Values are given for ambient conditions (AMTL) for a 24-hr exposure and for discharge conditions (DMTL) for a 15-min exposure in air, water and soil. It is assumed that DMTL= 100 x AMTL (based on 24 hr/15 min " 96) with some exceptions for irritants, toxicants known to produce late somatic effects and chemicals that persist in air or water. Separate values are also given for potential health effects, zero threshold considerations and potential environmental effects. Uncertainty or safety factors of 10,100 and 1000 are assigned depending on available data. The AMTL value for what is called a "nonthreshold" pollutant is defined as corresponding to an estimated human cancer risk of 10 s in 70 years. Based on the CAG values, the DMTL values correspond to a cancer risk of 10~3 in 70 years. For MTLs, cancer is taken to be independent of route of intake. A potency index, which equals (Incidence of Tumors in Animals)(dose)~,/3(100), serves to determine trigger levels for many nonthreshold chemicals. The values may be modified according to absorption factors for inhalation or ingestion. For the Monsanto Corporation, McMillin et al. (1980) used production volumes, emission factors, atmospheric half-lives and potency relative to B(a)P to rank atmos pheric pollutants. They decided that the potency of the possible and probable carcinogens would be determined, where possible, by the average of the mutagenic potency relative to benzo(a)pyrene using two different mutagenic tests and the carcinogenic potency relative to benzo(a)pyrene using two species or routes of administration. McMillan and co-workers also reviewed 125 chemicals and found that required data were unavailable for many compounds. Thus, they developed a more relaxed system to compare relative potencies. Features of this system included the following: (1) potency AP00020001 Toxicology and Industrial Health, Vol 1, No. 4,1985 219 was an average of one value for mutagenesis and one value for carcinogenesis; (2) missing data were treated using an average from other compounds within a chemical group; (3) B(a)P was used as the standard because of the availability of data from a large number of tests; (4) relative carcinogenic potencies were based only on data from the same species and route; (5) when several potencies were computed, the log average was used; (6) average mutagenic potencies were assigned to test compounds--even if those compounds were found to be negative by the test concerned; and (7) for test systems without (B(a)P data, secondary standards were used. Although the Monsanto system seems to be one of the better general-purpose scoring systems, several improvements are needed for many practical applications. Desirable modifications include: (1) quicker and simpler access to the toxicological literature-- McMillin reviewed 2300 publications to rank 125 chemicals; (2) potency factors should be based on the entire toxicological profile of an interviewing chemical or technology, (3) only real data should be used in preference to extrapolated values, i.e., average values from chemical homologues should be assigned for missing data on the chemical of interest only as a last resort (if all available test data on the chemical of interest are used, then it should not usually be necessary to extrapolate from chemical homo logues); (4) the computation of relative potencies should be carefully prescribed when much relevant data are available, but the algorithm should be subject to several less rigorous modes of comparison when very limited toxicological testing has been per formed; and (5) a meaningful measure of uncertainty should be provided to illustrate the consistency of the test data and the mode of comparison described in point 4. Thus, in this paper we propose to use all existing toxicity test data that can be evaluated by many different analytical techniques. The scatter of all the numerical evaluations attempted probably serves as a useful measure oftotal error in the system, i.e., random or statistical sampling error plus the much larger error descriptive of deviations between different experimental designs and the true scenario of interest. Because we plan to rely on readily available information--as exemplified by the NIOSH single-source document for basic toxicity information and for other data (i.e., RTECS)--many backup considerations are desirable to help flag concerns that may result from chemicals that test either as false negatives or false positives, as well as to help index overall uncertainty. The backup logic is especially attractive because the metabolism of a toxicant may differ between test species and humans in ways that produce false-negative or false-positive results with regard to possible human hazards. The appropriate test battery may be incompletely performed, but there may be other data, such as extensive information on the mechanisms ofaction in several species, to obviate a need for additional tests (NRC, 1984). Also, "it should be emphasized that calculating quantitative estimates of cancer risk does not require that an agent be a human carcinogen" (EPA, 1984). AP00020002 220 Jones, Walsh and Zeighami We have developed (but not published) a very complex, idealized and tiered logic tree (which selectively chooses from toxicity data) that would probably be our preferred approach if large amounts of biological test data were available. However, in practice, test data are usually very scarce. It is usually necessary to consider all existing test data for both acute and chronic effects that encompass both in vitro and in vivo test models, but this simple act helps greatly to obviate the false negatives and false positives. Even considering all levels oftoxicity data, many chemicals will be untested or tested in only one or a very few