In addition to these grouping categories being used for testing purposes, could these categories also be a useful starting place for supporting a mixtures risk assessment (e.g., risks can be estimated for each subgroup using a sentinel compound within each subgroup)? Note, this suggests that risks should NOT be assumed to be similar across terminal categories, but if sufficient information is available on at least one PFAS within a terminal category, risk for the entire category may be estimated. (Please explain your answer).

Answer Explanations

• No

  Expert 6

  If I have understood the question properly, what is being advocated is the development of group risk profiles based on a sentinel compound within that group. I would not support this approach. My preference is to develop sufficient toxicity data on mixture components to move towards a TEF or HI approach to estimating the group risk.

• No

  Expert 5

  If the strategy groups together into subgroups chemicals that affect the same organ or tissue, then dose addition can/should be applied. If it becomes known that there is dissimilarity in affected tissue(s) among components in the subgroup, then an alternate or additional subgrouping may be necessary. We need to start somewhere, and so the proposed grouping strategy represents a testable hypothesis for "the proposed grouping strategy is sufficient". SO my response of "No" is not set in stone, but represents "Not likely". Further, a "Yes" response was considered, with the caveat that it would have represented "Yes, but not very supportive. These points should be considered, and I was reluctant to "Not answer".

• I cannot answer

  Expert 4

  You need to have not only sufficient information on at least one PFAS in a category, but enough reason to expect that other PFAS in that category will behave similarly. Furthermore, you need at least some understanding of relative potencies of group members. The "one PFAS" you have info for needs to be representative of the class, and I don't think we have enough info right now to establish that most of the time.

• No

  Expert 1

  Is there data suggesting that these categories and physico-chemical properties are related to mechanism of action or potency? My guess is that there is very little data to support this. While I agree that a sentinel compound is useful in the absence of dose-response toxicity data for some chemicals, I believe that we should be more conservative and assume similar modes of action across terminal categories if we do not have data proving otherwise. When I saw 24-month carcinogenicity testing in the strategy, I though to myself: "Are we really going to wait 2-3 more years before we start addressing PFAS mixtures?" We are already way behind in terms of addressing elevated exposures through drinking water in dozens of communities, we need an approach that can be applied now while we generate additional data, and eventually be able to conduct a robust risk assessment (with groupings based on mode of action). With the EPA goal to release a validated method for total adsorbable organofluorine measurement in the Fall 2021, we should move forward with an AOF-based approach (using a sentinel compound for all PFAS) which will allow screening water systems and prioritize those that need filtration.

• Yes

  Expert 3

  I see no reason why these groups would not be useful as a starting point for mixtures assessment.

• Yes

  Expert 9

  This appears to be a reasonable approach to begin to understand how different characteristics of PFAS, as defined by sub-group categories, may interact as mixtures to induce toxicity. However, this approach also requires information on the relative contribution of individual constituents to a given mixture to ensure that the mixture is relevant to the human exposure experience. This is a well-acknowledge challenge of mixtures risk assessment. In the absence of environmental mixture information or a whole mixtures approach, this approach is probably better than doing nothing.

• No

  Expert 10

  As I said in response to charge question 4.2., the chain length cut-off needs rethinking, including a consideration of how perfluoroalkylether acids (e.g. HFPO-DA) can be compared to PFAAs. I can support a structurally-based grouping approach (as suggested) but this would have to much more carefully thought through than is currently presented. For example, I'd group PFAA precursors together with PFAAs (groups 2,3,5,6) because the primary health concern is with the PFAAs and the precursors will degrade/metabolize and form the PFAAs in the environment or within biota/humans.

• Yes

  Expert 2

  Yes, categories also be a useful "starting place" for supporting a mixtures risk assessment; starting place is the key phrase here.

• I cannot answer

  Expert 8

  Pragmatically, one could use these categories as a starting point. But beyond that the utility really depends on the purpose of the risk assessment. If we are talking about drinking water, then the categories should be limited to those that are present in drinking water. I would need to see how consistent the exposure, pk and toxicity data are for at least 2 members of the category (preferably at the low and high end of the category) to determine whether one could use data to assess the entire category. Risk assessment is an iterative process. These categories can be used to begin with, but are likely to be modified as the risk assessment proceeds for any number of reasons including new toxicity data, new exposure data etc.

• I cannot answer

  Expert 11

  This is a plausible hypothesis, but it needs be tested. It is not clear that trends in quantitative or qualitative toxicity will follow the measures of chemical similarity used to create the categories. What you need to see is empirical evidence of trends in toxicity that are a function of structure (or structure-based properties). We will not have that until the studies are complete. Two other problems with this approach are: 1) data are only generated on 24 of the 56 groups needing data, so you don't have the data to perform this analysis for all PFAS substances, and 2) the selected chemical may not be a good sentinel chemical for the group. A sentinel chemical is the chemical that is the most toxic of the group of chemicals that it represents. The tested chemicals may or may not be the most toxic members of their groups. The results of the testing program, however, can be used to answer the question "How conservative is a screening approach be that applies a worst-case PFAS toxicity to total absorbable/extractable organic fluorine?" (see Question 2.13). The data will show the range of toxicities that exist and how much lower the toxicity could be for a substance or groups of substances than the screening estimate of toxicity would indicate.