EPA Issues Guidance for Developing Data-Derived Extrapolation Factors
The U.S. Environmental Protection Agency (EPA) announced on October 3, 2014, the availability of a final document entitled Guidance for Applying Quantitative Data to Develop Data-Derived Extrapolation Factors for Interspecies and Intraspecies Extrapolation (DDEF Guidance). EPA states that the DDEF Guidance “lays out methods for calculation of factors compensating for the application of animal toxicity data to humans (interspecies) and for compensating for sensitive populations (intraspecies).” According to EPA, the use of data to conduct these extrapolations rather than relying on default values “advances EPA’s policy of considering relevant data first when conducting its chemical assessments.” As discussed below, the DDEF Guidance is a welcome addition to EPA’s growing collection of guidance documents designed to ensure actual data are used over default assumptions.
Historically, EPA has employed default uncertainty factors in its computation of toxicity values (e.g., reference concentrations (RfC) and reference doses (RfD)) to compensate for a lack of data. EPA states that it based default uncertainty factors on policy or regulatory positions rather than on empirical data applicable to the chemical of interest. The uncertainty factors used in EPA assessments include those compensating for a lack of information on how well animal models used in toxicity studies mimic humans and differences in response between the majority of the population compared with the sensitive individual. While the DDEF Guidance represents a response to recommendations by the National Academy of Sciences’ report Science and Decision: Advancing Risk Assessment, EPA states that, with the publication of An Examination of EPA Risk Assessment Principles and Practices, a “Staff Paper” published in 2004, and EPA’s Guidelines for Carcinogen Risk Assessment, published in 2005, EPA announced its policy of considering all relevant data before applying default values.
In 2011, EPA published Recommended Use of Body Weight 3/4 as the Default Method in Derivation of the Oral Reference Dose. EPA states in that guidance that it listed the optimal approach as using a physiologically-based pharmacokinetic or other biologically-based model versus the default approach using the ratio of body weights raised to the 3/4 power. According to EPA, the DDEF Guidance “lays out a computational process for using chemical-specific data on toxicokinetics (absorption, metabolism, distribution and excretion) and toxicodynamics (response of the tissue to the active form of the agent).” EPA notes that the DDEF Guidance “is the first EPA product to provide a method both for quantitative determination of relative sensitivity of the pharmacodynamic response in an assessment and for empirical determination of intraspecies sensitivity.” EPA states that, as such, “this method provides a valuable tool for identifying and quantifying sensitive populations and lifestages.”
The DDEF Guidance states that the goals of data-derived extrapolation factors (DDEF) are to maximize the use of available data and improve the scientific support for a risk assessment. According to the DDEF Guidance, a DDEF approach is an accepted approach for deriving RfCs, RfDs, or counterpart values, and is consistent with existing EPA guidance. The DDEF Guidance presents EPA’s approach “to identifying, justifying, and employing quantitatively useful data to develop nondefault values for inter- and intraspecies extrapolation. Moreover, this guidance will aid risk assessors and researchers in identifying data gaps and developing informative experiments to yield quantitatively valuable data.”
The DDEF Guidance describes the process for identifying pertinent data useful for quantifying inter- and intraspecies differences to serve as the basis for empirically determined DDEFs. The DDEF Guidance states that, when using DDEFs, inter- and intraspecies extrapolation factors are divided into two components representing toxicokinetic (TK) and toxicodynamic (TD) variability. According to the DDEF Guidance, key considerations include identifying an adverse health outcome, a measurable biological event associated with that adverse health outcome, and the concentration of the toxicant associated with the development of the biological event. Interspecies TK variability is quantified based on the external exposure that produces the same tissue concentration in animals and in humans. Intraspecies TK variability is defined as differences in tissue concentration attained from the same human external exposure. TD variability is quantified on the basis of differences in the tissue or in vitro concentration that produce the same response between animals and humans or among humans.
The DDEF Guidance focuses on identification and use of data that may be used to derive DDEFs to modify default factors for interspecies and intraspecies extrapolation in the development of RfD and RfC values. The DDEF Guidance does not address other default factors used in the RfD/RfC methodology, such as study duration or database deficiency. The interspecies extrapolation factor is designed to account for differences in sensitivity between animal(s) and humans. The policy adopted by EPA is that humans are — by default — more sensitive than laboratory animals to the toxic effects of a given chemical unless data show otherwise. The intraspecies extrapolation factor accounts for variability within the human population both in kinetic behavior and response, including the possibility that sensitive populations may not be adequately represented in available data sets.
The information requirements to derive the different components in the extrapolation factors consist of “sufficient” information on the mode of action of the chemical; toxicology data to identify target tissue(s) or organ(s); and kinetic data. There is no clarity as to what constitutes “sufficient” information about a mode of action, however.
Few chemicals are likely to have the required data set to implement fully the DDEF Guidance. The exception could be food-use pesticide active ingredients where a mode of action analysis may be possible with metabolism studies and kinetic data available as part of the registration package. For most chemicals, available data may support derivation of DDEF values for only some of the four components that make up the extrapolation factors.
Derivation of a DDEF for intraspecies extrapolation is likely out of reach for most chemicals because use of human data is inherently part of the process. The case studies used to illustrate this particular derivation address the TK component of the DDEF and involve chemicals (methyl mercury and boron) with human data sets. There was no illustration of the derivation of an intraspecies TD component. Derivation of a DDEF for interspecies extrapolation (TK and TD) appears to be more approachable because the required information is more likely to be included in established data sets for certain regulatory actions. An adequate animal toxicity study could be sufficient to derive the TK component by using the body weight to the 3/4 power scaling approach. The prospect of ToxCast data may eventually facilitate the derivation of DDEF values by contributing data to support identification of key events in the mode of action of chemicals, and by allowing comparison of responses in appropriate assays with animal and human tissues.
We look forward to the applied use of the DDEF approach as that experience will likely help to clarify the meaning of “robust” or “sufficient” studies and data sets that would be needed to apply the approach. We also take note of the statement in EPA’s policy of “considering relevant data first” when conducting chemical assessments. As a general matter, we agree that actual data and information should be relied upon when they are available, and that this approach should be applied generally by EPA in its assessments, including those conducted on new chemicals under the Toxic Substances Control Act (TSCA). In this regard, however, we have seen instances where EPA continues to fall back on the results of predictive approaches such as Quantitative Structure Activity Relationship (QSAR) analysis or exposure models despite data suggesting that toxicity concerns or the exposure/release levels are lower than have been suggested by the modeled results. We thus welcome efforts by EPA such as its recent workshop on chemicals that are “difficult to test” using standard aquatic toxicity test methods with an eye to developing approaches to generate useful and reasonably obtainable test data and thereby avoid default QSAR approaches. We encourage EPA’s New Chemicals Program to look for additional opportunities to demonstrate its commitment to “data over defaults.”