How should PFAS be defined for human health risk assessment purposes (e.g., derivation of drinking water criteria)? You may refer to a PFAS definition from a specific agency or publication, or provide your own definition

PFAS should be defined as a class of chemicals of commercial significance rather than drinking water or health criteria based definition. Like chlorinated pesticides or PCBs, PFAS is a class of chemicals and that is how they should be defined and the definitions made in Buck et al. 2011 is appropriate.

For risk assessment purposes one needs to start with a definition that is specific enough to target current substances of interest, while at the same time being broad enough to capture new substances. The definition should not be influenced by the availability of toxicological data for specific substances. The OECD (2021) definition is straightforward and meets the two above objectives.

OECD (2021) compiles thoughts from a wide range of scientists, with attention aimed at “PFAS” definition. Their definition is favored, and it is presented on page 21 of the report. This definition is agnostic to number of carbons, ring structure, branched or straight carbon chain, head or tail groups and the difference between “per” or “poly” with respect to the entire molecule.

“PFASs are defined as fluorinated substances that contain at least one fully fluorinated methyl or methylene carbon atom (without any H/Cl/Br/I atom attached to it), i.e.[,] with few noted exceptions, any chemical with at least a perfluorinated methyl group (-CF3) or a perfluorinated methylene group (-CF2-) is a PFAS.” - OECD, 2021.

According to Kwiatkowski et al. (2020), when chemicals have similar molecular structures, environmental properties, and/or biological hazards, a class-based approach can be an effective means of reducing adverse public health effects. Kwiatkiwski et al. (2020) goes on to acknowledge that while a class-based approach can challenge the traditional one chemical by one chemical approach, the extreme persistence of the vast majority of PFAS and the potential for harm from the many substances that are defined as PFAS (regardless of the definition used, the number of individual PFAS is in the thousands) warrants a class-based approach. A class-based approach can be more effective, efficient, and protective of public health. However, it also may be possible to identify multiple sub-groups of PFAS within the class, to link health risk assessment approaches directly with PFAS measured in environmental matrices that are derived from specific sources, for example industrial or aqueous film-forming foam (AFFF) activities. Such sub-groups could be derived not only with information from molecular structures, environmental properties, and/or biological hazards in mind, but with information on environmental occurrence to maximize the effectiveness of public health protective measures.

Clear definitions and strict application of established terms need to be maintained, these include: compound identity, physical chemical parameters for each compound, differentiation between measured and estimated properties. Not to start with already biased shortcuts.
Depends on constituency and definition.
As to constituencies and application: at least in the EU for POPs, food agencies do not include drinking water in human exposure (as food), so that they cannot regulate under the same law and the laboratories are different. Can be seen by the approaches made by DG Santé (food, human health; EFSA) and DG Environment (Chemicals agencies, Stockholm and Basel Conventions).
Drinking water is certainly not enough for human health risk assessment.
Under the chemicals and waste conventions, PFAS are defined as
• Perfluorohexane sulfonic acid (PFHxS), its salts and PFHxS-related compounds, defined as any substance that contains the chemical moiety C6F13SO2- as one of its structural elements and that potentially degrades to PFHxS (UNEP, 2018a, b)
• PFOA-related compounds can be substances containing a perfluorinated alkyl chain with the formula F(CF2)n- (n=7 or 8) and that is directly bonded to any chemical moiety other than a fluorine, chlorine or bromine atom or a phosphonic, phosphinic or sulfonic group (UNEP, 2017)
• PFOA-related compounds are any substances that degrade to PFOA, including any substances (including salts and polymers) having a linear or branched perfluoroheptyl group with the moiety (C7F15)C as one of the structural elements (UNEP, 2017)
• PFOS all substances that contain one or more C8F17SO2-groups
In the Basel technical guidelines, PFOS related compounds are defined to contain the moiety C8F17SO2 or C8F17SO3 and PFOA the moiety (C7F15)C as structural elements (UNEP, 2021b). However, the definition for PFOA is not correct since would not cover 8:2 FTS, which degrades to PFHxA. Note: According to decision SC-9/12, the characteristic moiety for PFOA-related compounds is (C7F15)C; however, it shall be noted that (C7F15)C as a degradation product (after defluorination and oxidation to carboxylic acid) can only be generated from compounds with the moiety (C8F17) (UNEP, 2021a).
For any fate and exposure assessment, differentiation into non-polymeric and polymer fluoro compounds and the latter ones differentiated into side-chain polymers and fluoropolymers will be necessary (Henry et al., 2018). Analytical approaches can help in the screening (Fiedler et al., 2021).

