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

Half-life is relatively short for both oral and inhalation exposure. Toxicokinetics appear to be non-linear with increasing exposure, that is, for inhalation in the mouse at or over 30 ppm. Saturation of metabolic pathways and /or depletion of GSH are contributing factors. TK appear to be similar between rodents and humans, although the descriptions are rather terse; I would like to see more explication of both the volunteer and worker studies .

Several of the graphs seem to be incompletely labeled on the y axis. Figure legends are incomplete.

Toxicokinetic (TK) concepts should be incorporated into the evaluation of safety and toxicity for any endpoint, including the issue of nonlinear TK. Toxicities occurring above the onset of nonlinear TK are highly uncertain and quantitatively of little relevance to real-world human exposures. Thus, use of an KMD concept in data interpretation and establishing a point of departure enables a more relevant and accurate risk prediction in humans.

The White paper presents clear evidence supporting the predominant role of a functional GSH conjugation pathway as a protective mechanism. Further, there is evidential support that GSH can be depleted in the lungs following inhalation exposure. Based on the majority opinion, the KMD is around 20 ppm or below (one reviewer indicated 61 ppm is a possibility).

In my opinion, the toxicokinetics issues most pertinent to the cancer WOE conclusion for 1,3-D are that:
1) Exposures are low from the relevant routes of exposure, both oral and inhalation;
2) 1,3-D is rapidly eliminated from the body, it does not accumulate, and the toxicokinetics following inhalation exposures is similar to that following oral exposures. Also the toxicokinetics in humans appears to be similar to those observed in rodents;
3) GSH based metabolic clearance seems to be saturated in the 10-30 ppm range based on the consistent results of several inhalation studies
4) The shift in linear clearance rates and changes in the ratios between the cis- and trans- isomers circulating in the blood support the use of a kinetically-derived maximum dose for repeated exposures at or below 30 ppm

The following points are offered in the context of my experience about how WOE is used to inform safety decisions at EPA, and, specifically with respect to how WOE is used by EPA in evaluating toxicokinetic studies.

Data from toxicokinetic and related studies are considered by EPA to understand a pesticide's absorption, distribution, bio- transformation, and excretion and to aid in understanding the mechanism of toxicity. Additionally, they are considered to evaluate the potential for accumulation of the test substance in tissues and/or organs as well as to evaluate the potential for induction of biotransformation as a result of exposure to the test substance.

To my read, the 1,3-D toxicokinetic studies indicate that the substance does not bioaccumulate, that it is metabolized and detoxified similarly by the relevant routes of exposure (oral and inhalation), and that it is metabolized in essentially the same manner in the rat, mice and humans.

Of prime importance to EPA is that toxicokinetic data can be used to assess the adequacy and relevance of the extrapolation of animal toxicity data (particularly chronic toxicity and/or carcinogenicity data) to human risk assessment. EPA recognizes that it is important not only to address questions regarding absorption, persistence, or distribution of the test chemical, but also to consider alterations in the metabolic profile occurring with doses that may be of toxicological concern, and that if such alterations do occur considerations of risk should be adjusted accordingly. The studies described in the 1,3-D white paper suggesting non-linear GSH metabolic clearance, and the identification of a KMD, are consistent with EPA's approach to weight of evidence evaluation.

Metabolism data can also be used to help determine whether animal toxicity studies have adequately addressed any toxicity concerns arising from exposure to plant metabolites, and in the setting of tolerances, if any, for those metabolites in raw agricultural commodities. This concern is of minimal importance for 1,3-D, because out of all the exposure scenarios from its use, the only post-plant food use is from vineyards. Further, according to the EPA HED Human Health Risk Assessment document (2008) human risk from oral exposure to 1,3-D is negligible in that it is less than 5% of the population adjusted dose (PAD) for all population groups for food plus water from surface water sources. (To clarify, for pesticide registrations under FIFRA, EPA derives acute or chronic population adjusted doses (PADs) using an FQPA Safety Factor mandated by the Food Quality Protection Act taking into consideration potential pre and/or post natal toxicity and completeness of the data with respect to exposure and toxicity to infants and children. In the majority of instances, the PAD and the RfD are the same. FQPA Safety Factors can account for uncertainties in the overall completeness of the toxicity database, extrapolation from subchronic to a chronic study duration, and LOAEL to NOAEL extrapolation.) In its evaluation of the WOE of the toxicokinetics of 1,3-D, EPA would clearly note that the combined worst-case oral exposures are less than 5% of the PAD, and thus would be considered to be of low to negligible risk for the general population.

Similarly, in their WOE evaluation of inhalation exposures to 1,3-D EPA would look at the modeled 50%, 90% and 95% risk calculations captured in table 31 of the 1,3-D white paper and the margins of exposure between these exposure scenarios and the chronic point of departure for 1,3-D of 3730 ug/m3 ambient air concentration. The greater than three orders of magnitude ratio between the chronic POD (or dose-response point that marks the beginning of low dose extrapolation) and the highest estimated exposure concentrations would also be considered to be low to negligible and likely lessen safety concerns during the WOE evaluation for the general population.

Similar assessments would be done for mixer/loader and applicator subpopulations, as well as any other identified key sub population that could potentially be at greater risk because of higher exposures.

The toxicokinetic studies are quite conclusive for similarities between mice, rats and humans. The non-linear region for dose response of clearance begins around 20 -30 ppm.

