1                             MEETING

 2                              OF THE









11                   255 SOUTH AIRPORT BOULEVARD






                       TUESDAY, APRIL 13, 1999
                              9:00 A.M.





24   Janet H. Nicol
     Certified Shorthand Reporter
25   License Number 9764



 1                           APPEARANCES


 3   Dr. John Froines, Chairman
     Dr. Roger Atkinson
 4   Dr. Paul D. Blanc
     Dr. Craig Byus
 5   Dr. Gary Friedman
     Dr. Anthony Fucaloro
 6   Dr. Stanton Glantz
     Dr. Peter S. Kennedy
 7   Dr. Hanspeter Witschi

     Mr. Lynton Baker, Staff Air Pollution Specialist
10   Mr. Bill Lockett, Deputy Ombudsman, Northern California
     Mr. Peter Mathews, Office of the Ombudsman

     Dr. George Alexeeff, Deputy Director for Scientific Affairs
14   Dr. Melanie Marty, Senior Toxicologist

     Mr. Paul Gosselin, Assistant Director (Via Telephone)
17   Dr. Lori Lim, Staff Toxicologist
     Dr. Jay Schreider, Primary State Toxicologist


20   Dr. Dale Hattis, Clark University
     Dr. Lorenz Rhomberg, Harvard University







 1                              INDEX

 2                                                           PAGE


 4   1    Discussion of margin of exposure (MOE) and          99
          reference exposure level (REL) approaches.
 5             DPR and OEHHA

 6   2    Scientific presentations regarding non-cancer
          risk assessment.
               Old problems and Some New Approaches in         3
 8             Non-Cancer Risk Assessment.
                    Lorenz Rhomberg
               Inter-individual Variability and               63
10             Quantitative Low Dose Risk Projections.
                    Dale Hattis
     3    Review of findings for DEF as a toxic air          140
12        contaminant.
     4    Update on the schedule of pesticide documents      161
14        to be submitted to the SRP.
               DPR staff
     Adjournment                                             175
     Certificate of Reporter                                 176











 1                      P R O C E E D I N G S

 2             CHAIRMAN FROINES:  Let's begin.  Welcome,

 3   everybody.

 4             The first thing I'd like to do is to introduce two

 5   people, Dr. Elinor Fanning and Lesley Dobalian, who are

 6   sitting back there.  Elinor and Lesley are working for the

 7   panel at this point.

 8             The load on this panel has increased dramatically,

 9   especially insofar as we are trying to plan workshops, we're

10   trying to address issues outside of having specific

11   documents, and we are having, as everybody knows, 51 acute

12   documents that we had to review the individual compounds.

13             And George Alexeeff tells me that we are about to

14   receive sometime in the next few months the chronic document

15   which will have how many chemicals?

16             DR. ALEXEEFF:  120.

17             CHAIRMAN FROINES:  120.

18             So we decided that we could use some help, and

19   Lesley and Elinor were kind enough to agree to work with us.

20   And so that, I think, is going to provide us with some

21   technical backup that will be useful.

22             We'll make sure the panel has information about

23   how to reach them and vice versa, so that questions that the

24   panel members have can be raised with them.

25             The second thing that I wanted to do today was to



 1   change the order a bit, and because we have been dealing

 2   with noncancer risk assessment, first with the 51 acute

 3   documents, soon with the 120, or what have you, chronic

 4   documents, and with the anticipated pesticides that we're

 5   going to be addressing, we have gotten more and more

 6   involved in issues of noncancer risk assessment, and we have

 7   had to be concerned with the methodologic implications of

 8   how one does that risk assessment.

 9             So what I did was to invite Lorenz Rhomberg from

10   Harvard University and Dale Hattis from Clark University to

11   talk about their research on trying to take some fresh looks

12   at noncancer risk assessment.

13             And I think that it would be better to start out

14   with Lorenz and Dale, because I think what they have to say

15   may have relevance for the discussion of the margin exposure

16   and reference exposure level approaches that DPR and OEHHA

17   use, and so I thought we would switch that.

18             So we didn't talk about who goes on first.

19             Dale, why don't you -- oh, Lorenz is going to go

20   first.  Okay.

21             Why don't we start with Dr. Rhomberg and

22   Dr. Hattis and then move to the margin of exposure

23   discussion.

24             And I had one question, housekeeping question.  Is

25   Paul Gosselin with us?



 1             DR. SCHREIDER:  I'm Paul Gosselin for the day.

 2   Paul has to appear in the Legislature.  For this week he's

 3   acting director and chief deputy and assistant.  So he will

 4   try to make it by telephone call.

 5             CHAIRMAN FROINES:  Well, that's what I was

 6   wondering.  Is there going to be a conference hookup?

 7             MR. MATHEWS:  Yes, there will be.

 8             CHAIRMAN FROINES:  So he'll pop in at some point

 9   and you'll let us know.

10             MR. MATHEWS:  I will.

11             CHAIRMAN FROINES:  I don't know if the panel has

12   any other issues to raise before we get started at this

13   point.

14             Hearing none, why don't, Lorenz, why don't you

15   begin.

16             And this is intended to assist the panel in

17   looking at noncancer risk assessment in the hopes that over

18   time we can develop new methodologic approaches that advance

19   the state of the art, so that we start to look at these risk

20   assessments with more confidence that they have -- that they

21   are more satisfactory.

22             For example, as you all know from reading the

23   Kenneth Crump article, there are problems with using NOELs,

24   and I won't go into that.

25             So good.



 1             DR. RHOMBERG:  Should I start?

 2             CHAIRMAN FROINES:  Yes, please.

 3             DR. RHOMBERG:  Thank you all for having me.

 4             Since I haven't participated in your meetings

 5   before, I'm at a little bit of a disadvantage of knowing

 6   about exactly where you are and what your questions are on

 7   this, so I'm hoping to keep this rather informal.  If I'm

 8   going off in the wrong direction, something that's too

 9   elementary or too beyond what you thought about, just tell

10   me and I'll try to make some on-the-fly adjustments.

11             I've titled my talk, "Some Old Problems and Some

12   New Approaches in Noncancer Risk Assessment."

13             CHAIRMAN FROINES:  Lorenz, let me stop you just

14   for a second.

15             DR. RHOMBERG:  Sure.

16             CHAIRMAN FROINES:  I think if the panel wants to

17   ask questions during the course of the presentation, that's

18   fine.

19             But I think we'll hold questions, George, from

20   members of the audience.

21             But after Lorenz has had -- has finished then I

22   think we'll also give your folks a chance to ask questions

23   if they have them.

24             DR. RHOMBERG:  Fine.

25             Well, as I said, this is a title I've chosen.



 1   Notice it doesn't say solutions.

 2             I think that we have some problems in noncancer

 3   risk assessment that are longstanding problems and there are

 4   people that are trying to grapple with them.  And I think

 5   what I'm really going to be talking about is the grappling,

 6   rather than exactly what the solution is and what's to be

 7   done.

 8             First, let's sort of start out where are we, what

 9   is the solution, what is the source of these old problems.

10             You know, really the basic approach to noncancer

11   risk assessment has been, one, to assume that these are

12   endpoints with thresholds, somehow that there is variation

13   and sensitivity among the population, and that's why you get

14   a dose response, that at some doses some will respond and

15   others will not, and at higher doses more will respond and

16   less will not, because of variations in sensitivity of their

17   individual thresholds.

18             That's a toxicologically reasonable thing for a

19   lot of kinds of endpoints, but again that's something that

20   probably could be thought about more in terms of underlying

21   rationale.

22             And then the other thing is the basic quantitative

23   approach that's been taken.  It's really a safety

24   assessment, rather than a risk assessment kind of approach,

25   that dates back to the Lehman and Fitzhugh paper in 1954



 1   that really established this method, and with minor changes

 2   or embellishments it's really gone on since then in more or

 3   less the same way.

 4             What Lehman and Fitzhugh did, they were working

 5   for FDA, and they wanted to have a method for assessing the

 6   safety of food additives.  Here were compounds that were

 7   added to foods in very small amounts, and so you really were

 8   just asking the question of can you really show that this is

 9   well below any levels that would can cause toxicity, and

10   therefore we are fairly sure that any kind of thresholds for

11   human toxicity that would be present for these compounds

12   that we are well below that.

13             And they argued that well 100-fold below a NOEL in

14   animals, a no effect level, in animal tests is probably

15   sufficient for that.

16             And they argued it mostly on the basis of

17   experience.  We don't really see anything too much worse

18   than this.  And they did mention that this could, in some

19   sense, account for the fact that humans and animals might be

20   different in the doses that give them toxic effects and

21   moreover that individual humans are different from one

22   another in their tolerances, and somehow that this 100-fold

23   could take care of all of that.

24             They really weren't all that explicit about saying

25   10-fold for this or 10-fold for that, but just 100-fold



 1   ought to be enough.

 2             By and large, that methodology is continuing

 3   today.  And here's what it looks like in a more modern

 4   guise, but it's basically the same thing.

 5             This is what the EPA does.  They call their output

 6   from this a reference dose or RfD.  You take a NOEL from

 7   animal tests, the most sensitive sex and species if you've

 8   got a variety of endpoints, and you usually want to have

 9   them to allow for the fact that different endpoints and

10   different sensitivities can pop up in different species.

11   And then you divide by a series of uncertainty factors,

12   which are usually a factor of ten, but not invariably, and

13   you can argue about their values.  And you might also divide

14   by something called a modifying factor or a data quality

15   factor to further hedge the allowable dose, if you think

16   that that's necessary, considering the quality of the

17   database, the quality of the data that go into it, the

18   completeness and so on.

19             And basically what this amounts to is modifying

20   the Lehman and Fitzhugh methodology to, one, recognize more

21   reasons, more extrapolations that need to be done and

22   therefore adding some more factors for them.  And, two, a

23   little bit more explicitly dividing the overall margin that

24   we have there into pieces that are sent out representing

25   various factors.



 1             So each of those factors is an extrapolation in

 2   some sense that we have to make from animals to humans in

 3   assessing the safety.

 4             And because we are uncertain about that

 5   extrapolation, we sort of divide by ten to account for that

 6   in some sense.  What we mean by that is something I'll get

 7   into in a minute.

 8             These are traditional ones.  UFA for animal-human

 9   extrapolation, because we're using animals we want to know

10   about human doses.

11             UFH for what is often called average human to

12   sensitive human.  As we'll talk about more, that's a little

13   bit of an oversimplified way to do it, but it's meant to

14   account for the fact that in the human population there may

15   be some that are very sensitive and we want to protect them

16   as well, so we divide by ten again.

17             Subchronic to chronic.  If you only have a

18   subchronic test and you want to know about what would be

19   safe for chronic lifetime human exposure, as is often the

20   application here, you might want to add something for that.

21             And if you have a test that doesn't actually

22   identify a NOEL you might want to say, well, we'll use a

23   factor in for that.

24             So the point is that you've added these things by

25   explicitly looking at the extrapolations that you want to do



 1   or the uncertainties in those extrapolations that you want

 2   to account for.

 3             Well, what's wrong with this methodology?  It's in

 4   some sense served us well, and often you get those

 5   statements, well, we've been doing it for 50 years, and we

 6   don't really have too many disasters on our hands, so maybe

 7   it's okay.

 8             But there are some difficulties.

 9             One is that we keep thinking of new things that we

10   might want to be aware of or careful about, and the one

11   that's being talked about these days is risk to children.

12             And there's a mandate in the Food Quality

13   Protection Act, for instance, that says, well, you have to

14   argue whether you do or don't need yet another factor of ten

15   for children as being yet more sensitive than even the

16   sensitive humans accounted for in UFH.  Or do you?

17             Another problem is that, you know, these factors

18   of ten, where do they come from?  There's a famous saying if

19   we had 12 fingers they would be factors of 12.  They're just

20   order of magnitude adjustments that we don't really have a

21   great idea about whether they're big enough or too big or

22   small enough or too small, or whatever, and we would like to

23   try to apply some data to those to see if in fact this kind

24   of safety that we presume is actually being accomplished.

25             Another problem is that in some sense if these



 1   represent the variable quantities, then it would be unlikely

 2   to be extreme on all of those quantities at once.  We're

 3   sort of assuming that we are by multiplying those things

 4   together, and some way we'd want to take into account that

 5   fact.

 6             So we have a problem here in that things are

 7   conservative to some degree, because we have allowed for

 8   uncertainties and variations in case to case that may or may

 9   not exist in our particular case.

10             We have a hard time saying quantitatively how

11   uncertain we really are.

12             So one set of problems.

13             And another set of problems has to do with what

14   we're asking risk assessment to do nowadays.  We're asking

15   something very different than Lehman and Fitzhugh were

16   asking.  Lehman and Fitzhugh had manifestly small exposures

17   that we just wanted to screen to see whether they were safe,

18   but there was no sense in which we were then going to start

19   jacking up the exposures to these things up to the safe

20   level.

21             But once you say you can determine what a safe

22   level is, there's a slippery slope here and we're slipping

23   down it rapidly over time.  Well, I suppose it's taken us 40

24   years, but we've continued to slip down it.  And the slope

25   goes like this.



 1             Well, here's a screening, sort of a one-way

 2   screening kind of pronouncement.  Below this level we're

 3   really fairly sure that we're safe, but above it we're not

 4   saying we're unsafe, we're just saying we're progressively

 5   less sure that we're safe.

 6             That doesn't really matter if you always have low

 7   exposures and you just want to make the argument this is

 8   okay.

 9             But once you think you can do that, then you start

10   to say, well, how much could I jack up the exposure, if

11   you're not applying that to food additives, but to something

12   else, how much emission will you allow from this factory,

13   how much cleanup will you require at this toxic waste site

14   or something like that until you get up to a level of risk

15   that will be acceptable.  So you can start titrating risk so

16   to speak.

17             That puts a very different interpretation on this

18   kind of level.

19             Another thing is that we are then, because we're

20   doing those things, we are then talking about exposures that

21   are sometimes in the range of this RfD or maybe even above,

22   and we say, well, we don't say that there is a risk above

23   the RfD, or the acceptable daily intake, or whatever like

24   that, we're just less sure that there isn't one.

25             And yet the interpretation of that becomes very



 1   important.  The more we're asked, for instance, to do things

 2   like cost-benefit analysis, you have to say what is the

 3   benefit of this regulation in terms of noncancer cases of

 4   toxic effects avoided.  How do you put any kind of measure

 5   on that, on being a little bit more sure than you were at a

 6   lower dose that nothing will be -- it will be little bit

 7   less sure that nothing will happen, but you don't really say

 8   how much is going to happen.

 9             So in some sense there's really no dose response

10   analysis in here.  There's no estimation of risks.

11             And more and more we're being asked to do that

12   kind of stuff.

13             So in some sense this methodology isn't serving us

14   as well as it used to, because the kinds of questions we're

15   asking about it, are asking it to cover, are different.

16             Here's a quick summary of some of the problems.

17             The first one I didn't even mention, I should just

18   mention that briefly, the nature of the NOEL.  I guess

19   you've talked about this before.  NOELs have to be one of

20   the doses that was tested and therefore the experimental

21   design that was chosen by the experimenter has a big

22   influence on what is determined as the NOEL.  It's not the

23   same thing as a threshold, and it's really more about the

24   experiment than it is about the toxicology of the compound.

25             Some of that can be gotten around by using the



 1   benchmark dose procedure that I'm sure you've talked about.

 2             I mentioned are the uncertainty factors of ten too

 3   big or too small.  There's a problem of compounding

 4   conservatism if each one is slightly conservative and you

 5   multiple them all together, you get a very conservative

 6   result at the end.

 7             Here's another one I didn't mention.  When you

 8   have pharmacokinetic data or something else that in some

 9   sense is accounting for some of the extrapolation and is

10   being specific about some of the uncertainty that you have

11   in one of these extrapolations, how do you then reduce the

12   uncertainty factor to account for the fact that you have

13   more information, if this is to account for uncertainty and

14   is to help you with the extrapolation and you have another

15   way of doing that better, somehow you should get rid of at

16   least some of that uncertainty factor.  We don't really have

17   a good idea of how much of it, because we haven't really

18   explicitly said what we think we are accomplishing with

19   those factors.

20             In the end the conservatism is unclear and it

21   isn't consistent.  Is this a very conservative procedure or

22   is it a slightly conservative procedure, and maybe not even

23   conservative enough, and we don't really have a good way of

24   making that argument other than by polemics.

25             Then as I say, this question of use and risk



 1   management, a one-way conclusion about safe that doesn't

 2   help us when we want to talk about exposures in the ranges

 3   where there might be risks that start to emerge and we want

 4   to look at costs and benefits or we want to say what we're

 5   accomplishing with our regulations.

 6             So what do we need then?

 7             We need a risk measure, not just a safety line,

 8   but a risk measure that will say something about what's the

 9   probability of responses at high doses, outcomes and the

10   probabilities of occurrences and presumably also something

11   about their severity.  That's a tough one, though.

12             Characterization of uncertainty.  We want to try

13   to separate our estimation of these uncertain quantities in

14   our characterization of that uncertainty from our decisions

15   of conservatism about what to do in the face of that

16   uncertainty.  It's really a question of separating risk

17   assessment and risk management.  The way we have it now,

18   it's sort of built in.  We define a safe level as something

19   that has these safety concerns built in and if those are

20   seen as excessive or not strong enough, somebody doesn't

21   have a real easy way of disentangling what we've estimated

22   about the compound from these risk management concerns, and

23   we want to be able to do that and the way to do that is to

24   fully characterize the uncertainty in the extrapolations.

25             We want a means to incorporate case specific data,



 1   pharmacokinetics and endpoint-specific pharmacodynamics,

 2   meaning models of the toxic effects when we have particular

 3   understanding of a compound, rather than just treating it in

 4   this generic way, and that somehow reduces some of those

 5   factors that we have to apply, but we don't have a good way

 6   of saying how, and so all together we want a sound framework

 7   for how to do it.

 8             Well, what can done to do that?  As I said, I

 9   wasn't really going to have solutions, but I'm going to show

10   you sort of a tentative idea towards a solution that was

11   proposed a few years ago by some of my colleagues at

12   Harvard, Sandra Baird and John Evans and some others.

13             And basically what they were suggesting in a paper

14   that's in Human and Ecological Risk Assessment is to think

15   of it in the problem this way.

16             These uncertainty factors are extrapolations, the

17   magnitude of which varies from one chemical to another, from

18   case to case, and we can get empirical data on how that

19   variation happens.  So if we can now describe a distribution

20   of the magnitude of the adjustment that's needed for each of

21   those factors, a statistical distribution, and then work

22   through all of those and call them adjustment factors now,

23   rather than uncertainty factors, because the uncertainty

24   arises in the question of where in the distribution you are

25   for the particular chemical you're looking at at the moment.



 1             As I say, an adjustment that varies from case to

 2   case.

 3             So you can say let's try to get empirical

 4   characterizations of those adjustments and then work through

 5   the joint distribution of all of those to come up with -- PT

 6   there is population threshold, meaning a dose that for a

 7   whole population should be below the level it's causing an

 8   effect.

 9             And we don't know that number for sure, but we can

10   say something about the distribution of where it would be

11   and how sure we want to be to be below it, it's something we

12   can do as a risk management decision.

13             And if you try to do that, and they tried to do

14   this with the way that they admitted was sort of

15   preliminary, because getting these distributions is not at

16   all easy, as a lot of our -- the following part of the talk

17   will be about, but if you do this, you get some sort of

18   estimates of how they do this.  I'll let you read the paper

19   and see how they actually propose to do all of these

20   things.

21             This is the actual result they got for acetone.  I

22   can't remember these.  Acetone.  Where you work through

23   those distributions with a Monte Carlo kind of approach and

24   they got this distribution of what would be the likely

25   population threshold and the arrow there shows where the EPA



 1   RfD was.

 2             And it's sort of where you would hope and expect

 3   that it would be in that it's sort of conservative compared

 4   to the whole of the distribution, with not too much chance

 5   of not being stringent enough, but a fair probability of

 6   being more stringent than would may be necessary, according

 7   to this kind of approach.

 8             Well --

 9             DR. GLANTZ:  What percentile is that in the

10   distribution?

11             DR. RHOMBERG:  I don't know offhand, but it's

12   somewhere close to a fifth.  So if you want to say this is

13   sort of in the 95th, you know, percent, lower bound on

14   something, you'd be pretty close, actually.

15             DR. GLANTZ:  Wouldn't that say that the current

16   approach actually is working?  Because you go through this

17   more complicated model and you end up about where you would

18   hope to be with the simple-minded approach.

19             DR. RHOMBERG:  I think it suggests that it was

20   working for the purposes that it was trying to do.  But as I

21   was trying to suggest to the earlier slides and the earlier

22   discussion, some of our -- we have some new purposes as

23   well, for which it wouldn't serve so well.  But for the

24   purpose of finding a reasonably conservative, but not unduly

25   conservative value, that's probably good for protecting,



 1   probably, most of the people.  Yeah, it worked pretty well

 2   in this case.

 3             There's a caveat there, though, and that is that

 4   in coming up with these distributions they were highly

 5   influenced by the existing factors of ten and the rationales

 6   for them.  They didn't really have a good way of getting

 7   good empirical data on all of these distributions, so in

 8   some sense they're just building in those distributions

 9   again, and it's not too surprising they got some of them

10   back.

11             So this is an interesting approach.

12             I'd like -- I like it as something you can do.  Of

13   course it is technically difficult, but it isn't without

14   some further issues and I just want to point out some things

15   that it doesn't do so well.

16             One is in some sense it's mixing uncertainty and

17   variability.  We have uncertainty and we're talking about

18   one compound here at a time for which we're doing this for

19   acetone, or whatever, and we're saying, now, well, there's

20   this distribution of what could be the relative toxicity of

21   acetone in the animals that were tested, which I don't know

22   offhand what they were, and in humans.

23             But, you know, one value from that distribution

24   applies to everybody, because that's sort of an

25   animal-to-human thing, we're talking about one species of



 1   animals and one species of humans, so we're taking one value

 2   from that.

 3             On the other hand, our adjustment for -- I pointed

 4   to the wrong one before --

 5             DR. GLANTZ:  Excuse me.  Wait.  You just lost me.

 6             I thought -- I don't see how you're taking one

 7   value.  I thought the whole point of this was that you're

 8   considering the whole distribution.

 9             DR. RHOMBERG:  You're considering the whole

10   distribution because you don't know which one value in that

11   distribution applies to acetone, but if you were sort of

12   omniscient, you would say this is a distribution over all

13   chemicals of relative toxicity in animals and humans, and

14   acetone is one of those chemicals and so it's somewhere in

15   that distribution.  Not knowing where it is, we choose, we

16   use the whole distribution, but, if we knew, it would be one

17   particular value.

18             DR. GLANTZ:  So you're saying then -- see, I had

19   assumed that each of the distributions applied to the

20   chemical of interest.  You're not saying that?

21             DR. RHOMBERG:  No, it doesn't.  It's on the, as I

22   say, it's on the logic that the needed adjustment for each

23   of these factors varies from one chemical to another.  And

24   we can get empirical data on how that happens among

25   chemicals and then for a particular chemical we don't have



 1   any such data, we don't know where it is, so we have to say,

 2   well, we'll assume it's drawn from the same distribution

 3   that applied to the other chemicals.

 4             But whatever it is, it's one of those things and

 5   so in some sense the one value applies to everything.

 6             That's true for that one, but in a way you're

 7   making the point that I'm trying to make here, which is that

 8   in this factor for sensitive humans, human sensitivity does

 9   vary amongst everybody for the one chemical and so everyone

10   isn't the most sensitive human.  So, in fact, the human

11   population that we're interested in assessing the risk that

12   was going to be spread through this distribution and the

13   whole distribution will apply, and we're mixing those things

14   together and maybe it would be a good idea to keep them

15   apart.

16             Dale will have a lot more to say about that in a

17   minute.

18             There are some other things.

19             The subchronic to chronic and the LOEL to NOEL

20   distributions, how you would actually do those empirically

21   there's some real questions about that I won't get into now.

22             What I'd really like to say is, though, this is a

23   good start, but then the question is how do you empirically

24   get those distributions, and how do you think about them.

25             So what I'd like to spend the rest of my time on



 1   is some questions about thinking about those distributions.

 2             Let's start with the animal-to-human distribution.

 3             Now, we have a factor of ten.  We say take the

 4   animal NOEL dose, or better maybe a benchmark dose, and

 5   divide it by ten to apply to humans.  Why do we do that?

 6   What do we think we've done by that?

 7             In my quizzing people about this, I've come up

 8   with three kinds of answers that people give of what they

 9   think they have done when they have done that.

10             And this diagrammatically shows those three

11   answers.

12             What I plotted here on the log scale across the

13   bottom is the human-to-animal ratio of equally toxic doses,

14   so that a factor of .1 means that humans will have the same

15   response to a dose as a tenth as much as animals, and ten

16   means they can have ten times more of the dose, on the dose

17   scale that you're using here, before they have the same

18   response.

19             And then on the vertical axis is relative

20   frequency over chemicals.  And there are some people who

21   say, well, I think that we do this factor of ten because

22   there's -- chemicals vary from one to another in what the

23   relative toxicity is.  There's some that humans are more

24   sensitive to, some that they're less sensitive to.  Maybe on

25   average in doses in milligram per kilogram per day are



 1   equally sensitive.