References:
Fiedler, H., Kennedy, T., and Henry, B.J. (2021). A Critical Review of a Recommended Analytical and Classification Approach for Organic Fluorinated Compounds with an Emphasis on Per- and Polyfluoroalkyl Substances. Integr Environ Assess Manag 17, 331-351

Henry, B.J., Carlin, J.P., Hammerschmidt, J.A., Buck, R.C., Buxton, L.W., Fiedler, H., Seed, J., and Hernandez, O. (2018). A critical review of the application of polymer of low concern and regulatory criteria to fluoropolymers. Integr Environ Assess Manag 14, 316-334.

UNEP (2017). Risk profile on pentadecafluorooctanoic acid (CAS No: 335-67-1, PFOA, perfluorooctanoic acid), its salts and PFOA-related compounds. UNEP/POPS/POPRC12/11/Add2.
UNEP (2018a). Annex I: Decisions adopted by the Persistent Organic Pollutants Review Committee at its fourteenth meeting
1. POPRC-14/1: Perfluorohexane sulfonic acid (PFHxS), its salts and PFHxS-related compounds 2. POPRC-14/2: Perfluorooctanoic acid (PFOA) its salts and PFOA-related compounds
3. POPRC-14/3: Evaluation of perfluorooctane sulfonic acid, its salts and perfluorooctane sulfonyl fluoride pursuant to paragraphs 5 and 6 of part III of Annex B to the Stockholm Convention, POPs Review Committee, ed. (Rome, Italy).
UNEP (2018b). Perfluorooctanoic acid (PFOA) its salts and PFOA-related compounds, POPs Review Committee, ed. (Rome, Italy).
UNEP (2021a). Guidance on best available techniques and best environmental practices for the use of perfluorooctane sulfonic acid (PFOS), perfluorooctanoic acid (PFOA), and their related compounds listed under the Stockholm Convention.
UNEP (2021b). Technical guidelines on the environmentally sound management of wastes consisting of, containing or contaminated with perfluorooctane sulfonic acid (PFOS), its salts and perfluorooctane sulfonyl fluoride (PFOSF) and perfluorooctanoic acid (PFOA), its salts and PFOA-related compounds. In Technical guidelines, Addendum, Conference of the Parties to the Basel Convention, ed.

All current definitions of PFAS (Buck et al. 2011, OECD 2021, etc.) are, in my opinion, too broad for use in human health risk assessment purposes. I would suggest limiting risk assessment to a few subgroups of PFAS, namely the perfluoroalkyl acids (PFAAs) and the perfluoroalkyl ether acids (PFAEAs, e.g. hexafluoropropylene oxide dimer acid (HFPO-DA)). The reasoning for limiting the scope of PFAS for human health risk asessment to PFAAs and PFAEAs is: 1) because these are the few PFAS for which toxicity and toxicokinetic data are available and 2) there are analytical methods for most substances in these subgroups of PFAS. Some consideration may also be necessary for the precursors to PFAAs/PFAEAs, but it's challenging to come up with a practica way of including them in risk assessment. Cousins et al. (2020) has suggested applying the total oxidizable precursor assay (TOPA), but they also points out the many current drawbacks.

I would suggest a definition for PFAAs as follows (Buck et al., 2011): the perfluoroalkyl sulfonic acids (CnF2n+1SO3H, n ≥ 4, PFSAs) and the perfluoroalkyl carboxylic acids (CnF2n+1COOH, n ≥ 3 and, PFCAs) and their corresponding anions. Maybe an upper limit of carbons should also be set for PFAAs; e.g. 10 for PFSAs and 14 for PFCAs. For the PFAEAs, a simple structural definition covering all relevant substances requires more thought because of the complexity of these structures. It could be simplest to just name those individual substances that are included in the grouping approach based on the availability of analytical methods and toxicity data.

Although I would suggest a limited number of PFAS to be grouped for human health risk assessment purposes, (e.g. for derivation of drinking water criteria), I have quite different opinions on how PFAS should be grouped for chemicals management purposes. I would instead suggest a broader restriction of PFAS in society going forward as being investigated in the European Union. The reasoning for these different grouping approached was outlined in Cousins et al. (2020). Decisions for how to group already emitted PFAS for the establishment of drinking water guidelines or environmental cleanup levels will have profound impacts on enforcement including costs and resource needs. It is probably therefore necessary, in resource-constrained settings, to more strictly prioritize cleanup levels on the basis of established toxicological risk for limited subgroups of PFAS. We should at the same time be cautious about the continued use and emission of highly persistent substances such as PFAS. It is concerning that we lack toxicological data for most of the PFAS currently used.

Finally, although I believe that human health risk assessment should be curently limited to those few substances for which there are exposure and toxicity data available (e.g. PFAAs and PFAEAs), in the future more data will become available and the scope of risk assessments can be expanded. Within the US, the focus of the US EPA and the US National Toxicology Program (NTP) is on developing high-throughput testing methods for PFAS to expand the scope of risk assessments. They recently selected 150 PFAS (expanded from 75) for high-throughput toxicity testing (e.g. in vitro assays) for multiple endpoints (see e.g. Patlewicz et al., 2019).