biological models that may be very poor analogues of man. METHODS Mechanisms of action: Usefulness of RTECS comparisons. Although the relative potency approach is used extensively in most chemical assessments, it is frequently and harshly criticized by some, who sometimes accept the general concept in a slightly different format--e.g., structure-activity comparisons. According to Saffiotti, potency is measured by an effect that is produced by an interaction ofthe agent with a host. Thus, effects are dependent on the conditions under which the interaction occurs. Therefore, we are measuring something that is a variable by itself. At the present time we find such a marked variation in the response, both at the inter-individual level, particularly as it applies to people, and at the species-to-species or biological system to biological system level, that we cannot really extrapolate the magnitude of an effect as measured in one case to the prediction of an effect in another (Saffiotti, 1980). The etiological molecular processes oflate somatic effects such as cancers or cardiovas cular diseases are much studied but incompletely understood. Correspondence between molecular interactions and human diseases have not been established; how ever, there is generally good correlation between DNA damage and initiation of primordial carcinogenic lesions (Ashurst et al., 1983; Rajewsky, 1972; Farber, 1973; Brooks and Lawley, 1964; Amlocher et al., 1977). Furthermore, there is strong and rapidly increasing evidence that compensatory cellular proliferation in response to toxic injury is a direct quantitative measure ofinduced carcinogenic promotion (Jones et al, 1983; Jones, 1984; Jones et al., 1984; Greenfield et al., 1984; Cohen et al., 1982; Argyris and Saga, 1981; Evans et al., 1978; Shami et al., 1982; Ying et al., 1981). Significant doses of most chemicals can cause irritation, focal necrosis, compensatory cellular proliferation and a general progression of toxic response symptoms ranging from acute transitory effects to late (orchronic) somatic effects. Because ofthese factors and because of the general correspondences outlined above, the relative potency of a chemical should maintain some degree of consistency when measured in various biological models, spanning molecular interactions to organoleptic processes, when pharmacological toxiflcation/detoxification processes have been taken into account AP00020003 Toxicology and Industrial Health, Vol. I, No. 4, 1985 221 (Carver et alM 1979; Meselson and Russel, 1977; Bridges, 1980; Heddle and Athanasiou, 1975; Ames, 1979; Bender, 1980). Ofcourse, some variability must occur depend ing on the pathological effect observed, dose level, dose rate, species, strainrage, nutrition, environmental conditions, pharmacological rate constants, enzyme inven tory, membrane permeability, route of chemical intake, chemical carrier or aerosol used and pathological protocol of diagnosis, among others. These and many others processes can induce variability in the potency of one particular chemical relative to a reference chemical. In many cases the range may be small, but in some cases the range may encompass orders of magnitude. Usually, in any particular biological study, the level of response is highly sensitive to only one or a few ofthe listed variables. Thus, one observes a fairly stable relative potency value instead of a potency value that has great variability. However, experimental and physical parameters can be adjusted to illus trate the extreme effect. We consider the range ofuncertainty to be one ofthe extremely usefulparameters ofhuman risk associated with a given chemical, in contrast to Dr. Saffiotti, who considered the variability as an impediment to risk studies. The range in response derived from variability in relative potency should be useful in addressing the range of response in man as estimated from extrapolations of test data and also the range ofindividual sensitivity ofanimals within a given biological test model. The other scoring methods which we have reviewed have no comparable measure ofuncertainty. The relative potency approach provides aframework for the use ofmultiple models and data bases to estimate the potential impacts of chemicals about which we know little. For example, if sufficient human exposure-response data exist, it is possible to make direct estimates of health risk in the exposed population. If sufficient human data are not available, the relative potency method can be used to consider all relevant biologi cal test data as long as the chemical of concern and the reference chemical have both been tested in the same biological model (preferably under the same experimental conditions). In this framework, we can also choose different models of dose response and judge the predictability of various subhuman systems as indicators for human health effects. Sources ofdata. There are many well-known ways to collect test data on a chemical of interest. Examples include: Chemical Abstracts, Carcinogenesis Abstracts, Medline, Toxline, Cancerline, Index Medicus, Documentation of the Threshold Limit Values (ACGIH, 1980a), Patty's Industrial Hygiene and Toxicology (Clayton and Clayton, 1981), Criteria Documents, Survey of Compounds which Have Been Tested for Carcinogenic Activity (NCI, 1961-1973), National Toxicology Program: Second Annual Report on Carcinogens (U.S. Department of Health and Human Services, 1981), etc. Data selection requires painstaking consideration in the Criteria Documents, CAG reports and the McMillin/ Monsanto method. It appears that the data selection effort offers the greatest opportunity to shorten the evaluation or scoring process. Because it is both difficult and controversial to select the most relevant test data given the condition that maximum