 2             But we want to allow for the fact that humans

 3   might be particularly sensitive for this chemical, and so a

 4   factor of ten, that UFA there, that I've drawn as sort of a

 5   factor of ten down from the one-to-one line, represents

 6   basically making sure out in the tail of that distribution.

 7   Okay.

 8             But obviously on average that's very conservative,

 9   because if on average the right answer is that they're about

10   equally toxic on these dose scales, that sometimes humans

11   are actually less sensitive and yet we're assuming that

12   they're ten times more sensitive.  You could be up to, say,

13   100-fold conservative on this.  On the average you'd be

14   about 10-fold conservative or so.

15             The second idea is that, no, humans really are

16   more sensitive in milligram per kilogram per day basis than

17   animals.  This is what we do in cancer risk assessment with

18   the so-called surface area scaling rule of the three-fourths

19   power of body weight scaling rule is now used at EPA, where

20   we're saying that at least on a milligram per kilogram per

21   day basis, humans would have a smaller does to get the same

22   response.  There's a systematic difference that we're

23   actually having to do.

24             So some people say, well, from mice to humans on

25   that surface area scaling is about a factor of 13 or 12,



 1   from rats to humans is about a factor of six or seven, so

 2   ten is sort of an approximation of that.  We're making that

 3   correction the way you do with surface area scaling of doses

 4   to get equally toxic doses in animals and humans.

 5             Under that interpretation you need that whole

 6   factor of ten, and maybe it isn't quite enough, just to get

 7   the equally toxic level in humans that you had in the

 8   animals.  It's not conservative at all, it's just getting

 9   you to the dose that has the same effect.

10             And then you allow, well, there's still variation

11   in that from chemical to chemical.  So sometimes maybe it

12   isn't as bad as all of that, this tail, but sometimes it's

13   maybe even worse.  So in some fairly large fraction of cases

14   that factor isn't enough to allow for the animal-human

15   variability.

16             A third kind of conception of this factor of ten

17   is -- well, before I get on to B here, obviously the whole

18   distribution has just shifted over by a factor of ten.

19   That's the idea is that's where the center of it is.

20             The third idea is that in some sense you're doing

21   both A and B.  You say, well, humans really are more

22   sensitive than animals, and chemicals really do vary from

23   one another to one another, but when you combine those, a

24   factor of ten is enough to make sure that you've cut off

25   most of the tail, so that the distribution is shifted over



 1   and it has some spread, but ten is still considered enough

 2   to cover both of those effects.

 3             Okay.  As I say, you talk to different people and

 4   you have different ideas about what you thought you were

 5   doing.

 6             And a lot of the discussion about what to do with

 7   the uncertainty factors, and some of the work that's been

 8   done elsewhere, not at Harvard, but elsewhere, where you try

 9   to turn the factor of ten into a log normal distribution for

10   which 10-fold is a 95 percent level, they're basically

11   trying to do something like this, except that they actually

12   draw it here starting at one and assuming that there are no

13   chemicals for which humans are more sensitive than animals,

14   and it's only a degree of how much less sensitive they are

15   with 95 percent of chemicals being 10-fold or less.

16             DR. BLANC:  Did you say that backwards?

17             DR. GLANTZ:  I'm totally lost.

18             DR. RHOMBERG:  I may have said it backwards.

19             DR. BLANC:  What you mean is that there's no

20   chemicals for which humans are not ten times as sensitive as

21   animals?

22             DR. RHOMBERG:  Yeah.  I think that's what I mean.

23   I wish I could really draw it.

24             DR. BLANC:  We're not ten times more resistant

25   than animals or --



 1             DR. RHOMBERG:  Right.  We're never more resistant

 2   than animals.  We're rarely more than ten times more

 3   sensitive.

 4             CHAIRMAN FROINES:  Rarely more than ten times.

 5             DR. RHOMBERG:  Right.  There are few chemicals.

 6   If we assume that we're ten times more sensitive than

 7   animals, we will cover most of the cases.  That's what I was

 8   trying to say.  I don't know what I actually said.

 9             DR. BLANC:  That is what you said, it's just that

10   it's a double negative, and I'm not quite sure I understand

11   how that curve would be drawn, because you're assuming,

12   therefore, that everybody -- that 10-fold would be good

13   enough, but you're not saying that you could draw the

14   distribution the other way.  What would that distribution

15   really look like?  Is that one of those up there?

16             DR. RHOMBERG:  I'm not sure I understand your

17   question.

18             You could draw it the other way out here and say

19   humans are usually more resistance on this dose scale.

20             DR. HATTIS:  You're referring to Paul Price's

21   work.

22             DR. RHOMBERG:  Right.

23             DR. HATTIS:  Paul Price draws it so that

24   essentially it would look like it would start from here and

25   then trail off down here, so it would drop to zero here.



 1             DR. RHOMBERG:  Right.  There would be nothing on

 2   this side, and there would be not very much on this side of

 3   one, so it would be a distribution.

 4             DR. BLANC:  I see.  Thank you.

 5             DR. RHOMBERG:  With all that, why was I saying all

 6   of that?

 7             DR. GLANTZ:  Excuse me.  You've like totally

 8   confused me.

 9             DR. RHOMBERG:  Oh, dear.

10             DR. GLANTZ:  It's fine.

11             DR. BLANC:  He's often confused.

12             DR. GLANTZ:  I'm often confused.

13             Anyway, could you just back up to B and just kind

14   of go over that again, and then C, and how they compare to

15   A, and make it like the dumbest possible terms.

16             DR. RHOMBERG:  Oh, dear.

17             B is the idea that humans really are on average

18   10-fold more sensitive than animals.  So you need that

19   10-fold because we're 10-fold more sensitive.

20             DR. HATTIS:  The reason why in part is that we

21   process things slower than animals.  So that if you've

22   gotten a given amount of milligram per kilogram in you, we

23   take a lot longer to get rid of it, more or less as the body

24   weight to the record.

25             DR. GLANTZ:  But then why is this, if I'm like



 1   boring everybody with these questions, stop me, but then why

 2   does -- do you have the distribution dropping down about at

 3   one?  I mean, why are you putting the distribution where you

 4   are?

 5             DR. RHOMBERG:  Well, I don't have a reason to put

 6   it in any particular place, because I don't know -- this B

 7   doesn't say anything about the width of that distribution,

 8   but it just acknowledges that there is some width to it.

 9             But basically in A you're correcting for, you're

10   allowing for the uncertainty.  You're allowing for the

11   spread of the distribution.  But you figure it's centered on

12   one.

13             B, you're looking for the centering.  You say it's

14   centered on one-tenth, but you don't really say anything

15   about the spread of the distribution.  But even though you

16   haven't said anything about it, it's still there, which

17   means that in some sense there will be some chemicals that

18   are yet worse than that and some chemicals that aren't as

19   bad as that.

20             DR. GLANTZ:  Let me try and restate what I think

21   you're saying and correct me if I'm wrong.

22             In A what you're saying is the humans and the

23   animals are equally sensitive, and that's why you have the

24   distribution centered on one?

25             DR. RHOMBERG:  Right.



 1             DR. WITSCHI:  Something got lost, and this is the

 2   scaling.  The top one assumes on a mig per kilo basis, and

 3   the second one the scaling should be done according to body

 4   surface, which then, according to the fact you have in your

 5   next thing, would bring it back to ten.  So a dose of one

 6   milligram per kilo in a mouse would correspond to a dose of

 7   .1 milligram per kilo in man.  That's if you scale by the

 8   body surface.

 9             DR. GLANTZ:  Okay.  But that's getting me even

10   more confused.

11             I just want to be one step at a time.

12             A, you're basically saying that the humans and

13   animals are on average equally sensitive.

14             DR. RHOMBERG:  Right.  On the dose scaling that

15   you have chosen to draw.

16             DR. GLANTZ:  The thing on the horizontal axis is

17   the dose to get a given effect.

18             DR. RHOMBERG:  Right.

19             DR. GLANTZ:  So what you're saying in B then is

20   that the humans are ten times more sensitive than the

21   animals.

22             DR. RHOMBERG:  Right.

23             DR. GLANTZ:  Because one-tenth of the dose creates

24   the same effect.

25             DR. RHOMBERG:  Right.



 1             CHAIRMAN FROINES:  Peter is saying that's because

 2   of inherent differences between the animal and the human.

 3             DR. GLANTZ:  Right.  Well, that's another

 4   question.

 5             But then so then what's the point you're making

 6   when you say there's an uncertainty?

 7             Oh, I see.  You're saying that uncertainty factor

 8   of ten in B takes into account the fact that the

 9   distribution is shifted like that?

10             DR. RHOMBERG:  Right.

11             These are two fundamentally different ideas about

12   what you've accomplished with your factor of ten.

13             The first one is that you've allowed for the fact

14   that there are some compounds for which humans are

15   particularly sensitive and you want to make sure you catch

16   all those.

17             The second one is you say all chemicals on average

18   humans are 10-fold more sensitive and you want to correct

19   for that to get to the same point, basically the equally

20   toxic dose.

21             But there's moreover distribution around that that

22   you haven't taken into account when you've applied the

23   factor of ten.  You've just dropped that issue from your

24   analysis.

25             DR. GLANTZ:  So then what does C do then?



 1             DR. RHOMBERG:  C says, well, we think that there's

 2   some of both of those in there, that humans maybe really are

 3   more sensitive than animals, but not by a factor of ten, by

 4   something less, and there is variation from chemical to

 5   chemical, so the distribution is shifted some, but that

 6   nonetheless this tail is such that the factor of ten cuts

 7   off most of it, that we've covered most of the chemicals in

 8   a factor of ten.

 9             DR. GLANTZ:  But in B it wouldn't?

10             DR. RHOMBERG:  Right.

11             Most people, and here when they are pressed to

12   something like C, but the problem with that is that then

13   that factor is partly a correction, a systematic correction

14   for extrapolation, the size of which we don't really have a

15   good handle on how big it is.  And partly an allowance for

16   uncertainty.

17             And once we try to get empirical about this, we

18   will want to not just -- these are mental models in these

19   graphs.  We want to get some real data to try to sort out

20   which of these things we're really doing, and what I'm

21   trying to point out is that we're sort of confused about

22   what we thought the thing meant in the first place, and

23   therefore what we are accomplishing by those things, and

24   therefore what kind of data would answer those as empirical

25   distributions.



 1             DR. GLANTZ:  Just continuing disrupting you.  So

 2   in getting back to what Pete was saying, what you're saying

 3   is the reason for going from A to B in this thing has to do

 4   with the dose scaling, is that what you're saying?

 5             DR. RHOMBERG:  Yeah.  I mean, one of the ways that

 6   you can handle that is if you think there's that systematic

 7   effect you try to take into account by dose scaling, and if

 8   you have done that, and then plot these doses in the scaled

 9   doses, then you're getting back to that.

10             So one way to handle that would be to scale the

11   doses and then apply the factor of ten.  And then you get

12   back to the A conception.

13             I'll probably go over this really quickly here.  I

14   just wanted to point out the different kind of dose scaling

15   methods that are usually used to point out that for cancer

16   we usually use surface area scaling in the past.  EPA is now

17   talking about body weight to the three-fourths power

18   scaling, which is rather similar, but that noncancer we've

19   traditionally talked about body weight scaling, and is that

20   a good idea.

21             And I just wanted to show a few little things to

22   at least cast some doubt in your minds on whether that's a

23   good idea.

24             This is an understudied area about scaling and

25   noncancer.



 1             CHAIRMAN FROINES:  Are you going to say anything

 2   about the two-thirds to three-fourths?

 3             DR. RHOMBERG:  A little bit later, but I'm not

 4   going to say lots.

 5             CHAIRMAN FROINES:  That's okay.

 6             DR. RHOMBERG:  There's a lot to talk about here.

 7             CHAIRMAN FROINES:  Dale, are you?

 8             DR. HATTIS:  Not much.

 9             DR. RHOMBERG:  I'll be happy to come back sometime

10   and talk about that, but I will talk about it some, unless

11   I've completely run out of time, which seems likely right

12   now.

13             On the top, this is work from Sandy Baird, and

14   it's plotting the fraction of the time that the various

15   kinds of experiments that you do in noncancer standard data

16   set end up being the most sensitive sex and species.

17             DC stands for dog chronic, RC, rat chronic, RR is

18   rat reproductive.

19             Basically when you have all of these studies in

20   dog, rats and mice and rabbits and so on like that, most of

21   the time the dogs are the most sensitive species.  This

22   should be telling you something.  The bigger the species is

23   the more often it's the most sensitive.  Somehow that's

24   suggesting that maybe we're not doing the dose scaling

25   right.  If larger species are systematically more sensitive



 1   and the data sets we see and the animals, doesn't that

 2   suggest that some sort of better scaling that allows for

 3   larger sensitivity in bigger species might be in order.

 4             If you use the surface area scaling in noncancer

 5   the way it isn't traditionally done, and then redecide which

 6   is the most sensitive sex and species, it's much more evenly

 7   divided amongst the different experiments that you do.

 8             CHAIRMAN FROINES:  Lorenz, can you, Ray or Elinor,

 9   can you folks in the back see this or do we need to dim the

10   lights?  You can see?

11             FROM THE AUDIENCE:  You can see okay.  I can't

12   read the details.

13             CHAIRMAN FROINES:  Peter, see if you can lower

14   them a little bit.

15             DR. RHOMBERG:  I apologize, but in fact the

16   details, you know, I'm using this as an illustration as a

17   general point.

18             CHAIRMAN FROINES:  That better?

19             FROM THE AUDIENCE:  Yeah.

20             DR. RHOMBERG:  I want to stand out of people's

21   way.  Well, okay.  Sorry.

22             The point is that the dog is usually the most

23   sensitive species, if you do scaling in the traditional

24   milligram per kilogram per day way, and if you do it by

25   surface area scaling, it's much more evenly divided amongst



 1   the various species, suggesting that there's maybe some

 2   systematic effect there that we haven't taken into account.

 3             Another relevant piece of information is the paper

 4   by Travis and White --

 5             DR. GLANTZ:  Is the surface area scaling got a

 6   power in it?  The two-thirds or --

 7             DR. RHOMBERG:  Yeah.  Two-thirds power of body

 8   weight.

 9             The other piece of evidence is this paper by

10   Travis and White in 1988 in risk analysis, where they looked

11   at 23, I think, antineoplastic drugs.  In these you have a

12   good measure of human toxicity, because you give these drugs

13   up to the level that are toxic in humans, clinically, as

14   part of treatment, and you can also get toxic levels in

15   animals from tests, and they basically -- this takes a

16   little bit of explanation to see what this graph is, they

17   actually made allometric regressions to see what power body

18   weight optimally scaled the animal toxic dose up to humans

19   and back and forth, among species, what equalized the toxic

20   doses among species the best, what power of body weight.

21             For each chemical they plotted that power, so

22   here's the power of body weight that is suggested by each

23   chemical and here are the chemicals.

24             You can see that, yeah, there's variation among

25   from chemical to chemical that we're talking about, but



 1   they're varying around the line that's the .75 power of body

 2   weight.  I think the actual number here is .74.

 3             DR. WITSCHI:  Actually, I object to using this

 4   paper as evidence for this, because the original paper who

 5   brought this up was written in about 1963, by, among others,

 6   Fryrick, and this is the first paper where the thing from

 7   the body scale it comes.  It's an old paper, about early

 8   '60s.  And those guys looked at all the toxicity of

 9   anticancer drugs and the then available toxicity in mice.

10   They came up with the scale in fact of two-thirds.  So

11   that's not new.

12             DR. RHOMBERG:  Oh, no.  First of all, this is

13   already 1988.  It's already not new.  But it wasn't new when

14   they did it, you're right.  Pyrite did it back in the '60s.

15             They just did it again with more data, more

16   compounds.

17             And since then, there have been compounds added to

18   this.

19             The point I want to make here is that here is the

20   suggestion for noncancer toxicity.  This is noncancer.  It's

21   not acute, but it's sort of five- or seven-day dosing,

22   subacute or whatever you want to call it, that the

23   three-fourths power of body weight is the weight to scale

24   these, not milligram per kilogram.

25             So maybe our assumption that doses are equally



 1   noncancer toxic in humans and animals in milligram per

 2   kilogram per day needs to be revisited.

 3             DR. WITSCHI:  There's no reason why it's surface

 4   is more appropriate, because surface is related to blood

 5   flow and blood flow is related to metabolism.

 6             DR. RHOMBERG:  Right.  I have lots of talks, I've

 7   written big papers on this myself about why such scaling

 8   ought to apply.

 9             I guess I will just refer to all of those right

10   now.

11             I just want to acknowledge the fact that there is

12   a question here about how the best way to do that is.

13             And I want to very briefly treat this here.

14   Basically, what Travis and White were doing was using almost

15   an allometric equation, where you relate the unscaled

16   factor, the size of the dose in milligrams necessary to

17   produce a toxic effect, or milligrams per day actually, to

18   some power of body weight where W is body weight.

19             And basically it's easier if you take the log of

20   through the both sides of this equation and you get a

21   straight line on a log log plot, where the log of the

22   feature is equal to a constant plus the slope of the line

23   times the log of body weight.

24             So basically I want to address the question what

25   does scaling doses really mean.



 1             Here's an example of scaling doses badly.  What I

 2   mean by scaling badly, I'm saying we do one of those log log

 3   plots here, here's the milligram per day, otherwise unscaled

 4   by body weight, toxic dose, that's equally toxic, the

 5   species A, B, C, D, E, F, G, plotted against the body mass

 6   of those species.

 7             And you can see that this line that you are saying

 8   let's scale it this way is inappropriate in that it's

 9   systematically overpredicting here and systematically

10   underpredicting here.  So that the actual line is flatter

11   than that.

12             What if you invoked this as the way you want to

13   scale the doses, that's a dose scaling theory that you would

14   invoke by this line.

15             DR. GLANTZ:  Where did that line come from?

16             DR. RHOMBERG:  You have it from a theory.  You say

17   I want to do milligram per kilogram per day scaling, because

18   I think that that's the right thing to do.

19             DR. GLANTZ:  So that would be the line for

20   milligram per kilogram scaling, is that what you're saying?

21             DR. RHOMBERG:  It's close to it.  Yeah, well.

22   Yes, it is the milligram per kilogram scale.  It's got a

23   slope of one.

24             The slope of that line will be the power of body

25   weight, so it's got a slope of one.



 1             So if that's --

 2             DR. FUCALORO:  Isn't that 1.2 to the 1.2 power?  I

 3   mean, I'm just reading.  It doesn't look right to me.

 4             DR. RHOMBERG:  It is 1.2.  I exaggerated this a

 5   little to make it look good from the back of the room, okay.

 6             But say somebody says I want to scale doses by the

 7   1.2 power of body weight, because I think that's the right

 8   thing to do and the point is that any such theory

 9   corresponds to a statement about what the slope of this line

10   is.  The slope of this line is now going to be 1.2.  Okay.

11             And if you've chosen badly your scaling, then you

12   get this kind of systematic bias around things.

13             Moreover, if you look at the deviations off of

14   this line from A down to the line here and C down to the

15   line and G up to the line and so on, those deviations, those

16   residuals around the line are the variations from species to

17   species in what's the toxic dose.

18             And so what we're aiming at now is some sort of

19   characterization of how different a dose's toxicity is from

20   one species to another, because we want to take one of those

21   species and extrapolate to another, namely, humans.

22             The point is that a badly chosen line will have

23   big residuals and therefore it will have a big variation in

24   the equally toxic doses when scaled according to this power

25   of body mass that you're invoking here, the 1.2 power of



 1   body mass.

 2             Moreover, it will be a bias such that the small

 3   species are at this end of the distribution and the big

 4   species are at that end of the distribution, because these

 5   residuals are on this side, and those residuals are down

 6   under on that side.

 7             So basically that shows that there will be a bias

 8   in here, so the distribution is wider than is necessary when

 9   you're scaling badly.

10             And also it introduces bias in that it's not just

11   going from any species to any other species.  If you're

12   going from a small species to a big species, you get

13   different kind of span here than from any pair of species.

14             So if you -- basically what I'm saying is if you

15   make an empirical distribution of relative toxicity in, say,

16   mice and rats and mice and dogs, or whatever like that, and

17   you say we'll just use milligram per kilogram scaling for

18   now, on the assumption that that's the right thing to do,

19   the choice of doing that scaling is making that distribution

20   of relative toxicity amongst species wider than it needs to

21   be and it is ruining your assumption that you want that you

22   want to be able to draw from that randomly when you are

23   doing the Monte Carlo analysis.

24             So you really have to settle the scaling question

25   before you do this.



 1             Here's scaling well the same data and now I've fit

 2   this line by regression, and it's an artificial example so

 3   it comes out to be the .75 power of body weight.  I say,

 4   okay, now if we realize this, and of course for any

 5   empirical set of data you might want to actually do the

 6   regression, and see what's the best thing.  It's got the

 7   properties of regression, the residuals are minimized.  And

 8   assuming that this is really linear on this log log scale,

 9   they will also be unbiased in terms of low here and high

10   there.

11             So you get a narrower distribution.  In other

12   words, you're more certain about your cross-species

13   extrapolation, you're optimally certain about that for any

14   kind of scaling, and you're also unbiased, and any other

15   solution will be less optimal.  You'll increase your

16   apparent uncertainty and will introduce bias.

17             So this is basically an argument why you have to

18   settle the cross-species scaling question before you really

19   grapple with even an empirical distribution of relative

20   toxicity across species.

21             Now, I've done some work on trying to do some of

22   this empirical stuff on one or two issues ago in risk

23   analysis on LD 50s, these acute single oral doses that are

24   lethal.

25             And I did lots and lots of species.  This is just



 1   showing rats and mice.  And you get this ratio of the log,

 2   and it shows that, yes, it does differ over species.  There

 3   are 4,600 and some chemicals in here.  It's sort of peaky

 4   compared to a real normal distribution, but in a way this is

 5   a common log, so it's basically saying the factor of ten

 6   either way is really pretty much covering this.

 7             And in fact that works pretty well for other

 8   species pairs as well.

 9             You can argue that this now is evidence that sort

10   of concept A that I had earlier on the graph that we were

11   arguing about is applying for this kind of noncancer

12   toxicity, namely, single dose lethality.

13             It also suggests that these are doses expressed in

14   milligrams per kilogram that maybe this is evidence that

15   milligram per kilogram scaling is working for noncancer.

16             But you will notice that now there's a conflict in

17   my results on single oral lethal doses, LD 50s.  I find

18   milligram per kilogram scaling working very well to predict

19   across species.

20             Travis and White found milligram for three-fourths

21   power to work well for the antineoplastic agents.

22             DR. BLANC:  Yeah, but you're looking at a mouse to

23   a rat.  They're really very similar in terms of their --

24   assume this area of the body weight times --

25             DR. RHOMBERG:  I'm just showing that --



 1             DR. BLANC:  What happens when you do mouse from

 2   monkey?

 3             DR. RHOMBERG:  Same thing.  Although there are as

 4   many data for monkey.  But I did mouse, rat, monkey, dog,

 5   hamster, pig.

 6             DR. BLANC:  How about mouse to dog, do you have

 7   that?

 8             DR. RHOMBERG:  I don't have it in my -- but the

 9   same pattern held across all pairs of species, no matter --

10   I did a regression of the size of the difference with the

11   size of the difference in body weight.  If there's scaling

12   by other than the one power of body weight, you should get a

13   trend that the acceptability of this milligram per kilogram

14   scaling would get worse and worse, the more different the

15   body weight was between the species compared, and there was

16   no trend.

17             If you look at all the species together in a big

18   regression it comes out to be the 1.01 power of body weight.

19             This is in risk analysis.  I'll let you read the

20   paper rather than go on about it.  But I'm just showing this

21   one, because I have a lot of data from mouse to rats, but I

22   have it for ten species in all pair-wise comparisons.

23             DR. GLANTZ:  Just, I just want to make sure I

24   understand this.  So what you're doing is the thing, the

25   distribution is over different chemicals, right?



 1             DR. RHOMBERG:  Yes.

 2             DR. GLANTZ:  You take chemical A and you find the

 3   LD 50 in the rat and in the mouse and you repeat the ratio.

 4             DR. RHOMBERG:  Right.

 5             DR. GLANTZ:  And then you take the chemical B and

 6   you do the same thing and those are --

 7             DR. RHOMBERG:  Right.

 8             If you want to see that, I do have a quick extra

 9   slide.  Here are the actual data where I plot the LD 50 in

10   the mouse versus the LD 50 in the rat.  You see, yeah, there

11   is variation, but a one-to-one line corresponding to

12   milligram per kilogram per day scaling works really well,

13   whereas the prediction of milligrams per three-fourths --

14   per kilogram per three-fourths per day, you know, clearly is

15   missing the middle of that cloud of points and is

16   systematically wrong, although there is variation around it

17   and there's the endpoints on either side of that line as

18   well.

19             It's only when you got a lot of data that you can

20   really tell things apart for species that are so similar in

21   body weight.

