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

Couple of comments:
1. Older NTP bioassays using test material containing epichlorhydrin (and different % impurities) that is not representative of current test material should not form the basis of conclusions regarding the carcinogenic potential of 1,3-D because these studies are confounded, and thus a clear interpretation is challenging. Further, some additional study quality issues were pointed out in paper/peer review. Guideline compliant studies with well characterize and representative test material should form the basis of conclusions.

2. Oral Route of Exposure: There is general agreement that the two mouse bioassays are clearly negative for both sexes (Redmond 1995 and Kelly 1997). Also, there was agreement that the rat bioassay by Kelly 1998 was negative for both sexes. There was an increase in hepatocellular adenomas in the rat bioassay by Scott 1995. I do not view this increase as providing convincing evidence of carcinogenicity given that the 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). At best, these results are equivocal.

3. Inhalation Route of Exposure: There was no evidence of carcinogenicity in either sex of the rat in the inhalation bioassay by Lomax, which used adequate dosing. With respect to the mouse inhalation bioassay by Stott, which used adequate dosing, there was no evidence of carcinogenicity in female mice. There was an increase incidence of bronchioavleolar adenomas in male mice, but no treatment related responses were found at other tissue sites. This response was weak, benign, late onset, only a little outside HCs, occurred in a genetically susceptible strain/sex and at a site with a high and variable background, and at the highest concentration tested (60 ppm which exceeds the KMD). There was not an increase incidence in lung tumors in female mice nor in both sexes of rats, and given that metabolic profiles are similar for rats and mice (and for both sexes), metabolic profile differences are unlikely to account for this male mouse lung response. This response does not provide convincing evidence of carcinogenicity, and is of questionable relevance to humans based on the susceptibility at the tumor site, evidence on the lack of in vivo mutagenesis, and information on nonlinear toxicokinetics.

In conclusion, the totality of evidence from oral and inhalation bioassays using representative test material does not provide convincing evidence of carcinogenicity nor a concern for oral or inhalation human exposures.

1) no clear positive bioassays as those reporting positive contained epichlorohydrin and/or had other confounding factors
2) positive results also seen in studies exceeding the MTD and/or not demonstrating treatment related preneoplastic effects
3) positive results at high doses in inhalation studies exceed the KMD and thus not relevant for human risk
4) Cancer bioassay subpanel agreement, with high levels of confidence, that considering the WOE 1,3-D is not likely to be carcinogenic to humans
5) The cancer bioassay subpanel also agreed that there is not a need to conduct additional studies to increase the level of confidence

The cancer WOE discussion for 1,3-D has the benefit of multiple chronic/carcinogenicity studies in rats and mice and by two routes of exposure. Later studies have allowed re-interpretation of the results of the earlier NTP studies where a genotoxic stabilizer, epichlorohydrin was used. Results from these early studies are consistent with results for epichlorohydrin alone both qualitatively and quantitatively (based on TD 50). Chronic exposure at high dose levels of 1,3-D causes increased incidence of benign hepatocellular adenomas in F344 rats only. Evaluation of the results on the CD-1 mouse bioassay provides no evidence that 1,3-D is associated with tumors although submucosal mesenchymal tumors were originally reported. These lesions are now considered benigm prolifereation and would provide no weight for assessing human carcinogenicity potential. Additionally, the two bioassays in CD rats and B6C3F1 mice, tested at equal to or greater than the tumorigenic doses in F344 rats, showed no tumorigenicity. Consequently, 1,3-D is associated with the formation of tumors in only one tissue (liver) and in only one test strain (F344 rat). The weight of these data is limited for human carcinogenicity potential given the fact that most liver tumors are adenomas (1 carcinoma) with no indication of progression. In addition, data suggest that the MTD was exceeded and that only in high dose males did the incidence of tumors exceed the range of historical controls.

Based on rodent and in vitro tests I believe the cancer bioassays are now sufficient to state that 1,3-D likely does not pose a cancer risk humans under its conditions of use. However, it has not been tested in non-human primates. It's effects in offspring of pregnant mothers is also unknown.

The older bioassays wherein there was (measured?) exposure to epichlorohydrin could be given less weight than rodent tests using 1,3-D with contemporary stabilizer. Generally, positive responses were observed at MTD or above. Note that one assay did not establish an MTD.

For oral exposure the results show increased incidence of hepatocellular adenomas in lifetime studies of F344 rats treated at (actually above) an MTD. The discussion of historical control data for these lesions provides useful context, with the caveat that such comparisons are relevant only within labs.