resources ate available (Anderson and EPA-CAG, 1983), we AP00020004 222 Jones, Walsh and Zeighami decided to use the entire toxicological testing profile as listed in an updated and supported single-source document. The RTECS is published annually, and the 1980 publication (in 1982) contained 45,156 substances. By presenting data on the lowest reported doses that produce effects by several routes of entry in various species, the Registry furnishes valuable information to those responsible for preparing safety data sheets for chemical substances. It is not the purpose of the Registry to quantitate a hazard through the use of toxic concentration or dose data that are presented with each substance. UNDER NO CIRCUMSTANCE CAN THE TOXIC DOSE VALUES PRESENTED WITH THESE CHEMICAL SUBSTANCES BE CONSIDERED DEFINI TIVE VALUES FOR DESCRIBING SAFE VERSUS TOXIC DOSES FOR HUMAN EXPOSURE (Lewis and Tatken, 1980.) However, in our opinion, tentative, permissible exposures for many different chemicals must be derived. Also, the authors of the RTECS most likely did not anticipate our particular use ofthe RTECS database. Risk assessments and chemical scoring must be performed. For this particular application we have chosen RTECS to be the most practical and convenient single-source reference, but our RASH method could use other sources of data. Rulesfor R TECS comparisons. According to the RTECS listings, one may find that x (mg/ kg) of a chemical has produced a particular effect such as LDLo in a particular species (i.e., low acute mortality) and y (mg/kg) of B(a)P or some other reference chemical tested by some other investigator was required to induce LDLo in the same species. The potency ofthe first chemical relative to the standard or reference chemical would be y/x. Thus, if the reference chemical was considered by NIOSH Criteria Documents, CAG, or some other regulatory agency to be safe at a concentration in water of 1 p$/ L, then the unregulated chemical could be limited to (1 Mil L)/(y/x). Another type of comparison of RTECS entries is that x (mg/kg) of a chemical produced an effect such as unscheduled DNA synthesis in 24 hr, but x (mg/ kg) of B(a)P caused unscheduled DNA synthesis in 3 hr. The relative potency of the interviewing chemical could then be taken as 3/24 that of B(a)P. Some comparisons will require both techniques. In this paper, we are simply using different biological test results; we are not suggesting that unscheduled DNA synthesis, delayed DNA synthesis, etc., are direct measures of carcinogenesis and mutagenesis. However, the array of different potency values stemming from all the different tests help to index the toxicity profile of any chemical of interest. When computing the array of relative potency values, one should use a consistent system of units for all comparisons, i.e., either mg or /zmol. Units should not be mixed when going from test to test. For chemicals that have 20 or more RTECS entries, it may be straightforward to match several RTECS entries (i.e., "exact" matches) between B(a)P and the interviewing chemical. Sometimes it is necessary to use secondary standards to supplement B(a)P. AP00020005 Toxicology and Industrial Health, Vol. 1, No. 4, 1985 223 The chemicals chosen as secondary standards should have several RTECS entries that match the interviewing chemical and several RTECS entries that match the primary standard, i.e., B(a)P in our case. In this paper, benzene, cadmium and DMNA (N-nitrosodimethylamine) were sometimes used as secondary standards. Their median potency values relative to B(a)P were 0.079, 0.0030 and 0.23, respectively. For some chemicals that have a small number of RTECS entries, it may be difficult to match the primary and secondary standards in an "exact" match. Often, however, "near'' matches are possible. For example, lethality endpoints ofLDso or LDLo may be matched across species. Also, certain routes of intake in the same species may be compared; e.g., intraperitoneal injections may be matched with subcutaneous injec tions, intramuscular injections, certain implants, etc., but may not compare well with intravenous or intracerebral injections. Many other examples of"near" matches can be given. The third and more difficult mode of comparison of RTECS entries requires a significant degree ofexperience with the toxicological literature and with dose-response modeling. This third mode is referred to as "reasonable" comparisons. These compari sons are needed for chemicals that have had limited testing or have been tested by a few novel assays that do not match tests of the primary and secondary standards. This mode of comparison is more subjective and can be as good or as bad as the experience of the person making the comparison. We feel that "reasonable" comparisons can produce good estimates ifthe logic is well based (e.g., see Jones and Walsh, 1983; Jones et al., 1984; Jones et al., 1983). Limited space prohibits a complete description of "reasonable" matches, but the data in Table 2 illustrate the mechanics of the relative comparisons. Basic relative potency comparisons were described in the first two paragraphs of this section. If secondary standards are used, the comparison is only slightly more complex. For example, in Table 2, LDLo for subcutaneous (scu) injec tions in rabbits (rbt) is 300 mg/kg for arsenic and 6 mg/kg for cadmium. The relative potency is 6/ 300. Because the potency ofcadmium relative to B(a)P is 0.079, the potency of arsenic relative to B(a)P is (6/300X0.079) = 1.6 x 10`\ Although many "reasonable"comparisons are difficult and subjective, the interquartile range is a measure of the success of the comparison, and there can be much stability through the "exact," "near" and "reasonable" modes of comparison--provided the modeler is careful. From