 Okay, welcome back. We finally got the slides uploaded. And we're going to start with a presentation by Chris. Chris and Holger have been working on a hazard index approach for determining the cumulative exposure to phthalates using biomonitoring data. And so we want to walk through this and open up discussion among the committee members. Chris? I'd like to start by saying this is just draft form. So please keep that in mind as we go through this. Okay, I'm pushing arrows. What's happening? Oh, I'll do that. Ah, there you go. Okay. So just to remind us, the form of the hazard index is given here. And so what we're going to do is to estimate the daily intake from NHANES data on a per subject basis. So we'll kind of step through that approach using some work that Holger has done. The reference dose is we're actually going to look at in two different ways, case one and case two. Case one is going to be based on the approach that Andreas presented to us on our first meeting and then a second approach that we kind of put together. Okay, so our objective is again to look at the distribution of the hazard index. So again, the approach is going to be not to estimate the daily intake at the median level or the 95th percentile, but it's on a per subject basis what their real exposures were as estimated from their metabolites. So we're going to start this in two ways. The first way, or the first thing we did is we included seven different phthalates. It's the six phthalates, but we added the diisobutyl phthalate, which is not, I think, on the list of our six. And then we're looking at additional androgen, anti-androgens, the bisphenol A butyl paraben and propyl paraben. Again, there may be others that we should add, but it's just a demonstration of what we could do to add additional chemicals to see the effect of additional chemicals besides just the phthalates. Okay, so we're going to use the NHANES 2005 and 2006. Now from the discussion this morning, I'm wondering should we actually go back and do this on the 2003 and 2004 data because of the fasting? I mean, that may be something that we can think about. Right now we're looking at 6 to 19. It's really less than 19 years old, so we're trying to focus on the children. And then the two cases, like I described before. Okay, so this is the meat of the work. The work from Holger has done, calculation of daily intake from these different metabolites. So maybe I can just hit some highlights here in Holger if you want to add things to that discussion. So the daily intake is estimated from the urinary excretion estimates that come from the actual metabolite values that we can measure from NHANES. The smooth creatinine excretions. We're doing this for children. So there's this paper Remer et al. So that the creatinine excretion rates are based on height and gender-based references for the smaller kids. And then we have this tabular molar ratio of metabolite excreted. So this fraction excretion factor we're going to look at. But some of these values we actually took from the literature. Others we just sort of set because, again, just to push the exercise through. And if we could get more information about those values that you set, I'll show you a table of that in just a minute. The body weight is given for an individual child. Molecular weights, we can calculate as well. Okay, so this is the list of the phthalite monoesters that we use. They're categorized by the components that would go towards a parent compound. And so these are the excretion factors that we use. Some of these are set at 10%. So if we could get informed values there. So essentially what we do is, for example, for DEHP, these would be summed to get a total for that parent compound. So we're going to use these values in our calculations. We also have some values here that we're assuming for these other anti-androgen. So we're going to look at this as a case with the phthalates and then with these additional. So a different case in addition to that. Okay, so this is a place that we could come back and inform what we're doing. Okay, so I've mentioned this a couple of times. But the second case here is where we looked at some of the work by Earl Gray and made some assumptions and we said, suppose that we assume that these four are at least approximately equipotent. And we look at the literature and find a NOAL for DEHP of 5 and then use a relative potency approach to get to the DINP value. When you do that, and these are now values that were published in this paper that we'll see in the next slide. So this side of the slide is essentially what's in Andreas' paper. I think we set these that wasn't in on, those were not considered in that paper. This is the second case where we assumed that equipotency and then we got to RFD values there. And I guess the DEP, we set it at 800. And standing here, I can't remember how we got the 800. So some of the details, maybe we can go back, I don't want to belabor that, but some of the details here maybe we could reconsider. Okay, so again, the approach is if we could look at fixed reference levels, reference doses, and then on a per subject basis estimate what their daily intakes are for each of the phthalates that we're looking at or the anti-androgens that we're looking at, and then come up with a hazard index per subject and then look at the distribution across subjects. And we can, I think an advantage of this approach is we can actually kind of go in and simulate things with lack of data that we know we have here. I think this approach is going to address variability across the population of kids that we're looking at. Again, this is not weighted with the in-hanes weights. These are just the values which I think some of the other speakers did this morning. But the particular subjects that are here we're looking at, it was 950 kids aged less than 19 years old, just 12 tonight, 12 is the youngest that in-hanes measured. 52% were 6 to 12, 48% are boys. The race, remember in in-hanes it's spiked, it's not just representative. And so you see the percentages are there. Okay, so what this is then is going through this calculation like I pointed out on a per subject basis. So it's like every child had their own hazard index estimated from the daily intakes. And what you see is a distribution of the hazard index here. It's off to the right or left whichever way you're looking at it. The values here may not be able to read, I can barely read them. The mean is like 0.18, the median is 0.08, I think. The maximum is 6. When I take the log base 10 of those values you get the distribution here. So now what's 1 there is 0 here. So we see that we are in the tail where beyond, if we're concerned about 1 is a cutoff. I don't know that we've actually identified that that's the point where we're mainly concerned if we like it. I think 1 is generally what people are compared to. So this is now the case where we use the reference doses from the Quartenkamp and Faust paper. So then the question is what happens if we change the reference doses because it changed the distribution drastically. So this next picture, these are the distributions now from that case too. So now the reference doses have changed slightly. But we still see somewhat similar shape in the tail part being above 1. The maximum value now is much bigger, instead of 6 it's 14. I don't know if we want to worry about that. Now this is an interesting slide I think that we can kind of look at. Let me see if I can teach you what's here. So if you think about an individual child has their own hazard index. So you can go back and figure out, well what's the percentage of that hazard index that came from DBP or that came from DEHP or that came from whatever. So there's a percentage per child from their hazard index. What you're looking at here is like the