22             But I have such plots for all of my comparisons.

23             Well, I wanted to just call attention to this sort

24   of contradiction between this result from milligram per

25   kilogram on single oral lethal doses and Travis and White's



 1   results of three-fourths power of body weight.

 2             What are we going to do for scaling for noncancer?

 3   Are we going to do it by milligram per kilogram or we going

 4   to do it by three-fourths power?  Why are those two things

 5   different?  Is one of us wrong?

 6             And my suggestion, and this is sort of speculation

 7   now, is that, no, we're both right, but there's a

 8   fundamental difference between single dose scaling and

 9   repeated dose scaling.

10             And I want to talk about why I think that might be

11   here briefly.

12             And if you don't follow this, I'm sorry, because

13   it gets a little bit complicated.  You have to sort of have

14   thought about this a while to sort of see what I'm getting

15   at.

16             This is a simplified hypothetical model, but it's

17   a real model and that there are real equations underlying it

18   that will have the properties that I'm describing.

19             Basically what I'm suggesting is that for a really

20   severe single dose endpoint like acute one dose lethality,

21   the real issue is how much reserve in some sense does the

22   body have to fight off this onslaught?  And what if it seems

23   reasonable to suppose that the reserves that you have are

24   proportional to your body size.  It's sort of your standing

25   level of capacity or ability to fight something off or



 1   absorb something without harm.

 2             So basically --

 3             DR. HATTIS:  Or just to dilute it.

 4             DR. RHOMBERG:  Or just to dilute it, right,

 5   exactly, to the point where the concentration is below that

 6   which is going to cause a harmful effect.

 7             If that's the case, then a human and a mouse

 8   should have about the same milligram per kilogram size to

 9   the volume that they're diluting it by dose for a single

10   lethal dose, even though the human, because of the slower

11   physiological approaches of humans, will do that slower,

12   absorbs it slower, and then there's recovery as this

13   depleted whatever it is or damage starts to repair itself.

14             Now, I built in a repair or recovery function

15   that's proportional to the loss from the control level in

16   here.

17             So the process takes a little longer in time, but

18   nonetheless how far you dip into your well of reserves is

19   the same depth, and in some sense the same milligram per

20   kilogram per dose should be equally toxic.

21             If you then scale by the three-fourths power of

22   body weight, the human could get off easier, because it has

23   to dip in that well less deeply.

24             DR. FUCALORO:  Excuse me.  I'm missing it.  The

25   concentration is the concentration of what?



 1             DR. RHOMBERG:  Compounds somewhere in the body.

 2             DR. FUCALORO:  Somewhere in the body.

 3             At time zero tell me what happens.

 4             DR. RHOMBERG:  At time zero you say gavage the

 5   animals or you give the single oral dose to humans, however

 6   you do it, I guess you wouldn't gavage, but something

 7   equivalent, and it's absorbed.  And as it's absorbed it's

 8   now depleting something or causing some damage.  I'm just

 9   called -- I'm sorry, I should have pointed out, I'm talking

10   about concentration of some substance in the body that is

11   depleted with interaction with --

12             DR. FUCALORO:  Because it looks like just the

13   opposite.  Not the concentration of the toxin, but some

14   concentration of a important chemical the body needs that is

15   depleted and then the body recovers.

16             DR. RHOMBERG:  Yes.  I'm sorry.  My apologies.

17   I'm trying to rush here and I'm leaving out details.  And

18   that's important.

19             The model is something like glutathione depletion

20   here, but it doesn't have to be that literally.  It can be

21   any kind of damage where going down is sort of damaging,

22   reducing some level of something that you need that

23   recovers.

24             DR. WITSCHI:  A model is nice, but we have lots of

25   what happens after an LD 50 in mice.  But I would wonder how



 1   many data we have for after LD 50 in man.  How many

 2   experiments do we have?  And now we're talking about some

 3   fundamental differences because the most trivial one of

 4   those is dose rate, and you only can compare what happens

 5   after an LD 50 in mouse, what happens in man if you have

 6   some real data, but you do not have.

 7             DR. RHOMBERG:  Well, there are some real data for

 8   Travis and White for the antineoplastic agents.  They aren't

 9   single doses.  They are multiple doses.

10             In fact, the upshot of my previous analysis on LD

11   50s, for which there are like four human LD 50s in this

12   database, suggests the sort of milligram per kilogram

13   scaling and that seems to be in conflict with these results

14   for multiple dose things on antineoplastic agents that

15   Travis and White has.

16             So what I'm trying to do is a conceptual model,

17   which I admit is conceptual and doesn't have data to say

18   what could be going on here to reconcile those two empirical

19   results.  The empirical results both exist.

20             DR. GLANTZ:  Now, would your graph of the human,

21   if you carried the time scale out, it would look like the

22   mouse, just slower?

23             DR. RHOMBERG:  Yes.  It would be just seven times

24   slower on the time scale.  And if you do that trick of

25   dividing time by the fourth power of body weight, those two



 1   curves would look the same, for those familiar with that.

 2             I just want to contrast that then with a repeated

 3   dose scenario.

 4             DR. FRIEDMAN:  Before you leave that, I'm not sure

 5   I understand.  Are you saying that this shows that the

 6   milligrams per kilogram scaling works?  This is a situation

 7   where it works?

 8             DR. RHOMBERG:  It would work, yes.

 9             DR. FRIEDMAN:  Why does it work if the mouse

10   recovers so much quicker and the human recovers so much

11   shorter?

12             DR. RHOMBERG:  This is on the hypothesis that it's

13   somehow that the depth of depletion of this depleted

14   substance or the amount of damage that's caused is what the

15   result of lethality.

16             DR. FRIEDMAN:  The distance that it drops down is

17   the key point?

18             DR. RHOMBERG:  Right.  So if you drop down to 50

19   percent of your background level of this glutathione, or

20   whatever like that, you have a toxic reaction.  If you don't

21   dip that far down, you don't.

22             I admit, it's hypothetical, but, as I say, I'm

23   trying to come up with a construct that would explain these

24   results and then it becomes a hypothesis to follow.

25             And so the subsequent part of the hypothesis, what



 1   happens on multiple dosing and the idea here is that in

 2   multiple dosing the issue is fundamentally different, it's

 3   not your reserve capacity that you carry around with you

 4   every day ready for whenever it may be challenged, but

 5   rather your ability to recover.  In some sense what you have

 6   is with repeated dosing some damage every day, and then in

 7   some sense recovery every night if you want to sort of split

 8   it into sequential parts.  Obviously, they're going on

 9   simultaneously.

10             So then the question is if humans are slower to

11   recover, why would that be, because of the slower

12   physiological processes and larger species that draws out

13   their time scale vis-a-vis rats and mice.

14             Then a daily dose of a certain amount in a mouse

15   that it can just recover from every day, so that it just

16   barely avoids dipping down or below that level that it's

17   toxic or just barely grazing that level of depletion that

18   can't be sustained.

19             The daily dose has to be in proportion to that

20   recovery rate, and humans with their slower recovery will

21   have the same amount of depletion every day and less

22   recovery every night, are going to have a bigger effect.

23             DR. FUCALORO:  We'd have difficulty, really, I

24   mean, I can see what you're drawing there clearly, but these

25   results from your model, your mathematical model, how you



 1   have your equations for depletion and recovery; correct?

 2             DR. RHOMBERG:  Yes.

 3             DR. FUCALORO:  And we haven't seen those.  I

 4   suppose you have some somewhere.

 5             DR. RHOMBERG:  Yeah, but they're hypothetical.

 6             DR. FUCALORO:  I know.  And you can change those

 7   models.  I don't know if they're kinetic, first order

 8   kinetic types or all sorts of types of modeling for

 9   depletion.

10             DR. RHOMBERG:  It's first order absorption, first

11   order reaction to the compound with this endogenous

12   compounds being depleted.

13             DR. FUCALORO:  That's what I was thinking.

14             DR. RHOMBERG:  It's got a recovery function that's

15   proportional to the amount of depletion.

16             DR. FUCALORO:  Right.

17             DR. RHOMBERG:  But that operates more slowly, and

18   all these things are operating more slowly in humans than in

19   rats.

20             DR. FUCALORO:  I guess what I'm saying is that by

21   showing the difference between the three-quarter scaling, I

22   mean, but by using different rate constants and so on, you

23   can get almost anything, any types of curves.

24             DR. RHOMBERG:  I suppose that's true, but two

25   things.



 1             One is this is referring back to the empirical

 2   results, the Travis and White, and the stuff on the LD 50s,

 3   that need to be reconciled somehow or other, so it's not

 4   just that I'm making this up.  There's a phenomenon that

 5   needs to be explained.  If this isn't the explanation, then

 6   we need another one.

 7             CHAIRMAN FROINES:  Let me interject here.

 8             I'm a little worried about time, because we've

 9   been going an hour now.  And we had thought this would be a

10   little briefer.

11             So would the panel hold their questions and let

12   him finish, and then we can have questions.

13             Go ahead, Gary.

14             DR. FRIEDMAN:  My only point is if we don't

15   understand the point --

16             CHAIRMAN FROINES:  Absolutely.

17             DR. FRIEDMAN:  Then I feel it's important to

18   interrupt, just to get some quick clarification.

19             CHAIRMAN FROINES:  If there's a point of

20   clarification, but otherwise I think it would be easier if

21   he could finish a conceptual.

22             DR. RHOMBERG:  Okay.

23             DR. HATTIS:  Let me help.

24             CHAIRMAN FROINES:  No, no, no, Dale.  Sit down.

25   Sit down.  You'll get your turn.



 1             DR. RHOMBERG:  Okay.  I just want to, if I could,

 2   just quickly follow through with this argument, and I hope

 3   that you will see the point.

 4             The point is that if you scale the doses by the

 5   three-fourths power of body weight for each day's dose, and

 6   in some sense the recovery process is scaling that way as

 7   well, then you're keeping the recovery ability in balance

 8   with the damage that's caused every day.

 9             So in some sense you have this contrast.  With a

10   single dose the issue I'm proposing with a single dose the

11   issue is how deeply do you dip into your well of reserves.

12   And the well is about the same depth in humans as it is in

13   animals, and so you get a milligram per kilogram dose to do

14   that equally.

15             Over the long term, the issue is how much on the

16   well analogy, how much do you draw out of the well every day

17   compared to the rate at which stuff is seeping back in to

18   replenish the water that was drawn away.  And I'm saying

19   that in humans that seeping back, that recovery,

20   replenishment, is slower systematically and therefore the

21   amount of stuff that you can draw out every day would be

22   smaller as well.

23             So if that's something like that were true, that

24   would explain this result.

25             Moreover, the things that I rely on in making the



 1   equations for these models are patterns of differences

 2   across species that people think exist.  It's not that I'm

 3   making them up in order to make this phenomenon happen.  I'm

 4   making the rates scale, the three-fourths power body weight

 5   of which there's tons and tons of data to suggest that's a

 6   reasonable way to approach this problem.

 7             So let's just leave it at that.  I'm trying to

 8   just sort of suggest a reason for this, but I don't want to

 9   detract from the original message, which was we somehow have

10   to decide how to do the scaling for noncancer before we can

11   get this empirical distribution across species, and so we

12   have to grapple with this issue, and it's not an easy one,

13   and the answer for single doses might be different than the

14   answer for multiple doses.

15             So that was the main point.

16             Oh, boy.  I'm going to skip over this one

17   entirely.

18             I want to briefly talk about the sensitivity among

19   humans and the question of dose response and this will be a

20   good lead-in to Dale, because that's what he's going to be

21   talking about mainly.

22             All this was just about the animal-to-human

23   factor, but another one of those factors that we always have

24   is the question about average humans to sensitive humans or

25   something like that.



 1             And I just want to point out that we have this

 2   notion here that the whole reason why there's a dose

 3   response is because there's variation in sensitivity or in

 4   capacity in some sense among the population.

 5             So here I'm basically just showing this

 6   diagrammatically.  You have a population of varying reserve

 7   capacity that is now sort of eaten away to some degree by

 8   exposure to a compound and it's eaten away more by bigger

 9   exposures than smaller exposures, so that when you draw

10   samples from that distribution to test at various doses, at

11   higher doses you've exceeded the capacity of a larger and

12   larger fraction of the population, and that's why you get a

13   gradual dose response curve in terms of percent responding,

14   rather than a step function which says doses below this are

15   okay and doses above this are not, even if there's a

16   threshold for individual behavior.  And we sort of have this

17   model for what's going on underneath here.

18             The point I want to make is that if you remember

19   way back when I started I said one of the things we need is

20   a risk measure.  You have to be able to do dose response

21   analysis for noncancer and not just safety analysis.

22             We also want to get a distribution of sensitivity

23   among humans for the factor UFH.  And I just want to point

24   out in some sense those are the same things, that on the

25   assumption that this kind of process is responsible for the



 1   dose response, that somehow looking at that process, tracing

 2   out the shape of the distribution, because, after all, this

 3   curve is basically the cumulative distribution of the

 4   distribution of individual reserve capacities here.

 5             Basically, in doing one, you're doing the other.

 6             And that's why I think it would be good to take

 7   that factor, and rather than just building it into this

 8   equation as another distributed uncertainty factor, to look

 9   at it explicitly as a dose response function and look at

10   empirical data on sensitivity among humans, of which Dale

11   has a lot, as a way of looking at the shape of the dose

12   response curve in humans.

13             The question then is there's a lot of questions

14   about how do you figure the properties of that dose response

15   curve, its location and so on like that from the animal

16   data.  That's part of the development that we're still

17   working on in this question.

18             One of the things you can do is to say, well, we

19   will just use the dose response curve in animals, but one of

20   the difficulties there is that, you know, we know that

21   humans are more variable amongst themselves than animals.

22   They're genetically heterogenous, they have different

23   lifestyles, young and old, other exposures and so on.

24             And so basically the slope of the dose response

25   curve we get in animals ought to be steeper than the one we



 1   get in human, because it's reflecting a narrower

 2   distribution of variability, and so in some sense we have to

 3   decide what are we going to do about the fact that sort of

 4   for a given center of a dose response curve, it's going to

 5   go down more slowly in animals -- in humans than we observed

 6   in animals.

 7             And one of the answers to that is to look at the

 8   human data directly, which Dale will talk about.

 9             Well, this took longer than I thought, but I hope

10   that I've been able to get across some of the ideas here.

11             The point is that you can imagine approaching this

12   problem by empirical distributions of these uncertainty

13   factors, but once you start to look at the individual

14   factors and look at what data you would apply to them, some

15   of these other kinds of problems about dose scaling and like

16   that pop up, so it shows it to be a complicated problem, but

17   nonetheless I think getting back to his original motivations

18   is one worth pursuing for the benefits that we would have to

19   make noncancer risk assessment be able to better answer the

20   tasks that we're trying to have it do for us.

21             Thanks.

22             I would say any questions, but there have been

23   plenty of questions already.

24             DR. WITSCHI:  Just what's the discrepancy, you

25   said, between the Travis data and your data?  They scale on



 1   a body surface basis and you find better congruence by

 2   scaling on a weight basis?

 3             DR. RHOMBERG:  Yes.

 4             DR. WITSCHI:  There's one thing you note and you

 5   mentioned this about having research and so on, but the

 6   toxicity that comes from cancer drugs, they're really going

 7   to people who are already sick.

 8             DR. RHOMBERG:  Yes.

 9             DR. WITSCHI:  These are not normal people,

10   otherwise they wouldn't have gotten the cancer drugs, so

11   they cannot necessarily be compared to healthy animals in

12   which you do the LD 50s.

13             DR. RHOMBERG:  That's a good point.

14             In fact, the measure of toxicity is hard to make

15   exactly congruent.

16             You know, there's a series of discussions on that

17   point in the Travis and White paper, and one of the things

18   they bring up is that, well, these are animals in their

19   cages without any kind of treatment, and the humans,

20   although they were sick, were also in a clinical setting

21   where they were being monitored very carefully.  In some

22   sense those are offsetting kinds of biases.

23             But it's true that it's hard to make any of these

24   comparisons really strictly.

25             I would only point out that it wasn't simply mice



 1   and humans in the Travis and White analysis, but they also

 2   had other species as well that were done more or less on a

 3   comparable basis, and humans were the only ones that were

 4   actually in a clinical setting.

 5             DR. BYUS:  Does it work for glutathione?  I mean,

 6   you brought up that example.  It would be a lot more

 7   comforting to me if you had an example like glutathione

 8   mediated toxicity to show that this scale is better.

 9             DR. RHOMBERG:  There are some data on glutathione

10   depletion time courses in animals.

11             DR. BYUS:  Are they faster at recovering than

12   humans?

13             DR. RHOMBERG:  I'm not sure.  I haven't found any

14   data yet on humans for this, but I'm sure there must be.

15             DR. BYUS:  There must be plenty.

16             DR. RHOMBERG:  Right.

17             DR. BYUS:  It would be more comforting to me if --

18             DR. RHOMBERG:  If you had a real example.

19             DR. BYUS:  You had an example.  And you brought

20   that up, that's why I am asking.

21             DR. RHOMBERG:  Right.  I made the thing as a sort

22   of general argument, and not simply about glutathione, but

23   glutathione is something that you sort of think as operating

24   sort of like my general argument.

25             DR. BYUS:  But still, a mechanism that's known and



 1   there are a lot of mechanisms known for a lot of drugs and a

 2   lot of toxic agents, if you took, somebody did, work through

 3   that mechanism as opposed to a different mechanism, maybe

 4   that may be the answer.

 5             DR. RHOMBERG:  Right.  I haven't published that

 6   model yet, because I don't have a good example to stick it

 7   with, and just as a model it does seem arbitrary.  Its main

 8   attraction to me is that it explains this phenomenon and

 9   it's sort of in accord with what people sort of think is

10   going on.  So it's a fruitful hypothesis, but at this point

11   that's what it is.

12             CHAIRMAN FROINES:  If there are no questions from

13   the panel, let's ask DPR and OEHHA and others if they want

14   to make comments.

15             DR. MARTY:  I just had one question.

16             This is Melanie Marty from OEHHA.

17             Do you have these data sets where you can look at

18   nonlethal endpoints?  The lethality part bothered me,

19   because there's obviously no recovery from the lethal

20   effects.  That might be more useful to look at a nonlethal

21   single dose endpoint to try to compare with the more

22   subchronic or chronic exposures.

23             You know, and also obviously the chemotherapy was

24   not meant to be lethal.  It may have been, but they try not

25   to kill you with the drug.



 1             DR. RHOMBERG:  But it was very severe toxicity.

 2             It's a good point.  It would be interesting to

 3   look at things of different severity and also at chronic as

 4   well as the short-term things.  When we're talking about the

 5   repeated dose studies, it's really only four or five days,

 6   it's not a lifetime or anything like that.

 7             The difficulty is finding data that are taken

 8   systematically enough that you can actually generate these

 9   databases that you need to do this kind of analysis.

10             The reason I did lethality is because there were

11   tons of data on LD 50s all organized and luckily LD 50,

12   lethality means the same thing in most labs, whereas

13   hepatotoxicity might not, and so it's really hard to find

14   data that are really comparable in different species.

15             Even though we have a test system that we use a

16   lot of the time, it's hard to find it in different species.

17   We've got a system that we run in mice and that we don't do

18   the equivalent endpoint in rats or in rabbits, because we're

19   already doing it in mice.  So finding the data to answer

20   this kind of thing is a real challenge and sometimes you can

21   say in the whole world of the literature, you ought to be

22   able to pull out those things and organize them, but it's

23   been a task that's bigger than I've had funding, let's say,

24   to try to accomplish.

25             CHAIRMAN FROINES:  George, last question.



 1             DR. ALEXEEFF:  George Alexeeff with OEHHA.

 2             I think there's a couple of things.

 3             I think you tried to discuss some of the issues

 4   regarding the uncertainty factors that you think can be

 5   better quantitated, because I think that's kind of the areas

 6   you're focusing on, because there's lots of other areas that

 7   we use those uncertainty factors for, that maybe they can be

 8   quantitated.

 9             Like, for example, the severity, the effect that

10   we see in the animal versus what we might see in the human,

11   or the effect that we see in the animal and the effect we

12   might see in the human, so part of the safety factor is for

13   that, because we're looking at maybe a severe effect in the

14   animal, and we're thinking maybe the humans might be

15   responding in less severe.

16             Or the fact that study design of the animal study

17   is just really poorly designed and we're trying to make up

18   for the fact that we're just not really that confident in

19   the actual data set because they're not as well structured

20   as the cancer studies where you have large groups of

21   animals.

22             And the other thing I thought was that those

23   distributions that you showed where you had the Baird paper,

24   I know she called them adjustment factors or something like

25   that, but really to me I just don't see whether or not



 1   uncertainty distributions, because really if they're not

 2   based on actual data, if they're basically modeled

 3   distributions for that particular uncertainty factor, then

 4   really they're just uncertainty distributions.

 5             DR. RHOMBERG:  I wouldn't disagree with -- taking

 6   your points in reverse order -- I wouldn't disagree.  I

 7   think that might have been a better term for it.

 8             To the first point, you're right, there are other

 9   uncertainties that aren't covered here and it's one of the

10   challenges.  As I say, you've got these factors that you

11   have vague notions about what you think you've accomplished

12   with them, and everything that needs accomplishing, somebody

13   is sort of saying, well, some part of that factor of ten is

14   to account for this, and when you try to pars these things

15   out and apply data to some of them and realize that others

16   are yet unaddressed, that is when it gets challenging, but

17   the only solution is to really think about what it was that

18   you're trying to do with those factors, what you think they

19   represent, what's in there and what's not, so that you can

20   disassociate them later on.

21             Good point.

22             CHAIRMAN FROINES:  Okay.  We're going to take a

23   five-minute break before Dale starts, to give the

24   stenographer a chance to rest her wrists.  So let's -- but

25   it is going to be very brief.



 1             (Thereupon a short recess was taken.)

 2             CHAIRMAN FROINES:  Can we get started again.

 3             My strategy when we're not reviewing documents is

 4   to try and get the meeting finished before lunch.

 5             And when we're reviewing documents, clearly that

 6   takes us longer, often, especially when we're doing more

 7   than one.

 8             But when we're having a meeting like this, which

 9   is more catch-up, if we can push ahead we can finish in a

10   reasonable time so everyone can leave.

11             And our next speaker is Dr. Dale Hattis, who is at

12   Clark University, and the panel knows Dale from the diesel

13   workshop when he presented the results of his work on

14   estimating risk for diesel exhaust.

15             DR. HATTIS:  I'm going to try to follow on to

16   Lorenz's talk with -- Lorenz gave you a good deal of the

17   problem and I'm going to give you a straw man, quote,

18   solution, unquote, and how you approximately you get there,

19   although I don't want to be -- I want to be clear that I'm

20   not going to be presenting you with tests of the solution,

21   but basically a concept of how we tell whether our NOEL

22   uncertainty factor procedures or something else that we can

23   come up with are doing what we could reasonably hope that

24   they would do.

25             And basically the idea behind that is by making



 1   some kind of a quantitative specification for the RfD, if

 2   only to provide a benchmark for trying to assess whether

 3   it's working or not.

 4             So basically this is what the title, which is

 5   slightly different than what it appears in the program, a

 6   quantitative definition, and my basic hope, this is

 7   basically the outline of the talk.

 8             First, we'll talk a little bit about what we have

 9   to gain by this kind of thing.  Lorenz has already talked

10   about that to a degree.  The hope is that the hundredth

11   anniversary of the Lehman and Fitzhugh paper is rapidly

12   approaching in the year 2054, and we need to build, with the

13   rate at which these things do change, we do need to start

14   now to prepare ourselves with some sort of a replacement, so

15   that we'll be ready when that day comes.

16             I will then talk to you a bit about the

17   difficulties and costs of trying to make a quantitative

18   definition of the RfD.

19             And then I'll specify the elements of this straw

20   man and I'm going to suggest to you is a starting point.

21             And then the requirements for a viable system,

22   eventually.

23             So among the benefits are, first, that we stop the

24   lying.  Basically the implication is made with the current

25   RfD approach that we are talking about population thresholds



 1   and a dose that we expect will have no effect in a diverse

 2   human population.

 3             And I think that there is some hope that that's

 4   right some of the time, but I think that as a general

 5   manner, I don't believe in population thresholds.  I think

 6   that there is good reason some of the time to believe in

 7   individual thresholds, because the general mechanism that

 8   people have for these kind of traditional toxic effects is

 9   the overwhelming of homeostatic systems, and that it follows

10   that every individual might have a different capacity for

11   absorbing and counteracting a small perturbation of

12   homeostatically controlled processes so that out of that

13   comes the idea that individuals have different amounts of

14   reserve capacity, and different people therefore might get

15   the effect or not when their individual reserve capacity is

16   depleted for counteracting a particular kind of perturbation

17   of one physiological parameter or another.

18             But if we have a broad population distribution of

19   reserve capacities including some people who are very

20   marginal or submarginal for a particular things, like my

21   87-year-old father had an episode of congestive heart

22   failure about a week ago, his heart wasn't pumping enough

23   blood to keep his lungs clear of fluid.

24             If you exacerbate that kind of problem in his

25   cardiovascular system a bit, he's going to be marginally



 1   worse off than he was in the absence of that perturbation.