the array of potencies for a particular chemical, we select the median potency as most characteristic of the interviewing chemical relative to the standard and the interquartile range as the most practical estimate of uncertainty. The range will probably span as many processes as are reflected in the tests used to make the comparisons, e.g., random statistical uncertainty, variables ofexperimental design and variables ofindividual response (e.g., enzyme levels). Thus, the characteristic toxicolog ical profile derived from many comparisons may in some cases be more representative of an interviewing chemical than data obtained from a carefully planned and analyzed AP00020006 Chntaicsl TABLE 2 Comparison of the Toxicity of Arsenic with the Toxidlj of B(a)P Aided by the Use of Cadmium, Benzene and N-nitrosodimethylamine as Secondary Standards ___[______________(symbob are the same as those in NIOSH-RTECS)_________________ Blbloglcal test* Type Of Bitlute DOM Toxic RTECB effect Reference ReE. 1 Potency BelstLve to B4e)P Jones, Walsh and Zeighatni AP00020007 ('Exact CoaMClaona*) Afsgnic CadllllM Arsenic B(a)P Cidaiai Arsenic taount ecu-tbt ipr-sua LDLo tdco ipr-gpg LDLo 304 ag/kg sg/leg 40 sg/kg/(preg. > 240 tag/hg/Ol-lSD preg.) 22pg/kg/(te preg.) 10 sg/kg 52? *g/kg -- TER TER TEX __ ASB1AL PBQTA TJADAB ADTOAS TJADAB CRSSAW RBTXAC 24, 442, 3B --. 55 15, 31A, 77 7, 2, 70 13, 3JA, 76 1, 144, ] 1. 42, SC 1.6 x 10"3 6.0 4.4 z 10-7 0.26 ('Rear CoaeariasnaM Irumtr Benten* B()P xrfeplc MtMM Arseoic Benzene ipr-awa Kt-BM Lsg-rat lpr-rat eo-gpg Ipr-gpg mco TDLO TDLO LDLo (.Me LOCO LOCO 00 ng/kg/tpeeg.) 2700 tag/kg/(13D prsg.) 240 sg/kg/Ul-ISD preg.) 20 sg/kg US0 sg/kg 300 Mg/kg 52? tag/kg TEE TER TBR -- -- __ -- TJADAB AflBMAO PSBBAA HCIUS TXAFA9 ABB1AL RBTXAC 14, J1A, 77 17, 2tS, 70 13S, 14, 70 43-64-BOS, SEPT. 70 1, 154, 59 24, 442, M 1, 42, 54 0.34 6.0 5.0 X 1C"2 0.0 s 10'3 ("Other Seasonable Couarlaona*) Arsenic Bsnssne DMHt DMA B(a)P Bensene Benzene CsAaius cu-rbe lvn-rbt lm-rbt ipr-aua Ipr-sui ipr-ret. lrp~9P9 ISS-IIM krmenfr Csdalun Csdslus Ipt-wu lwn-rst ivn-hsn 1DU> LDLo LDCo LOCo LDCo LDLo LDLo LDLO TDLo TDLO TDLO 300 sg/kg Mg/kg 40 tag/kg 9 sg/kg 500 sg/kg 1150 sg/kg 527 sg/kg 2S sg/kg 40 tag/kg/(prsg.) 1250Mg/kg/(9D preg.) 2 sg/kg/CID preg.) ___ -- -- --- -- -- TER TER TER A6BIAL JTEBDC BJIHAC TXMk? TXAfA* TXAPA9 HBTXAC MC1UB TJADAB EVHPAI tUlM 24, 442, M (Suppl. 2), 45, 77 11. 1ST, 54 21, 288, 72 23. Ill, 72 1, 154, 59 1, 42, 56 PI-41-64-BA4 IS, 31A, 77 20, 245, 79 25, 56. 69 1.5 x 10"3 3.2 i 10-2 7.2 x 10"3 1.7 1.9 X lO-2 0.0 1C'3 6.6 X 10-7 i 3.5 X ID-3 4.0 > 1a-3 Toxicology and Industrial Health, VoL 1, No. 4,1985 AP00020008 Cfceaic*! ACHalS Arsenic B(A>P Bansene 1MMM Cadaiua Cadalut BMNA MKA Arsenic Cadalua Cadalua unwu BIsJP DMMA DNM Arsen ie B(a)P BCalP B(a)P Bl)P M*)F B(a)P U)P OHM OHM OHM OHM OHM OMXA oraw OMU oraw TABLE 2 continued t Biological THta TtV* of Eat latte Bom ocl-aea ocl-cat ipc-aua ipr-rat scu-rbt laa-haa Ivn-rab Ipt-aua aw-m scu-rbt lM-h iva-rbt lpc-aii ivn-rat ipr-sus i^-rbt lpc-tu Use-rat actt-flua lm-au ipr-aus acu-aky acu-haa ipr-rat laa-rat ivn-cst ipr-aus ccu-au* ipt-eat ipr-au* Ipr-aus ipr-rat TOLo TDLO LM*a LDLo LHo LOLo LOLo L0LO LOLO LOLO LOLo LOLo LOLo LOLO LOLO LOLo LOLo TDLO TDLo TDLO TDLO TDLO TDLo TDLo TDLo TDLo TDLo TDLo TDLO TDLO TD TO TD TD 120 agAg/(pfg.> 1000 ag/kg/(preg.) 10 eg/kg 500 ag/kg USO ag/kg M agAg C ag/kg IS ag/kg 40 agAg 9 ag/kg MO ag/kg C ag/kg 25 ag/kg SB ag/kg SOO ag/kg 40 ag/kg ag/kg 74 JU9A9 14 ag/kg S ag/kg laoo ag/kg 10 ag/kg 200 ag/kg 40 ag/kg 400 ag/kg 20 ag/kg 14 ag/kg 14 ag/kg 7 ag/kg 70 ag/kg/]M*c 30 ag/kg/ioo-c 14 ag/kg 10 ag/kg 30 ag/kg TOlUC XTEC5 BEfect Reference TER TER -- -- -- -- -- -- -- -- -- -- -- -- -- --a a-- ETA ETA ETA ETA m CAR ETA KTA REO RTA m MGO HBO ETA BTA ETA REO TJADAB EXPEAM CRSBAH TXAPA9 TXAPA) JTEBM PR0TA MCI US BJIAAC TXAPA9 AS8XAL PBOTA HC1US JTMD4 Tuns BJIHAC TXAPAS iEKBAI 1KUI TP1TM TMHAV JIKIAH BJCAIl IBlttA cnnu aKAAI IMMQ HOMO CBIRAS ucjua GISAAA CHIIA3 JMCIAH BJCAAI Bef. 1 15, 31A, T7 20, 324, <4 11, l4, 14 23, 244, 77 1, im, as (Suppl. 2) 45, 77 --, ", 45 pR-4 3-44-144 11, 147, 54 23, 241, 72 24, 442, 34 --, --, SS pR-41-44-444 (Suppl. 2), 45, 77 23, 244, 72 11, 147, 54 23, 244, 72 52, 435, 42 12, 45, SS M, 31, 71 29, 109, 71 1, 225, 40 39, 741, 79 127, 594, 44 32, 140, 72 41, 345, 40 4, 37, 44 4, 37, (4 1, 395, 70 20, 171. 44 45(4), 17, 10 30, 11, 70 42, 1199, 79 2), 50. 74 Potency Relative to B(alP 4.3 50 0.54 4.4 s 10*2 4.7 k 10-2 0.20 0.94 0.22 1.4 > 10*3 4.4 10-3 1.5 k 10-3 1.7 3.3 1 lO-2 7.2 ID"3 0.21 4.7 * 10-J 1.1 K ID-2 8.13 2.7 D. 53 5.3 1 10-2 4.4 X lO-2 5.4 10"l s.i a to-2 2.2 X 10-3 0.22 9.4 * 10-3 4.S x 10-3 3.2 x lO-2 9.4 x 10-* ( 226 Jones, Walsh and Zeighami ORNL-OWQ 94-14342 MERCURY MERCURY. IACETATO) PHENYL MERCURY (It) CHLORIDE MERCURY. CHLOROETHYL MERCURY. O-CYANOQUANIDINO) METHYL 10"* 10-3 10-2 KT1 10 POTENCY RELATIVE TO 8U)P 10 + - MEDIAN VALUE l {-INTERQUARTILE RANGE FIGURE 1. A comparison of(he toxicity of metallic mercury to the toxicity offour important organic and inorganic mercury compounds. study of laboratory animals. Because we are using upper 95% estimates or concentra tions corresponding to a risk level of 10"5 from the Criteria Documents and CAG for our chosen standards (e.g., B[a]P) and are also using a relative potency scheme, the estimates of permissible concentrations derived by our RTECS comparisons should also be considered as very approximate upper estimates of risk for most of the chemicals being assayed. A larger safety margin could be attained if the upper limit of the interquartile range were used instead of the median value. However, for many practical concerns, this degree of added safety is unnecessary and expensive. RESULTS The Criteria Documents and other comparable