- Email: [email protected]
Evaluation of acute inhalation toxicity for chemicals with limited toxicity information
Regulatory Toxicology and Pharmacology 47 (2007) 261–273 www.elsevier.com/locate/yrtph
Evaluation of acute inhalation toxicity for chemicals with limited toxicity information Roberta L. Grant *, Bernard J. Kadlubar, Neeraja K. Erraguntla, Michael Honeycutt Texas Commission on Environmental Quality, P.O. Box 13087, MC168, Austin, TX 78711-3087, USA Received 19 July 2006 Available online 31 January 2007
Abstract A large reference database consisting of acute inhalation no-observed-adverse-effect levels (NOAELs) and acute lethality data for 97 chemicals was compiled to investigate two methods to derive health-protective concentrations for chemicals with limited toxicity data for the evaluation of one-hour intermittent inhalation exposure. One method is to determine threshold of concern (TOC) concentrations for acute toxicity potency categories and the other is to determine NOAEL-to-LC50 ratios. In the TOC approach, 97 chemicals were classified based on the Globally Harmonized System of Classification and Labeling of Chemicals proposed by the United Nations into different acute toxicity categories (from most toxic to least toxic): Category 1, Category 2, Category 3, Category 4, and Category 5. The tenth percentile of the cumulative percentage distribution of NOAELs in each category was determined and divided by an uncertainty factor of 100 to derive the following health-protective TOC concentrations: 4 lg/m3 for chemicals classified in Category 1; 20 lg/m3 for Category 2; 125 lg/m3 for both Categories 3 and 4; and 1000 lg/m3 for Category 5. For the NOAEL-to-LC50 ratio approach, 55 chemicals with NOAEL exposure durations 624 hour were used to calculate NOAEL-to-LC50 ratios. The tenth percentile of the cumulative percentage distribution of the ratios was calculated and divided by an uncertainty factor of 100 to produce a composite factor equal to 8.3 · 105. For a chemical with limited toxicity information, this composite factor is multiplied by a 4-hour LC50 value or other appropriate acute lethality data. Both approaches can be used to produce an estimate of a conservative threshold air concentration below which no appreciable risk to the general population would be expected to occur after a one-hour intermittent exposure. 2006 Elsevier Inc. All rights reserved. Keywords: Acute; Inhalation; Threshold of concern; Globally Harmonized System; LC50; Cramer; Air permit review; NOAEL
1. Introduction The Texas Clean Air Act (Chapter 382 of the Texas Health and Safety Code) authorizes the Texas Commission on Environmental Quality (TCEQ)1 to conduct air permit reviews of all new and modified facilities to ensure that *
the operation of a proposed facility will not cause or contribute to a condition of air pollution. Air permit reviews typically involve evaluations of best-available-control technology and predicted air concentrations related to proposed emissions from the new or modified facility. In a conservative evaluation, worst-case emission rates are
Corresponding author. Fax: +1 512 239 1794. E-mail address: [email protected] (R.L. Grant). 1 Abbreviations used: AEGL, Acute Exposure Guideline Level; ATSDR, Agency for Toxic Substances and Disease Registry; ESLs, effects screening levels; GHS, Globally Harmonized System of Classification and Labeling of Chemicals; GLCmax, maximum ground level air concentrations; GM, geometric mean; GSD, geometric mean standard deviation; hr, hour; LC50 or LD50, the concentration or dose of a chemical which causes the death of 50% of a group of test animals; ln, natural logarithm; LOAEL, lowest-observed-adverse-effect level; LTB, lower tolerance bound; MRL, minimal risk levels; NOAEL, no-observed-adverse-effect level; N–L ratio, NOAEL-to-LC50 ratio; NOEL, no-observed-effect level; OEHHA, California Office of Environmental Health Hazard Assessment; POE, point of entry; REL, reference exposure level; TC, toxicity category; TCEQ, Texas Commission on Environmental Quality; TOC, threshold of concern; UF, uncertainty factor; UTB, upper tolerance bound. 0273-2300/$ - see front matter 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.yrtph.2006.11.003
262
R.L. Grant et al. / Regulatory Toxicology and Pharmacology 47 (2007) 261–273
modeled to predict resulting short-term and long-term chemical-specific maximum ground level air concentrations (GLCmax). In the review of proposed emissions, federal/ state standards and chemical-specific effects screening levels (ESLs) developed by TCEQ toxicology staff are used for criteria and non-criteria pollutants, respectively. ESLs are chemical-specific air concentrations set to protect human health and welfare and are used in the air permitting process to assess the protectiveness of emission rate limits. Short-term ESLs are developed to evaluate acute intermittent exposures of one hour (hr) whereas long-term ESLs are developed to evaluate chronic exposures. The Texas Health and Safety Code is comprehensive. Therefore, ESLs are developed for as many air contaminants as possible, even for chemicals with limited toxicity data. There is a need to develop conservative, health-protective concentrations for chemicals with limited toxicity data for the review of short-term GLCmaxs taking into consideration that ambient air exposure is dependent on meteorological conditions and peak exposures that could occur several times per day. The purpose of this paper is to present two different approaches to establish health-protective air concentrations for chemicals with limited toxicity information for the general public via the inhalation route of exposure. This paper focuses on one-hour intermittent exposure scenarios that are normally the basis for air permit reviews. A method proposed by Munro et al. (1996) for establishing a threshold of concern (TOC) via oral exposure for chemicals in different toxicity potency classes will be discussed followed by an adaptation of a procedure proposed by Layton et al. (1987) and Venman and Flaga (1985) that involved the calculation of the ratios of no-observed-adverse-effect levels (NOAELs) to oral LD50 data for different chemicals. The LD50 is the dose of a chemical which causes the death of 50% of a group of test animals. Health-protective air concentrations derived using these two approaches are presented and compared to published short-term toxicity values. 1.1. A threshold of concern approach Munro et al. (1996) proposed a method for establishing a threshold of concern (TOC) based on grouping chemicals into three structural classes that correlated with chronic oral toxicity potency. Using a system devised by Cramer et al. (1978), organic chemicals were classified in Cramer Class I (least toxic class), II (intermediate class), and III (most toxic class). Munro obtained no-observed-effect levels (NOELs) from chronic, oral animal studies for over 600 chemicals and calculated the fifth percentile of NOELs for each structural class. The fifth percentile NOEL for each structural class was divided by an uncertainty factor (UF) of 100 to account for animal-to-human uncertainty and human variability to establish a conservative threshold dose below which no appreciable risk to human health would occur. The threshold dose could be used for chemicals with limited toxicity information to determine if toxic-
ity testing should be conducted if potential exposure levels were higher than the conservative threshold dose. The approach assumes that the proposed threshold dose for a chemical with limited toxicity information will not be significantly lower than the fifth percentile NOEL divided by an UF of 100. The TOC approach was initially used for food additives and for the evaluation of packaging material (FDA, 1995; Kroes et al., 2000, 2004; Munro, 1990; Munro et al., 1996). Other investigators have used this approach for other products such as food flavorings, personal and household care products and pharmaceutical compounds (JECFA, 1993, 1995, 1997; Kroes et al., 2004; Blackburn et al., 2005; and Dolan et al., 2005). The TOC approach has mainly been used to evaluate chronic oral exposure. However, the basic approach of Munro et al. (1996) could be used to develop a TOC for other routes of exposure such as inhalation exposure. Ford et al. (2006) investigated the inhalation TOC approach for the evaluation of tobacco additives (i.e., low molecular weight volatile compounds) where inhalation is the relevant route of exposure. They collected chronic inhalation NOAELs for approximately 350 chemicals categorized into the three Cramer structural classes. There was a good correlation with the oral database for those chemicals not causing respiratory tract point-of-entry (POE) effects but there was not a good correlation with the oral database for approximately 33% of chemicals that caused respiratory POE effects. Ford et al. (2006) concluded that the use of the chronic inhalation database, with the removal of those substances with NOAELs based on respiratory POE effects, allowed the determination of inhalation TOCs. Although the Cramer structural class categories were useful for establishing a TOC for the chronic inhalation route, a method to predict respiratory POE effects for chronic inhalation exposure must be developed and used in conjunction with the Cramer structural categories. The basic approach of Munro et al. (1996) was followed to establish a TOC approach for the review of short-term GLCmaxs. However, the basic approach was altered in that the acute inhalation database included NOAEL data, not NOEL data, and the chemicals were classified using acute lethality data and the Globally Harmonized System of Classification and Labeling of Chemicals (GHS) proposed by the United Nations (UN, 2005). The GHS classification system is designed to predict acute toxicity potency and is more appropriate for developing an acute TOC approach. 1.2. A NOAEL-to-LC50 ratio approach One of the first steps in determining toxic inhalation potency is to determine the concentration of a chemical in air which causes the death of 50% (one half) of a group of test animals (LC50). Acute lethality data have been used to determine concentrations that are predictive of an increasing likelihood of lethality for the general public (i.e., acute emergency exposure guideline levels (NRC, 2001) and temporary emergency exposure limits (Craig
R.L. Grant et al. / Regulatory Toxicology and Pharmacology 47 (2007) 261–273
et al., 2000)) and to establish short-term occupational limits such as immediately dangerous to life and health levels (NIOSH, 2005). Fay et al. (2005) investigated the use of LC50 data to extrapolate to safe exposure levels for emergency responders for acute hazardous chemical releases. In addition, Suda et al. (1999) investigated the linear relationship between animal LC50 data and occupational threshold limit values (time-weighted average) as well as short-term exposure limits and provided correlation equations that could be used to calculate threshold limit values from LC50 data. Layton et al. (1987) used oral LD50 data for estimating acceptable daily intakes for the evaluation of exposures to contaminants at hazardous waste sites. Venman and Flaga (1985) also proposed the use of LD50 data to establish provisional acceptable daily intakes for the evaluation of waste water contaminants. Both investigators calculated the ratio of NOAELs from chronic animal studies to oral LD50 data for different chemicals and determined the fifth percentile of the cumulative distributions of the ratios. The LD50 value for contaminants with limited toxicity data was multiplied by the fifth percentile ratio to derive a surrogate NOAEL. The surrogate NOAEL was divided by an UF of 100 in order to establish a conservative threshold dose below which no appreciable risk to human health would be expected to occur. We have used the basic approach of Layton et al. (1987) and Venman and Flaga (1985) for evaluating chronic oral toxicity and applied it to acute inhalation toxicity. Therefore, ratios of NOAELs from acute inhalation studies to LC50 data were calculated.
263
cal, methanol, where a calculated 95% lower confidence limit on the benchmark dose at the 5% response level was used as the NOAEL. The study types included those typically conducted in toxicology to investigate shortterm and acute health effects with an emphasis on 1–4 hr exposure duration, although longer-term studies as well as intermittent, multi-day studies with exposures generally up to 6–8 hr/day were included. The intermittent, multi-day studies were typically reproductive/developmental studies. There were two studies included in the database that determined concentrations that produced respiratory distress in 50% of the animals and two studies that evaluated cardiac sensitization in dogs that involved a 10-min exposure duration. The duration exposures of the original studies were not adjusted unless the primary investigator adjusted the duration because of the uncertainties involved with extrapolating exposure durations from longer exposure to shorter exposure durations (Jarabek, 1995). The database consisted mainly of studies in rodents and rabbits although a few studies used dogs and non-human primates. For the TOC approach, inhalation NOAELs for all 97 chemicals were included because of the need to include as many chemicals as possible into the different toxicity categories. However, for calculating the NOAEL-to-LC50 (N–L) ratio, intermittent, multi-day studies were excluded unless the original investigators noted effects after the first day of exposure. Data on 55 chemicals with NOAEL study exposure durations 624 hr were used to calculate N–L ratios. Studies that identified a lowest-observed-adverse-effect level (LOAEL) as well as a NOAEL were preferred. However, since inhalation exposure studies are more difficult to conduct than oral exposure, studies that identified only a LOAEL or NOAEL were included in the database in order to include as many chemicals as possible in the different toxicity categories. For four chemicals, some well-conducted studies only identified a NOAEL at the highest dose tested. For these four chemicals, the free standing NOAEL was conservatively chosen as the highest dose tested. In other instances, only LOAELs were identified and these studies were included if studies by other investigators supported the use of the LOAEL. In such cases, the LOAEL was converted to a NOAEL by the use of an UF, as described in the following section.
2.3. Adjustment of LOAELs to NOAELs 2. Methods 2.1. Acute inhalation reference database Ninety-seven organic and inorganic substances representing a range of industrial, environmental, and consumer chemicals were included in the acute inhalation reference database. Databases known to contain well-validated acute toxicological endpoints for a series of chemicals were consulted first since the quality of NOAELs and acute lethality studies would have been reviewed and, presumably, only acceptable studies would have been included in the toxicity assessments. These included acute reference exposure levels (RELs) from California Office of Environmental Health Hazard Assessment (OEHHA); acute (1–14 days) inhalation minimal risk levels (MRLs) from the Agency for Toxic Substances and Disease Registry (ATSDR); and Final and Interim Acute Exposure Guideline Levels (AEGLs). AEGLs are developed by the National Advisory Committee and reviewed by the Committee on Toxicology, a part of the National Research Council of the National Academies of Science (NRC, 2001). No studies on metallics, nerve gases, or non-structurally defined substances (such as gums, resins, oils, etc.) were included in the reference database, although inorganic compounds were included in the database. Munro et al. (1996) did not include inorganics in his study since the Cramer Structural Classes are for organic compounds.
2.2. Selection of the NOAEL based on animal studies and duration exposure adjustments For each of the 97 substances, the inhalation NOAEL with an endpoint that was relevant to humans was selected based on the most sensitive species, sex, and endpoint and was entered into the database. There was one chemi-
Typically, an UF of 10 was applied to the LOAEL to calculate the NOAEL. However, if the respiratory tract/irritancy health endpoint was considered to be mild as described in OEHHA (1999), an UF of 6 was applied to the LOAEL to calculate a NOAEL (Alexeeff et al., 2002).
