SRA Logo (print)

Society For Risk Analysis Annual Meeting 2009

Risk Analysis: The Evolution of a Science

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

Poster Platform: Toxicity in an Evolving World Sponsored by DRSG

Room: Salon A   1:30-3:00 PM

Chair(s): Margaret MacDonell

T3-C.1  13:30  The Update Project of the EPA's Integrated Risk Information System (IRIS) Program. Shoaf CR*, Foureman GL; US Environmental Protection Agency; US Environmental Protection Agency

Abstract: The IRIS Program develops Agency scientific positions on potential adverse human health effects from exposure to various chemical substances found in the environment. Health assessments developed by the IRIS Program are used by EPA program offices and regions, as well as state and international programs. A project to update IRIS values is underway. The project revisits all IRIS oral reference doses (RfDs), inhalation reference concentrations (RfCs), cancer slope factors, and inhalation unit risks posted more than ten years ago. The update process involves: (1) identification and prioritization of qualified values, (2) call for further technical information via a Federal Register Notice (FRN) and conduct of a literature search, (3) identification of new and relevant information, (4) development of draft documents, (5) review(s) of these documents, and (6) posting the updated assessments on IRIS. The update process uses current Agency methods (e.g., benchmark dose modeling, application of uncertainty factors, and animal to human dosimetry adjustments). In general, the update process differs from the standard IRIS Program in that it assesses batches of 6-10 chemicals, uses a Federal Standing Science Review Committee for interagency review, and uses the Science Advisory Board as a standing committee for external review. Also, a new document, the Summary of Updated Changes, is developed and integrated into the IRIS files to explain the updating process for each assessment. A preliminary group of chemicals has been chosen for reassessment. [The views expressed in this abstract are those of the authors and do not necessarily reflect the views or policies of the U.S. Environmental Protection Agency.]

T3-C.2  13:30  U.S. EPA framework for determining mutagenic mode of action for carcinogens. Schoeny R*, Owen R, McCarroll N, Kligerman A, Keshava N, Keshava C, Dearfield K, Cimino M, Putzrath RM, McMahon T; U.S. Environmental Protection Agency, U.S. Department of Agriculture

Abstract: U.S. EPA’s Guidelines for Carcinogen Risk Assessment (US EPA, 2005) specify that information on a chemical’s mode of action (MOA) can be of value in the risk assessment process. MOA analysis is useful for making decisions on the human relevance of animal tumors, on appropriate low-dose extrapolation, and whether age dependent adjustment factors can be used to account for early lifestage susceptibility. EPA recommends (in the absence of data to support lifestage-specific cancer slope factors) that during risk characterization of early lifestage exposure, adjustment factors are applied for carcinogens shown to have a mutagenic MOA. EPA has developed a draft framework for organizing and evaluating data to determine if a mutagenic MOA is applicable. This relies heavily on data from genotoxicity studies, but emphasizes that the MOA judgment uses all relevant data (e.g., pharmacokinetics, tumor data, structure activity relationships). The draft framework identifies four steps in the iterative MOA process. 1. Assemble relevant data. All available information on the mutagenic and genotoxic effects of a chemical is obtained to ensure as full a mutagenicity assessment as possible. 2. Evaluate data against current acceptance and quality criteria. It is recommended that all studies be included in a data comparison table that is described in the framework. 3. Judge the weight of evidence (WOE) that the chemical is mutagenic. WOE is judged based on several factors such as the type of assays, results, effects, endpoints, among others. The framework provides a hierarchy of preference for assays to be used in the WOE. 4. Once it is decided that the chemical or its metabolites are mutagenic, apply the EPA Cancer Guidelines framework for determining a MOA. Two brief case studies (cyclophosphamide and chloroform) will be used to demonstrate use of these steps. Disclaimer: The views represented in this abstract are those of the authors and do not necessarily represent the policies of EPA or USDA.

T3-C.3  13:30  Uncertainty modeling in dose response: Bench tests to support derivation of toxicity values. Cooke RM, MacDonell MM*; Resources for the Future (Cooke), Argonne National Laboratory (MacDonell)