activities often include chemical deriva tives with the precursor chemical when permissible concentrations are derived. Some times the daughter chemicals have similar pharmacological, absorption, transport, conjugation and elimination rates, and sometimes the rates are very different. The critical organ may be different, and the daughter may be more or less toxic than the parent: Figure 1 illustrates the toxicological profiles of elemental mercury and repre sentative mercury derivatives. Ofcourse, the overall toxicological potency will include pharmacological processes that can be greatly different for organic and inorganic compounds of mercury. The persistent problems in risk analysis activities are also common to the RTECS database. Frequent problems include: (1) in some cases very few biological tests are available for a particular chemical; (2) the entries (tests con ducted) for an interviewing chemical do not correspond to entries for the reference or comparison chemicals; (3) the biological test data for some chemicals, such as arsenic AP00020009 Toxicology and Industrial Health, Vol. I, No. 4,1985 227 ORNL--DWQ 34-14943 "EXACT COMPARISONS" <N - 4) 1__________ 1 + -1 "NEAR COMPARISONS" (N > 4) + "EXACT AND NEAR COMPARISONS" IN 8) + "OTHER REASONABLE COMPAR ISONS" (N - 39) + "ALL COMPARISONS" (N - 47) + --------------- ---------------- ---------------- 1 tO-3 ID-2 10-1 10 POTENCY OF ARSENIC RELATIVE TO B()P tO1 COMPARISONS ARE WITH THE PRIMARY STANDARD [B(a|P] AND SECONDARY STANDARDS (Cd, BENZENE, AND DMNA) + - MEDIAN VALUE | j INTERQUARTILE RANGE FIGURE 2. Variations in the estimated potency ofarsenic as different comparisons were made from RTECS entries. and bischloromethylether (and previously with benzene), may be quite ambiguous, or erroneous tests may have been conducted. Table 2 and Figure 2 illustrate the effect, on the median potency and the interquartile range, of different modes of relative compari sons of the RTECS listings. Arsenic is a difficult chemical to evaluate because, according to some (e.g., see Gori, 1980), it is a chemical needed for the biological processes of hemopoiesis and phosphorylation. However, according to others, it has yet to be demonstrated that arsenic is essential or needed in physiological processes in man. Arsenic is toxic at higher doses and induces tumors in man. Many other chemicals are essential at low concentrations, but can be toxic or carcinogenic at higher doses. For example, a toxic dose of selenium is only about twice the desired nutritional intake. Thus, dose-response curves that are of a nonsymmetric U-shape are typically envi sioned for a large number ofchemicals, and any chemical (even oxygen or water) can be used to stimulate a complex spectrum of toxicological effects that could include mutagenesis and enhanced tumor potentiation (Jones et ai. 1983). However, carcino genesis testing in animals under strict "one chemical treatment** protocols is negative. Figure 2 illustrates the high degree of stability in the somewhat subjective process of deriving relative potency values from RTECS entries. Orders of magnitude range in relative potency and/or in different biological responses should come as no shock to persons of diverse biological experience. However, to certain specialists, large ranges are often confusing. Large ranges are often forgotten at the end of assessments or standard setting activities. But uncertainty is ubiquitous to permissible concentrations, ADls, and risk assessments. AP00020010 228 Jones, Walsh and Zeighami ORNLDWGS4 I4M4 ARSENIC BENZAfe} ANTHRACENE BENZENE BENZIDINE 8ENZ0U1PYRENE BISCHLOHOMfiTHYLETHEfl CADMIUM CARBONTETRACH LOR IDE CHLOROFORM CHRYSENE ENTHYLENIMtNE FORMALDEHYDE HEXACHLOROBENZENE N-NITROSOOIMETHYLAMINE TETRACHLORETHYLENE trichloroethylene VINYL CHLORIDE 10' + CT 10 -5 C T *3+ +c W | CN M+ Uw T A STANDARD T 43,' C - 490. W 7400 CW M 4c WM 1t__________ * T W TM MT W Cj --CAlrt M WC C+W M 10 -4 WA MT I" 10>3 10-2 10 -1 POTENCY RELATIVE TO 6()P 10 10 + OUR ESTIMATE BASED ON RTECS I I - INTERQUARTILE RANGE A CAG - AIR W - CAG - WATER M MONSANTO C - CRITERIA DOCUMENT VALUES FROM S1TTIG 119801 N - EPA - MEG VALUE DERIVED FROM NIOSH ORDERING NUMBER T RATIO BASED ON TLV RELATIVE TO 0.2 mg/i*3 FOR PPAH *CAG VALUES ARE NORMALIZED TO SAME VALUE FOR At IN AIR AND WATER IN ORDER TO COMPARE CAG VALUES FOR OTHER CHEMICALS TO BU)P. FIGURE 3. Relative toxicities of 16 well-documented chemicals as estimated by EPA-CAG, Monsanto, Sittig (1980), EPA-MEG, ACHIG-TLVs, and our "New Health Effects Scoring Model." ~ There are two basic approaches for dealing with uncertainty. The first approach is to apply large safety factors to account for uncertainties in data and analytical models and, from this process, to recommend a precise estimate of an "acceptable" concentra tion. The second is to use currently available analytical models and data and to estimate an "acceptable" concentration and the range of possible uncertainty. The first method TABLE 3 Permissible Concentrations m Water and Air Based on (a) CAG Values for PNAs in Water, (b) a Permissible Risk of 10-5 and (be CAG Risk Estimator for Nitrosodimethybwiiiie in Air (lc. 23 X IQ'Vfe/uPI) and (c) Relative Potency Factors Derived from RTECS Entries Mater Air Chemical 1 <* tfi Ift/M fyjg/L)* Interquartll* Range of t yo/L>e number of ntrlea In unco 1900 Humber of KtBCs Cemparieonm Toxic effect In RT4CS* Valuea Derived from RTXCS Comipariaona (jg/*3)9 intarquactlle Range Jjg/m3)* Toxicology and Industrial Health, V ol l. No. 4,1985 AP000200I2 Anthracene Arsanlc Banal alanthcacene Benxene Benzidine Ben*o!) pyrene BnaaMOwnnt Biphenyl 4-Blphenylaalne Btacfeloromethylether B1 a < 2-ethylhexyl) - ptithalale Butyl-bnnxylphthalAte Cedmlem Caffeine Carbon tetrachloride Chloroform Chromium Chryaene 