2.4. Criteria for selection of acute lethality data Acute lethality data were used to assign chemicals to different acute GHS toxicity categories proposed by the United Nations (UN, 2005) (Table 1). LC50 data were also used to calculate N–L ratios. For many of the substances, more than one LC50 was identified from the literature, resulting from the fact that many substances were tested in more than one species and sex and/or at different exposure durations. This led, in some cases, to multiple LC50s or LD50s for individual substances. Typically, the lowest LC50 or LD50 data were used to categorize the chemicals and the lowest LC50 data were used to calculate the N–L ratios. Fig. 1 illustrates the steps that were followed to categorize chemicals into toxicity categories (TCs) and the criteria for selection of LC50 data to calculate the N–L ratios. First, LC50 data for all species 64 hr were obtained (Step 1). Values were adjusted to correspond to a 4-hr exposure duration because inhalation cut-off values for the GHS categories are based on 4-hr testing exposures (UN, 2005). Duration adjustments for LC50 data were made using Haber’s Law (Rinehart and Hatch, 1964) as modified by ten Berge et al. (1986). C n T ¼ K; where C is the exposure concentration, T is the exposure duration; K is a cumulative exposure constant, and ‘‘n’’ represents a chemicaland endpoint-specific exponent proposed by ten Berge et al. (1986). If a
c
d
a
Oral LD50 (mg/kg) LC50 Gas (ppmv) LC50 dust and mistsb mg/m3 LC50 vapoursc mg/m3 Tenth percentile composite TOC concentrationsd
b
>2,000a >5000 >5000 >20,000 1000 lg/m3 >300 to 62000a >2500 to 65000 >1000 to 65000 >10,000 to 620,000 125 lg/m3 >50 to 6300a >500 to 62500 >500 to 61000 >2000 to 610,000 125 lg/m3
Category 4 Category 3
Mild irritant
Category 2
Irritant 2A: Causes serious eye irritation 2B: Causes eye irritation >5 to 650a >100 to 6500 >50 to 6500 >500 to 62000 20 lg/m3
Category 1 most toxic
Corrosive Causes serious eye damage 0 to 65a 0 to 6100 0 to 650 0 to 6500 4 lg/m3 Skin irritation Eye irritation
Table 1 Globally Harmonized System of Classification and Labeling of Chemicals (GHS)
Oral LD50 data were not used to categorize a chemical for acute inhalation toxicity unless it was scientifically defensible to perform route-to-route extrapolation for that chemical. Liquids with a vapor pressure <0.1 mm Hg are treated as a mist. Solids are treated as dust. Liquids with a vapor pressure P0.1 mm Hg are treated as a vapor. Refer to Table 3.
R.L. Grant et al. / Regulatory Toxicology and Pharmacology 47 (2007) 261–273 Category 5 least toxic
264
chemical- and endpoint-specific value for ‘‘n’’ was available, it was used to adjust the LC50 data. If not, a default value of n = 1 was used. If all extrapolated values produced the same TC, no further action was required. If the extrapolated values produced different TCs, then the lowest TC was chosen. The quality of the experimental study, physical/chemical characteristics of the chemical, and other data such as eye/skin irritation, etc. were also used to decide the chemical’s TC. If LC50 data 64 hr were not available, then LC50 data >4 hr but 612 hr were obtained (Step 2). Duration adjustments were not performed on LC50 data >4 hr because of the uncertainties involved with extrapolating exposure durations from longer exposure to shorter exposure durations (Jarabek, 1995). If all values produced the same TC, no further action was required. If the extrapolated values produced different TCs, then the lowest TC was chosen. The quality of the experimental study, physical/chemical characteristics of the chemical, and other data such as eye/skin irritation, etc. were also used to decide the chemical’s TC. If LC50 data 612 hr were not available, then all other pertinent inhalation lethality data (i.e., LClo, LC33, etc.) were used to categorize a chemical (Step 3). This is generally a conservative approach because these values are lower than LC50 data. If inhalation lethality data were not available, then oral LD50 data were used to categorize a chemical using the GHS categorization scheme (step 4), but only if strict criteria for performing route-to-route extrapolation were met as discussed in USEPA (1994). Ethylene glycol was the only chemical included in the database based on LD50 data. Acute dermal LD50 data were not used to categorize chemicals because this data did not correctly categorized the chemicals for the evaluation of acute inhalation toxicity.
2.5. Assigning chemicals to Cramer structural classes Chemicals were assigned to Cramer structural classes using a computer program, Windows Toxtree based on the decision tree of Cramer et al. (1978). The software was obtained from the European Chemicals Bureau website (http://ecb.jrc.it/QSAR/). Organic chemicals were classified in Cramer Class I (least toxic class), II (intermediate class), and III (most toxic class).
2.6. The acute inhalation NOAEL and LC50 database Appendices, which are available as supplementary materials available on the Regulatory Toxicology and Pharmacology website, contain the following data: Appendix A provides data on LC50 or LD50 for chemicals by GHS category; Appendix B provides information on NOAELs for chemicals by GHS category; and Appendix C data on N–L ratios. An Excel spreadsheet of electronic data is available from the authors upon request.
2.7. Data and statistical analysis S-Plus statistical software (version 6.2) was used to perform statistical analysis. Distribution data were evaluated to determine whether they were normally distributed using the Kolmogorov–Smirnov test. The data were considered to be normally distributed if p P 0.05. Differences among GHS categories were evaluated with a one-way analysis of variance with a Scheffe´’s post hoc test. If there were two groups, statistical differences were evaluated with a standard two-sample t test. An approach presented by Berthouex and Brown (2002) was used to calculate parametric percentile values. The 95% tolerance bounds for the percentiles were calculated using procedures in Hahn and Meeker (1991) and StInt Software for Statistical Intervals (1993 version) available from the following website: http:// www.public.iastate.edu/~wqmeeker/stint.html. The fifth and tenth percentiles of the cumulative percentage distribution of NOAEL in each GHS category were divided by an UF of 100 to account for uncertainties due to animal-to-human extrapolation and human variability in order to derive values protective of the general population. The fifth and tenth percentiles of the cumulative percentage distribution of N–L ratios were divided by an UF of 100 to calculate a composite factor N–L ratio.
R.L. Grant et al. / Regulatory Toxicology and Pharmacology 47 (2007) 261–273
265
Fig. 1. Procedure for placing chemicals into GHS toxicity categories and for selecting LC50 data.
2.8. Published acute toxicity values based on human and animal studies Published acute toxicity values developed by OEHHA and ATSDR (i.e., acute RELs and acute MRLs, respectively) based on exposure durations of 624 hr from human and animal studies were entered in the database in order to assess whether the proposed health-protective concentrations developed using either the TOC approach or the composite factor N–L ratio approach were health-protective. It should be kept in mind that the toxicity values published by OEHHA and ATSDR were developed for different exposure durations and risk assessment purposes as discussed by Woodall (2005). Acute RELs are no-effect levels used to evaluate exposures based on an averaging time of one hour, except for reproductive/developmental toxicants which have an averaging time based on the number of hrs/day of the multiday-intermittent study (OEHHA, 1999). For example, the averaging time for a study with an exposure duration of 6 hr/day, gestational day 6–15, would be 6 hr. Acute MRLs for non-cancer health effects are used to evaluate exposure durations of up to one day to 14 days.
3. Results 3.1. Derivation of TOC concentrations for different GHS toxicity categories There were 97 chemicals that were classified in GHS acute toxicity categories based on acute lethality data (Appendix A). LC50 data were available for 86 chemicals, other types of inhalation lethality data were available for 10 chemicals, and LD50 data for one chemical, ethylene glycol. Forty-three percent of the acute lethality studies were based on a 4-hr exposure duration; 13% were <1 hr values;
35% were 1–3 hr values; and 12% were 6–10 hr values. Based on the LC50 ranges in the GHS categories (Table 1), 33 chemicals were classified in Category 1, 18 in Category 2, 23 in Category 3, four in Category 4, and 19 in Category 5. Data for the NOAELs based on acute inhalation data for 97 chemicals is shown in Appendix B. The NOAEL data for Category 4 were combined with the data for Category 3 since there were only four chemicals in Category 4 and the range of NOAEL data for these four chemicals corresponded to the range of NOAEL data in Category 3 (Appendix B).The distribution of NOAELs was not normally distributed; however, the natural logarithm (ln) of NOAELs was normally distributed. Therefore, the ln NOAELs were used to calculate percentiles instead of NOAELs. The cumulative percentage distribution of the ln NOAELs for the entire database with calculated 5th, 10th, 20th, 50th, 90th, and 95th parametric percentiles and their corresponding 95% lower and upper tolerance bounds (LTB-UTB) are shown in Fig. 2. The calculated parametric percentiles adequately describe the data set and the variation in the parametric percentiles is small. Fig. 3 presents the cumulative distributions of ln NOAELs of chemicals separated into GHS toxicity categories, the calculated parametric percentiles, and their corresponding 95% LTB-UTB. The calculated percentiles adequately describe the data set and the variation in the percentiles is small. The four category distributions were found to differ significantly from every other category. Table 2 presents the ranges of NOAEL values, geometric
266
R.L. Grant et al. / Regulatory Toxicology and Pharmacology 47 (2007) 261–273
Fig. 2. Cumulative percentage distribution of ln NOAELs for 97 chemicals in the acute inhalation database with calculated parametric percentiles and 95% lower and upper tolerance bounds (LTB-UTB).
mean (GM), geometric mean standard deviation (GSD) and number of chemicals for the GHS categories. Table 3 shows the parametric fifth percentile, tenth percentile, and 50th percentile of the four different GHS category distributions and the overall database of 97 chemicals. The fifth and tenth percentile of inhalation NOAELs for each acute toxicity category was divided by an UF of 100 to calculate the fifth and tenth percentile composite TOC concentration for each GHS toxicity category and their corresponding 95% LTB-UTB (Table 3). The uncertainty in the calculated tenth percentile value was greater for those categories with fewer number of chemicals (i.e., Categories 2 and 5). The difference between the LTB (most conservative value) and the tenth percentile was a factor of two for Category 1 (33 chemicals); a factor of four for Category 2 (18 chemicals); a factor of three for Categories 3 and 4 combined (27 chemicals); and a factor of four for Category 5 (19 chemicals). 3.2. Comparison of the fifth and tenth percentile composite TOC concentrations to published acute toxicity values Table 4 shows published REL and MRL toxicity values derived by OEHHA and ATSDR, respectively, for exposure durations 624 hr. All published toxicity values were higher (i.e., less conservative) than the fifth percentile TOC concentration for each GHS category. All toxicity values were higher than the tenth percentile TOC concentrations except for trans-1,2-dichloroethylene (Table 4). The tenth percentile TOC concentration was less than a factor of two higher than the published toxicity factor even though the averaging time for the ATSDR MRL for trans1,2-dichloroethylene is for 8 hr. Since the TOC concentrations were designed to evaluate one-hour intermittent exposure, these comparisons indicate that the tenth percentile TOC concentrations developed for the GHS categories should be adequately health-protective whereas the fifth percentile TOC concentration may be overly conservative.
Fig. 3. Cumulative percentage distribution of ln NOAELs grouped into GHS Categories with calculated parametric percentiles and 95% lower and upper tolerance bounds (LTB-UTB). A, Toxicity Category 1 (TC 1); B, TC 2; C, TC 3-4; and D, TC 5.
R.L. Grant et al. / Regulatory Toxicology and Pharmacology 47 (2007) 261–273
267
Table 2 Summary statistics of NOAELs for each GHS acute toxicity category 3
Range (mg/m ) GM ± GSD (mg/m3) Number of compounds
Category 1
Category 2
Categories 3 and 4
Category 5
0.09–236 5.4 ± 7.4 33
0.28–381 31 ± 8.6 18
1–3020 172 ± 7.7 27
79–167,000 1300 ± 7.2 19
Table 3 Parametric percentiles of NOAELs and proposed TOC concentrations based on the fifth and tenth percentile for each GHS acute toxicity category
50th percentile Fifth percentile Fifth percentile composite TOC concentrationa (95% tolerance bounds) Tenth percentile Tenth percentile composite TOC concentrationa (95% tolerance bounds)
Overall database
Category 1
Category 2
Categories 3 and 4
Category 5
57 mg/m3 0.56 mg/m3 NAb
5.4 mg/m3 0.20 mg/m3 2 lg/m3 (1–4 lg/m3) 0.41 mg/m3 4 lg/m3 (2–8 lg/m3)
31 mg/m3 0.91 mg/m3 9 lg/m3 (2–30 lg/m3) 2.0 mg/m3 20 lg/m3 (5–50 lg/m3)
172 mg/m3 6.0 mg/m3 60 lg/m3 (20–140 lg/m3) 12.6 mg/m3 125 lg/m3 (40–270 lg/m3)
1300 mg/m3 51 mg/m3 510 lg/m3 (110–1300 lg/m3) 104 mg/m3 1000 lg/m3 (280–2400 lg/m3)
1.6 mg/m3 NAb
a The fifth or tenth percentiles were divided by an UF = 100 and converted from mg/m3 to lg/m3 to calculate the composite TOC concentrations and 95% tolerance bounds for each separate category. b NA, not applicable.
