| Sponsored by: International
Society of Regulatory |
QUESTIONS FOR PANEL 1
DR. STARR: Let me ask the people who presented this morning whether or not they have any comments or questions to make, having heard what everyone had to say. DR. KERGER: I had a discussion with Rory Conolly about the pharmacokinetic modeling differences between the rat and the human studies at the break. I would like Rory to comment if he has anything to expand on as far as my explanation. Not being really from a PBPK background, I am really going on what I observe in general. DR. CONOLLY: Brent Kerger and I had a brief conversation at the break. One of the comments I made during my presentation was that PBPK modeling, in general, was, I thought, under-utilized in the assessment. There are good reasons to think that some of the issues that Brent discussed, such as maybe having a multi-compartmental or a multi-phasic elimination of dioxin could be an important factor to consider when you have to extrapolate back in time 20 or 30 years to estimate a dose at some point in the distant past. This is where a PBPK model could give you valuable information that just would be totally lost with the kinds of dose reconstruction methods that have been used. I think, yes, that is something where we could probably do a better job than we have done so far. I can make one comment that perhaps I didn't clearly state during my presentation. I think as kind of a general or over-arching comment, it is worth making and it may be obvious to everybody now. In spite of the amount of effort that has gone into this assessment and the length of time that it has taken to pull it together, that in many ways it is still very much a default-driven risk assessment. The fall back to the low dose linear extrapolations, in particular, is strictly a default approach to cancer risk assessment. We have become very interested, in recent years, in how that default may actually be inaccurate when you look at specific chemicals and examine their mechanisms of action. Again, we could spend all day talking about whether we could really escape from the defaults for dioxin or not, or whether we have to consider the defaults alongside other options. I personally think we certainly could consider the defaults alongside other options. It is just interesting to me that, after so much time and effort, we still are presented with what is largely a default-driven assessment. DR. STARR: I have one question of either Rory or Brent that relates to some of the points that were made. The question is on area under the curve as a measure that might be more appropriate. What time scale do you use when you extrapolate across species? Is there a rule of thumb that one would recommend? DR. KERGER: I don't think there is a rule of thumb. I think because almost all the data that we can extrapolate from, as far as dose-response, are going to be retrospective, that AUC, while it is probably the ideal estimator for knowing the dose-response relationships for receptor-based mechanism chemicals, is probably not a knowable number as much as perhaps the predicted peak exposure. We may end up using something like peak, like when we see at Seveso 76,000 PPT is a person's blood level soon after the exposure, use that as an indicator, just combining that with the generalized considerations of persistence and longer half life, it seems to me to make more sense as a dose-response metric, based on the historical data that has been collected already. DR. CONOLLY: Just to add to what Brent said, and I certainly agree with everything he said, I think the difficulty in a measure of dose like AUC or average body burden is that, if there are non-linear processes in how the body responds to an insult of dioxins -- and there certainly are -- sort of simplistic measures of dose are not going to capture that very well or predict it very well. That is where the most advanced PBPK-type modeling would be very useful. That will actually define dose on a minute-by-minute basis throughout every day or every week, every month, every year of an animal's life, whether it is a rat or a human being. One of the advantages of a more advanced approach to the dosimetry is that it would relax some of the simplifying assumptions that get made in the assessment about the description of tissue dose, such as AUC or body burden. MR. QUILL: If I may ask the panel a little less technical question -- this is to the entire panel and Brent, you went through this very well -- EPA sets forth criteria for including substances into its toxicity equivalence approach. As I mentioned before, that is the way the Agency is able to make the conclusion, part of the way the Agency makes a conclusion, that background exposures are at or near a level that poses adverse health effects. The four criteria are, again, age, receptor binding -- that is easy enough to mention -- that they are structurally related -- well, we know the structure of the compounds, that is pretty easy -- persistence -- I think your comments were right on and I think we are going to hear more from Steve Safe on that. Let's get to this whole idea of the fourth criterion, they induce a common spectrum of biological responses. I am wondering, if EPA had done a chart and, on one side, had all the congeners, the dioxins, the furans, the PCBs, and in fact, let's throw isoflavones and other cyclic compounds on there that bind the Ah receptor, if you have that on one side and you have all the potential toxic responses, what would that chart look like? Would it, in fact -- would these other substances in any way look like 2,3,7,8 TCDD? A second question is, would the isoflavones and the PAHs then look like some of the other congeners that are now included in the scheme? DR. KERGER: My only response to that would be that saying that it has a common range of responses becomes a very vague criterion, when you say that dioxin