0lbens(a,h}aattiracene Dt-n-butylphthalate Diathylphthalate 7-12,plnet by lbeni I a 1 - anthracene 2-4,Dlaetbylphenel Dl-a-octylphthalate Dodacme Btbyl alcohol Ethylene oxide etbylenlelne fluoranthene formaldehyde Hex ac hlo c obe* aene *.7 a iirJ 0.02 *.7 10"3 15 1.S7 * i<rJ *.7 a 10-3 -- -- -- 2.0 X 10-5 5, MO* 5.000* 10 _ 2.5 2.1 e.o > io-3 9.7 x 10-3 9.7 a iff*3 5,000* 5,000* 9.7 x IB-3 Minimise Contact 5,000* -- -- -- -- 200 -- 1.0 a 10-3 0.020 0.022 0.020 6.0 1.2 x 10-3 0.020 -- --- 3.0 t io- -- 0.26 -- 4.0 1.9 -a. 0.020 0.020 -- 0.020 -- -- -- -- -- -- 7.2 to IQ-3 0.20 a.ia 0.020 s.s 0.7 x 10-3 (0.0201 0.23 22 0.000 0.31 100 0.05 to 1.1 0.1 I 10-3 to 4.2 4.5 x 10-3 t1> 0.22 2.5 to 51 2.1 r 10-3 to 0.093 Standard 0.045 to o.as 12 to 25 4.7 x 10-3 to 0.37 0.010 to 4.7 34 to 050 7 9 20 31 22 05 12 2 18 14 14 120 0.35 1.1 13 5.0 7.0 x 10-3 0.13 0.020 20 25 0.023 0.1 to 370 0.25 to 0.39 0.035 to 4.5 4.1 to 53 4.1 to 7.0 2.2 k 10-3 to 0.014 O.05S to 5.4 9.3 r 10"3 to 0.20 1.1 to 250 4.7 to 390 2.1 t 10-3 to 0.044 1 10 54 34 29 3 8 35 S 4 77 4.0 0.97 to 14 4 200 33 to 2500 1.4 1.3 to 22 3< 14 to 175 4 1 40 1.3 2.2 0.40 0.05 1(9 0.0(2 to 2.0 0.024 to 2.3 0.1 to 4.5 9.0 X I#"4 to 2.3 0.47 to 4S 19 23 5 34 < 5 47 9 5 7 standard 9 3 13 11 IS I.H.M t.(),7 n.n.r l.l.N.R.T H,M,T M.M I,T N,I,T H.T t,I,f 4.3 5.2 0.43 170 0.20 (0.83) 4.9 440 2.4 9.2 3000 7* 3400 5 I,M,T 14 13 32 7 I,B,T 340 4 E,I,H,T 1T0 It (H) 0.23 ( h,h 3.1 IS M,*,T 0.43 LO B,T 30 13 1,1,T 754 27 R,I,K,H,T 0.19 S H.T 140 14 I,I,T 5900 3 <) 49 22 K.I.H, 1100 (H|,T 9 B,I,H,H,T 34 4 I.M.M.I 44 17 M,(M>,T 14 5 I,H,H,T 25 4 I,H.T 4(09 1.5 to 33 0.24 to 120 0.13 to 4.4 75 to 1,500 0.043 to 2.4 Standard 1.3 to 25 340 to 7S0 0.14 to 11 0.52 to 140 1000 to 25,000 240 to 11, OM 7.5 to 12 1.0 to 130 140 to 1600 130 U> 230 9.M to 0.42 1.7 to 170 0.20 to 5.9 52 to 7500 200 to 12,000 0.00 to 1.3 29 to 424 940 to 75,000 34 te 440 440 t 5200 2.4 to 59 0.43 to 49 5.2 to 130 0.027 to 69 20 to 1340 Jones, Walsh and Zeighami AP00020013 Mater TABLE 3 continued Air Cheaical value* 1 Muteier derived Of Meuber froa Inter Bntrlee of Toxic RTECS Inter quartile in RT8CS Effect Coatpar 1- quartile d> W= t<> Mange of t XT8CS Cooper1- in one Range l)|9/MB 1900 eona macs? [yg/u3> yg/a3)g Hercucy Mercury, (aeatatolphenyl Mercury, <Illchlorlde Mercucy, chloroethyl Mercury, (3-cyanegoani- dliwlaaetbyt 3-Mthylcbolanthrene Methylene chloride 1-Maphthylaalne 2-Maphthylmin* Nicotine h-KWswdxsetbjtoaisc Ptienanthrene Pentechlocophcnol pyttiM Pyridine Quinoline Resorcinol Tetrachloroethylene Thiophene Toluene 1,1,l-Trtchlscoethana Tr1chloroethylene vinyl chloride Xylene -xylene o-xylene p- xylene 0.2 0.2 0.2 0.2 0.2 0.7 to 10"* ___ *.7 x 10"* 0.7 to IC7-5 -- -- 9.7 X 10-3 140 *.7 s 10"3 -- 2 -- 12.4 -- 21 517 ___ -- __ -- .. -- -- -- -- 0.020 ~ 0.028 0.028 -- 0.014 0.028 -- 0.028 -- * -- 8 -- __ -- 27 20 __ _ -- 0.70 0.12 0.22 0.28 0.41 8.2 x 10*3 to 2.5 4 9 <K),T 21 0.24 to 75 8.032 to SO 12 12 3.6 0.94 to 1500 0.088 to 1.0 9B I,M,T 6.4 2.6 to 30 3.7 x 10'3 to 2.3 26 22 W,T 8.3 0.17 to 16 0.011 to 0.51 6 16 B,T 12 0.33 to 15 0.019 3.6 X I0"3 to 0.033 72 32 E,H,i,T 0.S5 0.17 to 0.98 11 2.3 to 85 1.4 0.67 to 13 23 17 IrM.OO.t 380 69 to 2S00 9 12 m.oo.t 42 20 to 300 0.014 0.035 to 0.074 20 H,M,T 0.42 1.0 to 2.2 0.074 2.9 to 10-3 to 1.8 34 IS B,T 2.2 0.086 to 52 0.12 0.011 to 1.6 80 22 H,N,T (3.4) 0.38 to 46 O.tSl 0.8 X 10-3 to 0.82 7 6 u,u,r 1.6 D.26 to 24 0.25 0.076 to 3.7 16 16 B, 1,(),T 7.5 2.2 to no 2.8 0.055 to 30 S4 I,K,i ) 83 1.6 to 900 3.7 1.3 to 9.3 12 IS I, H,T no 40 to 280 0.41 0.023 to 2.2 12 12 I,N,M,T 13 0.69 to 64 2.0 0.40 to 44 13 10 i,im ,T 59 14 to 1300 23 11 to 67 11 14 I.H.T 690 330 to 2000 1.1 0.33 to 3.9 2 ia T 32 9.7 to 120 7.4 3.9 to 22 13 11 1,T 220 115 to 640 22 14 to 82 2L 0 l.T 640 440 to 2400 36 11 to 74 32 12 I.N.M.T 1100 330 to 2200 90 4.7 to 120 30 0 M.M,T 2700 140 to 3600 4.0 1.3 to 70 18T 120 40 to 2100 0.0 5.7 to 29 11 12 l.T 240 17Q to 590 7.8 2.5 to 23 18 1.2 to 29 4 7 l.T 210 75 to 690 36 T 520 36 to 860 7.4 2.0 to 23 37 T 220 59 to 690 'ForRTECScoopariiOM.Bt^PwMtlKpnMiyaAndaRl awl betBOK, nkrowdradhytaxiioe, aed cadnmni amciecuNbxyjMndanls. Yataei|rve8 arc extra frfeiathai they aieiKM cotrwaed forpotemieBydiffcrcrMtUofpikiB factor* in lui,kit orut Aiattiiweapheikand water rcaflioa*arcnotoniMa(Kd.FofCMMgpte.bi*chtoroMKthylcthcrdoeompo*cmaier tad nuy otter eoapouads arc iaaolubfc, or only*M|fat)y aoluWe in purc water. *c -- Criteria Document value* from SMg (1980). *W = CAG value* forcaacer. \ s dfriird ftffu RTECS conpvwMi *Valuc in thi colema arc aomwlbed to pennieiitile cxpoaurc of 8-028 m/Lof PNA*. Fcr a differed contewtratioa x, giveti vahiee ehould be multiplied by (x/0.02>). 'E = Teracojcactii. I = Irritation; M = Mwaaeneaia; M = Neopiaiia;(N) = Pouibte Neopleiia, T = Toxkity. Vdue* in tM* cohma aie nonuefaed to a pwarimiNe expoaure of W */* cf ahraeodimcihytaiaine. For * differ** ooeccuiratioa y, gnwa valuta *odd be aukiplted by(yy3.4). "Difficiih to match RTECS eotrkt - ettiauUe very uncertain. y j i Toxicology and Industrial Health, Vol. 1, No. 4,1985 231 yields precise-appearing numbers that do not overlap, unless one considers the uncer tainty factors (often as large as 5000) that have been used in the calculations. The second approach helps us to remember the sizes of the different uncertainties and also to define the likely range of variability in humans. Relative potencies estimated from NIOSH Criteria Documents, EPA-CAG assess ments, Monsanto, MATE/ MEG, TLVs, ACGIH-TLVs and RTECS comparisons are shown in Figure 3. Estimated permissible concentrations in air and water are given for these and other chemicals in Table 3. As seen, the values derived from the RTECS comparisons agree quite well with the NIOSH and CAG values; however, we recom mend that our values