average percentage across the subject. But I think this part is also interesting, the max and the men. I think what it's telling us is although in case one 72% the average DEHP on a per child basis was a 72% of the hazard index. However, there was somebody with only 11%, there was somebody with almost all from DEHP. I think this gets to the variability that we were talking about earlier. Interestingly, when we change from case one to case two, what we see is that DEHP still ends up being the highest percentage but now DINP ends up showing up more. This range I think is just amazing that on the average it might be DEHP and DINP are the most common in terms of the hazard index but we're seeing children there that have the complete range for each of the chemicals. So I think that's kind of an interesting thing to think about. We can come back to these tables but I think these tables can be helpful because we can actually now simulate scenarios of if we alter some of these exposures what happens to the hazard index and things like that. So I'll show you a little bit of that coming up. I'm assuming you guys will ask me questions if I stumble too fast. Then the other thing we can do is look to see who are the children that have these extreme values. So this is on the log 10 scale, the hazard index, and what you're seeing is from case one. These are across the males and females. This is across the race. You may not be able to read that. Mexican American, other Hispanic, non-Hispanic white, non-Hispanic black, and other. So here again, the zero would be the reference line that would be corresponding to hazard index of one. So we're seeing just about in each of those groups we're having children above of value one. And here it is broken down from males and females across the races. And then here's the age group. So the younger children tend to have larger values. This again is on the log 10 scale, but you can see the distribution is up a bit there. Okay, so this is now where we can start to play with what's there. Back to case one. Now if we add the other three anti-androgen. So it's not just the phthalates. Now we're adding three additional ones. And this is the thing that concerns me is three is just a very small number based on what that could be. And I don't know how far you go and you keep adding. I mean you keep adding, it's going to get bigger. So the distribution, I think that the mean before was like 0.18. Is that what I said? Now it's like 0.207. So it's increased a smidge, not drastically, but it is increasing. Which I guess you would expect if you add more chemicals to the assessment. So the planning of what's going to be in here I think is going to be an important step. Okay, so now thinking about the fact that these are children 12 to 19. Sorry, six to 19. Sorry I said that wrong. So what we don't have are children less than six. So we said well suppose there are behavior patterns or whatever for these children. Could we actually try to simulate what would happen to children lower than six? And so we just said and this could be better informed. But for example we could say suppose that there's a 50% increase in the daily intake for DEHP based on behavior of mouthing or whatever. So what happens there? So the 50% came from a paper that Holger had published. So that didn't come right out of the air. But maybe this could be informed with other chemicals as well. So with that kind of hypothetical case we're saying now suppose this is now the daily intake for children smaller or younger than age six. And so now we can look again at the distribution and we see that it's increased again from up to .239 as a mean. And again the distribution is still over the value of one. And this is now the case where we said now if we could study some kind of a substitution effect. So we said well suppose DINP is actually substituting for DEHP and DEP is substituted for DNBP. Now how realistic that is we need to be informed about that. But so we just essentially changed values from DEHP to DINP and recalculated things. And these are values that have lower or I guess higher reference doses. And so now you see the distribution has actually shifted in the direction we'd like for it to be. So the point of this is this may not be a perfect example of how to do the substitution. But I think it is a way of saying we might be able to with the right input be able to simulate what would the effect be. What could the effect be and see how far we could really make it shift down the direction that we want it to shift. So there's still a lot of work to be done on this. I don't want you to think that this is a complete story. But I think the advantage of this is we do get to see variability from relevant real exposures in children. That we can do some simulations in terms of scenarios where we may not have complete data. We might be able to say for example suppose each of these chemicals a certain percentage comes from food and the other percentage comes from certain products. What's the effect of how much of the hazard index is actually based on the food part versus the other part. So we can look at things like that by having this kind of simulated case. Of course there are a lot of pitfalls. I just quickly just wrote down that we haven't corrected here anything for fasting. We can try to do that. We need more accurate inputs and there's probably a whole lot of other pitfalls. But from a case where we don't know much maybe this is an approach where we can start playing some, simulating some. That's that. Okay. So now discussion. Chop in. A couple things that I like about this. One is that the first is that you're not adding up the 95th percentile of each phthalate because that's probably a way over estimating risk. You're actually getting the 95th percentile of the total end result of exposure to multiple phthalates. The other that's the main thing the main advantage of this. But the other other thing is that we don't really know if people's exposures to different phthalates are correlated or not. But you can well your analysis doesn't require us to assume that they're independent or correlated. It's just it's what the data are on the other hand you can maybe test that. I'm not sure we need to do it but it might be of interest to know to look at the data and tease out are the different phthalates correlated. Because that's I think something came up in interpreting the epi studies. You know one possible confounder in one of the curious findings was that ethyl phthalate correlated with some of the health effects. But it's ethyl diethyl is not active in the animal bio assay. The question is is it really the diethyl or is it the other or is the diethyl just a marker for total phthalate exposure or something like that. So there's there's a lot of things about this a lot of advantages and it could be applied if we do modeling of exposures it can be applied to that. Yes. Great. Thank you very much for this helpful paper and helpful presentation. And my question follows on from Mike's remark but just maybe the answer is clear already but just to make this absolutely clear. The ananes database allows us to see what the urinary levels of phthalate of multiple phthalate metabolites are in one at the same urine sample. Is that correct? Yeah. Correct. Good. Then then that is indeed that's indeed a really a strong point. The only question I have relates to the potency values which you took from Grace data in your case number two. If you wish. Yeah. Just to to see. They're based you say on suppression of fetal testosterone synthesis. Do you have an explanation for the differences between case one and case two? I should probably know myself but I don't. As far as I remember what what Earl's Grace studies are they they are not designed to derive points of departure. That's