Abstract: Health risks from environmental contaminants are commonly assessed by combining exposure estimates with reference toxicity values to predict potential noncancer effects and cancer risk. Considerable uncertainty underlies the toxicity values used in these calculations, from inherent biological processes to the models applied in evaluating dose-response information. To address the latter, a workshop convened interdisciplinary experts to assess uncertainty in modeling dose-response data from animal bioassays. The objective was to promote better quantification of uncertainty for health risk analyses. The focus was on one part of the quantification challenge: fitting models to animal toxicity data within the observed range, with an emphasis on new approaches and environmental relevance. Various experts tested four data sets with approaches ranging from benchmark dose modeling to probabilistic inversion with isotonic regression, non-parametric Bayes modeling, and Bayesian model averaging. Results were reviewed by different experts representing three different perspectives: epidemiologist/toxicologist, statistician/mathematician, and risk analyst/regulator. Comments were also invited from other participants, and the modelers responded to the critiques. Subjecting results to these reviews not only strengthened the model explanations, it also provided valuable insights regarding their uses and limitations. The interactive format helped address two common complaints of dose-response modelers: statisticians do not understand what their models don’t do well and are not sufficiently mindful of biological problems inherent in data analyses. The contrast presented in the approaches of statisticians versus those who model dose-response heightened joint awareness of these issues and the importance of physiological parameters underlying dose-response relationships. This talk summarizes approaches, results, and insights gained across different options for characterizing this component of uncertainty in dose-response modeling for health risk analyses.

T3-C.4  13:30  Conundrums with Uncertainty Factors. Cooke RM*; RESOURCES FOR THE FUTURE

Abstract: The current practice of Uncertainty Factors (UFs) as applied to non-cancer endpoints in the IRIS data base lacks a valid probabilistic foundation. Several authors have proposed a “random chemical” interpretation under which UFs are random variables distributed as ratios of toxicity indicators (eg EDs, NOAELs, etc) for animals and humans, under various conditions, for toxic substances drawn randomly from a population of similar substances. The same interpretation underlies the recent NRC proposals for advancing risk assessment. Under prevailing assumptions, this leads to a log covariance matrix which is singular, meaning that there is a log linear relation between the variables of which uncertainty factors are ratios. Based on probabilistic arguments, several authors claim that the current practice of uncertainty factors is over-protective. These arguments use these same assumptions and lead to the same problems. To give an idea, let HS be a random variable denoting the toxic indicator value for Humans at Subchronic dosage, AC, the value for Animals at Chronic dosage, and let HC, AS be defined similarly. Then HS = AC*(AS/AC)*(HS/AS) := AC*(1/US)*(1/UH); where US is the UF for subchronic-to-chronic, and UH is the UF for Animal-to-Human. The assumption of independence of US and UH has often been noted. However, if we observe AC=a for some new chemical, then we need that the conditional distribution (HS|AC=a) is distributed as a*(AS/AC)*(HS/AS); and this requires additional independence assumptions. Compounding this is the assumption that, e.g. UH= AC/HC = AS/HS. The paper shows that these assumptions induce a singularity in the covariance matrix of {Log(HC), Log(HS), Log(AC), Log(AS)}. The practice of uncertainty factors is due for a thorough review. Two directions are briefly sketched, one based on standard regression models, and one based on non-parametric continuous Bayesian Belief Nets.

T3-C.5  13:30  Carcinogenic Potency Assessment for 2-Aminonaphthalene. . Naufal Z, Collie S*, Smith J, Wilson C; R.J. Reynolds Tobacco Company, Synergy Toxicology

Abstract: The compound 2-aminonaphthalene (2AN) is an aromatic amine classified by IARC as carcinogenic to humans (Group 1) based on sufficient evidence of carcinogenicity in humans and animals. The USEPA has not classified 2AN as to its carcinogenic potential. While human epidemiology studies are supportive of the carcinogenic potential of 2AN, they do not inform as to the quantitative dose for 2AN potency since all available studies are of mixed exposures including exposure to other known carcinogens. Therefore, risk assessment involving 2AN largely relies on potency values estimated from animal models. In the case of 2AN, the only available cancer slope factor (CSF) value (1.8 per mg/Kg-day) was derived by California EPA in an expedited assessment. A T25 value of 12.8 mg/Kg-day has also been derived for 2AN. In this report, available dose-response assessment data on 2AN were reviewed and two critical studies were identified as appropriate for developing a quantitative potency factor for 2AN based on USEPA guidelines. In these studies, rhesus monkeys and beagle dogs were exposed to 2AN via the oral route. Both animal species exhibited an increase in urinary bladder tumors congruent with results from human studies which suggest that the urinary bladder is the main target organ of 2AN. The multistage-cancer model was fitted to the tumor incidence data to determine a point of departure for low-dose linear extrapolation. An inhalation unit risk was subsequently calculated from the oral data. Ultimately, improved statistical fit and the closest non-human species were demonstrated with the rhesus monkey data. In conclusion, more thorough characterization of cancer potency values for 2AN potentially improves on cancer risk assessment of chemical mixtures where 2AN is detected.

