By Jay Goodman
This Dissenting Statement is written because I believe Committee’s Report should call for: 1) the use of mode of action (MOA) information to inform the shape of the dose-response curve for inorganic arsenic (iAs)-induced adverse effects; and 2) a discussion of the limitations/shortcomings of EPA’s proposed mathematical procedures
- The EPA’s decision to not use mode of action (MOA) data to define the shape of the dose-response curve is wrong, MOA data should be employed
The EPA’s decision to rely on epidemiological studies for use in dose-response analysis rather than employing MOA information (as indicated in Section 2.3.2 of the Protocol document) is wrong. A principal concern here is that while some epidemiology studies involve relatively low levels of exposure to iAs many of them have a focus on high exposure levels and, thus, responses observed at high doses will highly influence the estimated dose-response curves. Furthermore, as indicated in the Protocol’s Section 5.6, extrapolations may be made down to an order of magnitude below observed data. In my view, ignoring the rich body of inorganic arsenic (iAs) information regarding MOA in the Protocol, and iAs IRIS Assessment it is intended to result in, will lead to documents that are not scientifically credible. Biology trumps mathematics. MOA information should be employed to inform the shape of the dose-response relationships, including giving full consideration to the existence of thresholds for adverse effects of iAs. A recent paper by Tsuji et al. (Tsuji et al., Dose-response for assessing the cancer risk of inorganic arsenic in drinking water: The scientific basis for use of a threshold approach, Critical Rev. Toxicol., 2019, https://doi.org/10.1080/10408444.2019.1573804) provides a very good example of how this can be accomplished.
My position regarding the need to use MOA data is consistent with the National Research Council’s 2013 report entitled “Critical Aspects of EPA’s IRIS Assessment of Inorganic Arsenic.” The Summary, top of p. 6, states that, “Mode-of-action analyses should be used to inform dose–response modeling with respect to the shape of the curve, particularly in the low dose region, and the understanding of interhuman variability.” Additionally, the Summary of the Report’s 6-page Mode of Action Section, p. 75, states that, “Mode-of-action analyses permit a transparent assessment of the data supporting or refuting the human health effects of exposure to oral inorganic arsenic in the US population. Both cancer and noncancer effects can be integrated through a mode-of-action analysis. Such analysis permits an unbiased use of all the available data to examine the effects of arsenic exposure and exposure duration at different doses.” Furthermore, the Summary of the December 2, 2015 Meeting of the Committee on Inorganic Arsenic (Phase 2) includes a section which chides EPA for not indicating that MOA analysis will to be performed, “The NRC (2013) report recommended that MOA analyses be used to integrate data throughout the dose-response continuum to improve understanding of exposure-response relationships and interhuman variability of response, but no information was presented by EPA on how this will be accomplished” (Letter from Susan Martel, NAS, to Janice Lee, EPA, dated December 16, 2015).
Discuss limitations/shortcomings of mathematical procedures
The EPA should acknowledge and discuss the limitations/shortcomings of the mathematical procedures proposed in the iAs Protocol and the IRIS Assessment it will lead to. For example, Dr. Thomas B. Starr provided a Public Comment (Technical Comments on the USEPA Updated Problem Formulation and Protocol for the Inorganic Arsenic IRIS Assessment, June 27, 2019) that can help the Agency be more transparent in this regard. In particular, Dr. Starr indicates that the EPA’s proposed application of Bayesian Meta-Regression analysis to observational studies of heterogeneous populations which have experienced heterogeneous exposures [e.g., iAs epidemiology studies] can lead to inaccurate estimates. This underscores my view that EPA should use MOA data to inform the shape of the dose-response curve.
Letter from Susan Martel, NAS, to Janice Lee, EPA, dated December 16, 2015.
National Research Council, Critical Aspects of EPA’s IRIS Assessment of Inorganic Arsenic, 2013.
Starr, Thomas B., Public Comment, Technical Comments on the USEPA Updated Problem Formulation and Protocol for the Inorganic Arsenic IRIS Assessment, June 27, 2019.
Tsuji et al., Dose-response for assessing the cancer risk of inorganic arsenic in drinking water: The scientific basis for use of a threshold approach, Critical Rev. Toxicol., 2019, https://doi.org/10.1080/10408444.2019.1573804.
The NRC (2013) report recommended that MOA data be used to inform the shape of the dose-response curve when extrapolating below the range of observation. EPA acknowledged this recommendation. In fact, the statement about extrapolating down to an order of magnitude below the observed data in Section 5.6 of the IRIS iAs assessment plan is a reference to this recommendation. Since the NRC (2013) report was issued, however, the epidemiological database involving exposure to iAs at lower doses has increased, such that EPA states that “[extrapolation into the lower dose range] is not needed because EPA is modeling human data, and the lower range of exposures reported in the epidemiological studies are very near U.S. background exposure levels” (EPA 2019, p. 12). Given the stronger database that is now available, the committee finds it appropriate that EPA has chosen to perform dose-response meta-analyses to support its iAs IRIS assessment. EPA’s hierarchical, Bayesian meta-analysis approach is an accepted and contemporary technique that is still being refined to address the wealth of epidemiology data available for low-dose arsenic exposures. However, the committee shares the concern and recommendation that EPA needs better documentation of the methodology and transparent communication of the approach. In the opinion of the committee, this Bayesian technique provides an appropriate means of determining the shape of the dose-response curve using actual data from a body of evidence that includes low-dose exposures within the observed range of human exposures. Additionally, the committee has recommended several types of sensitivity analyses, including the removal of the high dose range or studies with only high doses, to evaluate their influence on the shape of the dose-response curve, which directly addresses one of Dr. Goodman’s concerns. The committee disagrees with him that dose-response meta-analyses and sensitivity analyses when data for low-dose effects exist are scientifically less credible than using MOA data and are fraught with limitations/shortcomings that will lead to inaccurate estimates. These are valid approaches for exploring factors that could influence how a body of evidence is interpreted, and have been used by other National Academies committees (for example, see meta-analyses and sensitivity analyses presented in NASEM ).
EPA (U.S. Environmental Protection Agency). 2019. Updated Problem Formulation and Protocol for the Inorganic Arsenic IRIS Assessment [CASRN 7440-38-2]. May 2019. EPA/635/R-19/049. Integrated Risk Information System,
National Center for Environmental Assessment, Office of Research and Development, U.S. Environmental Protection Agency, Washington, DC [online]. Available: https://cfpub.epa.gov/ncea/iris_drafts/recordisplay.cfm?deid=34395 [accessed May 28, 2019].
NASEM (National Academies of Sciences, Engineering, and Medicine). 2017. Application of Systematic Review Methods in an Overall Strategy for Evaluating Low-Dose Toxicity from Endocrine Active Chemicals. Washington, DC: The National Academies Press.
NRC (National Research Council). 2013. Critical Aspects of EPA’s IRIS Assessment of Inorganic Arsenic: Interim Report. Washington, DC: The National Academies Press.