not a criticism is not intended to be a criticism of Earl's work. It's just he would admit that freely himself. So that means that if someone went away and did a guideline study with sufficient numbers of animals on with that end point the estimates for point of departures would go down anyway. So but this is interesting to see that your estimates went down. But what can you just give a few details of Earl Grace study which you based this on? I think it actually came from some testimony that the Equipotent part came from a testimony. The five came. I don't remember what Holger do you remember where the five came from? So first of all the both cases have their pros and cons and studies you used basing your point of departure upon. They also are in study design. They have considerable differences. So that's the topic on the case one side on the case two side. So that's major mainly the studies by Earl Gray. They are based on this multi-dose EPA approach. And as you said they are not really designed to derive TDI or let's say a point of departure. But the findings at least from the effective doses show that and this is also the result from last sessions or last meetings. Earl Gray's presentation that he considers these studies as roughly equipotent. So this has been let's say a rather provocative approach on our side to present this rather let's say differing data basis data. But we thought it's just an illustration how potent the approach is that we can as soon as new data is present implement this data instantly and have a look what's coming out. Well and I think interesting too the distribution of the hazard index didn't change. It didn't like move an order of magnitude to the left or the right with very different values I think here. Now what the thing that changed the most I think was the distribution of which of the components was playing the major part. But the index itself was about in the same. I'm quite happy to see this and I don't want to I don't think it's worth me waxing lyrical about this. I mean we've said in this paper ourselves that very likely our potency estimates are not conservative enough and I guess what you've done bears this out. Well I think you know it might get to the point maybe the maybe I don't want to over interpret what Tom was saying but the whole issue of you know you're going to spend a lot of time worrying about tweaks here and there. What I'm seeing is the robustness of you know the index didn't change that much to two very very different places I think that. My very last question is in the light of this morning's debate did I don't think did you adjust those data for fasting. When I when we worked on this I didn't know how to do that yet and I'm still you know I think I'd be up for suggestions about how to do that. Again you know maybe what we can do is to present this in multiple ways adjusting for fasting not adjusting for fasting adjusting in different ways. You know what happens to the index. At the end of the day. Anti-indigens. But I think from this morning's presentations we can conclude that we have to incorporate a an adjustment factor of at least let's say two or three so we are in other words with this presentation here. Not dramatically but significantly underestimating exposure. Yeah. So but see that's my question so for the for the folks who have a very short fasting period is their value actually pretty good. It's only the people who have fasted a long time where we may be underestimating. Yeah. And we have seen that especially the high values the upper percentiles are influenced by fasting. I think that's both what Matt showed and what Rick showed. But if we went in and said you know for I mean there's multiple ways of doing this but suppose we went in and said if you have a fasting time bigger than four hours or eight out you know whatever the number is we're going to increase the exposure estimate for some of those chemicals that are based on food by right. And that's those are the kinds of steps that I think you know if we could agree have it how to do that. I think that would be let me try to walk through this as a as a risk assessor. These are based on animal studies which are feeding studies I think or are they gavage. I think they're feed and the ant. So those doses you're pointed departures and so on are based on average daily exposures. And with the NHANES data the thing about the adjusting for fasting if yeah if the people have fasted their metabolite levels are going to be lower but if you're going to compare to average daily value what you really think is the average daily exposure. And so when you adjust for fasting I mean are you adjusting to get the peak exposure or are you adjusting to get the average exposure. I think the first point I think we can agree on is that no matter no matter which approach we choose this data is underestimating exposure exposure. If we are underestimating peak or mean exposures we might have to talk about. But we definitely are underestimating exposure and we've seen that we are probably more dramatically underestimating exposure if we look at the upper percentiles. That has been the data shown by Rick and the data by Rick and there are two more points I would like to put the focus on. Currently our investigation stops at the age of six so we'll have to extrapolate the data to children younger than six and the data currently does not include pregnant women. There's only children. There are no pregnant women in NHANES is that true? No. But we have let's say adult data or we have data of women in their reproductive age. You're saying we do or? I think we don't. We don't have actually pregnant women in there but we have women in their reproductive age in NHANES. Definitely. Which in a sense maybe just even more important. Exactly. I mean we do have there's not a lot of data but there are some smaller studies where you have younger children and pregnant mothers or their mothers. And I think we should look at those data before we start extrapolating but we should get all those data together and take a close look at that. That's what I was going to say also. There's probably four or five data sets that would have maybe data on I would guess two thousand pregnant women. And maybe a hundred data points on children less than three or five I think. And what kind of data? Valley? No, not in NHANES. These are other publications that like Children's Centers, Columbia or other groups of Mount Sinai and others have published on phthalates in pregnant women during different trimesters and looked at outcomes. And I'm aware of I think at least three or four papers that have done that. And in those papers that are spot samples they don't ask the subject to fast but they also don't ask the subject when their last meal was or how long they were fasting or the time between their last meal and the sample. So that data I'm sure Hoga you're familiar with that too. Yeah I mean to have any data on children three and under is would be extremely helpful. Yeah well even maybe since then but. We might need to ask them for the raw data because for this approach we need the age, body weight and height of the children which is on an individual basis. We can definitely ask and I think there's a pretty good chance that someone will share those data. My question is really for Mike. Anticipating what you're going to be seeing later on this afternoon. Do the data that you have fit into this approach to help inform us of our answers, our recommendations? Well I think it complements this. It's the bot you know this is the top down and I'm going to talk about the bottom up and I think they're complementary and I think you need both. You know one thing with the children people do see higher levels, metabolite levels in the children than the adults. And the question is