For the purposes of conducting a robust risk assessment, how do you recommend the potential hazards/risks from PFAS mixtures be assessed?

Answer Explanations

• Assume dose additivity

  Expert 1

  Where total organic fluorine levels exceed guidance values, targetted methods could be used to quantify known PFAS. For all PFAS with a health-based criterion, hazard indices could be calculated and summed. A read-across approach could be employed for measured PFAS with no health-based guidance value, and an overall qualitative assessment could be performed accounting for i) PFAS with a health-based value, ii) PFAS measured in targetted analyses but withouth a guidance value, and iii) PFAS not measured in targetted methods but contributing to the total organic fluorine level.

• Use a whole-mixtures approach Other (please explain)

  Expert 9

  As discussed in Question 1.6, many PFAS are understudied or have not been studied at all with respect to their hazard identification, dose-response, and exposure pathways (however, it is likely that many PFAS track together with respect to exposure pathways, i.e., water contaminated with AFFFs will likely contain complex mixtures of 100s of PFAS). It may be possible to conduct robust risk assessments using whole-mixtures approaches for products that contain known PFAS such as food packaging. However, even in scenarios where information is complete, a whole-mixtures approach may be warranted Complex environmental mixtures of PFAS comprise a scenario where there is the potential for significant human exposure and there is some evidence for toxicity of PFAS as mixtures. These two characteristics indicate a need for a chemical mixture assessment (from: Chemical Mixtures: A Framework for Assessing Risks to Human Health, 2008).

• Other (please explain)

  Expert 10

  Scientists don't agree that liver hypertrophy is the appropriate endpoint, even for all PFAAs, and therefore it's unlikely that there will be agreement on application of the relative potency approach (RPF). Currently, there is not enough information to develop the RPF approach for other endonts.

  I believe that we could have a hybrid approach (not listed above) for risk assessment, based only on exposure, which could be coined an "internal dose additivity" approach. Instead of assuming all external exposures result in the same internal dose (the "dose additivity" approach as applied in current drinking water guidelines in the EU, Sweden, Denmark etc.), I'd account for differences in elimination half-lives where possible but still assume a comon toxicity for all PFAS based on the most toxic one (often PFOS or PFOA). It's a simplification of course, but better than assuming that even short-chain PFAAs bioaccumulate to the same extent as long-chain PFAAs. Such an approach would require agreement on human elimination half-lives for multiple PFAS, which might be problematic.

• Other (please explain)

  Expert 8

  As stated above, this depends both on the problem formulation statement, the PFAS compounds being evaluated, the state of knowledge of those specific compounds, the route(s) of exposure. Risk assessment is an iterative process where estimates of exposure and hazard are progressively refined. A more robust assessment will utilize PBPK if species specific half-life is an issue, will utilize more refined exposure estimates, will utilize where possible MOA considerations, will consider appropriate age groups, and so on. The state of knowledge of the specific PFAS group being evaluated will dictate the appropriate method.

• Other (please explain)

  Expert 7

  Seems to be contradictory or the same as 1.6.
  Based on this Task 1, terms of reference should be developed with aism and goals of the assessment including the application (to whom?; or message to scientists, to regulators, to waste management, to health, etc.)

• Other (please explain)

  Expert 6

  To move from a 'screening level" risk assessment to a more robust approach, all the above methods should be considered, where sufficient mechanistic data become available. Goodrum et al 2020 mentions a "hybrid integrated addition process" that could be explored. Goodrum states "...This approach applies concepts of both similar and independent MoA in that when there is a common apical endpoint, the multiple similar groups of component chemicals can have separate dose-additive assessments that are then combined via response additional into overall probabilistic risk estimates".

• Use of a relative potency approach Assume dose additivity

  Expert 3

  I suggest much the same recommendations as above, (see response to 1.6), with a couple of additional caveats. I suggest a thorough examination of the relative potency (RPF) approaches and results published by Goodrum (2020), ,and (Bil et al 2020). These RPF appear to be reasonable based on the data and methods available at the time of writing.

