HomeAbout SRAMembershipEventsNewsletterJournalOpportunitiesResourcesContact

Back to 2007 Meeting Main Page

Society For Risk Analysis Annual Meeting 2007

Risk 007: Agents of Analysis

Session Schedule & Abstracts

* Disclaimer: All presentations represent the views of the authors, and not the organizations that support their research. Please apply the standard disclaimer that any opinions, findings, and conclusions or recommendations in abstracts, posters, and presentations at the meeting are those of the authors and do not necessarily reflect the views of any other organization or agency. Meeting attendees and authors should be aware that this disclaimer is intended to apply to all abstracts contained in this document. Authors who wish to emphasize this disclaimer should do so in their presentation or poster. In an effort to make the abstracts as concise as possible and easy for meeting participants to read, the abstracts have been formatted such that they exclude references to papers, affiliations, and/or funding sources. Authors who wish to provide attendees with this information should do so in their presentation or poster.

Common abbreviations

M3-E Statistical Methods for Evaluating Environmental Chemical Mixtures Room: 205   2:00 - 3:30 PM

Chair(s): Glenn Rice, Lynn Flowers

M3-E.1    Sample size requirements for the 4 lab multigenerational nioassay of a complex mixture of drinking water Disinfection By-Products. Dingus C*, Teuschler LK, Rice GE, Simmons JE, Narotsky MG; Battelle; NCEA/ORD/U.S. EPA; NCEA/ORD/U.S. EPA; NHEERL/ORD/U.S. EPA; NHEERL/ORD/U.S. EPA   feder@battelle.org Abstract: Complex mixtures of disinfection by-products (DBPs), that occur at low concentrations in drinking water, need to be tested in toxicological bioassays for reproductive and developmental effects observed in some epidemiologic studies. There is a growing emphasis in toxicology to test complex chemical mixtures at environmentally relevant doses; these components and their relative proportions in the mixture should reflect those seen in environmental samples, and the impact of the unidentified materials in the mixtures should be considered. However, studying environmental mixtures such as DBPs is fraught with technical issues that preclude using standard protocols and bioassays. Study design, in particular power analysis, is important to ensure that experimental results are meaningful. This paper describes a statistical approach to calculating power for a multigenerational rodent bioassay (known as U.S. EPA’s 4 Lab study) to test for reproductive and developmental effects from exposure to a complex mixture of DBPs in drinking water concentrates. Due to physical and logistical constraints, the bioassay was divided into two blocks. The maximum number of dams per block was limited to 100. In this paper, a methodology for calculating power for non-independent observations using a two-block design is described. Two endpoints, prenatal loss and pup weight, are used to illustrate the methodology using data from a pilot study of the developmental effects of a complex mixture. The power analysis of the two-block study design with 100 animals in each block, a 40:60 ratio of control:treatment group animals, and a significance level of 0.05 provides approximately 90% power to detect pup weight decreases and 45% power to detect prenatal loss increases in the F1 generation, respectively. This abstract does not represent EPA policy.

M3-E.2    Evaluating sufficient similarity for drinking water disinfection by-product (DBP) mixtures with bootstrap hypothesis test procedures. Feder PI*, Ma Z, Bull RJ, Simmons JE, Schenck KM, Teuschler LK, Rice GE; Battelle; Battelle; MoBull Consulting; NHEERL/ORD/U.S. EPA; NRMRL/ORD/U.S. EPA; NCEA/ORD/U.S. EPA;NCEA/ORD/U.S. EPA    feder@battelle.org Abstract: The U.S. Environmental Protection Agency (EPA) has issued guidance documents that provide methods for assessing health risks from complex chemical mixture exposures. EPA’s 2000 Chemical Mixtures Guidance states “...If no data are available on the mixture of concern, but health effects data are available on a similar mixture...a decision must be made whether the mixture on which health effects are available is ‘sufficiently’ similar to the mixture of concern to permit a risk assessment…”. We apply the nonparametric "bootstrap" technique to evaluate the similarity of drinking water disinfection by-product (DBP) mixtures. The bootstrap extension of multivariate analysis of variance is discussed and illustrated with chemical data for DBP mixtures,which were reported in a study by EPA and the Association of Metropolitan Water Districts. Normal theory multivariate analysis of variance and its nonparametric bootstrap counterpart are applied to assess similarity of the composition of DBP mixtures obtained from 35 different drinking water utilities; the utilities obtained their waters from different sources and used different water treatment methods. The empirical distribution functions of the F-statistics corresponding to the bootstrap null distributions for testing the hypotheses of no overall utility effects, no treatment effects, and no water source effects are shown to have heavier tails than the corresponding normal theory F-distributions. Thus, observed significance levels based on the bootstrap test are less extreme than those based on the F-test, thus, normal theory procedures may overstate the strength of the statistical evidence concerning these effects. These results suggest that use of normal theory tests could potentially indicate departures from similarity of complex mixtures due solely to departures from the normality assumptions. Nonparametric procedures are more robust to departures from distributional assumptions and should be more routinely incorporated into risk assessment practice. (The abstract does not represent EPA Policy).

