After reviewing the output from the genotoxicity panel, please summarize your thoughts on the genotoxicity issues that are pertinent to the cancer WOE conclusion for 1,3-D carcinogenicity.

In my opinion, the genotoxicity issues most pertinent to the WOE conclusion for 1,3-D are that:
1) the extensive body of well conducted in vivo studies using characterized 1,3-D test material not containing epichlorohydrin are uniformly negative in tests covering the range of genotoxic endpoints of interest
2) 1,3-D does not bind to DNA
3) there are mixed results from the in vitro tests
4) addition of normal physiological levels of GSH to in vitro tests reduces or eliminates the genotoxic response in vitro
5) the fact that the genotoxicity review panel members did not agree about the genotoxicity of 1,3-D.

While the mixed results from the in vitro tests suggest a hazard potential, the white paper makes a compelling argument that from a WOE perspective 1,3-D is not likely to pose a genotoxic risk to the general population, especially given that 1,3-D does not bind with DNA and the large margin of exposure between the oral and inhalation exposures for the general population. Of course, there are other exposed subpopulations of concern, especially those potentially exposed to higher levels of 1,3-D such as mixer/loader and applicator worker subpopulations. Their exposure levels are not given in the white paper given to us but EPA has this information and will evaluate these, as well as other relevant subpopulations that have been identified in their assessment.

From the information provided us, I agree with the statement made in the white paper that the genotoxicity profile of 1,3-D is consistent with that of non-genotoxic compounds as identified by the "Framework for Determining a Mutagenic Mode of Action for Carcinogenicity: Using EPA's 2005 Cancer Guidelines and Supplemental Guidance for Assessing Susceptibility from Early-Life Exposure to Carcinogens" (US EPA 2007).

Couple of comments:
In addressing the issue of whether or not 1,3-D has the potential to be mutagenic in vivo (a hazard question), I agree that 1,3- D is not likely to act as a DNA reactive mutagen in vivo, and thus the evidence does not support a mutagenic concern for humans nor provides support in a weight of evidence evaluation of potential human cancer hazard/risk for oral or inhalation exposures. While recognizing limitations in certain studies (as pointed out in the peer review), all of the relevant in vivo evidence (e.g., Mouse Big Blue, Rat Big Blue, 32P, Mouse micronucleus, rat dominant lethal) collectively show coherence and internal consistency in support this conclusion. My opinion is based on the following weight-of-evidence approach:
1. Consideration of all pertinent evidence, including genetic toxicity studies, nature of tumor responses, metabolic profile similarities for different routes and species/sexes, and TK data on saturable metabolism;
2. Consideration of study quality (including peer review comments on certain studies);
3. Placing more weight on studies using well characterized and representative test material. (Studies using technical material that contains epichlorhydrin or studies that do not provide adequate information on purity or on the stabilizer used should not form the basis of conclusions);
4. Placing more weight on in vivo assays (vs in vitro) using mammalian systems that have intact/in vivo physiological conditions (including DNA repair and metabolism pathways) and on endpoints that are direct measures of mutations or chromosomal aberrations (versus indicators of genotoxicity, e.g., DNA strand breakage); and
5. Consideration of whether or not there is a consistent pattern of positive findings that would be expected for a DNA reactive mutagen.

B. In focusing on the explicit issue of whether 1,3-D acts via a mutagenic mode of action (MoA) that leads to certain tumors observed in the bioassays (i.e., rat benign liver tumors following oral exposure and mouse benign lung tumors following inhalation exposure), the totality of the relevant evidence does not support the occurrence of key events in vivo whereby DNA reactivity initiates the process of carcinogenesis. Key events in a mutagenic MoA for cancer would include reaction with DNA in target cells to produce DNA damage followed by misreplication or misrepair leading to critical mutations in target cells. All of these events must occur to lead to a neoplastic response. My specific observations for rat liver and mouse lung tumors follow:

Rat Hepatocellular adenomas following oral exposure (Scott, 1995)- The Big Blue mutation assay (Young 2018) and the 32P labeling assay (Stott 1997), which used adequate doses (i.e., overlapped with cancer bioassay), evaluated DNA adduct formation and mutation induction by the relevant route, and in the species/sex and tissue of interest. Together these assays provide evidence that 1,3-D does not react with DNA in target cells to produce DNA damage that leads to mutations. Reviewers considered the Big Blue rat study to be clearly negative and clearly relevant. Although some study limitations were pointed out in the peer review for the 32P assay, nonetheless, the lack of mutations in the oral Blue Big assay is consistent with the negative findings in the 32P labeling assay, and together these studies are supportive of the lack of promutagenic DNA adducts/lesions leading to mutation induction. Although the hepatocellular adenomas reported at the highest dose are not likely the result of a DNA reactive/mutagenic MoA, it is important to stress that further MoA research into alternative MoAs is not justified for this benign tumor response. Before conducting MoA studies, there needs to be a clear carcinogenic response that has a sound and rationale scientific basis to warrant such studies. This common rodent tumor response was benign only with no indication of progression, only slightly outside or near HCs (for males only), and at a dose above the MTD (based on BW).
Mouse Bronchioavleolar adenomas following inhalation exposure (Stott, 1987). The inhalation Big Blue mouse study (Gollapudi 1997) and the inhalation 32P labeling assay (Stott 1997) evaluated DNA adduct formation and mutation induction by the relevant route, and in the species/sex and tissue of interest. These assays used sufficient concentrations (i.e., overlapped with the cancer bioassay). The 60 and 150 ppm concentrations would have exceeded the KMD and the protective mechanism of GSH conjugation. Some study limitations were pointed out for the mouse Big Blue assay in the peer review; however it was also stated that it “essentially complies, with the relevant 488 OECD Guideline”. Despite limitations, it is unlikely that the mouse benign lung tumors are a result of a DNA reactive/mutagenic MoA. Given there are similar metabolic profiles for 1,3-D for the oral and inhalation routes exposure and for mice and rats (and both sexes), the totality of evidence from the relevant oral and inhalation genotoxicity assays supports this conclusion. Also, this conclusion is further strengthen by the observation that this response is not consistent with a DNA reactive carcinogen- ie., one sex (males), one species (mouse), and at one site. Again, MoA research into alternative MoAs is not justified because it was a weak response that was characterized as benign only and occurred in a genetically susceptible strain/sex at a site with a high and variable background, and only was only a little higher than the HCs at the highest concentration tested. No lung tumors were found in female mice or in both sexes of rats following inhalation exposure, nor in the 4 oral studies.

Some cytogenetic/DNA strand break studies are confounded with epichlorohydrin or used material not well characterized. Although there were some short comings pointed out in the peer review for the oral mouse bone marrow micronucleus test (Gallapudi, 1985) and the inhalation dominant lethal test (Gollapudi, 1998), both used representative material and were clearly negative. The overall weight of evidence (TK, genotoxicity, pattern of cancer bioassay findings) does not support a concern for clastogenicity in humans.

With respect to aneuploidy, the same study issues pointed out for the BM micronucleus test and the other cytogenetic studies apply. However, to my knowledge, it has not been demonstrated that chemically-induced aneuploidy is a " causal event " within a MoA that begins the process of cancer. It might contribute to genomic stability as a later secondary consequence of the cancer process. Moreover, aneuploidy induction does not involve direct interaction of a chemical (or its metabolites) with DNA, and thus, is expected to have a threshold. A POD based on an endpoint protective of the most sensitive and relevant toxicity should be protective.

C. Recommendation for White Paper: Before presenting the genotoxicity results on 1,3-D, I think it would be useful for the final White paper to provide some text upfront as to the explicit weight-of-evidence approach used for judging a potential for in vivo mutagenesis and a DNA reactive mode of action for cancer (ie how information is weighed and integrated). It appears that the White paper is weighing evidence as I have outlined above A. 1-5. But again this could be laid out clearly at the beginning of Section 3. Genotoxicity.

Genotoxicity of 1,3-D has been relatively well studied with in vivo studies which allows in vitro results to be put in perspective. Recent well-run studies confirm earlier observations and clarify in vitro results where a genotoxic stabilizer, epichlorohydrin, was present. While GLP-compliant studies are preferred, non-GLP studies can be interpreted and are consistent with our understanding of the potential genotoxicity of 1,3-D. Although 1,3-D appears to be genotoxic in vitro, it is reasonable to conclude from the data that 1,3-D is most likely not mutagenic in vivo based on direct testing and informations on similar chemicals. Findings regarding clastogenicity and aneugenticity are not as clear, although in vivo results suggest that neither is likely.