of permissible concentrations could be improved in a theoretical sense if corrected for differential absorption; e.g., GI (or inhalation) absorption of B(a)P divided by the GI (or inhalation) absorption ofchemical ofconcern. The GI and lung absorptions of B(a)P are certainly much less than unity, and if the absorption factors are ignored (Le., both taken as unity), then the concentrations in Table 3 would seem to be "extra" safe. In the event that projected human exposures exceed the values presented in Table 3, then somewhat relaxed, but still safe, permissible exposures could be calculated by taking differential absorption into account. The important point of Figure 3 is that (taken over the 16 comparison chemicals) the proposed method seems as accurate as any of the other six activities and requires orders of magnitude less analytical effort. DISCUSSION Ideal decision makingfrom permissible concentrations. For a single chemical and one transport medium (i.e., air or water in this paper), a projected exposure would be considered safe if it is less than the corresponding value in Table 2 (Le., Qe/Ql < 1 where Qe is the projected exposure and Ql is the derived permissible concentration). For exposures to complex mixtures, the composite biological response may be equal to, greater than or less than the sum ofthe individual component responses, depending on such factors as the chemical blend, the exposure protocol or the test receptor. Because the problem ofestimating the toxicity ofcomplex mixtures remains unsolved, the composite response is typically taken to be the sum of the component responses. Thus, for a mixture, the ideal assessment would consider ~ where i = all chemicals. 2 (Qe/Ql* *+5 (Qe/Ql* ii The higher the sum, the greater the potential for health impairment. This approach seems objective and ideal for establishing priorities when one is faced with several sites that compete for remedial actions. If the 2i (Qe/Ql* * + 2i (Qe/Ql* < 1, then less concern or remedial action would be warranted based on current data and available models, unless QeS or QlS were subject to significant changes in the future. AP00020014 232 Jones, Walsh and Zeighami Decision making without projected exposures. If projected exposures cannot be expressed in the same units as permissible limits, or if perhaps the exposure estimates are not available at the desired exposure site, then a relative comparison can still be made. In that case, the derived permissible concentrations can be ordered on some arbitrary numerical scale in the fashion of the Sax method (1979). For example, the estimated permissible concentrations in water given in Table 3 range from 10"6 for bischloromethylether to about 104 for zinc. This finding would suggest the use ofsome subjective scale such as y=4 - logio(x), where x is the permissible exposure from Table 2 and y is a toxicity index. Using this algorithm and permissible concentrations in water from the Criteria Documents, bischloromethylether would score about 8.7; B(a)P would score about 6.0; beryllium would score about 5.1; benzene, about 2.8; dichloro benzenes, about 1.6; zinc, about 0.3; etc. Of course, other such arbitrary scales can be evaluated from Table 3 depending on the nature of the problem to be ranked. ACKNOWLEDGMENT Research was sponsored by the Office of Health and Environment Research, U.S. Department of Energy, under contract DE-AC05-840R21400 with Martin Marietta Energy Systems, Inc. The help of M.D. Morris ofthe Mathematics and Statistics Research Department and the great document preparation skills of Waldean Richardson are greatly appreciated. REFERENCES ACGIH. (1980). Documentation ofthe Threshold Limit Values, 4th Edition. P.O. Box 1937, Cincinnati, Ohio. ACGIH. (1980). Threshold Limit Valuesfor Chemical Substances and Physical Agents in the Workroom Environment with Intended Changesfor 1980. P.O. Box 1937, Cincinnati, Ohio. ALBERT, RE., LEWTAS, J., NESNOW, S,, THORSLUND, T.W., and ANDERSON, E. (1983). Comparative potency method for cancer risk assessment: Application to diesel paniculate emissions. Risk Analysis 3:101-117. ALBERT, R.E., TRAIN, R.E., and ANDERSON, E. (1977). Rationale developed by the Environmental Protection Agency for the assessment ofcarcinogenic risks. J. Natl. Cancer Inst. 5&1537-1541. AMES, B.N. (1979). Identifying environmental chemicals causing mutations and cancer. Science 204:587-593. AMLOCHER, E., DANZ, M., STILLER, K.J., and RUDOLPH, C.L. (1977). Persistent action oflow dose ofcarcinogens on DNA-synthesis in comparison with protein synthesis in the regenerating rat liver. Experimental Pathology Board 14:227-231. ANDERSON, E.L. and CPA-CAG (1983). Quantitative approaches in use to assess cancer risk. Risk Analysis 3:277-295. ARGYRIS, T.S., and SLAGA, TJ, (1981). Promotion ofcarcinomas by repeated abrasion in initiated skin of mice. Cancer Research 41:5193-5195. i J AP00020015 Toxicology and Industrial Health, Voi 1, No. 4,1985 233 ASHURST, S.W., COHEN, G.M., NESNOW, S., DiGIOVANNI, J., and SLAGA, T.J. (1983). Formation of benzo(a)pyrene/ DNA adducts and their relationship to tumor initia tion in mouse epidermis. Cancer Research 43:1024*1029. BENDER, M.A. (1980). Cytogenetic bases for risk inference. In: Measurement ofRisks (G.G. Berg and H.D. Maiilie, Eds.). Plenum Press, New York. BRIDGES, B.A. (1980). An approach to the assessment of the risk to man from DNA damaging agents. In: Quantitative Aspect ofRisk Assessment in Chemical Carcinogenesis, Arch. Toxicol. Suppli 3:271*281. BROOKS, P., and LAWLEY, P.D. (1964). Evidence for the binding ofpolynuclear aromatic hydrocarbons to the nucleic acids of mouse skin: Relation between carcinogenic power of hydrocarbons and their binding to deoxyribonucleic acid. Nature 4934:781-784. CARVER, J.H., HATCH, F.T., and