M3-E.3    Evaluation of Proportional Response Addition Techniques for Analysis of Mixtures Data. Simmons JE*, Stiteler W, Hertzberg RC, McDonald A, Sey YM, Teuschler LK, Rice G, Colman J, Durkin P; U.S. EPA   Simmons.Jane@epa.gov Abstract: Progress has been made in development of methods/approaches for toxicological evaluation and risk assessment of chemical mixtures. However, few methods have been developed to evaluate whether the observed toxicity of a mixture is consistent with response addition. Under proportional response addition (PRA), the expected effect of the mixture is the proportionally weighted sum of the responses of the individual components at their concentrations in the mixture. A statistical approach consistent with PRA has been developed, based on the method of linear contrasts, which can be used to predict the effects of mixtures from single-chemical dose-response data. It provides for statistical evaluation of both deviation from additivity and its direction (greater or less than additive) without requiring an assumption of similarity either of mode of action or dose-response curve shape. The analysis approach is illustrated by comparison of the mixture response predicted under PRA to experimentally observed mixture data. Experiments were conducted with individual trihalomethanes (THMs) (chloroform, chlorodibromomethane, dibromochloromethane, bromoform) and their binary combinations. They were designed for PRA analysis, as each THM pair was tested at 2 different total mixture doses and 2 different THM ratios. Female CD-1 mice were exposed by daily aqueous gavage for 14 days. Hepatotoxicity was assessed by relative liver weight and serum levels of enzymes indicative of hepatic damage (SDH, ALT, AST). Generally, the experimentally observed and predicted mixture responses were not statistically different. As the variance in the responses increased, the ability (power) to detect departures from additivity decreased. In conclusion, a novel methodology for evaluating PRA has been developed; future work should be directed toward understanding the application of PRA methods in the low dose region. (This abstract does not reflect EPA policy.)

M3-E.4    Empirical evaluation of sufficient similarity in dose-response for mixtures of many chemicals. Stork LG*, Gennings C, Carter, Jr. WH, Teuschler LK, Carney EW; 1 Monsanto Co., St. Louis, MO; 2-3 Virginia Commonwealth University, Richmond, VA; 4 U.S. EPA/ORD/NCEA, Cincinnati, OH; 5 The Dow Chemical Company, Midland, MI    leanna.g.stork@monsanto.com Abstract: The U.S. EPA (Suppl Guidance for Conducting Health Risk Assessment of Chemical Mixtures, 2000) suggests that when toxicity data are not available for a mixture of concern, the risk assessment can be based on data for a surrogate mixture which is considered “sufficiently similar” in terms of chemical composition, component proportions, and toxicological properties. As a supplementary approach, using statistical equivalence testing logic we have developed methodology to define sufficient similarity in dose-response for mixtures of many chemicals containing the same components, but whose component ratios vary. The method assumes that dose-response data on a fixed-ratio mixture are available. Equivalence testing logic is applied to estimate boundary ratios for mixtures with dose-response relationships sufficiently similar to an observed mixture, based on a specified biologically meaningful dose-response region of similarity using expert judgment. To illustrate this method we use the data set of Rajapakse et al. (EHP, 2002), who tested a fixed-ratio mixture of 11 synthetic chemicals in a yeast estrogen receptor reporter gene assay. The 11-chemical mixture was combined with the endogenous hormone, estradiol in a 50,000:1 ratio. The degree to which this ratio could vary and still result in a sufficiently similar dose-response relationship (e.g., defined as +/- one log10 change in the EC50 and/or a +/- 10% shift in the maximum response) was then estimated statistically. Using the proposed method, mixing ratios ranging from 41,000:1 to 60,100:1 are estimated to yield sufficiently similar dose-responses to that of the evaluated 50,000:1 mixture. Statistical approaches such as this may help guide the application of sufficient similarity in dose-response concepts in chemical mixtures risk assessment. (Partially supported by NIEHS training grant #T32 ES007334. These data and methods are not associated with Monsanto. This abstract does not reflect EPA policy.)