For comparison purposes, the tenth percentile composite TOC concentrations for each category are included in Table 1 in order for a comparison to be made to the ranges of LC50 data for the different GHS toxicity categories. 3.3. Determination of the N–L ratios for different chemicals There were 55 chemicals for which acute LC50 data and corresponding acute NOAEL data for exposure durations of 624 hrs were available (indicated in Appendices A, B, and C). LCLo or other type of acute lethality data were not included. Appendix C shows the calculated N–L ratios for the 55 chemicals. The N–L ratios were not normally distributed; however, the ln N–L ratios were normally distributed. Therefore, the ln N–L ratios were used to calculate percentiles instead of N–L ratios. The cumulative percentage distribution of the ln N–L ratios for the entire database with calculated 5th, 10th, 50th, 90th, and 95th parametric percentiles and their corresponding 95% tolerance bounds are shown in Fig. 4. The calculated parametric percentiles appear to adequately describe the data set although the lower 5th–20th percentiles correspond better than the 50th–95th percentiles. The ranges, GM, GSD, the fifth and tenth percentiles for the distribution of N–L ratios are shown in Table 5. Layton et al. (1987) and Venman and Flaga (1985) did not include inorganic chemicals in their databases when calculating N–L ratios whereas there were 40 organic chemicals and 15 inorganic chemicals included in the present database (Appendix C). There were no significant
differences between the distribution of ln N–L ratios of organic or inorganic chemicals (p < 0.05). Venman and Flaga (1985) only used NOAEL and LD50 data from the rat to calculate N–L ratios. Layton et al. (1987) calculated N–L ratios using rat NOAEL data and LD50 data and compared it to N–L ratios calculated using rat NOAEL data and LD50 data from small mammals. In the present acute inhalation database, there were 26 chemicals where the same species was used to determine both NOAEL and LC50 data: 15 chemicals based on rat data, nine chemicals based on mouse data, and two chemicals based on guinea pig data (Appendix C). There were 29 chemicals where different species were used for NOAEL or LC50 data (Appendix C). There were no significant differences between the distribution of ln N–L ratios of the same species or different species (p < 0.05) although the variation in the ratios was greater when different species were used to calculate N–L ratios. Table 5 shows the calculation of the composite factor. The fifth and tenth percentile values were divided by an UF of 100 to derive a composite factor. The fifth and tenth percentile composite factors of 4.6 · 105 and 8.3 · 105, respectively, were less than a factor of two different from the LTB (more conservative value) of 2.4 · 105 and 4.7 · 105, respectively. The fifth and tenth percentile composite factors can be multiplied by a 4-hr LC50 value or other appropriate acute inhalation lethality data based on steps outlined in Fig. 1 to calculate an interim health-protective air concentration. Scientific judgment should be used to select the most appropriate and defensible acute lethality data.
268
Table 4 Comparison of estimated air concentrations to published toxicity values CAS no.
Toxicity value (lg/m3)
Sourcea
Adjusted duration (original study duration)
GHS category
TOC 5th Percentile (lg/m3)
TOC 10th percentile (lg/m3)
N–L ratio 5th percentile (lg/m3)
N–L ratio 10th percentile (lg/m3)
Acetone (2-propanone) Acrolein Acrylic acid
67-64-1 107-02-8 79-10-7
62,000 7 6000
MRL MRL REL
5C 1C 3C
510 2 60
1000 4 125
3500 2 130
6300 4 230
Acrylonitrile Ammonia
107-13-1 7664-41-7
220 3200
MRL REL
2C 3C
9 60
20 125
33 34
60 61
Arsine Benzyl chloride Carbon monoxide Chlorine Chloropicrin Dichlorobenzene, 1,4-
7784-42-1 100-44-7 630-08-0 7782-50-5 76-06-02 106-46-7
160 240 23,000 210 29 12,000
REL REL REL REL REL MRL
1C 1C 3C 1C 1C 4C
2 2 60 2 2 60
4 4 125 4 4 125
2 10 98 4 3 230
3 17 180 8 5 420
Dichloroethylene, trans-1,2 Dioxane, 1,4 Epichlorohydrin Ethylene glycol monobutyl ether Fluorine Formaldehyde Hydrochloric acid Hydrogen cyanide Hydrogen fluoride Hydrogen selenide Hydrogen sulfide Isopropyl alcohol Malathion Methanol Methyl bromide Methyl chloroform Methyl ethyl ketone Methylene chloride MTBE (methyl tert-butyl ether) Nitric acid Nitrogen dioxide Ozone Perchloroethylene Phenol
156-60-5 123-91-1 106-89-8 111-76-2 7782-41-4 50-00-0 7647-01-0 74-90-8 7664-39-3 7783-07-5 7783-06-4 67-63-0 121-75-5 67-56-1 74-83-9 71-55-6 78-93-3 75-09-2 1634-04-4 7697-37-2 10102-44-0 10028-15-6 127-18-4 108-95-2
800 3000 1300 14,000 16 93 2100 340 240 5 42 3200 200 28,000 3900 68,000 13,000 14,000 7200 86 470 180 20,000 5800
MRL REL REL REL MRL REL REL REL REL REL REL REL MRL REL REL REL REL REL MRL REL REL REL REL REL
4 hr 1 hr 1 hr (6 hr/d on 20 occasions; 6 hr adj. to 1 hr; ‘‘n’’ = 2) 24 hr (8 hr adj. to 24 hr) 1 hr (5 m to 120; m; all adj. to 60 m; ‘‘n’’ = 4.6) 1 hr 1 hr (2 hr adj. to 1 hr; ‘‘n’’ = 2) 1 hr 1 hr (30 m adj. to 1 hr; ‘‘n’’ = 1) 1 hr (10 m adj. to 1 hr; ‘‘n’’ = l) Not specified; occupational exposures 8 hr 1 hr (6 hr no adj.) 1 hr 1 hr (2 or 4 hr adj. to 1 hr) 24 hr (15 m adj. to 24 hr) 1 hr (3 hr adj. to 1 hr, ‘‘n’’ = 2) 1 hr (45 m adj. to 1 hr; ‘‘n’’ = 1) 1 hr (30 m adj. to 1 hr; ‘‘n’’ = 1) 1 hr 1 hr N/A(Mean of 1 hr data) 1 hr (4 m adj. to 1 hr) 6 hr 1 hr (75 m adj. to 1 hr) 1 hr (2 hr adj. to 1 hr) 1 hr (30 m adj. to 1 hr) 1 hr (2 hr adj. to 1 hr) 1 hr (90 m adj. to 1 hr) 24 hr (6 hr adj. to 24 hr) 1 hr (40 m adj. to 1 hr) 1 hr 1 hr 1 hr (3 hr adj. to 1 hr) 1 hr (8 hr adj. to 1 hr)
5C 5C 2C 3C 1C 2C 2C 1C 1C 1C 2C 5C 1C 5C 2C 5C 5C 3C 5C 1C 1C 1C 5C 2C
510 510 9 60 2 9 9 2 2 2 9 510 2 510 9 510 510 60 510 2 2 2 510 9
1000b 1000 20 125 4 20 20 4 4 4 20 1000 4 1000 20 1000 1000 125 1000 4 4 4 1000 20
4000b 2400 69 100 2 16 13 3 3 0.27 28 1800 2 1100 35 1200 1600 360 5500 5 5 1 1200 8
7100b 4300b 120 180 4 29 23 5 6 0.49 51b 3300b 4 2000 64 2100 2800 650 10,000b 8 9 3 2200 15
R.L. Grant et al. / Regulatory Toxicology and Pharmacology 47 (2007) 261–273
Chemical name
Abbreviations: adj., adjusted; C, category; d, day; GHS, Globally Harmonized System of Classification and Labeling of Chemicals (UN, 2005); hr, hour; m, min; N–L ratio, NOAEL to LC50 ratio; TOC, threshold of concern. a Source: MRL, ATSDR acute Minimal Risk Level; REL, OEHHA acute reference exposure levels. b Estimated concentrations that are greater than published values are highlighted in bold.
Phosgene Phosphorus, white Propylene oxide Styrene Sulfur dioxide Sulfuric acid Toluene Triethylamine Vinyl chloride Xylenes
75-44-5 7723-14-0 75-56-9 100-42-5 7446-09-5 7664-93-9 108-88-3 121-44-8 75-01-4 1330-20-7
4 20 3100 21,000 660 120 37,000 2800 180,000 22,000
REL MRL REL REL REL REL REL REL REL REL
1 hr 24 hr (5 m adj. to 24 hr) 1 hr (4 hr adj. to 1 hr) 1 hr 1 hr (varied 5–75 m) 1 hr (16 m adj. to 1 hr) 1 hr (6 hr adj. to 1 hr) 1 hr (8 hr adj. to 1 hr) 1 hr (7.5 hr adj. to 1 hr) 1 hr (30 m adj. to 1 hr)
1C 2C 3C 4C 2C 2C 5C 3C 5C 5C
2 9 60 60 9 9 510 60 510 510
4 20 125 125 20 20 1000 125 1000 1000
0.12 13 120 490 45 16 1200 190 6900 950
0.21 23b 220 880 82 29 2100 350 12,000 1700
R.L. Grant et al. / Regulatory Toxicology and Pharmacology 47 (2007) 261–273
269
Fig. 4. Cumulative percentage distribution of ln NOAEL-to-LC50 ratios for 55 chemicals in the acute inhalation database with calculated parametric percentiles and 95% lower and upper tolerance bounds (LTB-UTB).
3.4. Comparison of adjusted LC50 data based on the fifth and tenth percentile composite factor N–L ratio to published acute toxicity values Estimated threshold concentrations derived using the composite factor N–L ratio were compared to published acute toxicity values derived by OEHHA and ATSDR (Table 4). All published toxicity values were higher than the estimated threshold concentrations based on the fifth percentile composite factors except for trans-1,2-dichloroethylene based on an 8-hr exposure duration (Table 4). All published toxicity values were higher than the estimated threshold concentrations based on the tenth percentile composite factor except for six chemicals (Table 4). Three of the six chemicals were representative of exposure durations P8 hr: trans-1,2-dichloroethylene (8-hr exposure duration MRL); MTBE (24-hr exposure duration MRL); and white phosphorus (24-hr exposure duration MRL). The other three chemicals had exposure durations of one hour, but the difference between the estimated threshold concentrations were less than a factor of two: 1,4-dioxane (one-hour exposure duration REL); hydrogen sulfide (one-hour exposure duration REL); and isopropyl alcohol (one-hour exposure duration REL). Since the threshold concentrations based on the composite factor N–L ratio were designed to evaluate one-hour intermittent exposure, these comparisons indicate that use of the tenth percentile composite factor N–L ratio should be health-protective whereas the use of the fifth percentile composite factor N–L ratio may be overly protective. 3.5. Comparison of health-protective air concentrations estimated with the tenth percentile TOC approach and the tenth percentile composite factor N–L ratio approach For the tenth percentile composite factor, the estimated air concentrations from these two separate approaches
270
R.L. Grant et al. / Regulatory Toxicology and Pharmacology 47 (2007) 261–273
Table 5 Summary statistics for N–L ratios and proposed composite factor
Entire database Composite factora 95% Tolerance bounds
Number
Range
GM ± GSD
Fifth percentile
Tenth percentile
55
0.0009–0.83
0.068 ± 5.2
0.0046 4.6 · 105 2.4 · 105 to 7.5 · 105
0.0083 8.3 · 105 4.7 · 105 to 1.3 · 104
a The fifth or tenth percentile ratio for the entire database was divided by an uncertainty factor of 100 to account for animal-to-human uncertainty and human variability to produce the composite factor N–L ratio.