why is it because they eat more or some people think it's because they mouth toys but pound per pound they probably eat more too. And you know so that's something we want to get it would like to get a handle on. Well the estimates in terms of daily exposures the estimates we made for the toys are fairly small. But I'm not sure how they you know what percentage of the total lay account for. The problem with making those kinds of comparisons is if you have good data for one source and not so good data for another source those percentages could move around a lot so. But we can certainly try to do that. What kind of units are they in? Would they be comparable to this? Well I'm talking about micrograms per kilogram per day or you know we could do micrograms per day. So you could actually say them based on mouthing or based on exposure to is that what you have on a child basis? Well what we did is made population estimates for infants just mouthing for mouthing the toys. We estimated the daily exposures per parent compound. Well we only did it for one but yeah per parent compound we only did it for one but we can extend that to the others. So that would actually give us at least an estimate of a percent. It would be an estimate of what the exposure is from using the total for mouthing toys. I think that would be helpful. Yeah because it could be compared to that. But it's you know it's an it's an estimate I mean this is hard data. What I'm talking about is a model based on data. A model based on laboratory data. Combining that with I thought we were able to come up with percentages based on food per parent compound. Right is that true? Pretty accurate Matt I'm looking at you. Some have you know that can be done. Mark I don't I didn't see her paper but she I think she did that and then I could certainly do it for using the ACC database. But for example the exercise I do with DEHP suggested a large percentage was due and you could come up with that percentage. And you could probably continue that exercise with other phthalates and get an estimate again. I don't you know without any kind of validation you're just using what you have and you don't have from that database the consumer product inputs. People well yeah well that's the idea is to come in from both ends and see how they match up. If the bottom up if there's a gap it may mean you're overlooking a source but not necessarily. You know the the warmeth is I think the best example where they did both and they came up with estimates of how much comes from food and other other things. But they their estimates nothing's they're not hard numbers. Baldy can you see any other strategies that or information that Kristen Holger could use to form us bottom up and and how would we mean you talking about generating new data. No but OK OK or possibly expanding it a little bit. That's what I was getting at you don't know of any other sources of data that we might be able to tap into. That's that's what we need to know that you're the expert in that area. So if you say there's nothing else out there then you're more comfortable. Paul from your experience. Would you also advise us to look at each delay in the viewer at the 95 percentiles or maximum values on their own and add them up because that would. End up in a similar approach as you are suggesting. Adding up individual roots of significance. My apologies. I think doing it individually first is probably the wisest way. And then you can always then go back and decide to summit based upon the data we get from Michael's group and others about what in what products you find certain or all of these same. Rather than doing it blindly at the 95th percentile I think you have to be a little cautious. Yeah I think the universe of fallow eats and toys is is a little narrower. I mean well it's always changing and it depends on how broad you cast the net. But the toys we looked at 10 years ago they were it was one valley. Now there's a handful of valley substitutes. But it's it's not a big universe so that at least with these products I don't think you have to worry too much about adding 95th percentiles. That's only if you try to be inclusive and look at food and all the other sources. Looking at the data will be able to make better judgments. We can have a conversation based on facts. But it's something that it's I think it's possible to in a scenario based exposure a ground up. I think it's possible to do something similar to what you're proposing where you you take that into account. You do a probabilistic approach. Sure. And I think it's reasonable. Two other comments one in the spirit of the silver book that we heard about this morning and have read about. I assume colleagues of ours would see that this is a the approach is a forward looking approach. Consistent with the recommendations in there of ways to do this that are looking toward improvement of methods rather than stepping back and using old methods. The other comment or the other one is a question and that is considering what would remain assuming that we use this approach. We see that in the standpoint of when he was not going to be involved in doing all of this work. Is it feasible to look at the larger task that would come behind this considering what you've already done? Is it feasible to do this for the committee? Sorry to do what we're outlining here. Yeah. Yeah. It is. I think so. Yeah. I mean what I'd like to do is to minimize my interpretation of what people are thinking. If we could be kind of prescriptive about you know let's correct for fasting in this way or let's you know. And I don't mind doing it multiple ways but if you leave it up to me to make interpretations of what you guys are thinking then I'm afraid I might misunderstand or make mistakes. I think we can offer different ways of possible approaches to interpret the data. Show what happens if we substitute one of the late with another one or resubstitute it again if we say well let's lift the ban and then everything goes back to the start again so we can play around with the data. That's clearly one of the strengths of this approach. Well and the other thing is you know should we be looking at 2003 and 2004 data also? I think Russ is perfectly right. We should first look into the young children and pregnant women data before we do this rather complex approach with and Haynes fasting non-fasting and so on. I think we have should have an individual look on this data there. Olga I agree with you I think don't go complicated. Use a more simple approach because you may find that the value added for going complex may not be worth the time and effort. Well that data I think will probably date back to around 2000 or so because these were all part of the I think it was the NIEHS or EPA children's centers which I think all were funded right around 98 to 2000. So that's when the urine samples were taken. Well unless the vanguard centers of the children's study can get some data. We're not going to be really starting in the field again until 2012. January 1st that's when they expect us to be back in the field. It is quite intuitive to begin with children but let me remind you that those biological data from Earl and others would actually be more suitable to look at women in reproductive age because what this models is the exposure during the critical male programming window. So I hear already toxicological purists slaughtering the panel if we use those reference values for for making assumptions about possible hazards or risks to children. You see my point. Yeah exactly and I think I think the approach is sound. We can revisit the details like where which reference doses and how they were derived in Andreas makes a good point. These data are probably best suited for the mothers. Or the women of reproductive age. You might