SciPi 353: Grouping of Per- and Polyfluoroalkyl Substances (PFAS) for Human Health Risk Assessment
Based on a consideration of scientific merit, please rate the following grouping strategies on a scale of 1-10 (1=lowest, 10=highest)
Results
1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | Total | |
---|---|---|---|---|---|---|---|---|---|---|---|
Production/Use | 0.00% 0 | 36.36% 4 | 0.00% 0 | 0.00% 0 | 9.09% 1 | 18.18% 2 | 18.18% 2 | 0.00% 0 | 0.00% 0 | 18.18% 2 | 11 |
Potential exposure route | 0.00% 0 | 18.18% 2 | 27.27% 3 | 0.00% 0 | 0.00% 0 | 18.18% 2 | 0.00% 0 | 0.00% 0 | 9.09% 1 | 27.27% 3 | 11 |
Biological half-life | 0.00% 0 | 0.00% 0 | 9.09% 1 | 0.00% 0 | 18.18% 2 | 9.09% 1 | 36.36% 4 | 9.09% 1 | 0.00% 0 | 18.18% 2 | 11 |
Physical-chemical properties | 0.00% 0 | 0.00% 0 | 0.00% 0 | 18.18% 2 | 9.09% 1 | 18.18% 2 | 9.09% 1 | 9.09% 1 | 9.09% 1 | 27.27% 3 | 11 |
Carbon chain length/chemical structure | 0.00% 0 | 0.00% 0 | 0.00% 0 | 0.00% 0 | 18.18% 2 | 36.36% 4 | 18.18% 2 | 0.00% 0 | 9.09% 1 | 18.18% 2 | 11 |
Toxicological endpoints | 0.00% 0 | 0.00% 0 | 9.09% 1 | 9.09% 1 | 0.00% 0 | 9.09% 1 | 0.00% 0 | 18.18% 2 | 27.27% 3 | 27.27% 3 | 11 |
Mode of action | 0.00% 0 | 9.09% 1 | 9.09% 1 | 0.00% 0 | 0.00% 0 | 0.00% 0 | 0.00% 0 | 18.18% 2 | 18.18% 2 | 45.45% 5 | 11 |
Other (please explain below) | 0.00% 0 | 0.00% 0 | 0.00% 0 | 0.00% 0 | 0.00% 0 | 0.00% 0 | 16.67% 1 | 16.67% 1 | 16.67% 1 | 50.00% 3 | 6 |
Answer Explanations
- Expert 5
Grouped by: 1 2 3 4 5 6 7 8 9 10 Production/Use 0 1 0 0 0 0 0 0 0 0 Potential exposure route 0 0 1 0 0 0 0 0 0 0 Biological half-life 0 0 0 0 0 0 1 0 0 0 Physical-chemical properties 0 0 0 0 0 1 0 0 0 0 Carbon chain length/chemical structure 0 0 0 0 1 0 0 0 0 0 Toxicological endpoints 0 0 0 0 0 0 0 0 0 1 Mode of action 0 0 0 0 0 0 0 0 1 0 Other (please explain below) Risk is a function of exposure and hazard. Production has been applied as a surrogate for human exposure. However, the relationship between production and exposure is likely inconsistent across PFAS chemicals for reasons including differential environmental persistence and bioaccumulation. Further, the wide range of potencies demonstrated, even within a given PFAS class, establishes an additional level of uncertainty between exposure and hazard. For these reasons, production and use seems a poor choice for the basis for grouping to establish health risk.
Whether potential exposure route means oral, dermal, inhalation; or exposure pathways from source to receptor does not matter much. This grouping strategy provides an approach likely no more pertinent that production/use.
Molecular/structural attributes of PFAS chemicals do exert some influence on environmental fate and transport and biological interactions, but do not, themselves, have a reliable measure of toxicological predictivity, given our present state of knowledge. The development of such knowledge may require an amount of testing (likely in vitro testing) that is likely to produce enough (in vitro) effect data to make moot the use of such data in the establishment of sufficient understanding of structural alerts as reliable, quantitative indicators of risk. Nonetheless, some structural characteristics do have some predictive value at a very gross level.
Physical-chemical characteristics may include persistence, bioaccumulation potential, environmental mobility, and may be quantified to some extent. The use of octanol-water partition coefficient and protein binding are tow attributes that seem to offer some quantitative value in estimating the relationship between environmental exposure and internal dose (toxicokinetics, clearance, biological half-life).