What is DD-92 (Kelly 1998)? White Paper doesn't say. Also I would prefer that the White Paper show the cancer incidence data for the NTP studies for relevant organs. Note that rats are not tumorigenic, but rather, tumor-bearing. Discussion of TK in section 4.1.6.2 was unclear as to effect on the conclusions.

For inhalation exposure bronchioalveolar adenomas were observed in male B6C3F1 mice treated with 60 ppm 1,3-D, but not in female mice or in F344 rats. Tumors observed may be a portal of entry effect, as no lung tumors were found from oral exposure. Comparison with historical control data provides useful context albeit with caveats above. Table 24 ought show percentages. Studies of lung toxicity could be described in a MOA evaluation.

The interpretation of the various bioassays is obviously the crux of this exercise; the white paper and panel responses to date are problematic in that they distort the evidence base.

Most generally, there are various references to “six” bioassays in total, and Question 4.2 lists six of them—but of course, there are EIGHT in total—two of them by inhalation rather than by ingestion, and two others (ingestion) by NTP that used epichlorohydrin as a stabilizer. So depending on the context, the discussions selectively omit two of the eight assays, and *never* deal with all eight as a whole, with their strengths and weaknesses arrayed together. The white paper also is not careful to avoid misinterpretations of which NTP studies are being cited—for example, on p 34 it states that epichlorohydrin caused “the same [forestomach] lesions observed in the two NTP studies,” but does not mention that the major finding in the NTP mouse study was bladder cancer (see below).

There is nothing inherently wrong with an advocacy document (the white paper) taking a science-policy position (here, that 1,3-D cannot cause tumors below a threshold dose by ingestion or by inhalation) and attempting to support it. But two aspects of this advocacy cross the line here: (1) each individual study’s weaknesses are dissected, but without its strengths also even-handedly discussed [in particular, the single most worrisome finding—that of a high incidence of bladder carcinomas in exposed female mice against a zero control rate, is NEVER mentioned]; and (2) the result of casting some doubt on every one of the positive face-value findings (even if that was successfully accomplished, which I don’t believe it was) is NOT the same as demonstrating that the chemical is not a human carcinogen—in other words, a series of unrelated “positive but for this factor” arguments (again, even if all the doubt was justified) does not necessarily add up to a negative finding, as I will explain.

I will discuss each of the main “face-value” positive findings of the bioassays in turn, summarize how the white paper and/or panel comments to date have dismissed each finding, and offer my reactions to the merits of the dismissal.

(1) The rat liver adenomas (Scott 1995; male F344 rats) are dismissed (a) as not “really” elevated above the highest possible control rate during a 22-year period; (b) as “late onset” in nature; (c) as only statistically significantly elevated at one dose (25 MKD); and (d) as not accompanied by toxicity of the liver.

(2) The mesenchymal tumors of the bladder (Kelly 1997; female CD-1 mice) are dismissed as benign lesions.

(3) The various neoplasms (NTP 1985; F344 rats) are dismissed as a result of epichlorohydrin contamination.

(4) The bladder carcinomas and bronchioloalveolar adenomas (NTP 1985, B6 mice) are dismissed as a result of epichlorohydrin contamination, to the extent they are mentioned at all.

(5) The bronchioloalveolar adenomas (Stott 1987; B6 mice) are dismissed as benign, as “portal of entry” related, and as not “really” elevated above historical control rates.

In the same order, here are some additional observations:

(1) Table 16 in the white paper represents extreme cherry-picking. The historical control rate for male F344 rats at Dow in the year of the study is 2% (pooled 5/250). The complete control rate in Table 16, for all years, is 4.6% (40/865). Only by picking the highest possible control rate for males (16% in one year) and adding an irrelevant value for females (8%) was the white paper able to inflate the control rate by an amount large enough to make the 18% incidence at 25 MKD seem POSSIBLY not significant. As for “late onset,” perhaps someone in the bioassay group could provide additional explanation as to why a disease (cancer) whose very hallmark is age-related incidence should be discounted when it appears late in the lifespan (!) The incidence at 12.5 MKD is not significant at p=0.05, but a p-value of 0.13 should not be completely ignored—this means that there is “only” an 87% chance that the increase is not due to chance. In addition, the white paper (p 38) states that the p-value at 25 MKD is “greater than 0.01,” as if p=0.026 can be dismissed—but in fact, the correct p-value is 0.014, which takes account of the one carcinoma as well as the 9 adenomas. It takes some real contortions to argue that p=0.014 is not significant.