M3-E.5    Proposed Derivation of Relative Potency Factors (RPFs) for Individual Polycyclic Aromatic Hydrocarbons (PAHs) and Characterization of Uncertainty . Carlson-Lynch H*, Stickney J, McClure P, Gehlhaus M, Flowers L; H Carlson-Lynch, J Stickney, and P McClure: Syracuse Research Corporation; M Gehlhaus and L Flowers: National Center for Environmental Assessment, U.S. EPA   hclynch@syrres.com Abstract: U.S. EPA’s Integrated Risk Information System (IRIS) Program is undertaking a health assessment for PAH mixtures that considers an RPF approach, which estimates cancer risk for a PAH mixture by summing the carcinogenic potential of the component PAHs relative to an index compound such as benzo(a)pyrene. To identify data suitable for RPF derivation, dose-response data sets were extracted from studies of carcinogenicity or cancer-related endpoints in which at least one PAH was compared with benzo[a]pyrene. Relative potency estimates were calculated from each data set, using benchmark dose modeling to estimate a point-of-departure for slope estimation when possible, and a point estimate approach when modeling was not feasible. One or more relative potency estimates were calculated for 50 PAHs. The proposed approach for deriving a single RPF for individual PAHs uses a decision tree to identify potentially carcinogenic PAHs. For these PAHs, a draft RPF was selected from the relative potency estimates using a ranking framework to assess the relevance of the underlying endpoint to tumor formation and study quality. Each draft RPF was given a relative confidence rating based on the quality and consistency of the overall database, as well as the quality of the principal study(s) on which the draft RPF is based. An extensive data array was developed to capture the variability and uncertainty in the database for each of the 50 PAHs. Some of the areas of uncertainty in the approach include: (1) possible contributions to carcinogenicity from PAHs without adequate data for developing RPFs; (2) possible non-additive interactions among components of whole PAH mixtures; and (3) extrapolating human cancer risks from cancer responses in laboratory animal studies and in vitro cancer-related data. (The views expressed are those of the authors and do not necessarily reflect the views or policies of the U.S. EPA).

M3-E.6    The relative potency factor (RPF) approach and polycyclic aromatic hydrocarbon (PAH) mixtures. Flowers L*, Carlson-Lynch H, Gehlhaus M, Keshava C, McClure P, Rice G, Stickney J, Strong J, Teuschler L; USEPA, Syracuse Research Corporation   gehlhaus.martin@epa.gov Abstract: An RPF approach is being developed by U.S EPA’s Integrated Risk Information System (IRIS) Program to provide risk assessors with methodology to estimate cancer risks from exposure to PAH mixtures. In the RPF approach, the carcinogenic potential of selected individual PAHs is compared to that of an index compound, e.g., benzo[a]pyrene (BaP), yielding a PAH-specific RPF. The specific PAH dose is multiplied by its RPF yielding an index chemical equivalent dose. The index chemical equivalent doses are summed for all selected PAHs yielding an equivalent dose for the mixture. The equivalent dose is multiplied by the index chemical’s slope factor to estimate the cancer risk from the quantified PAHs in a PAH mixture. An extensive literature search of available in vivo and in vitro cancer-related data for individual PAHs was conducted. BaP was the only PAH which was identified as a possible index chemical. Dose-response information was extracted from studies in which at least one PAH was studied in addition to BaP, resulting in data for one or more endpoints for 50 PAHs found in environmental mixtures. One of the major issues associated with the use of the RPF approach for PAH mixtures is that a common mode of action for carcinogenicity is assumed and is the underlying basis for the assumption of dose additivity. Issues to be discussed include: (1) the potential for multiple modes of carcinogenic action; (2) known interactive effects among PAHs and between PAHs and other chemicals; and 3) the use of a response addition approach to address carcinogenic contributions from non-PAH components in environmental PAH mixtures. (The views expressed are those of the authors and do not necessarily reflect the views or policies of the U.S. EPA).