 2             So with a diverse population with different

 3   amounts of reserve capacity, and in some sense subminimal

 4   for the excellent function, I think that one should believe

 5   that there are effects.

 6             And by trying to specify what these distributions

 7   are in much more empirical ways, we have a chance of

 8   approaching what the real truth is likely to be.

 9             The second is the idea is to reconcile what we do

10   in noncancer assessment with to some extent with cancer

11   assessment, rather opposite to the way that EPA is now

12   wanting to do it.  EPA wants to merge noncancerous -- some

13   parts of cancer risk assessment into noncancer ones.  I

14   think that we ought to make the noncancer ones more

15   quantitative, like the cancer potency assessment.

16             Provide a basis to quantitively assess risk for

17   input to policy decisions.  Lorenz mentioned this already.

18   Partly for those times where you want to juxtapose the cost

19   and benefits of policies to control exposures, and, second,

20   to facilitate judgments of the equity and fairness of the

21   burden of health risks and benefits potentially imposed on

22   different subgroups.  You can only do that if you talk in

23   some more quantitative terms about, well, how large do you

24   think that burden is likely to be, how often.

25             Finally, this would allow some comparable analyses



 1   of uncertainties among exposure and toxic potency,

 2   potentially leading to some value of information analyses

 3   that allows you to have a chance of directing your resources

 4   to the places where you have in fact larger uncertainties,

 5   and those that might change your decision making.

 6             There are serious disadvantages and cost to trying

 7   to go this direction.

 8             Among those are there some need for the experts to

 9   assess and publicly defend past choices of acceptable

10   intakes and risks, and I can just see that you spent 20

11   years developing a series of regulatory actions based upon

12   good faith analyses of data, and now we're going to revisit

13   that whole 20 years' worth of work.  That's a cost.

14             Second, there's some difficulty of social

15   acceptance of finite risks.  There's a need for explicit

16   decision making on uncomfortable tradeoffs that is not posed

17   by the NOEL safety factor procedure in the same way, that

18   the NOEL safety factor pretends at least that you have no

19   risk, and I just think it's not -- at least sometimes not

20   likely to be right, but nevertheless it's a cost to face

21   those tradeoffs squarely.

22             Third, of course, numerical expressions will lead

23   some to imagine that the estimates of risk are more precise

24   than they are.  These estimates of risk that we can now

25   produce, or are likely to be able to produce in the next



 1   decade or so, are going to be highly uncertain.  We need to

 2   be able to represent that uncertainty in a fair way to our

 3   audience.

 4             Fourth, there is going to be significant

 5   controversy over some technical details that are important

 6   technical details.  What sorts of distributions of human

 7   susceptibility should we assume based upon our limited data?

 8             Okay.  We will not have specific measurements in

 9   as many as ten to the fifth people.  We're going to have to

10   use data on some more limited sample of those to be able to

11   project to a larger universe.  We need to have rules about

12   how to do that.  There's no way to avoid that.

13             The elements of this proposal are tentatively, I

14   would suggest, that the RfD for cases where you want an RfD,

15   that is a dose that you're going to tentatively accept

16   without thinking about benefits and other things, as the

17   more restrictive value of either the daily dose rate that's

18   expected with 95 percent confidence to produce less than one

19   in 100,000 excess risk incidence over background, okay.  So

20   in increments of background of a minimally adverse response

21   in a standard general population.

22             So the A relates to a standard general population,

23   ten to the minus fifth risk, which resembles what's been

24   thought of for California as the standard for notification

25   for a much more serious outcome that is cancer, so this



 1   would be rather more protective in the sense that we're

 2   talking about not cancer, about a minimally adverse

 3   response.  Okay.

 4             B, to take into account the possibility that there

 5   might be some recognizable minority of people with a lot

 6   more sensitivity.  In those cases, basically, we want to say

 7   there might be a daily dose that's expected, again with 95

 8   percent confidence, to produce less than a one in 1,000

 9   excess risk incidence, excess incidence over background of a

10   minimally adverse response in this definable sensitive

11   subpopulation.

12             So we don't want to, even if there's a one in

13   10,000 group of people out there that we know about to be

14   highly sensitive, we don't want to cause them to have,

15   without thinking about it very seriously, to have a

16   particularly large burden.  And basically you can play

17   around with these numbers as you like to reflect the social

18   consensus, but I think this is a starting point for

19   calculations, okay, to be made that I think is not grossly

20   inconsistent with what you might think of as policies, as

21   the precedence from the cancer side.

22             Now, how would we actually do this?  Okay.

23             First, let's say we're starting with animal data,

24   which is often the case.  Start with people data, that

25   avoids some of the problems, but not all of them.



 1             First, we would --

 2             DR. GLANTZ:  I just had one quick question.

 3             DR. HATTIS:  Yeah, sure.

 4             DR. GLANTZ:  Why did you pick 100,000 in a

 5   thousand?

 6             DR. HATTIS:  I picked 100,000 because of the

 7   analogy with Prop 65 standards.  That's the one number that

 8   I'm aware of in California law, and it seemed like it wasn't

 9   as extreme as one in a million, and basically that's --

10   well, that's the answer.

11             And why one in a thousand, well, I didn't want to

12   get too high, because if you get started up to one percent,

13   then you start to look at, say, okay, you start looking in

14   the faces, 1,000 is still not too many -- well, it's my

15   guess, and I would defer to our elected representatives the

16   final policy choice of what the number ought to be.

17             But that was my guess as to what would be a

18   sensible recognition of the possibility of minority

19   subpopulations.

20             DR. BLANC:  But by simple long division, if I

21   understand correctly, even if the standard did protect one

22   in 100,000, and if the people in the sensitive population

23   were a hundred times more responsive than the people in the

24   general population, then the same standard would still

25   protect them at a one in a thousand.



 1             DR. HATTIS:  If they were only one in a thousand

 2   of the population, if the sensitive population were no more

 3   common than one in a thousand in the population.  That's why

 4   I said that if that second requirement under this tentative

 5   scheme would only kick in if you have a substantially rarer

 6   group that was -- see, if you have a group that's as common

 7   as one in hundred --

 8             DR. BLANC:  No, they could still be as common as

 9   one in a hundred, couldn't they, if -- and be a hundred

10   times more sensitivity and wouldn't you still be protecting

11   them --

12             DR. HATTIS:  No.  If it were as common as one in a

13   hundred, which is ten to the minus two, and they were a

14   hundred times more sensitive, then essentially that would

15   put us in general population risk of one to ten to the

16   fourth, which would break the general population standard.

17             So they have to be both rare and extraordinarily

18   more sensitive in order for that --

19             DR. BLANC:  And you could balance them -- you

20   could make your own calculations if they were a thousand

21   times more responsive, but they were only --

22             DR. HATTIS:  Sure.  Right.

23             So most of the time the general population

24   standard is going to govern, but this second case was put in

25   there just so that in case we have something, someone that's



 1   really very -- quite a bit more sensitive and very rare,

 2   that we could take that into account.

 3             CHAIRMAN FROINES:  The ten to the minus fifth

 4   value standard was first defined by Governor Deukmejian in

 5   the mid '80s, and so that was essentially a policy decision.

 6   There's nothing -- it's not cast in stone.

 7             DR. HATTIS:  Moreover, for technical purposes it

 8   would be easier if the standard were higher, but I don't

 9   think that we want to make that -- we don't want to choose

10   technical convenience over public health policy, even though

11   the ten to the minus fifth makes things inconvenient for us.

12             Let's start with some putative animal data on a

13   particular kind of response.  One of the things, obviously,

14   that gets done under both the existing procedure and my

15   suggestion is that you need to define some candidates

16   anyhow, for what the critical effect might be and the

17   critical studies that show that effect.

18             And I would depart from the standard procedure at

19   this stage a little bit in that I would want to keep in the

20   analysis multiple candidates if they have a reasonable

21   chance of being significant contributors to an overall risk

22   or effect.  So it's conceivable that you could have

23   multiple effects that would sum up to an overall risk that

24   you would care about.

25             Second, for each effect I want to choose a dose



 1   response family for projection, and there are two broad

 2   categories of these families.

 3             One is the one we've been talking about

 4   implicitly.  The homeostatic system overwhelming idea that

 5   leads to individual thresholds for response, but where the

 6   responses of a mixed population give you a population dose

 7   response that tells you something about the distribution of

 8   thresholds in that population.  Okay.  That's one category.

 9             But there is maybe another category of mechanism

10   that you could use where, in fact, what you're doing is

11   perturbing some parameter in animals that has some analogy

12   with some human parameter for which you have good human data

13   on its individual significance or its group significance for

14   ultimate risk.

15             Examples of this are, for example, sperm counts as

16   a mediator of male fertility reduction or other sperm

17   quality parameters.

18             We have all kinds of data in animals showing

19   testicular atrophy and the point at which the dose response

20   in the animals for reductions of sperm determine.

21             I wouldn't then want to go and assume that the

22   fertility response in people mirrors the fertility response

23   in animals, because the physiology and the reproductive

24   physiology are very different.  And moreover we have good

25   human data, at least we have some human data, on how often



 1   people with different amounts of sperm counts with standard

 2   sets of partners succeed in achieving conception per month,

 3   and then basically we can use those data.

 4             So once we have the sperm count effect modeled in

 5   the animals, we would then want to go to the people to

 6   interpret the significance of that.

 7             There's several other examples of that as a

 8   possibility of fetal growth inhibition as a predictor of

 9   birth weight reduction and strong associations between birth

10   weights and infant mortality.

11             For example, this upper curve here is a curve that

12   shows the relationship between birth weights and infant

13   mortality.  That's the probability that the baby dies in the

14   first year of life.  It's a very strong and more or less

15   continuous relationship between these parameters extending

16   over orders of magnitude.  Okay.

17             So notice that nearly all the infants that are --

18   that are 500 grams or less die, this is in 1980, actually

19   some of them survive now, but nevertheless there is no sharp

20   defining distinction between those who meet the conventional

21   definition low birth weight and those that are slightly

22   above that.  It's over a broad continuous range, it's a

23   little bit worse to be born a little earlier and a little

24   lighter.  Okay.

25             Probably not because babies directly die of being



 1   too light, but because their birth weight reflects some --

 2   is a proxy for some developmental state, some developmental

 3   steps that they have taken to protect themselves against

 4   either infections early in infancy or forgetting to breathe

 5   or something of that sort.

 6             But nevertheless we have responses in animals that

 7   are fetal growth inhibition.  We have fetal growth

 8   inhibition in humans based upon data on very small amounts

 9   of cigarette smoking and things of that sort that suggests

10   that this is a very sensitive parameter.

11             So I would say that for things that operate this

12   way with this risk factor mode, we should be going to those

13   kinds of modes, rather than the sensitivity, partly because

14   we have the advantage of taking into account human data on

15   the relationships between these intermediate parameters and

16   difficult to directly -- things that are difficult to

17   directly measure that we care about.

18             So anyhow other examples of this are destruction

19   of neurons or oocytes that don't regenerate, changes in

20   cardiovascular risk factors, blood pressure, serum

21   cholesterol, that sort of stuff, where we have human data on

22   prospectively collected on the relationships between these

23   intermediates and risk, and we don't have to be very cheery

24   about that.

25             There is another slide.  Right.



 1             Third step is to select a specific analytical

 2   approach for the animal dose response data, now going back

 3   to the threshold branch of the tree.

 4             We want to have some -- we want to select some

 5   point of departure for projection, and because of the

 6   phenomenon that Lorenz illustrated in some of his slides, I

 7   don't trust the human variability to be the same as the

 8   animal variability.  My guess is that as a general matter

 9   the animals are more uniform than the human, the exposed

10   human population is likely to be, because the animals tend

11   to be relatively genetically uniform, they're treated

12   usually at an early adult age, they are treated under

13   conditions where you exclude lots of infections and moreover

14   if you look at the, at least the animal lethality data that

15   have been the basis of Sandy Baird's distributions and other

16   things, they are lot more narrow than the distributions of

17   sensitivity that I infer from the human data.

18             Now, my data are not for lethal effects usually,

19   so there is a possible difference in the comparison that I

20   can make between distributions of sensitivity for very

21   severe effects, which are typically measured in the animal

22   systems, and distributions of sensitivity for such things as

23   methyl choline sensitivity, the amount of methyl choline

24   that it takes to cause you to be -- to decrease in your

25   lungs the amount of air that you can exhale in one second by



 1   20 percent or things of that sort.

 2             Anyhow, I've got a lot of measurements in people

 3   of not very severe effects to moderately severe effects, and

 4   I don't have much in the way of human LD 50, although

 5   there's just a little bit.  There's lethality information.

 6             And I'm -- but in any event I don't think that

 7   it's a general matter we should be assuming that the animal

 8   distribution of sensitivities is similar to the human.  I

 9   would rather go and use a generic assumption based on my

10   human experience and use the animal data only to get an idea

11   of what the chemical can do and what the -- in general the

12   potency of the chemical in the animal systems.

13             So therefore I think it's just as well to go from

14   something like an ED 50 for my point of departure for the

15   projection of the human response.

16             For biomarker mediated effects, the idea is to

17   model the biomarker response in animals as a continuous

18   function, and then translate to people before the

19   interpretation of the ultimate effect parameter in quantal

20   form.

21             Fourth, then I need to define my needed

22   adjustments, as Lorenz has just done.  These adjustments

23   are -- it often happens that the study that was conducted in

24   the animals is not exactly the right length compared to the

25   human exposure, that's the acute chronic projection usually,



 1   and the dynamics of elimination repair in people.

 2             There's incompleteness of the database, not all of

 3   the studies have been done, and that was the part of

 4   Lorenz's presentation.

 5             There is an animal-to-human factor where one can

 6   make a general expectation that on average maybe for chronic

 7   effects you have this metabolic scaling factor of body

 8   weight to the three-quarters, but there's clearly

 9   variability among that, and any chemical that I take is a

10   random draw from that distribution about that factor and we

11   need to say okay to that.

12             And then finally there's severity effect and human

13   variability where the human variability gets done on the

14   basis of how many standard deviations do I need to get from

15   the ED 50 and projected ED 50 in people down to this ten to

16   the minus fifth type of incidence.

17             And anyhow, basically we combine these basically,

18   it's different adjustment factors with a Monte Carlo

19   simulation which is basically a fancy way just to draw

20   random samples from each of these things with some caveats,

21   right.

22             We want to worry about whether there are

23   dependencies among these, the different factors, but

24   probably there ought not to be in general.

25             And then calculate how many of our -- how often do



 1   we expect, based upon the empirical distributions of each of

 2   these factors done for the kind of chemical, hopefully, and

 3   kind of response that we're studying, do we expect it to

 4   be -- do we expect our adjustments to achieve our goals.

 5             So that basically the 95th percentile confidence

 6   that I referred to is based on this kind of Monte Carlo

 7   simulation, hoping that we can develop empirical

 8   distributions or some other distributions that reflect our

 9   real uncertainty in each of these areas that we need to

10   cover from a particular set of data.

11             Now, what are the requirements for anybody to take

12   this kind of thing seriously?  To be a viable replacement

13   for the current RfD I think a numerical definition needs to

14   be a plausible representation of risk management values.  It

15   needs to be estimatable with no greater amount of chemical

16   specific information than is traditionally collected or

17   could be easily augmented with existing things.

18             It needs to be subjected to a series of

19   comparisons with existing RfDs, not because the existing

20   RfDs are a gold standard, but because the first question

21   that any risk manager is going to ask you is, well, how is

22   this going to change my overall standards as a general

23   matter, and we have to have, in order to even be considered,

24   we have to have that comparison made with enough cases where

25   we are pretty sure we know the general tenor of the changes



 1   that are being suggested.

 2             And, finally, it needs -- and this is actually the

 3   part of the benefit of the alternative, is it needs to be

 4   able to accommodate more advanced types of technical

 5   information.

 6             One of the difficulties with the current system is

 7   that there's no way to use variability information in a very

 8   orderly way anyhow, and therefore this hasn't been a very

 9   great incentive to collect it.

10             This kind of alternative procedure should allow

11   you to plug in your empirical factors in place of a default

12   set of distributional factors based upon empirical

13   distributions from other chemicals, and therefore have a

14   predictable way to use the new kind of information on

15   comparative pharmacokinetics, et cetera, in the thing.

16             I mentioned where there is a plausible

17   representation of risk management values issue.

18             I already mentioned that part of the way I got to

19   the starting point here was an analogy with the standard for

20   notification for cancer hazard under California's

21   Proposition 65.

22             The minimally adverse severity level specified

23   makes this somewhat more protective, given the similar

24   incidence level specified for a less serious outcome, that

25   is if we do the calculations correctly.



 1             Estimatable with no greater chemical specific

 2   information than traditionally collected.

 3             Obviously, we need some generic assumptions on the

 4   distributional forms for variability and uncertainty.  I've

 5   built, spent a fair amount of time building some of the

 6   basis for this in the paper I presented in Bologna last

 7   September that I think you have up here.  I would, as I

 8   said, be happy to inflict some of those data on you, but I'm

 9   paying attention to your injunction, John, to not present

10   large tables of logarithms until provoked.

11             The basic calculations, however, assumed some log

12   normal variability among people.  I think there's some

13   reason that we might want to relax that in some cases,

14   because at least there's some evidence in the data that

15   there might be some amount of bimodality in the log normal

16   distribution some of the time.

17             I can't resist --

18             DR. WITSCHI:  Excuse my ignorance, but log normal

19   distribution has a tail only on one side, right?

20             DR. HATTIS:  Yeah.  At least much more extended

21   tail on one side than the other.

22             DR. BLANC:  Which side?

23             DR. HATTIS:  The side that's the bad side.

24             DR. BLANC:  The sensitive side?

25             DR. HATTIS:  Yeah.



 1             DR. BLANC:  It's on this side?

 2             DR. WITSCHI:  Wouldn't it be the other side, the

 3   resistance side?

 4             DR. HATTIS:  In principle it's possible, but I'm

 5   viewing it the other way, but I'm trying to figure out why I

 6   think it's -- basically, if you look at in log space it's

 7   completely symmetrical.  And I'm viewing it as --

 8             DR. FUCALORO:  You mean more symmetrical?

 9             DR. HATTIS:  Well, in theory it's completely

10   symmetrical, but at least on average, yeah.

11             What I have done is to arrange my data in always

12   in the same orientation with the worst values of the

13   parameter, the values of the parameters that would lead to

14   greater risk on one side of the thing than the other.

15             And when I do that, at least for the, for example

16   for the pharmacokinetics parameter, I would put low volume

17   of distribution on the worst side and low clearance rates on

18   the worst side, and bigger area under the curve of a

19   concentration times time product on the worst side.

20             And so in that way I get -- I do all the log --

21   the analyses in log space, so that I only translate them

22   back into linear space.

23             DR. WITSCHI:  So you're a pessimist.

24             DR. HATTIS:  You always get -- the problem is

25   that -- yeah.  Right.  Anyhow, you basically -- right.



 1             I don't have -- I have to think about that

 2   question, because I'm obviously dancing around it, a proper

 3   response to your question.  Basically if you look at it in

 4   log space there's no problem about which side, as long as

 5   you keep the orientation respect to sensitivity consistent

 6   on the database.

 7             Anyhow, one can do technical sensitivity

 8   calculations and maybe some uncertainty calculations

 9   assuming mixtures of two or more log normal distributions

10   and figure out what the impact of a risk analysis would be

11   of that.

12             The most easily quantified uncertainty is

13   reflected in the observed variability among chemicals of log

14   normal variability estimates for overall sensitivity in my

15   case or for in Lorenz's case the other adjustment factors.

16             There's also, I think, on the scientific side some

17   potential for learning the mechanistic related groupings

18   predicted with different amounts or forms of variability for

19   either chemicals that are processed in different ways or

20   chemicals that have different kinds of actions in the body.

21             I'm finding for example that my chronic systemic

22   neurotoxicants have rather less variability than such things

23   as acute responses to locally acting things.

24             I'm finding more variability in that case than in

25   the chronic systemic neurotoxicants.



 1             I'm worried about whether I'm also comparing

 2   different severities of effect.  It's possible that I have

 3   more -- less variability for very severe effects and more

 4   variability for less severe types of effects.

 5             But all of that is subjected -- is subject to

 6   empirical study, essentially, and so what I'm suggesting

 7   that one of the benefits of this general proposal is that it

 8   opens up the regulatory system to a sense to influence to

 9   being informed by empirical information.

10             And with that, I think I'll -- I will talk, if you

11   like, a little bit more about how I come up with my

12   estimates of variability and what they are, but I think at

13   this point I think I can open it up to some questions.

14             CHAIRMAN FROINES:  Do you have any examples of --

15             DR. HATTIS:  Yes, I do.

16             CHAIRMAN FROINES:  -- outcomes that you actually

17   have done?

18             DR. HATTIS:  Right.

19             CHAIRMAN FROINES:  Why don't you give one just as

20   an --

21             DR. HATTIS:  All right.  Yeah.  Now, this is a way

22   of plotting things essentially that turns a log normal

23   distribution into a straight line.  Essentially what I'm

24   doing is plotting in this, some of the time I'm plotting the

25   Z score.  I didn't invent this method.  This is called



 1   probit analysis.  This was invented in the '30s, right.

 2   This is not something that I came up with myself.

 3             This is essentially the Z score of the percent

 4   response essentially and the Z score essentially is the

 5   number of standard deviations that you have to go from the

 6   mean of a distribution to have the area under the curve of

 7   the distribution correspond to the percent of the people who

 8   respond at a given level.

 9             So essentially if ten percent of the -- if five

10   percent of the people respond with -- to a given

11   concentration of chromium with skin sensitivity, then we say

12   that five percent of the people have thresholds below that

13   concentration.

14             And if 50 percent of the people responded some

15   higher concentration, we say that that corresponds to a Z

16   score of zero.

17             The five percent level corresponds to a Z score of

18   minus 1.64, which is essentially in a normal distribution.

19             If you go 1.64 standard deviations below the mean,

20   then five percent of the people will be even farther from

21   the mean than that, below that.

22             Anyhow, so if I plot this kind of Z score

23   transformation versus the log of the chromium concentration

24   that it gave responses in this group of tested people,

25   basically this is 102 tested, people were tested.  54 of the



 1   them responded to the highest dose level, just a little bit

 2   over a Z score of zero, okay, so just a little bit more than

 3   the Z.

 4             What you see is that the points can be arranged in

 5   a way that is reasonably close to a straight line, which

 6   means that these data are reasonably -- suggest a reasonably

 7   log normal distribution of the individual thresholds of

 8   these people for this kind of response.

 9             The same -- and basically the slope of that line

10   relates to the amount of individual variability that you

11   have.

12             In this case, the slope is rather shallow and

13   corresponds to a lot of variability.  Basically for those of

14   you who think in log normal statistics, this is a geometric

15   standard deviation of ten.  Okay.

16             What that means is that 95 percent of the

17   population would have the threshold spread out over four

18   orders of magnitude, essentially from a 100-fold less than

19   the median, minus two standard deviations, to a 100-fold

20   more than the median.

21             So this is an example of a response, and this is

22   based on individual clinical measurements.  So I'm not

23   talking about anything that's a little faky here.  This is

24   talking about clinical measurement that is quite highly

25   variable.



 1             What basically most of the things are not as

 2   variable as that, but there are quite a few things that are

 3   almost as variable.  Among the things that are almost as

 4   variable -- let me get a couple of them.  This I'm choosing

 5   partly because it's a large study, and I'm told that one of

 6   you is a lung physician.

 7             This distribution, this is a distribution of --

 8   this is a plot in the opposite direction.  This orientation,

 9   the shallow slope, means less variability.  But don't get

10   too confused about that.

11             This is the log normal plot of the concentrations

12   that cause, I believe, a 20 percent increase -- 20 percent

13   decrease in FEV 1, the amount of air that you can breathe

14   out in one second, down in a very large epidemiological

15   study over 5,000 smokers with mild to moderate air flow

16   obstruction.

17             And what you see again is a pretty decent

18   correspondence of the points to the theoretical predictions

19   of this straight line, which represents the log normal

20   distribution.

21             You see a pretty good amount of variability.  This

22   is -- this is the log -- the geometric standard deviation

23   would be 10 to the .6, which is about, doing it quickly in

24   my head, six or seven or so.  And that's, you know, that's

25   enough variability so that if you wanted to get from a five



 1   percent effect level to a ten to the minus fifth effect

 2   level, you need about 3.3 standard deviations to do that.

 3             3.3 standard deviations is three times .6, meaning

 4   1.8 orders of magnitude.  So you need something like, oh,

 5   50-fold to do the job that you would expect the 10-fold

 6   safety factor to do if what you want the 10-fold safety

 7   factor to do is to get you from a five percent effect level

 8   consistent with a no observed effect, to ten to the minus

 9   fifth effect.

10             So that's essentially the implication of that kind

11   of variability.

12             DR. BLANC:  Say that statement one more time.

13             DR. HATTIS:  Right.

14             DR. BLANC:  This is not the key statement you're

15   trying to make through all of this?