consider using other reference values for children or other adults or so on. But I'm not sure you could come up with any necessarily come up with any better values. So basically this is what we have what you're saying. And but at least we have to identify the uncertainties around that. Or just take a I mean this you did a great job illustrating the process. We just need to step back and take a look at the you know are these the reference doses we want to use or are there better ones or so on. Are there any that are children specific. No. Okay so basically again that's an uncertainty we have to deal with. Yeah I mean the panel can derive any factors that they if the data are available. I'm not so sure the data are. Okay. But are the uncertainty factors chosen because I mean is there an additional value of 10 to these because of the. Well I think what I got from Earl Gray's data is that the fetus is the most sensitive. And then after that the children. Or are you the younger males and then even adults there are effects but less and less. Let the potency is lower so are the sensitivities lower so. I think it would be okay to use our reference doses based on the anti-androgenicity for everybody but just now understand what it means. I also think those data the data they have on the developing fetus they have good recent data. I think the data on the juvenile animals is not really comparable. It may not be as good so it may be possible to come up with child specific RFDs based on juvenile animals but the data probably aren't there. I'm not sure but it's something we should keep in mind and you know maybe take a look at before we go too far. Not sure I'm understanding this and I want to. Can we in your in the table that you showed us with case one case two that was all derived from animal data. And we have I assume similar data relative to the different sensitivities of the fetus versus. The dam versus I don't know what but could we could we use those differences in the same way that we use this these data to to get at that. No. No that's that's unfortunately not not easily done. What is clear is that you see those those studies this work was inspired by exploring number one. First of all defining the so-called male programming window that's been done by Richard Sharp and Grant Paul Foster. And secondly then to to first of all qualitatively rather than in the sense of deriving points of departure to see whether these satellites have actually the the effect that was hypothesized. And that is true that was the case now. So you can on the basis of this make some assumptions about the sensitivity of the developing male during the male programming window. To go from there if you if you just directly in a one to one fashion applied those kind of data to children. Then strictly speaking you would need to know what exposure their mothers experienced. But that's often not not available the only bridge I can see is very recent data from Richard Sharp's lab where he investigated what happens when when you dose. When you continue to expose male fetuses that were exposed in utero after they were born whether the effect gets worse or stays the same etc etc. But that's still that that would provide additional information which would still not allow us to make an assumption about sensitivity of children or the risk zone of the hazards to children derived from that kind of data. But I don't understand why not. I mean we're not we're not saying that this is the end of the story. We're saying that we're going to take what information we have and it's very limited and it comes from rats. But we're going to apply it to the human situation and we may be we may be off by a factor of whatever. That's essentially what we're doing here isn't it. Yeah I know what you're saying. But the point is that the the toxicological literature the toxicological data define a period of vulnerability which is not the young child but rather I understand that. I mean we're arguing about we're saying that the fetus is more sensitive than the newborn is more sensitive than a child six years older older. And don't we have some data from rats to kind of support that. We know what what doses will create a problem when when the fetus is exposed at the appropriate stage of development. And I thought Earl talked about data where they had phthalate exposures postnatally. That's right. That's right. That's what I mentioned. These are data from Richard Charp's lab but maybe Earl has some as well. No my point is this if you hypothetically if you expose a child to levels of phthalates that if received by pregnant mother during the first semester. Well if that child hasn't experienced these levels while developing in the womb it is probably safe. That's the point. So because we don't know this really the the best way to move forward would first of all be to to look at women in their reproductive age. But there's no problem with that. No in terms of choosing women in in Haynes that are of a certain age. But would you are you suggesting to use the same reference doses here as. I mean is that the point where we were right that if we had better information we could have actually used different reference doses for different situations. But we don't have they don't exist. I don't think that's right. I don't think so. I'm saying is the what you've done is wonderful and these reference doses here can be used for for women in their reproductive age. That would be the most appropriate thing to do. But it would be bought for another curveball. Would you. I'll throw it. Let me I'll throw a knuckleball. Would it be reasonable to assume also that pregnant mother or the mother the woman of childbearing age would be exposed to the same phthalates. As with the child after they're born because the child after the born is. Products X whereas the mom to be could be exposed to products why and you may be comparing apples. That's why that's why it's a straight ball versus a knuckleball. Yeah. Yeah. You're absolutely right. Paul very likely the child in you know a one year old child wouldn't use cosmetics. No no but their mother does. But I think they're in lies the fact that we get to too much detail. I'm trying to say is that the farther the more finally we try to carve this up in terms of what they've done with Chris and over done. I think we're going to be faced with the prospect that we're going to be looking at things that we have more uncertainty about other factors. And then what they did and why they chose the the fact the the RFDs they did. Yeah. Because I don't see a problem going ahead with where we are. My point is this is a good start and we should use it. My question is just this is perfect for women of childbearing age. I don't think it's too much of a stretch to apply it to young children. OK. Beyond that it's getting dicey. It would be probably conservative but it's getting dicey. I just want to point out that the calculation model Chris presented is right now limited to children from the age of three to 18. Because for this subpopulation we have data to extrapolate from the urinary levels to daily intake. We right now have no calculation basis to extrapolate from urinary levels of children younger than three to daily intake. Probably would go via urine volume per day. And so we will have to recheck what data we have. And for women in their reproductive age or childbearing age we would have to use a different formula of calculation. Just a single part is different but it would be a different formula just to point that out. We wouldn't move. Yeah. Exactly. We would need different values there. No problem. But we would have problems with children younger than three. Because there's very little data basis not in terms of metabolites in urine but in terms of extrapolating from the urinary level to