Should potential risks/hazards of PFAS be managed as a single homogenous group, or divided into different subgroups?

Answer Explanations

• Other (please explain below)

  Expert 10

  For human risk assessment (e.g. for drinking water guidelines), in my opinion limited subgroups should be considered (see above answer). For assessment of PFAS based on their intrinsic probematic properties (as practised in the European Union), e.g. for restricting the use of PFAS in society going forward, a single group approach can be appropriate based on their extremely high persistence. Uses can be exempted from restriction based on their essentiality.

• Multiple subgroups

  Expert 2

  Toxic mechanisms and mode of toxicity of PFAS are still not known. If there is a single/common toxic mode that can be applicable for all PFAS then it can be grouped as a single group. Toxic mechanisms and toxic potential differ widely among PFAS; bioaccumulation potential and metabolism vary widely among PFAS. Therefore, it is important categorize them based on some common properties, if possible toxic mechanisms. While persistence may be only common criterion many groups think about grouping them all into one category, toxicity, bioaccumulation, metabolism/toxico-kinetics vary among PFAS.

• Other (please explain below)

  Expert 7

  Cannot be mixed. In principle, MUST start with single compound characterization and assessment.
  Hazard is an inherent property of the chemical and risk is a derived parameter and takes dose into consideration (which means toxicity and exposure pathways).
  This can be done on a single compound basis only. If common denominators are found, grouping may be possible.
  Terms such as “subgroup” and “homogenous” need to be defined (chemically, use, mode of action, endpoint, analytical strategy)?
  I would warn on just taking/accepting earlier assessments rather go back to the original literature and review with fresh eyes and from scratch for the purpose of this work.
  In addition, differentiation has to me made between scientific technical assessment and regulatory risk management, which would take a practical approach based on science but not describe single experiments and cannot be used to extrapolate to other single compounds. Example: The TEF for OCDD cannot be used to derive a TEF for a potential new dioxin-like compound like a pentachloronaphtahlene.
  In context to traditional (chlorinated but also brominated) persistent organic pollutants (POPs), PFAS are different due to water solubility (which does not exist for the fluoropolymers) but scientists/agencies keep the other attributes including persistence (and then be inconsistent and refer to precursor compounds and degradation pathways), toxicity, and bioaccumulation.
  With PFAS there is a clear lack of “hard” data. I am not aware that many experiments have been undertaken on the molecule such as exposure to uv light, degradation studies under defined conditions. Relevant concentrations have to be taken into account as well.
  I suggest starting with the two model compounds, PFOS and PFOA, find all necessary hazard information and all necessary toxicological information - facts (not modelling or back-tracing from monitoring observations with bad analytical data).
  With PFOS, there is already the problem with the linear and the branched isomers that cannot be neglected and I wonder what toxicologists have tested or if effects differ based on structure.
  A critical question would be how to handle the statement that "PFOS" is protein-bound" in the (human) body. If not free circulating, no effect?
  Careful reading/formulation of the legal text is necessary to include or exclude compounds.
  TOF, TOP, EOF outcomes cannot be used for risk assessment.

• Multiple subgroups

  Expert 4

  The question focuses on management, not assessment, of risks, and so does my response. The risks/hazards should be managed in subgroups.

• Multiple subgroups

  Expert 6

  From a practical viewpoint, aggregating all PFAS into a single group for risk assessment provides a more holistic approach, but it ignores the potentially significant differences in toxicological properties/MoA etc. Significant toxicological potency differences between PFAS carboxylates and sulfonates of similar chain lengths have been suggested (e.g. see Ali et al, Andersen et al and Luz et al Reg Tox Pcol 2019), so differentiation based on this characteristic could be considered. Otherwise, grouping according to common toxicological endpoints or MoA is a more conventional approach (see answer to 1.3).

• Other (please explain below)

  Expert 11

  As discussed above, when a group is not defined on the basis of a specific mode of action, multiple modes of action are possible. In addition, PFAS have a wide range of physical and chemicals properties that can affect exposure potential and toxicokinetic behaviors. Multiple modes of action and exposure routes would be expected to warrant different management strategies structure for various sub-categories of PFAS. The potential for multiple modes of actions increases when broader definitions of PFAS are used. For example, PTFE does not have a polar moiety and its toxicity would be expected to be different than PFOA or PFOS.

  That said, there is always great value in developing screening approaches for eliminating concerns for low-risk exposures. As a result, I would recommend that effort be made to develop multiple tiered approaches for managing risk that used a range of ways of characterizing hazard some based on screens (total fluorine in a sample, the sum of all PFAS, sum of certain PFAS, etc.) and some higher tiers that are based on the amounts of specific substances. Such an approach is discussed in section 1.3.

• Other (please explain below)

  Expert 3

  Risks and hazards are two different concepts. Hazard characterization or effects characterization includes identification of likely adverse health effects associated with an agent, as well as the dose response (potency) of that agent. By contrast, risk is defined as a function of both hazard and exposure. There is no risk in the absence of exposure. And an exposure is without risk if there is no hazard at that level, type, or route of exposure.

  It would make sense to divide PFAS into groups based on knowledge of exposure and of hazard. As noted below -- depending on the purpose of the assessment -- one might focus on a group of high production PFAS, or legacy PFAS, or observation of particular chemicals at a waste site or other source to environmental media. For a ranking exercise these groupings may be particularly useful. Or at another extreme for an assessment of a Superfund site (an RI/ FS) it would be most important to include those only PFAS amenable to analysis and/ or which have been identified at the site.