16             DR. HATTIS:  Yeah.  What I'm saying is that the

17   thrust of Lorenz's and Sandy Baird's stuff is that several

18   of the uncertainty, the tradition uncertainty factors do

19   appear to be conservative, that is that they build in some

20   more protective factor than would be needed, at least for

21   the average chemical, which is what was the intent after

22   all.  But this one may not be as a general matter.

23             DR. BLANC:  Because the range that you would have

24   to go, to go from five percent, which would be the no effect

25   level in humans to --



 1             DR. HATTIS:  To a ten to the minus fifth.

 2             DR. BLANC:  To ten to the minus fifth was how many

 3   logs?

 4             DR. HATTIS:  That often you need --

 5             DR. BLANC:  In that particular case.

 6             DR. HATTIS:  In that particular case you might

 7   need about 50-fold, rather than 10-fold.  You might need

 8   another factor of five or so, if the population is really

 9   log normal.  Okay.

10             DR. BLANC:  Wouldn't we be going a 100-fold,

11   because wouldn't we be going from the lowest to the values

12   in the no effect level, which was assuming -- okay --

13             DR. HATTIS:  Okay.  Some of the time you have --

14   you don't have perfect log normal distributions.  Some of

15   the time you have -- now, this is a classic case of a

16   genetically determined variability.  This is a distribution

17   of -- this is from the original paper, propyl thiouracil

18   taste sensitivities.  Now this is not an adverse effect,

19   okay.  So I must say that.  But this is a different kind of

20   response.  Each of these blocks from minus three to minus

21   two, four of those represent one order of concentration of

22   propyl thiouracil.  But what you do see is that there is now

23   we look like there's actually two distinct subpopulations.

24   But nevertheless there's still variability in each of the

25   subgroups.



 1             DR. GLANTZ:  What are the open bars?

 2             DR. HATTIS:  The open bars -- the open bars are

 3   the total population.  The shaded bars are the things that

 4   he characterized as tasters and nontasters.

 5             And I -- one of my pet peeves is how biologists

 6   have a tendency to dichotomize.  I mean, I do believe that

 7   there's two humps to this distribution, but by saying that

 8   these are the tasters and these are the nontasters, you're

 9   giving the wrong impression.  These are the more sensitive

10   and the less sensitive people.  Okay.  It's not that --

11             DR. FUCALORO:  Biology has a tradition of

12   taxonomy.

13             DR. HATTIS:  I like taxonomy.  I think taxonomy is

14   good.  I think the overuse of dichotomization is a disease.

15   I mean, there are people who talk of the normotensives and

16   the hypertensives.

17             DR. FUCALORO:  It's either a bird or a dinosaur.

18             DR. HATTIS:  That's right.  And you either have

19   chicken pox or you don't.

20             But it's not true that you either have -- right?

21   But so there are some things where you should dichotomize, I

22   think, but you don't dichotomize everything.

23             And you do have to make decisions about who you

24   test for -- who you treat for blood pressure, but it isn't

25   true that there is -- blood pressures are a broad continuous



 1   distribution, and to draw an arbitrary line and say these

 2   people are qualitatively different than those people is just

 3   without basis.  It's only based upon my habits and

 4   prejudices.  Anyhow.

 5             DR. BYUS:  You should have been here for lead.

 6   You would have loved that discussion.

 7             Don't write that down.

 8             DR. HATTIS:  This isn't the first time that I've

 9   encountered this problem, actually, which is why I'm

10   venturing to be pejorative about a whole class of people

11   that includes myself.

12             Anyhow, nevertheless, what I'm trying to say as a

13   general matter is that sometimes we have to worry about

14   bimodality of the distribution.

15             But it's often the case that we have variability,

16   particularly for these less severe responses that clearly

17   spans a great deal more in the order of magnitude.

18             With that, I'll turn it back, or if you want I

19   will again go into a little bit more detail about how I'm

20   trying to figure out, reconstruct distributions of overall

21   variability.  Basically I've got a database of something

22   like, it's over 200 observations now, and each observation

23   is a data set or a number of data sets where I'm looking at

24   variability in human populations.

25             The human populations are not always ideal for



 1   projections in the way we're trying to do it.  Sometimes

 2   they're restricted to normal healthy adults.  In fact,

 3   rather often they're restricted to normal healthy adults,

 4   often young, normal, healthy young adults of one gender, and

 5   there are other difficulties, but nevertheless the idea is

 6   by putting together variability observed for a number of

 7   portions of the pathway between external exposure and

 8   biological response, classifying the biological responses

 9   one way or another, I'm trying to develop the empirical

10   distributions that one would need to say, okay, for this

11   kind of chemical, for this kind of response, what should we

12   expect.

13             But absent the subcategorization we can use the

14   existing distribution for some beginning calculations.

15             CHAIRMAN FROINES:  George or people in the

16   audience have questions to ask now?

17             DR. ALEXEEFF:  George Alexeeff with OEHHA.

18             I was thinking back to the earlier part of your

19   presentation and the issue about different endpoints.  And

20   one of the things that we struggled with with our acute

21   document was different endpoints with different severities.

22             So, for example, reproductive toxicity that might

23   occur at a slightly higher dose than respiratory irritation.

24             So theoretically we have two different sensitive

25   subpopulations.  One is for the respiratory irritation.



 1   Let's say we can have asthmatics, but then for the

 2   reproductive effects or depending on how you define pregnant

 3   women, or potentially pregnant woman as a sensitive

 4   subpopulation.

 5             So I'm just wondering how one can look at your --

 6   the calculation.  I was thinking of the one in a thousand

 7   versus the one in 100,000, and if one then has different --

 8   you're talking about carrying forward the critical effects

 9   of multiple critical effects, if your thought was to then be

10   able to calculate differential impact to different

11   subpopulations.

12             I don't know if I explained it.  But to me that

13   seemed to be maybe a power of carrying forth those critical

14   effects.

15             DR. HATTIS:  I think it's simpler to keep them

16   separate and do the complete calculation in parallel terms,

17   because you don't know at the early step, you could have a

18   difference of severity which would require you to have an

19   adjustment factor that you needed to determine later for

20   that to adjust back to a minimal adverse severity level, if

21   that's the way you're doing it.

22             Or, you know, at some point you might even say if

23   we can't adjust the severity, say we've got a quantal type

24   phenomenon, either you've got a cleft lip or you don't, and

25   there is no such thing -- or at least I guess both -- then



 1   you might say we don't want to tolerate as much as one in

 2   100,000 of that.  You might want to have a stronger -- we

 3   want to be more confident that as a general matter that you

 4   don't have that effect or you want to have an equal level of

 5   confidence for a smaller incidence than --

 6             DR. ALEXEEFF:  Well, I was thinking less in terms

 7   of adjusting for severity and more simply that calculation.

 8   I don't know what the percentages of how one would define

 9   the sensitive subpopulation for reproductive effects.  If

10   you're including all women of reproductive age or only those

11   women that are currently pregnant at the time.

12             And the same thing on the asthmatic side.

13             You would have different proportions instead of --

14   I mean, it could go anywhere from one in two, or one in

15   three, to maybe a much -- if you assume women of

16   reproductive age might be one in four of all people.  I

17   don't know what the amount is, but some very high number.

18             And if that could end up impacting or actually

19   causing a lower level than you would for the respiratory

20   effects, even though it's more mild, because you have a

21   larger population that could be impacted.

22             That's actually I was wondering if you had done

23   those kind of calculations where you had differential -- I

24   don't know if I'm making sense -- but differential impacts,

25   differential sizes of the sensitive subpopulations that



 1   could then end up causing the actual final effect level to

 2   overlap.

 3             DR. HATTIS:  I think that that's an interesting --

 4   the short answer to your question is that I haven't tried to

 5   do that.  I think that that's where we need to have some

 6   serious dialogue with the risk managers and the affected

 7   communities as to exactly what the rules are for defining

 8   sensitive subpopulations of that kind.

 9             And I don't have a quick and easy answer to

10   suggest that, after all, I'm just a poor technical person, I

11   don't do risk management.

12             I do think that it's -- I think that one of the

13   key things that a policy person would say is that if you can

14   narrowly define a population, and it's easy for them to do

15   ex ante, if it's easy for the person to say I'm pregnant,

16   I'm an albino, and therefore I know that I'm much more

17   sensitive than you typified in your calculation.  It might

18   be reasonable to deal with them.

19             I mean, but I don't know exactly how far you take

20   that.  I don't know how fair that is.  I mean, you have

21   people who are very highly sensitive to iron in the diet and

22   that came up as an issue in the allowing of iron

23   supplementation of bread.

24             So I don't know exactly how to do that.

25             CHAIRMAN FROINES:  Thanks, Dale.



 1             Lorenz.

 2             DR. RHOMBERG:  Lorenz Rhomberg.

 3             Dale, I was wondering have you ever given any

 4   thought to the notion that you might want to adjust the ten

 5   to the minus fifth or ten to the minus whatever level that

 6   you are trying to protect, based on the size of the

 7   population exposed, so that ten to the minus fifth is very

 8   stringent if you're only exposing four or five people.  The

 9   chance that one of those might be among them is very small,

10   but if you're exposing the whole population of the US, then

11   it's not very stringent at all.

12             You can imagine something that says I'm going to

13   protect sort of with some assurance the most sensitive or

14   next to most sensitive in the group or something like that,

15   and it could have it slide depending on how many were

16   exposed.

17             DR. HATTIS:  These calculations are all in a

18   context of an equity framework.  Is it unfair to impose this

19   as a general summary thing before we talk about tradeoffs,

20   essentially.

21             I think that there's another whole set of

22   calculations of where one is juxtaposing cost and benefit

23   where the size of the population exposed in affected matters

24   centrally.  And so that I think for purposes of the equity

25   discussion, I think maybe it's most significant to talk



 1   about these individual risk things, but for purposes of

 2   saying how do we best direct our resources, how do we set

 3   priorities for social intervention, there's certainly room

 4   for the size of the population to enter into that whole

 5   other set of discussions about juxtaposing cost and benefit,

 6   which I think is its own thing.

 7             I'm not absolutely sure whether that I want to

 8   automatically allow as an equity matter.

 9             Certainly at the point where you get to very large

10   risks, as much as one in ten, you start to get into the

11   realm of criminal law.  So you can't, regardless of the size

12   of the population, you can't go as high as that.

13             And I think it certainly matters whether you're

14   exposing one in a million people or ten people to a ten to

15   the minus six or ten to the minus fifth risk.  I think

16   that's more a discussion I guess of the social priority that

17   should be allocated to those.

18             DR. FUCALORO:  Weren't your comments based upon a

19   population large enough that you reached a sampling is equal

20   to probability, essentially?  That the population is so

21   large that when you say one in 100,000, at least a million

22   people or more to -- you're certainly not talking about a --

23             DR. HATTIS:  That was my general idea, but still

24   I'm wondering, you know, if you have a population of a

25   hundred houses in a neighborhood of a facility of some sort,



 1   would you want to do a different calculation, would it be

 2   equitable to expose them to a greater amount of risk just

 3   because they're relatively fewer of them.

 4             I think in an equity terms maybe not.

 5             On cost-benefit terms, then that certainly does

 6   matter.

 7             So I think it matters what's the social policy

 8   framework that is governing the particular choice.

 9             CHAIRMAN FROINES:  Thank you, Dale, and Lorenz.

10   Thank you very much.

11             (Applause.)

12             CHAIRMAN FROINES:  I think we would have to

13   restructure the nature of this panel to deal with the issues

14   that Dale just raised.  We'd have, since we're now mixing

15   risk management and risk assessment --

16             DR. HATTIS:  This is firmly in risk management,

17   but the point is I'm presuming to raise it as a technical

18   person is that I'm convinced that the risk management

19   community will only work -- will only start to think about

20   it if the technical community opens up some choices for

21   them.

22             CHAIRMAN FROINES:  We have three more items on the

23   agenda.  One is the MOE REL approach, one is the findings

24   and one is the update.

25             I think all of those are going to go relatively



 1   quickly.

 2             So what I think we should do, if everybody agrees,

 3   is to take a ten-minute break for the stenographer and then

 4   to go until we finish those three items and then call it a

 5   day, if that's acceptable.  But I think she would like to

 6   have another break.

 7             (Thereupon a short recess was taken.)

 8             CHAIRMAN FROINES:  Melanie and Jay Schreider.

 9   There he is.

10             DR. MARTY:  I think I was tasked with just a very

11   brief overview of what we do for reference exposure levels.

12             The panel has heard all of this in the

13   deliberations on the technical support document for

14   determining acute reference exposure levels.  So I'm going

15   to be very brief.

16             OEHHA has developed a number of acute and chronic

17   reference exposure levels, and you've seen the acute and

18   you're going to see the chronic.

19             We used either the NOAEL divided by uncertainty

20   factor approach, which you just heard a whole lot about, or

21   the benchmark dose divided by uncertainty factor.

22             We have used uncertainty factors in a way that's

23   quite similar to the US EPA approach.

24             Essentially the uncertainty factors used with the

25   benchmark concentration approach vary whether the study



 1   subject is animals or humans or sensitive human population,

 2   and factors have been used for intraspecies variability, as

 3   well as interspecies variability.

 4             The uncertainty factors used with the NOAEL

 5   approach are fairly similar to that used with the benchmark

 6   dose approach, although we have a couple of other issues and

 7   that is we may not have a NOAEL, we may be stuck with LOAEL

 8   and we have to do that extrapolation when there's one of

 9   those uncertainty factors would account for extrapolating

10   from a low observed adverse effect level to a no observed

11   adverse effect level.

12             Also I should throw in for chronic reference

13   exposure levels sometimes you don't have a chronic study,

14   you have a subchronic study, so there's an additional

15   uncertainty factor that would be incorporated into your

16   reference exposure level calculation.

17             CHAIRMAN FROINES:  What is that value?

18             DR. MARTY:  Generally ten for subchronic to

19   chronic.

20             Also for the NOAEL approach it's going to the

21   uncertainty factor you use is going to depend on whether

22   it's an animal study or a human study or a study in

23   sensitive humans.

24             This table is taken out of the acute reference

25   exposure level document, basically gives an idea of the size



 1   of the uncertainty factors used for the different

 2   uncertainties.

 3             Essentially if you start out with an animal study

 4   and you use the benchmark concentration where you have a

 5   regression analysis from -- to get you to the benchmark

 6   concentration for a five percent response rate, we would use

 7   a 10-fold intraspecies species variability factor and a

 8   3-fold interspecies uncertainty factor, for a total of about

 9   30.

10             If you start out with a human study, but it didn't

11   involve sensitive individuals, for intraspecies we have used

12   either three or ten.

13             We use ten in the case of formaldehyde, even

14   though it was a benchmark concentration, because there was

15   good evidence for a wide variability in the response to the

16   irritancy produced by formaldehyde.

17             And then if you have a human study that was done

18   in sensitive individuals, then we have used an interspecies

19   variability factor as low as one.

20             Next slide, Jim.

21             DR. BLANC:  How would you comment on that in terms

22   of the data we heard earlier that for example for certain

23   human sensitivities, maybe 50 would be better and that the

24   range should be from three to 50 instead of from three to

25   ten?



 1             DR. MARTY:  Yes.  That is a possibility.  And we

 2   did see that with the -- Dale showed a slide of people

 3   reacting on a patch test to hex chrome and in my mind that

 4   is the type of response where you would anticipate seeing a

 5   lot of variability because, correct me if I'm wrong, but

 6   it's essentially a hypersensitivity response.

 7             So I think that we have been using ten, and we're

 8   still using ten, and for some endpoints we are probably not

 9   being health protective enough.

10             DR. BLANC:  George.

11             DR. ALEXEEFF:  George Alexeeff with OEHHA.

12             Probably the example that we have is the

13   formaldehyde example, because in that case, although Melanie

14   just quickly went through it, we took a benchmark

15   concentration which we already calculated the response, at

16   least within the study population, the five percent response

17   rate, and then on that we add an additional 10-fold

18   uncertainty factor.  The reason for that is because we're

19   looking at all the other formaldehyde data, there was a very

20   very wide distribution of effects, still some individual

21   effects reported at some lower levels.  So that seemed to be

22   an example that might be consistent with what Dale was

23   saying where it's a very very broad range.

24             DR. MARTY:  I think a lot more research needs to

25   be done on variability by endpoint.



 1             CHAIRMAN FROINES:  But the implication of what

 2   Dale was saying is that to the degree that data does exist,

 3   you could incorporate that data into an estimate of your

 4   uncertainty factor.  You're not -- nobody said it has to be

 5   as small as ten.  Nobody said it has to be ten.

 6             DR. MARTY:  Right.

 7             CHAIRMAN FROINES:  So to the degree that issue has

 8   merit in terms of existing data, and Dale has 300,000 log

 9   probit graphs that he can give you, at least that's what I

10   think he said.  It's something to consider in terms of

11   certain chemicals as they come up.

12             DR. GLANTZ:  Actually, I think Dale has three

13   times ten to the fifth graphs.

14             DR. ALEXEEFF:  Actually, I think that what Lorenz

15   was saying that part of the original ideas of these

16   uncertainty factors was when you went from animal studies

17   you had animal factor and the human factor, and it could be

18   that part of the animal factor was really helping to deal

19   with the human variability, but just overall the factor was

20   a hundred.

21             So I think the situation that we might be

22   concerned with are those where we're basing it actually on

23   human data and the variability within humans might be much

24   broader than ten, because we don't have any other factors

25   we're adding in.  So if there were ways that we can develop



 1   or understand what would those situations be, that would

 2   greatly help us in terms of making sure we're providing

 3   protective levels.

 4             DR. MARTY:  This table is also taken partly from

 5   the truncated table from our acute REL document.

 6             When we use the NOAEL method, we have both

 7   interspecies and intraspecies uncertainty factor, and they

 8   have generally been ten.

 9             If you have a human study, then obviously you

10   don't have an interspecies adjustment that needs to be made,

11   but you may want to make an intraspecies adjustment,

12   particularly if your study does not include sensitive

13   subjects.

14             We also have the case where we may have to

15   extrapolate from a low observed adverse effect level, and we

16   have done an analysis of the subset of chemicals where we

17   looked at the LOAEL to NOAEL ratios and for mild effects the

18   95th percentile was six.  So we ended up using that for mild

19   effects.

20             For other than mild effects, we used a factor of

21   ten.

22             DR. FUCALORO:  Excuse me.  Define mild effect.

23             DR. MARTY:  We have a whole definition in that

24   document.

25             DR. FUCALORO:  I know you do.



 1             But that is -- that's essentially the type of

 2   effect, I mean, as opposed to being just a low dosage, am I

 3   correct?

 4             DR. MARTY:  Right.  The severity of the endpoint.

 5             Anyway, in sum, that's what we've been doing to

 6   develop reference exposure levels.

 7             CHAIRMAN FROINES:  Move ahead.

 8             DR. MARTY:  Jay Schreider is going to talk about

 9   what DPR has been doing.

10             DR. SCHREIDER:  My name is Jay Schreider.  I'm

11   from Department of Pesticide Regulation.

12             And I think what we do is many respects very

13   similar to what OEHHA does.  We tend to use in a lot of

14   instances a margin of exposure approach which is similar to

15   what US EPA does, in least in the Office of Pesticide

16   programs, and that's frequently because we are looking, or

17   at least initially, was the focus was on occupational

18   exposure, and that we have a no observed adverse effect

19   level over the exposure, and that's contrasted with the REL,

20   or in our case what we sometimes use is the reference

21   concentration, where you have the no observed adverse effect

22   level over a specific uncertainty factor.

23             The endpoints of the studies obviously would be

24   the same, that is there's no difference in how we're

25   treating the endpoint or how we're judging the endpoint.



 1             One difference you can see at the top where we --

 2   again the margin of exposure approach or what used to be

 3   called the margin of safety.

 4             With a reference concentration we employ a

 5   breathing rate correction or breathing rate for along with

 6   the uncertainty factor to adjust for breathing rates between

 7   adults and children and then also between the experimental

 8   animals and man, assuming a higher breathing rate for

 9   children and that frequently as a result becomes the driving

10   factor in the risk assessment.

11             DR. GLANTZ:  Can I just ask a question.

12             Would I be correct in saying that the -- what

13   you're calling the reference concentration and what OEHHA is

14   calling the reference exposure level are the same?  Is that

15   true?  No?  Okay.  Except for the breathing rate.

16             DR. SCHREIDER:  Except for the breathing rate they

17   would be the same.

18             DR. GLANTZ:  Is that --

19             DR. SCHREIDER:  I mean, there may be differences

20   on individual basis on the uncertainty factor, but the ideas

21   would be the same.

22             DR. GLANTZ:  Well, let me ask some more precise

23   questions, because assuming that you did the uncertainty

24   factors the same way, and if you didn't take breathing rate

25   into account, then what you're calling the reference



 1   concentration would be the same as what OEHHA is calling the

 2   reference exposure level?

 3             CHAIRMAN FROINES:  No.  Because in the chronic

 4   document OEHHA uses breathing rate adjustments.  So that

 5   depends on whether we're talking about acute versus chronic.

 6             DR. GLANTZ:  Okay.

 7             CHAIRMAN FROINES:  In this, he must be referring

 8   to a chronic.

 9             DR. SCHREIDER:  We would use the breathing rate

10   regardless of whether we're talking about acute, subchronic

11   or chronic.  We apply a lot of our data or no effect levels

12   from acute studies, and we would still use the breathing

13   rate correction.

14             DR. GLANTZ:  Does the breathing rate correction --

15   the breathing rate correction makes the reference

16   concentration bigger or smaller?

17             DR. SCHREIDER:  Lori.

18             DR. LIM:  This is Lori Lim from DPR.

19             I think we need to show another transparency.

20             DR. GLANTZ:  Am I getting ahead of you here?

21             DR. LIM:  It's just a secret.

22             DR. GLANTZ:  The secret transparency.

23             DR. LIM:  Just in case.

24             This shows exactly what I think you're talking

25   about in comparison, whether you adjusted it or not.



 1             DR. SCHREIDER:  The only difference would be going

 2   over the reference exposure at the top, and the reference

 3   concentration at the bottom where we had got the first

 4   adjust after the animal study which was using six-hour

 5   exposure, adjust that for 24 hours, and for the breathing

 6   rate of the animal, and then convert that for the breathing

 7   rate of the child at the bottom, and we would come up with a

 8   reference concentration of 40 micrograms per cubic meter,

 9   versus taking the concentration from the animal study and

10   dividing it by the uncertainty factor would give reference

11   exposure of 122 micrograms per cubic meter.

12             DR. GLANTZ:  So I'm just trying to harmonize here.

13   Anyway, this is in honor of Rick Becker.

14             So if OEHHA was given the same NOEL, okay, would

15   you come up with the same numbers?

16             DR. ALEXEEFF:  Well, there's --

17             DR. GLANTZ:  If not, could somebody explain to me

18   the difference.

19             DR. ALEXEEFF:  There's a couple of different

20   things going on in this slide, and it almost -- we almost

21   have to have a couple more slides just to sort of tease a

22   couple things out.

23             For example the six-hour to 24-hour adjustment

24   there, okay, that is a weighted adjustment.  So if we were

25   making a 24-hour level, we would do the same adjustment.  So



 1   that's right there a factor of four.  Okay.  That seems to

 2   be kind of -- so if you were -- the top one is basing it on

 3   a six-hour reference exposure level and the bottom one is a

 4   24-hour reference exposure level.

 5             Is that right, Lori, just to be sure?

 6             So in one sense that adjustment is kind of

 7   confusing.  The other adjustments, although they're showing

 8   how they make the calculations, which is what the purpose of

 9   this slide is, but to make the -- understand the question

10   you're -- to answer the question you're asking, if we were,

11   like I say, we would always, if we were looking for a

12   24-hour level, then we would make the six over 24 adjustment

13   just like they did.  So what our number would look like was

14   basically 122 divided by four.  So it actually would be like

15   30.  So that would be the actual difference in this case if

16   we were both coming up with 24-hour levels, our number would

17   be 30 something and theirs is 40.

18             CHAIRMAN FROINES:  How do we --

19             DR. GLANTZ:  I'm now totally confused.

20             DR. ALEXEEFF:  There's two adjustments happening

21   in here, which is their standard practice.

22             One is the time adjustment, because they're

23   developing a 24-hour standard.

24             The other one is the breathing rate adjustment.

25             You asked a question about the breathing rate



 1   adjustment.  In order to understand that, how that impacts

 2   it, you have to sort of ignore the time adjustments.

 3             So I was trying to normalize the time adjustment

 4   issue, which we would also do if we -- remember, we had that

 5   concentration time formula?  So it would depend upon the

 6   exact situation what the exact calculation was.

 7             CHAIRMAN FROINES:  But, George, but that's a

 8   specific point.  You're attempting to develop acute RELs for

 9   a one-hour exposure and you're making a determination that a

10   one-hour exposure is precisely an acute exposure.

11             Now, for any time averaging that's done, there

12   should be a reason for it.

13             So the question I think Stan is getting at in

14   terms of the six divided by 24 hours is what's the

15   toxicologic basis for that particular adjustment.

16             Is that what you're saying?

17             DR. ALEXEEFF:  That's the kind of adjustment we

18   would make, let's say, in our cancer documents or our

19   chronic documents do that exact same adjustment.  It's just

20   that regular averaging you would be considered concentration

21   times time to the concentration of the N where N is one.