were similar (Table 4). For approximately 70% of the chemicals, the differences between the estimated threshold concentrations based on the two different approaches were within a factor of three. This is a reflection of the fact that both approaches were developed using similar acute inhalation toxicity data for the chemicals in both databases. The approach based on N–L ratios produced estimated concentrations that were generally higher than the TOC approach (i.e., less conservative). This may be due to different numbers of chemicals and grouping of chemicals in the databases. While there were 55 chemicals in the N–L ratio database and only one group, the TOC approach using the GHS categorization system (UN, 2005) had 97 chemicals and four separate groups with smaller number of chemicals in each group (Table 2). In addition, the TOC approach included animal studies with exposure durations P24 hr and intermittent, multi-day studies in order to include as many chemicals in the different categories as possible. This illustrates a possible advantage of the N–L ratio approach when compared to the TOC approach. A larger number of chemicals are needed to develop TOC concentrations for several different toxicity potency classes. 3.6. Target organ or critical effect and exposure durations of the NOAEL inhalation studies Table 6 provides a summary and Appendix B provides detailed information on the target organ/critical effects of the NOAEL inhalation studies in the database of 97 chemicals. The database contained a variety of different endpoints, although 48% of the inhalation studies produced POE effects such as irritation, respiratory tract, or dermal/ocular effects. Interestingly, the great majority of the chemicals that produced POE effects occurred in Catego-
Table 6 Summary of critical effects of acute toxicity studies used as the basis for NOAELs Critical effect
Number of compounds
Respiratory tract Irritation Dermal/ocular/mixed Respiratory and Systemic Neurological Developmental Hematological Hepatic or renal Cardiac
25 18 3 3 17 17 4 7 3
ries 1 and 2: 76% of chemicals in Category 1 produced POE effects; 78% in Category 2; 22% in Categories 3 and 4 combined; and 11% in Category 5. The exposure durations of the NOAEL animal studies for the 97 chemicals varied (Appendix B): 35% of studies were 61 hr; 24% of studies were >1 hr to 66 hr; and 7% of studies were >6 hr to 624 hr. Thirty-four percent of the studies were intermittent multi-day studies, mainly representing developmental studies. For the subset of 55 chemicals used for the N–L ratio approach (indicated in Appendix B), 52% of NOAEL studies were 61 hr; 26% of studies were >1 hr to 66 hr; and 22% of studies were >6 hr to 624 hr. 3.7. Distributions of NOAELs based on Cramer Structural Classes Appendix B shows the Cramer Structural Classes of the chemicals in the database. A total of 76 organic chemicals were classified using Cramer structural classes. There were 17 organic chemicals in Class I, four in Class II, and 55 in Class III. The cumulative distributions of NOAELs for Cramer Structural Classes I and III did not differ significantly from each other (data not shown). There were not enough chemicals in Cramer Class II to include in the analysis. The Cramer classification scheme developed for chronic oral exposure does not adequately define acute inhalation toxicity potency, possibly due to the fact that many of the chemicals included in the acute inhalation database produced POE effects (Table 6). 4. Discussion 4.1. Uncertainty and variability of LC50 and NOAEL data Both LC50 and NOAEL data are subject to uncertainty and variability due to inter- and intraspecies differences; differences in test protocols, laboratory practices, and measurement technologies; variation according to the age, weight, sex, and health of the test animals; and diet, housing conditions, and ambient temperature (Layton et al., 1987). Therefore, the LC50 and the NOAEL are not biological constants. The NOAEL is the highest exposure level at which there are no biologically or statistically significant increases in the frequency or severity of adverse effect between the exposed population and its appropriate control. The NOAEL approach has been criticized because it does not
R.L. Grant et al. / Regulatory Toxicology and Pharmacology 47 (2007) 261–273
use the full dataset of the dose–response curve and is dependent on spacing between doses. For studies with greater variation in endpoint measurement or smaller sample size, the NOAEL approach tends to result in a higher NOAEL being determined based on statistical comparison to the controls. In addition, some experimental studies only identify a LOAEL which increases the uncertainty in the dose–response assessment. Alternately, a mathematical model can be used to fit the entire dose–response data for a chemical with a non-linear mode of action so that the threshold concentration corresponding to an estimated incidence or change in level of response can be determined (i.e., benchmark concentration as well as the lower confidence limit on the benchmark concentration from a dose–response model). The advantages of benchmark concentration modeling are it uses the full dataset of the dose– response curve, accounts for the greater uncertainty due to smaller sample size or greater variation, and can estimate the threshold concentration when a NOAEL can not be established. Benchmark concentration modeling is preferred over the NOAEL approach although not all dose– response data can be adequately modeled. Travis et al. (2005) discusses the advantages and disadvantages of both the benchmark concentration modeling and NOAEL approach. When gathering data for the acute inhalation database, there was only one chemical, methanol, where data obtained from benchmark concentration modeling was available. In the future, more researchers will conduct dose–response assessments using benchmark concentration modeling and the uncertainty in NOAEL data will be decreased. The LC50 is calculated from a statistical analysis of the fractions of animals dying after being exposed to different concentrations of a chemical. Layton et al. (1987) provides a discussion of three different interlaboratory studies that determined variation of LD50 data. They concluded that ‘‘the actual variation in LD50 values for a given species was surprisingly low, given the various sources of uncertainty associated with lethal toxicities’’ (Layton et al., 1987). However, a similar interlaboratory study on LC50 data has not been performed so the uncertainty in LC50 data due to interlaboratory variation cannot be determined. Because of the uncertainty and variability inherent in determining an LC50 or NOAEL, a statistical method was used to determine the cumulative percentage distribution of a range of acute inhalation NOAELs in each toxicity category or of a range of N–L ratios. The tenth percentile of the distributions was calculated and a total UF of 100 was used to calculate TOC concentrations or a composite factor N–L ratio. These are conservative procedures. There was greater uncertainty in the TOC concentrations for Category 2 and Category 5 when compared to Category 1 and Categories 3–4 because there were fewer number of chemicals in the categories. Conservative measures were used to compile the database, as discussed in the following section. In addition, a
271
comparison of published toxicity values available from OEHHA and ATSDR based on human and animal studies indicate the tenth percentile composite value derived for both methods is conservative to evaluate one-hour exposure. There is less than a 10% probability that toxicological studies of any other substance not included in the database would result in a N–L ratio or TOC concentration lower than the tenth percentile. If human exposure to a chemical is below the estimated health-protective air concentration, the substance can be judged, with reasonable confidence, to present a low probability of risk, even if there is limited toxicity information. 4.2. Conservative measures used to compile the database Ideally, the database of NOAELs used to develop a TOC approach or N–L ratio approach for one-hour exposure durations would contain only studies of one-hour exposure durations. In order to include as many chemicals in the database as possible and to ensure only high-quality studies were included in the database, studies based on exposure durations longer than one-hour were included which is a conservative procedure. As mentioned previously, the exposure durations of NOAEL data were not adjusted to a one-hour exposure duration because of the uncertainties involved with extrapolating exposure durations from longer exposure to shorter exposure durations (Jarabek, 1995). For four chemicals, free-standing NOAELs were included, which is conservative since an actual NOAEL could be much higher. The TOC approach used studies of longer exposure durations such as multi-day intermittent exposure whereas the N–L ratio approach used NOAELs from exposure durations 624 hr. As a consequence, the air concentrations determined using the TOC approach were generally more conservative than air concentrations derived using the N–L ratio approach. The N–L ratio approach has the advantage of having 55 chemicals in one category as opposed to the TOC approach where there were essentially four categories with fewer chemicals in each category. 4.3. Health-protective air concentrations based on the TOC approach The basic approach that has previously been used to establish a chronic oral TOC dose (Munro et al., 1996) has been used to establish health-protective human exposure threshold concentrations for the evaluation of one-hour intermittent inhalation exposures. The Cramer structural class system (Cramer et al., 1978) was not used since it did not correctly predict acute inhalation potency although a structural class system would be preferred over the use of acute inhalation data since the only information needed to categorize a chemical is its structure. Acute inhalation lethality data is not available for all chemicals. However, if chemicals are grouped according to acute inhalation toxicity potency as defined by GHS (UN,
272
R.L. Grant et al. / Regulatory Toxicology and Pharmacology 47 (2007) 261–273
2005), a distribution of NOAELs that are statistically different from each other can be generated for different categories. When appropriate UFs are applied to the category-specific tenth percentile of the distribution of NOAELs, health-protective air concentrations are derived for each category. If the only toxicity information for a chemical is LC50 data, then the chemical can be placed in an appropriate GHS toxicity category and the categoryspecific TOC concentration can be used as an interim health-protective air concentration. 4.4. Health-protective air concentrations based on the composite factor N–L ratio approach The basic approach that has previously been used by Layton et al. (1987) and Venman and Flaga (1985) to estimate a chronic oral benchmark has been used to estimate healthprotective human exposure threshold concentrations for the evaluation of one-hour intermittent inhalation exposures. An estimated air concentration calculated by multiplying the tenth percentile composite factor of 8.3 · 105 by 4-hr LC50 data or other appropriate acute lethality data (Fig. 1) can be used as an interim health-protective air concentration for a chemical that has limited toxicity information. Venman and Flaga (1985) developed a chronic oral N–L ratio using NOAEL and LD50 data from only one species, the rat, and suggested their acute-to-chronic application factor be applied to oral rat LD50 data only. Layton et al. (1987) calculated a chronic oral composite N–L ratio of 5 · 106 when both NOAEL and LD50 data were obtained from the rat and another chronic oral composite N–L ratio of 1 · 105 when NOAEL data were obtained from the rat and LD50 data were obtained from small mammals. Layton et al. (1987) suggested that values from 5 · 106 to 1 · 105 per day be used to establish interim acceptable daily intakes based on scientific judgement. Since the rat is not the best model for toxic responses in humans nor is it the most sensitive species to any given toxic response, NOAEL and LC50 data from the most sensitive species were included in the acute inhalation database. Approximately half the ratios were obtained from the same species and half from different species. The inclusion of these different groups increased the variation within the tenth percentile composite N–L ratio but is based on a larger dataset of 55 chemicals and can be applied to LC50 data from various species based on criteria in Fig. 1 to calculate health-protective air concentrations. 4.5. The evaluation of one-hour intermittent inhalation exposure for chemicals with limited toxicity information Acute lethality data are used in both the tenth percentile category-specific TOC concentration approach and the tenth percentile composite factor N–L ratio approach so it is possible to determine health-protective air concentrations using both approaches. Since both approaches are conservative, the decision on which approach to use for a
chemical should be based on scientific judgment and an analysis of all available data including physical/chemical parameters, proposed mode of action, toxicity of structurally similar chemicals or chemical classes, data quality of the acute lethality study, etc. For example, if there is limited information on the quality of the study used to determine a LC50 value, using a range of LC50s as defined by the GHS system to categorize the chemical and adopting a TOC concentration for the category may be more defensible than using the N–L ratio approach. In the future, acute inhalation lethality may be evaluated using the fixed dose procedure, the acute toxic class method or the up-anddown procedure (Lipnick et al., 1995; Botham, 2004) instead of determining a LC50 value. In that case, a TOC approach may be preferred. Both methods should be regarded as useful tools to predict health-protective air concentrations for a chemical with limited toxicity information in conjunction with a weight-of-evidence approach on the toxicity assessment of the chemical. 4.6. Conclusion If a chemical has limited acute inhalation toxicity information, then the use of the TOC approach or composite factor N–L ratio approach can be used on an interim basis until additional acute toxicity information becomes available for a chemical. Air concentrations derived based on the TOC or N–L ratio approach will be used as generic ESLs in Texas for chemicals that have limited toxicity information in order to evaluate short-term one-hour exposures. The generic ESLs will be used in the TCEQ air permitting process to assess the protectiveness of substance-specific emission rate limits for facilities undergoing air permit reviews. If the GLCmax, a worst case concentration resulting from a worst case emission rate, is below the generic ESL for that chemical, the substance can be judged, with reasonable confidence, to present a low probability of risk. Acknowledgments The authors gratefully acknowledge the significant contribution of Dr. Valerie Meyers who classified the chemicals in the database into Cramer structural classes and Mike Aplin who reviewed the manuscript. Appendices A, B, and C. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.yrtph.2006. 11.003. References Alexeeff, G.V., Broadwin, R., Liaw, J., Dawson, S.V., 2002. Characterization of the LOAEL-to-NOAEL uncertainty factor for mild adverse effects from acute inhalation exposures. Regul. Toxicol. Pharmacol. 36, 96–105.
R.L. Grant et al. / Regulatory Toxicology and Pharmacology 47 (2007) 261–273 Berthouex, P.M., Brown, L.C., 2002. Estimating percentiles. In: Berthouex, P.M., Brown, L.C. (Eds.), Statistics for Environmental Engineers, second ed. Lewis Publishers, Washington, D.C., pp. 71–75. Blackburn, K., Stickney, J.A., Carlson-Lynch, H.L., McGinnis, P.M., Chappell, L., Felter, S.P., 2005. Application of the threshold of toxicological concern approach to ingredients in personal and household care products. Regul. Toxicol. Pharmacol. 43, 249–259. Botham, P.A., 2004. Acute systemic toxicity—prospects for tiered testing strategies. Toxicol. In Vitro 18, 227–230. Craig, D.K., Davis, J.S., Hansen, D.J., et al., 2000. Derivation of temporary emergency exposure limits (TEELs). J. Appl. Toxicol. 20, 11–20. Cramer, G.M., Ford, R.A., Hall, R.L., 1978. Estimation of toxic hazard— a decision tree approach. Food Cosmet. Toxicol. 16, 255–276. Dolan, D.G., Naumann, B.D., Sargent, E.V., Maier, A., Dourson, M., 2005. Application of the threshold of toxicological concern concept to pharmaceutical manufacturing operations. Regul. Toxicol. Pharmacol. 43, 1–9. Fay, M. Luukinen, B., Holler, J., 2005. Extrapolation of LC50s to safe exposure levels—insights for emergency responders. In: Society of Toxicology 2005 Annual Meeting, Toxicologist 44, 302 (Abstract #1481). FDA (Food and Drug Administration). 1995. Food additives; threshold of regulation for substances used in food-contact articles. Federal Register 60 (136) 36582–36596. Ford, R., Davis, R., Gardner, D., 2006. Development of inhalation thresholds of concern for the evaluation of tobacco additives. In: Society of Toxicology 2006 Annual Meeting, Toxicologist 45, 497 (Abst. #2317). Hahn, G.J., Meeker, W.Q., 1991. Methods for calculating statistical intervals for a normal distribution. In: Hahn, G.J., Meeker, W.Q. (Eds.), Statistical Intervals A Guide for Practitioners. John Wiley & Sons, Inc., Canada, pp. 53–74. Jarabek, A.M., 1995. The application of dosimetry models to identify key processes and parameters for default dose–response assessment approaches. Toxicol. Lett. 79, 171–184. JECFA (Joint FAO/WHO Expert Committee on Food Additives). 1993. Evaluation of certain food additives and contaminants (Fortieth report of the Joint FAO/WHO Expert Committee on Food Additives). WHO Technical Report Series, No. 832. JECFA (Joint FAO/WHO Expert Committee on Food Additives). 1995. Evaluation of certain food additives and contaminants (Forty-third report of the Joint FAO/WHO Expert Committee on Food Additives). WHO Technical Report Series, No. 855. JECFA (Joint FAO/WHO Expert Committee on Food Additives). 1997. Evaluation of certain food additives and contaminants (Forty-sixth report of the Joint FAO/WHO Expert Committee on Food Additives). WHO Technical Report Series, No. 868. Kroes, R., Renwick, A.G., Cheeseman, M., Kleiner, J., Mangelsdorf, I., Piersma, A., Schilter, B., Schlatter, J., van Schothorst, F., Vos, J.G., Wurtzen, G.European branch of the International Life Sciences Institute, 2004. Structure-based thresholds of toxicological concern (TTC): guidance for application to substances present at low levels in the diet. Food Chem. Toxicol. 42, 65–83.
