Please identify the primary strengths and weaknesses of the USEPA (2021) definition

EPA's definition is pragmatic in narrowing down a long list of PFAS of concern. Listing nine primary structural categories for risk assessment is a good approach to start assessing risks of this class of chemicals; this is in line with most of us discussing in the previous rounds about creating 'subgroups' of PFAS for risk assessment. However, a major weakness is removal of chemicals for which vapor pressure cannot be determined renders this definition less useful in identifying a most of the PFAS in commerce. The rationale for selecting vapor pressure as a parameter for inclusion/removal of PFAS is not clear. In fact, most PFAS escaped regulatory scrutiny in the past because they were portrayed/deemed as polymeric substances (high molecular weight). Breakdown of polymers, resulting in small molecules that are persistent and bioaccumulative, was not thought out in those days for regulatory assessments. In fact, even most commonly studied PFAS such as PFOS and PFOA are expected to be present in ionic form in the environment and human tissues, because of their low pKa values. Thus, real VP for most PFAS cannot be measured even under laboratory conditions. Nevertheless, EPA's definition is one way to start assessing risks of PFAS, under various subgroups as indicated under nine structural categories, but it should also state that this definition is not exclusive and should be made more inclusive, as new information becomes available for inclusion of a new structural category or a chemical property that will fall in the category of PFAS.

Perhaps the primary strength is that the definition is not overly restrictive. While this opens the door for the inclusion of "any number" of fluorinated compounds or a remarkably wide set of molecular characteristics, the assessment will (or should be) restricted to only those chemicals that comprise the exposure suite. It may be expected that such a broad definition will result in the inclusion of a remarkably varies set of mixture constituents, which may stimulate the conception and development of additional mixtures risk assessment methods/procedures. .

Perhaps the primary weakness is that the definition does not restrict the definition to two or more fully fluorinated carbons in the molecule. The present concern for these chemicals is based on toxicity associated with chemicals having more than 2 fully fluorinated chemicals. The PFAS issue has not arisen from pharmaceuticals, therapeutics, etc, but from a very short list of fluorinated chemicals having a relatively consistent set of molecular characteristics, and are derived from a relatively consistent manufacturing process.

The main strength is that it is prescriptive, and everyone will be bound by it. That is a good thing. However, the EPA definition excludes -- in my opinion inappropriately unnecessarily -- a variety of substances that could be of concern, e.g. pharmaceuticals. This is a key weakness.

This definition seems broad enough to include PFAS with cyclic structures and fluorotelomeres (which were not included in some of the definitions we evaluated previously). I would tend to agree with this definition.

No definition or grouping based thereupon will be perfect. This proposal has the advantage of being well-defined and logical. The rationales for grouping are clearly stated. I found the provided discussion and example of tertiary structural categories to be clear and reasonable. It certainly seems to be inclusive, given that more than 6500 PFAS met the initial criteria and were amenable to grouping.

Note that the Problem Formulation for the USEPA strategy is entirely different than that we have been considering for this exercise. USEPA is seeking to write test rules under TSCA based on their strategy. I think it is reasonable to assume that programs under both TSCA and SDWA will use the fruits of the strategy.

A strength of the definition is that it is clear and provides definitive boundaries around what types of PFAS will or will not be included in the "working definition." A weakness of the definition is that it excludes a large number of chemicals that can be defined as PFAS.

One of the strengths of the USEPA (2021) definition is listed above in the charge question. This strength is that many substances are eliminated in the US EPA definition including many pharmaceutical and agricultural chemicals. The number of substances captured by a structural definition increases with the number of fluorinated carbons included in the structural definition. For example, recent searches using the online search tool SciFinder revealed that the number of substances (CAS numbers) including CF2 alone was 20 772 063 (4 430 726 commercially available. 16 341 337 not commercially available). When the structural fragment was increased to CF2 CF2, 4 667 078 substances were identified (266 086 commercially available, 4 400 992 not commercially available). And finally, for a structural fragment of CF2 CF2 CF2, 1 188 469 substances were identified (31 393 commercially available, 1 157 076 not commercially available). Similar numbers were obtained with another online tool, PubChem. While many of the substances are probably not used in high tonnages (and maybe simply listed in patents), the high number of the substances with one CF2 fragment is daunting.