References cited:

Buck, R.C., et al. (2011) Perfluoroalkyl and Polyfluoroalkyl Substances in the Environment: Terminology, Classification, and Origins. Integrated Environmental Assessment and Management, 7, 4, 513-541.

Cousins, I.T. et al. (2020) Strategies for grouping per- and polyfluoroalkyl substances (PFAS) to protect human and environmental health. Environ Sci-Process Impacts, 22, 1444-1460.

OECD (2021) Reconciling Terminology of the Universe of Per Recomme ndations and Practical Guidance Publishing, Paris. , and Polyfluoroalkyl Substances: OECD Series on Risk Management, No. 61, OECD Publishing, Paris.

Patlewicz, G.; Richard, A.M; Williams, A.J.; Grulke, C.M; Sams, R.; Lambert, J.; Noyes, P.D.; DeVito, M.J.; Hines, R.N.; Strynar, M.; Guiseppi-Elie, A.; Thomas, R.S. (2019) A Chemical Category-Based Prioritization Approach for Selecting 75 Per- and Polyfluoroalkyl Substances (PFAS) for Tiered Toxicity and Toxicokinetic Testing, Environ. Health Perspect., 127 , 5.

The term “PFAS” has been used inconsistently through the years and this inconsistency has led to a lot of concern over the potential hazards associated with exposure. The toxicological database is quite extensive for the two legacy PFAS chemicals, PFOS and PFOA, and the concerns associated with these two compounds have often been extrapolated (wrongly so) to the broader PFAS world.
Buck et al 2011 provided an extensive overview of this broad series of compounds and offered harmonized terminology for the broad definition, as well as for the subcategories. This has recently been revised by the OECD (2021) and they have defined PFAS broadly as “fluorinated substances that contain at least one fully fluorinated methyl or methylene carbon atom (without any H/Cl/Br/I atom attached to it), i.e. with a few noted exceptions, any chemical with at least a perfluorinated methyl group (–CF3) or a perfluorinated methylene group (–CF2–) is a PFAS.” However, OECD (2021) notes that this definition “is based only on chemical structure, and the decision to broaden this definition compared to Buck et al. (2011) is not connected to decisions on how PFASs should be grouped and managed in regulatory and voluntary actions.” The OECD (2021) provides excellent examples in Table 1 of the correct and incorrect uses of specific PFAS terminology.

Human health risk assessment is generally conducted to determine if there is a potential concern for an adverse health outcome associated with a particular exposure, and if so, what can be done to reduce the exposure and therefore the risk. From a regulatory point of view, the exposures must be something that can be managed. It is therefore crucial that one define the specific exposure scenario to be addressed and limit the definition of “PFAS” to that context. Simply saying that there are thousands of “PFAS” out there only leads to unnecessary hysteria, and is not useful for developing a scientifically reasonable approach to the “mixtures” issue, or risk management.

The first step in defining “PFAS” for the purposes of human health risk assessment is to define the specific question or problem that is being addressed; this is known as problem formulation. The definition of “PFAS” has to be put into the context of the specific risk assessment question being addressed. If the question is “what are the potential health effects associated with levels of “PFAS” in drinking water?”, then one first has to know which PFAS are present in the drinking water. For example, the EPA collects data every 5 years on 30 potential contaminants in drinking water through the Unregulated Contaminant Monitoring Rule. In the last cycle which monitored drinking water facilities from 2013-2015, EPA collected data for six PFAS including PFOA, PFOS, perfluorononanoic acid (PFNA), perfluorohexane sulfonate (PFHxS), perfluoroheptanoic acid (PFHpA) and perfluorobutanesulfonic acid (PFBS). These data then can inform the definition and “world” of PFAS to be considered in the assessment. This question might be further refined to include specific age groups, such as what is the risk to 2 year old children from the drinking water. Similarly, if a novel “PFAS” is being considered, then one would need to look at potential uses, potential exposures through those uses, potential release to the environment, toxicity etc. all of which may influence how one approaches the assessment process and eventually risk management.

In summary, the specific definition of “PFAS” should be dependent upon the specific problem formulation statement, and this should be narrow enough such that exposure reduction measures can be implemented if necessary.

A difficult question! Definitions based on chemical structure (e.g. Table 1 in PFAS Background document) should capture all the legacy PFAS of concern and relevant PFAS replacement chemicals, but such a grouping would inevitably include PFAS chemicals that may not share common toxicological properties, environmental handling and metabolic/clearance characteristics. The simpler definitions in Table 1 (e.g. Buck et al 2011, OECD 2018) are probably best.

Please note that my answers are all somewhat informed by the question I propose in the last charge question: "What is the purpose for which the risk assessment being done?"