273
Kroes, R., Galli, C., Munro, I., Schilter, B., Tran, L., Walker, R., Wurtzen, G., 2000. Threshold of toxicological concern for chemical substances present in the diet: a practical tool for assessing the need for toxicity testing. Food Chem. Toxicol. 38, 255–312. Layton, D.W., Mallon, B.J., Rosenblatt, D.H., Small, M.J., 1987. Deriving allowable daily intakes for systemic toxicants lacking chronic toxicity data. Regul. Toxicol. Pharmacol. 7, 96–112. Lipnick, R.L., Cotruvo, J.A., Hill, R.N., Bruce, R.D., Stitzel, K.A., Walker, A.P., Chu, I., Goddard, M., Segal, L., Springer, J.A., Myers, R.C., 1995. Comparison of the up-and-down, conventional LD50, and fixed-dose acute toxicity procedures. Food Chem. Toxicol. 33, 223–231. Munro, I.C., Ford, R.A., Kennepohl, E., Sprenger, J.G., 1996. Correlation of structural class with no-observed-effect levels: a proposal for establishing a threshold of concern. Food Chem. Toxicol. 34 (9), 829– 867. Munro, I.C., 1990. Issues to be considered in the safety evaluation of fat substitutes. Food Chem. Toxicol. 28, 751–753. NIOSH (National Institute for Occupational Safety and Health). 2005. Documentation for Immediately Dangerous to Life or Health Concentrations (IDLHs). US Department of Health and Human Services, Centers for Disease Control and Prevention, National Institute for Occupational Safety and Health, Cincinnati, OH. Available at:
Evaluation of acute inhalation toxicity for chemicals with limited toxicity information Roberta L. Grant *, Bernard J. Kadlubar, Neeraja K. Erraguntla, Michael Honeycutt Texas Commission on Environmental Quality, P.O. Box 13087, MC168, Austin, TX 78711-3087, USA Received 19 July 2006 Available online 31 January 2007
Abstract A large reference database consisting of acute inhalation no-observed-adverse-effect levels (NOAELs) and acute lethality data for 97 chemicals was compiled to investigate two methods to derive health-protective concentrations for chemicals with limited toxicity data for the evaluation of one-hour intermittent inhalation exposure. One method is to determine threshold of concern (TOC) concentrations for acute toxicity potency categories and the other is to determine NOAEL-to-LC50 ratios. In the TOC approach, 97 chemicals were classified based on the Globally Harmonized System of Classification and Labeling of Chemicals proposed by the United Nations into different acute toxicity categories (from most toxic to least toxic): Category 1, Category 2, Category 3, Category 4, and Category 5. The tenth percentile of the cumulative percentage distribution of NOAELs in each category was determined and divided by an uncertainty factor of 100 to derive the following health-protective TOC concentrations: 4 lg/m3 for chemicals classified in Category 1; 20 lg/m3 for Category 2; 125 lg/m3 for both Categories 3 and 4; and 1000 lg/m3 for Category 5. For the NOAEL-to-LC50 ratio approach, 55 chemicals with NOAEL exposure durations 624 hour were used to calculate NOAEL-to-LC50 ratios. The tenth percentile of the cumulative percentage distribution of the ratios was calculated and divided by an uncertainty factor of 100 to produce a composite factor equal to 8.3 · 105. For a chemical with limited toxicity information, this composite factor is multiplied by a 4-hour LC50 value or other appropriate acute lethality data. Both approaches can be used to produce an estimate of a conservative threshold air concentration below which no appreciable risk to the general population would be expected to occur after a one-hour intermittent exposure. 2006 Elsevier Inc. All rights reserved. Keywords: Acute; Inhalation; Threshold of concern; Globally Harmonized System; LC50; Cramer; Air permit review; NOAEL
1. Introduction The Texas Clean Air Act (Chapter 382 of the Texas Health and Safety Code) authorizes the Texas Commission on Environmental Quality (TCEQ)1 to conduct air permit reviews of all new and modified facilities to ensure that *
the operation of a proposed facility will not cause or contribute to a condition of air pollution. Air permit reviews typically involve evaluations of best-available-control technology and predicted air concentrations related to proposed emissions from the new or modified facility. In a conservative evaluation, worst-case emission rates are
Corresponding author. Fax: +1 512 239 1794. E-mail address: [email protected] (R.L. Grant). 1 Abbreviations used: AEGL, Acute Exposure Guideline Level; ATSDR, Agency for Toxic Substances and Disease Registry; ESLs, effects screening levels; GHS, Globally Harmonized System of Classification and Labeling of Chemicals; GLCmax, maximum ground level air concentrations; GM, geometric mean; GSD, geometric mean standard deviation; hr, hour; LC50 or LD50, the concentration or dose of a chemical which causes the death of 50% of a group of test animals; ln, natural logarithm; LOAEL, lowest-observed-adverse-effect level; LTB, lower tolerance bound; MRL, minimal risk levels; NOAEL, no-observed-adverse-effect level; N–L ratio, NOAEL-to-LC50 ratio; NOEL, no-observed-effect level; OEHHA, California Office of Environmental Health Hazard Assessment; POE, point of entry; REL, reference exposure level; TC, toxicity category; TCEQ, Texas Commission on Environmental Quality; TOC, threshold of concern; UF, uncertainty factor; UTB, upper tolerance bound. 0273-2300/$ - see front matter 2006 Elsevier Inc. All rights reserved. doi:10.1016/j.yrtph.2006.11.003
262
R.L. Grant et al. / Regulatory Toxicology and Pharmacology 47 (2007) 261–273
modeled to predict resulting short-term and long-term chemical-specific maximum ground level air concentrations (GLCmax). In the review of proposed emissions, federal/ state standards and chemical-specific effects screening levels (ESLs) developed by TCEQ toxicology staff are used for criteria and non-criteria pollutants, respectively. ESLs are chemical-specific air concentrations set to protect human health and welfare and are used in the air permitting process to assess the protectiveness of emission rate limits. Short-term ESLs are developed to evaluate acute intermittent exposures of one hour (hr) whereas long-term ESLs are developed to evaluate chronic exposures. The Texas Health and Safety Code is comprehensive. Therefore, ESLs are developed for as many air contaminants as possible, even for chemicals with limited toxicity data. There is a need to develop conservative, health-protective concentrations for chemicals with limited toxicity data for the review of short-term GLCmaxs taking into consideration that ambient air exposure is dependent on meteorological conditions and peak exposures that could occur several times per day. The purpose of this paper is to present two different approaches to establish health-protective air concentrations for chemicals with limited toxicity information for the general public via the inhalation route of exposure. This paper focuses on one-hour intermittent exposure scenarios that are normally the basis for air permit reviews. A method proposed by Munro et al. (1996) for establishing a threshold of concern (TOC) via oral exposure for chemicals in different toxicity potency classes will be discussed followed by an adaptation of a procedure proposed by Layton et al. (1987) and Venman and Flaga (1985) that involved the calculation of the ratios of no-observed-adverse-effect levels (NOAELs) to oral LD50 data for different chemicals. The LD50 is the dose of a chemical which causes the death of 50% of a group of test animals. Health-protective air concentrations derived using these two approaches are presented and compared to published short-term toxicity values. 1.1. A threshold of concern approach Munro et al. (1996) proposed a method for establishing a threshold of concern (TOC) based on grouping chemicals into three structural classes that correlated with chronic oral toxicity potency. Using a system devised by Cramer et al. (1978), organic chemicals were classified in Cramer Class I (least toxic class), II (intermediate class), and III (most toxic class). Munro obtained no-observed-effect levels (NOELs) from chronic, oral animal studies for over 600 chemicals and calculated the fifth percentile of NOELs for each structural class. The fifth percentile NOEL for each structural class was divided by an uncertainty factor (UF) of 100 to account for animal-to-human uncertainty and human variability to establish a conservative threshold dose below which no appreciable risk to human health would occur. The threshold dose could be used for chemicals with limited toxicity information to determine if toxic-
ity testing should be conducted if potential exposure levels were higher than the conservative threshold dose. The approach assumes that the proposed threshold dose for a chemical with limited toxicity information will not be significantly lower than the fifth percentile NOEL divided by an UF of 100. The TOC approach was initially used for food additives and for the evaluation of packaging material (FDA, 1995; Kroes et al., 2000, 2004; Munro, 1990; Munro et al., 1996). Other investigators have used this approach for other products such as food flavorings, personal and household care products and pharmaceutical compounds (JECFA, 1993, 1995, 1997; Kroes et al., 2004; Blackburn et al., 2005; and Dolan et al., 2005). The TOC approach has mainly been used to evaluate chronic oral exposure. However, the basic approach of Munro et al. (1996) could be used to develop a TOC for other routes of exposure such as inhalation exposure. Ford et al. (2006) investigated the inhalation TOC approach for the evaluation of tobacco additives (i.e., low molecular weight volatile compounds) where inhalation is the relevant route of exposure. They collected chronic inhalation NOAELs for approximately 350 chemicals categorized into the three Cramer structural classes. There was a good correlation with the oral database for those chemicals not causing respiratory tract point-of-entry (POE) effects but there was not a good correlation with the oral database for approximately 33% of chemicals that caused respiratory POE effects. Ford et al. (2006) concluded that the use of the chronic inhalation database, with the removal of those substances with NOAELs based on respiratory POE effects, allowed the determination of inhalation TOCs. Although the Cramer structural class categories were useful for establishing a TOC for the chronic inhalation route, a method to predict respiratory POE effects for chronic inhalation exposure must be developed and used in conjunction with the Cramer structural categories. The basic approach of Munro et al. (1996) was followed to establish a TOC approach for the review of short-term GLCmaxs. However, the basic approach was altered in that the acute inhalation database included NOAEL data, not NOEL data, and the chemicals were classified using acute lethality data and the Globally Harmonized System of Classification and Labeling of Chemicals (GHS) proposed by the United Nations (UN, 2005). The GHS classification system is designed to predict acute toxicity potency and is more appropriate for developing an acute TOC approach. 1.2. A NOAEL-to-LC50 ratio approach One of the first steps in determining toxic inhalation potency is to determine the concentration of a chemical in air which causes the death of 50% (one half) of a group of test animals (LC50). Acute lethality data have been used to determine concentrations that are predictive of an increasing likelihood of lethality for the general public (i.e., acute emergency exposure guideline levels (NRC, 2001) and temporary emergency exposure limits (Craig
R.L. Grant et al. / Regulatory Toxicology and Pharmacology 47 (2007) 261–273
et al., 2000)) and to establish short-term occupational limits such as immediately dangerous to life and health levels (NIOSH, 2005). Fay et al. (2005) investigated the use of LC50 data to extrapolate to safe exposure levels for emergency responders for acute hazardous chemical releases. In addition, Suda et al. (1999) investigated the linear relationship between animal LC50 data and occupational threshold limit values (time-weighted average) as well as short-term exposure limits and provided correlation equations that could be used to calculate threshold limit values from LC50 data. Layton et al. (1987) used oral LD50 data for estimating acceptable daily intakes for the evaluation of exposures to contaminants at hazardous waste sites. Venman and Flaga (1985) also proposed the use of LD50 data to establish provisional acceptable daily intakes for the evaluation of waste water contaminants. Both investigators calculated the ratio of NOAELs from chronic animal studies to oral LD50 data for different chemicals and determined the fifth percentile of the cumulative distributions of the ratios. The LD50 value for contaminants with limited toxicity data was multiplied by the fifth percentile ratio to derive a surrogate NOAEL. The surrogate NOAEL was divided by an UF of 100 in order to establish a conservative threshold dose below which no appreciable risk to human health would be expected to occur. We have used the basic approach of Layton et al. (1987) and Venman and Flaga (1985) for evaluating chronic oral toxicity and applied it to acute inhalation toxicity. Therefore, ratios of NOAELs from acute inhalation studies to LC50 data were calculated.
263
cal, methanol, where a calculated 95% lower confidence limit on the benchmark dose at the 5% response level was used as the NOAEL. The study types included those typically conducted in toxicology to investigate shortterm and acute health effects with an emphasis on 1–4 hr exposure duration, although longer-term studies as well as intermittent, multi-day studies with exposures generally up to 6–8 hr/day were included. The intermittent, multi-day studies were typically reproductive/developmental studies. There were two studies included in the database that determined concentrations that produced respiratory distress in 50% of the animals and two studies that evaluated cardiac sensitization in dogs that involved a 10-min exposure duration. The duration exposures of the original studies were not adjusted unless the primary investigator adjusted the duration because of the uncertainties involved with extrapolating exposure durations from longer exposure to shorter exposure durations (Jarabek, 1995). The database consisted mainly of studies in rodents and rabbits although a few studies used dogs and non-human primates. For the TOC approach, inhalation NOAELs for all 97 chemicals were included because of the need to include as many chemicals as possible into the different toxicity categories. However, for calculating the NOAEL-to-LC50 (N–L) ratio, intermittent, multi-day studies were excluded unless the original investigators noted effects after the first day of exposure. Data on 55 chemicals with NOAEL study exposure durations 624 hr were used to calculate N–L ratios. Studies that identified a lowest-observed-adverse-effect level (LOAEL) as well as a NOAEL were preferred. However, since inhalation exposure studies are more difficult to conduct than oral exposure, studies that identified only a LOAEL or NOAEL were included in the database in order to include as many chemicals as possible in the different toxicity categories. For four chemicals, some well-conducted studies only identified a NOAEL at the highest dose tested. For these four chemicals, the free standing NOAEL was conservatively chosen as the highest dose tested. In other instances, only LOAELs were identified and these studies were included if studies by other investigators supported the use of the LOAEL. In such cases, the LOAEL was converted to a NOAEL by the use of an UF, as described in the following section.
2.3. Adjustment of LOAELs to NOAELs 2. Methods 2.1. Acute inhalation reference database Ninety-seven organic and inorganic substances representing a range of industrial, environmental, and consumer chemicals were included in the acute inhalation reference database. Databases known to contain well-validated acute toxicological endpoints for a series of chemicals were consulted first since the quality of NOAELs and acute lethality studies would have been reviewed and, presumably, only acceptable studies would have been included in the toxicity assessments. These included acute reference exposure levels (RELs) from California Office of Environmental Health Hazard Assessment (OEHHA); acute (1–14 days) inhalation minimal risk levels (MRLs) from the Agency for Toxic Substances and Disease Registry (ATSDR); and Final and Interim Acute Exposure Guideline Levels (AEGLs). AEGLs are developed by the National Advisory Committee and reviewed by the Committee on Toxicology, a part of the National Research Council of the National Academies of Science (NRC, 2001). No studies on metallics, nerve gases, or non-structurally defined substances (such as gums, resins, oils, etc.) were included in the reference database, although inorganic compounds were included in the database. Munro et al. (1996) did not include inorganics in his study since the Cramer Structural Classes are for organic compounds.
2.2. Selection of the NOAEL based on animal studies and duration exposure adjustments For each of the 97 substances, the inhalation NOAEL with an endpoint that was relevant to humans was selected based on the most sensitive species, sex, and endpoint and was entered into the database. There was one chemi-
Typically, an UF of 10 was applied to the LOAEL to calculate the NOAEL. However, if the respiratory tract/irritancy health endpoint was considered to be mild as described in OEHHA (1999), an UF of 6 was applied to the LOAEL to calculate a NOAEL (Alexeeff et al., 2002).