22             CHAIRMAN FROINES:  I understand.

23             DR. ALEXEEFF:  Just the regular averaging time in

24   that case.

25             So this is assuming sort of a, what do you want to



 1   call it, lineal extrapolation over concentration to the nth

 2   power, where N equals one.  And we would do that kind of

 3   calculation if we were doing adjusting from like an

 4   occupational exposure.

 5             DR. FUCALORO:  So in other words, six hours a day

 6   of exposure, that all you're doing, right?

 7             DR. LIM:  What we are assuming is that person

 8   would be exposed for 24 hours, so we amortize it to 24

 9   hours.

10             DR. FUCALORO:  What's the concentration, because

11   you ultimately have to meet the standard of milligrams -- I

12   got myself a little confused.

13             You really ultimately have to get the standard of

14   micrograms per kilogram per day.

15             DR. LIM:  Maybe I should just take it through

16   slowly.  It's kind of complicated.

17             The whole conversion is analogous to dietary

18   experiment where we have certain amount of food, but the

19   animal is not going to take it all, so you have to put in

20   food consumption rate.

21             So in this case we're saying we're putting in

22   breathing rate factor into consideration.

23             So the animal is exposed to NOEL in terms of air

24   concentration in this amount.

25             The rat was breathing at this amount and for six



 1   hours and you amortize it over 24 hours, the dosage for that

 2   animal for the no effect level in terms of milligram per

 3   kilogram per day is this level.

 4             So when we back calculate it for the child to have

 5   the same dosage, breathing at this rate and putting the

 6   factor of ten, what air concentration should it be that it

 7   cannot be higher than this particular air concentration.

 8             DR. FUCALORO:  So 40 micrograms per meter cubed

 9   and a child breathes six hours a day will put him or her at

10   the 2.9 milligrams per kilogram per day, 24 hours.

11             CHAIRMAN FROINES:  Let's go ahead.

12             DR. SCHREIDER:  Next one.

13             In terms of the uncertainty factors that we use,

14   there again are similar and I'd like to point out though

15   that these are adult uncertainty factors so that when

16   obviously there is more information that either humans are

17   less sensitive or more sensitive than assumed for the

18   10-fold or that the variability may be more or less, that

19   would be factored in, so that again the starting from a no

20   effect level or no observed adverse effect level from an

21   animal study, we would use an intraspecies uncertainty

22   factor of ten to get to humans and then intraspecies factor

23   of 10-fold again for as a default for the human variability.

24             The second case would be where we have the no

25   effect level for human study, and then in that case we would



 1   go back to what we're talking about before, and use the

 2   interspecies uncertainty factor of ten.

 3             That actually may be coming more of an issue as we

 4   start seeing more data submitted where in fact no effect

 5   levels are generated in humans and we're starting to see

 6   some of those with some of the organophosphates, with those

 7   studies done to generate a no observed effect level for

 8   cholinesterase inhibition, so that that starts becoming an

 9   issue.

10             CHAIRMAN FROINES:  Tony.

11             DR. FUCALORO:  I was just wondering, I don't

12   understand the heading for that where you say the

13   uncertainty factors used by DPR in calculation of MOE or

14   reference concentration.  Nowhere did I see in your

15   equations MOE using an uncertainty factor.

16             Am I missing something?

17             DR. SCHREIDER:  Yeah.  When we calculate, let's

18   say we calculate an uncertainty factor of 40 for a given

19   situation, that's compared with -- I'm sorry I wasn't clear.

20   That was compared with a benchmark, and that benchmark would

21   be uncertainty factor.  In that case it goes from an animal

22   study to benchmark of a uncertainty factor of 100.

23             Would then be for management to determine whether

24   or not the adequacy of the -- of that exposure situation.

25             DR. FUCALORO:  It's not MOE, you're just talking



 1   about a reference exposure.

 2             DR. LIM:  In that --

 3             DR. FUCALORO:  Maybe I'm missing.  I don't see

 4   anywhere where an MOE uses that.

 5             DR. LIM:  Precisely.  It is not in the equation.

 6   What I meant to say is that in our discussion in our

 7   documents in the risk appraisal section we usually generally

 8   say have some discussion where the default ten would be

 9   adequate or not for interspecies or intraspecies.  That's

10   all I meant.  It didn't mean that it would physically be in

11   the equation.

12             CHAIRMAN FROINES:  This is a very important point.

13   This is an extremely important point, precisely because the

14   MOE in the DEF document is the basis for the determination

15   of whether something is a toxic air contaminant or one of

16   the bases.

17             But there is no explicit incorporation of

18   uncertainty factor in the MOE calculation.  I'm being very

19   clear.  I'm saying there is no explicit incorporation.

20   There may be an implicit incorporation, but that's what

21   we're trying to get you to tell us about.

22             DR. GLANTZ:  See, and I just want to jump on that,

23   because I had been -- I mean, this is actually very helpful

24   because I'm beginning to understand these distinctions

25   finally.



 1             But the problem with using the MOE is this point

 2   about the lack of an explicit consideration of the

 3   uncertainty factor, because, as you know from attending

 4   these meetings, a huge amount of discussion goes into the

 5   question of how to handle the uncertainties, and by not

 6   explicitly accounting for that, what you're doing when you

 7   come up with an MOE, let's say you come up with an MOE of

 8   9600, which is the DEF number.  Okay.

 9             Well, that could be a very big margin of -- if

10   there was not a high level of uncertainty associated with it

11   or it could be not a very big margin if there was a lot of

12   uncertainty.

13             So that I think what we need to be shooting toward

14   is actually going more toward the idea of the REL in these

15   pesticide documents where the REL that we come up with and

16   sort of bless, explicitly accounts for the uncertainty

17   factors, and then if you want to talk about a margin of

18   exposure based on the REL, that I think would make sense

19   because then what you're doing is you're coming out and

20   saying the margin of exposure is a thousand, what you're

21   saying is that the exposure levels are a thousand times

22   below the REL.

23             CHAIRMAN FROINES:  Let me make a point about that.

24             DR. GLANTZ:  Because, see, the problem you have

25   with this -- this is what confused me about this all along,



 1   the problem that you have with these, the margin of

 2   exposures is that you end up with an apples and oranges

 3   problem, because you're not taking into account the fact

 4   that one margin of exposure may be associated with something

 5   with a lot of uncertainty and another one might be

 6   associated with something where there isn't a lot of

 7   uncertainty.

 8             CHAIRMAN FROINES:  I want to follow up.

 9             DR. GLANTZ:  And then I have another question.

10             CHAIRMAN FROINES:  I want to make one point about

11   this.

12             The child MOE in the DEF document is 9600.

13             Now, part of the problem with that is the

14   perception.  Perception weighs heavily in all this.

15             So 9600 looks to be very safe.  So everybody says,

16   wow, we don't have to worry about DEF because the number is

17   9600.

18             Well, now we have a different NOEL for DEF than we

19   did when this document was written, which is now .6 and so

20   when you take the .6 versus the 12.2, and then if you add an

21   uncertainty factor of 100 in, it turns out that your MOE

22   incorporating the uncertainty factor into the NOEL

23   calculation turns out to be 20.  That's a very small number.

24             And one then has to say when you have that number

25   of 20 is the public appropriately protected at that level.



 1             And I think we might all agree that they are and

 2   you might disagree that they are.

 3             And the fact of the matter is once when you start

 4   to get these numbers of 9600, it's because if you divide by

 5   a hundred, you're down to 96.  And therein lies the problem,

 6   I think, everybody is concerned about.

 7             It is precisely that by not incorporating the

 8   value of 100 into an uncertainty factor calculation it gives

 9   an appearance of being different than in fact it really is,

10   let alone the science that we can talk about.

11             DR. SCHREIDER:  Yeah.  I think there would be a

12   couple points.

13             One is that we're moving in the direction of doing

14   the RELs, uncertainty for these documents, generating a

15   reference concentration.

16             And also I think part of this may be due to how we

17   started out, which was, aside from air toxics of having a

18   set concentration or set pesticide and starting with a

19   concentration on occupational exposure and saying, okay, is

20   exposure from all routes adequate or not adequate, and some

21   of this may be residue of that, and certainly the FQPA we're

22   still having to look at aggregate risk.

23             But I think certainly we're moving in the

24   direction of generating both values.

25             And part of it is also do we generate it



 1   numerically, which you're suggesting explicitly and

 2   numerically, or do you handle it in the risk appraisal

 3   section where it's described.

 4             And I think I clearly hear you say, no, ought to

 5   be really cut and dry so that it doesn't give the impression

 6   even if the --

 7             CHAIRMAN FROINES:  I've been putting words in your

 8   mouth, because I'm assuming as you go along in your

 9   presentation you're going to say that the MOE calculation

10   incorporates somehow what you haven't really said yet, the

11   100-fold factor.

12             So why don't you go back to where you were.

13             DR. GLANTZ:  Before you do, let me continue

14   interrupting here.

15             Because I'm just trying -- I'm just trying to

16   understand how what DPR and OEHHA are doing different,

17   because my goal is in the end to have you using the same

18   procedures, the same definitions.

19             DR. BYUS:  The definitions.

20             DR. GLANTZ:  The best definitions available and

21   what's good for -- but when you look at the uncertainty

22   factors, which you just you had up, or you recited, I went

23   back and looked at the slides that Melanie presented, and it

24   looks to me that you're using the same uncertainty factors

25   that OEHHA is.  Is that true?



 1             DR. SCHREIDER:  Basically.

 2             DR. GLANTZ:  Okay.  Basically?

 3             DR. SCHREIDER:  We look at a given date set --

 4             DR. GLANTZ:  Let me be more precise.  The default

 5   values that you're using are the same; is that correct?

 6             If there's data to not use a default factor,

 7   neither you -- both you and OEHHA would use the data.  But

 8   in the absence of specific information you're using the same

 9   uncertainty factors and you're using uncertainty factors for

10   all the same things too, right?

11             DR. SCHREIDER:  Right.

12             DR. GLANTZ:  If I handed a NOEL to Melanie or to

13   you and said go compute the reference exposure, the

14   reference exposure level, the reference concentration, you

15   both get the same number.  Is that correct?

16             DR. SCHREIDER:  Correct.

17             DR. FUCALORO:  But I guess the problem, and maybe

18   I'm missing some of what you're saying, I think this is what

19   you're saying, John, is that the MOE, the REL is independent

20   of what's actually out in the environment.  It's something

21   that's a pure figure that people have worked through and

22   working through statistics and studies, whereas MOE of

23   course has exposure in it.

24             And I guess, am I correct in assuming this is what

25   you object to, essentially?  Am I correct?



 1             And that -- and the difference between them

 2   algebraically and I think that's obvious in your first

 3   slide, is the MOE is equal to the REL times the factor times

 4   UF over exposure.  I mean, that's just algebraic.

 5             So the answer to your question is you can always

 6   calculate, if they use the same set of uncertainty factors.

 7             I mean, I don't know, is there anything more

 8   complicated?

 9             I do I think agree with Dr. Froines in saying that

10   the REL is a much purer number in the sense that it doesn't

11   have the effect of exposure which can change, after all,

12   from area to area and from year to year.

13             DR. SCHREIDER:  Correct.

14             DR. FUCALORO:  Is that all there is to the

15   discussion or am I missing something?  It seems --

16             CHAIRMAN FROINES:  There's two parts of the

17   discussion, I think.

18             One part is let's assume we did what you said,

19   which is I think what we should do.  I think we should take

20   the REL as -- you should establish an REL and then look at

21   your exposure as the second factor.

22             DR. FUCALORO:  That's what I thought you said.

23             CHAIRMAN FROINES:  Then the question becomes what

24   limit, what definitions do you put on your exposure and what

25   constitutes -- is an exposure equal to the REL, is that



 1   considered safe, not safe.  In other words, is there an

 2   exposure uncertainty that we then apply?

 3             They're using the 95 percent upper confidence

 4   limit on the distribution and so one could argue that that's

 5   a reasonable number and adds some conservatism and so one

 6   could take the upper 95 percent confidence limit as the

 7   exposure number and divide that into the REL, and if you

 8   have one that would seem like a reasonable value.

 9             And so but in each case all I'm saying is each

10   case you have explicitly defined what you're talking about.

11             Whereas here you're not.  You've got this exposure

12   which actually includes their uncertainty factors built into

13   that, so it's not explicit.

14             DR. FUCALORO:  When they get an MOE of 9600, you

15   have to ask yourself how does that compare to uncertainty

16   factors and I think that's therein lies a bit of confusion.

17   It's much cleaner to use the REL, since --

18             CHAIRMAN FROINES:  We take the current, we take

19   the current NOAEL for DEF and go through the numbers, then

20   it turns out that the value that you calculate is 20.  And

21   then you have to make a judgment about how important you

22   think 20 is.

23             DR. BYUS:  They just corrected it.  I kept looking

24   at the numbers.  I still don't get it.  I was going to ask

25   you --



 1             CHAIRMAN FROINES:  They add up now.  The number

 2   should be -- the denominator should be 303.5, right?

 3             DR. GLANTZ:  Nanograms.

 4             DR. BYUS:  I thought there was another --

 5             CHAIRMAN FROINES:  So go ahead.  We're now

 6   disrupting your presentation.

 7             DR. SCHREIDER:  Not at all.

 8             When we calculate the margin of exposure that does

 9   not in itself, that is in effect then compared to what would

10   be the appropriate uncertainty factor or total safety

11   factor.  That's done in the risk appraisal section where,

12   depending on what words were used, it may be that because

13   this was derived from a low effect level, not a no effect

14   level, we should have an uncertainty factor that would

15   normally be a benchmark of 10-fold for LOEL to NOEL.  Those

16   are the sorts of things that would go into a verbal

17   description in the risk appraisal section and, again, that

18   typically was from what we were asked to do for overall

19   assessments of the chemical is the exposure to this chemical

20   acceptable or not and to give information to risk management

21   to make that decision.

22             MR. GOSSELIN:  We can't hear anybody.

23             CHAIRMAN FROINES:  Can you get closer to the

24   microphone.

25             DR. SCHREIDER:  Can you hear me now, Paul?



 1             MR. GOSSELIN:  Yes.  I don't know if people aren't

 2   speaking into the mikes or not.

 3             DR. SCHREIDER:  I'm now speaking into it.

 4             And I think the other point is that in the case of

 5   the air toxics that we're moving in the direction of

 6   calculating the RELs as a reference concentrations.  That's

 7   going to be a standard practice.

 8             And, Paul, you can correct me if I'm wrong.

 9             Paul?

10             MR. GOSSELIN:  Yes.

11             DR. SCHREIDER:  I was indicating that we are in

12   fact for the air toxic compounds going to be generating

13   reference concentrations or RELs.

14             MR. GOSSELIN:  That is correct.

15             DR. SCHREIDER:  So that that will be generated in

16   addition to the margin of exposure calculation and part of

17   that is so that we can then combine the exposure from all

18   the routes of exposure, whether it's going to be inhalation

19   in the case of the air toxics, but there may also be

20   occupational, depending on how the pesticide is used around

21   the home, maybe dermal exposure, so that those can be

22   combined to try and determine overall margin of exposure to

23   that compound.

24             So to that extent, the reference concentration

25   would not be applicable -- or be particularly suited to that



 1   overall concentration or the overall calculation of what the

 2   overall exposure was.

 3             So I think we are going to be moving in the

 4   direction and we'll be including both values.

 5             DR. MARTY:  I think I need to add something to the

 6   discussion that might clear it up a little more.

 7             The way we use a reference exposure level in

 8   site-specific risk assessment in the hot spots program, that

 9   reference exposure level is compared to the modeled ground

10   level concentrations that are predicted using air dispersion

11   models and information about the facilities' emissions.

12             The ratio of the ground level concentration to the

13   reference exposure level is called the hazard index for that

14   chemical.

15             If the hazard index is one or higher, it triggers

16   risk management in the hot spots program.

17             So that would almost be equivalent to an MOE of

18   100 or 1,000, depending on the uncertainty factors that were

19   put into the reference exposure level.

20             CHAIRMAN FROINES:  I missed something.  I didn't

21   get what you said.

22             DR. GLANTZ:  I got it.  We'll explain it to you

23   later, John.

24             DR. FUCALORO:  We're getting cruel now.

25             DR. BYUS:  It's the coffee.



 1             CHAIRMAN FROINES:  Coffee, and people are getting

 2   hungry.

 3             DR. ALEXEEFF:  George Alexeeff.

 4             Let me explain it a little bit differently.

 5             DR. FUCALORO:  George always fills the breach.

 6             DR. ALEXEEFF:  When we are developing a reference

 7   exposure level, we're developing a concentration with the

 8   uncertainty factors built in.  So it's simply a risk

 9   assessment value.  And we are assuming that the uncertainty

10   factors we apply are appropriate, that there are risk

11   assessment issues issued on variability or taking into

12   account subpopulations.

13             So our reference exposure level is calculated as a

14   risk assessment value.

15             That value is then used by a risk manager, let's

16   say, they look at their exposure, and then they look at the

17   ratio of that.  If the ratio is greater than one, that is to

18   say the REL over the exposure, if it's greater than one --

19   actually it's flipped.  You're right.  If the ground level

20   exposure over the reference exposure, if it's greater than

21   one, that means your exposure is exceeding the reference

22   concentration, and then the risk managers have to think

23   about what they -- if they need to do anything.

24             So that is analogous to the MOE.

25             What my understanding of the MOE, and this might



 1   help in the discussion, is that it's basically a tool more

 2   for the risk manager where the REL is basically a risk

 3   assessment product, a risk assessment output.  And I think

 4   that's kind of one difference.

 5             They're setting up a system so the risk manager,

 6   as I said, DPR can make a decision as to whether it's a

 7   toxic air contaminant or there's some management issue to

 8   do.

 9             I don't know if there's -- as opposed to MOE

10   actually being a discrete risk assessment issue.

11             CHAIRMAN FROINES:  I have two questions.

12             One of which is if, George, if your ratio was .1,

13   then would the risk managers worry about that or would that

14   be basically he would, he or she would consider it

15   negligible risk?  In other words, where does the negligible

16   risk enter?

17             DR. ALEXEEFF:  They would consider it negligible

18   risk.

19             Usually it's under one, and depending upon the air

20   districts, how far below one is kind of their decision.

21   It's a little bit on the accuracy of how they measure.  Some

22   say .5, some might be .99.  I'm not sure.  Depends on the

23   different districts.

24             If it's below one, it's considered negligible.

25             CHAIRMAN FROINES:  Let me make my second point,



 1   because -- that was just bookkeeping.

 2             The more important point is this panel has

 3   historically disagreed with DPR on the determination of a

 4   compound as a toxic air contaminant.  That is, we've always

 5   believed a chemical should be determined a toxic air

 6   contaminant irrespective of the level of exposure.  And

 7   that's a fundamental disagreement which got us into some

 8   very contentious discussions in the '80s, and Stan remembers

 9   them well.

10             DR. GLANTZ:  Just for the record, I was just

11   observing John being contentious.

12             CHAIRMAN FROINES:  That's the -- so there is a

13   major underlying history and disagreement.

14             Now we're into the situation where we're not

15   debating that issue for at least for the moment, but the MOE

16   comes in.

17             And I think that this discussion comes up because

18   of this notion that once you get into the exposure as being

19   determining the determination of whether this is a toxic air

20   contaminant, then I still would feel better if we had an

21   explicit risk assessment process that then the exposure

22   determination became a next step in.

23             DR. GLANTZ:  Well --

24             CHAIRMAN FROINES:  And so I don't agree with the

25   notion of having it be essentially the third piece.  What



 1   you have is the risk assessment, the exposure assessment and

 2   then this hazard characterization, and then the risk

 3   appraisal.  And so that the uncertainty factors get brought

 4   in your process as the risk appraisal, which is at the very

 5   end of this document, and that's the problem.

 6             DR. BYUS:  I agree.  I didn't realize it at all

 7   until you were just going over it.

 8             I'm not sure what the MOE real advantage of it.

 9   Who uses the MOE?

10             CHAIRMAN FROINES:  US EPA.

11             DR. SCHREIDER:  US EPA.

12             DR. GLANTZ:  This is one more way that we can do

13   better than they do.

14             FROM THE AUDIENCE:  I think I know --

15             CHAIRMAN FROINES:  Excuse me.  I think that what

16   should happen is if your leadership wants you to come up and

17   speak to it, he should have you come up, but I'd rather you

18   don't call up and come walking up.

19             FROM THE AUDIENCE:  The MOE is based on 100, not a

20   one.  That's billion in 100, MOE 100 acceptable.

21             DR. GLANTZ:  Well --

22             CHAIRMAN FROINES:  That didn't help.

23             Dale, what you were going to say?

24             DR. HATTIS:  I think that part of the difference

25   is in fact, I think, perceptual.  And the very fact of



 1   having a big number being exhibited as the difference

 2   between the exposure that's okay or not okay in the exposure

 3   that exists, has a policy implication, to put it in the most

 4   neutral way I can.

 5             And that one of the ways in which this can be

 6   confusing to people is if in fact some of that number is in

 7   fact what should really be an adjustment factor for real

 8   average differences between the exposure circumstance in the

 9   experimental animals and exposure circumstances --

10             CHAIRMAN FROINES:  I think that's particularly

11   important to the lay public when you take -- when they're

12   looking at this thing and they see something called a no

13   effect level, a no effect level, and then you divide it by

14   some exposure and you get 10,000, the average layperson

15   would say let's go home and forget this chemical.

16             And that perception is a problem, because the

17   reader is first seen as the no effect level, well, if you're

18   10,000 a no effect level, you must be safe.  So that the

19   person who doesn't understand some of these more complicated

20   elements is going to have a very different perception and

21   it's going to create a level of belief that the compounds

22   are safe, that may or may not be entirely accurate.

23             DR. FUCALORO:  To beat a dead horse of course, a

24   MOE of a million does not tell you whether or not the

25   compound is very toxic.  I mean, it doesn't matter.  Really



 1   the REL tells you how toxic it is and then you look at

 2   ambient concentrations and see if you want to control this.

 3             I mean, we're saying the same thing, the algebra

 4   is very simple.  We're just wondering, I guess, as a panel,

 5   I don't know how much influence we have, whether or not

 6   documents can be recast in such a way as to introduce

 7   something like REL and in order to talk about the inherent

 8   toxicity of a material, rather than looking at the effect on

 9   the population, given some ambient concentration that people

10   think we have in California.

11             MR. GOSSELIN:  Might I interject?  This is Paul

12   Gosselin.  Can everyone hear me?

13             DR. FUCALORO:  We sure miss you.

14             MR. GOSSELIN:  Miss you too.  I'd rather be there

15   than here.

16             You know, I caught the tail end of this and I

17   think Jay said it's our intent to have our documents be in a

18   format that OEHHA -- you're used to seeing from OEHHA.

19             So that's the direction we're moving in.  It's our

20   intent to recraft our documents in more of an REL type

21   presentation versus an MOE.

22             CHAIRMAN FROINES:  Stan.

23             DR. GLANTZ:  I think that's wonderful.

24             What I would -- what I'd like to suggest we do,

25   just to formalize this, is that we say to DPR that we like



 1   the documents to come forward with the REL in them, and also

 2   since you, in order to better express what we know about the

 3   exposure compared to the REL is also include the hazard

 4   index, which OEHHA uses for -- in their things.

 5             And then if you want to also put the MOE in

 6   because you like it, I don't mind if it's there.

 7             But I think what -- so I would suggest that the

 8   panel actually say that's how we want the documents

 9   presented, which I don't think is that different from what

10   you're saying, Paul.

11             MR. GOSSELIN:  No.  I would actually say that if

12   you want to move to endorse our efforts to do that --

13             DR. GLANTZ:  Okay.

14             DR. FUCALORO:  I second.

15             DR. GLANTZ:  I so move.  And Fucaloro seconded.

16             CHAIRMAN FROINES:  Do you want language or --

17             DR. GLANTZ:  I think I'll make a motion, if you'd

18   like a motion, and that is that we direct or that the

19   pesticide documents have an REL presented in them, a hazard

20   index to provide a better measure of the risk, the magnitude

21   of the risk, and in also the controlling for exposure, and

22   also that OEHHA and DPR use the same uncertainty factors,

23   same default uncertainty factors, in that analysis, so that

24   we will standardize things.

25             Obviously, if there's -- that's the end of my



 1   motion.

 2             And then I'll just add the comment that if there

 3   are data that would lead you to something beside the default

 4   uncertainty factors, that obviously you use the data driven

 5   uncertainty factors.

 6             But so that's the motion I'm making.

 7             DR. FUCALORO:  My second stands.

 8             CHAIRMAN FROINES:  So I think, let's just take --

 9   is there any discussion?

10             MR. GOSSELIN:  I would add that the uncertainty

11   factor issue, suggest that it might be taken out, because

12   one of the things we do when we go to a peer review with

13   OEHHA is that those issues could and should come up during

14   that process, and that if we do have legitimate reasons to

15   be different, we can explain that in our comments and our

16   report back.

17             But I think that categorically taking a blanket

18   statement that the two institutions are going to immediately

19   bridge scientific perspectives, I think it's probably kind

20   of a leap.