That having been said, the definitions in the document for review seem OK to this non-chemist. I like this one for simplicity: ITRC, 2021 PFAS, Naming
Convention Considerations3
“PFAS include only fluorinated aliphatic (carbon chain) substances.
PFAS do not include fluorinated compounds that contain aromatic
(carbon ring) features in their structures (for example, active
pharmaceutical ingredients, crop protection agents, or
chlorofluorocarbons [refrigerants]).”

Based purely on some prior experience with PAH, it might be useful to include only those PFAS with 3 or more carbon atoms.

I am not an expert on PFAS toxicity or exposure (seen my response to question 2.1) and do not have specific recommendations for defining PFAS. However, here are some general thoughts on defining groups of chemicals.

The definition of a group of chemicals is fundamentally an institutional decision that is taken by an organization in order to facilitate meeting the needs/goals of the organization. It can be based on technical information but also consider economic and management considerations. As such, the act of defining a group is a societal act not a scientific act. It can be criticized as not allowing the achievement of the intended goals, but it cannot be overturned strictly on technical findings or consistency of arguments.
If one starts with the assumption that the purpose of the definition of PFAS should be to facilitate the protection of public health, then the definition of the structures that are included in PFAS would need to aid in the management of the assessments, assist in defining the hazard of the chemicals, and in the characterization of exposures to the compounds that pose a hazard. The management of the chemicals is assisted by rules that are unambiguous in determining which chemicals were included in the group (objective) and have a clear justification (transparent). The definition should simplify (or at least not complicate) the exposure assessment, hazard findings, and risk management processes.
For example, including more chemicals in a group (to assure that the additional chemicals will be considered in the assessment) is a conservative assumption. But such a definition may increase the cost of measurement (and thus decrease the number of samples that could be analyzed on a given budget), increase the complexity and cost of risk assessment, include populations and sources of uncertain concern, and reduce/distract focus on chemical exposures expected to be of the greatest concern. Increasing the scope of the definition of PFAS will also increase the likelihood that PFAS chemicals will operate by multiple modes of action and will need to be subdivided into multiple assessment groups.
The selection of chemicals for a group can be defined by chemical characteristics (PAHs, dioxins, PCBs) by the ability to cause a specific toxicity endpoint (endocrine disrupters) or a toxicological mechanism (acetyl cholinesterase inhibition). If it is defined by the chemistry, then often there is an archetypal compound(s) that defines the toxicological hazards of the group (2,3,7,8 TCDD or benzo(a)pyrene) and the physical and chemical properties of group members. Inclusion in the group is then defined as all substances that have the specific chemical, physical, and toxicological properties of the archetypes.
In the case of PFAS as I understand it the archetypical chemicals are PFOA and PFOS, the key properties of these chemicals are:
• A portion of the molecule is dominated by CF3 and CF2 moieties and result in that portion of the molecule having strong hydrophobic and lipophobic properties.
• The molecules are highly persistent in the environment.

• Very long half lives in some but not all organisms.

But there is a confusion in that PFAS has also been defined as including perfluorinated oligomers and polymers that do not have a polar moiety(ies). This was justified because high volume perfluorinated polymers contained significant levels of PFOS. These chemicals are not likely to have the same exposure potential, ADME, or toxicodynamics as PFOA and PFOS. As a result, definitions of PFAS that include perfluorinated polymers should immediately bifurcate into those with a hydrophilic moiety and those that do not.
Based upon the definitions of the structures that flow from the properties, rules can be defined that determine the inclusion and exclusion of specific chemicals from PFAS. For example, properties like specific toxicity endpoints or a vapor pressures, solubility, and bioaccumulation potential are not essential and can vary across PFAS. Other moieties are allowed in PFAS when consistent with the above criteria.
Whatever the rules are for defining PFAS, the rules should link back to the properties of the archetypical chemicals and an acknowledgement of the quality of the data that supports a linkage. When the linkage is based on the absence of data (we don’t know enough to exclude this type of molecule from the category of PFAS) rather than based on an empirical finding (we have data that justifies inclusion) expect that the status of the chemical will change as more data are developed.
In the case of PFAS, the relevant mode of action and their relationship to the above properties are not yet defined (unlike OP pesticides, dioxin-like chemicals, and PAHs) and there is evidence that multiple modes of action exist. It is also interesting that the basis for the interest in the chemicals are driven as much as the exposure properties of the molecules (forever chemicals that may have large bioaccumulation factors) as much as the toxicology of the chemicals. This focus on exposure related properties is not unique it also occurred with the halogenated aromatic hydrocarbons. In these cases, multiple modes of action would be expected to occur in the group (i.e., dioxin-like PCBs and non-dioxin-like PCBs).

The definition should be as broad as possible to encompass all current and future PFAS. My understanding is that the OECD definition is the most flexible, although I have not tested it against different groups of PFAS.