Biological half-life seems to be of most value in its relationship to exposure, with risk being a function of exposure and hazard. Grouping strategies to this point have focused on Exposure, rather than Hazard, or Risk.
The two most risk-relevant grouping strategies are mode of action and toxicological endpoints, with toxicological endpoints being the most important. Mode of action information can be used to discriminate between chemicals that affect the same target organ/tissue/system, and absolutely serve as the basis to defend the choice of Dose Additive or Response Additive mixtures approaches to the Cumulative Assessment Group. Even for chemical affecting the same organ/tissue/system, those with different modes of action are places in different subgroups, their risks determined for each subgroup, and the risks (not the doses) are summed. Absent a reliable level of MOA understanding, chemicals are typically grouped according to affected organ/tissue/system, and the mixture response is determined via dose additive approaches (e.g., hazard index, relative potency factors). While MOA infers a more detailed knowledge of the toxicity, determining a MOA typically infers knowledge of the impact on a given tissue/organ/system, knowledge of the impacted organ/tissue/system is a more readily determined piece of information.
- Expert 10
Grouped by: 1 2 3 4 5 6 7 8 9 10 Production/Use 0 0 0 0 0 0 0 0 0 1 Potential exposure route 0 0 0 0 0 0 0 0 0 1 Biological half-life 0 0 0 0 0 0 0 0 0 1 Physical-chemical properties 0 0 0 0 0 0 0 0 0 1 Carbon chain length/chemical structure 0 0 0 0 0 0 0 0 0 1 Toxicological endpoints 0 0 1 0 0 0 0 0 0 0 Mode of action 0 0 1 0 0 0 0 0 0 0 Other (please explain below) 0 0 0 0 0 0 0 0 0 1 This is a poorly defined question. Grouping strategies depend on the purpose of grouping and this is not stated in the question. I could rank all of the above as 10 depending on the end purpose. For example, grouping all PFAS based on high persistence would be a good strategy for restricting the use/emission of PFAS going forward. Elimination half-life can be useful for determining which PFAS should be grouped to determine the highest internal exposure (e.g. this is why long-chain PFAAs could be grouped separately from short-chain PFAAs). Toxicologial end points and mode of action are important for risk assessment if taking toxicologically-strict strategy about which PFAS should be grouped for risk assessment. Given the lack of information on endpoints and mode of action, I'd personally group for risk assessment purposes based only on exposure considerations (i.e. which PFAS would result in the highest internal exposures in humans based on external exposure and elimination half-life). I realize that this is a simplification, but it's a precautionary one that I could support scientifically.
- Expert 8
Grouped by: 1 2 3 4 5 6 7 8 9 10 Production/Use 0 0 0 0 0 0 0 0 0 1 Potential exposure route 0 0 0 0 0 0 0 0 0 1 Biological half-life 0 0 0 0 0 0 0 0 0 1 Physical-chemical properties 0 0 0 0 0 0 0 0 0 1 Carbon chain length/chemical structure 0 0 0 0 0 0 0 0 0 1 Toxicological endpoints 0 0 0 0 0 0 0 0 0 1 Mode of action 0 0 0 0 0 0 0 0 0 1 Other (please explain below) I find this question difficult as there is no problem formulation statement and therefore no context as to the reason for grouping. I do not think the science supports grouping of all PFAS. Grouping of a subset of PFAS will depend on the issue being assessed, and all of these are important depending on the scenario one is interested in. If one were looking at exposure from furniture treatments, then production/use, exposure, leaching from the furniture, pchem properties would be important for grouping, and would be a major driver for the assessment as it would answer the question of whether there was exposure. Then if there is exposure, one would know the subset of PFAS and then half-life, MOA, toxicity etc would become important. Again, risk assessment is an iterative process which constantly refines exposure and hazard. If one were looking at drinking water, then one might be looking at a different subset of PFAS. Production/use might be less important in this scenario, whereas half-life, toxicity, MOA would be of greater importance for grouping. Therefore I have rated all of them the same.