21             DR. GLANTZ:  Well, I don't agree.  Well, that's

22   okay, jump.

23             DR. WITSCHI:  Do we have a discussion by a

24   non-panel member on a motion that's made by the panel?

25             DR. GLANTZ:  Okay.  By a what?



 1             DR. BLANC:  It's useful to -- Paul Blanc here.  I

 2   think it's useful to hear Paul's perspective on this, but

 3   now that we're heard that perspective, I think it would be

 4   useful to hear from the panel members, and useful for you to

 5   hear from the panel members their take on that.

 6             I think what you're going to hear, though, is from

 7   people that the consensus here is that OEHHA and your agency

 8   should not come to us with different uncertainty factors for

 9   the same chemical.

10             I think that OEHHA's already stated that there

11   will be some variability within their own agency on certain

12   uncertainty factor adjustments, varying between three and

13   ten, depending on the chemicals.  So that does leave you

14   some leeway, but I don't want to be in a situation, and I

15   think the other panel members feel the same way, that for

16   the same chemical, let's say for some reason you were coming

17   to us with a document related to chlorine as a biocide, we

18   wouldn't want to have you using an uncertainty of three and

19   OEHHA using an uncertainty factor of ten for that same

20   chemical.  And similarly if the range of uncertainty factor

21   from OEHHA varies from three to ten for sensitive

22   populations, let's say, we wouldn't want you to come in with

23   a one that had adjustment of two or 1.5.

24             CHAIRMAN FROINES:  If one can define a reason why

25   an uncertainty factor should differ.



 1             DR. BLANC:  Within that range.

 2             CHAIRMAN FROINES:  Within that range, then they

 3   could petition and give a scientific argument.

 4             But I think that the general principle is that

 5   reasonable people should reasonably agree on uncertainty

 6   factors.

 7             DR. GLANTZ:  Yeah.  I think we spent a lot of time

 8   discussing this issue in the context of the acute reference

 9   exposure document.  And I'm pretty comfortable with the

10   uncertainty factors that we came up -- the default, I want

11   to keep saying default, uncertainty factors that we came up

12   with there.

13             And I mean the numbers which were presented here

14   by DPR as the defaults are in fact the same ones that OEHHA

15   is using anyway.  So I don't think that should be a

16   controversial point.

17             Again, obviously if there's data that would lead

18   you to conclude that one -- that you shouldn't use the

19   default number, then you should use the data.  The data is

20   always better than the default.

21             CHAIRMAN FROINES:  This is -- but Stan's right,

22   that's a good point, because where -- you know where our

23   history is.  The studies that DPR uses are often these

24   industry studies that have a kind of mysterious quality to

25   them because we never see them and so they --



 1             DR. BYUS:  We can see them if we want to.

 2             CHAIRMAN FROINES:  I know.  If we go to the

 3   library, I understand.

 4             But the point is that DPR may, based on those

 5   studies, decide that they want to establish a NOEL based on

 6   study X from Eastman Kodak, for example.

 7             George has been working with data on the same

 8   chemical, but he's been doing it all with the peer reviewed

 9   literature, for example, just as a hypothetical situation.

10             He decides that his NOEL is based on a different

11   set of data than what DPR decides to do.

12             But it seems to me if that's the case, then we

13   really do need to work to get the two groups reading the

14   same studies to make -- and they shouldn't be making

15   different determinations based on different scientific

16   studies, which means that what they decide to use for the

17   NOEL should be reasonably consistent or else we ought to

18   find out why it's not consistent.

19             Does that make sense?

20             MR. GOSSELIN:  I agree with that.

21             DR. FUCALORO:  Let's call the question.

22             CHAIRMAN FROINES:  So we'll call the question, all

23   in favor raise your hands.

24             (Panel members raise hands.)

25             CHAIRMAN FROINES:  So that passes.



 1             DR. GLANTZ:  Unanimously.

 2             CHAIRMAN FROINES:  Unanimously.

 3             Now, there's one other issue that we haven't

 4   talked about, but I think we're not going to talk about

 5   today, but it is really an issue of major magnitude.

 6             DR. GLANTZ:  Before you do that, just to pound one

 7   last nail into this coffin, I don't think this is

 8   controversial, but I would appreciate it if the OEHHA and

 9   the DPR people would just get together and look over the

10   default uncertainty factors and the definitions and all that

11   stuff that we spent so much time talking about in the

12   context of the REL document, and just come back at the next

13   meeting and just say to us, we looked at them, and we are

14   agreed on the definitions and we're agreed on the

15   uncertainty factors.

16             I mean, that's what you've already said, but I

17   just want to make absolutely sure that nothing is falling

18   through the cracks.

19             It should be a short conversation, but I just want

20   to be absolutely positive that everybody is on the same

21   page.

22             If you just check, if you guys could just get

23   together, look over that and come back to the next meeting

24   and say, yeah, we did that, and everything is under control.

25             I think it is.



 1             DR. ALEXEEFF:  We can certainly do that with

 2   regard to the, let's say, the acute document, which has

 3   already been processed through.

 4             I don't know what we do if we have a difference,

 5   but let's presume we don't.

 6             But then the chronic document will be different.

 7   There's additional uncertainty factors in the chronic

 8   exposure issues.  So we'd almost have to do it again at that

 9   point or --

10             DR. GLANTZ:  Well, I think then you should,

11   because what I want to see happen is a standard protocol

12   that everybody is using.  And so if we need to revisit that

13   in the chronic document, we should.  And I would hope in

14   developing that that you'll talk to DPR so that when that

15   comes forward, any issues will have been resolved by the

16   time it comes to us.  And if they haven't been, then we can

17   help resolve them.

18             DR. ALEXEEFF:  And then along similar lines, the

19   DPR develops seasonal values, which we don't develop, so

20   that would be some discussion of that at some point.

21             CHAIRMAN FROINES:  That's what I'm coming to.

22             DR. ALEXEEFF:  Three different types of values

23   that we develop, and depending on some slight modifications

24   of those uncertainty factors in those cases.

25             MR. GOSSELIN:  I was raising an invisible hand



 1   here.

 2             One thing, a couple things is that I agree with

 3   Stan on the point, because I, you know, I walked away after

 4   the discussion on the acute documents and my staff reviewed

 5   it, and largely acknowledged how well written the document

 6   is, how clear, and actually pointed to a need for us at DPR

 7   to go back, take a look at what our existing policies are

 8   and actually, in light of the acute documents, start to

 9   document what our policies are and start to follow down

10   similar paths.

11             I think if particularly since both agencies are

12   going to be dealing with the panel, if there's a process to

13   prepare similar document on our chronic endpoints, I would

14   offer that we'd like to get involved with that in tandem

15   with OEHHA early on, and with the panel, to participate in

16   that.

17             But I think one of the things I walked away from

18   the acute document is that we're going to be -- we are

19   actually going to start internally to go through a process

20   of identifying what policies we need to revisit and sort of

21   recraft and then start sending those out to OEHHA, the

22   panel, and getting peer review on them before we finalize

23   them.

24             CHAIRMAN FROINES:  Good.  Thanks, Paul.  That's

25   very helpful.



 1             The issue I want to comment about that the final

 2   issue about the MOE that is in many respects by far the most

 3   crucial, because coming up with a NOAEL is not an easy task,

 4   but it's certainly a doable task, but I think that the

 5   actual determination of, quote, what we mean by exposure is

 6   a major task that's very difficult.

 7             And we have talked about it, Lyn Baker is here,

 8   and we've talked about it in the past.  Roger Atkinson is

 9   now on the committee and hasn't been part of all of those

10   discussions, but I think it will be interesting when we take

11   it up more.

12             And so I think that one of the key issues is going

13   to be how we actually define protocols for doing exposure

14   measurements, what are we trying to measure, are we trying

15   to do dermal, are we trying to measure metabolite, I mean,

16   rather breakdown products, air chemistry.  What are we

17   actually trying to measure and coming up with these exposure

18   values.

19             And so I make that as a major issue which we need

20   to develop speakers and actually have probably a half day

21   session talking about the whole issue of exposure.

22             And so we'll defer it for now, but I think it's a

23   fundamental issue that we have to deal with on this one.

24             So let's go over, if there's nothing else more on

25   this, we can -- I just want to bring you up to date.



 1             Does everybody have the DEF findings now?

 2             These findings have taken a while and they've

 3   taken a while because we were cavalier at a meeting not long

 4   ago where we went through, and you can tell everybody had

 5   had a lot of coffee, because everybody was jumping -- if you

 6   look at the transcript, everybody was jumping into the

 7   discussion and people held positions very strongly and

 8   wanted changes made and so on and so forth.

 9             And then that -- but we didn't write everything

10   down that everybody was saying, so we then had to go back to

11   the transcript and went through the transcript to see what

12   the changes in fact were, and when we looked at the

13   transcript to make the changes, we realized that there were

14   enormous inconsistencies in what we had done, and especially

15   in relationship to the document itself.

16             Well, I think that we have basically addressed all

17   the inconsistencies and all the problems, and so I think the

18   draft that you have is one that we would propose go forward.

19             I should say that a very major series of

20   discussions occurred because what we do at these meetings is

21   we always work with OEHHA and say -- and on lead Stan went

22   and took about three hours going through line by line

23   recommending changes.  If everybody remembers that.

24             And OEHHA sat there and made the notes down and

25   they then dutiful went out and made the changes and seemed



 1   at least in appearance's sake happy to do that.  We never

 2   know what was inside their head, but they were smiling.

 3             DR. GLANTZ:  At least some of them were happy.

 4             CHAIRMAN FROINES:  Some of them were.

 5             And they went and made the lead changes.

 6             DR. FUCALORO:  You should see the little dolls

 7   they have of you with pins in it.

 8             CHAIRMAN FROINES:  But we moved so quickly on the

 9   DEF document that the transcript reads that we want to do

10   this and we want to do this and we're going do this and

11   we're going to do this, and no place in that transcript does

12   it ever -- does Paul Gosselin's voice ever come in and say,

13   okay, we'll do that.

14             So we left with the situation where DPR really

15   hadn't been part of the discussion in a way that OEHHA is

16   normally part of the discussion when we're making changes.

17             And I think one rule out of that is when we go

18   through and make changes, we have to get agreement on the

19   part of the agency as we go through just to make sure that

20   everybody is on the same page.  And if Paul or other

21   representatives have major problems, that's when they should

22   state them.

23             So DPR then had to consider whether they wanted to

24   make the changes while I was rewriting the findings.

25             For example, on the designation of DEF as an



 1   NOAEL, as the values for the NOAEL, and so on and so forth,

 2   and Paul and I finally resolved it yesterday at 7:30 in the

 3   morning, and he agreed and DPR agreed, that those changes

 4   would be acceptable to DPR, and they will go back to their

 5   document and make the changes that now are -- that will make

 6   their document consistent with our findings.

 7             So that's the outcome.  It's, I think, a very good

 8   outcome.  It avoided a lot of potential controversy.

 9             So they feel that they were able to make those

10   changes and they felt comfortable with them as a matter of

11   science.

12             Paul, did you hear what I said?

13             MR. GOSSELIN:  Yeah.  Actually, I got called out

14   on a phone, but I heard the last couple minutes and, yeah, I

15   largely agree.  I think part of it too is that I think over

16   the past year just the rebuilding of a relationship between

17   us and the panel, you know, on better communication of

18   getting at least through the DEF document process.

19             CHAIRMAN FROINES:  Okay.  So I think these

20   findings, as they currently exist, are acceptable to DPR,

21   are acceptable to the panel, I mean, except people are

22   welcome to make suggested changes, but this is what will go

23   forward.  And I think the process worked out, although it

24   wasn't as easy as everybody thought it would be.

25             There is one thing I do want to say, everybody now



 1   has the cholinesterase document, which I downloaded.

 2             DR. GLANTZ:  Just a point.  Do you want us to vote

 3   on these or anything?

 4             CHAIRMAN FROINES:  Yeah.  I'll come back to that.

 5   Let me make one comment.

 6             I think one thing that is very important is that

 7   we made a decision on defining a number, and that number we

 8   defined as a no observable adverse effect level for DEF.

 9             I do want to make the point that from a matter of

10   science that we did not define policy on blood plasma or RBC

11   cholinesterase.  We said that in the case of DEF that the

12   plasma and RBC cholinesterase is a reasonable surrogate for

13   an adverse effect.  That's the decision we made.

14             But I just want to emphasize that we'll have to

15   make that decision on every chemical that comes up, because

16   it's not going to be true that plasma cholinesterase is an

17   adverse effect in every case.  It's going to be defined by

18   the toxicokinetics of the compound for the most part.

19             And I feel very strongly about that.

20             So we haven't made, as far as I'm concerned, we

21   haven't made a policy decision.  We have made a

22   determination based on the science associated with this

23   particular pesticide, and the next pesticide that comes up,

24   we'll have to revisit the issue of plasma cholinesterase in

25   the context of looking at the science.



 1             And we're going to have to ask DPR to do a much

 2   better job in the future in developing the toxicokinetics of

 3   these pesticides, because they are going to differ very much

 4   in terms of whether or not they're reaching the brain,

 5   whether or not they are having an effect in the brain, and

 6   whether what you measure the plasma has anything to do with

 7   what's happening in the rest of the body.

 8             I'm waiting for Paul, who's just got a --

 9             DR. BYUS:  I'm not sure I agree with John.

10             DR. BLANC:  Is there some reason that you'd rather

11   not share that you're making this statement?  Am I missing

12   something?

13             CHAIRMAN FROINES:  About what?

14             DR. BLANC:  Why you're making that statement.

15             CHAIRMAN FROINES:  I just want to make clear that

16   we, at this point, have not made a policy decision about

17   blood cholinesterase as an adverse effect, that we've made a

18   decision about blood cholinesterase relative to DEF.

19             DR. BLANC:  You think it has implications for how

20   we might approach other --

21             CHAIRMAN FROINES:  No, I personally think that

22   every chemical is going to be different.  And some

23   chemicals, a plasma cholinesterase inhibition may have no

24   significance whatsoever and in some cases it may be an

25   adverse effect, but I think you can only determine that by



 1   looking at the science associated with the determination.

 2             DR. FUCALORO:  Didn't Hanspeter point out the

 3   problem with plasma cholinesterase two times ago saying that

 4   it's a -- I hate to use the term -- generic effect, that

 5   regardless of how it's depleted it can have, for sensitive

 6   people, some very negative health effects.

 7             I'm not a toxicologist.  I just -- you were

 8   talking about the surgery or something.

 9             DR. WITSCHI:  This was a famous case was king of

10   Morocco in '56 or something like this, who didn't tolerate

11   the anesthetic for a trivial operation because he had a

12   deficient plasma cholinesterase and he never woke up out of

13   the narcosis, and that's how people became aware.

14             It also happened in Switzerland, that's why I

15   know.

16             CHAIRMAN FROINES:  Well --

17             DR. BLANC:  We'll revisit that.  I just was

18   curious if there was something I was supposed to be getting

19   that I wasn't getting.

20             CHAIRMAN FROINES:  No.  I just think that we may

21   find that clearly a plasma cholinesterase is a surrogate for

22   something else going on and in a quantitative sense you can

23   have lots of plasma cholinesterase inhibition going on

24   without necessarily brain cholinesterase being --

25             DR. BLANC:  I'll just say that, John, because most



 1   cholinesterase inhibitors in commercial use are lipid

 2   soluble, I think that it would be the unusual situation I

 3   think that you're describing, and that I do view what we did

 4   as precedent setting, and I do think that DPR should not get

 5   the misimpression that cholinesterase inhibition is going to

 6   be presumed benign until proven otherwise.

 7             So, clearly, every time that we take a decision it

 8   is not a blanket policy for all chemicals.

 9             On the other hand, it would be, I think,

10   misinterpreting the thrust of this group were DPR to somehow

11   think that this was an aberrancy in terms of cholinesterase

12   inhibition.

13             CHAIRMAN FROINES:  I think that's fine.  All I'm

14   saying is that I think as a matter of policy we haven't made

15   that decision.

16             DR. BLANC:  That's always true, isn't it?

17             CHAIRMAN FROINES:  No.  We could define policy if

18   we chose to.

19             DR. BLANC:  But we didn't.

20             CHAIRMAN FROINES:  But we haven't though.

21             DR. BLANC:  No.

22             CHAIRMAN FROINES:  I sat down with eight

23   toxicologists and pharmacologists and discussed this at UCLA

24   recently and there wasn't a single person who in giving some

25   examples from some certain organophosphates would have



 1   called that an adverse effect whatsoever.

 2             So I think that there's very strong scientific

 3   points of view that may differ very widely on this issue.

 4             I think we have to look at it as the issues come

 5   forward to us.

 6             And everybody may vote to call it an adverse

 7   effect, but I think we have to look at it as an individual

 8   situation rather than a generic situation.

 9             DR. BLANC:  That's fine.  I'm just saying that in

10   general it's going to be the burden of argument is going to

11   be -- have to be that it's not an adverse effect, in

12   general.  It's a very suspect outcome variable.

13             CHAIRMAN FROINES:  Yeah, I agree.

14             And I also think that what you find in the plasma

15   is different than what you find in the red blood cell and

16   that raises some other issues.  So that we'll have to talk

17   about the whole ball of wax as it comes forward.

18             Finally, the pesticide documents.

19             DR. FRIEDMAN:  Do you want discussion of the DEF?

20             CHAIRMAN FROINES:  Oh, yes.

21             DR. FRIEDMAN:  I obviously have not read the whole

22   thing, since we just received it, but I feel there's

23   something explicit missing from the conclusions.

24             If you read 31 and 32 on page five, it says for

25   something to be a toxic endpoint, pesticides with risk



 1   greater than ten to the minus seven should be identified as

 2   such, and then it just in 32 it gives the levels which are

 3   obviously greater than ten to the minus seven, or some of

 4   them.

 5             But I'd like to see that 32 have a statement,

 6   therefore DEF is a toxic air contaminant.  I don't think we

 7   ever say that explicitly.

 8             DR. GLANTZ:  That's on No. 34.

 9             DR. FRIEDMAN:  I didn't see that.  Sorry.  I

10   missed that.

11             DR. GLANTZ:  I move we adopt them.

12             CHAIRMAN FROINES:  I thought there was one other

13   question somebody had.  I thought Peter had a question.

14             DR. WITSCHI:  No.

15             CHAIRMAN FROINES:  You're okay.

16             DR. WITSCHI:  Okay.

17             DR. FRIEDMAN:  I just feel a little uncomfortable

18   getting this document and saying -- with no chance to look

19   at and saying we should adopt it.  Could you just at least

20   run through the major changes that were made?

21             CHAIRMAN FROINES:  Sure.  No problem at all.

22             The reason you're just getting it is we just

23   resolved it and everybody -- the vote has been taken on this

24   issue.  So what you're really saying is what you'd like to

25   know is --



 1             DR. BLANC:  Does this reflect all the comments as

 2   you best as you could adjudicate.

 3             CHAIRMAN FROINES:  Yeah.

 4             The important factor is originally the major

 5   difference -- no, there are two major differences.

 6             One is Roger Atkinson was quite concerned with

 7   discussion of a number of elements of it, and particularly

 8   around air chemistry and he agreed to provide information to

 9   DPR that they could incorporate in their document and we can

10   make that available what he wrote to you.  It all

11   strengthens the document.

12             DR. FRIEDMAN:  Where does it appear in the

13   document or what in the document was changed?

14             CHAIRMAN FROINES:  Going to go in this document.

15             DR. FRIEDMAN:  I see.  It isn't in --

16             CHAIRMAN FROINES:  No.

17             DR. FRIEDMAN:  Okay.

18             CHAIRMAN FROINES:  And so the document -- so the

19   document will now have Roger's input.

20             The second major area was that the value -- look

21   under 18, originally DPR had proposed a NOEL of 12.2

22   milligrams per cubic meter.

23             We recommended that two changes.  One that the

24   NOEL not be a NOEL, it be an NOAEL.  And we also recommend

25   that that be 2.4 milligram per cubic meter based on blood



 1   cholinesterase inhibition.

 2             So if you'll look under 18, those values have been

 3   incorporated.  So now the report identified an acute air

 4   NOAEL of 2.4 milligrams per cubic meter.  So that represents

 5   the major change that occurred.

 6             Then there was discussion about the NTE, the

 7   neurotoxic esterase, which -- and Paul was commenting on

 8   that and I was.  And basically we've changed that so that it

 9   now states the inhibition of NTE in sensitive species is a

10   biomarker that correlates with the induction of OPIDN, you

11   know, delayed neuropathy.  So that change has been made.

12             And we've taken out the sort of the implication

13   that NTE is a mechanistic finding rather than a biomarker.

14   And I think that's important.

15             Then one of the interesting things that happened

16   was under 26 we added a sentence which says the SRP requests

17   DPR and OEHHA to evaluate the hen delayed neurotoxicity

18   model and determine how it may be used to perform

19   quantitative risk assessments, because as of now we don't

20   have the -- we have not incorporated the NOAEL from the hen

21   feeding study into the document as a NOAEL.  And the

22   question is how might we do that.

23             And that's important also, because this issue is

24   going to come up again and again to the degree that we deal

25   with compounds that display a delayed neuropathy.



 1             And finally on 27 I took out an entire paragraph

 2   and just put this in, which I think is very important.  And

 3   that reads the NOAEL of .4 milligram per kilogram day, based

 4   on brain cholinesterase inhibition for the rat oral toxicity

 5   study, is close to the NOAEL of 0.6 milligrams per kilogram

 6   day based on blood cholinesterase inhibition for the rat

 7   inhalation study.

 8             I take that as being a very important sentence,

 9   precisely because it shows that there is a quantitative

10   relationship between the brain cholinesterase and the blood

11   cholinesterase inhibition.  And so that the dose response

12   for the brain and the dose response for the blood are in the

13   same ballpark, and so one feels more comfortable defining

14   the NOELs and NOAEL, given the proximity of those values.

15             So that's pretty much it.  Those are the major

16   points.

17             So the document has now, as I say, it has a

18   significantly lower NOAEL and that changes the MOE to what I

19   calculate this morning as about MOE of 20, if you

20   incorporated the -- no, I think it is --

21             DR. BYUS:  I have three at 7,000 for No. 21.

22             CHAIRMAN FROINES:  What?

23             DR. BYUS:  It's written down as 3,000 to 7,000,

24   the MOE.  No. 21.

25             CHAIRMAN FROINES:  Well, we'll have to change



 1   those.  I just didn't do that calculation.  So I'll do the

 2   calculation and we'll change it.

 3             DR. FUCALORO:  Just two things.

 4             First, in item 32 is just a repeat of 23, and

 5   that's what you intended, is that correct?  It's a

 6   conclusion and really it's just the same, identical.

 7             And the other is the MOEs that you're prepared to

 8   change, I'm not quite sure I know why.  Maybe you should

 9   talk about it.

10             CHAIRMAN FROINES:  The MOE of -- let's see here.

11   In the document the MOE -- well, let me go back and check,

12   because in the document the MOEs range from 9600 to much

13   higher values.  So that 3,000 to 7,000 may reflect the new

14   values, and I'll just double check that.

15             When I say it gets down to 20, that's if you

16   included a 100-fold uncertainty factor in the calculation.

17             DR. FRIEDMAN:  Is there some reason that the

18   complete was crossed out, the bottom of 33?

19             CHAIRMAN FROINES:  Yeah.  I feel -- this is my

20   judgment call.  I feel that this document still has

21   significant limitations, that if we could it would be good

22   to see a rewritten document that was organized differently

23   that dealt with some of these issues like the hen values.

24   And I think there's things missing from this document.  And

25   we're not going to be able to correct them and we'd like to



 1   move forward on this.

 2             So I felt that the panel's finding would be a more

 3   accurate statement of the way we think about this if we said

 4   that the document represents a balanced assessment of our

 5   current scientific understanding, because I frankly don't

 6   think this document represents a complete assessment of our

 7   scientific understanding.

 8             DR. FUCALORO:  And I would say just one more

 9   thing, John, I don't mean to harp on this.

10             CHAIRMAN FROINES:  No, no, it's fine.

11             DR. FUCALORO:  I guess I actually do mean to harp

12   on it.

13             In item 21 where you're thinking of incorporating

14   uncertainty factors into their computed MOEs, that really

15   lends confusion, because the MOEs are not defined with

16   uncertainty factors.

17             CHAIRMAN FROINES:  I'm just going to check these

18   numbers to see if they're okay.

19             DR. FUCALORO:  Okay.  We may want to put in RELs,

20   which are fine with me, I have no problem, and they're

21   purely computational.

22             DR. GLANTZ:  The RELs are in the findings.

23             DR. FUCALORO:  They're in the findings, right.

24             CHAIRMAN FROINES:  No, Gary is exactly right, and

25   Tony is.  Any changes are -- this is your document.  So I'm



 1   not trying to speed us through.

 2             DR. BYUS:  I just have one question.  It says in

 3   No. 20, so the NOAELs include an uncertainty factor.  Is

 4   that -- this is what it says, the NOAEL, it incorporates

 5   100-fold uncertainty factor to address potentially increased

 6   sensitivity in humans, and then we have the MOE, which is

 7   the NOAEL divided by the exposure.