causes all cancers and that it causes all these other diverse range of effects in exposed children, all the way up to the occupational cohorts. I could see that -- you know, the level of evidence for PAHs certainly is more specific, even though it is theoretically an Ah-active compound. For isoflavones, I think the evidence for human effects is really a whole different story as far as the level of certainty that we have about those compounds. I don't have a good answer for you other than that idea of those words being a criterion is sufficiently vague that it is hard to make a distinction on, in my view. DR. STARR: I have a speculation about that matrix, Terry, and that is that it would be filled with a lot of empty space. A lot of the congeners, including truly dioxin-like ones, really have not been tested for the full spectrum of effects that are sort of identified as possibly being caused by any one of them. It may be that the requirement that any one of them demonstrate the full spectrum is asking too much, in the absence of good data, to indicate that. DR. SAFE: Because it is a receptor-mediated pathway, you will never get the full spectrum of responses for dioxin-like compounds as compared to non-dioxin-like compounds. Most of the non-dioxin-like compounds are weak agonists. In order to get a strong agonist response, you have to overload that animal so much that you can't see it. Because it is receptor mediated, the differences between weak and strong agonists can be pretty important. However, it is also true that within the dioxin series -- and this has never, I don't think, been taken into account, the octa and the hepta chlorinated dioxins are also weak agonists. It is probably true that they won't mediate the same spectrum of toxic responses. They may not mediate cancer because you can't get enough of them into the animal. That has really never been taken into account. JEFF SLOAN, Chlorine Chemistry Council. I guess my question is mostly for Tom Starr. EPA makes the distinction between mechanism of action and mode of action. In the reassessment, it acknowledges that there is not enough knowledge to define the exact mechanism of action but does draw some strong conclusions about the mode of action. I am wondering what your thoughts are on EPA's distinction between the two and what they say about what is known about the mode of action? DR. STARR: It is my opinion that we don't know enough to identify a single mode of action. Part of it, receptor binding, is certainly something that makes sense and there is a lot of evidence for. Once you get beyond simple responses like enzyme induction as a consequence of receptor occupancy, we haven't got the foggiest idea of what modes might be involved. As I showed today, even cell proliferation, which is thought to be the mechanism by which promotion produces tumors as a consequence of dioxin exposure is not well correlated with exposure. So, it seems to me like there are a lot of speculative hypotheses of various modes by which dioxin may cause cancer and none of them are well supported by data. Would anybody else like to chime in or comment? DR. KERGER: Ditto. DR. SCHEUPLEIN: As most of you have mentioned, one of the new things about the dioxin reassessment model 2000 is an emphasis on a body burden rather than an emphasis on daily intake. Dr. Conolly basically had said that he believed that they could have used more pharmacokinetic information in their risk assessment. I would like to ask him if he really believes that there is data that is sufficient now to make that an effective addition. I hope there is, but it almost would seem to me -- this is an old concern. We have known that the dioxin half life was seven to eight years for a long time and we have also known that most of the dioxin gets into the fat. I remember Renata Kimbrough saying at one meeting, well, that probably means it is sequestered and that the human is protected a bit because we don't think there is anything going on in the fat cell except one fat cell. The question is, can your pharmacokinetics basically estimate, from the data in fat what the data in tissues is going to be, what the data in the cells is going to be. Those are the concentrations that the receptor is going to be proportional to. DR. CONOLLY: That is exactly what physiologically based pharmacokinetic models do. I mean, they compartmentalize the body according to tissue type. So, you have a fat compartment and a liver compartment, kidney and so on. You describe each of those compartments according to physiological characteristics such as volume and blood flow. You also describe them according to things like the solubility of dioxin. In the model, you distinguish the great capacity of fat to store dioxin versus other tissue like liver. You can also, in these models, describe things like the induction of the cytochromes that bind dioxin and may well account for a longer half life at lower levels of concentration, the kind of thing that Brent showed. So, I think that the answer to your question is yes. This kind of technology can do a lot to help us, both interpret existing data and, just as important, to make predictions for risk assessment in a way that is more closely tied to our understanding of the pharmacokinetic mechanisms. DR. SCHEUPLEIN: Do the data exist? DR. CONOLLY: Sure, I would say it exists. DANIEL BYRD, Consultants in Toxicology, Risk Assessment and Product Safety. Having one foot in both camps, toxicology and epidemiology, I would like to ask a question for all the panel, but it is a little different question for the two groups. It goes as follows. It is true, looking retrospectively at exposure is a hard thing to do. It is true that we can bring more to it than has been done so far. Having gone through several epidemiology studies myself, you can almost always generate, for each person or each group in the study an estimate of average