[small prefatory comment: I like to emphasize important words/clauses with italics, underlining, or boldface, which this interface does not seem to support—so please forgive the very occasional use of ALL CAPS—I am not trying to “shout”…]

The white paper and the subpanel discussions make a convincing case that GSH conjugation is a driver for 1,3-D dose-response in rodents—but they fail to demonstrate whether or how this knowledge should affect hazard classification and risk assessment for 1,3-D in HUMANS.

First, the logical connection between metabolic saturation and the presence AND LOCATION of a threshold has not been established at all here. Basic mathematics (see, e.g., Hattis 1990—Risk Analysis 10(2): 303-316) dictates that even when exposure is time-invariant, saturation of one of two pathways does not guarantee that the dose-response function will show a threshold. It may cause the function to have two different slopes (which one could “connect” and say that the entire function is convex, or sublinear), but the low-dose behavior may still be linear or at least monotonic. More importantly, the tests to date are only relevant for steady-state kinetics, NOT for intermittent exposure with the same long-term average. As the first panelist’s answer under Question 3.6 states well, “the KMD may significantly vary with the dosing scenario.” For example, even if the carcinogenic metabolites of 1,3-D are only produced at all when GSH has been completely depleted, we currently have no idea how many metabolites might be produced if “30 ppm” on a given day/week/month resulted from continuous unvarying exposure or from 120 ppm for ¼ of that day/week/month and zero exposure at all other times. It is unacceptable to claim that “the risk only exists when exposure exceeds some KMD” without exploring what happens when transient exposures above the KMD contribute to an average dose below the KMD.

In addition, there is zero information in the white paper or panel comments to date about how we might analogize from rodents to humans here. (see footnote at end of this section). Even if ppm is the proper dose metric to equate responses across species, what about the substantial interindividual variability in human GSH kinetics (see, e.g., Takamatsu and Inaba (1994), Toxicology, 88:191-200, finding 20-fold human variability in GST activity towards various substrates)?? Surely 300 million outbred humans exhibit far more variation than a couple of dozen inbred rodents. So what is the DISTRIBUTION of KMDs among people? We have no idea, and therefore no idea how to assess human risk in such a way as to provide acceptably low levels to 50%, 90%, 95%, or any other fraction of the human population.

I also find the drawings on pp. 11-13 of the white paper to be quite unconvincing. Where is the goodness-of-fit analysis that proves that a “bent” function is significantly superior to a linear one? Where is the uncertainty analysis that shows the confidence limits on the LOCATION of the alleged “kink” in the function? And of course, all the points in the previous two paragraphs also apply.

Finally, the interesting observation on p 36 of the white paper—that in rats, particular doses by gavage were non-tumorigenic whereas identical doses in diet were—deserves more thought, as it could explain much about the disparate bioassays. But this doesn’t explain the comparison (same strain this time) between Kelly 1998 (negative by gavage) and NTP 1985 (I will write about the NTP bioassays in another section), or between Redmond 1995 (negative by diet in mice) and NTP 1985 (positive in mice by gavage). In any event, there are two possible and *opposite* explanations for the lack of tumorigenicity in some dietary studies: yes, effective doses may be lower from dietary exposure (in which case the negative bioassays suffered from lack of power), but it is also possible that the “more severe toxicity” via diet interfered with the cancer bioassays. And the lack of TK studies by inhalation is a serious deficit.

Note-- there is one paragraph in the white paper about human TK, on page 9. But all it references is one human study for 6 hours at 1 ppm. I see no way that this exposure could shed any light whatsoever on GSH **saturation** at doses much higher than 1 ppm, so I discount this one minor reference to human kinetics.

ADDENDUM ADDED MAY 1 BASED ON ROUND 4:

The commentary has firmed up my prior belief that the White Paper convincingly establishes the GSH pathway as of primary importance, and that GSH can be saturated *in rodents* at exposures around 50 ppm. But the comments have only reinforced my belief that the White Paper does NOT:

a) provide any basis for deriving a KMD for the typical human being, let alone a "KMD with variability" that would allow risk assessment to account for outbred humans rather than inbred mice and rats;

b) acknowledge that the analysis of Figure 4 *at most* establishes that there is an important nonlinearity in the dose-respose for 1,3-D-- it does *not* establish that there is a THRESHOLD for 1,3-D carcinogenicity. The White Paper conflates these two very different concepts.

c) account for any possible impact of intermittent exposure in humans. IF GSH saturation is required for the dose-response relationship to be steep rather than shallow (again, I am not at all convinced that the dose-response has a threshold, only that it has a nonlinearity or "kink" in it), then a risk assessement useful for humans MUST model the effect of intermittent exposures above saturation in the context of a lifetime average dose that is well below the saturation point. The exposure estimates at the end of the Paper do suggest that average exposures are very low relative to exposures of concern, but they say nothing about transient exposures.

It is reasonable to assume that 1,3-D interacts with and is eliminated by conjugation with GSH. Data support this view for both portal of entry and systemic doses. Data include information on tissue depletion of GSH in target tissues and by-products of metabolism in the urine. This metabolism results in non-linearity of clearance and increased dose to targets if GSH is depleted. This is an important finding for selection of doses that have not exceeded the KMD. Evaluation of the data suggest non-linearity below 30 ppm based on blood data and lung tissue depletion. While this observation is important for evaluating the animal cancer results, it may have limited use in cross-species extrapolation where human data on tissue levels of GSH and elimination kinetics will be required. required