- Expert 6
Grouped by: 1 2 3 4 5 6 7 8 9 10 Production/Use 0 1 0 0 0 0 0 0 0 0 Potential exposure route 0 0 1 0 0 0 0 0 0 0 Biological half-life 0 0 0 0 1 0 0 0 0 0 Physical-chemical properties 0 0 0 1 0 0 0 0 0 0 Carbon chain length/chemical structure 0 0 0 0 0 1 0 0 0 0 Toxicological endpoints 0 0 0 0 0 0 0 0 1 0 Mode of action 0 0 0 0 0 0 0 0 0 1 Other (please explain below) Conventional grouping methods for risk analysis emphasise commonality of toxicological endpoint or mode of action if additivity is to be assumed and a Hazard Index (HI) approach employed. The problem is that there may be insufficient toxicological information across the spectrum of relevant PFAS, or available Reference Doses (RfD) for specific PFAS, so alternative grouping strategies (e.g. chemical structure/chain length; physico-chemical properties) may be a more pragmatic approach. There is a useful diagram (Fig 1) in a paper by Wang et al (2021) in Environmental Science & Technology (not in list of references provided) that groups >3000 PFAS into sub-groups based on chemical structures and shows the relative abundance of published data on each sub-class.
- Expert 9
Grouped by: 1 2 3 4 5 6 7 8 9 10 Production/Use 0 1 0 0 0 0 0 0 0 0 Potential exposure route 0 1 0 0 0 0 0 0 0 0 Biological half-life 0 0 0 0 1 0 0 0 0 0 Physical-chemical properties 0 0 0 0 0 0 0 0 1 0 Carbon chain length/chemical structure 0 0 0 0 1 0 0 0 0 0 Toxicological endpoints 0 0 0 1 0 0 0 0 0 0 Mode of action 0 1 0 0 0 0 0 0 0 0 Other (please explain below) 0 0 0 0 0 0 0 0 1 0 Persistence (somewhat based on the physical-chemical property of the vast majority of PFAS).
- Expert 11
Grouped by: 1 2 3 4 5 6 7 8 9 10 Production/Use 0 0 0 0 1 0 0 0 0 0 Potential exposure route 0 0 1 0 0 0 0 0 0 0 Biological half-life 0 0 1 0 0 0 0 0 0 0 Physical-chemical properties 0 0 0 0 0 0 1 0 0 0 Carbon chain length/chemical structure 0 0 0 0 0 0 1 0 0 0 Toxicological endpoints 0 0 0 0 0 0 0 0 1 0 Mode of action 0 0 0 0 0 0 0 0 0 1 Other (please explain below) 0 0 0 0 0 0 0 0 0 1 This question is difficult to answer as posed. You appear to be asking us to rank the relative value of each strategy as a sole basis for creating subcategories of PFAS substances. While some of the criteria given in the individual strategies could support grouping on a stand-alone basis, most would have greater value when used in combined with other criteria. In addition, some would provide a basis for screening assessments but not for higher tier assessments. The following are comments on each of the strategies.
Production
Production is necessary but not sufficient for a prediction of risk. Where a chemical is not created in the environment (degradation product of another chemical or naturally occurring) then the lack of production (current and historical) can be used to exclude a chemical from concern. There would be no need of testing of the chemical’s hazards or properties (unless it was selected as an input for in silico models) and no need to monitor for the chemical in the environment or in individuals.
But the production of a chemical cannot justify the assumption that toxicity testing or monitoring is necessary for a chemical since some chemicals are intermediates that are not released to the environment in large amounts and may have characteristics the limit exposure (high molecular weight).
Because of the value of a “no production” finding to eliminate the potential of the presence of a specific chemical I have ranked the factor as “5”.Use – Potential exposure route – Biological half life
Patterns of use, potential exposure route, and biological half-life, are closely intertwined factors. All three are components of the determination of the Exposure Fraction (EF) of a chemical (the fraction of a chemical produced that reaches the target site of the molecular initiating event). Use of a chemical determines the fraction that is released to the environment, the location of the release, and the relevant media. The use defines the starting point for near- and far-field exposures and the relevant exposure routes. PFAS telomers that are covalently bonded to polymers that are in turn present on fabric (Gortex) will have a lower EF for both near field (clothing user) and far field (leachate from disposed of the fabric in a landfill or residential wastewater from laundering the fabric) than PFAS used in fire-fighting foams. Once a PFAS reaches the individual or reaches an individual’s food chain, then biological half-lives become important. Biological half-lives in humans will also play a significant role in is setting permitted doses (RfDs or ADIs) but not as a standalone criterion.
As a result, none of the three criteria make a good strategy for stand-alone grouping. I have ranked all three as a “3” (modestly valuable).