exposure, using something like, you know, average exposure times time, and some estimate of peak exposure. To me, a minimum requirement for fitting any model, even if it is a policy-based model, to such data is to inquire whether you have a better goodness of fit with peak estimates of exposure versus average estimates of exposure. I am not aware that that has been done for any of the epidemiology studies that you all have reviewed this morning. So, that is my question to Dr. Mandel and Ms. Aylward. My question to the toxicologists is, if those kind of data exist and we look at the goodness of fit, shouldn't we make something of them toxicologically, based on the data we have about half lives? MS. AYLWARD: I will start on that. I think there are two parts to the answer. The first is that the epidemiology is fairly rough and the estimates that you make of exposure per person or per subcohort are quite variable. The second part of the answer is that those various measures -- peak, AUC, average concentration, for a given type of exposure are very highly correlated. You don't wind up with a high peak and a low AUC for a given person or subcohort, they are very highly correlated dose measures. You cannot distinguish very much between a fit based on an AUC dose metric versus a peak dose metric versus an average dose metric. I think it is important to keep those various dose metrics in mind when you are looking at the different types of studies, though. When you are comparing studies of an accidental exposure, for instance at Seveso where you have a very high peak exposure and then a long elimination period versus a steady-state background exposure where you don't experience a high peak, where your exposure accumulates gradually over time and you have a very flat curve, those two are very different and may have very different physiological consequences even though they may not, over time, show up to be quite as different as you might think. Looking at peak exposures is important from that point of view, but I don't think the epidemiology is a precise enough tool, and those various dose measures are so highly correlated that you are not going to see biologically meaningful differences in fit in general. GEORGE CLARK, Xenobiotic Detection Systems. I want to say that I agree with the panel, that PBPK modeling should be done. I want to bring out one technical aspect that makes it difficult for epidemiology studies. You have got to realize that the dioxin concentrations are in parts per quadrillion in human blood. To be able to even measure it, you have to use 50 mls of blood. There aren't too many people who want to get stuck every few moments for 50 mls of blood, and it costs a lot to be able to analyze it. So, there is a tough technical difficulty that makes it also very difficult to do the PBPK modeling. BOB GOLDEN, ToxLogic. This is a question for Jack Mandel. We all know that one of the sort of odd things that has been done with the dioxin epidemiology is this cancer at all sites combined. My question is, what if you look at some other chemicals that are classified as known-human carcinogens because of the association with cancer at a particular site, and do the same sort of exercise? I mean, they are not viewed as pan carcinogens, but if you do whatever is done to get a cancer at all sites, would you find that sort of thing? Could you do that? DR. MANDEL: Actually, we just completed a review of all the studies that have been done on chemical workers in the United States and western Europe back to 1964. I think there are 481 studies we reviewed and did a meta-analysis on. In fact, as you would expect, because there are some known carcinogens in the work place -- arsenic and lung cancer, aromatic amines and bladder cancer -- you see some expected elevations. The overall cancer rate is not elevated. I believe the overall SMR was something like .99. I don't know of any chemical -- and we tried to identify one -- that causes all cancer elevations. Now, someone, I think in fact IARC suggested smoking, but smoking is not one chemical. There are multiple chemicals involved in tobacco. The other one they mentioned was radiation, which is not a chemical. So, the answer I would have to give is no, I don't know of anything. DAVID FISCHER, Chlorine Chemistry Council. My question, I guess, is perhaps to Tom or anyone else who would like to chime in and answer. It seems as if the TEF scheme trumps the cancer guidelines. That is, the cancer guidelines make an effort to -- and EPA's application of the cancer guidelines makes an effort to distinguish between the carcinogenicity of TCDD and the other dioxin-like compounds. On the other hand, the TEF scheme essentially says that all compounds within the TEF scheme have the same health endpoints. It is just a matter of how potent one compound is to another. I guess my question is, how can the reassessment essentially reconcile this tension? I think it was sort of picked up at the July peer review group, although unfortunately it wasn't really discussed at length. There clearly is a tension there. I don't know how it is reconciled and I don't really know what there is to do about it, when you have cancer guidelines that you are supposed to apply, and on the other hand, it is tough to reconcile that with the TEF scheme that EPA also embraces. DR. STARR: There is one way to reconcile it and that is to call TCDD a likely human carcinogen. Then we can all go home. I don't understand what the reasoning is for pushing so hard on calling TCDD a known-human carcinogen. You get into this quandary where Ellen Silbergeld challenged Bill Farland, how can a mixture that is 10 percent known-human carcinogen be classified only as a likely human carcinogen? It is an indefensible position and I don't know how it can be resolved with the position that EPA wants to take, but that is what they do. [Click Here to View Slides] |