T3-C.6  13:30  Use of life-stage adjustment in oral risk assessment for 4,4-methylene dianiline. English JC*, Ball GL, McLellan CJ; NSF International, Ann Arbor, MI

Abstract: An acceptable level of 4,4’-methylene dianiline in drinking water was determined using NSF/ANSI 61 Annex A (2008) and U.S. EPA (2005) risk assessment guidelines. Accidental human exposure to 4,4′-methylene dianiline has been associated with hepatotoxicity. The key study was a chronic drinking water bioassay for 4,4′-methylene dianiline dihydrochloride in which thyroid and liver tumors were observed in male and female F344 rats and B6C3F1 mice (NTP, 1983). Since the weight of evidence suggests that 4,4’-methylene dianiline is genotoxic in vitro and in vivo, a mutagenic mode of action was considered. Due to the potential for greater susceptibility to tumor development when exposures occur in early life stages as compared with later life stages for carcinogens that act through a mutagenic mode of action, the cancer risk level was modified to incorporate potency adjustments for early-life exposures. Default adjustment factors (10 for the risk during the first two years of life; 3 for the risk for ages 2 through < 16; and 1 for ages 16 until 70) were applied due to the lack of toxicokinetic or toxicodynamic data for 4,4′-methylene dianiline during early life exposure. Benchmark dose analyses were conducted assuming both linear (mutagenic) and non-linear (growth stimulation) modes of action in the thyroid, and a linear, non-threshold mode of action in the liver. The combined incidence of hepatocellular adenomas and carcinomas in female mice was the most sensitive test system and endpoint, since these data produced the lowest BMDL10 of 0.67 mg/kg-day. The life-stage adjusted 10-5 cancer risk level of 0.000024 mg/kg-day, that was extrapolated from this BMDL10, corresponds to a Total Allowable Concentration (TAC) of 0.8 ug/L. This drinking water level is protective of public health since it was based on chronic oral data from the most sensitive endpoint and laboratory animal species, and takes into account susceptibility from early life exposure.

T3-C.7  13:30  Independent Expert Peer Workshop for the Toxicological Assessment and Development of RfDs for Acetanilide Degradates: A Workshop Using the Alliance For Risk Assessment (ARA) Collaborative Model. Parker A*, Gadagbui B, Dourson M, Christoper J, Maier A, Willis A; 1 Toxicology Excellence for Risk Assessment (TERA), 2 California Environmental Protection Agency

Abstract: The workshop was organized to provide a diverse group of independent experts’ review of toxicity data to develop RfDs for tertiary-ethanesulfonic acid (t-ESA) and tertiary-oxanilic acid (t-OXA) degradates of acetochlor and alachlor. The workshop was open to the public; interested persons were invited to attend the meeting either in person or via a real-time Internet webcast. Observers were provided the opportunity to provide written or oral technical comments before and during the meeting. TERA compiled the data into a series of comprehensive tables which included key data summary tables, critical endpoint summary tables, benchmark dose modeling results, potential RfD summary tables, and copies of key references for the parent chemicals (acetochlor and alachlor) and their degradates (acetochlor t-ESA, acetochlor t-OXA, alachlor t-ESA, and alachlor t-OXA). The panel assessed the critical effect, appropriate NOAEL, LOAEL, or BMD and recommended uncertainty factors for the above acetanilide degradates using the compiled tables and supporting materials during the 2-day workshop. The panel had an active discussion on the adequacy of the data, critical effects, mode of action, uncertainty factor selection, and derivation of reference values. A summary of the panel’s interpretation of the data, selected uncertainty factors and resulting oral reference dose (RfD) values will be presented. This project was conducted under the auspices of the Alliance for Risk Assessment (ARA), a collaboration of stakeholders representing government, academic, industry, environmental and consulting perspectives. As an ARA project, this assessment was vetted for scientific relevance and was conducted by an independent, nonprofit organization, using state-of-the-science chemical risk assessment methods to protect public health. ARA risk assessments are performed in a transparent manner, and made publicly available upon completion.

T3-C.8  13:30  Human health risk from exposure to perfluooctanoic acid (PFOA). Bastaki M*, Calkins M, Lawrence C; State College

