The Sponsor of this review concluded that a threshold-based cancer risk assessment is supported for both oral and inhalation routes of exposure to 1,3-D. To what degree do you agree with this conclusion (1=strongly disagree; 10=strongly agree)? [please explain)

I agree with a threshold approach. 1,3- D does not act as a DNA reactive mutagen in vivo and the cancer bioassay data with relevant test material is largely negative (the benign liver and lung responses are of questionable relevance). The quantification of cancer risk is not conducted for a “not likely” classification . Thus, a noncancer endpoint/PoD that is considered to be the most relevant and sensitive is selected to assess the chronic exposure scenario using a threshold approach. Protection of human health is achieved by limiting exposures below the selected PoDs/UFs.

Even if one considered there is “suggestive evidence of carcinogenicity”, typically a quantification of cancer risk is not done because the nature of and uncertainties around the tumor data would not reliably support a dose response/quantitative assessment (see EPA 2005). And only a noncancer assessment is conducted (as described above).

A default linear approach would be taken for a "likely to be carcinogenic to humans" conclusion (in absence of MoA understanding, even if in vivo mutagenesis is ruled out), but in this case, the WoE does not support such a strong conclusion.

First, this is a bit harder to answer because I'm disappointed that the whole exercise tends to conflate two very different reviews-- the appraisal of the carcinogen classification versus the risk assessment. For example, I might (hypothetically, and not w/reference to 1,3-D) conclude that some chemical is both a "known human carcinogen" AND that it poses negligible risk to humans (because of exposure, potency, or both). Similarly, I might conclude that some other chemical is only a weak "possible" but that it should be stringently controlled to lower current exposure levels (because of exposure, potency IF truly carcinogenic, or both). The white paper, unfortunately, mixes the two concepts at key points in the narrative.

Similarly, "threshold" is a precisely-DEFINABLE concept that this exercise has failed to define well at all. Hypothetically, I might conclude that some chemical has a threshold that dramatically changes its risk at relevant doses-- but I might also conclude that the chemical has a threshold, but that it makes no difference at all. In other words, that the risk at current levels, assessed either via a linear assumption of dose-response or a threshold-based assumption, is either unreasonably high or trivially low regardless.

With these general caveats stated, I think it is usually unwarranted to claim that ANY carcinogen should have its risk assessed via the NOAEL/factor approach that has been used for non-carcinogens. Indeed, I agree with the National Academy of Sciences 2009 panel who concluded that even *non-carcinogens* should be assessed via continuous dose-response functions that allow for probability of harm at any dose to be estimated (even if that probability at a particular dose is near zero or trivially close to zero). So given this recommendation, surely all carcinogens, "genotoxic" or not, should be so assessed.

So I reject a "threshold based cancer risk assessment" entirely, UNLESS proponents of it can establish WHERE that threshold actually lies, rather than, as is done here, make theoretical arguments that a threshold exists SOMEWHERE. And that showing must at least attempt to account for the fact that each PERSON'S threshold will differ based on human interindividual variability. I am not arguing that a threshold can't be highly decision-relevant, only that it is wrong to assume that it MUST be relevant, sans evidence.

Moreover, in this case I don't think the proponents have even established that the exposures they say are too low to pose any risk actually are so. All they have done is provide some evidence that a KMD exists for mice and rats, without accounting at all for the power of the available studies to rule out a lower KMD, for the uncertainty in converting a rodent KMD to a human one, and for the interindividual variability WITHIN humans that would force science-policy judgments about what fraction of the human population should be afforded acceptable risk under that assumption and what fraction shall be left inadequately protected. In particular, the statements on p 56 that the particular values of 2.5 MKD (oral) and 5 ppm (inhal) are non-toxic and "human health protective" are wholly unsupported. Everyone knows (or should) that a NOAEL for a substance known to have a threshold is NOT a level with no effect, but (as the name indicates) a level with no OBSERVED adverse effect-- given the sample sizes common to toxicology, the NOAEL is usually thought to reflect a risk no higher than about 5%.

That said, the short section on exposure assessment, if true, would make a good case that prevailing RISKS are low, without having to rely on a controversial and ultimately unsupported claim that the risks are zero.

ADDENDUM ADDED MAY 1 BASED ON ROUND 4:

I renew my suggestion that the final section of the White Paper link together better the information on exposure with the information (and speculation) about hazard classification and dose-response. I have no problem with the sponsor stating that the MOEs appear to be quite large given what we (I assume...) know about exposure to 1,3-D. My concern remains, though, that no circular reasoning be employed: if 1,3-D is "not a concern as long as exposures are low," that does NOT mean that it should be classified as "not carcinogenic to humans" and therefore able to be released in much greater amounts/concentrations. The MOST that can be said from the White Paper is that 1.3-D is a "high-dose only carcinogen," which means that we must be concerned about doses becoming high. (I still have some concerns about intermittent doses, human variability, bladder cancers, etc., but here I'm just referring to the confusion between dose-response and hazard classification).

The sponsor of the review has not provided a MOA analysis, beyond trying to make a case for lack of direct DNA reactivity. For at lease some protocols (e.g. US EPA 2005 Cancer Guidelines), the lack of a MOA requires use of a default approach to low dose extrapolation involving linearity from a point of departure (POD). The US EPA Office of Pesticides Program as well as other groups may consider a Margin of Exposure (MOE) from an appropriate POD to measured or estimated human exposure. The POD need not be an exposure at which tumors were observed in a target organ. The White Paper authors spend considerable effort to exclude potential target organs as associated with 1,3-D exposure. I did not find the arguments sufficiently convincing as written.

Assuming that the exposure models and / or measurements are conservative, and that a POD based on tumors in rodents is relatively high dose effect, it may be difficult to establish that the calculated MOE is adequate to protect human health. Note also, however, that uncertainty factors are not used by US EPA in determining a MOE. The MOE is simply the distance between the estimated or measured human exposure and the POD expressed as an equivalent human dose or exposure. Factors such as variability in the human population may be considered in determining the adequacy of the MOE.

I would quite strongly agree that use of a point of departure to assess chronic toxicity based on these chronic/carcinogenicity studies is called for given the available data on genotoxicity, and despite the lack of understanding of a possible mode of action for toxic effects. However, we should be clear that, for a chemical that is found to be highly unlikely to be a human carcinogen, no quantitative assessment approach is needed. The value of using a threshold-based assessment is to suggest that in the face of some uncertainty in the finding, existence of an adequate margin of exposure (MOE) adds to our confidence that no carcinogenic harm is likely.

1) the nonlinear kinetics of response
2) responses were seen at doses exceeding the MTD
3) the lack of preneoplastic lesions
4) 1,3-D is not an in vivo genotoxicant
5) inhalation toxicokinetic study showing linear dose proportional below 30 ppm supporting the presence of a KMD
6) WOE indicating that 1,3-D is not a tumorigen at doses below 12.5 mg/kg/d via oral exposure or 20 ppm via inhalation

The data for rodents is strong, but it is not known how well it extrapolates to humans.