2.4. Criteria for selection of acute lethality data Acute lethality data were used to assign chemicals to different acute GHS toxicity categories proposed by the United Nations (UN, 2005) (Table 1). LC50 data were also used to calculate N–L ratios. For many of the substances, more than one LC50 was identified from the literature, resulting from the fact that many substances were tested in more than one species and sex and/or at different exposure durations. This led, in some cases, to multiple LC50s or LD50s for individual substances. Typically, the lowest LC50 or LD50 data were used to categorize the chemicals and the lowest LC50 data were used to calculate the N–L ratios. Fig. 1 illustrates the steps that were followed to categorize chemicals into toxicity categories (TCs) and the criteria for selection of LC50 data to calculate the N–L ratios. First, LC50 data for all species 64 hr were obtained (Step 1). Values were adjusted to correspond to a 4-hr exposure duration because inhalation cut-off values for the GHS categories are based on 4-hr testing exposures (UN, 2005). Duration adjustments for LC50 data were made using Haber’s Law (Rinehart and Hatch, 1964) as modified by ten Berge et al. (1986). C n T ¼ K; where C is the exposure concentration, T is the exposure duration; K is a cumulative exposure constant, and ‘‘n’’ represents a chemicaland endpoint-specific exponent proposed by ten Berge et al. (1986). If a
c
d
a
Oral LD50 (mg/kg) LC50 Gas (ppmv) LC50 dust and mistsb mg/m3 LC50 vapoursc mg/m3 Tenth percentile composite TOC concentrationsd
b
>2,000a >5000 >5000 >20,000 1000 lg/m3 >300 to 62000a >2500 to 65000 >1000 to 65000 >10,000 to 620,000 125 lg/m3 >50 to 6300a >500 to 62500 >500 to 61000 >2000 to 610,000 125 lg/m3
Category 4 Category 3
Mild irritant
Category 2
Irritant 2A: Causes serious eye irritation 2B: Causes eye irritation >5 to 650a >100 to 6500 >50 to 6500 >500 to 62000 20 lg/m3
Category 1 most toxic
Corrosive Causes serious eye damage 0 to 65a 0 to 6100 0 to 650 0 to 6500 4 lg/m3 Skin irritation Eye irritation
Table 1 Globally Harmonized System of Classification and Labeling of Chemicals (GHS)
Oral LD50 data were not used to categorize a chemical for acute inhalation toxicity unless it was scientifically defensible to perform route-to-route extrapolation for that chemical. Liquids with a vapor pressure <0.1 mm Hg are treated as a mist. Solids are treated as dust. Liquids with a vapor pressure P0.1 mm Hg are treated as a vapor. Refer to Table 3.
R.L. Grant et al. / Regulatory Toxicology and Pharmacology 47 (2007) 261–273 Category 5 least toxic
264
chemical- and endpoint-specific value for ‘‘n’’ was available, it was used to adjust the LC50 data. If not, a default value of n = 1 was used. If all extrapolated values produced the same TC, no further action was required. If the extrapolated values produced different TCs, then the lowest TC was chosen. The quality of the experimental study, physical/chemical characteristics of the chemical, and other data such as eye/skin irritation, etc. were also used to decide the chemical’s TC. If LC50 data 64 hr were not available, then LC50 data >4 hr but 612 hr were obtained (Step 2). Duration adjustments were not performed on LC50 data >4 hr because of the uncertainties involved with extrapolating exposure durations from longer exposure to shorter exposure durations (Jarabek, 1995). If all values produced the same TC, no further action was required. If the extrapolated values produced different TCs, then the lowest TC was chosen. The quality of the experimental study, physical/chemical characteristics of the chemical, and other data such as eye/skin irritation, etc. were also used to decide the chemical’s TC. If LC50 data 612 hr were not available, then all other pertinent inhalation lethality data (i.e., LClo, LC33, etc.) were used to categorize a chemical (Step 3). This is generally a conservative approach because these values are lower than LC50 data. If inhalation lethality data were not available, then oral LD50 data were used to categorize a chemical using the GHS categorization scheme (step 4), but only if strict criteria for performing route-to-route extrapolation were met as discussed in USEPA (1994). Ethylene glycol was the only chemical included in the database based on LD50 data. Acute dermal LD50 data were not used to categorize chemicals because this data did not correctly categorized the chemicals for the evaluation of acute inhalation toxicity.
2.5. Assigning chemicals to Cramer structural classes Chemicals were assigned to Cramer structural classes using a computer program, Windows Toxtree based on the decision tree of Cramer et al. (1978). The software was obtained from the European Chemicals Bureau website (http://ecb.jrc.it/QSAR/). Organic chemicals were classified in Cramer Class I (least toxic class), II (intermediate class), and III (most toxic class).
2.6. The acute inhalation NOAEL and LC50 database Appendices, which are available as supplementary materials available on the Regulatory Toxicology and Pharmacology website, contain the following data: Appendix A provides data on LC50 or LD50 for chemicals by GHS category; Appendix B provides information on NOAELs for chemicals by GHS category; and Appendix C data on N–L ratios. An Excel spreadsheet of electronic data is available from the authors upon request.
2.7. Data and statistical analysis S-Plus statistical software (version 6.2) was used to perform statistical analysis. Distribution data were evaluated to determine whether they were normally distributed using the Kolmogorov–Smirnov test. The data were considered to be normally distributed if p P 0.05. Differences among GHS categories were evaluated with a one-way analysis of variance with a Scheffe´’s post hoc test. If there were two groups, statistical differences were evaluated with a standard two-sample t test. An approach presented by Berthouex and Brown (2002) was used to calculate parametric percentile values. The 95% tolerance bounds for the percentiles were calculated using procedures in Hahn and Meeker (1991) and StInt Software for Statistical Intervals (1993 version) available from the following website: http:// www.public.iastate.edu/~wqmeeker/stint.html. The fifth and tenth percentiles of the cumulative percentage distribution of NOAEL in each GHS category were divided by an UF of 100 to account for uncertainties due to animal-to-human extrapolation and human variability in order to derive values protective of the general population. The fifth and tenth percentiles of the cumulative percentage distribution of N–L ratios were divided by an UF of 100 to calculate a composite factor N–L ratio.
R.L. Grant et al. / Regulatory Toxicology and Pharmacology 47 (2007) 261–273
265
Fig. 1. Procedure for placing chemicals into GHS toxicity categories and for selecting LC50 data.
2.8. Published acute toxicity values based on human and animal studies Published acute toxicity values developed by OEHHA and ATSDR (i.e., acute RELs and acute MRLs, respectively) based on exposure durations of 624 hr from human and animal studies were entered in the database in order to assess whether the proposed health-protective concentrations developed using either the TOC approach or the composite factor N–L ratio approach were health-protective. It should be kept in mind that the toxicity values published by OEHHA and ATSDR were developed for different exposure durations and risk assessment purposes as discussed by Woodall (2005). Acute RELs are no-effect levels used to evaluate exposures based on an averaging time of one hour, except for reproductive/developmental toxicants which have an averaging time based on the number of hrs/day of the multiday-intermittent study (OEHHA, 1999). For example, the averaging time for a study with an exposure duration of 6 hr/day, gestational day 6–15, would be 6 hr. Acute MRLs for non-cancer health effects are used to evaluate exposure durations of up to one day to 14 days.
3. Results 3.1. Derivation of TOC concentrations for different GHS toxicity categories There were 97 chemicals that were classified in GHS acute toxicity categories based on acute lethality data (Appendix A). LC50 data were available for 86 chemicals, other types of inhalation lethality data were available for 10 chemicals, and LD50 data for one chemical, ethylene glycol. Forty-three percent of the acute lethality studies were based on a 4-hr exposure duration; 13% were <1 hr values;
35% were 1–3 hr values; and 12% were 6–10 hr values. Based on the LC50 ranges in the GHS categories (Table 1), 33 chemicals were classified in Category 1, 18 in Category 2, 23 in Category 3, four in Category 4, and 19 in Category 5. Data for the NOAELs based on acute inhalation data for 97 chemicals is shown in Appendix B. The NOAEL data for Category 4 were combined with the data for Category 3 since there were only four chemicals in Category 4 and the range of NOAEL data for these four chemicals corresponded to the range of NOAEL data in Category 3 (Appendix B).The distribution of NOAELs was not normally distributed; however, the natural logarithm (ln) of NOAELs was normally distributed. Therefore, the ln NOAELs were used to calculate percentiles instead of NOAELs. The cumulative percentage distribution of the ln NOAELs for the entire database with calculated 5th, 10th, 20th, 50th, 90th, and 95th parametric percentiles and their corresponding 95% lower and upper tolerance bounds (LTB-UTB) are shown in Fig. 2. The calculated parametric percentiles adequately describe the data set and the variation in the parametric percentiles is small. Fig. 3 presents the cumulative distributions of ln NOAELs of chemicals separated into GHS toxicity categories, the calculated parametric percentiles, and their corresponding 95% LTB-UTB. The calculated percentiles adequately describe the data set and the variation in the percentiles is small. The four category distributions were found to differ significantly from every other category. Table 2 presents the ranges of NOAEL values, geometric
266
R.L. Grant et al. / Regulatory Toxicology and Pharmacology 47 (2007) 261–273
Fig. 2. Cumulative percentage distribution of ln NOAELs for 97 chemicals in the acute inhalation database with calculated parametric percentiles and 95% lower and upper tolerance bounds (LTB-UTB).
mean (GM), geometric mean standard deviation (GSD) and number of chemicals for the GHS categories. Table 3 shows the parametric fifth percentile, tenth percentile, and 50th percentile of the four different GHS category distributions and the overall database of 97 chemicals. The fifth and tenth percentile of inhalation NOAELs for each acute toxicity category was divided by an UF of 100 to calculate the fifth and tenth percentile composite TOC concentration for each GHS toxicity category and their corresponding 95% LTB-UTB (Table 3). The uncertainty in the calculated tenth percentile value was greater for those categories with fewer number of chemicals (i.e., Categories 2 and 5). The difference between the LTB (most conservative value) and the tenth percentile was a factor of two for Category 1 (33 chemicals); a factor of four for Category 2 (18 chemicals); a factor of three for Categories 3 and 4 combined (27 chemicals); and a factor of four for Category 5 (19 chemicals). 3.2. Comparison of the fifth and tenth percentile composite TOC concentrations to published acute toxicity values Table 4 shows published REL and MRL toxicity values derived by OEHHA and ATSDR, respectively, for exposure durations 624 hr. All published toxicity values were higher (i.e., less conservative) than the fifth percentile TOC concentration for each GHS category. All toxicity values were higher than the tenth percentile TOC concentrations except for trans-1,2-dichloroethylene (Table 4). The tenth percentile TOC concentration was less than a factor of two higher than the published toxicity factor even though the averaging time for the ATSDR MRL for trans1,2-dichloroethylene is for 8 hr. Since the TOC concentrations were designed to evaluate one-hour intermittent exposure, these comparisons indicate that the tenth percentile TOC concentrations developed for the GHS categories should be adequately health-protective whereas the fifth percentile TOC concentration may be overly conservative.
Fig. 3. Cumulative percentage distribution of ln NOAELs grouped into GHS Categories with calculated parametric percentiles and 95% lower and upper tolerance bounds (LTB-UTB). A, Toxicity Category 1 (TC 1); B, TC 2; C, TC 3-4; and D, TC 5.
R.L. Grant et al. / Regulatory Toxicology and Pharmacology 47 (2007) 261–273
267
Table 2 Summary statistics of NOAELs for each GHS acute toxicity category 3
Range (mg/m ) GM ± GSD (mg/m3) Number of compounds
Category 1
Category 2
Categories 3 and 4
Category 5
0.09–236 5.4 ± 7.4 33
0.28–381 31 ± 8.6 18
1–3020 172 ± 7.7 27
79–167,000 1300 ± 7.2 19
Table 3 Parametric percentiles of NOAELs and proposed TOC concentrations based on the fifth and tenth percentile for each GHS acute toxicity category
50th percentile Fifth percentile Fifth percentile composite TOC concentrationa (95% tolerance bounds) Tenth percentile Tenth percentile composite TOC concentrationa (95% tolerance bounds)
Overall database
Category 1
Category 2
Categories 3 and 4
Category 5
57 mg/m3 0.56 mg/m3 NAb
5.4 mg/m3 0.20 mg/m3 2 lg/m3 (1–4 lg/m3) 0.41 mg/m3 4 lg/m3 (2–8 lg/m3)
31 mg/m3 0.91 mg/m3 9 lg/m3 (2–30 lg/m3) 2.0 mg/m3 20 lg/m3 (5–50 lg/m3)
172 mg/m3 6.0 mg/m3 60 lg/m3 (20–140 lg/m3) 12.6 mg/m3 125 lg/m3 (40–270 lg/m3)
1300 mg/m3 51 mg/m3 510 lg/m3 (110–1300 lg/m3) 104 mg/m3 1000 lg/m3 (280–2400 lg/m3)
1.6 mg/m3 NAb
a The fifth or tenth percentiles were divided by an UF = 100 and converted from mg/m3 to lg/m3 to calculate the composite TOC concentrations and 95% tolerance bounds for each separate category. b NA, not applicable.