daily intake. And I assume it's not 24 hour urine data. It's spot urine. And you know how it is with creatinine in children half a year a year of age. Very low. So it's difficult to use that. We would probably go via the urine volume the daily urine volume. And then we need some data on that. But again we can be very open in and explain everything. And it's it's up to everybody to decide on the reliability uncertainty. I think when we go through this the uncertainties are going to pop right out. Yes. Any other comments questions not like. Can I just ask so are we saying we're going to try to get data from other people to do this and pregnant women or children or are we going to how where are we with that. If assuming you're still co-workers willing to make estimates. I mean we certainly ask for the data on children and or pregnant women. And we can ask for the individual data. If you're willing to do the calculations. We'll have to discuss first what kind of data we request or what is the data basis we need like age of the woman body weight. So yeah exactly. We need to check whether all this data is available. This equation the correct equation for pregnant woman. I mean other than other than the creatinine part. There are no studies with pregnant women in terms of urinary metabolism. You know they have never been dosages of pregnant women. Of course. Primaster versus third. I mean I would think creatinine excretion is going to change a lot or it's known to. In addition to body weight. Yes. That usually goes up. So but I mean are there ways of informing. I mean do we would we other than adjusting the creatinine the smooth creatinine excretion. Are there other corrections we would make to such a formula for pregnant woman. We just don't know how reliable it is. Is that what we're saying because those studies haven't been done. Okay. Comments. I just think that we should just move ahead and learn about what Mike has in terms of data that's available to us. Could help. I think Taylor some of the thoughts that Chris and Holger have been toying with in their approach. Okay. This is the work that we did in about 2000. And it's all about DINP. And this is I think well this is I think mostly about the methodology in the approach that we used. But I think it's still still relevant to the current situation. But it's only oral exposure. In 1998 almost everything that we tested was contained DINP. And these are essentially these are Tether's rattles and soft plastic toys sold in the U.S. 99 98 90% of them contain DINP. And that's just a little bit of history. First there was DHP was replaced by DINP and so on. And of course DINP is relatively unique because it's it's a mixture of many different isomers which sort of complicates everything. So it has very low vapor pressure lower than most of the other phthalates. It's less soluble and more hydrophobic than even most of the other phthalates. So what did we do? We surveyed the soft plastic Tether's and toys for DINP in the amount. We measured migration rates in the laboratory. Method it was calibrated against adult volunteers. We also did an observation study with children up to 36 months old which looked at mouthing behavior. And then did a probabilistic analysis for children by different age and in different products. These are just some of the toys that we're dealing with. This is the apparatus. This was first developed by the Dutch and later adopted by the European Commission. You take a disc, a 10 square centimeter disc punched out from the product. You immerse it in a saliva simulant or saline solution, whatever. It's tumbled in this apparatus at 60 RPMs for a certain amount of time, half an hour. You collect the solvent. You add fresh solution to repeat it. And then when you're done you combine them, measure the amount of DINP in the solution. And this was validated, well I'll explain that in a minute. This just shows the migration rates as a percentage of DINP. There's a lot of scatter. You would expect that all things being equal, that the migration rate would increase with the concentration. But obviously with DINP not all things are equal. There are different manufacturing processes. There are different attitudes. I'm not sure what's actually accounting for the difference except that it's there. These were validated against or calibrated against studies with volunteers, human adult volunteers. There were several of these studies done. The Dutch did the first. A typical study has 10 or 20 adult volunteers. They're given a PVC disc which they mouth for one hour. Now the instructions are to mouth suck or gently chew this disc. It's broken up into four 15-minute sessions and you collect the person's saliva and analyze for the phthalate. There's also in the 15-minute sessions there are breaks and there's also a session with a Teflon blank. Some of these use actual toys. Some of them used a standard disc that was prepared that contained a known amount of DINP. So this shows some results from four studies. These are the migration rates are scaled divided by the mean to make them equivalent. But it's just remarkable that they're very consistent once you make that observation or make that correction. It just struck me as very consistent in their, of course, log normally distributed. So the migration data, now this is, you know, 10 years ago they were almost all DINP by this time. 42% of them, only 42% of the toys that we tested contained DINP. Get a migration rate and then we also, what we did, we added a factor. The Dutch and the European Commission developed this method for measuring migration in the laboratory. They adjusted the condition so that the migration rate in the laboratory would correspond to the 95th percentile in the human studies. And that was their, that was their choice. So here what we did was we took data from one of the volunteer studies with the standard disc. Got the migration rate with the volunteers and a migration rate by the laboratory method came up with a ratio of 0.28. So in other words, we're adjusting to the mean to the mean instead of the 95th percentile to the mean. The observation study, 169 children from three to 36 months randomly selected in two metropolitan areas. We had trained observers, there's a couple of studies where they have the parents make all the observations. We sent trained observers to the children's either their homes or their daycare, wherever they were. We did a total of 12, 20 minute observations over a two day period. They recorded everything that the child mouthed, including the frequency and the duration. We did have the parents record the time because they were the only ones present for the entire day. The time that the child was awake and not eating, presumably because they're not mouthing while they're asleep and of course when they're eating they can't. So what we ended up with, or this is, okay, so for all exapacifiers, the youngest children averaged 70 minutes per day and it declines as they get older. When you start looking at specific products like plastic tethers and rattles or soft plastic toys, the times actually are very surprisingly low on the order of a minute per day. What the children mouth on most more than anything is their fingers, then pacifiers, then tethers and toys and then after that it drops off. So previously we had been using estimates of hours per day of mouthing time and some people are still using for risk assessment purposes, assuming an hour per day or more. This is just for soft plastic toys, for an age group, the means, couple of minutes per day, here the 95th percentile is still under 10 minutes. The medians in fact were zero because on a given day less than half of the children were actually mouthing or observed to mout those particular toys. So we calculated the daily exposure based on the