Abstract: Perfluorooctanoic acid (PFOA) is a stable synthetic perfluorinated carboxylic acid and fluorosurfactant. It is considered a member of a class of emerging persistent pollutants and emerging endocrine disrupters. The consumption of contaminated food is the main exposure pathway, followed by the ingestion of dust and inhalation of air. The highest dose was obtained in high-exposure scenario for toddlers (128ng/kg/day), while average adult dietary intake from whole diet in adults is estimated at 70ng/kg/day. PFOA is readily absorbed but not metabolize in animals and humans, and has an estimated mean serum half-life 3.8 years (95% CI 3.1-4.4) in humans. It can cross the placental barrier and it accumulates in plasma, liver and lung tissues. We estimated the risk for non-cancer endpoints using a reported NOAEL of 0.06 mg/kg/day based on increased liver weights and liver hypertrophy. This is chosen over the less sensitive endpoint of body weight reduction. A total of an uncertainty factor of 100 was applied to derive a RfD of 0.6ug/kg/day. While the hazard index (HI) for the average adult exposure and high exposure scenario for toddlers are 0.1167 and 0.213, respectively and imply low risk for non-cancer liver toxicity, the choice of endpoint remains rather gross and does not account for more subtle effects including reproductive or developmental. Given the long biological half-life and the potential to accumulate, along with the chronic nature of exposure scenarios and the developmental stage in young children, more sensitive endpoints must be employed in assessing the toxicity of PFOA, including assessment of endocrine function endpoints.

T3-C.9  13:30  Development of human health toxicological criteria for five organic acids. Neuber K*, Durda J, Bradley A; Integral Consulting Inc.

Abstract: Site-specific human health toxicological criteria were developed for dimethyl phosphorodithioic acid (DMPT), diethyl phosphorodithioic acid (DEPT), o-phthalic acid, p-chlorobenzenesulfonic acid (pCBSA), and bezenesulfonic acid (BSA) using existing animal toxicology studies and toxicological surrogates. These five compounds were detected in the groundwater at a former industrial facility in Nevada but no published toxicological criteria were available with which to interpret the potential significance of these findings. For four of these organic acids, chronic oral reference doses (RfDs) were developed based upon existing toxicity data for assigned toxicological surrogates. DMPT and DEPT are metabolites of organophosphate chemicals that inhibit acetylcholinesterase (AChE) in mammals. Dimethoate and phosalone, both AChE inhibitors, were originally proposed as site-specific toxicological surrogates for DMPT and DEPT, respectively. However, there is sufficient evidence in the literature to support the assertion that DEPT and DMPT are not AChE inhibitors. Therefore, diisopropyl methylphosphonate was selected as a toxicological surrogate for DEPT and isopropyl methylphosphonic acid was selected as a toxicological surrogate for DMPT. Similarly, o-phthalic acid is a metabolite of various phthalic acid esters in humans and mammals. Di-(2-ethylhexyl) phthalate (DEHP) was originally proposed as the toxicological surrogate for o-phthalic acid. There is sufficient evidence in the literature to support the conclusion that DEHP has significantly different toxicological endpoints than o-phthalic acid. Instead, phthalic anhydride was selected as the toxicological surrogate for o-phthalic acid. For BSA, a toxicological surrogate of p-toluenesulfonic acid (pTSA) was selected. For pCBSA, a chronic oral RfD was derived from an existing subchronic duration rat study. For DMPT, DEPT and o-phthalic acid, oral RfDs established by USEPA for the assigned surrogates were used.

T3-C.10  13:30  Dose-response assessment for influenza A virus based on the datasets for its vaccine strains. Watanabe T*, Bartrand TA, Omura T, Haas CN; The University of Tokyo

Abstract: Quantitative risk assessment of current-circulating H1N1 and H3N2 influenza A viruses would be helpful in preparing for such influenza pandemics. This study was aimed at developing a dose-response assessment for influenza A virus as a key step of risk assessment. Since reported datasets on infection of human volunteers challenged with wild-type influenza A virus at graded doses are few, we first developed dose-response models based on the datasets for 11 vaccine strains of its virus. Dose-response relationships for all strains could be described by a beta-Poisson model. Pooling analysis of the models with respect to virus subtype (H1N1 or H3N2), attenuation method (cold-adapted or avian-human reassortment) for vaccine development and host age (adults or children) demonstrated that virus subtype and host age are significant factors to determine the infectivity of influenza A virus. The attenuation method affected only the infectivity of H1N1 subtype to adults. We also tried to pool the datasets on infection of adult volunteers challenged with wild-type and vaccine strains in order to compare their infectivities. Those datasets for H3N2 virus could be successfully pooled and the pooled datasets suggested that gene reassortment would attenuate wild-type H3N2 virus by 770 times. Considering this degree of attenuation, 10% infectious dose of wild-type H3N2 virus for adults was estimated as 25 TCID50 (95% C.I. = 12 to 49 TCID50). The infectivity of wild-type H1N1 virus was still unknown since the datasets pooling was failed. These findings in dose-response relationships for influenza A virus would contribute to understanding the virulence of H5N1 or other subtypes of influenza A virus which have capacity to cause the pandemic, just like the current influenza pandemic, and to controlling local and global transmission of influenza via environmental reservoirs or human-to-human communication.

[back to schedule]