(2) I agree that these lesions are of less concern than many other neoplasms, but they should not be dismissed out of hand; for example, Halliwell 1998 (Toxicologic Pathology 26:128-136) concludes that “Each risk assessment of a drug or chemical is an individual task. The incidence and biology of SMT of the mouse urinary bladder should be evaluated with other related factors to provide a risk assessment decision.”

(3) and (4) The epichlorohydrin issue is pivotal, and I don’t think it’s properly handled here. It is certainly regrettable that the NTP bioassays were compromised in this way—but NTP itself clearly acknowledged this at the time. The white paper fails to mention that a sizable peer review panel of NTP Technical Report No. 269 discussed the epichlorohydrin issue, and concluded (see p 14 of TR 269) that the bioassay found clear evidence of tumorigenicity at several sites OTHER THAN those where epichlorohydrin had ever been found associated with tumors. The NTP panel conceded that epichlorohydrin could well have contributed to the forestomach tumors, but I note that there has never been any evidence that epichlorohydrin is associated with tumors of the bladder. We should also consider the mathematics and probabilities here. EPA has computed a unit risk factor for epichlorohydrin by ingestion—
9.9x10-3 per MKD. Now consider that a 1,3-D dose of 50 MKD was contaminated with 1 PERCENT epichlorohydrin: that puts the epichlorohydrin level at about 0.5 MKD, which would make the excess lifetime cancer risk to the test animals, from the epichlorohydrin alone, roughly 5x10-3 (9.9x10-3 times 0.5). This is 5 chances per thousand, which means that in a group of 50 animals, **the expected number of additional tumors due to the epichlorohydrin impurity would be less than one single tumor.** (50 times (5/1000) equals 0.25). This cannot possibly completely explain even the 9 forestomach tumors in the male rats given 1,3-D, let alone the 21 bladder tumors.

**I wish to strongly emphasize that it is simply improper to ignore 21 carcinomas of the bladder, which NTP called “clear evidence of carcinogenicity in female mice,” because of a trace amount of an impurity that has never been associated with any bladder carcinomas, even at non-trivial doses.** The epichlorohydrin contamination is a concern, but cannot be invoked via handwaving to explain ALL of the concerning findings in both NTP bioassays. To say, as the white paper does, that (p 6) both NTP studies have “no relevance” in their entirety is a major deficiency in the report.

(5) The white paper uses some of the same dubious logic to inflate the control response as was done with the Scott 1995 bioassay. The concurrent control rate of 18% adenomas is near the exact midpoint of the range of 6.3-36% over the 17-year period reported in Table 23. To be sure, if the true control rate was at its historical maximum (36%), the incidence in all but the 60 ppm group would not have been elevated—but why not also mention that if the rate was at its historical minimum, ALL THREE DOSE GROUPS would have shown an increase? Answer: selective and self-serving use of uncertainty. As for the “portal effect,” I would appreciate some elaboration from the bioassay panel, as I’m not sure why a carcinogen acting at the first site of its encounter in the body is artifactual. Would you call a melanoma in a lifeguard a “portal effect” and ignore it? Finally, the white paper does not mention the clear (right or wrong) guidance from EPA to treat adenomas as capable of progressing to malignancy. The authors may believe that these lesions should not be so treated, but they should not fail to mention the sound reasons for the precautionary science-policy judgment that they may be of concern.

ADDENDUM ADDED MAY 1 BASED ON ROUND 4:

There seems to be some more agreement with the point I made in earlier rounds-- that it is improper to summarily dismiss the NTP bioassays because of a mere assertion that the epichlorohydrin impurity could explain all the positive findings therein, ESPECIALLY the bladder carcinomas which are simply ignored. Commenters asked if data were available to estimate the possible excess tumor risk of 1% epichlorohydrin, and I beleve I demonstrated that it's highly unlikely that amount could have been responsible even for the non-bladder tumors.

I continue to believe strongly that the invocation of selective partial statistics about historial control rates is inappropriate. Surely the rate in the actual study is the most important data point, followed closely by the pooled rate in ALL prior studies. But to pick selected years where the rate was at its highest is no less wrong than picking selected years when the rate was at its LOWEST to claim that otherwise non-significant increases suddenly become "significant"!

I believe that several commentors answered my questions about "portal of entry" to support my concern that lung tumors in an *inhalation study* should not be ignored by invoking this "explanation."