 8             Does the NOAEL include the uncertainty factors or

 9   not?  Someone care to answer me?

10             CHAIRMAN FROINES:  This document, what are you

11   referring to?  20?

12             DR. BYUS:  I'm referring to 20, yeah.  It says the

13   NOAEL value is the air concentration below which there is no

14   anticipated health risk.  It incorporates a 100-fold

15   uncertainty.

16             CHAIRMAN FROINES:  I think that this is a good

17   point.  I think what you mean here is that the it actually

18   refers to the air concentration level.  Am I not correct?

19             DR. SCHREIDER:  The reference concentration level

20   incorporates the 100-fold --

21             CHAIRMAN FROINES:  That's right.  The NOAEL

22   doesn't.  And you're absolutely right the it is referring --

23   I'll correct it.

24             DR. FUCALORO:  It's referring to the REL, right?

25             CHAIRMAN FROINES:  It's referring to the air



 1   concentration level.

 2             DR. FUCALORO:  The REL.

 3             CHAIRMAN FROINES:  No, it's referring to the air

 4   concentration level.

 5             DR. FUCALORO:  Which is the 8.8 micrograms per

 6   meter cubed?

 7             CHAIRMAN FROINES:  Listen, it says an air

 8   concentration --

 9             DR. FUCALORO:  I see, I see, I see.

10             CHAIRMAN FROINES:  Then you've got that statement

11   about the REL in parentheses, so that the it below is

12   referring to the air concentration level.

13             DR. BYUS:  That's --

14             CHAIRMAN FROINES:  We'll clean it up.

15             DR. FUCALORO:  Equivalent to the usage to an OEHHA

16   REL.

17             I'm reading like mad.

18             DR. FRIEDMAN:  So I understand that if we do vote

19   to adopt this that you will go through it word by word and

20   make the changes that incorporate some of the things that

21   have been brought up today or anything else that you might

22   find that's incorrect?

23             CHAIRMAN FROINES:  Yeah.  We'll correct any other

24   errors that we find, but it's better to find them now

25   because we won't have to go back.



 1             DR. FUCALORO:  In any event, when you do clean it

 2   up and if we do vote on it today, you will send us a final

 3   copy so we can take a look at and see if there are any

 4   problems.

 5             CHAIRMAN FROINES:  This is it.  Let's do it today.

 6   I don't want -- I'll send you a copy which has been sent

 7   forward.  I don't want -- do you guys realize how long we

 8   have been pursuing this?

 9             DR. BLANC:  Also, how many hours you must have

10   spent to go through this thing.  I'm sure it must have been

11   incredible.

12             CHAIRMAN FROINES:  The problem with it was is when

13   we got the first findings, we got SRP findings that DPR had

14   helped craft and we had OEHHA findings and they were

15   completely different.  So it's actually been a number of

16   different elements to it.

17             But so let's try and find everything we can right

18   now.

19             DR. ALEXEEFF:  Dr. Froines.

20             CHAIRMAN FROINES:  Yes.

21             DR. ALEXEEFF:  George Alexeeff.

22             Just going back to Dr. Byus' comment, I think the

23   sentence he pointed out still needs to be corrected, because

24   it is -- I don't think -- I think you are referring to the

25   next sentence, which required clarification of the word it,



 1   but the previous sentence is also incorrect.  It needs to be

 2   revised.  So because the previous statement says the NOEL

 3   value is the air concentration below which.

 4             CHAIRMAN FROINES:  You're right.

 5             DR. ALEXEEFF:  That has to be changed.  It should

 6   be --

 7             CHAIRMAN FROINES:  REL.

 8             DR. ALEXEEFF:  Yes.  The REL or the air

 9   concentration standard, whatever is being used.

10             DR. BYUS:  I'm not sure what you mean.  It doesn't

11   sound right.

12             My other question is about the MOEs on No. 21.  Do

13   we want to say the MOEs are significantly above the target

14   value for prompting regulatory action?  Is that a

15   informative statement, are we making that finding?  Do we

16   think -- is the MOE is not much value in determining that

17   statement.  That's what I'm asking.

18             DR. FUCALORO:  Because the statement like that

19   implies we have some uncertainty factor in mind in order to

20   make that statement.

21             CHAIRMAN FROINES:  We should take it out for

22   another reason.  That's a risk management statement.

23             DR. BYUS:  Take it out.  That's not our finding.

24             CHAIRMAN FROINES:  Our job is not to prompt -- our

25   job is to hopefully prompt regulatory action, but not to say



 1   that we're prompting regulation.  So that's good.

 2             DR. FUCALORO:  Let me ask what's on some people's

 3   minds as they're furiously reading this.  I read this

 4   furiously too.

 5             Do we -- you would like to get a motion and

 6   passage of this document with all the changes noted prior to

 7   my comments, and perhaps after my comments, but we don't

 8   have any mechanism for mail ballot after seeing the final

 9   document or maybe that is much too cumbersome.

10             CHAIRMAN FROINES:  We can do whatever you like.

11             DR. FUCALORO:  Or we can wait until next time.

12             CHAIRMAN FROINES:  No, no, no.

13             DR. FUCALORO:  You refuse to do that.

14             DR. GLANTZ:  I'd like to suggest that maybe we

15   take a break, let people read it carefully for ten minutes

16   or something, and then make any final changes and vote on

17   it.

18             CHAIRMAN FROINES:  Okay.

19             DR. GLANTZ:  We should just stop and let everyone

20   go through it and just deal with it, because this has

21   dragged on for a very long time, and I think John has done a

22   very good job, but there may be a few dregs of things that

23   weren't edited properly, but I think just take, maybe take a

24   break, we've done this before, and let people just read

25   through it so they don't have to be listening to whatever



 1   everyone else is trying to say, and then just go through it

 2   and finish it.

 3             CHAIRMAN FROINES:  I would rather finish it today.

 4             DR. FUCALORO:  Okay.  I'm not going to insist.  I

 5   just wanted to, if people were not speaking up I wanted --

 6             DR. GLANTZ:  That way people will be able to just

 7   stop and --

 8             DR. FUCALORO:  I meant to crystallize a solution

 9   to what we're going to do, and this seems reasonable to me.

10             CHAIRMAN FROINES:  There's no -- these are -- I

11   marked four places where we made changes and all of them are

12   quite reasonable changes and they're changes we would like

13   to have made.  So that there are -- and they're changes that

14   we clearly missed the last time we went through this

15   document.

16             DR. FUCALORO:  Let me make one -- let me go one

17   step further.

18             I understand what you're saying, Stan, but are

19   people prepared to vote now?  If people are prepared to vote

20   now, then let's scrap the break and get on and finish up the

21   meeting.

22             DR. FRIEDMAN:  I think, may I propose that we vote

23   now, then you'll send us the final document with the

24   understanding that if there's some minor things in the final

25   thing that you send us, after you've made the recalculations



 1   and corrections that you will --

 2             CHAIRMAN FROINES:  I'll go with one caveat.  I'll

 3   send you the document and we will give you a return date on

 4   comments, which will be within 24 hours or --

 5             DR. FUCALORO:  Make it 48, but certainly a date

 6   and time certain.

 7             CHAIRMAN FROINES:  No matter how many hours we

 8   give you, you'll do them towards the end of those hours.

 9             DR. FRIEDMAN:  Sometimes we don't even get -- we

10   may not get the mailing.  We may be out of the office for a

11   day.

12             CHAIRMAN FROINES:  We'll fax them.

13             DR. FRIEDMAN:  We may be out of the office for a

14   day or two.

15             DR. FUCALORO:  There's natural variability in

16   human behavior.

17             CHAIRMAN FROINES:  We've gone this far.  We'll

18   give you 72 hours.  At the end of 72 hours if you haven't

19   got them back, it's over.  And that's a major compromise.

20   You guys all owe me now on the 72-hour time.

21             DR. BYUS:  We owe you.

22             Could on 24, could we define what NOEL is as

23   opposed to NOAEL, just so that we're clear we mean NOEL?

24             DR. FUCALORO:  One is Christmas.

25             DR. BYUS:  Right, I know.  We define NOAEL, we



 1   should define NOEL.

 2             DR. FUCALORO:  Adverse and not adverse.

 3             With that in mind, a motion to approve is in

 4   order, Mr. Chairman?

 5             I move to approve.

 6             DR. GLANTZ:  Second.

 7             CHAIRMAN FROINES:  All in favor?

 8             You notice I didn't ask for any discussion.

 9             (Panel members raise hands.)

10             DR. GLANTZ:  Can we have 72 hours of discussion,

11   please?

12             DR. FUCALORO:  By yourself.

13             CHAIRMAN FROINES:  The final thing --

14             DR. GLANTZ:  Just for the record it was a

15   unanimous vote.

16             CHAIRMAN FROINES:  The final issue for us is to

17   hear from DPR about their schedule.  And I actually have

18   three questions about that.

19             One, is I'd like to hear from DPR about their

20   schedule.  I actually have three questions about that.  One

21   is I'd like to hear from DPR about their schedule.  I'd like

22   to hear from Melanie about her schedule.  And I'd like to

23   know what we're doing in May.

24             MR. GOSSELIN:  You want me to go first?

25             CHAIRMAN FROINES:  Sure.



 1             MR. GOSSELIN:  We were scheduled to bring methyl

 2   parathion back in May.  There's a question of staff on

 3   actually having the document completely rewritten for final

 4   consideration.  We may still achieve that, but, if nothing

 5   else, we're going to have the lion's share of the issues

 6   that were presented during the workshop and the issues

 7   raised by the panel to be able to be presented then and also

 8   a recalculation of REL format and a more complete emphasis

 9   on blood and brain cholinesterase inhibition, so we can at

10   least schedule the discussion of the methyl parathion

11   document for May.

12             CHAIRMAN FROINES:  Okay.

13             DR. BYUS:  Pardon me, Paul, this is Craig Byus.

14             We are going to present it in May?  You are going

15   to do it in May?  Is that what you said?

16             I just talked to Ruby yesterday and she thought it

17   was going to be difficult to get done in May.

18             MR. GOSSELIN:  No.  And she sent me an e-mail.

19   It's kind of a question as to whether it's going to be in a

20   final format.  But I think even if it's not, I think she'll

21   be prepared to talk about the issues that she's worked on,

22   because I think if we don't -- if we're not able to finalize

23   it in May, I think just given the lapse in time from, I

24   think it was November, that it would probably be good to use

25   that meeting to bring the document back up, go over the



 1   changes that have occurred since then, and get prepared to

 2   wrap it up in June.

 3             DR. BYUS:  I think that's good.  I had a nice

 4   conversation with her about it yesterday.  She was just

 5   worried she wouldn't be able to get it to me in time to look

 6   at.  I think my discussions with her are very positive and

 7   what she's doing, I think, is very good, incorporating some

 8   new data and also doing the REL calculations, as well as

 9   calculating it on a variety of different parameters is a

10   good one.

11             If she doesn't get it to me -- I told her, I

12   encouraged her, correct me if I'm wrong, to try and bring it

13   up to the panel in May if possible.  At least we can go over

14   some of the issues.  If she gets it to me too late I don't

15   have time to read the entire document in detail, we can

16   still do that and bring it back up in June.

17             CHAIRMAN FROINES:  I have a question about that.

18             Lesley, if anybody knows, Ray, Lesley, Elinor, is

19   methyl parathion one of the 14 EPA risk assessments that

20   were just released?

21             MR. GOSSELIN:  Yes, it is.

22             CHAIRMAN FROINES:  It is one of them.

23             So we clearly need to get to the panel that EPA

24   risk assessment, so they have that to look at before they

25   get the DPR document.



 1             MR. GOSSELIN:  We'll attempt to get the latest

 2   version from EPA and get that out.

 3             CHAIRMAN FROINES:  Okay.  And has OEHHA written

 4   comments on your document yet?

 5             MR. GOSSELIN:  Yeah.  Those were presented back in

 6   November, and our response to those, we've already gone

 7   through that step.

 8             I believe OEHHA has drafted up their findings.

 9             But what I would suggest is that at this point we

10   work on probably just time wise to start taking a look at

11   OEHHA's findings and sort of the current shape of the report

12   besides crafting up closure to the document.

13             CHAIRMAN FROINES:  You've drafted your -- Melanie,

14   I'm sorry.  You've drafted your methyl parathion findings?

15             DR. MARTY:  Yes.

16             DR. ALEXEEFF:  George Alexeeff with OEHHA.

17             We've drafted them, and we've approved them

18   internally.  We just have not yet transmitted them to ARB or

19   DPR at this point.  But they're very similar to the previous

20   findings we had before.

21             CHAIRMAN FROINES:  So I think that we -- Paul,

22   we're going to plan to have methyl parathion in May.

23             MR. GOSSELIN:  Except that I think staff wise I

24   don't think we're going to have the rewritten report ready

25   in time.  It might be tight.  But I think if nothing else it



 1   would be a good point to at least go over the document where

 2   changes will be in the document and start to go over some of

 3   the findings, and then bring closure in June.

 4             CHAIRMAN FROINES:  Okay.  And I think Craig

 5   needs -- Craig is the lead, and so he needs to see where

 6   you've gotten to as soon as possible, because if he has any

 7   major problems, that could be a sticking point.

 8             MR. GOSSELIN:  Yeah.  That's why I'm trying to

 9   build in, and I don't think trying to squeeze all this in

10   May may work out, to give Craig time to -- a chance to go

11   over the document and bring up some final issues he might

12   have, but I think by the May meeting we would have reached

13   that point in that discussion to at least start having a

14   broader discussion with the panel.

15             CHAIRMAN FROINES:  Yeah.  Just I don't want to

16   take it up if we're not going to be able to really go

17   through something that we can bring to closure in our minds.

18   We don't want to create another one of these rocks rolling

19   up and down the hill again.

20             MR. GOSSELIN:  I know.  I was just thinking if we

21   try to bring this entire document up to the panel all at

22   once at the end, let's say we resolve all the issues with

23   the lead, the panel may bring issues up and the findings may

24   have some further discussion, and we may need to come back

25   in June anyways.



 1             CHAIRMAN FROINES:  What will then follow and when?

 2             MR. GOSSELIN:  You mean the next step?  I think

 3   it's going to take maybe about two weeks or couple weeks to

 4   get some of the rewrite and amendments to the report.

 5             Then that's going to be transmitted to Craig and

 6   for review.

 7             And then there will be some final interaction and

 8   hopefully we will resolve all the issues.

 9             And then the document itself with appropriate

10   flagging of the changed issues will be brought to the entire

11   panel.

12             CHAIRMAN FROINES:  No, yeah, I hear that.  But I'm

13   saying in terms of other pesticides besides --

14             MR. GOSSELIN:  Oh, oh.  Molinate is, that document

15   is talked about in December is being significantly

16   rewritten.  It's going to take some time, so we're looking

17   summer to having that document be brought back in a new

18   format.

19             The other issue is MITC.  We have the initial

20   draft document that we're going to start initiating

21   discussions with the lead on.

22             CHAIRMAN FROINES:  Who is our lead on MITC?  Roger

23   and Peter?

24             DR. WITSCHI:  Yeah.

25             CHAIRMAN FROINES:  You're the lead.



 1             DR. WITSCHI:  I'm not going to be available in May

 2   or June.

 3             CHAIRMAN FROINES:  This is something that is going

 4   to come up in the summer.

 5             So you're talking about a draft MITC document

 6   sometime over the summertime?

 7             MR. GOSSELIN:  You mean once we get through the

 8   leads?  Probably, yeah.

 9             CHAIRMAN FROINES:  And then what follows MITC?

10             MR. GOSSELIN:  Molinate will be brought back in

11   the summer.  We've already gone through one cycle with that.

12             And I think next up is azinphos methyl.

13             CHAIRMAN FROINES:  And that will be in the fall?

14             MR. GOSSELIN:  Yes.

15             CHAIRMAN FROINES:  And then what?

16             MR. GOSSELIN:  Offhand I don't have the rest of

17   the list with me.

18             CHAIRMAN FROINES:  Okay.  I'm interested in trying

19   to see if we can look at those 14 organophosphate risk

20   assessments and think about whether or not we can approach

21   them as a group and avoid ourselves, if there was some basis

22   that you folks and OEHHA could review those risk

23   assessments, and they were good and we could deal with

24   various elements that are missing that are necessary for

25   California, that we could think in terms of having 14



 1   pesticides move forward based on the US EPA risk assessments

 2   but --

 3             MR. GOSSELIN:  What I could do is have my staff

 4   get together with OEHHA, pull together those 14 US EPA risk

 5   assessments.  And would you want us to initially go over

 6   sort of the status and format of them with a lead?

 7             CHAIRMAN FROINES:  Yeah.  I think what we need to

 8   do is to see if we can approach them collectively.  I mean,

 9   George gives us 120 compounds at one time, so 14 should be a

10   piece of cake.

11             The problem is, I'm looking at Lyn Baker, there

12   may -- what information is available on them in terms of

13   pesticide use and air monitoring may be one of the gaps that

14   we have to worry about, and there may be others, but since

15   we have risk assessments already prepared, it behooves us to

16   take a look at those 14 and see how we can make that process

17   move more efficiently.

18             MR. GOSSELIN:  We've been working closely with EPA

19   on these 14 and they match up in many respects with what

20   we're working on.

21             If you want to get a lead or two, we can sort of

22   evaluate the status of the 14 with OEHHA and the lead and

23   report back in May.

24             CHAIRMAN FROINES:  Okay.

25             MR. BAKER:  Dr. Froines, Lyn Baker with ARB.



 1             Just thought you might be interested to know that

 2   we have done monitoring for eight of those 14.

 3             CHAIRMAN FROINES:  Well, that's good.  I mean, if

 4   we can do something to bite off a big chunk of pesticide,

 5   that would really be good, because the anticipated 12

 6   clearly never worked out.  And so but if EPA has done a good

 7   job on the risk assessment, and I'm not presuming that they

 8   have, they may be of some value.

 9             Melanie, what are you doing in May and June and

10   July?

11             DR. MARTY:  I hear my staff laughing.

12             We have two more hot spots related documents that

13   we need to bring to the panel.

14             One of them is the technical support document for

15   determining chronic reference exposure levels.  And we

16   mentioned that that has 120 chemicals or so in there

17   summarized with RELs.

18             What we wanted to do to make it more palatable is

19   to bring forth the methodology in the first 40 or so

20   chemicals first.

21             CHAIRMAN FROINES:  First what?

22             DR. MARTY:  First bring forth the methodology in

23   the first 40 chemicals and then feed in the rest of the

24   chemicals over time.

25             And we hope to have the methodology in the first



 1   40 chemicals to the panel in June.  So that's to you for

 2   review.

 3             We also have the exposure assessment and

 4   stochastic analysis technical support document which we want

 5   to bring forward to the panel this summer.  I'm guessing the

 6   earliest we would bring that to you would be July.

 7             CHAIRMAN FROINES:  Do we have a June meeting,

 8   Bill?

 9             MR. LOCKETT:  Yes.

10             CHAIRMAN FROINES:  Do we have a July meeting?

11             MR. LOCKETT:  No.

12             CHAIRMAN FROINES:  Good.  Let's not.

13             This is -- we're meeting with reasonable

14   frequency.  July would be a good time to take a break.

15   Maybe even August.

16             DR. MARTY:  We would anticipate that the rest of

17   the chronic RELs would come to you September-ish,

18   November-ish.  Something like that.

19             We also have a Governor's Executive Order to

20   evaluate ethanol as an oxygenate in gasoline, which we need

21   to do an assessment with reference to gasoline without

22   oxygenate and gasoline with MTBE, and that document needs to

23   be peer reviewed, and we were thinking that you all would be

24   the peer reviewers.

25             We have an absolute deadline of December 31st to



 1   get that to the Environmental Policy Council.

 2             CHAIRMAN FROINES:  What about Elinor and I put our

 3   hats on and MTBE hats on and go back to work.

 4             Are you going to look at combustion products?

 5             DR. MARTY:  Yes.

 6             CHAIRMAN FROINES:  And atmospheric chemistry

 7   products?

 8             DR. MARTY:  Yes.

 9             CHAIRMAN FROINES:  Because the key question has to

10   do with what are we going to think about in terms of acid

11   aldehyde and PAN, at least.

12             DR. ALEXEEFF:  Actually it's a fairly

13   comprehensive analysis of which the Air Resources Board is a

14   large part.

15             There's also a component from the Water Resources

16   Control Board in terms of breakdown products.

17             It hasn't been worked out exactly how the peer

18   review is going to work.  Clearly something at least the Air

19   Resources Board and our parts would certainly fit with the

20   activity that this panel does in terms of the atmospheric

21   chemistry breakdown, the types of health effects we're

22   looking at.

23             But we're not sure how it's fitting with the Water

24   Board issue and how we're going to get that component

25   reviewed.



 1             Along similar lines, there's a question in terms

 2   of understanding ethanol and its byproducts and relative

 3   toxicity.  There's a question as to how much we have to

 4   compare it with the existing fuel, which is the MTBE fuel.

 5             So under that circumstance we may also bring MTBE

 6   to the panel to review, which we have a document that's

 7   already been prepared.

 8             And there's also the UC report which can also be

 9   relied upon.

10             Simply to nail down the reference levels and the

11   potencies, so that when the calculations are done with the

12   ethanol document, we have agreement previously on what

13   numbers we're comparing them to.

14             CHAIRMAN FROINES:  I assume that you would have

15   to -- what would you do?  Modify your PHG document on MTBE

16   to bring it to the panel?  That has the quantitative risk

17   assessment.

18             DR. ALEXEEFF:  Right.

19             CHAIRMAN FROINES:  Our document doesn't have a

20   quantitative.

21             DR. ALEXEEFF:  It would be the PHG document, and

22   that's our Public Health Goal water document.  And that

23   document was written with the idea that we may have to have

24   an air risk number, so it has an air analysis as well.

25             But the advantage of the UC document is that it



 1   does incorporate, I believe, does incorporate, is my

 2   understanding, the Air Resources Board information on MTBE

 3   as well as the self-generated information on MTBE that UC

 4   did in terms of combustion analysis.

 5             So I think you'd have a full exposure assessment

 6   without having to wait until we wrote another document, is

 7   what I'm trying to think of.

 8             I'm not -- that is something that's going to be

 9   worked out with us and the Air Resources Board and the Water

10   Board.  We may be bringing you those documents later this

11   year as well.

12             CHAIRMAN FROINES:  We did an exposure assessment

13   but those data is a little bit out of date at this point.

14             Elinor, what's the last date that we have dated

15   from ARB?

16             MS. FANNING:  Gosh, I know we had some '96 data

17   and I believe -- I can't remember if we ever got the first

18   half of the '97 data incorporated into that document.  We

19   had issues with sort of getting, compiling it.  I would redo

20   that exposure assessment.

21             CHAIRMAN FROINES:  Are we going to see the chronic

22   document in May or June?  I'm sorry.

23             DR. MARTY:  We'll bring it to the panel in June,

24   but that's -- you guys are going to get it in June for your

25   review.  If I'm remembering correctly, the meeting is set up



 1   for the 16th.  So you won't have had enough time to review

 2   everything by the time the meeting comes around.

 3             CHAIRMAN FROINES:  So we have nothing from OEHHA

 4   right now in May.

 5             And we have methyl parathion in May.

 6             And it seems to me that if there's a problem with

 7   methyl parathion such that Craig doesn't think it's ready to

 8   come to the panel, then we should not have a May meeting and

 9   take up everything in June.  Makes more sense.

10             So Craig needs to be the litmus test on whether or

11   not we should plan to take up methyl parathion in May.

12             DR. BLANC:  The June meeting is in Los Angeles.

13             CHAIRMAN FROINES:  I guess.  I don't know.

14             So right now we will --

15             DR. GLANTZ:  You've got a lunch offer over here,

16   Craig.

17             CHAIRMAN FROINES:  Right now we assume that we

18   will have a meeting in May on methyl parathion, but it's

19   also possible there won't be a meeting in May, and in June

20   we'll take up methyl parathion and the chronic RELs.

21             DR. BLANC:  I would like to make a motion that we

22   adjourn.

23             CHAIRMAN FROINES:  Got a second?

24             DR. FUCALORO:  Did you make the motion or just

25   like to?



 1             DR. BLANC:  I'm making the motion that we adjourn.

 2             DR. FUCALORO:  I second.

 3             CHAIRMAN FROINES:  All in favor.

 4             Thanks, everyone.

 5             (Thereupon the meeting was adjourned

 6             at 1:45 p.m.)
























 3             I, JANET H. NICOL, a Certified Shorthand Reporter

 4   of the State of California, do hereby certify that I am a

 5   disinterested person herein; that I reported the foregoing

 6   meeting in shorthand writing; that I thereafter caused my

 7   shorthand writing to be transcribed into typewriting.

 8             I further certify that I am not of counsel or

 9   attorney for any of the parties to said meeting, or in any

10   way interested in the outcome of said meeting.

11             IN WITNESS WHEREOF, I have hereunto set my hand

12   this 21st day of April 1999.




                                     Janet H. Nicol
17                                   Certified Shorthand Reporter
                                     License Number 9764