migration rate of the product measured in the laboratory. This is the correction factor, this is the standard disc migration rate in the human volunteers versus the laboratory. This is the mouthing time in minutes per hour that the child mouth the particular article and the time that they're awake each day and divided by the body weight. So you basically do a sampling Monte Carlo kind of an analysis. Now some of these are the mouthing time, the minutes, its minutes per hour is dependent on the product, the child's age and months in the product. The exposure time depends on those things as well and the body weight depends on the child's age. In months. So we did all the sampling and came up with these estimates for children 3 to 11 months old. We're talking about mean, even 95th, even 99th percentile exposures going up to on the order of microgram per kilogram per day. Now this is assuming 40% of the toys contain DINP. And you can do the same thing as assuming do the calculation assuming that they all contain DINP, which was the case some time ago. So well, the key assumptions. Well, we applied migration rates from one product to another because I didn't think that really mattered. You know, in all of these studies in the human studies and interpreting them, you're collecting their saliva. So we're assuming that absorption through the oral mucosa is negligible. And we also, well, we didn't estimate dermal exposure in this case. The limitations, well, there's a lot of variability. It's not so much uncertainty as variability in the migration rates and mouthing times. And well, for our overall conclusions, we were only looking at one phthalate in a narrow, a narrower range of products. So now here we are. We're here. It's in now winter. So I want to just quickly bring up another a few more slides that shows the more recent work that we did. Gonna skip the flame retardants. Yeah, I can't. But there's, yeah, I can't, I don't see slideshow maybe. Well, in that case. I accidentally copied the PDF instead of the, yeah, that's what I'm going to do. Okay, I hear you. Okay, so we already know this. We already know the, but the three permanently prohibited in the interim prohibition. And so let's look at exposure studies. We collected 63 toys and childcare articles. And we collected these a couple of months before the the ban actually became effective. And we identified the plastics and plasticizers. And we measure the plasticizer concentration and migration. So we don't really need that. And 10 years ago, almost everything was PVC and had DINP. You can see now that PVC is now about a third of the plastic toys, soft plastic toys that we tested. There's a variety of other plastics, most of which don't require a plasticizer of the PVC products. Of course, now that the phthalates are banned, and this is actually just before the ban. This is one DINP product, which would be still allowed because it's not, because it can't fit in a child's mouth. The DEHP is one that now would be banned. And as far as the plasticizers go, ATBC, Cetal Tributal Citrate is probably the biggest, followed by, this is the para isomer or terephthalate version of DEHP. DINCH is, if you take DINP and reduce the ring, so instead of an aromatic ring, you have a cyclohexane. TXIB is a plasticizer, it's not really a plasticizer, it's used in combination with other plasticizers. It's not used by itself, but it was present in a significant portion of the products, a few benzoates. And in fact, most of the products that we tested had more than one of these, some of them had as many as three. You've already seen that. These are the migration rates. This was interesting because DINCH is actually pretty similar to what DINP was. And the other thing that struck me here, there is some difference in the migration rates among the different ones. For most of these, the concentration of the plasticizer seems to correlate pretty well, or the migration rate seems to correlate pretty well with the percentage. But DINCH is more like we found with DINP, it's kind of all over the place. The others are much more better correlated. And let's see, I didn't show all the data. This is really all that I have on these slides. But the other thing that we did besides the migration data that I've shown, we did wipe samples of the toys to estimate dermal that we could use to estimate dermal exposure. So we've got a set of toys, and I'm just presenting the summary data. We've got the individual toys in any level of detail you want. Toys and other kinds of things that the kids, things that would qualify as childcare articles under the new regulations. And the question is whether we need more data or some other kind of migration study that we don't have, but it's a start. And the issue we had with the last chap is the data weren't ready until the chap was already done. So this way we wanted to do, at least get started on it before you came together. So at least you would have some data to work with. Comments, questions? Olga? Yes. Mike, probably I asked with the question last time already, what is the reason for the human versus lab correction factor? When they developed the study, that method, they adjusted the conditions of the laboratory, the time and volume and so on to correspond to the 95th percentile or the upper end of the volunteers. And you saw it was very skewed distribution. Here we did that factor to, so it would be a one to one basically. And then we, because we were doing a probabilistic exposure assessment, so we didn't have to worry about being conservative. So would the lab have underestimated the exposure as it has been found in the human studies? Well, by this method, it's nominally the same. The one possible source was with the humans. If they were, if some of the DINP were absorbed through the lining of the mouth, then that could cause an underestimation. Well, we convinced ourselves that that wasn't likely, but that's not a fact. Did you check for metabolic activity in the saliva? No. You know that saliva is full of esterases, for example? So you might produce monoesters in saliva? At that time, no one tested for that? Because I think one should at least measure the monoesters to check whether, because my suspicion is that a lot of the DINP, DHP or DINCH will be broken down already if you chew it for 50 minutes in the saliva. Even in 15? Yeah, pretty fast. So I just want to add these comments to maybe 50. Yeah, yeah, yeah. Definitely. You put up the slide again that had the, didn't you have estimates in micrograms per kilogram per day, what those values are? I'll put this one up. This is assuming that all of the toys and tethers and so on contain DINP. So I just, I could, there's a problem with doing things quickly and I could have made an error, but what I just did while you were talking was look at the estimated daily intake for DINP from the calculations that I showed you all. Now this is for older kids, six to 18, six through 18, but there's somebody with a very high value, you know, 4,800 micrograms. But the median is actually 7.1. The mean is huge, so the mean is 18.5. But the median, so if you think about 7.1 as the median daily intake for DINP, that might be... Micrograms? Micrograms per kilogram per day. Yeah, of course this is just for mouthing toys. But the point is, so here the average exposure, sorry, the average daily intake for DINP is estimated at 7.2. But, you know, a third of that is the 99 percentile for 2012 to 2030 months, and that's different ages than all, you know, I know I'm apples and oranges. Assuming that this is right. If you use the mean and 3 to 11 month old, right, and use the higher level, it's about 10 percent of the total daily intake, right, because you said it was 7. The median is 7. The mean is 18.5. So it's even less. I just would like to cite