BRANSCOMB, E.W. (1979). Estimating maximum limits to mutagenic potency from cytotoxic potency. Nature 279:154-156. CLAYTON, G.D., and CLAYTON, F.E., Editors. (1981). Patty's Industrial Hygiene and Toxicology, Third Revised Edition. John WQey & Sons, New York. COHEN, S.M., MURASAK1, G., FUKUSH1MA, S., and GREENFIELD, R.R (1982). Effect of regenerative hyperplasia on the urinary bladder. Cancer Research 42:65-71. DUDNEY, C.S., WALSH, P.J., JONES, T.D., CALLE, E.E., and GRIFFIN, G.D. (1983). On the use of relative toxicity for risk estimation. In: Proceedings of Conference on Short-term Biossays in the Analysis of Complex Environmental Mixtures HI (Waters, Sandhu, Lewtas, Claxton, Chemoff, and Nesnow, Eds.) pp. 499-513. Plenum Publishing Corp., New York. EVANS, M.J., DEKKER, N.P., CABRAL-ANDERSON, L.J., and FREEMAN, G. (1978). Quantification of damage to the alveolar epithelium by means of type 2 cell proliferation. Am. Rev. of Respir. Disease 118:787-790. FARBER, E. (1973). Hyperplastic liver nodule. Methods in Canoer Research 7:345-375. GORI, G.B. (1980). The regulation of carcinogenic hazards. Science 208:256-261. GREENFIELD, R.W., ELLWEIN, LB., and COHEN, S.M. (1984). A general probabilistic model of carcinogenesis: Analysis of experimental urinary bladder cancer. Carcinogenesis 5:437-445. HEDDLE, J.A, and ATHANASIOU, K. (1975). Mutation rate, genome size, and their relation to the rec concept. Nature 258:358-361. INTERDISCIPLINARY PANEL ON CARCINOGENICITY. (1984). Criteria for evidence of chemical carcinogenicity. Science 225:683-687. IARC. (1979-1983). IARC Monographs on the Evaluation ofCarcinogenic Risk ofChemicals to Man. World Health Organization Publication, Geneva, Switzerland. JONES, T.D. (1984). A unifying concept for carcinogenic risk assessments: Comparison with radiation-induced leukemia in mice and men. Health Physics 47:533-558. JONES, T.D., DUDNEY, C.S., and WALSH, PJ. (1984b). A unifying concept for carcino genic risk assessments: Comparison with tumors in the EDO! study of 2-acetylaminofiuorene, in review for journal publication. JONES, T.D., GRIFFIN, G.D., and WALSH, P.J. (1983). A unifyingconcept for carcinogenic risk assessments. J. Theor. Biol. 105:35-61. JONES, T.D., and WALSH, P.J. (1985). Animal studies and prediction of human tumors can be aided by graphical sorting ofanimal data: Neoplastic risk from B(a)P, benzene, benzidine, and chromium. Am. J. of Industrial Medicine 7:185-217. APb6620bi6 234 Jones, Walsh and Zeighami KINGSBURY, G.L. and CHESSIN, R.L (1983). Monitoring Trigger Levels for Process Characterization Studies. U.S. EPA, Research Triangle Park, work in progress. KINGSBURY, G.L., SIMS, R.C.,and WHITE, J.B. (1979). Multimedia Environmental Goals for Environmental Assessment, Vols. 3,4, MEG Charts and Background Information Summaries (Categories 1-26), EPA-600/7-79-176a,b. U.S. Environmental Protection Agency, Washington, D.C. LEWIS, RJ., SR., and TATKEN, R.L. (1980). Registry of Toxic Effects of Chemical Substances. U.S. Department of Health and Human Services, February 1982. McMILLIN, C.R., MOTE, L.B. and DeANGELIS, D.G. (1980). Potential Atmospheric Carcinogens, Phase /. Identification and Classification. EPA-600/2-80-015, U.S. EPA, Washington, D.C. MESELSON, M., and RUSSELL, K. (1977). Comparisons of carcinogenic and mutagenic potency. In: Origins of Human Cancer, pp. 1473-1481, Cold Spring Harbor. NCI. (1961-1973). Survey of Compounds which Have Been Testedfor Carcinogenic Activity, prepared by Tracor Jitco, Inc., for NCI, Public Health Service Publication No. 149, U.S. Department of Health, Education, and Welfare, 7 volumes. NRC. (1984). Toxicity Testing: Strategies to Determine Needsand Priorities. Steering Commit tee on Identification of Toxic and Potentially Toxic Chemicals for Consideration by the National Toxicology Program, National Academy Press, Washington, D.C. OECD. (1980). Chairman's Report on the First Working Party Meeting. F. Schmidt-Bleek, Chemical Group and Step Systems Group Working Party on Exposure Analysis. October 13-15, 1980, Expo.-80.121D (working draft). OFFICE OF SCIENCE AND TECHNOLOGY POLICY. (1984). Chemical Carcinogens; Review of the Science and Its Associated Principles. Federal Register 49:21594-21661. RAJEWSKY, M.F. (1972). Proliferative parameters of mammalian cell systems and their role in tumor growth and carcinogenesis. Zekrebsforsch 78:12-30. SAFFIOTT1, U. (1980). Identification, classification and regulation ofpotential occupational carcinogens. Occupational Safety and Health Administration, Federal Register 45(15). SAX, N.I. (1979). Dangerous Properties ofIndustrial Materials, 5th Edition. Van Nostrand Reinhold Co., New York. SHAMI, S.G., THIBODEAU, L.A., KENNEDY, A.R., and LITTLE, J.B. (1982). Prolifera tive and morphological changes in the pulmonary epithelium of the Syrian golden hamster during carcinogenesis initiated by I10Po 7-radiation. Cancer Research 42:1405-1411. SITTIG, M., Editor. (1980). Priority Toxic Pollutants: Health Impacts and Allowable limits. Noyes Data Corp., Park Ridge, New Jersey. SYNTHETIC FUELS CORPORATION. (1983). Final environmental monitoring plan guide lines. Federal Register 49:46676-46685. U.S. DEPARTMENT OF HEALTH AND HUMAN SERVICES. (19B1). National Toxicol ogy Program Chemical Registry Handbook. Produced by the Specialized Services Division of the National Library of Medicine, PB82-142977. WALSH, P.J., JONES, T.D.. GRIFFIN, G.D., DUDNEY, C.S., CALLE, E.E., and EAS TERLY, C.E. (1982). Risk assessment approaches: General definitions, limitations, and research needs. J. Environ. Sci. Health A17:541-552. Y1NG.T.J., SARMS, D.S.R., and FARBER, E. (1981). Role of acute hepatic necrosis in the induction of early steps in liver carcinogenesis by diethyl-nitrosamine. Cancer Research 41:2096-2102. t r AP00020017