  Any description of RPF should explain the limitations and the range of applicability. Unlike the toxicity equivalence factors (TEF) developed for dioxin-like compounds, the RPFs are grounded in an understanding of particular MOA for a specific adverse outcome. By contrast the TEF for dioxin-like compounds are based on the hypothesis that a singular molecular initiating event (MIE) is involved in all adverse outcomes observed for this group. The RFPs for PFAS should be limited to consideration of only the toxicological effect / organ toxicity on which the potencies were based. Goodrum et al (2020) present a good argument for separating PFOS and PFHxS from their other studied PFAS based on the dissimilarity of dose response curves.
  The Bil et al. (2020) analysis indicates that the RPFs can be ranked as a function of chain length, with the greatest potency seen in those PFAS with between 7 and 12 carbon chains. This grouping could be useful in establishing the conceptual model for a PFAS risk assessment.

• Other (please explain)

  Expert 11

  In a perfect world a biologically based dose response model would be created for each substance being assessed. The model should be based on a quantitative AEP-AOP network. This network would be validated using whole mixture test data.

  More realistically, I would like the collection of data on MIEs and early steps in the PFAS AOPs. Such data would determine if there is a potential for dose additivity and could provide a basis for relative potency factors. I would also focus on the chemicals that are known to drive mixture toxicity in exposed populations. Studies of additive models of cumulative exposures in populations of humans or environmental receptors have shown that one or two chemicals typically drive mixture risks (see the literature on the Maximum Cumulative Ratio).

• Use a whole-mixtures approach Use of a relative potency approach Assume dose additivity Assume response additivity

  Expert 4

  The key word in the question is "robust." In general, a robust risk assessment requires robust data, which are not available for most PFAS. In the absence of data, the reliability of any of the above approaches is unknown. Probably the best that can be done in that situation is to use all of the approaches and assumptions, and provide risk managers with the range of possible risks generated, and the degree of confidence in the estimates.

• Other (please explain)

  Expert 5

  Because mixtures may contain several groups distinguishable on the basis of target organ/tissue/system and/or mode of action, an approach not specifically identified here should be considered. This hybrid approach is called by some “integrated Additivity” and it has been presented in the context of risk assessments for mixtures of (often poorly characterized) drinking water disinfection byproducts (US EPA, 2003) and described by Teuschler et al. (2004). Integrated addition combines aspects of dose addition and response addition. The mixture components are first grouped according to toxicological similarity (affecting the same organ/tissue/system), then components within these groups are assessed to determine whether they act via different modes of action. If so, they are segregated further. If dissimilarity of MOA cannot be determined, then the group of toxicologically similar components is not further divided (into distinct MOA groups). Once grouped, chemical components of each of the final groups are assessed via an RPF approach, and the risks from each group are combined via a response addition across the whole mixture. Notably, for some group risks not attaining a pre-set level of risk, risk for those groups may be eliminated from further consideration. An important aspect of the USEPA’s approach is that it is based on internal doses, rather that environmental concentrations or applied doses. In the case of drinking water disinfection byprodutcs, this was necessary to address a multi-route exposure scenario (oral, dermal and inhalation). However, the prescription of using internal doses (e.g., absorbed doses, tissue concentrations) should apply to mixtures risk assessment for PFAS chemicals, as well, given their substantially divergent bioconcentration factors, octanol-water partition coefficients and biological half-lives.

  Teuschler, LK; Rice, GE; Wilkes, CR; Lipscomb, JC; Power, FW. (2004). A feasibility study of cumulative risk assessment methods for drinking water disinfection by-product mixtures. J Toxicol Environ Health A 67: 755−777. http://www.ncbi.nlm.nih.gov/pubmed/15192867

  U.S. EPA (Environmental Protection Agency). (2003). The feasibility of performing cumulative 30 risk assessments for mixtures of disinfection by-products in drinking water (pp. 435). (EPA/600/R-03/051). Cincinnati, OH: Office of Research and Development, National Center for Environmental Assessment. https://cfpub.epa.gov/ncea/risk/recordisplay.cfm?deid=56834