Do you agree with this proposed grouping strategy for the purposes of defining PFAS categories that are most likely to share similar human health risk (toxicity and exposure) profiles?

Answer Explanations

• I cannot answer

  Expert 2

  Vapor pressure based classification will not inform additivity or mechanisms of toxicity. Similarly, the toxicity is not related to structure of these chemicals and many large molecules may fall under polymer category and escape from scrutiny. Risk cannot be assumed similar across groups.

• Yes

  Expert 5

  The predominant strength is that is objective and transparent. Developing a grouping strategy is necessary, and we seem to have agreed that grouping based on health endpoint or mode of action is optimal, yet perhaps the least likely to be successfully implemented (owing to lack of information). The predominant weakness of this approach is that it does not address the health endpoint - and perhaps it does not/is not based on health endpoint is exactly because such data are likely to be unavailable. To the extent that this grouping strategy is workable, optimal or not, it is workable. It can be followed. To that extent, "Yes".

• I cannot answer

  Expert 4

  Since I disagree with the selected definition, it's difficult to answer yes or no. Notwithstanding my concerns about the definition, this strategy is as good as any other in the context of the available, very limited data. You have to start somewhere.
  I note that phys/chem attributes may predict likelihood of exposure better than they predict human toxicity. Disclaimer: I don't have any experience with the Morgan fingerprinting method, so don't know if it is appropriate or not in this context.

• No

  Expert 1

  I am not convinced that the parameters used to create these categories are necessarily tied to toxicity and exposure. PFAS in different categories could have additive effects, and possibly be found in combination in contaminated drinking water. The risk characterization is not presented in the document, so it's impossible to comment on the way mixtures will be handled. In addition, it looks like only 29 PFAS will be measured as part of UCMR 5 - this will clearly overlook some PFAS that may be present in drinking water. Finally, some categories will require additional toxicity testing, which will be time-consuming and considerably delay risk assessment and exposure mitigation strategies in contaminated communities. Overall, I believe the grouping strategy based on structural/physchem parameters leads to too many categories, which may separate PFAS which share a common mechanism/mode of action.

• I cannot answer

  Expert 3

  There is some information to justify an initial assumption that PFAS with similar adverse outcomes can be expected based on structural grouping (as was discussed in rounds 1 and 3). It is also reasonable to assume some degree of similarity of exposure profile among members of the same group. Testing the hypotheses will be challenging, but it would appear do-able if high throughput tests can be validated and used.

• No

  Expert 9

  I'm not sure if I agree with the idea of sub-grouping, but if the Agency is going to pursue a grouping approach, parts of this approach appear reasonable and parts appear to be based on assumptions that will leave out important contributors to the PFAS exposome. First, a definition that excludes pharmaceutical and agricultural PFAS potentially ignores the contribution of these substances to the global PFAS burden and thus potential contributors to the human PFAS exposome. Further using vapor pressure to eliminate certain PFAS (i.e., fluoropolymers) also ignores that these PFAS also contribute to the human PFAS exposome if the life-cycle perspective is considered. But, for testing purposes, for better understanding exposure potential and toxicity, these approaches are somewhat defensible.

• No

  Expert 10

  I don't agree with excluding fluorinated polymers because there are huge human health concerns related to the lifecycle of these substances.

  I don't agree with dividing substances into long-chain and short-chain based on 8 carbon atoms (presumably 8 fluorinated carbon atoms?). It is well known that perfluorohexane sulfonate (PFHxS), which has only 6 fluorinated carbons is slowly eliminated from humans and is grouped together with PFOA, PFNA and PFOS for human risk assessment in the recent European Food Safety Authority (EFSA) guideline in Europe. I'd prefer not to have such a simple cut-off because surprises are likely. For example, HFPO-DA (hexafluoropropylene oxide, also known as GenX) has recently been found to be more of a human health risk than previously thought in the US EPA's own assessment. If the US EPA want a cut-off then it probably needs to be more carefully thought through. For example, a cut-off of 6 fluorinated carbons would be appropriate for perfluoroalkane sulfonates, and some consideration needs to be given to how ether groups in perfluoroalkylether acids (e.g. HFPO-DA) contribute to human elimination kinetics and toxicity, etc.

• Yes

  Expert 6

  The US EPA grouping strategy makes sense to me. It includes evaluation of structural characteristics along with consideration of existing toxicity data. Exposure profile data appears to make little or no impact on the grouping strategy and I have no problem with that.

• I cannot answer

  Expert 8

  The grouping strategy is certainly a starting point. But beyond that, I honestly can’t answer this question without seeing the toxicity information that OPPT currently has for members of the categories, or the data on potential uses of the chemicals. Exposure may in part be related to persistence in the environment, but if the uses do not result in release to the environment, then there is little potential for exposure. I would also anticipate that there will be further subcategories once toxicity information and pk information are available. For example, simply categorizing PFAAs based on chain length is inadequate, as the MOA for the sulfonates such as PFOS differ from the MOA for the carboxylic acids such as PFOA. Therefore, one could argue that they should be in different subcategories. However, having said this, the strategy used has to reflect the problem formulation. The OPPT strategy was developed for the purpose of generating basic toxicity information. This strategy may need modification for use for other purposes and may need modification once data are collected.

• I cannot answer

  Expert 11

  I think the idea of the grouping strategy was a bit different then this question implies. EPA states "the Strategy aims to identify a representative substance(s) for each chemical category where categories have been constructed to span the landscape of PFAS of interest." I understood this to mean that the tested chemicals needed to be uniformly spread across the chemical structure space of PFAS substances so that if structure-based subcategories with higher toxicity exist they could be identified. I could also speculate that the data set was being designed to support in silico assessments (i.e., any PFAS substance would have several "nearest neighbors" to support a read across). It was not clear to me that the groups were assumed to be similar in toxicity, EPA only states that they are similar in structure. I found the discussion of exposure in this section to be a bit muddled. This exercise was about understanding the range of toxicities across PFAS structures not exposure potential. It was not clear how exposure fit into the groupings, but dividing the PFAS into low and high volatility would separate PFAS with a potential for significant air transport from PFAS that did not have this potential.