For comparison purposes, the tenth percentile composite TOC concentrations for each category are included in Table 1 in order for a comparison to be made to the ranges of LC50 data for the different GHS toxicity categories. 3.3. Determination of the N–L ratios for different chemicals There were 55 chemicals for which acute LC50 data and corresponding acute NOAEL data for exposure durations of 624 hrs were available (indicated in Appendices A, B, and C). LCLo or other type of acute lethality data were not included. Appendix C shows the calculated N–L ratios for the 55 chemicals. The N–L ratios were not normally distributed; however, the ln N–L ratios were normally distributed. Therefore, the ln N–L ratios were used to calculate percentiles instead of N–L ratios. The cumulative percentage distribution of the ln N–L ratios for the entire database with calculated 5th, 10th, 50th, 90th, and 95th parametric percentiles and their corresponding 95% tolerance bounds are shown in Fig. 4. The calculated parametric percentiles appear to adequately describe the data set although the lower 5th–20th percentiles correspond better than the 50th–95th percentiles. The ranges, GM, GSD, the fifth and tenth percentiles for the distribution of N–L ratios are shown in Table 5. Layton et al. (1987) and Venman and Flaga (1985) did not include inorganic chemicals in their databases when calculating N–L ratios whereas there were 40 organic chemicals and 15 inorganic chemicals included in the present database (Appendix C). There were no significant
differences between the distribution of ln N–L ratios of organic or inorganic chemicals (p < 0.05). Venman and Flaga (1985) only used NOAEL and LD50 data from the rat to calculate N–L ratios. Layton et al. (1987) calculated N–L ratios using rat NOAEL data and LD50 data and compared it to N–L ratios calculated using rat NOAEL data and LD50 data from small mammals. In the present acute inhalation database, there were 26 chemicals where the same species was used to determine both NOAEL and LC50 data: 15 chemicals based on rat data, nine chemicals based on mouse data, and two chemicals based on guinea pig data (Appendix C). There were 29 chemicals where different species were used for NOAEL or LC50 data (Appendix C). There were no significant differences between the distribution of ln N–L ratios of the same species or different species (p < 0.05) although the variation in the ratios was greater when different species were used to calculate N–L ratios. Table 5 shows the calculation of the composite factor. The fifth and tenth percentile values were divided by an UF of 100 to derive a composite factor. The fifth and tenth percentile composite factors of 4.6 · 105 and 8.3 · 105, respectively, were less than a factor of two different from the LTB (more conservative value) of 2.4 · 105 and 4.7 · 105, respectively. The fifth and tenth percentile composite factors can be multiplied by a 4-hr LC50 value or other appropriate acute inhalation lethality data based on steps outlined in Fig. 1 to calculate an interim health-protective air concentration. Scientific judgment should be used to select the most appropriate and defensible acute lethality data.
268
Table 4 Comparison of estimated air concentrations to published toxicity values CAS no.
Toxicity value (lg/m3)
Sourcea
Adjusted duration (original study duration)
GHS category
TOC 5th Percentile (lg/m3)
TOC 10th percentile (lg/m3)
N–L ratio 5th percentile (lg/m3)
N–L ratio 10th percentile (lg/m3)
Acetone (2-propanone) Acrolein Acrylic acid
67-64-1 107-02-8 79-10-7
62,000 7 6000
MRL MRL REL
5C 1C 3C
510 2 60
1000 4 125
3500 2 130
6300 4 230
Acrylonitrile Ammonia
107-13-1 7664-41-7
220 3200
MRL REL
2C 3C
9 60
20 125
33 34
60 61
Arsine Benzyl chloride Carbon monoxide Chlorine Chloropicrin Dichlorobenzene, 1,4-
7784-42-1 100-44-7 630-08-0 7782-50-5 76-06-02 106-46-7
160 240 23,000 210 29 12,000
REL REL REL REL REL MRL
1C 1C 3C 1C 1C 4C
2 2 60 2 2 60
4 4 125 4 4 125
2 10 98 4 3 230
3 17 180 8 5 420
Dichloroethylene, trans-1,2 Dioxane, 1,4 Epichlorohydrin Ethylene glycol monobutyl ether Fluorine Formaldehyde Hydrochloric acid Hydrogen cyanide Hydrogen fluoride Hydrogen selenide Hydrogen sulfide Isopropyl alcohol Malathion Methanol Methyl bromide Methyl chloroform Methyl ethyl ketone Methylene chloride MTBE (methyl tert-butyl ether) Nitric acid Nitrogen dioxide Ozone Perchloroethylene Phenol
156-60-5 123-91-1 106-89-8 111-76-2 7782-41-4 50-00-0 7647-01-0 74-90-8 7664-39-3 7783-07-5 7783-06-4 67-63-0 121-75-5 67-56-1 74-83-9 71-55-6 78-93-3 75-09-2 1634-04-4 7697-37-2 10102-44-0 10028-15-6 127-18-4 108-95-2
800 3000 1300 14,000 16 93 2100 340 240 5 42 3200 200 28,000 3900 68,000 13,000 14,000 7200 86 470 180 20,000 5800
MRL REL REL REL MRL REL REL REL REL REL REL REL MRL REL REL REL REL REL MRL REL REL REL REL REL
4 hr 1 hr 1 hr (6 hr/d on 20 occasions; 6 hr adj. to 1 hr; ‘‘n’’ = 2) 24 hr (8 hr adj. to 24 hr) 1 hr (5 m to 120; m; all adj. to 60 m; ‘‘n’’ = 4.6) 1 hr 1 hr (2 hr adj. to 1 hr; ‘‘n’’ = 2) 1 hr 1 hr (30 m adj. to 1 hr; ‘‘n’’ = 1) 1 hr (10 m adj. to 1 hr; ‘‘n’’ = l) Not specified; occupational exposures 8 hr 1 hr (6 hr no adj.) 1 hr 1 hr (2 or 4 hr adj. to 1 hr) 24 hr (15 m adj. to 24 hr) 1 hr (3 hr adj. to 1 hr, ‘‘n’’ = 2) 1 hr (45 m adj. to 1 hr; ‘‘n’’ = 1) 1 hr (30 m adj. to 1 hr; ‘‘n’’ = 1) 1 hr 1 hr N/A(Mean of 1 hr data) 1 hr (4 m adj. to 1 hr) 6 hr 1 hr (75 m adj. to 1 hr) 1 hr (2 hr adj. to 1 hr) 1 hr (30 m adj. to 1 hr) 1 hr (2 hr adj. to 1 hr) 1 hr (90 m adj. to 1 hr) 24 hr (6 hr adj. to 24 hr) 1 hr (40 m adj. to 1 hr) 1 hr 1 hr 1 hr (3 hr adj. to 1 hr) 1 hr (8 hr adj. to 1 hr)
5C 5C 2C 3C 1C 2C 2C 1C 1C 1C 2C 5C 1C 5C 2C 5C 5C 3C 5C 1C 1C 1C 5C 2C
510 510 9 60 2 9 9 2 2 2 9 510 2 510 9 510 510 60 510 2 2 2 510 9
1000b 1000 20 125 4 20 20 4 4 4 20 1000 4 1000 20 1000 1000 125 1000 4 4 4 1000 20
4000b 2400 69 100 2 16 13 3 3 0.27 28 1800 2 1100 35 1200 1600 360 5500 5 5 1 1200 8
7100b 4300b 120 180 4 29 23 5 6 0.49 51b 3300b 4 2000 64 2100 2800 650 10,000b 8 9 3 2200 15
R.L. Grant et al. / Regulatory Toxicology and Pharmacology 47 (2007) 261–273
Chemical name
Abbreviations: adj., adjusted; C, category; d, day; GHS, Globally Harmonized System of Classification and Labeling of Chemicals (UN, 2005); hr, hour; m, min; N–L ratio, NOAEL to LC50 ratio; TOC, threshold of concern. a Source: MRL, ATSDR acute Minimal Risk Level; REL, OEHHA acute reference exposure levels. b Estimated concentrations that are greater than published values are highlighted in bold.
Phosgene Phosphorus, white Propylene oxide Styrene Sulfur dioxide Sulfuric acid Toluene Triethylamine Vinyl chloride Xylenes
75-44-5 7723-14-0 75-56-9 100-42-5 7446-09-5 7664-93-9 108-88-3 121-44-8 75-01-4 1330-20-7
4 20 3100 21,000 660 120 37,000 2800 180,000 22,000
REL MRL REL REL REL REL REL REL REL REL
1 hr 24 hr (5 m adj. to 24 hr) 1 hr (4 hr adj. to 1 hr) 1 hr 1 hr (varied 5–75 m) 1 hr (16 m adj. to 1 hr) 1 hr (6 hr adj. to 1 hr) 1 hr (8 hr adj. to 1 hr) 1 hr (7.5 hr adj. to 1 hr) 1 hr (30 m adj. to 1 hr)
1C 2C 3C 4C 2C 2C 5C 3C 5C 5C
2 9 60 60 9 9 510 60 510 510
4 20 125 125 20 20 1000 125 1000 1000
0.12 13 120 490 45 16 1200 190 6900 950
0.21 23b 220 880 82 29 2100 350 12,000 1700
R.L. Grant et al. / Regulatory Toxicology and Pharmacology 47 (2007) 261–273
269
Fig. 4. Cumulative percentage distribution of ln NOAEL-to-LC50 ratios for 55 chemicals in the acute inhalation database with calculated parametric percentiles and 95% lower and upper tolerance bounds (LTB-UTB).
3.4. Comparison of adjusted LC50 data based on the fifth and tenth percentile composite factor N–L ratio to published acute toxicity values Estimated threshold concentrations derived using the composite factor N–L ratio were compared to published acute toxicity values derived by OEHHA and ATSDR (Table 4). All published toxicity values were higher than the estimated threshold concentrations based on the fifth percentile composite factors except for trans-1,2-dichloroethylene based on an 8-hr exposure duration (Table 4). All published toxicity values were higher than the estimated threshold concentrations based on the tenth percentile composite factor except for six chemicals (Table 4). Three of the six chemicals were representative of exposure durations P8 hr: trans-1,2-dichloroethylene (8-hr exposure duration MRL); MTBE (24-hr exposure duration MRL); and white phosphorus (24-hr exposure duration MRL). The other three chemicals had exposure durations of one hour, but the difference between the estimated threshold concentrations were less than a factor of two: 1,4-dioxane (one-hour exposure duration REL); hydrogen sulfide (one-hour exposure duration REL); and isopropyl alcohol (one-hour exposure duration REL). Since the threshold concentrations based on the composite factor N–L ratio were designed to evaluate one-hour intermittent exposure, these comparisons indicate that use of the tenth percentile composite factor N–L ratio should be health-protective whereas the use of the fifth percentile composite factor N–L ratio may be overly protective. 3.5. Comparison of health-protective air concentrations estimated with the tenth percentile TOC approach and the tenth percentile composite factor N–L ratio approach For the tenth percentile composite factor, the estimated air concentrations from these two separate approaches
270
R.L. Grant et al. / Regulatory Toxicology and Pharmacology 47 (2007) 261–273
Table 5 Summary statistics for N–L ratios and proposed composite factor
Entire database Composite factora 95% Tolerance bounds
Number
Range
GM ± GSD
Fifth percentile
Tenth percentile
55
0.0009–0.83
0.068 ± 5.2
0.0046 4.6 · 105 2.4 · 105 to 7.5 · 105
0.0083 8.3 · 105 4.7 · 105 to 1.3 · 104
a The fifth or tenth percentile ratio for the entire database was divided by an uncertainty factor of 100 to account for animal-to-human uncertainty and human variability to produce the composite factor N–L ratio.