Physical-chemicals properties
Physical-chemical properties are not a single criterion. Some properties like vapor pressure or solubility would be relevant for those PFAS that that reach individuals by air or water pathways. Because this strategy includes many endpoints which if used in combination could be more effective, I have ranked it as “7”.Carbon chain length/chemical structure
Carbon chain length/chemical structure is not a single criterion. Some structure related measurements affect multiple steps in exposure and toxicity pathways and will be very useful for exclusion (e.g., very large molecular weights suppress exposure from air and water and reduce internal doses by reducing transport across membranes). These could be an effective part of a multi-strategy approach but would be less effective as a stand-alone criterion. Because this strategy includes many endpoints which if used in combination could be more effective, I have ranked it as “7”.Toxicological endpoints
Toxicological endpoints are highly relevant for predicting risk. Endpoints defined in vivo serve to integrate all steps in a molecule’s toxicokinetics and toxicodynamics. However, the finding that two chemicals cause the same endpoint does not indicate that the operate by the same mode of action, have the same MEI, would follow dose additivity, or would be the basis for RPFs.Mode of action
As discussed in “Knapen, D., Angrish, M. M., Fortin, M. C., Katsiadaki, I., Leonard, M., Margiotta‐Casaluci, L., ... & Villeneuve, D. L. (2018). Adverse outcome pathway networks I: development and applications. Environmental toxicology and chemistry, 37(6), 1723-1733.” and “Price, P. S., Jarabek, A. M., & Burgoon, L. D. (2020). Organizing mechanism-related information on chemical interactions using a framework based on the aggregate exposure and adverse outcome pathways. Environment International, 138, 105673”, dose additivity (whether measured by HIs or RPFs) is limited to chemical mixtures that operate by a common molecular initiating event (MIE). Chemicals with a common apical effect but a different AOPs operate by response additivity but not dose additivity. Chemicals with separate MIEs but are on a common convergent AOP network will have a complex interaction that could be dose additive, antagonistic, or synergistic. But only at, or above, doses that are sufficient to cause each chemical’s MIEs. Below these doses the chemicals would not interact.
Because of this, knowledge of the mode of action in particular the initiating mechanism and how multiple chemicals interact with each other, and the initiating mechanism provides the strongest basis for grouping. This is also the position taken by the Office of Pesticide in the determination of cumulative risk.
This suggests that testing programs should focus on identifying the biological activity that initiates the adverse outcome pathways for PFAS substances. PFAS substances that affect the same MEI should be the basis for assessment groups for PFAS mixtures. Defining AOPs and AEPs for PFAS would allow the creation of AEP-AOP networks. Such networks can be used to define the potential for kinetic and dynamic synergistic interactions (Price et al. 2021).Other
Grouping strategies for assessing risks could include multiple strategies or could include conservative assumptions that would simplify groupings and mixture risk assessments. The following is an example of a tiered system of screening mixture assessment. The simplest screen is based on total fluorine and evaluates all chemicals based on their weight fraction of fluorine. The higher tiers use an HI approach that conservatively assumes that all substances operate through a common mode of action and cause the same adverse effect. The highest tier uses information on mode of action to create a biologically based dose response model for the substances that have MEIs on a common AEP-AOP network.
In this approach, a decisionmaker would have a series of tired risk assessment approaches to determine if levels of PFAS substances in a sample of media or biomonitoring sample was a concern. As an initial screen, a weight fraction of fluorine could be set based on the assumption that all of the fluorine was in the form of a highly toxic PFAS chemical that (based on production and use) reasonably could be expected to occur in the media/biomonitoring sample. The analyst would only need to measure of total fluorine in the sample. If the sample passed, then the analyst could conclude that any PFAS in the sample would pose no risk.
If it failed, the sample could be evaluated by a second-tier screen. The sample could be tested or total fluorine after by excluding inorganic fluorine and fluorine in high molecular weight perfluorinated polymers (e.g., Teflon). If it passed the fluorine weight fraction, then the analyst could conclude that any PFAS in the sample would pose no risk.
If it failed, the sample could be tested in a third-tier screen. In this screen the range of carbon chain lengths in the PFAS in the sample are measured. If it was C5 or less, then a higher amount of fluorine could be present and not be a concern since the most toxic PFAS would be excluded.
If it still failed, then a fourth tier a Hazard-Index analysis could be performed based on the identification of the actual chemicals present in the sample. The toxicity values would be taken from available data or estimated using read across. See the following example of a method of performing such an analysis (Price, P. S., Hollnagel, H. M., & Zabik, J. M. (2009). Characterizing the noncancer toxicity of mixtures using concepts from the TTC and quantitative models of uncertainty in mixture toxicity. Risk Analysis: An International Journal, 29(11), 1534-1548.). This approach would add the hazard quotients for all PFAS irrespective of the chemicals’ mode of action.