something. The library glands also secrete the library lipase. A more important form of lipase to start for fat digestion. Lipase plays a large role in fat digestion in newborns, as the pancreatic lipase is still sometimes developed. So I think if we look at newborns and young children, you have to assume that they have strong lipase activity in the mouth, and therefore any extracted DINP is fastly metabolized to MINP or other monoesters. And we would have to assume that in some way additional to the mucose resorption. Well, I think that is exactly my opinion. But in an infant, once it migrates, we're assuming that the infant, this migrates out, and we're assuming that the infant swallows all of that. And it doesn't matter at once it's in the mouth. I mean, once it's in the mouth, it doesn't matter if it gets metabolized in the saliva or in the gut or whatever. What's a concern is in the adult studies, it's a concern. For the kids, it doesn't matter. It's just an additional point of remark. So you would have to correct for the lipase activity, or at least you would have to measure the monoester in these studies too. I don't understand why. I mean, you're putting a level of detail on here that's unnecessary. No, you are doing a big mistake if you only measure the diester in the saliva, because it might already... What he's saying here is this is the estimate of exposure from the toys before. Anything's metabolized. Anything. Yes, it is based on the human saliva data where you measured the diester. If 50 or 70% of the diester is already cleaved to the monoester, you underestimate the result. They're measuring the diester, and it's quickly cleaved, probably almost instantaneously in an infant's or child's mouth, to the monoester. So the chemical analysis will miss the monoester because it's measuring the diester. This is a hypothetical measurement. This is not real. This isn't saliva. These are phosphate buffered saline or something. That's what it is. The adults, it's saliva. Right, the adult one with the saliva. With the adults, that's an issue. Not in this data. Exactly, Paul. That's exactly why I asked the question, what is the difference between the lab and the human study? Okay, yeah. Okay, if you're asking that question, yes, but I think what you're interpreting from it is not... The lab study, there's no light-paces. Exactly. Right. So I was interested in the difference. Obviously, there has been a difference which needed to be corrected for. I would have been interested in the reason for this. Well, I mean, I think I misunderstood the question, but what they did was they actually did the human studies first and then developed a method that would approximate the human studies. A laboratory method. If the assumption has been wrong in the beginning, you cannot develop a lab method which is based on the wrong data. Exactly. Yeah. Here in the saliva is under-measured because you're not measuring... I mean, a lot of the diesters convert it to the monoester and then you're using that to model the... In the human study. Yeah. And then you're using that to model the hypothetical or done-by-machine, I guess. Right? That was just one point to... Before you get off that one point, I still know where you're going with this. I mean, this is a hypothetical DIMP exposure from plastic toys. Right? Explain to me what your point is with respect to those numbers. Okay. Do it stepwise. Yes. I'd like to hear something that you're confusing me. Yeah. If you're confusing me, you'll confuse others. Yeah. So my introductory question was why there was this factor relating lab to human results. Obviously, there is a difference. Yes. A big difference in the human approach compared to the lab model is that the lab saliva is missing lipase. Mm-hmm. So in the human study, if you assumed lipase activity, you would assume that the diester is cleaved to the monoester. In the human study, only the diester has been measured. If you don't measure the monoester, you're underestimating the migration of the diester from the product. How fast is that migration? How fast does that transfer occur? It's just hydrolysis, right? So it's seconds. Once there's contact between the enzyme and the diester, it's a simple chemical reaction. I would think it would occur. There's often kinetics involved, but let's face it, you moved this object for a couple of minutes at 37 degrees ideal temperature for enzyme activity. Has anyone done that experiment? That's the issue. Probably nobody has. We're making an assumption. One way or the other. I think you're making the wrong assumption from the beginning on that you don't assume the cleavage of the diester because that would be the first and most logical step to occur, in my opinion. Or us. I think there is lipase activity in saliva. I brought it up because Mike said they modeled the membrane take up, or the mucous membrane, that the mucous membrane would play a role. I would say if you introduced the mucous membrane, you would definitely have to introduce the lipase activity. I think we should go back to the original report and see what they did. I want to find out what their assumptions were, whether they made any corrections. Because right now what we're doing is debating hypotheticals. We need to look at the original approach that they used and why they did, why the ratio they selected is what it is. Let's get back to some of the discussion items on our draft agenda. Who will do what? It sounds like one of the things that Mike will provide protocols and data studies to Holger and Chris. What other talked about data for pregnant women? So Russ, you're going to provide these references? The contact directly from Mike and CPSC. But I could work with you in identifying the three or four cohorts and asking for original data. And then as Holger mentioned, we'd want to know which variables, in terms of age and body weight, et cetera, that they would supply with the data. I can do that. Yeah, there are assignments that we've explicitly made or explicitly made. Here, Chris. So I'm going to push it down. So we look at, have someone look at those tables that we're assuming values from and see if there's any improved corrections or vary? I mean, do we want to just go with what we've got? Do we want to try to improve those values, the ones that we set? Are there better values for those? I mean, did you calculate them from Earl Grey's data or from? It is literally taken from either Earl Grey's data, Earl Grey's testimony, or Earl Grey's presentation at the second meeting. Okay. So it was five milligrams per kilogram per day and that they were equivalent. That's what they all have five. And then the relative potency was there as well, the .15 relative potency. Did that come from them? Yes. That's what he said, and that's what in the testimony. I'm trying to read. The five, was that a no-well, low-well, or it was a no-well? No-well. The question is on the table from Chris as to whether we just want to go with this in our scenario modeling or do we want to entertain more discussion about whether we should change these assumptions or these values? We might again ask Earl for his opinion. We might ask Andreas again for his opinion. I think this was very helpful, productive and instructive. I think we should go ahead along those lines. But I'm a little mindful of the other aspect of this morning's discussions. That is back to the problem definition step. I think we need to do a little more looking at the charge. And I guess we need to reflect a little more on what precisely Chris's and Tolga's work is good for, and what way does it help us address the charge? Mike, do you have slides you can put up? Charge for the charge. I might. Let's have a coffee break. Okay. While Mike is looking for this. Let's try to be back in 10, 15 minutes, please.