Some substances that meet the OECD PFAS definition are mineralizable which is also problematic for implementing a class regulation of PFAS. While PFAS are known to be diverse in structure and properties, it is often considered that the one commonality is their high persistence due to high strength of the C-F bond. However, substances with a CF3 group attached to a heteroatom (O, S, N) are known to be degradable (fully mineralizable), but are considered to be PFAS under the OECD definition. These substances appear not to be included under the proposed US EPA definition.

A weakness is that some problematic substances will not be included in the US EPA definition, including all fluorinated polymers. Many low molecular PFAS are released during the lifecycle of fluoropolymers and the stability of side-chain fluorinated polymers is uncertain. Side-chain fluorinated polymers could be precursors to PFAAs through cleavage of the side chains and they also contain residual PFAAs, which are known to leach out of the products.

OPPT has a lot of experience in binning large lists of chemicals. In this particular exercise, the emphasis has focused almost entirely on chemical structure. OPPT has binned the chemicals into these categories for the purpose of generating some basic toxicity information. The strengths of their methodology are basically that they have eliminated chemicals that are not under the authority of TSCA, those that are not currently commercialized (although they don’t explicitly state that), large polymers, and chemicals that are unlikely to cause toxicity via the perfluoro moiety. This is logical. They have further subdivided the categories into those chemicals with more or less than 8 carbons. This is logical IF all the categories follow a similar pattern of toxicity as is observed with PFAAs. What seems to be lacking is a crosswalk with chemical use or some other measure of potential exposure. I am guessing that such a crosswalk would result in a much shorter list. However, given all that, this seems to be a logical approach for the PURPOSE of gathering some toxicity data. It should be noted, however, that TSCA section 4 test rules do not have a good track record in generating data in a timely fashion!

Publication of the US EPA strategy appears to overlap (and possibly overtake) the work of this Panel and confirms much of what has been discussed so far. It uses a definition for PFAS that is consistent with views expressed by a majority of this Panel and the sub-groupings suggested by this definition make a lot of sense. It further recommends a tiered testing regime that progresses from basic physico-chemical, metabolic, toxicokinetic and receptor binding properties and to more specialised testing, with tier 3 testing focussed on specific toxicity targets and an ability to develop dose-response relationships. The costs and onus on testing is transferred to industry sponsors. The only potential weakness is the timeframe over which the strategy can be implemented. I suspect this Panel has been looking for a grouping and mixture analysis strategy that can be implemented more quickly.

Looks difficult would be very confusing:
The headings must have the same type of definition: volatile = vp>100 mg HG (note: SI units must be used; thus, hPa, mm Hg is not acceptable) and non-volatile = vp<100 mm Hg. These dividers cannot be associated with C8 chains; for example C8 compounds, like PFOS and PFOA, have been identified in ambient air at all parts of the world.
In addition, vp is strongly temperature-dependent; thus, an additional parameter needs to be defined.
Using the C8 chain length would be helpful only for chemical characterization; hen, the third column is not needed. Further, the structure must be known.

The same inconsistency applies to the 9 rows, where the definitions are too vague and overlap. Some are defined using characteristic chemical groupings but inconsistent since is not clear where the group is placed (alkane, sulfonamide), other are acids only (what is with salts, ethers, etc?), precursors and derivatives does not have any chemical description and therefore, cannot be further used.

It shall be noted that the classification cannot be used for toxicity purposed nor for uses (production or sources).

Addressing this question is beyond my expertise. But, I was pleased to see that after the definition was used the proposed process immediately divided the PFAS into subcategories with more homogeneous properties.