Finally, the highest tier would be the risk posed by the mixture of PFAS would be evaluated using models of chemical interactions based on the relevant modes of action for specific chemicals. The assessment groups would be based on chemicals that affect a common MEI or common apical effect. These models could use dose addition, response addition, threshold response addition, or models of synergy/antagonism. - Expert 4
Grouped by: 1 2 3 4 5 6 7 8 9 10 Production/Use 0 0 0 0 0 0 1 0 0 0 Potential exposure route 0 0 0 0 0 0 0 0 1 0 Biological half-life 0 0 0 0 0 0 1 0 0 0 Physical-chemical properties 0 0 0 0 0 0 0 1 0 0 Carbon chain length/chemical structure 0 0 0 0 0 1 0 0 0 0 Toxicological endpoints 0 0 0 0 0 0 0 0 1 0 Mode of action 0 0 0 0 0 0 0 0 1 0 Other (please explain below) 0 0 0 0 0 0 1 0 0 0 Carbon chain length/structure is a predictor of phys/chem properties. It would be preferable to group by phys/chem properties, which are potentially a better predictor of both hazard and exposure. The reality, however, is that there is little or no information on most PFAS (regardless of how they are defined). The only thing known for virtually all PFAS is there structure. So features like structure may, in the end, be the most practical way to start the grouping process.
Production/use might inform which specific substances should be evaluated, but isn't a good way of grouping for managing risk, since risk doesn't scale predictably with production/use (it scales with hazard or exposure).
There is merit in grouping by potential exposure route (e.g. ingestion of drinking water or food; inhalation of air), since risk management actions are are often media-specific.
- Expert 1
Grouped by: 1 2 3 4 5 6 7 8 9 10 Production/Use 0 1 0 0 0 0 0 0 0 0 Potential exposure route 0 1 0 0 0 0 0 0 0 0 Biological half-life 0 0 0 0 0 0 0 1 0 0 Physical-chemical properties 0 0 0 1 0 0 0 0 0 0 Carbon chain length/chemical structure 0 0 0 0 0 1 0 0 0 0 Toxicological endpoints 0 0 0 0 0 0 0 1 0 0 Mode of action 0 0 0 0 0 0 0 1 0 0 Other (please explain below) Should the data on half-life/toxicology become available for a suite of past- and current-use PFAS, the most robust approach would use this data for grouping PFAS and assessing health risks. However, generating this data in a timely manner seems unrealistic and unsustainable.
- Expert 3
Grouped by: 1 2 3 4 5 6 7 8 9 10 Production/Use 0 0 0 0 0 1 0 0 0 0 Potential exposure route 0 0 0 0 0 1 0 0 0 0 Biological half-life 0 0 0 0 0 1 0 0 0 0 Physical-chemical properties 0 0 0 0 0 1 0 0 0 0 Carbon chain length/chemical structure 0 0 0 0 0 1 0 0 0 0 Toxicological endpoints 0 0 0 0 0 1 0 0 0 0 Mode of action 0 0 0 0 0 0 0 1 0 0 Other (please explain below) 0 0 0 0 0 0 0 1 0 0 As noted in other answers, I think there will be iterations of subgroups based on the purpose for the risk assessment.
Production / use / identification in environmental media can be a reasonable first cut; again, if there is no exposure, there is no risk.For a full blown (high tier) mixtures risk assessment, grouping by mode of action (MOA) is preferred and is best supported for judging "sufficient similarity of mixtures". These data for PFAS are limited, albeit becoming more available. Accepted MOAs for PFAS based on well characterized adverse outcome pathways (AOP) are even more limited. It is clear from the literature provided as part of the review package that there is not a single MOA that applies to all adverse outcomes associated with PFAS. It is also clear that a well-characterized MOA for rodent hepatic tumors induced by PFAS (mediated through PPAR alpha) is of limited relevance to effects in humans.
In the hierarchy of information used to determine sufficient similarity of mixtures, grouping by toxicological endpoints is the next preferred after MOA, followed by grouping based on affected organ. Note that several authors have developed relative potency factors (RPF) for PFAS based on a few toxicological endpoints (Goodrum et al 2020; Borg et al. 2013; Bil et al. 2020).