  Likewise the grouping may depend on the population potentially at risk by virtue of a type of exposure. For example, episodic, rather than continual low-dose chronic exposure, may raise concerns about developmental effects over liver toxicity; thus the in utero or childhood population would be of particular. The consideration of population exposure and susceptibility raises further questions as to the applicability of PFAS data across diverse endpoints.

  I expect that even once the purpose of the assessment has been defined and a conceptual model described, there will be iterations of subgroups based on several considerations.

• Single group

  Expert 9

  Given the widespread occurrence of PFAS in the environment, their persistence, mobility, potential to bioaccumulate, potential and known toxicities, and/or continuation of exposure, PFAS risks/hazards should be managed as a single group. While sub-grouping approaches could be considered, it may be challenging to arrange sub-groups due to the high number of individual PFAS that have been studied minimally or not studied at all. Sub-grouping by more traditional structure-activity relationships and/or molecular initiating events, for example, would not be possible for a high number of PFAS. Cousins et al. (2020) compared and contrasted several sub-grouping approaches that could be applied to PFAS and with all but the "P-sufficient" approach, data requirements would likely prohibit the placement of understudied/unstudied PFAS into sub-groups. With the P-sufficient approach, no data are required as the vast majority of PFAS are persistent and even precursors that are not persistent degrade to stable PFAS end products that are persistent. Grouping all PFAS using a P-sufficient approach for management would require no additional data and is based on the very long environmental half-lives shared by the vast majority of PFAS.

• Single group

  Expert 1

  The approach to risk assessment for mixtures of PFAS needs to match available analytical methods. Given the wide variety of PFAS and the limited number of analytes measured in current targeted chemical analyses, it would be most useful to determine screening guidance values for adsorbable organic fluorine (AOF) or extractable organic fluorine (EOF). Given the lack of toxicological studies on most PFAS, translating an existing health-based guidance value (e.g., PFOA/PFOS) into a AOF/EOF could be useful to quickly evaluate the risks of a given total organic fluorine dose/concentration.

• Other (please explain below)

  Expert 8

  The "PFAS" which would be included in the broad OECD definition should not be considered as a group for the purposes of risk assessment. It is clear from the data on the legacy PFAS that there are vast differences in toxicity, MOA, half-life, exposure etc, and this will even be more complex as the PFAS world is extended beyond the legacy PFAS. The management of PFAS again depends upon the initial problem formulation and the risk management options that are available. For most known endpoints, the data do not support grouping. For example, both PFOS and PFOA cause neonatal mortality in mice but the MOA is totally different - neonatal mortality caused from PFOA is PPARa dependent whereas that caused by PFOS is PPARs independent. Risk management of novel "PFAS" already considers use, exposure, potential for release to the environment, pchem properties, and toxicity (either through data or read across) and largely limits any potential human exposure. There is greater concern for legacy PFAS which did not undergo a similar regulatory evaluation process before entering the market. Risk management of PFAS that are already present in the environment will likely involve single chemicals or perhaps subgroups. Risk management may also take a more pragmatic approach. For example, in drinking water the overall toxicity of a group of PFAS will likely be driven by those PFAS with the longest half-life. One might propose that the total level of PFAS allowed in drinking water is 1 unit and this can be filled by any of the PFAS as long as the total does not exceed 1 unit.
  For legacy PFAS as well as some "newer" PFAS, it is also important to consider the NHANES data which would reflect exposure from all routes. Although only a limited number of PFAS have been screened by in NHANES, these data may help inform which PFAS are of potential concern to the general US population. Obviously there may be hot spots where exposures are higher. But, for example, PFBS has been reported in serum of humans in the general population. In American Red Cross samples collected in 2015, 8.4% had a quantifiable serum PFBS concentration; the majority of samples were below the lower limit of quantitation (4.2 nanograms per milliliter [ng/mL]). The National Health and Nutrition Examination Survey (NHANES) 2013–2014 data reported the 95th percentile for PFBS at or below the level of detection (0.1 ng/mL). Thus, exposure to PFBS is quite low in the general population, and may not be of top priority.

• Other (please explain below)

  Expert 5

  Assessment via multiple subgroups is warranted when data indicate different (multiple) targets of toxicity and/or multiple modes of action.

  Optimally, assignment of a chemical to a given subgroup should reflect knowledge of target organ/tissue/system and mode of action. Without mode of action knowledge, subgroups should be based on target organ/tissue system. This approach would enable dose additive approaches based on hazard index when both exposure and acceptable level (e.g., Reference Dose) are known; or on relative potency, when an acceptable level is established for the index chemical and measures of potency can be developed (in vivo or - with caveats - in vitro) for members of the subgroup. Because risk is a function of exposure and hazard, and because differences in biological longevity (half-life, clearance, bioaccumulation) relate to exposure, differences in biological longevity should not be considered as a basis for grouping. Grouping should be accomplished on the basis of hazard (target organ/tissue/system) and refined if mode of action data are available and demonstrate different modes of action are evident within a given target.

  Modes of action should be determined/estimated for effects demonstrated at the animal point of departure and should not be speculated-on without knowledge of the relationship between the test dose/concentration and the dose/concentration pertinent to the animal point of departure. This information will be important, should discussions of the potential human relevance of the mode of action be instigated.