It appears that 1,3-D is genotoxic in certain in vivo tests, but this has not translated into positive results for in vivo tests. Likely because detoxification pathways predominate.

As with the toxicokinetic evaluation, the genotoxicity write-ups provide strong evidence that a traditional default assumption may not hold true for 1,3-D. It does seem that several much older studies overstated the likelihood that 1,3-D is genotoxic. However, the Lawlor 2009 assay (apparently Ames-positive, without any genotoxic impurities) needs to be better explained in the Appendix. I also note two panelists’ answers to question 2.5, which point out that epichlorohydrin itself is negative in many assays, at concentrations far above what would have been impurities in the 1,3-D assays.

But the *theory* behind why the positive in vitro findings can be “eliminated completely” (white paper, p 16) by in vivo findings needs to be better thought through. If GSH is “completely protective,” then all of my concerns in the toxicokinetics section apply: this protection must be dose-related, and must depend on pharmacodynamics that is highly variable from person to person. So 1,3-D may be a “high-dose genotoxin,” and that would require much more information about what exposure levels could we confidently assume were truly “low” enough for most people. It is also somewhat illogical (also p 16) to claim that 1,3-D does not bind to DNA, when the issue is of course whether one or more *metabolites” of 1,3-D can do so.

I also find it odd that in the Big Blue mouse assay, the positive controls were allowed to live for 54 weeks and then assayed, whereas the mice exposed to 1,3-D were sacrified and assayed after only 17 days—wouldn’t this tend to inflate the control response or depress the exposed response?

But the main concern I have about this whole section of the white paper and panel impressions to date is that is ASSUMES a direct connect-the-dots between (if true) nongenotoxicity and "less concern." Where does this assumption come from? There are dozens of nongenotoxic carcinogens (see, e.g. Hernandez et al, (2009) Mutation Research 682:94-109), and scads of situations in which low additional exposures to (say) a tumor promoter are a GREATER concern for public health than additional exposures to some initiator.

The National Academy of Sciences committee that most recently reviewed EPA’s (and others’) methods for cancer risk assessment (NAS 2009, Science and Decisions in Risk Assessment) strongly recommended that the artificial distinctions between “safety assessment” for nongenotoxic substances and quantitative dose-response assessment for genotoxic ones be wiped away, in favor of a “unified” theory that strives to estimate risk (i.e., population response at a given dose) for ALL toxicants. This recommendation is founded on the substantial theory and evidence that even for mechanisms of action for which every individual in the human population has a “safe dose” threshold, the POPULATION exposure-response function may have no threshold or one sufficiently low that the function is monotonic (linear, sublinear, or supralinear) at all exposures of public-policy relevance. This review, to date, has failed to establish EITHER that 1,3-D is not genotoxic, or that if it was, it should rationally be assessed as less risky than before.

ADDENDUM ADDED MAY 1 BASED ON ROUND 4:

The commentary is persuasive that (with the exception of clastogenicity), most of the evidence points to 1.3-D having a nongenotoxic MOA. However, neither the White Paper nor the commentary has explained much at all how/why this is RELEVANT to risk assessment. Nongenotoxic carcinogens may well have a sublinear or threshold dose-response, but that would be evidence-based, and all we have here is evidence of the qualitative MOA, not of its risk or exposure-risk properties.

The in vitro results are not particularly germane to the evaluation of contemporary 1,3-D formulations. There is a strong likelihood that epichlorohydrin contributed to the observed positive responses in the older assays. The older assays were not done according to revised OECD guidelines; I would recommend giving less weight to these. I feel that the White Paper authors did not present a case for discounting assays that used DMSO as a solvent. Note that several tables show "epichlorhydrin" without the second "o", and that "Gollapudi" is misspelled in the Round 1 report.

The first paragraph of the White Paper section on Genotoxicity gives the impression of cherry-picking studies rather than of a systematic review with a priori criteria for selecting or discounting studies.

There is a reasonably good array of relevant in vivo assays. I consider the dominant lethal assay to be of less weight than the transgenic mouse results. I agree with the panelists that aneugenicity has not been ruled out; however, I see no strong indication that this should be given more serious consideration. I would prefer that the White Paper show the results of the 32P Post Labeling Assay by Stott (1997).