- Expert 7
Grouped by: 1 2 3 4 5 6 7 8 9 10 Production/Use 0 0 0 0 0 0 1 0 0 0 Potential exposure route 0 0 0 0 0 0 0 0 0 1 Biological half-life 0 0 0 0 0 0 1 0 0 0 Physical-chemical properties 0 0 0 0 0 0 0 0 0 1 Carbon chain length/chemical structure 0 0 0 0 0 0 0 0 1 0 Toxicological endpoints 0 0 0 0 0 0 0 0 0 1 Mode of action 0 0 0 0 0 0 0 0 0 1 Other (please explain below) 0 0 0 0 0 0 0 0 0 1 All of the above have their value at certain stages of the assessment; however, once a decision was made, then subsequent conclusions must respect these and not omit those that are not convenient.
For toxicological endpoint, clear definition is needed as to the chemical and how the endpoint has been determined. Chemists and toxicologists should be aware of the experimental set-up. For example, the paper by Han et al. (Han et al., 2003) defines PFOA as protein-bound (others state the same for PFOS); on the other hand, environmental chemists and biomonitoring determine “free” PFOS or PFOA, i.e., after digestion, thus, not the protein-bound. I understand that protein-bound PFA molecules do NOT exhibit the toxic properties. If correct, the exposure scenario has to be adapted.
I consider production/use as starting point to determine which kinds of PFAS exist but it will not be possible to make quantitative assessments. Needs differentiation between non-polymeric and polymer fluoro compounds and the latter ones differentiated into side-chain polymers and fluoropolymers.
Expert 3
09/15/2021 13:36None of these approaches is bad. But I appreciate the comments that several of these criteria for grouping are linked.
Expert 3
09/15/2021 13:40I found reviewer 11's comments to be particularly useful and comprehensive. I may differ, however, whether additivity should be limited only to those PFAS with the same MIE. I agree that interconnecting AOP tend to be complex (connections or branches at other than the first key event), but I am uncertain as to how that affects the likelihood of interaction among components of a mixture. Are not most interactions noted at high exposures?
Expert 3
09/15/2021 13:42I agree with reviewer 10 that 1.3 is a poorly designed question.
Expert 5
09/16/2021 14:24Regarding scientific merit as focussed here on risk, I find the ratings reassuring in that toxicity parameters seemed to receive generally high marks. It must be noted, however, and may be addressed in further discussions that such data may not be available. Thus, some hierarchy of grouping strategies might be sought.
Expert 11
09/17/2021 20:34In response to reviewer 3, I would argue that true additivity, where many small doses can add to cause an effect, is limited to chemicals that operate on a common MIE. But chemicals with multiple MIEs on a common AOP network can at high concentrations also follow a dose additive model, cause synergy, or cause antagonism. But these interactions will only occur when both chemicals occur at concentrations that trigger their respective MIEs. It is the subsequent events on the AOP that interact and not the chemicals. If neither chemical (or if only one of the chemicals) triggers an MIE then there are no separate effects in the AOP network to interact and the mixture will not follow dose additivity. This can be thought of a threshold in dose additivity, that there are doses of such chemicals below which the chemicals do not add.
Now MIEs can be triggered at concentrations below the doses that cause apical effects, so in practice it is hard to say when a predicted concentration is not large enough to trigger an MEI. But it does suggest that if you have a mixture of a PFAS where one compound predominates and 20 other PFAS compounds are at much lower low doses, then even if all compounds caused a common endpoint in practice they may not add.
Expert 7
09/20/2021 01:24I agree with expert 4 that sectoral approach is necessary.
Further it shall be noted that some of the parameters above govern exposure, others effect. Both must be combined and weighted.
A database containing the values for all of the above grouping parameters should form the basis and then being weighted as is typically done for criteria of high confidence (measured/proven in real life situations by many experiments/different groups/ etc ) down to QSAR- or otherwise derived data.
Expert 9
09/20/2021 11:36The Cousins et al (2020) paper addresses many of the concerns raised by the experts here and could provide some further discussion points that may enhance the depth of the debate.
Expert 10
09/22/2021 03:27It seems that several experts agree that the question was not well formulated. Different approaches can be used depending on the purpose of the grouping exercise.
Expert 4
09/23/2021 07:01Whether or not the question was well-designed, it has led to some good discussion. After reading everyone's thoughts, I conclude that the most appropriate grouping approach will vary by purpose, and by practical considerations (budget, time constraints, availabiillity of suitable data). I agree with reviewer 10 that any of these approaches could be best, depending on the situation.