M3-E.7    Overview of U.S. EPA’s Integrated Risk Information System (IRIS) assessment for polycyclic aromatic hydrocarbon (PAH) mixtures. Gehlhaus M*; USEPA   gehlhaus.martin@epa.gov Abstract: PAHs are a class of aromatic hydrocarbons which are products of the incomplete combustion and pyrolysis of organic substances including coal, wood, and petroleum products. PAHs occur in the environment as complex mixtures, such as coal tar and coke oven emissions. Under the Clean Air Act, PAHs are listed within the polycyclic organic matter (POM) group and constitute the major risk component associated with this group. Under the Clean Water Act, sixteen PAH compounds are on the Priority Pollutant List generated in the 1970s. These sixteen compounds are now listed on the Contract Laboratory Program Target Compound List for the Superfund Program and are routinely sampled for at hazardous waste sites. Current U.S. EPA PAH health assessments were entered on the IRIS database around 1990 and include those for 15 non-methylated PAHs with three or more rings (e.g., benzo[a]pyrene) and three PAH-containing mixtures (coke oven emissions, diesel engine emissions, and creosote). The current IRIS assessments do not address the environmental occurrence of PAHs as complex mixtures and do not reflect the most recent research on PAHs. For these reasons, the IRIS Program is undertaking a health assessment for chronic environmental exposure to PAH mixtures. Three approaches described in the U.S. EPA Guidance for the Health Risk Assessment of Chemical Mixtures are potentially useful for PAH mixtures health assessment, including two whole mixture approaches, the comparative potency and surrogate mixture approach, and a component approach utilizing relative potency factors. The latter approach has been undertaken for further development. (The views expressed are those of the authors and do not necessarily reflect the views or policies of the U.S. EPA).

M3-E.8    Development of Relative Potency Factors (RPFs) for Polycyclic Aromatic Hydrocarbon (PAH) Mixtures: Implementation Issues for Risk Assessment. Teuschler LK*, Rice GE, Flowers L, Gehlhaus M, Keshava C, Strong J; U.S. Environmental Protection Agency   teuschler.linda@epa.gov Abstract: The development of RPF values for PAHs on U.S. EPA’s Integrated Risk Information System (IRIS) database is planned as a significant first step in providing scientifically sound risk assessment information for this important class of environmental mixtures; whole mixture surrogate and comparative potency approaches are also in preparation. Methodology is needed for applying the RPF approach with attention to specific issues impacting the certainty and confidence of the application. Such issues include: 1) selection of benzo[a]pyrene (BaP) as the index chemical and the strength of the BaP dose-response information; 2) support for the assumptions of a common carcinogenic mode of action and dose additivity; 3) applying the RPF approach to PAH mixtures with a combination of low and high confidence RPFs; 4) assessing the composition of the PAH mixture, including the %BAP in the mixture and the % of the total PAH mixture not accounted for by the RPF approach; 5) extrapolating across routes or estimating multiple-route risks; and, 6) estimating risk and analyzing uncertainty when applying central tendency estimates and ranges of PAH RPFs. The RPF approach is a useful, flexible method for assessing risk associated with exposure to environmental mixtures. However, it may not always be optimal for assessing the wide variety of PAH mixtures that occur in the environment; thus, methodology is needed to identify appropriate situations for application of RPFs vs. other component or whole mixture approaches. (The views expressed are those of the authors and do not necessarily reflect the views or policies of the U.S. EPA).

[back to schedule]