were similar (Table 4). For approximately 70% of the chemicals, the differences between the estimated threshold concentrations based on the two different approaches were within a factor of three. This is a reflection of the fact that both approaches were developed using similar acute inhalation toxicity data for the chemicals in both databases. The approach based on N–L ratios produced estimated concentrations that were generally higher than the TOC approach (i.e., less conservative). This may be due to different numbers of chemicals and grouping of chemicals in the databases. While there were 55 chemicals in the N–L ratio database and only one group, the TOC approach using the GHS categorization system (UN, 2005) had 97 chemicals and four separate groups with smaller number of chemicals in each group (Table 2). In addition, the TOC approach included animal studies with exposure durations P24 hr and intermittent, multi-day studies in order to include as many chemicals in the different categories as possible. This illustrates a possible advantage of the N–L ratio approach when compared to the TOC approach. A larger number of chemicals are needed to develop TOC concentrations for several different toxicity potency classes. 3.6. Target organ or critical effect and exposure durations of the NOAEL inhalation studies Table 6 provides a summary and Appendix B provides detailed information on the target organ/critical effects of the NOAEL inhalation studies in the database of 97 chemicals. The database contained a variety of different endpoints, although 48% of the inhalation studies produced POE effects such as irritation, respiratory tract, or dermal/ocular effects. Interestingly, the great majority of the chemicals that produced POE effects occurred in Catego-
Table 6 Summary of critical effects of acute toxicity studies used as the basis for NOAELs Critical effect
Number of compounds
Respiratory tract Irritation Dermal/ocular/mixed Respiratory and Systemic Neurological Developmental Hematological Hepatic or renal Cardiac
25 18 3 3 17 17 4 7 3
ries 1 and 2: 76% of chemicals in Category 1 produced POE effects; 78% in Category 2; 22% in Categories 3 and 4 combined; and 11% in Category 5. The exposure durations of the NOAEL animal studies for the 97 chemicals varied (Appendix B): 35% of studies were 61 hr; 24% of studies were >1 hr to 66 hr; and 7% of studies were >6 hr to 624 hr. Thirty-four percent of the studies were intermittent multi-day studies, mainly representing developmental studies. For the subset of 55 chemicals used for the N–L ratio approach (indicated in Appendix B), 52% of NOAEL studies were 61 hr; 26% of studies were >1 hr to 66 hr; and 22% of studies were >6 hr to 624 hr. 3.7. Distributions of NOAELs based on Cramer Structural Classes Appendix B shows the Cramer Structural Classes of the chemicals in the database. A total of 76 organic chemicals were classified using Cramer structural classes. There were 17 organic chemicals in Class I, four in Class II, and 55 in Class III. The cumulative distributions of NOAELs for Cramer Structural Classes I and III did not differ significantly from each other (data not shown). There were not enough chemicals in Cramer Class II to include in the analysis. The Cramer classification scheme developed for chronic oral exposure does not adequately define acute inhalation toxicity potency, possibly due to the fact that many of the chemicals included in the acute inhalation database produced POE effects (Table 6). 4. Discussion 4.1. Uncertainty and variability of LC50 and NOAEL data Both LC50 and NOAEL data are subject to uncertainty and variability due to inter- and intraspecies differences; differences in test protocols, laboratory practices, and measurement technologies; variation according to the age, weight, sex, and health of the test animals; and diet, housing conditions, and ambient temperature (Layton et al., 1987). Therefore, the LC50 and the NOAEL are not biological constants. The NOAEL is the highest exposure level at which there are no biologically or statistically significant increases in the frequency or severity of adverse effect between the exposed population and its appropriate control. The NOAEL approach has been criticized because it does not
R.L. Grant et al. / Regulatory Toxicology and Pharmacology 47 (2007) 261–273
use the full dataset of the dose–response curve and is dependent on spacing between doses. For studies with greater variation in endpoint measurement or smaller sample size, the NOAEL approach tends to result in a higher NOAEL being determined based on statistical comparison to the controls. In addition, some experimental studies only identify a LOAEL which increases the uncertainty in the dose–response assessment. Alternately, a mathematical model can be used to fit the entire dose–response data for a chemical with a non-linear mode of action so that the threshold concentration corresponding to an estimated incidence or change in level of response can be determined (i.e., benchmark concentration as well as the lower confidence limit on the benchmark concentration from a dose–response model). The advantages of benchmark concentration modeling are it uses the full dataset of the dose– response curve, accounts for the greater uncertainty due to smaller sample size or greater variation, and can estimate the threshold concentration when a NOAEL can not be established. Benchmark concentration modeling is preferred over the NOAEL approach although not all dose– response data can be adequately modeled. Travis et al. (2005) discusses the advantages and disadvantages of both the benchmark concentration modeling and NOAEL approach. When gathering data for the acute inhalation database, there was only one chemical, methanol, where data obtained from benchmark concentration modeling was available. In the future, more researchers will conduct dose–response assessments using benchmark concentration modeling and the uncertainty in NOAEL data will be decreased. The LC50 is calculated from a statistical analysis of the fractions of animals dying after being exposed to different concentrations of a chemical. Layton et al. (1987) provides a discussion of three different interlaboratory studies that determined variation of LD50 data. They concluded that ‘‘the actual variation in LD50 values for a given species was surprisingly low, given the various sources of uncertainty associated with lethal toxicities’’ (Layton et al., 1987). However, a similar interlaboratory study on LC50 data has not been performed so the uncertainty in LC50 data due to interlaboratory variation cannot be determined. Because of the uncertainty and variability inherent in determining an LC50 or NOAEL, a statistical method was used to determine the cumulative percentage distribution of a range of acute inhalation NOAELs in each toxicity category or of a range of N–L ratios. The tenth percentile of the distributions was calculated and a total UF of 100 was used to calculate TOC concentrations or a composite factor N–L ratio. These are conservative procedures. There was greater uncertainty in the TOC concentrations for Category 2 and Category 5 when compared to Category 1 and Categories 3–4 because there were fewer number of chemicals in the categories. Conservative measures were used to compile the database, as discussed in the following section. In addition, a
271
comparison of published toxicity values available from OEHHA and ATSDR based on human and animal studies indicate the tenth percentile composite value derived for both methods is conservative to evaluate one-hour exposure. There is less than a 10% probability that toxicological studies of any other substance not included in the database would result in a N–L ratio or TOC concentration lower than the tenth percentile. If human exposure to a chemical is below the estimated health-protective air concentration, the substance can be judged, with reasonable confidence, to present a low probability of risk, even if there is limited toxicity information. 4.2. Conservative measures used to compile the database Ideally, the database of NOAELs used to develop a TOC approach or N–L ratio approach for one-hour exposure durations would contain only studies of one-hour exposure durations. In order to include as many chemicals in the database as possible and to ensure only high-quality studies were included in the database, studies based on exposure durations longer than one-hour were included which is a conservative procedure. As mentioned previously, the exposure durations of NOAEL data were not adjusted to a one-hour exposure duration because of the uncertainties involved with extrapolating exposure durations from longer exposure to shorter exposure durations (Jarabek, 1995). For four chemicals, free-standing NOAELs were included, which is conservative since an actual NOAEL could be much higher. The TOC approach used studies of longer exposure durations such as multi-day intermittent exposure whereas the N–L ratio approach used NOAELs from exposure durations 624 hr. As a consequence, the air concentrations determined using the TOC approach were generally more conservative than air concentrations derived using the N–L ratio approach. The N–L ratio approach has the advantage of having 55 chemicals in one category as opposed to the TOC approach where there were essentially four categories with fewer chemicals in each category. 4.3. Health-protective air concentrations based on the TOC approach The basic approach that has previously been used to establish a chronic oral TOC dose (Munro et al., 1996) has been used to establish health-protective human exposure threshold concentrations for the evaluation of one-hour intermittent inhalation exposures. The Cramer structural class system (Cramer et al., 1978) was not used since it did not correctly predict acute inhalation potency although a structural class system would be preferred over the use of acute inhalation data since the only information needed to categorize a chemical is its structure. Acute inhalation lethality data is not available for all chemicals. However, if chemicals are grouped according to acute inhalation toxicity potency as defined by GHS (UN,
272
R.L. Grant et al. / Regulatory Toxicology and Pharmacology 47 (2007) 261–273
2005), a distribution of NOAELs that are statistically different from each other can be generated for different categories. When appropriate UFs are applied to the category-specific tenth percentile of the distribution of NOAELs, health-protective air concentrations are derived for each category. If the only toxicity information for a chemical is LC50 data, then the chemical can be placed in an appropriate GHS toxicity category and the categoryspecific TOC concentration can be used as an interim health-protective air concentration. 4.4. Health-protective air concentrations based on the composite factor N–L ratio approach The basic approach that has previously been used by Layton et al. (1987) and Venman and Flaga (1985) to estimate a chronic oral benchmark has been used to estimate healthprotective human exposure threshold concentrations for the evaluation of one-hour intermittent inhalation exposures. An estimated air concentration calculated by multiplying the tenth percentile composite factor of 8.3 · 105 by 4-hr LC50 data or other appropriate acute lethality data (Fig. 1) can be used as an interim health-protective air concentration for a chemical that has limited toxicity information. Venman and Flaga (1985) developed a chronic oral N–L ratio using NOAEL and LD50 data from only one species, the rat, and suggested their acute-to-chronic application factor be applied to oral rat LD50 data only. Layton et al. (1987) calculated a chronic oral composite N–L ratio of 5 · 106 when both NOAEL and LD50 data were obtained from the rat and another chronic oral composite N–L ratio of 1 · 105 when NOAEL data were obtained from the rat and LD50 data were obtained from small mammals. Layton et al. (1987) suggested that values from 5 · 106 to 1 · 105 per day be used to establish interim acceptable daily intakes based on scientific judgement. Since the rat is not the best model for toxic responses in humans nor is it the most sensitive species to any given toxic response, NOAEL and LC50 data from the most sensitive species were included in the acute inhalation database. Approximately half the ratios were obtained from the same species and half from different species. The inclusion of these different groups increased the variation within the tenth percentile composite N–L ratio but is based on a larger dataset of 55 chemicals and can be applied to LC50 data from various species based on criteria in Fig. 1 to calculate health-protective air concentrations. 4.5. The evaluation of one-hour intermittent inhalation exposure for chemicals with limited toxicity information Acute lethality data are used in both the tenth percentile category-specific TOC concentration approach and the tenth percentile composite factor N–L ratio approach so it is possible to determine health-protective air concentrations using both approaches. Since both approaches are conservative, the decision on which approach to use for a
chemical should be based on scientific judgment and an analysis of all available data including physical/chemical parameters, proposed mode of action, toxicity of structurally similar chemicals or chemical classes, data quality of the acute lethality study, etc. For example, if there is limited information on the quality of the study used to determine a LC50 value, using a range of LC50s as defined by the GHS system to categorize the chemical and adopting a TOC concentration for the category may be more defensible than using the N–L ratio approach. In the future, acute inhalation lethality may be evaluated using the fixed dose procedure, the acute toxic class method or the up-anddown procedure (Lipnick et al., 1995; Botham, 2004) instead of determining a LC50 value. In that case, a TOC approach may be preferred. Both methods should be regarded as useful tools to predict health-protective air concentrations for a chemical with limited toxicity information in conjunction with a weight-of-evidence approach on the toxicity assessment of the chemical. 4.6. Conclusion If a chemical has limited acute inhalation toxicity information, then the use of the TOC approach or composite factor N–L ratio approach can be used on an interim basis until additional acute toxicity information becomes available for a chemical. Air concentrations derived based on the TOC or N–L ratio approach will be used as generic ESLs in Texas for chemicals that have limited toxicity information in order to evaluate short-term one-hour exposures. The generic ESLs will be used in the TCEQ air permitting process to assess the protectiveness of substance-specific emission rate limits for facilities undergoing air permit reviews. If the GLCmax, a worst case concentration resulting from a worst case emission rate, is below the generic ESL for that chemical, the substance can be judged, with reasonable confidence, to present a low probability of risk. Acknowledgments The authors gratefully acknowledge the significant contribution of Dr. Valerie Meyers who classified the chemicals in the database into Cramer structural classes and Mike Aplin who reviewed the manuscript. Appendices A, B, and C. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.yrtph.2006. 11.003. References Alexeeff, G.V., Broadwin, R., Liaw, J., Dawson, S.V., 2002. Characterization of the LOAEL-to-NOAEL uncertainty factor for mild adverse effects from acute inhalation exposures. Regul. Toxicol. Pharmacol. 36, 96–105.
R.L. Grant et al. / Regulatory Toxicology and Pharmacology 47 (2007) 261–273 Berthouex, P.M., Brown, L.C., 2002. Estimating percentiles. In: Berthouex, P.M., Brown, L.C. (Eds.), Statistics for Environmental Engineers, second ed. Lewis Publishers, Washington, D.C., pp. 71–75. Blackburn, K., Stickney, J.A., Carlson-Lynch, H.L., McGinnis, P.M., Chappell, L., Felter, S.P., 2005. Application of the threshold of toxicological concern approach to ingredients in personal and household care products. Regul. Toxicol. Pharmacol. 43, 249–259. Botham, P.A., 2004. Acute systemic toxicity—prospects for tiered testing strategies. Toxicol. In Vitro 18, 227–230. Craig, D.K., Davis, J.S., Hansen, D.J., et al., 2000. Derivation of temporary emergency exposure limits (TEELs). J. Appl. Toxicol. 20, 11–20. Cramer, G.M., Ford, R.A., Hall, R.L., 1978. Estimation of toxic hazard— a decision tree approach. Food Cosmet. Toxicol. 16, 255–276. Dolan, D.G., Naumann, B.D., Sargent, E.V., Maier, A., Dourson, M., 2005. Application of the threshold of toxicological concern concept to pharmaceutical manufacturing operations. Regul. Toxicol. Pharmacol. 43, 1–9. Fay, M. Luukinen, B., Holler, J., 2005. Extrapolation of LC50s to safe exposure levels—insights for emergency responders. In: Society of Toxicology 2005 Annual Meeting, Toxicologist 44, 302 (Abstract #1481). FDA (Food and Drug Administration). 1995. Food additives; threshold of regulation for substances used in food-contact articles. Federal Register 60 (136) 36582–36596. Ford, R., Davis, R., Gardner, D., 2006. Development of inhalation thresholds of concern for the evaluation of tobacco additives. In: Society of Toxicology 2006 Annual Meeting, Toxicologist 45, 497 (Abst. #2317). Hahn, G.J., Meeker, W.Q., 1991. Methods for calculating statistical intervals for a normal distribution. In: Hahn, G.J., Meeker, W.Q. (Eds.), Statistical Intervals A Guide for Practitioners. John Wiley & Sons, Inc., Canada, pp. 53–74. Jarabek, A.M., 1995. The application of dosimetry models to identify key processes and parameters for default dose–response assessment approaches. Toxicol. Lett. 79, 171–184. JECFA (Joint FAO/WHO Expert Committee on Food Additives). 1993. Evaluation of certain food additives and contaminants (Fortieth report of the Joint FAO/WHO Expert Committee on Food Additives). WHO Technical Report Series, No. 832. JECFA (Joint FAO/WHO Expert Committee on Food Additives). 1995. Evaluation of certain food additives and contaminants (Forty-third report of the Joint FAO/WHO Expert Committee on Food Additives). WHO Technical Report Series, No. 855. JECFA (Joint FAO/WHO Expert Committee on Food Additives). 1997. Evaluation of certain food additives and contaminants (Forty-sixth report of the Joint FAO/WHO Expert Committee on Food Additives). WHO Technical Report Series, No. 868. Kroes, R., Renwick, A.G., Cheeseman, M., Kleiner, J., Mangelsdorf, I., Piersma, A., Schilter, B., Schlatter, J., van Schothorst, F., Vos, J.G., Wurtzen, G.European branch of the International Life Sciences Institute, 2004. Structure-based thresholds of toxicological concern (TTC): guidance for application to substances present at low levels in the diet. Food Chem. Toxicol. 42, 65–83.
273
Kroes, R., Galli, C., Munro, I., Schilter, B., Tran, L., Walker, R., Wurtzen, G., 2000. Threshold of toxicological concern for chemical substances present in the diet: a practical tool for assessing the need for toxicity testing. Food Chem. Toxicol. 38, 255–312. Layton, D.W., Mallon, B.J., Rosenblatt, D.H., Small, M.J., 1987. Deriving allowable daily intakes for systemic toxicants lacking chronic toxicity data. Regul. Toxicol. Pharmacol. 7, 96–112. Lipnick, R.L., Cotruvo, J.A., Hill, R.N., Bruce, R.D., Stitzel, K.A., Walker, A.P., Chu, I., Goddard, M., Segal, L., Springer, J.A., Myers, R.C., 1995. Comparison of the up-and-down, conventional LD50, and fixed-dose acute toxicity procedures. Food Chem. Toxicol. 33, 223–231. Munro, I.C., Ford, R.A., Kennepohl, E., Sprenger, J.G., 1996. Correlation of structural class with no-observed-effect levels: a proposal for establishing a threshold of concern. Food Chem. Toxicol. 34 (9), 829– 867. Munro, I.C., 1990. Issues to be considered in the safety evaluation of fat substitutes. Food Chem. Toxicol. 28, 751–753. NIOSH (National Institute for Occupational Safety and Health). 2005. Documentation for Immediately Dangerous to Life or Health Concentrations (IDLHs). US Department of Health and Human Services, Centers for Disease Control and Prevention, National Institute for Occupational Safety and Health, Cincinnati, OH. Available at: