' / r
'*•;.-, 02.
•"".•
••*•;-, - -
jrary
(202) 234-4433
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
WASHINGTON. D.C. 20005
(202) 234-4433
-------
INDEX
Welcome and Opening Remarks 3
Dr. Steven K. Galson
Introduction to Program 6
Dr. Deborah Norris
Integrating Neurotoxicology Assessments into 9
a Pollution Prevention Program
Dr. John O'Donoghue
Neurobehavioral Evaluations of Binary and 37
Tertiary Mixtures of Chemicals: Lessons Learned
Dr. Virginia Moser
The Toxicity Profile of a Mixture of Two 60
Anticholinesterase Insecticides, Chlorpyrifos
and Diazinon in Adult Male Rats
Dr. Stephanie Padilla
Neurotoxicity Chemical Mixtures: Example of 75
Ototoxicity Produced by Combined Solvents
Dr. Will Boyes
Integrating Dose-Response and Kinetic 93
Properties of Pesticides During Cumulative
Risk Assessment
Dr. Gregory Christoph
Aggregate and Cumulative Exposure 138
Dr. Mohamed Abou-Donia
Dose Additivity?: The Interaction of Drugs 155
on a Behavioral Baseline
Dr. Anthony Riley
USEPA Mixtures Risk Guidelines and the 180
Mixtox Database: Lessons Learned
Dr. Richard Hertzberg
OECD Guidelines for Health Risk Assessment 201
Dr. Herman Koeter
(202) 234-4433
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
WASHINGTON. D.C. 20005
(202) 234-4433
-------
1 PROCEEDINGS
2 Time: 9:10 a.m.
3 MS. SJOBLAD: Good morning, everyone. I'm
4 Beverly Sjoblad from the Office of Pesticide Programs.
5 We would like to welcome each and every one of you to
6 the symposium on the neurotoxicity of mixtures,
7 jointly sponsored by the OPP Technical Training
8 Committee and the OPP Scientific and Technical
9 Training Committee.
10 A full day has been scheduled. So without
11 delay, I'd like to begin our program by introducing
12 Dr. Steven Galson, Director of the OPP-TF Office of
13 Science, Coordination and Policy. Dr. Galson.
14 DR. GALSON: Thank you very much, and I
15 wanted to welcome all of you to the symposium on
16 neurotoxicity of mixtures. I wanted to particularly
17 thank the organizers in the Technical Training
18 Committee of OPPT and the Scientific and Technical
19 Training Committee in OPP, particularly Debbie Norris,
20 Beverly Sjoblad, Trish Coleman, Ethel Brown and John
21 Blowen.
22 It's a very important event to get this
23 group of people together. We need to do it more, and
24 I thank you for your efforts to bring us here together
25 on this important topic.
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
l I also want to thank the speakers who are
2 here from industry, from university centers, and give
3 a particular thanks to Herman Koeter for leaving the
4 springtime in Paris. Where are you, Herman? It's
5 probably raining there, too, but Herman is truly one
6 of the most patient people in the continent of Europe,
7 I'm convinced, and we're very grateful to him for the
8 important work that he is doing on harmonizing
9 guidelines in the OECD.
10 These guidelines are becoming increasingly
11 important as the world and the chemical regulation
12 world becomes more globalized, and he's really the one
13 who facilitates the involvement of so many EPA staff
14 and American government officials and scientists in
15 this important work. So thank you, Herman.
16 Why is this symposium important? For the
17 first thing, it's sponsorship. These training
18 committees that got together to provide this
19 opportunity are doing work that we don't have enough
20 of. There isn't enough education and career
21 development for the scientists in our part of the
22 agency.
23 As you probably know, we're the largest
24 group of scientists in EPA outside of the Office of
25 Research and Development, and it's critically
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202} 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
1 important for us to get together as a group to see
2 each other's faces and get to talk about issues,
3 particularly pressing issues, relevant issues such as
4 this.
5 The second major reason this is important
6 is that neurotoxicity is a very, very critical
7 scientific issue. It's received more and more focus
8 over the last few decades with the realization that
9 even small amounts of neurotoxins, particularly during
10 sensitive periods in infancy and childhood and fetal
11 life, can result in profoundly deleterious effects.
12 It's very important that we get risk
13 assessment for neurotoxins right and that we keep up
14 with the state of the science. So the relevancy of
15 this topic is really, really perfect.
16 The other -- third reason that it's
17 important is that the adjectives that were placed on
18 the title, the analysis of mixtures, the analysis of
19 cumulative risks, aggregate risks, these are all areas
20 that, I think we would all agree, are very important,
21 that we don't know enough about. We don't have enough
22 ; focus.
23 We know that multiple simultaneous
24 exposures to neurotoxins clearly have different
25 €;ffects than exposure to those chemicals alone. Yet
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
1 we don't have good methods to do risk assessment in
2 these areas.
3 Under the Food Quality Protection Act
4 mandate that OPPTS received, we have really gone
5 farther than anybody else in the country and anybody
6 else in the world on developing these kind of risk
7 assessment methods and guidelines. So it's really
8 important that we talk about this and that everybody
9 is aware of what's going on,
10 We are really on the cutting edge in this
11 area. So it's, again, a perfect reason to be here.
12 So I look forward to hearing at least some
13 of the day today, and I wish all of you a good,
14 thoughtful learning experience, and I'm really happy
15 to be here and happy that we are doing it. Thanks.
16 DR. NORRIS: Thank you, Dr. Galson, and
17 thank you all for being here today. Welcome and good
18 morning.
19 First, I must tell you that today at our
20 symposium on the neurotoxicity of chemical mixtures,
21 we're not here in any way to review or establish EPA
22 policy. We are here rather for the pleasure and
23 opportunity of scientific exchange, which means that
24 we can discuss scientific concepts, ideas, and data
25 for the sheer joy of it, and it is a great joy to me
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON. D.C. 20005 (202) 234-4433
-------
1 to do that.
2 As a self-proclaimed expert on the brain,
3 I was talking yesterday with a self-proclaimed expert
4 on the human mind, Dr. Tara Brock. Dr. Tara Brock
5 told me that the mind has about 80,000 thoughts a day.
6 Knowing the brain and how it functions, I added
7 another thought to my cumulative total for the day,
8 and that was that the brain probably processes even
9 more than that each day.
10 Then Dr. Brock added that 98 percent of
11 these thoughts we have each day are ones we've
12 probably already had. I hope to change those
13 statistics today. I think that we will greatly
14 broaden our concept of the science of neurotoxicology
15 and the neurotoxicology of mixtures.
16 Today we have the opportunity to hear from
17 some of the world's experts in neurotoxicology and in
18 the regulatory guidelines that we use and consider in
19 evaluating toxicity. I have the pleasure of
20 introducing to you this morning's speakers, and I will
21 introduce them all to you now and then briefly one at
22 a time as they speak:
23 Dr. John O'Donoghue who is Director of the
24 health and Environmental Laboratories for Eastman
25 Kodak Company, and I know him as the editor of several
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON. D.C. 20005 (202) 234-4433
-------
8
l volumes on the neurotoxicity of industrial chemicals
2 which I use as references frequently.
3 Dr. Ginger Moser, who is a
4 neurotoxicologist from the Neurotoxicology Division of
5 EPA's laboratories in RTF. Ginger has received EPA's
6 highest scientific achievement award for her work on
7 EPA test methods and guidelines.
8 Dr. Stephanie Padilla is Chief of the
9 Cellular and Molecular Toxicology Branch, and she is
10 also from the Neurotoxicology Division of EPA's
11 laboratories at RTP.
12 Dr. Will Boyes -- have you arrived yet?
13 No. The last I heard, Dr. Boyes' plane had landed in
14 Baltimore, and due to the weather or due to some
15 unforeseen event. So I've asked -- If Dr. Boyes
16 doesn't arrive, in an effort to stay on schedule, I've
17 asked Dr. Abou-Donia if he would come forward to this
18 morning's talks, in case that happens.
19 Dr. Abou-Donia is Professor of
20 Pharmacology and Cancer and Neurobiology at the Duke
21 University Medical Center.
22 Dr. Greg Christoph, who I'm so happy
23 you're here with us today, because he is retiring
24 tomorrow as head of Neurotoxicology at DuPont Company.
25 So we are very lucky and delighted to have
NEAL R. GROSS
I- COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
these speakers with us this morning. Thank you, and
I will let Dr. O'Donoghue begin our program.
DR. 0'DONOGHUE: Good morning. I'm very
happy to be here. I have a little bit of trepidation,
because every time I've been moving my digital slides
between PCs, things have changed, and nothing has
migrated since the last change.
As Debbie said, I'm John O'Donoghue with
Eastman Kodak. What I'd like to talk to you about
this morning is how we integrate neurotoxicology into
our pollution prevention program.
My company is not in the pesticide
business. So we are not really going to talk very
much about OPs or pesticides, but rather some
relatively common and mundane materials. Our focus is
oftentimes, when we find a material that has
neurotoxic properties, is to find what the limits of
the effect are. In many cases, it's to eliminate it
from the supply chain.
So in many cases, we are not trying to do
a quantitative assessment. We are trying to think
through the process of identifying materials,
identifying safe levels, and elimination through a
pollution prevention project.
Now the types of materials I want to talk
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
10
1 about are constituents of relatively common materials,
2 paints, adhesives, inks and greases. We use a lot of
3 different mixtures in our everyday activity, and what
4 we are concerned about is how these materials might
5 interact.
6 This is just an overhead of what we would
7 consider the manufacturing process to be. So within
8 the box you actually have the manufacturing process.
9 Then what I've done highlighted in gold are those
10 aspects of the process where we are looking at
11 assessing chemicals.
12 So the process isn't just to look at
13 products or co-products. We are looking at raw
14 materials that come in. We are looking at the
15 manufacturing process, whether heat is supplied to
16 those materials, whether they become aerosolized, how
17 they are used, and if the product is to be reused or
18 recycled, is there anything in the product that is
19 going to be potential during the recycling process.
20 Then we look at what are our potential waste due to
21 air emissions, water emissions or other types of
22 releases.
23 The main thing we don't look at is on the
24 left, which is kind of earth, wind and fire, which is
25 going back to actually the mines and looking at some
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
11
1 of the raw materials. So we are looking at lots of
2 different types of materials used in lots of different
3 ways.
4 Now the process we use is a fairly
5 straightforward process, and it actually has a lot of
6 similarities to the process that's used for assessing
7 materials under TSCA. One of the things we look at
8 very early on is analogous materials. Do we have
9 chemicals in our historical toxicology file that can
10 tell us something about a new material or how a
11 material might interact with the mixture.
12 We look at exposure, production, and how
13 people are going to use the materials. Based on that,
14 we do a preliminary chemical evaluation. We decide at
15 that point if there is data that indicates there is
16 interaction or there is neurotoxicity. We may
17 eliminate the material at that point.
18 If not, then the material moves on to what
19 we call a testing strategy. We decide what types of
20 tests we might run on those materials, conduct the
21 test, evaluate it, and then cycle back through.
22 We also have programs to look at what I've
23 put down as process control and health assessments.
24 We have industrial hygiene data available to us. We
25 have employee medical information, not for the
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
12
1 individual employees but for groups of employees, and
2 we also operate an 800 number for customers or
3 consumers, if they have a question.
4 So this is part of the process. If we
5 were to detect something post-marketplace, we have a
6 process then to go back in and reevaluate the
7 material.
8 Now one of the interesting -- or latest
9 aspects of this is we've been working with Bill Law
10 from your office to integrate the pollution prevention
11 framework into our assessment process. This doesn't
12 have specific aspects of it that are related to
13 neurotoxicity, but there are some general pieces of
14 information that are in the pollution prevention
15 software that we integrate into our process.
16 So, for example, we are able to estimate
17 chemical and physical properties of the materials
18 based on structure activity relationships. We look at
19 housing and the environment, and how was this material
20 apt to be transformed, and is it going to be
21 transformed into something which is a neurotoxin,- and
22 we look particularly at cancer hazard potential. Is
23 there something about the structure of this chemical
24 that would have some alerts from the point of view of
25 producing cancer in the nervous system?
NEAL R. GROSS
(202) 234-4433
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
WASHINGTON. D.C. 20005
(202) 234-4433
-------
13
l So this has become part of our overall
2 assessment process. Now the types of effects that
3 we're talking about fall in a variety of different
4 categories. Chemicals can have a number of potential
5 biological effects. Some of them are relatively
6 direct in that they can cause irritation to sensory
7 organs, irritation to the eyes, to the nose.
8 There can be some general C&S toxicity due
9 to effects on, say, the liver or the kidney. Then
10 there are some specific effects that chemicals have on
11 the nervous system.
12 When we're talking about the mixtures, I'm
13 talking primarily about pharmacologic and neuro-
14 degenerative effects that chemicals or chemical
15 mixtures can have, not the indirect effects through
16 C&S depression or irritation.
17 There are also potentially effects that
18 are neuro-oncologic effects that can be caused by
19 chemicals. These are relatively rare, but it is
20 something that has to be considered in assessing a new
21 chemical.
22 Now the possible relationships that
23 materials have to each other with regard to
24 neurotoxicity are not that different from other organ
25 systems, in that we are concerned about if you have
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
14
l two materials that have the same effect, they both
2 produce peripheral neuropathy, if there is a combined
3 exposure, is that effect additive?
4 Are there synergistic effects in that two
5 chemicals will have the same effect, but when a co-
6 exposure occurs, the effect is much greater than what
7 the additivity effect would be.
8 Potentiation is a situation where we have
9 a mixture which contains a neurotoxin and other
10 materials which are not neurotoxic, and whether or not
11 those materials enhance the neurotoxicity of the main
12 agent.
13 Then inhibition: Although we see
14 inhibition less commonly than some of the other
15 effects, we do see materials that, in fact, inhibit
16 neurotoxicity.
17 Now what I've done on this slide is put
18 together some factors that affect how chemicals
19 interact in a mixture situation that we have to be
20 concerned about. The first is the rate and extent of
21 some of the pharmacokinetic properties of the
22 material.
23 For example, absorption: If you've got a
24 mixture and you've got two neurotoxic materials, if
25 they are absorbed at different rates, in fact, the
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
15
effects may not impact on each other. They may not be
additive or synergistic.
For example, hydroquinone, which is a
common black and white developer, at high doses has
effects on the nervous system that cause tremor and
convulsions if the dose is high enough. However, the
material is very water soluble. It's absorbed very
rapidly. The effects are produced within about 15
minutes, and the material is excreted with a half-time
of about 17 minutes.
So if you have a mixture that has two
materials, one is absorbed rapidly, metabolized and
excreted, and the other material is not absorbed as
quickly, what may happen is, yes, the effects on paper
could be additive, but in fact they are not, because
of the pharmacokinetic parameters.
That's also true of metabolism, the rates
of metabolism. It could very well be that two
materials have exactly the same effect, but they
produce that effect through a mechanism which requires
metabolism to a common metabolite, and if the extent
or the rate of that metabolism isn't similar with the
two materials, there may not be an interaction.
Another factor which has an impact is the
exposure route. We have seen materials like 3-
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
{202! 234-1433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
16
1 heptanone. 3-heptanone is a material which is present
2 in many foods as a flavoring agent. It's in perfumes,
3 and we've looked at 3-heptanone from the point of view
4 of trying to produce neurotoxicity with it, because if
5 you look at the structure, it ought to be metabolized
6 to a common neurotoxic gamma diketone.
7 In fact, with 3-heptanone what you find is
8 that, by inhalation, you can't get enough material
9 into the body to produce the gamma diketone which
10 would result in a neurotoxicity. However, if you
11 gavage the animals, you can produce the effect. The
12 reason we think this is the case is that, by giving a
13 very large bolus dose, we're increasing the likelihood
14 of a first pass metabolic effect in the liver that
15 results in neurotoxicity.
16 So if you are assessing the material, and
17 the data you have is oral, it may not be relevant to
18 the inhalation situation or vice versa. You may find
19 that, if you test it by inhalation, you see no
20 neurotoxicity. You test it by the oral route, and you
21 do.
22 Mechanisms of action: Common mechanisms
23 of action are obviously situations in which additivity
24 is likely to occur. I put down -- The last issue is
25 the blood-brain barrier, because there are a couple of
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
17
l aspects of this which are important.
2 One is that some materials can damage the
3 blood-brain barrier. So you may have a material
4 which, when tested alone is not neurotoxic, but if
5 tested with a material which somehow or other alters
6 the blood-brain barrier, it would allow that material
7 to get into the brain and, therefore, produce
8 neurotoxicity.
9 Now there are also some slight differences
10 in various species of animals that tell us that there
11 are conditions where the blood-brain barrier is a
12 limiting factor in neurotoxicity; for example,
13 iramectin.
14 Iramectin is a parasiticide, and it's been
15 --It can be used in most strains of dogs without
16 producing any neurotoxicity, but in some of the
17 smaller collie breeds the blood-brain barrier is not
18 as effective as other brains of dogs. What you end up
19 seeing is neurotoxicity in those dogs.
20 It's the same material with the different
21 breeds, but in fact, because there's a defect in the
22 blood-brain barrier in one breed versus the other, you
23 are seeing the effect. So there is a concern about
24 whether the chemicals would alter the blood-brain
25
barrier.
(202) 234-4433
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
WASHINGTON, D.C. 20005
(202) 234-4433
-------
18
1 Now what I've done is I have put together
2 some scenarios for materials that we've seen and had
3 some concern about how they might interact with regard
4 to a mixture. So the first one, I think, is a
5 relatively simple scenario.
6 That is that we have chemicals which are
7 similar. They have the same metabolites, and they
8 have the same molecular target. So we would expect
9 here that the effects that we would see would be of an
10 additive nature.
11 An example of these materials is shown on
12 this slide. Now on the upper lefthand corner is a
13 chemical diagram of hexane. In the upper righthand
14 corner there is a diagram of methyl N-butyl ketone.
15 These materials are similar except for the carbonyl.
16 In the body both of these materials are
17 metabolized to common metabolites. In fact, N-hexane
18 can be metabolized to methyl N-butyl ketone. So that
19 we end up with a common metabolite, 2,5-hexanedione;
20 and a 2,5-hexanedione itself is the most neurotoxic of
21 these metabolites.
22 Now we have tested these materials in rats
23 and looked at them from a comparative point of view.
24 On this slide what I have is the first line shows the
25 methyl N-butyl ketone tested at 6.6 millimolar.
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
(202) 234-4433 WASHINGTON. D.C. 20005 (202) 234-M33
-------
19
1 Now in comparing these materials, we use
2 millimolar doses as compared to milligram per kilogram
3 doses, because the body -- These chemical reactions
4 occur on a molar basis, not on a milligram per
5 kilogram basis. So doing the comparisons, it makes a
6 lot more sense to use a millimolar basis for this.
7 What we have here is rats with neuropathy.
8 That is, they are showing some clinical sign, some
9 sort of weakness or some sort of sensory change or
10 they have some histologic evidence in their peripheral
11 or central nervous system.
12 Then we listed the number of days it took
13 for those animals to reach the endpoint, and the
14 endpoint in this case was that at least one of the
15 legs of the animals was paralyzed by the material. So
16 it's a pretty severe endpoint, but it's one that
17 doesn't change very much.
18 What you find with these materials is
19 that, as the animals are dosed, the clinical state
20 varies slightly from day to day. So it's hard to
21 pinpoint a point in time when you can say the effect
22 is quite similar. By choosing this endpoint, which is
23 stable, we can do a better comparison.
24 Then we've calculated what I refer to here
25 as a neurotoxic index. This neurotoxic index is based
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON. D.C. 20005 (202) 234^433
-------
20
1 on the number of days that it took for the MNBk
2 animals to be paralyzed. So because it took 55.8
3 days, any group that had its endpoint at that same
4 time would have a neurotoxic index of one. Any group
5 that took less time would have an index of greater
6 than one or be more neurotoxic. So it's a basis for
7 comparing the relative effects here.
8 Now with these two materials, you might
9 expect that the effect would be additive, but we also
10 dosed these animals with an equi-molar dose of Ki-
ll hexane, and what you find out with N-hexane is you see
12 no neurotoxicity with this material.
13 Now these dose levels correspond on a
14 milligram per kilogram basis to somewhere around 600-
15 700 milligrams per kilogram. So the N-hexane dose
16 here is quite a substantial dose. It's not a minor
17 dose but, in fact, there is -- Even though the
18 metabolites are the same and you're giving the same
19 molar amount of the material, the effect you're going
20 to get here is not additive.
21 Now we doubled the dose of N-hexane to
22 13.2 millimolar, and that wasn't effective. Then we
23 i raised it up to 46.2, and 46.2 is as much material as
24 we could get into the animal. That was the physical
25 limitation here. It's about four grams per kilogram
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-M33 WASHINGTON, D.C. 20005 (202) 234-4433
-------
21
1 of the material.
2 when you give that dose, you do get
3 clinical effects. Here we got three of the five
4 animals had effects, and four of the five had
5 histologic effects; but it took almost twice as much
6 time as with MNBk. So the effects here -- You have
7 the same metabolites. The effects are not additive.
8 We also looked at a material called
9 practical grade hexane. Practice grade hexane is
10 actually what most people are familiar with. This is
11 actually a mixture of hexanes. It's about 40 percent
12 N-hexane. It has cyclohexane in it, methylpentane in
13 it, and a variety of other materials.
14 Our original concern with this mixture is
15 we were concerned that this mixture would, in fact,
16 increase metabolism of N-hexane and be more
17 neurotoxic. In fact, what we found was that the
18 neurotoxicity was no more than with hexane alone. So
19 these mixtures of hexanes didn't have any interaction.
20 Now there is another way to look at these
21 materials. Hexane, MNBk, 2-hexanol and 2,5-
22 hexanedione are all commercial -- have been commercial
23 materials. N-butyl ketone now, there's a significant
24 new use rule on, and 2,5-hexanedione itself has had
25 relatively minor use. But in fact, these materials
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON. D.C. 20005 (202) 234^433
-------
22
1 could be used in combinations in some products.
2 So we looked at how we would go about
3 trying to compare these materials and how to figure
4 out on a relatively simple basis how to judge the
5 relative activity of these materials if they are
6 combined.
7 Now this data is very similar to the data
8 that I showed you before on the neurotoxic index, in
9 that for each of these materials we gave them the same
10 millimolar dose, and we then looked at the onset of
11 neurotoxicity. We divided that by the time it took
12 for the MNBk animals.
13 So MNBk here has a neurotoxic index of
14 three. N-hexane, because it took a higher dose and a
15 much longer period of time to create neurotoxicity,
16 has a neurotoxic index of -- I think it was less than
17 .1. Then 2, 5-hexanedione, which is the last material
18 at the top of the screen, it took a third less time
19 for the animals to develop neurotoxicity. So, in
20 fact, it was three times more neurotoxic than
21 hexanedione.
22 The other part of this relationship is the
23 serum concentration of 2,5-hexanedione, because you
24 can see that there is a relationship between how the
25 neurotoxic index in the peak concentration -- the more
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON. O.C. 20005 (202) 234-4433
-------
23
1 2,5-hexanedione was formed by these materials, the
2 more neurotoxic it was.
3 So this allowed us a basis for actually
4 combining these exposures and for looking at exposures
5 to other materials, because we now knew that the
6 effect was based on the serum 2,5-hexanedione level.
7 We could measure that in blood. We could measure it
8 in serum, and we could measure it in urine.
9 Therefore, it became a convenient
10 biomarker of exposure, and also a convenient biomarker
11 of effect, because you can actually quantitate at what
12 level in the animals you will see neurotoxicity based
13 on a 2,5-hexanedione level.
14 This parameter is one that's recommended
15 now by ACGIH as a biological index for exposure to
16 hexane and related materials. So it is possible to
17 calculate how these similar materials are going to add
18 together, but they don't add together in a simple
19 fashion. It's not a matter of adding one mole of
20 hexane to one mole of methyl N-butyl ketone and coming
21 up with two. It's a matter of understanding what the
22 metabolic pathways are and how they can be combined
23 based on that common pathway.
24 Now the second scenario I put together is
25 one in which there are similar classes of chemicals,
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
(202; 234-4433 WASHINGTON. D.C. 20005 (202) 234-4433
-------
24
1 and in this particular case there are ketones and
2 other similar alkane solvents that have different
3 metabolites. They are not metabolized in the same
4 way, but there are metabolic interactions that occur
5 with the mixture.
6 This is a mixture that was proposed for
7 use in paints and for plastics. This mixture was sent
8 to us for evaluation, because there was a low level of
9 MNBk in it. It's less than one percent MNBk. Our
10 first thought with this was, well, that's probably not
11 going to be enough to cause the solvent to be
12 rieurotoxic. There is also some 2-heptanone in it,
13 which was a little bit of concern, because there are,
14 at least in theory, ways in which 2-heptanone could
15 form a gamma diketone.
16 So our initial read on this mixture was
17 this probably isn't a problem, but we did some
18 screening level tests on it and found that one of the
19 components, 5-nonanone which was in there at 12
20 percent, when we actually tested the mixture, we found
21 that 5-nonanone was metabolized to methyl N-butyl
22 ketone.
23 I haven't shown you the whole pathway
24 here. The methyl N-butyl ketone then is metabolized
25 to 2,5-hexanedione. So the potential pathway here
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
25
1 between this and MNBk is very similar. The
2 biochemistry is very similar except in this particular
3 case we're getting decarboxylation, which makes the
4 material a lot more neurotoxic than you might expect
5 otherwise.
6 So before we had discovered this, we
7 started off looking at the components of the mixture.
8 This table shows again the chemicals that we looked at
9 and the mixtures on the left, the day of onset of
10 neuropathy, the number of days it took before we saw
11 any evidence of clinical neuropathy, the most severe
12 clinical neuropathy, and a +3 here is the animal is
13 dragging one feet. A +1 is that the animal doesn't
14 completely extend the foot, and a +2 indicates that
15 the animal is having some problems with placing the
16 foot. They tend to trip or misplace the foot, but
17 they are able to walk.
18 Then the next column is the number of
19 animals with clinical neuropathy at the end of 90
20 days, and the last column is the number of animals
21 that had histologic changes at the end of 90 days.
22 So with the original material, this
23 commercial grade methylheptylketone --it was given at
24 2,000 milligrams per kilogram -- we got a neuropathy
25 at about 60 days. It was severe, and all the animals
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-M33 WASHINGTON, D.C. 20005 (202) 234-4433
-------
26
1 eventually had clinical signs or histopathology.
2 Now we also looked at the main ingredient
3 of this material, which was 5-methyl-2-octanone, and
4 we tested it at 2,000 milligrams per kilogram. In
5 fact, we saw no effect with this material from a
6 neurotoxicity point of view.
7 The other material that we looked at here
8 was 5-nonanone. It's shown here as N-O-N, and the
9 high dose that we used was 2,000 milligrams per
10 kilogram, and we got neuropathy in 11 days. It was
11 severe, and in this case the screening level animals
12 all had it. They all had -- Both of them had clinical
13 signs, and both of them had neurotoxic effects.
14 Now we dropped the dose down, and we got
15 a similar effect. We dropped the dose down to 233
16 milligrams per kilogram. This equates to the amount
17 of 5-nonanone that was in the original mixture. So the
18 12 percent 5-nonanone in the commercial
19 methylheptylketone equates to 233 milligrams per
20 kilogram, and here we saw no neuropathy in the
21 animals, even though when we looked at them
22 histologically there was a very slight neuropathy.
23 So there wasn't enough 5-nonanone in this
24 mixture to account for the neurotoxicity that we were
25 seeing. There wasn't enough MNBk in the mixture to
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
(202) 234-4433 WASHINGTON. D.C. 20005 (202) 234-4433
-------
27
1 account for that or enough of the 2-heptanone.
2 So what we did is we went back, and we
3 reconstituted the mixture with the two materials we
4 had the greatest concern about, and that was 5-methyl-
5 2-octanone and 5-nonanone. So we recreated the
6 mixture and dosed the animals at 2,000 milligrams per
7 kilogram.
8 In fact, what we saw was that we produced
9 a neuropathy that was similar to the commercial
10 mixture. All the animals had histologic changes, but
11 they didn't all have clinical signs. So by combining
12 the two materials, we got an effect that was similar
13 to the mixture, but wasn't even quite as severe as the
14 mixture. It was slightly less.
15 So there are other components in this
16 mixture that are probably contributing to this.
17 Probably the MNBk and the 2-heptanone are adding --
18 have an additive effect, but the effect that we're
19 seeing here is a promotion of the metabolism of 5-
20 nonanone to its ultimate neurotoxic endpoint, which is
21 again a diketone.
22 Now because 5-methyl-2-octanone would do
23 this, we were concerned about other mixtures. Having
24 studied other mixtures of ketones, we were aware that
25 methylethylketone, which is a very common solvent --
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
28
1 methylethylketone will potentiate the neurotoxicity of
2 methyl N-butyl ketone. It enhances the production at
3 2.5 hexanedione.
4 So our thought here was, well, gee, would
5 methylethylketone potentiate the effects of 5-
6 nonanone. In fact, what we found, it didn't. We
7 dosed these animals in a comparable way to the animals
8 that got 5-methyl-2-octanone, and it didn't.
9 The reason we believe this is the case is
10 because what the 5-methyl-2-octanone is doing is not
11 promoting the metabolism of MNBk, but it's promoting
12 the metabolism of 5-nonanone to MNBk. It's affecting
13 the decarboxylation reaction but not the oxidation of
14 2, 5-hexanedione; and because MEK doesn't do that, you
15 don't see the potentiation in this particular case.
16 Now this is a summary of a number of the
17 mixture studies that we've done. We've looked at MEK,
18 and it potentiates N-hexane. It potentiates methyl N-
19 butyl ketone. It does not potentiate 5-nonanone.
20 We've looked at 5-methyl-2-octanone which
21 potentiates the effect of 5-nonanone, but it doesn't
22 potentiate the effects of another material which
23 produces a gamma diketone, which is ethyl N-
24 butylketone.
25 Others have looked at the effects of
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
(202} 234-W33 WASHINGTON. D.C. 20005 (202) 234-4433
-------
29
1 toluene, and toluene, in fact, inhibits the
2 neurotoxicity of N-hexane. This is an interesting
3 inhibition, because I'm sure Dr. Boyes is going to
4 talk about toluene. Toluene itself is neurotoxic. It
5 produces auditory problems.
6 Here you have two neurotoxic materials,
7 one that produces an auditory change, N-hexane which
8 produces a peripheral neuropathy. You give them
9 together, and you don't get the peripheral neuropathy.
10 So you might think that, because they are neurotoxic,
11 they would have some kind of interaction. In fact,
12 they do, but it's the opposite of what you would
13 actually expect.
14 Now this scenario I put together involves
15 significantly different chemicals. On a molecular
16 basis, the chemicals are very different. They have
17 different metabolites. They have a different
18 molecular target. They don't operate by exactly the
19 same mechanism. Yet they produce the same effect in
20 the animal.
21 I'm sure later in the day people are going
22 to talk about organophosphates . I put this cartoon up
23 here in case you're not familiar with the
24 neuromuscular junction. This is a cartoon of the
25 neuromuscular junction.
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
30
1 This is the presynaptic area. This is the
2 synaptic cleft and the postsynaptic area. Now what
3 happens here is the action potential comes down. You
4 get calcium influx into the neuromuscular junction.
5 You get fusion of the synaptic vesicles with the
6 presynaptic membrane and release of acetylcholine into
7 the presynaptic cleft, and then you get binding on the
8 postsynaptic membrane, and you get an action potential
9 in the muscle. The muscle contracts.
10 Now what causes this to reverse is
11 acetylcholinerase, which breaks down the acetylcholine
12 and reduces its effect on the postsynaptic membrane.
13 Now these membranes then are -- These vesicles then
14 are recycled back to the cell body.
15 Now with this next material, this is 4-
16 nitropyridine-n-oxide. It was an R&D material that
17 people were pretty excited about, and we were
18 concerned about it, because pyridines can have
19 neurotoxic effects.
20 So we did some screening level studies
21 with this material, and this is the acute toxicity
22 data that we got from this material, which caused us
23 to be very concerned. You see the dermal LD£0? The
24*! dermal LDSO is a gram per kilogram. The oral LD50 is
25 50-100.
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-M33 WASHINGTON, D.C. 20005 (202) 234-4433
-------
31
l Now if you put it in the eye, and the eye
2 is unwashed, the LDSO here is greater than 10
3 milligrams per kilogram. You, in fact, see clinical
4 effects in the animals, but none of the animals die.
5 The effects that you start to see are muscle
6 twitching, and then you see massive vesiculation in
7 the skeletal muscle, seizures, and then if the animals
8 get a high enough dose of this material, they stop
9 breathing and die.
10 What you're seeing is an acute cholinergic
11 effect in these animals due to application to the eye.
12 Now if you put this material in the eye and you wash
13 it out, it has an LDSO, we say, between one and ten.
14 We've never actually been able to calculate precisely
15 what it is, but we've had animals die with one
16 milligram per kilogram of this material in the eye.
17 We think what happens is that, when you go
18 to wash it out, what you in fact do is solubilize it,
19 and it's very rapidly absorbed through the eye and
20 into the CNS.
21 Now this material, which is a test
22 I material on the left, is very similar to 4-
23 aminopyridine. These pyradines are used
24 pharmacologically to enhance the release of
25 acetylcholine. People that have Eaton-Lambert
NEAL R. GROSS
(202) 234-4433
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
WASHINGTON, D.C. 20005
(202) 234-4433
-------
32
1 syndrome, for example, have very weak muscle activity,
2 and one way in which to enhance the activity is to
3 provide them 4-aminopyridine.
4 4-aminopyridine affects the potassium pore
5 in the membrane. What happens is the action
6 potential, when it comes down to the endplate, is
7 prolonged. So you get a long action potential, and
8 the prolongation of the action potential allows
9 release of significantly more acetylcholine.
10 So what you end up with is a cholinergic
11 crisis. Now our concern with these materials, if they
12 • are in a mixture with a material which is also a
13 cholinesterase inhibitor, you would have two materials
14 that produce a cholinergic crisis by two different
15 mechanisms. In one case it enhances the action
16 potential. It releases more acetylcholine. In the
17 other case, it inhibits the breakdown of that
18 acetylcholine.
19 So you get an enhancement of the effect.
20 This enhancement occurs by two different -- at two
21 different molecular targets. The effect in the animal
i
22 ! is similar, and the effects are additive, even though
23 the mechanism of action isn't additive.
24 Now the fourth scenario I put down here
25 are we have significantly different chemicals. They
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-*433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
33
1 have different metabolites, but they have the same
2 molecular target.
3 This cartoon is to help you understand
4 some of the neurobiology associated with materials
5 that we're concerned with here. The axon is the
6 mechanism -- the pathway by which the cell body
7 communicates with the periphery, whatever that
8 periphery happens to be, whether it happens to a
9 neuromuscular junction or happens to be a sensory
10 endpoint.
11 There is movement of some relatively large
12 structures. The organelles that are formed in the
13 cell body are transported to the neuromuscular
14 junction. So the synaptic vesicles, for example, are
15 moved from the cell body to the endplate and back
16 again.
17 This cellular movement is very important
18 to the life of that axon. If there is anything that
19 interferes with that intracellular movement, you, in
20 fact, get some form of axonopathy. The materials that
21 we have been concerned with have been materials such
22 as acrylamide, such as MNBk, such as N-hexane.
23 We've looked at the mechanism of action of
24 these materials. This schematic is a schematic I put
25 together about how acrylamide, in fact, can result in
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202)234^433 WASHINGTON. D.C. 20005 (202)234-4433
-------
34
1 a peripheral neuropathy. In this particular case, the
2 acrylamide is absorbed and distributed.
3 We know it undergoes hepatic metabolism.
4 We're not totally sure whether the metabolites in
5 acrylamide itself can have a neurotoxic effect, but we
6 know that from binding acrylamide to various targets
7 in the axon you get interference with axonal
8 transport. Then you get some degree of axonal
9 degradation.
10 Now depending on the dose level, you
11 actually also get axonal regeneration. So there is a
12 balance here between very low levels of exposure axons
13 being damaged and axons regenerating. If, in fact,
14 the damage occurs at a faster rate than regeneration,
15 you get clinical deficits.
16 Now what we've been concerned about is
17 that these materials all have some very common
18 similarities in that they all have effects on
19 glycolytic enzymes. They have effects on glycolytic
20 enzymes which has an impact on energy metabolism in
21 the axon and can affect axonal transport.
22 They have effects on the microtrabeculae
23 system in that they tend to stabilize the axonal
24 proteins, and the stabilization can result in reduced
25 axonal flow, and that will slow down transport.
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
35
1 They can also interact with the active
2 transport proteins, and they can interfere with
3 movement, having an impact on retrograde flow, and in
4 the end they all result in this dying back neuropathy.
5 They cause a neuropathy by impacting axonal flow.
6 So our concern has been that there is a
7 common mechanism of action. Even though these
8 chemicals are very different, there is no reason that
9 we see that these materials should not interact and
10 cause peripheral neuropathy at -- I won't say low
11 doses, but our concern has been that, if we give
12 subneurotoxic doses of some of these materials, are
13 those effects additive?
14 From what we have seen, we think they
15 probably could be additive, although for the most
16 part, we have not had a reason to actually create such
17 mixtures.
18 So in summary, there are interactions
19 among chemicals that are affecting neurotoxicity.
20 These interactions are modulated by pharmacokinetic
21 parameters, absorption, metabolism excretion. There
22 are metabolic interactions that occur outside of the
23 nervous system -- we see them primarily in the liver -
24 - that affect neurotoxicity.
25 There are also nervous system specific
NEAL R. GROSS
(202) 234-4433
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
WASHINGTON, D.C. 20005
(202) 234-4433
-------
36
1 vulnerabilities, either with the blood-brain barrier
2 or with the transport system in the nervous system,
3 that offer unique opportunities for interactions of
4 chemicals.
We see mechanisms of action that are
6 common for a number of neurotoxic chemicals, and thus
7 offer the opportunity for an additivity effect.
8 So in summary we can increase or decrease
9 the likelihood of observing neurotoxicity with a
10 particular mixture through any of these various
11 parameters. Thank you. Do you want to take questions
12 now, Debbie, or later? Questions?
13 (Applause.)
14 (INAUDIBLE QUESTION)
15 DR. O'DONOGHUE: The way we've determined
16 additive effects is to go back in and reconstitute the
17 mixture at various concentrations and try to see if we
18 can reproduce that additive effect with the mixture.
19 That's what we tried to do with some of the ketone
20 solvents.
21 (INAUDIBLE QUESTION)
22 MR. O'DONOGHUE: No. We tried to -- He
23 was asking how we went about doing the additive
24 studies. What we actually do is try to understand
25 what the chemical mixture is and then recreate that
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON. D.C. 20005 (202) 234-4433
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
37
mixture and then retest the mixture . We are not doing
it by a modeling program.
Does that answer your question?
I think, actually, when you talk about
cholinesterase inhibition later on, we'll probably
talk more about how those things are actually modeled.
The reason we don't do a lot of that is because when
we find one of these materials, our main objective is,
frankly, to get rid of it. It's not to figure out a
way to use it .
We are not intentionally using neurotoxic
materials. We're trying to understand what the limits
of their safe use are and, in many cases, get rid of
the materials, if possible. So our goal is quite a
bit different.
DR. MOSER: I wanted to thank Debbie and
the organizers for inviting me to be here today. I
appreciate this opportunity.
I'm going to switch gears a little bit
from what John was covering, and I'm going to talk
more specifically about one study and also get into
some of the more statistical analysis on that study.
As we all know, and the reason that we are
all here, I think, is that most environmental chemical
exposures do occur in multiple. So we are not exposed
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
38
l to just single chemicals. So a lot of the information
2 that we have about single chemicals may not be as
3 appropriate to understanding mixtures.
4 For now, there is just simply no way of
5 predicting when or what kind of interaction may be
6 produced when there is exposure to multiple chemicals,
7 and trying to model these interactions in the
8 laboratory requires a lot of different types of
9 considerations, and also the statistical analysis of
10 these interactions becomes very difficult very
11 quickly.
12 What I want to present here today is the
13 data from a large, rather a huge study that we've
14 conducted quite a few years back now. I want to try
15 to present some data that you may find interesting on
16 the mixture, and I also want to try to illustrate the
17 complexity of this type of study, both in terms of its
18 statistical analysis and the data interpretation.
19 In this particular study we wanted to go
20 past binary combinations, which are typically the way
21 that people look at chemical mixtures in the
22 laboratory. We wanted to go all the way from two to
23 three chemicals.
24 I The project was initiated as part of a
25 Superfund project. We wanted to study chemicals that
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON. D.C. 20005 (202) 234-4433
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
39
were frequently found in Superfund sites and were
found to co-occur.
We used a full factorial design to study
these interactions of these three chemicals, which
means we had three chemicals and five dose levels of
each chemical. One of those dose levels was zero, and
then there were four other dose levels.
We looked at all possible combinations.
so it was a five to the three study. Five to the
three gives you 125 treatment groups. In the study we
used ten rats for each treatment. So you can see why
I say it's such a huge study.
The study was actually conducted in
collaboration with some other people at EPA, and we
did neurotoxicological evaluations on the animals. In
the same animals we did general toxicity studies, and
we also at the end of the study took organ -- did
organ histopathology and clinical chemistries.
The study was run in conjunction with the
developmental toxicity study using the same chemicals
and similar dose levels. The results of the
developmental tox study have been published already,
but this is the debut for the neurotox data.
I'm going to briefly describe what
response surface modeling is. It's a method that aims
NEAL R. GROSS
(202) 234-4433
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
WASHINGTON, D.C. 20005
(202) 234-4433
-------
40
l to mathematically describe the surface of the
2 response, the response to the combination of
3 chemicals, all combinations of chemicals.
4 It uses a factorial design where chemicals
5 are present -- all the different chemicals are present
6 in different levels, dose levels. Some of the
7 advantages are that it does use all the experimental
8 data. There is no mathematically derived numbers like
9 ED50.
10 It does estimate the response surface over
11 a range of doses. It provides estimates for both the
12 individual chemical as well as the interactions
13 between the chemicals. Theoretically, there is no
14 limit to the number of chemicals that can be studied
15 this way, even though you can only picture -- you
16 know, draw a graph for combinations of two chemicals,
17 mathematically you could have an infinite number of
18 chemicals.
19 The other advantage to this type of
20 modeling that we did was that there would be data
21 available to take out subsets of the data, so that we
22 could analyze the subsets of data and see whether we
23 could predict the same result using a reduced dataset
24 compared to the full dataset.
25 Now as I say, we did the
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
41
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
neurotoxicological evaluations in these animals. We
used the functional observational battery or FOB,
which probably most of you all are familiar with.
It's a battery of tests that rapidly evaluates
neurological function of the animals.
We also used motor activity using an
automated assessment of locomotor activity. The FOB
is a neurological exam. It does evaluate
neuromuscular dysfunction, sensory deficits, autonomic
changes, and changes in activity levels and reactivity
or excitability of the animal.
The standardized series of open field and
manipulative tests that are well standardized are used
widely now to screen for hazard identification for
potential neurotoxicity. So these were the tests that
we used in this study.
Protocol here shows that we used Fischer-
344 female rats. The reason for this was, as I said,
we did this study in conjunction with the
developmental tox study. So, of course, we had to use
females. We used the Fischer-344, because that was
the standard strain that's used in the Chernov-Kavlok
assay.
We did neurobehavioral testing, as I've
already said. We tested the animals before dosing
(202) 234-4433
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
WASHINGTON. D.C. 20005
(202) 234-4433
-------
42
1 began, four hours after the first dose, 24 hours after
2 the first dose. We dosed the animals repeatedly for
3 ten days, and then we also tested them at four and 24
4 hours after that tenth dose.
5 Because of the number of treatment groups,
6 obviously, we couldn't test all the animals at one
7 time. We had to split up the study into replicates.
8 We had the statistician draw up the replicates for us
9 so that we had five replicates of 250 animals in each
10 one. Of course, we can't test 250 animals at a time
11 either. So we had to break down the replicates into
12 blocks so that essentially we tested 50 animals at a
13 time.
»
14 The chemicals that we chose to study are
15 listed here. The first one is heptachlor. It's a
16 cyclodiene pesticide that's very persistent in the
17 environment. It does act on the nervous system. It
18 blocks the GABA-ergic system. So it causes
19 excitation.
20 The next chemical is diethylhexylphthalate
21 or DEH. It's a liver toxicant, and it's not known to
22 act, on the nervous system at all.
23 v. Then the last one is trichloroethylene,
24 which is a volatile organic solvent. The mechanism
25 for trichloroethylene isn't exactly known, but it is
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C, 20005 (202) 234^433
-------
43
l known as a general type of CNS depressant.
2 i want to point out that only two of these
3 chemicals did act on the nervous system, and the
4 mechanisms of actions are very different for the three
5 chemicals.
6 The doses that we used are listed here.
7 For heptachlor -- or for all the compounds that we
8 tested, we chose the doses in order to span the range
9 of effectiveness. We wanted to be sure that the high
10 dose was effective and that the low dose had
11 essentially no effects, and we wanted to be able to
12 construct a good dose response curve with these four
13 doses.
14 One thing to point out is that with the
15 response surface analysis you don't need to have your
16 dose basing be any specific way. So you can see that
17 the dose basing was very different for the different
18 compounds. We could select our doses in order to
19 construct the best dose response curve. We were not
20 restrained in any way.
21 The other thing to point out is that the
22 components were mixed together into the dosing
23 solution, so that the rats only received one dose a
24 day instead of getting the three different doses. So
25 we had basically at any one time 50 different doses
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
44
1 being given in a day, and they were all dissolved in
2 corn oil, which was our vehicle.
3 - Now for the data analysis. This was
4 really the trickiest part of the whole study. The FOB
5 produces different kinds of endpoints. Some of the
6 data are binary or just simply yes/no data. Some of
7 them are continuous like body weight or activity
8 count, and some of them -- most of them, actually, are
9 ordinal. They are ranked scores.
10 There are published methodologies for
11 analyzing binary and continuous data using response
12 surface analysis. So far, even now, there is no
13 analysis that is developed for ordinal types of data.
14 So for those data we had to make transformations
15 before we could actually put the numbers into the
16 response surface analysis.
17 With the data analysis we get parameters
18 that will -- you can get parameters for an overall
19 test of interactions. So it will test whether or not
20 there is a significant deviation from additivity,
21 significant less than probability value .05.
22 You can also get individual parameters for
23 the effects of the chemicals by themselves. You can
24 look at the binary combinations of each three
25 chemicals. So you could see whether there is
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON. D.C. 20005 (202) 234-4433
-------
45
1 interactions between just two of the chemicals, and
2 then you can also look at the tertiary interaction or
3 the three-way interaction.
4 We could also analyze -- Because of the
5 way that we set up the replicates, we could also
6 analyze whether there were replicate effects or not.
7 We did do a reduced design -- a reduced analysis.
8 What we did was take the zero level and the data from
9 the next to the low dose and the high dose. So that
10 the same type of analysis was run using just two doses
11 of each compound instead of the four doses of each
12 compound.
13 We were hoping that, if this was
14 predictive of the full factorial design, this would be
15 a more efficient way to do the study.
16 Then because of the time restraints, we
17 had to limit the analysis to just one time point, and
18 the time that we chose was the four-hour data after
19 the tenth dose.
20 First, I'll go over the effects of the
21 individual compounds, what's listed here are the --
22 The measures of the functional observational battery
23 are listed here. As you can see, trichloroethylene,
24 being a CNS depressant, does tend to decrease or
25 depress most of the nervous system functions.
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
46
1 We get decreases in grip strength. We get
2 gait changes, decreases in activity levels. Kind of
3 interesting, we did get increases in alertness, but
4 basically decreases on most of the other
5 responsiveness tests like sensory motor responses.
6 We saw some general health changes with
7 piloerection and lacrimation, and we saw body weight
8 decreases and a small amount of lethality at the high
9 dose.
10 Heptachlor, as I've already said, inhibits
11 the inhibitory transmission. So it actually produces
12 increases in some of the nervous system functions. We
13 get increases in grip strength, gait changes again.
14 We do see decreases in locomotor activity with
15 heptachlor and these types of chemicals, but we get
16 increases in responsiveness to both being handled and
17 some of the sensory tests.
18 We also got lacrimation, some salivation,
19 decreases in body weight, and a small amount of
20 lethality. As I've already said, DEHP does not act on
21 the nervous system, and we got absolutely no effects
22 on these tests with that chemical by itself.
23 This is the results of the statistical
24 analysis. There is actually a lot on this slide. So
25 I may need to take a minute.
NEAL R. GROSS
(202) 234-4433
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
WASHINGTON. D.C. 20005
(202) 234-4433
-------
47
1 Again, these are the endpoints that we
2 tested. This first column is just whether or not
3 there was a significant interaction or significant
4 deviation for additivity. As you can see, for almost
5 all the endpoints there was a significant interaction.
6 In fact, just about every single one of them showed a
7 significant interaction.
8 Now what's listed here is the algebraic
9 sign of the parameters, where there was a significant
10 parameter. So that means that, if the parameter is
11 significantly different from additivity, the algebraic
12 sign is listed.
13 What that means is a positive sign shows
14 that the effect is increasing with dose. So you have
15 a positive going to this response curve, and a
16 negative sign is a decreasing dose response curve.
17 So for instance, with hind limb grip
18 strength we decreased grip strength with TCE by
19 itself, and we increased grip strength with heptachlor
20 by itself. Now in cases like gait score -- I want to
21 point this one out -- we did see gait changes with
22 both TCE and heptachlor. So we had a positive
23 parameter for both of those, but then the interaction,
24 the trichloroethylene by heptachlor interaction had a
25 negative parameter.
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
48
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
That means that the dose response curve is
either decreasing or else it's not increasing as much,
and that's interpreted as being antagonism. So
whenever the sign is opposite from the signs of the
individual parameters, that's antagonism; whereas, if
it was the same direction, then that would be
considered synergism, because the effect would be
greater than the two compounds alone.
As you can see, for most of these
characterizations we did have antagonism.
Interestingly, with lethality we did see significant
effects of both individual compounds. When you looked
at trichloroethylene and heptachlor together, we had
antagonism.
We also had antagonism with the DEHP and
the heptachlor as a binary combination. Remember that
DEHP had no effect on its own. But then if you look
at the three-way interaction, the three chemicals
together actually produce synergism.
So this was kind of the worst case
scenario where for most of the neurological endpoints
we got antagonism, and then for lethality, which is
probably the most important toxicity, we got
synergism.
(202) 234-4433
Then we have instances that we can't
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
WASHINGTON, D.C. 20005 (202) 234-4433
-------
49
1 really characterize what kind of interaction it was,
2 because none of these interaction parameters were
3 significant. So you couldn't really say whether it
4 was antagonism or synergism, and that actually
5 occurred for half of the endpoints where we got
6 effects.
7 So we compared the results of the overall
8 analysis, the full factorial design, with the data
9 from the reduced analysis. So this is just the same
10 column that I just showed with the number of yeses and
11 noes, showing whether or not there was a statistically
12 significant interaction, and then compared to the
13 reduced dataset analysis.
14 For the most part, they did not match up.
15 In fact, there were very, very few cases where they
16 did give the same result. So it's obvious that,
17 looking at just the two doses the way that we did, you
18 could not predict the outcome that you would have
19 gotten when you looked at the full dose response
20 curves.
21 So now I'm going to try to show a little
22 bit of the data. This is the data from gait score.
23 Gait score is a subjective assessment of the
24 abnormality of gait by the animals moving around in an
25 open field. It's scored from one to four, with one
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
50
1 being normal and four being the most severe
2 abnormality.
3 What I've got plotted here are the
4 individual compounds, and the circles indicate the
5 average gait score. So this is the average rank. You
6 can see, with the TCE and the heptachlor, that there
* 7 is an increasing severity of gait.
8 Then the triangles indicate incidence or
9 the number of rats that show an abnormal gait. What
1 10 we had to do, because there is no statistical model to
11 handle rank data, we had to convert all the data to
12 incidence. So either the animals had abnormal gait or
13 they didn't.
14 So you can see what the dose response
15 curves look like for the incidence for
16 trichloroethylene alone, heptachlor, and DEHP
17 basically didn't have any effect on gait.
18 So the first place to look at is to start
19 looking at the binary combinations. One of the best
20 ways to look at binary combinations is to construct an
21 isobologram. What you do is you construct dose
22 response curves for, say, chemical A in the presence
23 of many different doses of the other chemical or
«. 24 chemical B.
25 You can actually then derive ED50 for
NEAL R. GROSS
(202) 234-4433
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
WASHINGTON, D.C. 20005
(202) 234-4433
-------
51
1 chemical A, and there will be different ED50,
2 depending on the dose, of B -- depending on what dose
3 B was. Then you can plot it.
4 So this line shows the EDSO for chemical
S A, the ED50 for chemical B by itself. Just draw a
6 line, and that's the theoretical line of additivity.
7 If you give a dose, say, of four of chemical B and
8 then do a dose response with chemical A, if the ED50
9 falls right about here, then it falls on the line of
10 additivity, and that's taken to be a statistical --
11 That's interpreted to be -- Statistically, that's
12 interpreted to be additivity. But if the ED50 is much
13 less, it falls in this area which is synergism, and if
14 it's much higher than the ED50 would have been, it
15 would be antagonism.
16 This is just a standard way of looking at
17 binary combinations. Then this shows the analysis
18 with gait score. What I did was calculate the EDSO
19 for heptachlor alone, which was about 14, ED50 for
20 trichloroethylene alone, which was about 1,000 --
21 these are milligrams per kilogram -- and connected the
22 line, and then looked at the dose response for
23 heptachlor in the presence of all the different doses
24 of trichloroethylene and the dose response for
25 trichlorethylene in the presence of all the doses of
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
52
1 heptachlor.
2 That's what these EDSO points are that are
3 plotted. You can see that they all fall in the area
4 of antagonism, and that was kind of a relief to me,
5 because that's what the response surface analysis had
6 said also, that there was antagonism between these two
7 chemicals.
8 Now the problem with the isobolograms is
9 that it doesn't show all the data. You have no idea
10 what the dose response curves look like, and these
11 EDSO are mathematically derived numbers.
12 So this 3-D graph actually shows all the
13 data for the combination of heptachlor and
14 trichloroethylene. Note that this all in the presence
15 of zero dose of DEHP. So this is just a binary
16 combination right now.
17 This kind of purple looking bar showed the
18 dose response for trichloroethylene in the presence of
19 zero heptachlor. So that's trichloroethylene alone,
20 just like you saw a couple of graphs back. Then
21 heptachlor alone is this set of bars right here.
22 Now if you, say, look at the high dose of
23 trichloroethylene by itself and then you start adding
24 heptachlor to it, you actually get less abnormal
25 gaits, especially at this mid-dose of heptachlor.
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
53
That's where the antagonism is showing up.
In the same way, if you look at the high
dose of heptachlor, you've got about 50 percent of the
animals showing abnormal gait, but as you add low
doses of trichloroethylene, that percentage drops
down. Then even at the highest dose of the two
combinations, the two chemicals in combination, you're
not getting much more effect than you got with the
individual compounds.
So this kind of dose response here where
you see all the data, you can actually see where the
antagonism is.
Now you can get into the really hairy
data. This is the three-way interaction, and this is
the data for lethality. As I said, there was a
significant interaction between trichloroethylene and
heptachlor, between DEHP and heptachlor, and also
between all three compounds with lethality.
So this graph here is the same as what you
saw with the gait score with a zero level of DEHP. As
you can see, the chemicals by themselves produce very
little lethality at the high doses. In the low dose
range here, we've got low doses of trichloroethylene,
the incidence of lethality in the heptachlor high dose
was actually less. So you can see where the
NEAL R. GROSS
j (202) 234-4433
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
WASHINGTON. D.C. 20005
(202) 234-4433
-------
54
1 antagonism is coming from.
2 Again, this is all dose dependent kinds of
3 interactions, which makes interpretation even harder.
4 Now if DEEP had no effect on the
5 interaction between trichloroethylene and heptachlor,
6 then all five of these graphs would look the same or
7 essentially the same. If you just kind of jump from
8 one to the next, you can see that they don't look the
9 same.
10 So we've got the low dose of DEHP. Then
11 it goes up a dose, and then this is the highest dose.
12 Now you recall I said that when you looked at the
13 three chemicals together, you actually had synergism,
14 and the best way of looking at that is to look at this
15 high dose combination.
16 This is the highest level of DEHP, the
17 highest dose of trichloroethylene and the highest dose
18 of heptachlor, and we got 100 percent lethality in
19 those animals. So this shows where the synergistic
20 responses actually are.
21 So this just kind of summarizes the study.
22 Basically, we had deviations from additivity in 77
23 percent of the endpoints. So a great number of the
24 endpoints did show significant interactions.
25 The reduced dataset predicted this kind of
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-M33 WASHINGTON, D.C. 20005 (202) 234-4433
-------
55
1 level of interactions for only three of the ten. So
2 it was not very efficient in being predictive. The
3 interactions mostly involved trichloroethylene and
4 heptachlor, which kind of makes sense; because those
5 are the two that act on the nervous system, in the
6 first place. But some of the interactions did include
7 DEEP, which has no effect on its own.
8 Some of the interactions were
9 characterized as antagonism. Other ones could not be
10 characterized. Then, of course, lethality showed
11 synergistic effects, and some of these effects were
12 dose dependent. The interactions were dose dependent.
13 So the bottom line is that the outcome
14 could not be predicted on the basis of the functional
15 effects of these chemicals, and they could not be
16 predicted based on the mechanisms of action.
17 So, basically, there's just no substitute
18 -- Based on these data, there's no substitute for
19 experimental testing, if you're looking at compounds
20 with different mechanisms of action.
21 Some of the lessons learned from the study
22 is that these kinds of factorial designs are not
23 practical, I don't think, for general laboratory
24 testing. It took 1,250 rats to generate all these
25 data, and it was about a year and a half to two years
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-*433
-------
56
l to collect the data and then another year and a half
2 to two years to actually analyze the data. So it's a
3 tremendous drain on resources.
4 There's still analyses that could be run.
5 There are other time points. As I said, we only
6 analyzed that one time point. We actually should go
7 back and analyze the data using the pre-dosing data as
8 a covariate, because some of these endpoints, the
9 response does depend on what the baseline data were.
10 Then there is still a method that needs to
11 be developed to analyze the ranked data. Converting
12 the ranked data to incidence data is probably not the
13 best way, but it's the only way we have of doing it.
14 Then just to end, I need to acknowledge
15 that this work was conducted under contract to EPA.
16 The laboratory work was conducted by ManTech
17 Environmental, and that was a contract to the HERL
18 back when we were HERL back before we became a
19 national lab.
20 Joe Elder and Bob Dyer were the contacts
21 at the EPA and helped us a lot with setting up these
22 studies and keeping them going, and Billie did most of
23 the testing for the neurotox testing, and Mike did all
24 the general tox testing as well as preparing all these
25 125 dosing solutions three times for each study. So
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
57
1 that was quite a feat in itself.
2 The data analysis were all conducted at
3 ~he Medical College of Virginia under the supervision
4 of Chris Gennings, and she had a graduate student,
5 Carol, who did all this work and, I think, finished
6 her whole thesis and probably killed herself after
7 finishing all this analysis.
8 That's it. So if we have time, I'd like
9 to answer some questions.
10 (Applause.)
11 DR. NORRIS: Can I remind people to please
12 use the microphone if you have questions, and we have
13 just a few minutes.
14 QUESTION: Could you speak a little bit
15 more about some of the interactions you saw at the
16 lower doses,- you know, the very lowest doses that you
17 tested.
18 DR. MOSER: For the other endpoints, you
19 mean?
20 QUESTION: Yes.
21 DR. MOSER: Well, for the most part, if
22 you actually graphed out all the data and looked at
23 it, which generates millions of these 3-D plots, it
24 looked pretty much the same kind of pattern that you
25 saw with lethality and with gait score.
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
58
l If you started adding low doses of the
2 second chemical, the effects of the first chemical
3 tend to go away. So it looked more like there was
4 antagonism for most of the endpoints. Then when you
5 start getting up to higher levels again, that goes
6 away, and you start getting more effects.
7 It just is a clear case of the antagonism
8 being dependent on the dose of the chemicals. I don't
9 know if that's enough to address, but we don't have
10 any other characterization of it besides that.
11 QUESTION: I wanted to ask you a question
12 in terms of synergism of lethality. Number one, were
13 the deaths very shortly after you gavaged them or were
14 they more long term, and how many -- I actually have
15 three questions. The second question is how long
16 after the tenth day of dosing did you hold the animals
17 or did you sacrifice them right after you did the
18 neurological testing?
19 Third, in terms of the lethality issue,
20 would the volume of the dose being important?
21 DR. MOSER: Okay. If I can remember all
22 three, the first one was, no, we never got death
23 immediately after dosing which, of course, is a very
24 important thing when you're talking about this number
25 of animals and gavaging. But the technicians that we
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
59
1 had doing this were very good and never did lung any
2 animals.
3 The death actually occurred during the
4 last couple of days of dosing, and we found and have
5 shown with heptachlor other times, too, that when you
6 dose heptachlor repeatedly, it tends to build up and
7 you start to see toxicity after days of repeated
8 dosing that you don't see on the first day.
9 So all the deaths occurred on about the
10 seventh, eighth and ninth and tenth day of dosing, and
11 usually what would happen is we would come in the next
12 morning and find them dead. We rarely ever saw them
13 die during the day.
14 Because the study was run with the general
15 tox, what we did was we tested them at 24 hours after
16 that tenth dose, and then immediately after that there
17 were sacrificed and we took the organs out, and
18 weighed the organs, prepared them for pathology and
19 took blood for clinical chemistries and that sort of
20 thing. So we never held them past that eleventh day.
21 Oh, the volume. I'm trying to remember.
22 I think the volume that we used was 5 milliliters per
23 kilogram, which is a very reasonable level for oral
24 dosing. We had to kind of hold some of these doses
25 down to make sure we could do that, but I think that's
NEAL R. GROSS
(202) 234-4433
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
WASHINGTON, D.C. 20005
(202) 234-4433
-------
60
1 what it was, 5 mis per kilogram.
2
3 Any other question? Thank you.
4 DR. NORRIS: Thank you, Ginger. I'm
5 trying to be quick with questions at this point and
6 stay on schedule. We will have time for more
7 questions during the panel discussion.
8 Our next speaker is Dr. Stephanie Padilla.
9 Thank you.
10 DR. PADILLA: Let's change gears yet
11 again. This is a group of studies that Wendy Haines,
12 who is the second author here, who is a graduate
13 student in my laboratory, started about a year and a
14 half ago. Basically, her doctoral dissertation is
15 going to be looking at mixtures of OP pesticides, OPs
16 and carbamates, and we'll probably also do an OP and
17 a pyrithroid pesticide.
18 We first started out with these two
19 compounds, chlorpyrifos and diazinon. I'm sure these
20 structures are probably familiar to many people in the
21 room. We chose them, because we thought we would get
22 a more than additive interaction on the basis of their
23 structures, their metabolism, and their mechanism of
24 action.
25 They are both anticholinesterase
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
61
inhibitors or they are both anticholinesterase
compounds. So they both inhibit acetylcholinesterase
as their mechanism of action. They are both converted
from their parent compound here. This sulphur is
replaced by an oxygen, and this becomes chlorpyrifos
oxone, and this one becomes diazoxone, which are very
potent cholinesterase inhibitors.
This conversion takes place by the P-450
enzymes, mostly in the liver. So they could interact
at that level. They are both detoxified via
carboxylesterases, stoichiometric binding by the
carboxylesterases, and they are also both substrates
for the A-esterases and can be hydrolyzed by the A-
esterases.
So their metabolism is extremely similar
in the animal. What we wanted to do was not construct
an isobologram, but what we wanted to do was to look
at the interaction of these two compounds using
multiple endpoints, and I'll get into that a little
bit here. Well, I'll go back.
Our quality -- Our hypothesis was that the
interaction was going to be more than additive,
because we thought that the interaction with the
detoxification enzymes, that the presence of one or
the other would inhibit the detoxification of the
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON. D.C. 20005 (202) 234-4433
-------
62
1 other enzyme and, therefore, create more toxicity of
2 the mixture than you would predict from looking at the
3 two by themselves. So we predicted that their
4 interaction would be supra-additive.
5 We used a dose additive model. Let me
6 take a little bit of time with this, and I also have
7 some examples graphically after this.
8 We spent some time reviewing the mixture
9 literature, and it seemed to be that, if you wanted to
10 look at compounds that had the same mechanism of
11 toxicity, that are homergic --is what they are called
12 in the literature -- that you needed to employ the
13 dose additive model.
14 To do this, what you do is you determine
15 the ED50 for whatever endpoint you've decided to look
16 at. In our case, we looked at two dose levels. We
17 looked at what I would consider a very low dose level,
18 which was the amount of the compound that would cause
19 50 percent inhibition of red blood cell
20 acetylcholinesterase.
21 Then we looked also -- I did another group
22 of experiments that looked at a higher dosage level,
23 which was the amount of the compound that would cause
24 50 percent inhibition of brain acetylcholinesterase.
25 So we looked at it at two different dosage
NEAL R. GROSS
(202) 234-4433
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
WASHINGTON. D.C. 20005
(202) 234-4433
-------
63
1 levels, but in general what you do is you determine
2 the ED50 for whatever your target endpoint is for both
3 of the compounds. Then you take half that dose for
4 each one of the compounds and put that in your mixture
5 and look to see what the effect of the mixture is.
6 Your results can look like this, depending
7 on what the interaction is. This is completely made-
8 up data. That's why it looks so good. If you go up
9 here at the very top, this is what your results would
10 look like if you had an additive interaction.
11 So in this case, if we had -- This is the
12 dose of chlorpyrifos that causes 50 percent inhibition
13 in whatever tissue we're looking at. This is the dose
14 of diazinon that causes 50 percent inhibition of that
15 same tissue.
16 Then if you take half of those doses and
17 mix them together and dose the animal, you should get
18 --If the interaction is additive, you should get 50
19 percent inhibition.
20 Now this is what it would look like if it
21 was more than additive or supra-additive or
22 potentiation or synergy, depending on who you read and
23 how they define it. But this again would be the dose
24 of chlorpyrifos that caused 50 percent inhibition.
25 This would be the dose of diazinon, but together you
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON. D.C. 20005 (202) 234-4433
-------
64
1 would get more inhibition if you used half of the
2 dose. This down here, of course, is if you actually
3 had antagonism or infra-additivity. There would be
4 less inhibition in the groups.
5 So this was the model that we chose to use
6 with these two compounds. The first thing you have to
7 do is construct a very good dose response. You have
8 ~o know what dose of chlorpyrifos or what dose of
9 diazinon causes 50 percent inhibition in the brain or
10 50 percent inhibition in the red blood cells. That's
11 what you've got here.
12 The x axis here is approximately the same
13 for both of them, and you can see that there is a very
14 different pattern here for the two compounds. The
15 upper is chlorpyrifos. The filled symbols are brain,
16 cholinesterase. The open symbols are red blood cell
17 acetylcholinesterase.
18 You can see here that chlorpyrifos
19 inhibits red blood cell acetylcholinesterase at lower
20 doses than it does brain acetylcholinesterase. This
21 is not news.
22 The same thing is true for diazinon. The
23 various symbols, the various shapes are different
24 experiments that we did. This is just to give you --
25 We had to repeat -- We had some data from other
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON. D.C. 20005 (202) 234-4433
-------
65
1 studies. Then we did newer studies to combine with
2 that data, and the same thing is true down here for
3 diazinon.
4 This line right here is the 50 percent
5 line. So if you look at diazinon here, you can see
6 that at this dose of about 10 milligrams per kilogram,
7 you get about 50 percent inhibition of the red blood
8 cell acetylcholinesterase, but at a dose at about --
9 this is about 75 milligrams per kilogram, you get 50
10 percent inhibition of the brain acetylcholinesterase.
11 So there's a lot of work to be done up
12 front with these type of studies, because you've got
13 to have a really good dose response curve. Even if
14 you had a really good one, it doesn't always turn out
15 like you want it to.
16 So these are all the compounds by
17 themselves. These are all acute dosing in corn oil in
18 adult male rats. The animals were sacrificed at the
19 time of peak effect, which conveniently for these two
20 compounds is about the same time, which is about three
21 to four hours after dosing.
22 Our endpoints: We had three different
23 flavors of endpoints. We basically looked at
24 cholinesterase inhibition, of course, in the brain, in
25 the red blood cell, but we also did other
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
66
l cholinesterase determinations in other tissues.
2 We did toxicokinetics. We're set up to
3 look at the toxicokinetics of chlorpyrif os. So we
4 looked at the distribution of chlorpyrifos in various
5 tissues to answer the question of whether co-dosing
6 with diazinon would cause a different distribution of
7 chlorpyrifos and/or its metabolites in the liver and
8 the brain.
9 Then in collaboration with Ginger Moser,
10 she did behavioral assessment on these animals. So
11 she ran an abbreviated functional observational
12 battery on the animals.
13 So in the first experiment, which is the
14 lower dose experiment, we used male -- adult hooded
15 male rats, and our target endpoint was red blood cell
16 inhibition -- red blood cell acetylcholinesterase
17 inhibition of approximately 50 percent.
18 We determined the ED50 for chlorpyrifos to
19 be approximately one milligram per kilogram, and
20 that's the dose of chlorpyrifos that produces 50
21 percent inhibition in the red blood cell by itself,
22 and diazinon to be ten milligrams per kilogram.
23 So what we did was we -- in our mixture
24 group -- the mixture group consisted of .5 milligrams
25 per kilogram of chlorpyrifos and 5 milligrams per
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
(202) 234-4433 WASHINGTON. D.C. 2000S (202) 234-4433
-------
67
l kilogram of diazinon. We had five dosage groups. We
2 had a control that got nothing but vehicle.
3 We had the two groups by themselves. We
4 had chlorpyrifos by itself. We had diazinon by
5 itself. We had the mixture of the two at half, but
6 then we also -- In order to compare the
7 pharmacokinetics, we needed to have the chlorpyrifos
8 at half the dose, at .5 milligrams per kilogram,
9 because we needed to have that group to compare to the
10 group, the mixture group, to see if the amount of
11 chlorpyrifos and metabolites in those two groups were
12 the same or had been altered.
13 You can't just assume that the amount of
14 chlorpyrifos in the animal at one milligram per
15 kilogram is going to be twice that as what you should
16 see at five. So we needed -- I mean the .5. We
17 needed the .5 to compare with the one. So that's a
18 general outline.
19 So this is just to show you where we are
20 in the dose response curve with regard to those doses.
21 So we had about one milligram per kilogram
22 for chlorpyrifos and the mixture group. So that's the
23 dose that we predicted would produce about 50 percent
24 inhibition of acetylcholinesterase in red blood cells
25 and about ten for the diazinon. So you can see where
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
(202) 234-W33 WASHINGTON. D.C. 20005 (202) 234-4433
-------
68
1 we are here on the dose response curve.
2 So here are the results, same bars as I
3 was showing you guys before, only this is real data.
4 This is chlorpyrifos by itself at one milligram per
5 kilogram. We didn't get exactly 50 percent
6 inhibition, and diazinon at ten milligrams per
7 kilogram, again a little bit higher than 50 percent.
8 But you can see here that the mixture is right in
9 between the two, which would indicate that we have an
10 additive interaction at this low dose.
11 Same thing is true for plasma. We had a
12 bit more inhibition by the two compounds in the plasma
13 by themselves, but the mixture here produced a value
14 that was in between these two, not higher or lower.
15 We did not get any significant inhibition in the
16 brain, which is what you would expect from our
17 original dose response curves.
18 Now if you look at the toxicokinetic
19 results, we're looking at -- At this low dose, we were
20 unable to see any chlorpyrifos or any of its
21 metabolites in the brain, but we did see
j
22 trichlorpyridinol, which is a metabolite TCP of
23 chlorpyrifos, in the liver.
24 if you look here, this is how much
25 trichlorpyridinol is in the liver. This is an
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
69
1 nanograms per gram of tissue. This is how much
2 trichlorpyridinol is in the liver of the animals that
3 were dosed with one milligram per kilogram, and this
4 is how much, and it's just about, conveniently, half.
5 You couldn't have predicted that.
6 This is how much is in the liver at .5
7 milligrams per kilogram, and this bar here is how much
8 trichlorpyridinol that was in the liver of the animals
9 that received the mixture.
10 Now this is a different paradigm here.
11 What you're looking at here is how closely these two
12 bars resemble each other, because these animals got .5
13 milligrams per kilogram of chlorpyrifos, but they also
14 got 5 milligrams per kilogram of diazinon. But you
15 can see here that these bars aren't significantly
16 different, and having that amount of diazinon on board
17 did not change the distribution of the
18 trichlorpyrydinol, presumably, and the chlorpyrifos in
19 those animals.
20 There was no behavior to measure in those
21 animals. They were not showing enough over-toxicity
22 or behavioral alterations at those dosages for there
23 to be any assessment in those animals. So I don't
24 have any behavioral results for those animals.
25 Now in the second experiment -- This was
NEAL R. GROSS
(202.) 234-4433
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
WASHINGTON, D.C. 20005
(202) 234-4433
-------
70
1 basically the same type of animal, Long-Evans hooded
2 adult male rates. Our endpoint here is 50 percent
3 inhibition of cholinesterase in the brain, not the red
4 blood cell.
5 It was determined that the EDSO in this
6 case was 20 milligrams per kilogram for the
7 chlorpyrifos, which is 20 times higher than it was for
8 the red blood cell, and 75 milligrams per kilogram for
9 the diazinon, which is about 7.5 times higher than it
10 was for the red blood cell.
11 Again, we had five dosage groups. We had
12 a control. We had each one by itself. We had the
13 mixture, and then we had animals that received
14 chlorpyrifos at half the dose of it by itself so we
15 could compare the distribution.
16 Again, our dose response curves here were
17 at about ten for the chlorpyrifos for the brain
18 acetylcholinesterase, and we're down here at 75 for
19 the diazinon and for the brain acetylcholinesterase.
20 Here are the results from that group of studies.
21 This is cholinesterase inhibition on the
22 y axis. This is chlorpyrifos by itself. We did not
23 have 50 percent inhibition in the brian. We only had
24 about 90 percent -- 80 percent -- Sorry. We only had
25 about 15-20 percent inhibition in the brain.
NEAL R. GROSS
(202) 234-4433
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
WASHINGTON, D.C. 20005
(202) 234-4433
-------
71
1 The diazinon, we did have about 40 percent
2 inhibition in the brain, and the mixture was between
3 the two. So, again, the interaction is additive.
4 We've also looked at the retina. The chlorpyrifos
5 produced about 60 percent inhibition in the retina of
6 those same animals. Diazinon produced about 40
7 percent inhibition, and again the mixture group is in
8 between them, indicating that the quality of
9 interaction here is additive, even though we've used
10 a much higher dose.
11 Diaphragm for peripheral tissue --we just
12 decided to look at this -- is basically the same
13 thing. We had a lot less inhibition in the brain than
14 we did in the diaphragm of the chlorpyrifos dosed
15 animals than we did in the diazinon dosed animals all
16 by themselves, but again the mixture was right in
17 between.
18 So we don't see anything here in the
19 biochemical results that would indicate that we've got
20 anything more than just an additive interaction of
21 these two, either at a low or a higher dose.
22 Now if you look at the toxicokinetics, we
23 were able in the liver to see both chlorpyrifos, the
24 parent compound, and trichlorpyridinol, the
25 metabolite. Here again, you're looking at -- You're
NEAL R. GROSS
(202) 234-4433
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
WASHINGTON, D.C. 20005
(202) 234-4433
-------
72
1 comparing these two bars.
2 In the liver the level of
3 trichlorpyridinol in the mixture animals was basically
4 not any different than if you had given the same
5 amount of chlorpyrifos by itself. These two are not
6 significantly different statistically, but it does
7 look like, if you did this maybe with 200 more
8 animals, you might be able to get yourself a
9 statistical difference between here. But all this is
t
10 telling you is that maybe the presence of diazinon at
11 this dose is inhibiting the conversion of the
12 chlorpyrifos to its metabolites. But right now this
13 is not significantly different and, therefore, we've
14 got no reason to suspect anything more than additive.
15 We actually did see some trichlorpyridinol
16 in the brains of these animals, and again these two
17 bars are not significantly different. So the presence
18 of diazinon did not change the distribution of
19 trichlorpyridinol in the animals.
20 Here are the behavior results. We saw
21 something a little bit different here. This is the
22 motor activity in these animals, same method of
23 presentation here. The chlorpyrifos animals alone saw
24 about a 40 percent inhibition in motor activity. The
25 diazinon animals you saw a slight inhibition in motor
NEAL R. GROSS
(202) 234-4433
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
WASHINGTON, D.C. 20005
(202) 234-4433
-------
73
l activity.
2 The mixture group is basically between
3 these two bars, and so you have an additive
4 interaction. The mixture is where you would expect it
5 if you gave half of the dose of these two bars.
6 Up here, this is the ataxia ratings that
7 Ginger's group does. You can see here -- We've
8 actually done this twice, because it was quite
9 interesting the first time. This is the control
10 animals. These are the chlorpyrifos by itself. These
11 are the diazinon by itself.
12 In each case here you had one animal that
13 showed a strange -- You've got a four or five point
14 scale here. So this is slight ataxia in these
15 animals. But in the mixture group we had four
16 animals, five animals, that showed this effect, and
17 each time we saw this.
18 I don't want to make anything too much out
19 of this except that it was repeatable. so we've got
20 here is an interaction at either a low or high dose
21 that shows an additive interaction, the low dose being
22 the dose that inhibited the brain -- I mean the red
23 blood cell acetylcholinesterase, and the higher dose
24 being the animals that had approximately 50 percent
25 brain acetylcholinesterase inhibition.
NEAL R. GROSS
(202) 234-4433
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
WASHINGTON, D.C. 20005
(202) 234-4433
-------
74
1 In neither one did we see any type of --
2 any evidence that would lead us to believe that the
3 interaction could be considered anything but additive
4 except for some aspects of the behavioral testing.
5 These are the people that did all the
6 work. Debbie Hunter does all the pharmacokinetic
7 analysis. Lynn Lassiter and Renee Marshall help out
8 in the lab with assessing the animals and also
9 collecting tissues. Kathy McDaniel, Ginger Phillips
10 and -- I mean Pam Phillips and Ginger Moser work with
11 the behavioral assessment of the animals.
12 I'll be glad to take any questions, if
13 anybody has got any questions. Nope? Okay.
14 (Applause.)
15 DR. NORRIS: Thank you, Stephanie. We'll
16 take a quick 15 minute break, and we'll reconvene.
17 Thank you.
18 {Whereupon, the foregoing matter went off
19 the record at 10:49 a.m. and went back on the record
20 at 11:14 a.m.)
21 DR. NORRIS: Okay. We keep getting off
22 and on schedule. We'll try and stay closer to on
23 schedule.
24 I'll introduce our next speaker who made
25 it from Baltimore, thank goodness. In spite of the
NEAL R. GROSS
(202) 234-4433
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
WASHINGTON, D.C. 20005
(202) 234-4433
-------
75
1 weather? Dr. Will Boyes is the Chief of
2 Neurophysiology and Toxicology Branch and the Acting
3 Division Director for the Neurotoxicology Division
4 right now at RTF in North Carolina.
5 I see folks still coming in. I did
6 major introductions before you arrived to this
7 morning's session, but I'll let you take it from here,
8 Will. Thank you.
9 DR. BOYES: Well, thank you, Debbie, and
10 I'm very happy to be here, especially given our
11 episode on the plane this morning. We approached the
12 Washington airport runway twice, and then the pilots
13 said they couldn't see the runway. So we were going
14 to Baltimore. I was looking at my watch wondering if
15 I was going to make it or not, but they put us in cabs
16 and we got here in good shape.
17 I haven't heard the morning's introductory
18 talk. So if I repeat something, please forgive me.
19 One more apology: I'm sorry I didn't put a biography
20 in the thing. I didn't realize that it was going to
21 be published like that, but I work in the same
22 division as Stephanie and Ginger and, if you want to
23 contact me, I'm in the EPA directory.
24 I am Acting Director for the Division at
25 the moment while our division director is up here
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
76
1 learning what risk assessment is all about. So if I'm
2 a little bit confused, I have a lot of excuses
3 already.
4 Let's go ahead and get started now.
5 People are sitting down. This is a set of studies
6 that was done almost a decade ago, like what Ginger
7 presented to you. At that time the EPA Superfund
8 office was very interested in mixtures and funding
9 research in the Office of Research and Development,
10 and this is one of the projects that we ran with that
11 funding.
12 This one was done with me as the project
13 officer in collaboration with Chuck Rebert at the
14 Stanford Research Institute. All of the data were
15 collected out in Stanford at SRI.
16 We were interested in solvents. Before we
17 designed the studies, though, we started looking at
18 the literature. At that time, and I think it's
19 probably still true today, the literature on mixtures
20 in neurotoxicology is very sparse. There are very few
21 studies published.
22 The ones that have been published have
23 used largely in vitro preparations and, because of
24 some of the complexities that I think you've already
25 heard about, it's more practical to do it that way.
NEAL R. GROSS
(202| 234-4433
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
WASHINGTON. D.C. 20005
(202) 234-4433
-------
77
1 But it leaves a lot to be desired in terms of
2 applicability to in vivo situations.
3 Those in vivo mixture studies that you
4 find are largely, if they are not cancer studies, they
5 are hepatotoxicity studies. There's very few in the
6 nervous system, and there's even fewer with
7 environmentally relevant compounds.
8 So we wanted to address some of those
9 problems, but first why is that the case? It's a
10 difficult thing to do mixture studies, and it's a
11 difficult thing to study the nervous system.
12 The nervous system is a very heterogeneous
13 structure. It has multiple target sized complex
14 systems. You can have many outcomes. I think you
15 saw this in Ginger's presentation. In some cases,
16 there are very steep dose response curves, which can
17 make the analysis difficult.
18 The difficulties with chemical mixture
19 studies we're all hearing about today, that there are
20 many chemicals, each -- If you look at complex
21 mixtures, each complex site, each Superfund site, each
22 source if you're talking about air pollution, might
23 have a unique mixture, and how do you generalize from
24 one complex mixture to another?
25 The designs, when you start looking at
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
78
1 multiple chemicals and multiple interactions, get very
2 complex, and so do the analyses. So this is what
3 you're faced with when you try and do these kinds of
4 studies. But we forged ahead bravely.
5 We wanted to test the additivity
6 hypothesis, because in the absence of other
7 information, the EPA risk assessment assumes that
8 compounds are additive.
9 We wanted to do this in whole animals with
10 relevant compounds . So based on the interest from the
11 Superfund program, we looked at a series of volatile
12 organic solvents. They are not just of interest, of
13 course, to Superfund. They are major components of
14 EPA's portfolio for just about all the program
15 offices.
16 With organic solvents in the nervous
17 system, in humans the primary concern is usually
18 cognitive function, but there has not been good models
19 for cognitive function in animal studies. So we
20 thought we would focus on a very definitive outcome
21 that has been reported. It was actually discovered by
22 Chuck Rebert in his collaborative work prior in the
23 mid-eighties. That's damage to hearing, ototoxicity.
24 I'll talk more about that in just a minute.
25 We wanted to use whole animals by
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
(202) 234-4433 WASHINGTON, O.C. 20005 (202) 234-4433
-------
79
l inhalation route, and to do this, though, we started
2 with an abbreviated experimental design that's
3 similar, in fact, to what Stephanie presented in the
4 last talk.
5 This is a scanning electromicrograph from
6 several rate cochlea that I stole from the
7 dissertation of Ann Christian Johnson. These are
8 stereocilia on three rows of hair cells. This is
9 after just a couple of days of exposure to toluene.
10 I think the exposure for 18 hours a day for three days
11 to 1400 ppm high dose and extended exposures.
12 This is a couple of days later, and you
13 can see that here they look normal and healthy, and
14 here you can see the stereocilia beginning to break
15 up, fall apart. Some of the cells are actually
16 missing.
17 This is a couple of days after the
18 exposure stopped. You can see there are whole areas
19 where the hair cells, these outer hair cell, have
20 disappeared, and there are other cells infiltrating to
21 take their place a couple of weeks later, and there
22 are whole regions of the cochlea that are devoid of
23 outer hair cells. This will produce a profound
24 hearing loss in those frequency bands in the cochlea.
25 This is from a review article by Gordon
NEAL R. GROSS
(202) 234-4433
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
WASHINGTON. D.C. 20005
(202) 234^433
-------
80
l Pryor. A lot of solvents were tested by Rebert and
2 Pryor through the eighties and nineties, and this is
3 from a review article that he published.
4 It's interesting. There is a very unusual
5 and so far not well figured out structure activity
6 relationship. The compounds that do produce hearing
7 loss, and these are high dose, again, phenomena, but
8 a lot of substituted benzenes, methylbenzene which is
9 toluene, ethyl and propylmythoxybenzene and mixed
10 xylenes. There are three xylene isomers, but when you
11 test the isomers separately, only the para-isomer is
12 ototoxic. The ortho and the meta- isomers are not.
13 Styrene and other substituted benzene is
14 ototoxic, and then some -- and then monopyribenzene
15 also, carbon disulfide, a little bit different solvent
16 from these, and then some alkanes. Trichloroethylene
17 is ototoxic and N-butanol. But then interestingly,
18 these did not produce noticeable ototoxicity. Benzene
19 itself, even though all the substituted benzenes --a
20 lot of substituted benzenes do, these substitutes
21 don't, isopropyl, 1,2-dimethyl, 1,3, the xylene
22 isomers.
23 Then comparing with trichloroethylene,
24 dichloromethane, trichlorethane, tetrachloroethylene
25 or perf doesn't. N-butanol is. 2-propanol is not,
NEAL R. GROSS
(202) 234-4433
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
WASHINGTON, D.C. 20005
(202) 234-M33
-------
81
1 and ethyl alcohol and N-hexane is not. N-hexane is
2 the well known compound for producing peripheral
3 neuropathy, like I imagine John O'Donoghue talked
4 about, but I didn't hear him, because I was in
5 Baltimore.
6 Okay. So I mentioned that we wanted to
7 use an abbreviated experimental design. This sort of
8 schematically illustrates what I'm talking about.
9 It's very important when you look at chemical mixtures
10 to do a dose response curve for your constituents of
11 the mixture.
12 That's because, if you think of compound
13 A here as just a typical dose response curve, if you
14 do a mental experiment of combining a dose level like
15 this of compound A with the same thing of compound A,
16 a mixture of the compound A with compound A, and then
17 that would be this dose plus this dose. So it would
18 put you out here, and you could get a very profound
19 effect by mixing compound A with itself, much greater
20 than you would ever expect, and it's only because
21 j you've moved into a steep part of the dose response
I
22 curve.
23 So if you don't understand the shape of
24 the dose response curve, then you can fallaciously or
25 spuriously conclude that you are getting a greater
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
82
1 than additive effect, when in fact you've just moved
2 into a nonlinear part of the dose response curve.
3 So if you don't understand where you are
4 on the dose response curve for your individual
5 compounds, then it becomes almost impossible to
6 interpret a mixture of different compounds.
7 This is the design that Stephanie talked
8 about where she took a dose response curve for
9 compound A and for compound B, took an equal effective
10 dose of these, and then you add proportions of those
11 doses together. If you add this EDa to EDb, then you
12 would expect to get a very large effect. You can
13 predict that based on the shape of the dose response
14 curve.
15 We actually used these as our target dose
16 in our experiments where we added fractions of the
17 effective dose of compound A and fractions of the
18 effective dose of compound B together, because we
19 wanted a target on the steep part of the dose response
20 curve so that we would be able to see changes that
21 were both greater than or less than additivity.
22 So this is our basic strategy. This is a
23 typical isobologram which you would take relative
24 proportions of the compound A and compound B and add
25 them together at different dose levels, so that you
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
83
1 have these different rays coming out, and then if they
2 are additive, you would get a straight line between
3 them at the point which produces an equal effect all
4 along here. This is an additivity isobole.
5 If you get -- if it takes larger and
6 larger doses of these compounds to produce the same
7 effect, then that's considered to be less than
8 additive or some people have used the term
9 antagonistic. If it takes lower doses to produce the
10 same effect, then that's termed greater than
11 additivity or synergism.
12 We didn't have the ability to do that with
13 these large animal studies. So what we've done is
14 focused on the line of additivity, which you can see
15 here, and take a 50/50 mixture of the doses that
16 produced these effective doses, 75:25 and 25:75, and
17 then 100/0 and 0/100 or vice versa, and add these --
18 use these dose combinations to test our additivity.
19 So if the compounds are, in fact,
20 additive, then they should all produce the same
21 effect. If they are greater than or less than
22 additive, they should be statistically different from
23 this linear extrapolation between the effective dose
24 of A and the effective dose of B.
25 So that's the general strategy for our
NEAL R. GROSS
(202) 234-4433
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
WASHINGTON. D.C. 20005
(202) 234-4433
-------
84
1 experimental design.
2 This just sort of differentiates
3 traditional from our simplified isoboles. Many of
4 those levels for each compound when you apply the iso-
5 effective concentration combinations, and it's
6 practical in vitro, but it's very difficult to do in
7 vivo unless you have very simple measures.
8 So we used our simple dose group at
9 relative proportions, and then we're plotting them to
10 see if the effects are different from the iso-
11 effective. You can do this more practically with the
12 kind of studies we want to do.
13 Nonadditivity will be demonstrated by
14 whether or not the effects we see are different from
15 that linear prediction.
16 This is kind of a repeat of just what I
17 said. Dose response curve for each compound. Our
18 mixtures are 0:100, 75:25, 50:50, 25:75 and 0:100
19 proportion of the iso-effective concentrations.
20 We picked five solvents from the list of
21 positive ototoxic solvents: Styrene,
22 trichloroethylene, toluene, mixed xylene isomers, and
23 chlorobenzene. We did the same solvents on this side.
24 When you do binary combinations of these, you come up
25 with ten experiments that you can run.
NEAL R. GROSS
(202) 234-4433
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
WASHINGTON. DC. 20005
(202) 234-4433
-------
85
1 We ran all ten of these combinations in
2 this set of studies, first doing dose response curves
3 for the individual solvents, picking the iso-effective
4 levels, and then going back with a second study with
5 five dose groups for each of these solvents.
6 We had about eight rats in each of these
7 dose levels or -- I'm sorry, about 40 animals plus a
8 clean air control for about 50 animals, 48 animals,
9 for each of these ten experiments. So you can see,
10 even with these simplified designs, when you look at
11 multiple chemicals, it becomes a very large endeavor.
12 The animals were inhalation exposed five
13 days, eight hours per day, Monday through Friday.
14 It's enough to produce permanent ototoxicity. We
15 waited about ten days before we started testing their
16 hearing.
17 This is how we tested hearing. We didn't
18 dissect out the cochlea and do scanning
19 electromicrographs like you saw in that beautiful
20 slide of Ann Christian Johnson's. We have a much more
21 efficient way to do this, and this is the
22 electrophysiological method called the brain stem
23 auditory evoke potential or some people refer to it as
24 the auditory brain stem response, ABR.
25 To do this, you anesthetize the animal so
NEAL R. GROSS
(202) 234-4433
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
WASHINGTON, D.C. 20005
(202) 234-4433
-------
86
1 that it doesn't move. That's the main thing you want.
2 You can put needle electrodes under the scalp,
3 maintain body temperature so that's not a concern, and
4 then stimulate the ears with tone pips.
5 You can make the sound loud or you can
6 change the frequency from low to high. Now what we
7 report is a response that's the mass discharge of
8 cells in the auditory system. There's a lot of work
9 that's gone into the generators of this potential, and
10 we know that the first peak we measure is generated --
11 There's a little response in the cochlea. The first
12 major peak is in the auditory nerve.
13 Here you can see the eighth nerve or the
14 acoustic nerve. Then the next peaks are generated
15 along the line of the ascending auditory pathway, and
16 the neural generators for each of these are pretty
17 well identified, the cochlea nucleus, superior
18 auditory complex, lateral aniscus, medial tinnicular.
19 So we can follow the ascending auditory
20 signal as it goes up the auditory pathway. This
21 basically says what I just said. Anesthetize the
22 rats. Stimulate varying and loudness in pitch,
23 recording the electrical activity.
24 Then we needed a simple measure. There
25 are a lot of measures you can take off of this, but
NEAL R. GROSS
(202) 234-4433
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
WASHINGTON, D.C. 20005
<202) 234-4433
-------
87
1 because of our complex design, we wanted a single
2 dependent variable. What we settled on was the
3 integrated amplitude of the response between 55 and 85
4 decibels. I'll show you this in just a minute.
5 This gives us a very nice and regular
6 measure that's sort of a global indicator of the
7 neural function in the auditory system. Let me show
8 you the next slide, and that will become clear.
9 These are examples from groups of animals
10 averaged together at 25 to 95 decibels sound level.
11 In the control animals you can see that it starts out
12 at 25 decibels with a very small response, and it
13 grows in amplitude -- this is voltage on this scale,
14 and this is time on this scale -- and becomes earlier
15 in latency as the response -- as the stimulus gets
16 louder.
17 PI is the peak generator in the auditory
18 nerve, and you can see that it follows up and down
19 with a nice relationship to intensity. What we did
20 for the dependent measure was to take between here and
21 here, rectify it so that it was all positive, and then
22 do an area under the curve. It gives us just a
23 summary measure for the function of the auditory
24 system.
25 [ Chuck decided to it. I think it was
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
88
1 between 25 and 85 decibels. So he summed all of these
2 up. I think he left off 95 decibels, because of the
3 possibility of auditory recruitment, which happens in
4 some animals with hearing loss.
5 What's interesting here is the effect of
6 styrene. Here you can see after 2000 PPM styrene for
7 five days many things. First of all, the amplitudes
8 are much smaller, but more importantly, the threshold
9 for eliciting response is tremendously changed.
10 What you see, about 25 decibels in the
11 control, you're seeing at around 65 decibels in the
12 exposed animals. This is a profound hearing loss,
13 about a 40db loss of hearing. So that shows you the
14 kind of an effect that we're talking about.
15 Now I didn't bring all the data. I wasn't
16 sure how much time it would take me to go through that
17 part of it. But this just samples dose response
18 curves for the individual compounds alone. This is
19 chlorobenzene, and this is toluene. This is kind of
20 the best and the worse.
21 The toluene looked very nice. We had a
22 nice linear response curve. Chlorobenzene --we had
23 a little trouble with this particular compound, not
24 the others, in that at high doses the animals lost a
25 lot of weight. So we couldn't go much higher, and we
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
89
1 had a couple of uneffective dose levels. So we didn't
2 have a lot of room on the chlorobenzene dose response
3 curve to work out things.
4 Let me show you the next slide now. This
5 is sort of -- We ran ten experiments. One of them, we
6 had to throw out because the solvent generation
7 apparatus basically failed, and the animals got
8 exposed to what they weren't supposed to. But the
9 other nine all gave us nice data, and we felt very
10 confident about that they were exposed to what they
11 were supposed to be.
12 We did have some trouble with some of the
13 experiments, and I'll talk about that in a minute.
14 This is what we expected to see. This is
15 trichloroethylene and toluene. The control animals
16 had a very large integrative amplitude, about 200.
17 The trichloroethylene alone produced about a half a
18 maximal effect. This was exactly what we were
19 shooting for, and the toluene produced about the same
20 thing.
21 You can see that the 25:75, 50:50, and
22 75:25 combinations produced effects that were
23 equivalent to those of the compounds alone. This is
24 exactly what we expected to see with additivity, and
25 this is the equivalent to the graphs that Stephanie
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
90
1 was showing in the last talk.
2 The problem that we had was that we saw
3 drift in the dose response curves for the individual
4 compounds alone. Here, xylene produced about what we
5 wanted to by itself, but here trichloroethylene at
6 2800 PPM is less effective than it was here at 2600
7 PPM. But still, the combinations between these two
8 produced a linear relationship.
9 Dave Sensgaard did a lot of statistical
10 analysis with us, and demonstrated that, even though
11 our individual compounds were not iso-effective, that
12 the linear relationship between them was indicative of
13 an additive effect.
14 Here you can see a similar effect. Here
15 the xylene was not as effective as we thought, but
16 again in combination with chlorobenzene, it was a
17 linear relationship between the effects of the two
18 compounds.
19 Similar here, this is chlorobenzene at
20 2000. This was chlorobenzene at 2400. So again you
21 can see there's quite a drift in the different
22 attempts to ascertain the dose response curve for the
23 individual compounds. But the simultaneously run
24 mixtures here with toluene showed again an additive
25 interaction.
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
91
1 This is four of the ten experiments, but
2 without showing you all the data, what we found was
3 essentially in all the experiments there was a linear
4 relationship between the different combinations. so
5 we did not detect any nonadditive interactions.
6 Now let me go back and make one more point
7 from that slide. One of the things that I think is
8 our lessons learned is that, obviously, you need to
9 have --It would be nice to have stable dose response
10 curves to do this kind of work.
11 Another thing is that it's not clear to me
12 how big a difference from additivity we would have
13 been able to detect. So it would have been nice to
14 have compounds that we knew were nonadditive to see if
15 our design was sensitive enough to detect them.
16 We did some mathematical calculations, but
17 it would have been nice to have some positive control
18 compounds.
19 So a couple of conclusions: For the
20 ototoxic solvents we saw no deviations from
21 additivity. This suggests that the compounds are
22 substitutable for each other, that they cause the same
23 thing. And given that they are all organic solvents
24 and they produce a fairly similar hearing loss, one
25 possibility is that this suggests a common mechanism
NEAL R. GROSS
(202) 234-4433
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
WASHINGTON, D.C. 20005
(202) 234-M33
-------
92
l or mode of action. It doesn't prove it, but that is
2 a possibility.
3 Lessons for the experimental design: I
4 think that our simplified analysis of the isobolograms
5 was effective at allowing us to test binary mixtures.
6 We had, as I pointed out, problems with the
7 instability of our dose response curves for the
8 individual compounds and, as I mentioned before, it
9 would have been nice to have a positive control to
10 differentiate between our ability to detect additivity
11 and nonadditivity.
12 We weren't able to follow these studies
13 beyond this. But one of the things that I would have
14 liked to have been able to do was to look at noise.
15 Noise is a very common cause for hearing loss in
16 occupational settings, although it's not currently
17 something EPA is concerned about. But other people
18 have looked at workers exposed to solvents in the
19 presence of noise.
20 The conclusion has been that it does tend
21 to cause a greater than expected hearing loss for
22 noise when you are exposed to solvents. So I think
23 that's something that is a very interesting outcome
24 from this work.
25 I'd be happy to take any questions. Thank
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON. D.C. 20005 (202) 234^433
-------
93
1
2
3
4
you.
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
drifted?
(Applause.)
QUESTION: Why do you think the baseline
DR. BOYES: I don't have the answer to
that. We did take simultaneous blood levels from all
of the animals during all the exposures, and they were
fairly constant. It doesn't look like it was exposure
of the animals to the solvent or absorption into the
blood.
It was something in the different batches
of animals or the different time of year or whatever
it was when we ran the studies that the animals
differed in their sensitivity to the solvent. I wish
I knew that, because I don't.
DR. NORRIS: Thank you, Will. Our next
speaker, Dr. Greg Christoph.
DR. CHRISTOPH: Thank you, Deborah, and
thanks to the organizers for inviting me here to talk,
as I'm just preparing to leave the industry. It was
mentioned earlier that I am retiring tomorrow, and
actually I'm not going to retire. I'm going to --
Well, I am retiring, but I'm going to start a
different career with a different kind of industry and
a different kind of regulatory agency.
NEAL R. GROSS
(202) 234-4433
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
WASHINGTON, D.C. 20005
(202) 234-4433
-------
94
1 So this is my opportunity just to say
2 exactly what I think. Well, indeed, I think I've
3 always said exactly what I thought. At least I hope
4 I have, and perhaps that's one reason why I'm retiring
5 as a director instead of a senior vice president. But
6 I do work for the DuPont Company as of today, and I'm
7 sure all of you are aware that DuPont is a very, very
8 large chemical company and makes a number of the kinds
9 of chemicals we've been talking about here today.
10 In particular, since I am going to be
11 talking about pesticides, DuPont does have an
12 agricultural chemicals business, and they do make lots
13 of different kinds of pesticides. In particular, they
14 make an organophosphate cholinesterase inhibitor.
15 They make carbamate cholinesterase inhibitors. They
16 make a pyrethroid sodium channel opener that is an
17 insecticide, and soon they will be producing a
18 pyraziline insecticide which is a new and different
19 kind of mechanism of action which, nonetheless,
20 affects sodium channels.
21 So we make a lot of different chemicals
22 that are sort of, by definition, neurotoxic or have
23 neurotoxic potential in animals anyway because of
24 their known mechanism of action.
25 I'm going to be talking primarily about
NEAL R. GROSS
(202) 234-4433
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
WASHINGTON. D.C. 20005
(202) 234-4433
-------
95
1 theory, very little data here, in fact, no data that
2 I've produced myself, some tiny amount of data from
3 the literature; but it's really the ideas about
4 cumulative risk assessment that I would like to talk
5 to you about, particularly in the context of the Food
6 Quality Protection Act and how cumulative risk
7 assessment is envisioned in the context of the Food
8 Quality Protection Act.
9 Basically, the steps in cumulative risk
10 assessment are fundamentally no different, I think,
11 from those in any kind of a risk assessment analysis.
12 There's four fundamental categories of activity.
13 First is hazard identification. There's
14 dose response characterization, exposure estimations,
15 and finally the risk assessment itself where all that
16 information is pulled together. The nature of the
17 events or the specific events within each of those
18 categories are a little bit different, however, for
19 the cumulative assessment process.
20 First of all, the hazard identification
21 step -- really, the analog of that in cumulative risk
22 assessment has to do with deciding what the common
23 mechanism of action is for the chemicals. It is
24 generally decided that something like inhibition of
25 the cholinesterase enzyme is a common mode of action,
NEAL R. GROSS
(202) 234-4433
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
WASHINGTON, D.C. 20005
(202) 234-4433
-------
96
1 and that itself poses a hazard.
2 So that kind of hazard determination,
3 while there's a lot of discussion about that and what
4 needs to do into that, is typically -- When that
5 determination about the common mechanism of action or
6 common mode of action is made, that's essentially the
7 hazard identification step associated with the group
8 of chemicals.
9 The dose response characterization step
10 really becomes an issue in deciding what the identical
11 critical endpoint in the same species for all
12 compounds is, and those are kind of ideal kind of
13 statements. It would be very, very -- It would be
14 best is, for all the compounds in the risk cup, if the
15 same endpoint were used. So red cell, cholinesterase
16 inhibition in female rats, for example, is common
17 critical endpoint.
18 That's not always going to be possible for
19 certain groups of compounds, for pyrethroids, for
20 example. One may have a more sensitive endpoint which
21 might be muscle fasciculations in one case, and it
22 might be a mild limb tremor in another. You have to
23 decide whether those kinds of endpoints are similar
24 enough to be counted as the same thing and do the
25 modeling and the risk evaluation activities off those
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON. D.C. 20005 (202) 234-4433
-------
97
1 kinds of qualitatively similar or probably
2 mechanistically related sort of data points.
3 Dose normalization is something that
4 happens, is absolutely necessary to do within
5 cumulative risk assessment, and it's blue here because
6 that's what I'm going to spend most of my time talking
7 about today or at least one of the things I'm going to
8 spend most of my time talking about.
9 Exposure estimation is a huge area. In
10 fact, all these things are huge areas and require a
11 lot of discussion, and I'm only going to talk about a
12 few of them today. But basically, the way exposure
13 estimation works is some kind of a Monte Carlo routine
14 which assembles data associated with the foods people
15 really eat, the kinds of pesticides that are used on
16 those foods, the kinds of quantities of pesticides
17 that potentially occur on those foods, and the kinds
18 of other activities that people get involved in,
19 because the cumulative risk assessment process
20 involves both an aggregate and a cumulative risk
21 analysis, all combined into one thing.
22 Then, of course, there is the addition of
23 the normalized dose units, and I'm going to spend some
24 time talking about that process.
25 There's some knowledge of how we are going
NEAL R. GROSS
(202) 234-4433
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
WASHINGTON, O.C. 20005
{202) 234-4433
-------
98
1 to apply uncertainty factors to the analysis. I'm
2 sure everyone is familiar with the way uncertainty
3 factors are used in risk analysis or risk assessment
4 for a single compound. The way those uncertainty
5 factors are used, I think, is a little bit different,
6 and certainly it's more complicated in the context of
7 a group of compounds that share a mechanism of
8 toxicity.
9 I'm going to spend some time talking about
10 the time frame, the appropriate time frame over which
11 exposures ought to be considered for addition, and
12 certainly, the kinetics of the individual compounds in
13 the risk cup matter in that context.
14 Finally, the risk assessment itself in
15 which all the data are assembled together and some
16 kind of an outcome is handed over to a risk manager.
17 Now my personal opinion is that the kind of
18 information that the risk manager should get is the
19 frequency of their expected or potential
20 potentially expected frequency of adverse effects in
21 the population.
22 So many people out of a large number, like
23 265 million, might be expected to experience an
24 adverse effect due to the presence of this combination
25 of pesticides used in the marketplace.
NEAL R. GROSS
(202) 234-4433
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
WASHINGTON, D.C. 20005
(202) 234-4433
-------
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
99
While I think that's kind of where we
should be driving, the analysis, as I understand where
it's going, is a little bit different than that in
that it basically is a more sort of a point analysis.
That is, well, this group of pesticides has an
unacceptable margin of exposure. So we have to do
something about that.
I think we can make a more intelligent
decision if we knew something about the actual
frequency of adverse effects and are actually worried
about calculating that kind of value.
As I said, all these are very long
conversations in themselves, and I'm simply going to
focus on the ones highlighted in blue here.
First of all, dose normalization:
Essentially, this is the theory underlying justifying
how we are going to be adding doses together. So it's
worthwhile to talk about that theory.
Essentially, on the top figure there on
the right we have two pesticides, and let's say they
are cholinesterase inhibitors and the experiments are
similar, say 90 day rat studies in which we have
sampled red cell cholinesterase inhibition, and we
have a dependent variable there listed on the
abscissa, and -- no, let's see, ordinate. The
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202)234-4433 WASHINGTON. D.C. 20005 (202)234-4433
-------
100
1 dependent variable on the ordinate is -- let's say
2 that's cholinesterase inhibition. This is clearly
3 totally invented numbers that fit a very pure
4 mathematical function, in this case the logistic
5 function, which is fit to them.
6 The two compounds have different potency
7 in this regard. The compound on the left is ten times
8 more potent than the compound on the right. So how
9 are we going to add exposures of these things
10 together?
11 Well, we have to normalize the doses --or
12 we have to normalize them somehow to make them appear
13 as though they had comparable potency. So there is an
14 adjustment that's done.
15 The one that's shown here on the lower
16 panel, the left figure in the lower panel, shows what
17 happens if we compute the EDSO for the compound on the
18 left and divide all the doses used in the actual
19 experiment by that EDSO. We can express the dose for
20 that chemical in EDSO units.
21 So, two, we take the ED50 of the compound
22 on the right and divide all the doses in that
23 experiment by the EDSO, and then we express the doses
24 then in ED50 units. So then we can pull those two
25 things together, and in the lower lefthand figure
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
101
o
l there you see what happened.
2 The data points are probably maybe,
3 unfortunately, too small to see, but the four data
4 points for the two chemicals making eight data points
5 altogether now appear on a single merged superimposed
6 function, when the doses are expressed in common
7 xinits, in this case common ED50 units. The dose
8 response functions are perfectly superimposable.
9 Indeed, this is an assumption of the
10 analysis, that the dose response curves are parallel.
11 parallel is really the wrong word, because they are
12 not linear functions. They are sigmoidal functions or
13 curvilinear functions, and in this case the best way
14 to -- more accurate way to describe it is the dose
15 response functions share the same slope parameter, but
16 I think you know what I mean. The dose response
17 functions are parallel.
18 Under those conditions, expressing the
19 dose according to a common effect level, in this case
20 ED50, has the effect of superimposing the dose
21 response curves.
22 Now in this case, what happens is that the
23 -- one can take any number of EDSO units of any of the
24 chemicals in the mixture and add them to another one,
25 and that result, that sum, will lead to a predictable
NEAL R. GROSS
(202) 234-4433
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
WASHINGTON. D.C. 20005
(202) 234-4433
-------
102
1 effect. So for example, one could take .1 EDSO units
2 of chemical A and .1 EDSO units of chemical B and .1
3 plus . 1 is .2, and we can simply march up the dose
4 response function in the lower lefthand panel and say,
5 okay, where does .2 ED50 dose units bring us in terms
6 of the predicted effect.
7 Perhaps it would be better to deal with a
8 slightly larger set of numbers like, let's say, we had
9 -- If we had .5 ED50 units of compound A plus .5 ED50
10 units of compound B, the total is 1 EDSO unit. We
11 would expect 50 percent inhibition in that case.
12 This is true regardless of the number of
13 the chemicals in a risk cup. If there are ten or 34
14 chemicals in the risk cup, if we're adding up and they
15 all have the same slope and all have the same parallel
16 dose response functions, when we normalize the doses
17 according to a common -- the dose associated with a
18 common effect level, that has the effect of
19 superimposing the dose response curves and making it
20 justifiable that we're adding doses to lead to a
21 common effect.
22 This is fundamentally what we're doing in
23 any kind of a dose addition cumulative risk analysis.
24 We don't always say we're marching up the dose
25 response curve like that, because very often we are
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-M33 WASHINGTON. D.C. 20005 (202) 234-4433
-------
103
l looking for some kind of a point threshold, whether
2 you exceed it or not, that might be important. But
3 essentially what we're doing is adding doses and
4 tracking up the dose response curve.
5 There is a tendency to say, well, we can't
6 really express doses on EDSO because the data aren't
7 good enough to actually get our handle on a good
8 estimate of the ED50 for all the compounds in the risk
9 cup. That's probably true. Probably the actual
10 regulatory data that you have in your hands that are
11 submitted to you by companies probably is not always
12 good enough to produce an accurate estimate of the
13 ED5C or an ED10, for that matter.
14 So it is tempting and probably necessary
15 as a practical matter to, instead of using a common
16 effect level like ED10 or EDSO and to normalize dose
17 units, rather instead to use the no effect level of
18 the two compounds. That's shown here.
19 So the experiment here has a no effect
20 level of .1 mgs per kilogram. The compound on the
21 right has a no effect level of, looks like, 3 mgs per
22 kilogram, and then we could normalize the doses by
23 dividing each of the doses in the experiment by the no
24 effect level dose. Now we're expressing doses in
25 terms of no effect units, NOEL units, basically.
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
(202) 234-4433 WASHINGTON. D.C. 20005 (202) 234-4433
-------
104
1 So one NOEL unit in both cases is a no
2 effect level for both compounds. Note that what
3 happens, because of the arbitrariness of how the
4 experiments are conducted and dose selection and all
5 the different kinds of rationales that go into exactly
6 how the experiment is conducted and statistics and
7 everything else, when such a transformation of dose is
8 made, the dose response curves are no longer -- are
9 not superimposable.
10 In essence, when we're doing that, now
11 when we're adding up NOEL units of compound A and
12 compound B, we don't really know which line to track
13 up to know what the predicted level is going to be,
14 and that's simply a consequence of the fact that the
15 functions are not superimposable like they are on the
16 left.
17 So for this reason, it's technically
18 inappropriate to use something like a NOEL, but
19 obviously, until we get better data to the agency, I
20 think we are probably going to be forced as a
21 practical matter to do that.
22 Now on the positive side of doing that,
23 probably the errors are not going to be too great,
24 because where we are located down here on the low end
25 of the dose response curve is -- things are going to
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
(202) 234-4433 WASHINGTON. D.C. 20005 (202) 234-4433
-------
105
l be reasonably similar to each other most -- well,
2 we're pretty sure that doing such an action certainly
3 would not put the public at risk. It's simply not
4 totally scientifically appropriate to do it.
5 There are other issues pertaining to non-
6 parallel dose response functions. I was in a hallway
7 conversation earlier: What do you do in the case
8 where the dose response functions are not parallel?
9 It's an interesting problem, because --
10 Well, technically, we really shouldn't be adding the
11 doses together in that case. As a practical
12 resolution to that, instead of comparing -- or
13 normalizing doses around EDSO for example, what one
14 might do there would be to normalize doses around a
15 lower effect level such as an ED10, and now express
16 the effects in terms of ED10 dose units.
17 The reason for doing that is because, when
18 non-parallel dose response functions are normalized
19 around an EDSO, essentially they become superimposed
20 at the inflection point of the sigmoidal dose response
21 function around the EDSO point. So one could imagine
22 one compound having that slope and then the normalized
23 function intersecting with that, but following that
24 sort of trajectory.
25 What that does is displace things on the
NEAL R. GROSS
(202) 234-4433
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
WASHINGTON, D.C. 20005
(202) 234-4433
-------
106
1 low end a lot more than if one were to normalize
2 around an ED10 in which case the compounds of
3 dissimilar slopes are now fairly close together on the
4 low end where we are actually going to be adding
5 things up where things really matter, and are more
6 disparate at the higher dose levels where people
7 aren't really exposed to those kinds of levels anyway.
8 If we get up to 50-60 percent inhibition, we know we
9 are close to that in human beings. So that would be
10 how to handle that problem.
11 Well, in kind of the real world, how does
12 this work, this idea of normalizing dose response
13 functions to something like an EDSO? Here we have
14 seven organophosphate cholinesterase inhibitors. I
15 just labeled them A,B,C, and so on here. But in fact,
16 you know, they are things like chlorpyrifos and your
17 basic list of typically used organophosphates.
18 These are dose response functions for red
19 blood cell cholinesterase inhibition, all taken from
20 90-day rat dietary studies. You see here that the
21 seven compounds have -- They vary almost a factor of
22 100 in potency. In general, the dose response
23 functions are fairly similar in slope.
24 They are not so markedly different than
25 one would say those are really different. Certainly,
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
(202) 234-4433 WASHINGTON. D.C. 20005 (202) 234-4433
-------
107
l they are not statistically different, although
2 interestingly, there are two groups of disparate --
3 two groups of different slopes here, but they are not
4 that different. Actually, I think it's the ones with
5 the white data points that mathematically end up
6 having slightly less steep slopes than the others.
7 Let's go through that exercise of now
8 determining the EDSO for each of these compounds and
9 expressing the doses of that compound in EDSO units
10 and going on through that entire process for all seven
11 compounds, and then assembling all the data together
12 in the same graph and seeing what happens.
13 That's what's going on in panel B down
14 here in which we have percent inhibition of
15 cholinesterase as a function of dose expressed in EDSO
16 units. You can see that the data points from the
17 seven different experiments all pretty much fit
18 reasonably nicely a logistic curve function, and here
19 are the parameters associated with that logistic
20 function.
21 One can compute a 95 percent confidence --
22 set of 95 percent confidence intervals associated with
23 that logistic function, and from that 95 percent
24 confidence interval one can determine that an ED20, a
25 dose that causes 20 percent inhibition of red cell
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, O.C. 20005 (202) 234-4433
-------
108
o
l cholinesterase, is 0.34 EDSO units, and that's the
2 intersection of the upper bound of the 95 percent
3 confidence limit with the 20 percent inhibition level.
4 So regardless of the OP that we have here
5 that differ by a factor of 100, in general 0.34 ED50
6 units, whatever the EDSO of the particular compound
7 is, is going to produce about 20 percent inhibition.
8 If one does not like ED20 as some kind of
9 a point of departure for risk analysis, you might like
10 ED5 or ED10 or some other number. What I'm just
11 showing you is simply a matter of whatever number you
12 like. If you like ED10, you can determine what the
13 intersection of the upper bound of the 95 percent
14 confidence limit with the ED1D is, and determine that
15 that is 0.2 ED50 units.
16 Once you have that number, then one can
17 sort of very readily move forward with a dose addition
18 j process. So when you know that an apple has X EDSO
19 units on it, then you could add that to Y EDSO units
20 of a pear -- of a pesticide that's on a pear, a
21 different pesticide that's on another piece of food,
22 for example.
23 So I just mentioned the process or the POD
24 -- that is, the point of departure from the
25 experimental data. I simply want to define that for
NEAL R. GROSS
(202) 23*4433
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
WASHINGTON. D.C. 20005
(202) 234-4433
-------
109
1 you. That's the number from which all further
2 analysis is going to occur, and whether that point of
3 departure is a NOEL or an ED10 or an ED20, we're simply
4 going to call that a point of departure to have some
5 language that's useful in the analysis.
6 We need to have some kind of a measure or
7 assumed or assigned values associated with the
8 pesticides that people consume. So we're going to
9 have to know the amount of pesticide residues on foods
10 or be able to estimate that in some way.
11 We need to know how much food is consumed
12 by an individual and the kinds of foods that they eat,
13 and when they eat it; and we certainly need to know
14 their body weight.
15 Basically, these kinds of things occur in
16 the context of a Monte Carlo algorithm that assigns
17 them based on distributions of their occurrence in the
18 real world. Then we're going to go through a process
19 of computing the normalized units of exposure for each
20 of the compounds.
21 So a person might eat an apple that has 12
22 tnicrograms per kilogram of chlorpyrifos on it, and
23 we're going to have to convert that amount of
24 chlorpyrifos to normalized exposure units and then
25 determine what the dose in normalized exposure units
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
110
l for that compound on that apple are for that
2 individual.
3 This whole process can -- You know, now we
4 have to express the cumulative exposure, and there are
5 a number of different ways of getting there. The
6 mathematical -- Well, the arithmetic -- It's not
7 mathematics, really; it's arithmetic -- by which one
8 expresses these cumulative exposures can look really
9 different, depending on which kind of metric one
10 chooses.
11 One way to do it is to calculate a margin
12 of exposure. Another way is to calculate the
13 milligram per kilogram equivalents of a particular
14 index chemical, and there's a lot of interest in
15 saying that chlorpyrifos, for example, is an index
16 chemical. It's certainly a widely used chemical
17 that's been fairly well studied, and we can express
18 all other organophosphate cholinesterase inhibitors in
19 the same -- as milligram per kilogram chlorpyrifos
2 0 equivalent s.
21 Alternately, we can say how that exposure
22 fills up the risk cup. So we have a risk cup which is
23 allowed to hold 100 percent of one's permitted
24 allotment of organophosphate cholinesterase inhibitor,
25 and as one eats more foods that contain more different
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-M33
-------
Ill
1 kinds of pesticides, more different kinds of OPs,
2 we're filling up that risk cup.
3 Arithmetically, the calculations look like
4 quite different. Mathematically, they are absolutely
5 identical. So it doesn't really matter if we're
6 talking margin of exposure or relative potency factors
7 in milligram per kilogram equivalents or we're filling
8 up a certain percentage of the risk cup. The numbers
9 are quantitatively exactly the same, because there is
10 a mathematical identity underlying all those
11 computations.
12 The reason that the answers might be
13 different is if one is not careful about handling the
14 uncertainty factors that figure into the equation. If
15 one handles the uncertainty factors in the same way,
16 all the values will come out exactly the same. They
17 are all directly translatable to each other in a
18 precise quantitative way.
19 Well, let's go through a hypothetical
20 exposure scenario here, just to show you how this
21 might work in the context of a margin of exposure kind
22 of analysis, and I've just made up these numbers.
23 These numbers have no meaning at all.
24 First of all, we are going to define the
25 margin of exposure as the rat POD, the normalized rat
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON. D.C. 20005 (202) 234-4433
-------
112
1 POD, divided by the human exposure, the normalized
2 human exposure, sort of expressed in the same kind of
3 unit.
4 So a rat POD might be one - - or . 1 EDSO
5 units. Actually, I said it was .34 EDSO units before
6 when I was going through my exercise, and that would
7 be a .34 ED50 units provides 20 percent red cell
8 cholinesterase inhibition in the rat.
9 Well, what is the actual human exposure on
10 that apple, for example? What is the normalized human
11 exposure? What is the ratio of those two? That's the
12 margin of exposure. Clearly, the higher that number,
13 the larger the difference is between what it takes to
14 affect a rat and what the human receives.
15 We can calculate an MOE for the entire
16 chemical, and that's simply done by adding the
17 reciprocal MOEs and taking the reciprocal of that. In
18 this particular example, I've had a person eating --
19 On a given day, I had them eat one kiwi fruit, 12
20 grapes, four slices of bread, two slices of pizza, ten
21 ounces of cornflakes, 16 ounces of green salad, l
22 ounces of spaghetti with tomato sauce, and 14 apples.
23 Then we have chemical A, B, C and D. So
24 if we want to determine what the MOE for a particular
25 chemical is, we compute what the MOE is for the kiwi
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
113
1 fruit, what the MOE is for the apples, what the MOE is
2 and so on. Then we can figure out what the total MOE
3 for the chemical is.
4 In this case in my hypothetical example
5 here, the person has consumed -- has a 386 margin of
6 exposure. That would -- Whether that's safe or not or
7 harmful or not depends on where the uncertainty
8 factors are associated with that and whether there
9 will be some kind of a threshold above which or below
10 which we would no longer regard that exposure as safe.
11 In general, at this point in time and
12 probably it will change soon, is changing as we speak,
13 right now an acceptable margin of exposure might be
14 regarded as 100. That incorporates two uncertainty
15 factors, animal to human, uncertainty associated with
16 the animal to human extrapolation, and a factor of ten
17 for variation in sensitivity within the human
18 population.
19 So ten times ten is 100. That number
20 ! might go up if, one has more uncertainties about
21 things like unstudied effects in developing animals,
22 but for now let's just say it's 100. So 386 is a
23 larger value than 100. So one would say, well, the
24 MOE for that chemical, that particular chemical A, is
2 5 okay.
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
114
1 So we go through the process and simply
2 now look at the exposures associated with these
3 different foods which have chemical B, C, D and E and
4 so on, and indeed this would actually extend quite a
5 bit further. There are 34 OPs registered for use in
6 the United States, and one could imagine -- You know,
7 one has to deconstruct things like pizza. Well, pizza
8 has tomato sauce. It might have green peppers and
9 onions on it. It has dough, flour, which is grain
10 which is -- pesticides are used on that.
11 A slice of pizza might have five or six
12 different pesticides on it altogether. Spaghetti is
13 made with flour as well. It's got tomatoes in it. So
14 a particular food -- and the Monte Carlo algorithm
15 does all this deconstruction of foods based on
16 standardized data that comes out of the -- I think
17 it's USDA.
18 So once one figures out how the values
19 ought to be assigned -- and here the exposures are
20 expressed in MOE units -- one gets an MOE for each
21 chemical. Then one has to determine what the total
22 cumulative margin of exposure is, and that's a similar
23 kind of equation except now we're adding up the MOE
24 values or adding up the reciprocals of the MOEs to get
25 the cumulative margin of exposure.
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON. D.C. 20005 (202) 234-4433
-------
115
1 In this particular example, the cumulative
2 MOE is 54. That is less than 100 or we've filled up
3 184 percent of the risk cup. This particular
4 combination of events would probably be regarded as
5 unsafe.
6 That is really just one iterative step in
7 the Monte Carlo routine that assigns pesticides and
8 pesticide levels to foods that are also assigned to
9 individuals. That iterative process is executed
10 10,000 times or more, and from that, the total MOE,
11 one obtains then the distribution of MOE values.
12 Here I've simply invented a distribution
13 of MOE values to show you how this might work, in
14 which case we have the average, in this case, total
15 MOE is on the order of 900 or so. So the average
16 person is probably pretty safe. But there is a small
17 group of people down here --in fact, it's .82 percent
18 of the population -- that might be expected to have an
19 MOE less than 100.
20 The risk manager getting this information
21 would then want to decide whether something needs to
22 be done about this. He can either look at the events
23 that went into that .82 percent of the population,
24 what was it that they were doing? Were they eating
25 apples? Was it pizza that did it? You need to
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
116
1 maintain the linkages in the Monte Carlo routine to
2 get back to the specific chemical events that led one
3 there.
4 So that's kind of how the process works
5 now, as we understand it. I want to take a short
6 aside here and mention that the -- about getting to
7 the frequency of the population that's affected.
8 This is right now what would go to the
9 risk manager. .82 percent of the population has an
10 MOE of less than 100; what do we need to do about
11 that? I think it might be worthwhile if we take that
12 extra step and try and figure out what the frequency
13 of the affected individuals in the population is.
14 To know that, it's not .82 percent of the
15 population that's affected. I mean, that's a lot of
16 people in a country the size of the United States.
17 Let's remember that the underlying assumptions that go
18 into the risk factor analysis -- or the uncertainty
19 factor analysis in this -- One of them, for example,
20 there's a tenfold variation in human sensitivity. How
21 frequent is that sensitive person in the population?
22 The bit of thinking I've done about that
23 problem, it's actually a pretty rare individual. It's
24 certainly as rare as one in 1,000 persons is going to
25 be ten times more sensitive than the average person.
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
117
1 It's probably more like one in a million people is
2 going to be ten times more sensitive than the average
3 person. That's based on an analysis of a lot of
4 pharmacological data, because the drug companies have
5 a lot of information on that.
6 So given your standard drug, what is the
7 variation introduced -- or in human sensitivity to
8 drugs? Those numbers are on the order of a factor of
9 ten has a probability of something like one in a
10 million or certainly more than one in 1,000 or less
11 than one in 1,000, probably something more like one in
12 a million.
13 So when we're looking at .82 percent of
14 the population, that's .82 percent of the population
15 that's one in 1,000 or one in a million in itself. So
16 it's a very small fraction of a very small fraction.
17 We're not talking about millions of people being
18 affected by this .82 percent. It's quite a bit less
19 than that.
20 Okay. In a few minutes I'll say something
21 about the kinetics and trying to get to an analysis of
22 the time frame that one needs to go through in this
23 process.
24 If our data were based on that 90-day
25 study -- remember, that's a repeated dose, 90-day
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON. D.C. 20005 (202) 234-4433
-------
118
1 study; the animals receive that dose every day Cor 90
2 days, and we determine what the POD or the NOEL was in
3 that experiment.
4 What I just went through with the Monte
5 Carlo analysis really involved just a single day --
6 Okay? -- analyzing the dietary life in a single day
7 of this hypothetical individual and generating a
8 normalized distribution in that single day. We ended
9 up concluding that .82 percent of the population in
10 the hypothetical example might have had a problem.
11 Okay. Does that .82 percent of the
12 population really have a problem, because in fact
13 that's just happening on one day? Remember, that
14 individual was eating 12 apples or 14 apples, was
15 eating a ton of spaghetti, very, very large salad, and
16 I guess, just sort of apply some common sense here.
17 How many days in a row are you going to eat 14 apples?
18 I submit, not many.
19 So what we really need to do is to start
20 thinking about how those daily exposures really occur
21 in time. If one is at the 99.9 percentile of the
22 exposure distribution on one day because they eat 14
23 apples, is it likely they are going to be there on the
24 next day and the next day and the next day; because,
25 let's face it, in the context of a long acting set of
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
119
1 compounds like organophosphates what's going to happen
2 is that it's the daily exposure that accumulates over
3 time that builds up to that sustained daily exposure
4 that causes the cholinesterase inhibition to build up
5 to a particular level.
6 I've modeled this process extensively.
7 The modeling -- The details of the modeling are
8 important if you were to accept my quantitative
9 conclusions, and I really don't have time to talk
10 about sort of the detailed model here. But again just
11 sort of -- It's the qualitative conclusion which is
12 the take-home message here.
13 So I think what we really need to do here
14 is to just ask, well, what is the accumulated
15 cholinesterase inhibition? Here we have percent
16 inhibition as a function of time in hours. Zero to
17 1500 hours is two months.
18 So a compound like chlorpyrifos which has
19 an acute half-life of about 175 hours, one consumes a
20 little bit one day one. Then that produces some
21 cholinesterase inhibition which decays with a half-
22 life of 175 hours. On day two, another amount of
23 chlorpyrifos is added to the diet. That is going to
24 increment the amount of residual cholinesterase
25 inhibition that occurred from the day before, and so
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
120
1 on and so on and so on.
2 Clearly, you can model this process.
3 That's what I've done here to generate this kind of
4 function. This basically describes the progression of
5 cholinesterase inhibition in a rat administered
6 chlorpyrifos every day over a period of two months.
7 One can see that it takes about
8 actually, about a month to get to 95 percent of the
9 steady state. So that's an animal that's eating one
10 POD unit -- That rat is eating one POD unit to get --
11 and POD is defined as the amount it takes to get to 20
12 percent inhibition in a subchronic study. That rat --
13 By eating that one POD unit every day for two months,
14 in fact, that's what it takes to build up to that
15 level.
16 Now what if the person were to eat -- not
17 person -- that rat were to eat one POD unit on a
18 single day, never had any on any other day? Well,
19 that's shown right here. In fact, what that does is
20 produce a very small two percent change in the
21 dependent variable.
22 Basically, the way the risk analysis is
23 set up now, as I understand it, we're basically
24 treating that two percent effect as if it's a 20
25 percent effect, because we are only focusing on one
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
121
l day of exposure. We need to focus on the multiple day
2 to day to day exposure scenario, because that's the
3 experiment that's driving our analysis, not a single
4 day of exposure.
5 So we need to open up that process. In
6 fact, if you compute what it would take for a single
7 day exposure to get to 20 percent cholinesterase
8 inhibition, which is kind of where our hypothetical
9 point of departure is for this experiment, it takes
10 four and a half POD units.
11 So one POD unit might be the amount of
12 pesticide residue that's, say, on 20 apples. If you
13 are never exposed to chlorpyrifos, you could probably
14 eat 4.5 times 20, some 90 apples, on a particular day,
15 and you would still be below the threshold that might
16 be regarded as adverse.
17 So it's simply by examining the kinetics
18 and modeling that process, one can get at least a
19 qualitative flavor that we need to focus on, land on
20 more than a single day of exposure.
21 Clearly, the individual kinetic properties
22 of the compounds in the risk cup make a difference.
23 It takes a different amount of time to get to the
24 steady state level. Here we have three different OPs.
25 A compound like azinfosmethyl has a half-life of 35
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
122
1 hours. Chlorpyrifos is 175. Chlorthoxifos is 125
2 hours, and you can see that these different kinetics
3 mean that steady state is established after different
4 periods of time.
5 For chlorpyrifos it takes about 30 days.
6 Other compounds with quicker half-lives will get there
7 quicker. In a cumulative analysis the time frame that
8 we need to sort out is equal to that which has the
9 longest -- well, the compound with the longest half-
10 life in the analysis.
11 So probably what we would be needing to do
12 in the case of organophosphates is to focus on 30 days
13 of dietary -- daily dietary consumption and try and
14 figure out what the process over 30 days looks like.
15 What would happen there? What's the right
16 metric. Once we've determined that the right time
17 frame for OPs is something on the order of 30 days,
18 what kind of metric do we do within that 30 days?
19 Do we choose the very, very highest value that occurs
20 in 30 days or do we choose the mean value?
21 Well, the mean value is certainly a lot
22 closer to the one we should be using rather than the
23 P. Here is an example of that. I've gone ahead and
24 now computed what the cholinesterase inhibition, the
25 expected cholinesterase inhibition would be over time.
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
123
1 Here we have cholinesterase inhibition over two months
2 of continuous administration or daily administration
3 anyway.
4 Here on the graph on the right we have the
5 exposure units that go into this. Again, I've just
6 programmed my computer to make up values to introduce
7 some variation in them. The average over this 30 day
8 period is about .39 POD units. So that's like .39 of
9 a NOEL is administered to this rat here.
10 Now under those conditions where one
11 receives kind of an average of .39, you see that we
12 build up to a steady state which looks like about
13 three percent inhibition here under those conditions,
14 with even a spike of activity I forced in here of two
15 and half POD units. That would be like going from
16 the 20 apples to 50 apples. So eating 50 apples on a
17 day, one still does not get anywhere close to the
18 level of cholinesterase inhibition that would be
19 regarded as adverse.
20 Now turning this kind of analysis to the
21 more cumulative case, in this case we have three
22 different OPs, A, B, and C. Here is the exposure
23 scenario. Everything is randomized, so different
24 amounts of compound A on every day of the two months,
25 different amounts of compound C on every day of the
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
(202) 234^433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
124
1 two months, and different amounts of compound B in
2 every day of the two months.
3 During this two month exercise of daily
4 diet, you can see here -- Let's just focus on the last
5 month of exposure here. We have one, two, three,
6 four, five, six, seven, eight, nine -- nine out of the
7 30 days in which this animal in this -- While this is
8 modeled after animals, one could think about this as
9 the one in 1,000 person that's more sensitive than the
10 average person and also assuming that the average
11 person is ten times more sensitive than the average
12 rat. Under those kind of conditions, you can think
13 about these data as being a human exposure scenario.
14 So nine out of these last 30 days contain
15 an exposure which is -- in the present context, would
16 be regarded as excessive, but the degree of cumulative
17 inhibition is just getting to the point where it's 19
18 percent inhibition. It does not exceed our threshold
19 of 20 percent.
20 So even though we've busted the line on
21 exposure, the process of actually modeling the effect
22 according to using more of an average like .8 -- the
23 average of these values are less,- 30 days is .8 POD
24 units, which brings us right there.
25 So, clearly, I don't think we should be
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
(202) 234-4433 WASHINGTON. D.C. 20005 (202) 234-4433
-------
125
1 using an individual daily peak which is driven by rare
2 and unusual events and with long acting compounds
3 don't have the ability to have that immediate punctate
4 reaction or impact on the system.
5 On the other hand, using the mean itself
6 is probably a little bit too -- not conservative or
7 not conservative enough, because if we look at this
8 for our mean over this last month is .8, and we're
9 really driving right up to a level that we would
10 regard as -- we would start regarding as dangerous.
11 Certainly, if the mean were 1, we would be
12 over that danger level, and we probably wouldn't want
13 to do that. So the mean isn't quite the right answer.
14 The mean exposure, the mean daily exposure over that
15 30 days is not quite the right answer. It's got to be
16 something less than that to be safe.
17 Exactly how much less it should be, I
18 think, kind of relates to the inherent variation, how
19 often excursions and large spikes and unusual events
20 can occur,- and we've got some ideas about how to
21 approach that.
22 My point in bringing this out here is to
23 stimulate folks in the agency to think about the right
24 way to solve this problem. The analyses I've done
25 suggest that the right answer is not to use the mean
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
126
1 but rather the mean plus a fraction of the standard
2 deviation, and the fraction of the standard deviation
3 -- the amount of the fraction depends upon the
4 individual kinetics, the individual half-life of the
5 individual compounds in the risk cup.
6 In fact, that sort of analysis works, and
7 it would be a useful way to go about protecting the --
8 both protecting the public and preventing the analysis
9 from being more conservative than it needs to be.
10 So to summarize here, we've talked about
11 the process of dose normalization. I've urged you to
12 recognize the kinetic difference among chemicals in a
13 risk cup can be as important as potency differences.
14 Simply because we' re normalizing compounds
15 according to potency doesn't mean they are equivalent
16 compounds. Very large kinetic differences can occur.
17 That's especially true if we're dealing with something
18 like carbamates which have a half-life on the order of
19 30 minutes or 45 minutes, and adding them in with
20 something like OPs which have half-lives on the order
i
21 ! of 100 or 150 hours.
22 So we need to be very aware of those
23 kinetic differences and factor them into the analysis.
24 We need to select the correct time frame for the
25 compounds in the risk cup, and my suggestion is to use
NEAL R. GROSS
(202) 234-4433
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
WASHINGTON. D.C. 20005
(202) 234-4433
-------
127
1 something like four half-lives of the longest acting
2 compound in the risk cup. For a group of compounds
3 like OPs, that would be on the order of four weeks.
4 Finally, use a realistic exposure metric
5 that accumulates temporally separated exposures in a
6 manner consistent with the kinetic properties of those
7 chemicals. Thank you very much.
8 (Applause.)
9 DR. NORRIS: Do you have some questions?
10 The other thing we can do is briefly have all the
11 speakers come up and start the panel discussion now,
12 since we are running a little bit behind. Then if you
13 would just stand right there, you can ask your
14 question when they get here.
15 Okay, we are ready. Thank you.
16 QUESTION: The example you used where you
17 were looking at the cumulative risk from the OPs, you
18 used red blood cell cholinesterase inhibition. I
19 think that people would generally agree that
20 cholinesterase inhibition and toxicity is perhaps
21 measured by other parameters, signs; and other
22 anticholinesterase type --or cholinergic symptoms is
23 not necessarily correlated with the red blood cell
24 cholinesterase inhibition.
25 My question to you is: You had compounds
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
(202) 234-4433 WASHINGTON. D.C. 20005 (202) 234-4433
-------
128
1 there from your company, and I'm just wondering when
2 you ranked -- You ranked the potencies -- With what
3 you did, you ranked potencies by cholinesterase
4 inhibition. But if you looked at the same chemicals
5 in terms of what you know about their toxicity in
6 terms of clinical signs and cholinergic type
7 responses, were the rankings similar or did you find
8 big differences in potency if you would look at
9 toxicity by that other way?
10 The reason I'm asking that is because, of
11 course, the agency is going to have to look at risk
12 mitigation issues, and we are going to be getting
13 chemicals from different companies. So one chemical
14 could emerge from a process such as what you have
15 looking very bad in terms of cholinesterase inhibition
16 but not necessarily seeming to make any sense by other
17 criteria of toxicity, i.e., clinical signs and that
18 sort of thing.
19 DR. CHRISTOPH: Yes. I focused on red
20 cell cholinesterase inhibition, because that's where
21 the agency seems to be going. My personal opinion is
22 that we should focus on brain cholinesterase
23 inhibition, but that's just what I think. It just
24 makes more sense to me, because basically we're
25 talking about red cells, and essentially the red cell
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
129
1 cholinesterase inhibition is really a biomarker of
2 exposure, to me, more than an adverse effect.
3 The rationale, of course, is that it's
4 correlated with an adverse effect elsewhere and is a
5 surrogate for peripheral cholinesterase inhibition.
6 I think Stephanie would be a more appropriate person
7 to comment on that.
8 I really think -- You know, there was, of
9 course, a huge discussion about what's adverse in
10 terms of cholinesterase inhibition, and I think the
11 agency has just decided that it is cholinesterase
12 inhibition itself that is. In my mind, the thing
13 that's adverse about it is not inhibition of the
14 enzyme. It's the consequences of inhibition of the
15 enzyme, the consequences of the kinds of things you
16 were talking about, signs.
17 Well, in fact, those are fairly
18 insensitive. The things we use to measure those are
19 our eyeballs, for the most part, and the kinds of
20 things that would occur that are pretty easy to see up
21 to really high concentrations in high levels of
22 cholinesterase inhibition, things like tremor and so
23 on, are simply not particularly sensitive instruments
24 to measure those kind of things.
25 In my mind, the real adverse effect is
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON. D.C. 20005 (202) 234-4433
-------
130
1 disruption of the fidelity of neurotransmission.
2 That's kind of the bottom line on this. At what point
3 is there so much acetylcholine in the synapse that the
4 fidelity of neurotransmission is degraded?
5 That endpoint, no one is measuring, to my
6 knowledge. I guess that's -- I would like to see work
7 focused on that kind of problem.
8 So the nature of your question was how do
9 these other kinds of behavioral things correlate? In
10 fact, we really haven't studied that very much or
11 studied that particular question very much, because,
12 frankly, it's not important to us from the context of
13 regulations. The agency drives the regulations, and
14 the agency has decided that red cell cholinesterase
15 inhibition is where we're going to be driven.
16 So I'm sorry that I can't offer a more
17 detailed answer than that.
18 (INAUDIBLE QUESTION)
19 DR. CHRISTOPH: I really can't answer the
20 question in detail. I'm sorry. I don't know.
21 QUESTION: Yes. I think, for the most
22 part, those of us who are in the business of the risk
23 mitigation part of it, and not to cloud the issue, but
24 when you speak of that one in a million individual
25 with higher levels of sensitivity or one in 1,000, if
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
131
l we break it down even further and we correlate or we
2 use an example of children one to six, I think for the
3 most part for us, we try to regulate on the side of
4 safety.
5 I think our greatest challenge is to make
6 sure that we're using sound science. But by doing
7 that, if you take that one in a million individuals
8 and, you know, we use that age group of one to six,
9 just for example, I think we are probably in a -- you
10 know, it's safer to make sure that we are regulating
11 on the side of safety for that particular age group
12 for the various reason of, in general, the chronic
13 effects over the long term that they could be exposed
14 to various chemicals, OPs in particular.
15 So that is one of the things that, you
16 know, we primarily try to keep in mind when we look at
17 mitigation for various OPs.
18 DR. CHRISTOPH: If I can could just
19 comment on that, companies are not interested in
20 poisoning people. We have the same interests you do
21 in terms of ensuring the safety of our products.
22 Throwing around figures like one in a
23 million -- I actually believe we are talking about
24 those kind of risk factors here with the use of
25 pesticides. That's my personal belief, as I've kind
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
132
l of gone through and looked at it. There are certainly
2 people who would disagree and think that the risks are
3 much higher than that.
4 You are absolutely right that the highest
5 risk group that always falls out of all these analyses
6 are children one to six.
7 QUESTION: I have got an anecdote. In a
8 former life I was a Food and Drug inspector, and a
9 bunch of us were sitting around sniggering one day
10 about how we were going out and looking for swordfish
11 based on numbers, mercury numbers, that had to do with
12 eating swordf ish every day. That was in
13 Massachusetts. Swordfish was a luxury good.
14 Then one of the women in the group looked
15 sort of stricken. She was on some kind of diet. She
16 was eating tuna every morning -- every evening. That
17 was her big thing, was tuna.
18 My guess is you're sort of assuming that
19 the 12 apple person -- eating 12 apples is like being
20 struck by lightning, and I think that's wrong. I
21 think that people who eat 12 apples one day are going
22 to eat 12 apples the next day. This is someone who
23 has a serious apple thing, and you do have to think
24 about that person in the course of that risk
25 assessment.
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 2344433
-------
133
1 DR. CHRISTOPH: I agree with that. I
2 would just like you to think about the transition
3 rules. You know, there may be an individual there
4 that eats 12 apples day after day after day. How
5 frequent is that individual? You know, think about
6 that when you build a Monte Carlo is my point.
7 Sure, build that person into the analysis,
8 and build that person into the analysis with a
9 frequency that they actually occur in the population.
10 I don't know exactly what the day to day transition
11 rules are in the Monte Carlo. I mean, it's a
12 complicated question, and there are some data from
13 three consecutive days of people, real people, eating
14 food. That can certainly help in building that one
15 month of exposure.
16 Something like the three-consecutive day
17 data in the database would help us get to understand
18 how often that person eats -- how frequent that
19 individual is that eats 12 apples a day.
20 DR. NORRIS: One more question. Thank
21 you.
22 QUESTION : Okay. I had a question about
23 your normalization of data. If you were using two
24 compounds that were not equally efficacious, would you
25 use the same methodology to normalize data or would
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
(202) 234-1433 WASHINGTON, D.C. 20005 (202) 234^433
-------
134
1 you have to adjust it, since the EDSO would be
2 somewhat different?
3 DR. CHRISTOPH: Well, I think the whole
4 point of normalization is that they are not equally
5 efficacious. They differ in potency or differ in
6 efficacy. Is that what you mean by efficacy here?
7 QUESTION: Well, in terms of efficacy, I
8 mean their ability to induce a response. When you
9 presented your example of how you would normalize
10 data, both of the compounds appeared to produce the
11 same amount of effect.
12 DR. CHRISTOPH: right.
13 QUESTION: See what I'm saying? So if you
14 were using two compounds that did not produce the same
15 strength of effect, how would you normalize the data
16 then?
17 DR. CHRISTOPH: So your idea, one compound
18 would be from zero to 100, another compound might be
19 zero to 50 and not go any higher than that?
20 QUESTION: Correct.
21 DR. CHRISTOPH: In that case we are
22 : dealing with a case of non-parallel dose response
23 functions, which bring us into a whole special
24 category of events.
25 I think the first response when you end up
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
135
1 with non-parallel dose response functions, you should
2 ask are these really compounds that belong together in
3 the same risk cup? Are they truly common mechanism
4 chemicals or is there something different about them
5 that makes them work real differently?
6 There may be mathematical ways to handle
7 that. One way to think about it is, since we are
8 really concerned about the low dose side of this
9 equation anyway, is the lower half of the dose
10 response curves -- are they pretty similar to each
11 other? And if those are pretty similar to each other
12 --We can forget the stuff that happens above 50
13 percent cholinesterase inhibition, because we are not
14 going to drive any human to 50 percent cholinesterase
15 inhibition by this process, by eating food.
16 So you know, if it's that case, let's deal
17 with the lower half of the dose response function, see
18 what's going on down there. Can we somehow convince
19 ourselves that the lower halves of the functions are
20 superimposable after a normalization step?
21 DR. NORRIS: Greg, I didn't mean for you
22 to end up having to answer all the questions, if
23 anybody else has anything to say. I would like to add
24 that I think, if I ate 90 apples in one day, I would
25 have other problems.
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
136
1 Well, let's go to lunch. If anybody is
2 not from around here and needs to know where to go to
3 lunch, apparently, I've been told there's restaurants
4 all over. I can see the pesticide people saying, yes,
5 there are restaurants everywhere. I hope we'll find
6 them.
7 Thank you. We'll see you back here at two
8 o'clock.
9 (Whereupon, the foregoing matter went off
10 the record at 12:41 p.m.)
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
137
AFTERNOON SESSION
Time: 2:05 p.m.
DR. NORRIS: All right. Most of us are
back. Others will probably come trickling in. If I
can have your attention, please, we'll begin the
afternoon session of our program.
Beverly and Ethyl, I've been asked to
change the name of our program by management again.
They would like to make sure it's the neurotoxicity of
chemical mixtures. Perhaps the management is
concerned that there will be the neurotoxicities of
the mixtures of the colors in my dress or something.
I'm not sure. We aim to please.
We have four speakers this afternoon, and
I'll introduce them all at this time and then briefly
as they get up to speak.
Dr. Abou-Donia is Professor of
Pharmacology and Cancer Biology and Professor of
Neurobiology at the Duke University Medical Center.
Dr. Anthony Riley is Director of the
Behavioral Neurosciences and Professor and soon to be
chair of the Psychology Department at American
University.
Dr. Rick Hertzberg initiated the EPA
Mixtures Risk Assessment Research Program here at EPA
NEAL R. GROSS
(802) 234-4433
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
WASHINGTON. D.C. 20005
(202) 234-4433
-------
138
1 and chaired the initial Mixtures Risk Guidelines. He
2 has also developed the first version of EPA' s mixtures
3 database, Mixtox; and I hope you're finding some more
4 data to add to your database today, Rick.
5 Dr. Herman Koeter: Dr. Koeter is the
6 Principal Administrator at the Paris based OECD
7 Environmental Health and Safety Division. He is in
8 charge of test guidelines, harmonization, the
9 endocrine disrupters program, and the animal welfare
10 policies.
11 We are very proud and pleased to have all
12 of you with us this afternoon. Thank you. I won't
13 take anymore of your time. I will let Dr. Abou-Donia
14 begin.
15 DR. ABOU-DONIA: Thank you, Dr. Norris.
16 My talk this afternoon is on chemical/chemical
17 interactions which we have been hearing about all
18 morning. As we know, most individuals are exposed to
19 multiple chemicals. We are not exposed to just one
20 chemical.
21 FDA requires testing of drug-drug
22 ; interactions. The testing that the FDA requires
23 actually is very intensive, and they require
24 particularly the metabolic profile of each
25 interaction, including the specific enzymes of
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-1433 WASHINGTON. D.C. 20005 (202) 234-M33
-------
139
l cytochrome P450 and the effect of each chemical on the
2 induction or inhibition of the enzyme as well as of
3 the MDR peak glycoprotein, the drug transporter in the
4 body.
5 Unfortunately, there is no requirement for
6 drug-pesticide interaction or pesticide-pesticide
7 interactions neither by the EPA or the FDA. When we
8 think about it, chemical companies discovered many
9 years ago that combined exposures to chemicals is more
10 toxic. That's why it's very rare that anybody would
11 use one insecticide in the field. Usually, it's used
12 much more than one insecticide.
13 On many occasions I have asked the
14 questions to some of my friends in the chemical
15 industry, do you have any data on the effect of
16 combined exposure and health effects? They said no.
17 Why? Because this is not required. So I think this
18 is an area that should really be -- should be looked
19 at.
I
20 The other thing is drug-pesticide
21 interaction, which would be a joint venture between
22 the FDA and the EPA. All of us are exposed to
23 pesticides, one way or the other, and we are also
24 exposed to drugs. We have prescription drugs and
25 over-the-counter drugs, and we get exposed to both of
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
140
1 them at the same time, but we don't really know if
2 this would cause any interaction.
3 That's why our specific -- our study was
4 to investigate the interaction between a drug, which
5 is pyridostigmine bromide and insect repellant called
6 a DEBT, and an insecticide, permethrin. These three
7 chemicals were presumably present in the environment
8 during the Gulf War, and they might be involved in the
9 Gulf War diseases or illnesses.
10 The pyridostigmine is a carbamate that has
11 positive charge. So presumably, this carbamate does
12 not cross the blood-brain barrier. It only acts on
13 the peripheral system. Pyridostigmine bromide is used
14 - - is actually approved for use for myasthenia gravis
15 at very, very large doses that range between 200 to
16 almost 1,000 milligrams per kilogram dose. However,
17 it is used -- It was used during the Gulf War as a
18 prophylactic treatment against possible death by nerve
19 gas.
20 The way it acts, as we know, it will
21 shield the acetylcholinesterase in the peripheral
22 nervous system so that when there is exposure to
23 certain of the nerve gases, there will be protect ion ,-
24 and then when the gas goes away, the enzyme will
25 eventually spontaneously recover, and the person will
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
141
survve .
1
2 This chemical is fairly toxic. It has a
3 LDSO of 61.6, which is less than half of that of the
4 chlorpyrif os which you have been talking -- hearing
5 about today. It's about 150. Of course, this is a
6 reversible inhibitor of the cholinesterase .
7 The other chemical which we have been
8 using is called DEBT, which is -- This was developed
9 in 1940s by the military to be used for military
10 personnel when they go in the tropical areas.
11 However, it's now available in the market, and an
12 estimated 50 million Americans use it every year.
13 This chemical is used as an insect
14 repellant and has very, very low acute toxicity, 3,000
15 milligrams per kilogram. However, in the literature
16 there are several reports of death resulting from
17 exposure to DEBT. So even though it has very low
18 toxicity, it is still lethal at certain dose levels.
19 The third chemical we have been using is
20 permethrin, which is -- This is a pyrethroid
21 insecticide, and it acts by disrupting the sodium
22 channel and the axon. Now this chemical is even less
23 toxic. The LD50 is 9,000 milligrams per kilogram in
24 rats. However, the permethrin is very widely used for
25 human use.
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
142
1 It is used in many shampoos for treatment
2 of lice, for control of lice. It is also used to
3 impregnate carpets, mattresses, many of the linen
4 blankets. For those people who are actually sensitive
5 to mites and other insects, that would reduce some
6 kind of allergy.
7 So this chemical is very widely used.
8 During the war it is estimated several hundred
9 thousand American personnel used uniforms that were
10 impregnated with permethrin to control insects. So
11 potentially, a soldier that was in the Gulf War might
12 have been exposed concurrently to pyridostigmine
13 bromide, DEBT and permethrin at the same time.
14 Some years ago, 1994 or '95, we did some
15 study by actually trying the effect of each one of
16 these chemicals alone at a very high dose and then in
17 combination. With the high dose, that caused no
18 effect. When we used them in binaries, unlike single,
19 there was minimum toxicity. Binary combinations was
20 greater toxicity. When we used three chemicals
21 together, they produced paralysis and sometimes death.
22 Later we applied for a grant from the
23 Department of Defense, and we are using the same
24 combinations to study their action at a real life
25 I constant dose level that presumably the soldiers were
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
143
exposed to.
In this study we are using -- we are
actually studying both stress as well as chemicals.
the chemicals that we used are the pyridostigtnine
bromide, DEET and permethrin.
The dose that we are using -- These are
the doses that we obtained from DoD. The dose was 1.3
milligrams per kilogram. This is the exact dose that
the soldiers were taking. They were given -- They
were supposed to take three 50-milligram pills a day.
That's 90 milligrams, which divided by 70 would be 1.3
milligram per kilogram per day orally.
DEET, they told us the dose was 40
milligram per kilogram dermally. So we used that
amount. Permethrin, .13 milligram per kilogram in
ethanol dermally. We treated the animals, rats,
Sprague-Dawley rats, for 28 days.
Also another group of animals was given
stress by simply placing the rat in a plexiglass
restraint for five minutes every morning. The design
of the experiments was like this. We have four
groups. One was chemicals, the three chemicals, the
same doses I just gave, and stress, chemicals under
stress, and the control. Controls were given water
orally as well as ethanol dermally. The water was
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
144
1 given at the one milliliter per kilogram.
2 So what we did, we had male Sprague-Dawley
3 rats treated 28 consecutive days. Animals were weighed
4 weekly. We had subgroups of five rats that were used
5 for enzymatic analysis, to analyze for brain
6 acetylcholinesterase and plasma acetylcholinesterase.
7 We determined the binding of m2 muscarinic
8 acetylcholinesterase muscarinic receptor to [3HJAF-
9 DX384. Now we determined that the integrity of the
10 blood-brain barrier, which we heard some about it this
11 morning from Dr. O'Donoghue --we determined that a
12 similar way --a couple of ways.
13 One, we determined the uptake of
14 triturated hexamethonium iodide, a chemical that has
15 positive charge, is not supposed to cross the blood-
16 brain barrier. The horseradish peroxidase as well was
17 used in this experiment. Then we looked at the
18 histopathology in the brain as well as the liver.
19 Well, this treatment for 28 days resulted
20 in -- There was not clinically different -- The
21 treated animals are not much clinically different than
22 control. As a matter of fact, they might have done a
23 little better, because they were a little bit lighter
24 or gained less weight.
25 The weight gain was -- All of the animals
NEAL R. GROSS
(202) 234-4433
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
WASHINGTON, D.C. 20005
(202) 234-4433
-------
193
1 The diesel here was not -- let's see. It was not used
2 in the calculation of the line. It's an actual data
3 point, but the line is based upon the other mixtures
4 Of PAHS.
5 So in this case, it looks like there is
6 some consistency, some proportionality between mouse
7 skin tumors and human lung cancer risk. That's kind
8 of the basic idea here. If you have a way to scale up
9 from in vitro assays, demonstrate it, and then as we
10 did with the interaction patterns before, you have a
11 way to sort of generalize for untested chemicals or
12 untested mixtures.
13 Another revelation we had, as I mentioned
14 briefly, in looking at the data and the published
15 studies was too many definitions of interaction, too
16 many definitions of additivity. So being a regulatory
17 agency, we have the power to make our own definitions.
18 So we decided, first of all, to simplify.
19 That was to have dose addition as our
20 default, as our no-interaction, so we don't talk about
21 additive interactions. We just talk about dose
22 addition as no interaction. Anything higher than that
23 is synergism. Anything lower than that is antagonism.
24
25 That made it extremely simple, and in a
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
194
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
regulatory sense it's very useful that way. We don't
need to know the other things. We just need to know,
if we're making a mistake, which direction are we
erring in -- on, toward?
I think I'll move on. Another thing we
found in looking at the data is that synergism wasn't
reported that often. We thought that, since academics
do most of this kind of research, you need to publish.
You need to publish positive results.
So you would want to have a selection of
chemicals that would show you some interaction.
Surprisingly, it did not happen.
When we looked at trying to estimate
magnitude for the handful of studies that were
actually useful for doing this, we had high dose
interactions of tenfold to twentyfold or more. We had
low dose interactions that were two, three, five, in
that range.
So our default so far for our mixtures
guidance is a magnitude of five. It's a fivefold
decrease in effective dose, if you're talking about
synergism.
This is the hazard index. It's the basis
of dose addition.
Superfund has this in their
guidance. The bottom expression is what's used for
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
195
1 the dioxins where you replace the mixture by its
2 surrogate or index chemical, the equivalent dose in
3 that chemical. It's the same formula, but you
4 translate to each other.
5 We had one improvement, and this is where
6 I'll give you some more lessons learned. In '92 we
7 came up with an approach to replace the conservatism
8 in using the reference dose, the reference
9 concentration as a scaling factor for relative
10 potency.
11 The problem with those is that they are
12 bounding estimates. They are supposed to be the --
13 representing the critical effect. If you are trying
14 to do some common mode of action that is some other
15 effect, then you are going to be over-regulating or
16 overestimating the risk, because you're using
17 acceptable doses or dose scaling that's way, way too
18 low.
19 So we could just do the same kind of
20 process but do it for the effective concern. If we're
21 looking at neurotoxicity, then make sure that all of
22 these pseudo reference doses are now based only upon
23 neurotoxicity data. Then you don't have the over-
24 conservatism.
25 Well, so we wrote this up, and we got
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
196
l slammed hard by our regions and our program offices,
2 because EPA has no procedure in place for developing
3 these target organ toxicity doses. So why advertise
4 this as a great new method if we can't do it? So
5 that's my lesson learned number three or four, I
6 guess, by this point.
7 Good ideas need to be published, but if
8 you put them in guidance, then someone expects you to
9 actually do it.
10 Now once again we're back in the case of
11 not having enough information. Here you see again a
12 lot of places, not as many as we were concerned about,
13 but you see that in many cases we don't have enough to
14 do a TTD so we can resort back to using the reference
15 dose. Okay, we have five or six cases here where we
16 are essentially going back to defaults.
17 So a good approach. Just doesn't have
18 enough information to do it.
19 This was raised several times. So what do
20 you do if you do have interactions? How do you
21 account for them? We talked to a lot of people who do
22 interaction studies. We looked at a lot of literature
23 on pharmacokinetic modeling and mechanisms of
24 interaction.
25 It seemed that we had some general
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W
(2021234^433 WASHINGTON. D.C. 20005 (202)23^4433
-------
197
l principles that seemed to come out. One was something
2 that, if you didn't have your chemicals at equally
3 toxic doses where one starts to dominate, then the
4 interaction often starts to die off. So we would want
5 to have some kind of way to reflect that, and that's
6 what this function does down here. As the chemicals
7 get more dissimilar, you have a curve that goes like
8 this. One chemical becomes more dominant. The
9 interaction goes down.
10 These are just general qualitative
11 characteristics. What we have here, basically, is
12 this part over here is the hazard index, dose
13 addition, and we just took every hazard quotient and
14 modified it by a function that represents the
15 interactions to that chemical.
16 So this is all based upon pair-wise data.
17 That's weight of evidence considerations, or the Bs
18 right there. So this is a way we can kind of
19 incorporate the data we do have on interactions,
20 quantify it, end up with a change in the risk
21 estimate.
22 What we need is more information so that
23 all of these functions in here can be replaced by real
24 dose interaction/magnitude relationships and not just
25 sort of plausible relations.
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
198
1 Okay. This just kind of explains the
2 different pieces.
3 So here's my summary of our lessons.
4 First of all, we have to be flexible in our methods.
5 We can't require testing or new data for every mixture
6 we encounter. There are too many mixtures and too
7 many varieties of exposure situations, changes in the
8 proportions, in total dose.
9 You saw the influence of changes in
10 sequence, changes in repeated exposures as opposed to
11 a single shot. All those things have to be somehow
12 reflected. When you have a mixture and you're trying
13 to do something to build up from component
14 information, the component information is not the same
15 for each chemical.
16 So that's what I mean by flexibility. You
17 have to have a way to use the different varieties of
18 information.
19 We don't have a lot of experience yet with
20 mixtures. Superfund is our only main office that's
21 done this for any length of time. So approaches such
22 as the acceptable daily intake that ran FDA and EPA
23 for a long time and build up a history, we don't have
24 that with mixtures. So we don't have a lot of
25 procedures set in concrete yet that we can rely on as
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
199
1 passing the public sector.
2 This is some examples of the problems of
3 multiple data types and qualities. This is one of the
4 concerns, I know, with OPP when they put out their
5 guidance, was what do you do when you have one
6 chemical that's just dominating the whole mixture
7 assessment because it has a huge uncertainty factor.
8 I haven't heard whether that got resolved,
9 but there's some nice suggestions on how to fix that.
10 So I'm looking forward to the next version to see if
11 it's included in there.
12 Incorporating interactions: There are
13 lots of pairs. A mixture of ten chemicals has -- what
14 is it, 1,000 pairs or so? We can't information on all
15 those pairs. We know that. So you're going to have
16 to have these data gaps and have some way to fill in
17 those holes with something.
18 Then this is typical of the government, is
19 that you will have defaults so you can make a decision
20 in spite of lack of data. When you have defaults, you
21 have to combine it with real information. How do you
22 do that in a nice, scientific fashion without having
23 the defaults just dominate everything?
24 If they dominate, then there's no point in
25 gathering the better information, because it no longer
NEAL R. GROSS
(202) 234-4433
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
WASHINGTON, D.C. 20005
(202) 234-4433
-------
200
1 plays a role. We found this with NOELs, doses based
2 upon NOELs.
3 If you have other chemicals that have
4 really good ED10 or ED01, it doesn't matter a whole
5 lot, because the uncertainty surrounding the NOELs is
6 so much higher that it just swamps the extra accuracy
7 that you get from the other chemicals that have ED10.
8 We are all hoping we can move toward
9 pharmacokinetic modeling, but we need to have human
10 data. There is already a lot of questions being
11 raised about overuse of PBPK models based only on the
12 rat. So we need to have some better validation, I
13 think, there. As I said before, interaction magnitude
14 -- we're really in the dark there.
15 Okay. We do have a lot of good ideas.
16 It's really exciting for me to work with the Pesticide
17 Office and watch them go through their haranguing to
18 put together guidance after we spent so many years and
19 still hadn't had any new guidance come out.
20 A lot of new ideas came out. There are a
21 lot of ideas just today. A nice advantage, I guess,
22 of being one of the last speakers is you see many new
23 concepts being put forth, but to have them in their
24 practical method, a guidance for an agency to use, we
25 need more, and we have to have ways to know what to do
NEAL R. GROSS
(202) 234-4433
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
WASHINGTON, D.C. 20005
(202) 234-4433
-------
201
1 when those pieces of information are not present.
2 The bottom line -- It's still there. This
3 is the one we had hoped for 15 years ago. So any of
4 you that have any insight on how to do these studies
5 at the really low environmental levels, start showing
6 us where these thresholds occur, where below that
7 level we don't have to worry about interactions. It
8 would really make our job much, much easier.
9 Thank you.
10 (Applause.)
11 DR. NORRIS: Are there questions now?
12 Thank you.
13 Our next speaker is Dr. Herman Koeter, and
14 I'll make another announcement after that. Thank you.
15 What I'd like to suggest is that, in lieu of taking a
16 break and then coming back, if our speakers wouldn't
17 mind, if we could talk to you for a few minutes during
18 the break over cookies and soda in the back room
19 there. We can skip the panel discussion and just have
20 a more intimate conversation in the back over our
21 break, and then we will excuse ourselves from there.
22 Thank you very much. Now we'll be set up here in a
23 moment for Dr. Koeter.
24 DR. KOETER: Thank you very much for the
25 invitation to speak to you here today, and I consider
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
(20?) 234^*433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
202
1 that a privilege to have an opportunity to explain a
2 little bit more about the work of OECD and, in
3 particular, with respect to the work that we do on the
4 classification of mixtures and the development of test
5 guidelines.
6 I apologize for the sort of prehistorical
7 way of presenting my information. The slides that I
8 use are --It only needs a bit of focusing. Here we
9 go.
10 Well, first of all, the OECD, as some of
11 you do know and many of us probably don't, is an
12 organization for economical cooperation and
13 development, and the E for economical is quite an
14 important one. What that means actually that we try
15 to improve the economies of our member countries, and
16 that's the main goal of that organization.
17 Now you may wonder what chemicals have to
18 do with economy in that respect. By the way, these
19 are the current members of OECD, and I say current
20 because, well, there is sort of a continuous debate on
21 the adherence. A lot of countries do OECD. The ones
22 that you see up here, some of them have been members
23 since the early sixties when OECD was established.
24 Others like Mexico, Korea are members that joined the
25 organization only recently. That means in the last
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) £34-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
203
1 couple or four or five years.
2 As I said, it's an economic organization,
3 and chemicals do play an important role in such a way
4 that we feel that the management of chemicals in
5 member countries, if we could harmonize that in one
6 way or another, would save a lot of money, and we
7 published a document on that that shows that slowly,
8 because of the work done by one division on chemical
9 safety and the harmonization of methodologies, that
10 that would save a numerous amount of dollars. As a
11 matter of fact, it's around approximately $56 million
12 in OECD member countries on a yearly basis just
13 because we share information between countries. We
14 share registration procedures, and we have harmonized
15 testing methodology.
16 As you all know, the risk assessment
17 paradigm consists of the hazard identification, and
18 from there on further down to risk assessment. In our
19 work on harmonization of classification, I would like
20 to start to show you what role mixtures play in our
21 work and to give you a little bit of an impression how
22 much work it is to harmonize certain things.
23 We identify these steps here, the hazard
24 identification steps, which we would like to harmonize
25 between countries and among countries, and then after
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
204
1 that and based on that hazard identification, the
2 classification of chemicals and mixtures, for that
3 matter and, when that is done, the hazard
4 communication.
5 Well, the latter part is a part where we
6 share the work with other international organizations
7 such as WHO and ILO. You can imagine that
8 communicating a certain hazard and the detail of a
9 hazard to people that are illiterate, live in
10 countries where cultures are completely different from
11 western countries or from OECD member countries in
12 that respect is quite a challenge in itself.
13 Well, in order to organize that work, we
14 started to establish a comity, a task force, we call
15 that. Unfortunately, in OECD we have rules and
16 regulations what you will call a group, whether that
17 will be a working group or a task force or whatever.
18 But that is very clearly defined, and a task force is
19 a relatively higher level. That means there's most
20 definitely a policy component to that as well, not
21 only technical.
22 Here you see the task force that we set up
23 for the classification and labeling work. It consists
24 not only of member country representatives. They are
25 all nominated by their governments, but also the
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W
(202) 234-4433 WASHINGTON. D.C. 20005 (202) 234-4433
-------
205
1 European Commission representative. That is in
2 addition to members from the European member states.
3 We have chairs from other groups that do
4 related work. You see the task force of NCs. That
5 stands for national coordinators of the TGP, which is
6 the test guidelines program, and I will talk a little
7 bit about that in a minute.
8 We have the chair of our working group on
9 pesticides . That is a working group that works on the
10 harmonization of pesticides work. Then we have a
11 whole lot of acronyms that I will not explain, but
12 they all deal with international organizations that,
13 in one way or another, are involved in classification.
14 You can imagine, classification is not only for
15 consumer use and for pesticides. It's also used in
16 transportation quite a lot and in marine
17 transportation, the big tankers, it makes a really big
18 difference whether their contents is classified as
19 hazardous or nonhazardous, and within that hazardous
20 category also further details.
21 So IMO, International Marine Organization,
22 transport of dangerous goods, U.N. organizations.
23 BIAC and TUAC stands for the official organizations
24 that represent industry. That is BIAC, Business and
25 Industry Advisory Committee to the OECD, and TUAC is
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W
(202) 234-4433 WASHINGTON. D.C. 20005 (202) 234-4433
-------
206
1 worker unions, international worker unions. They all
2 play a role in that work, and they all disagree
3 usually.
4 We recognize a couple of steps to sort of
5 facilitate and organize the work, because we deal with
6 classification systems that are around, that are
7 substantially different between countries, and we
8 first thought, well, as a first step we need to know
9 what exactly do these classification systems look like
10 in the various member countries. Where do they
11 differ, and why do they differ, and do they all work
12 in the same way if you would compare them, run
13 chemicals through one system into another, do you end
14 up with different classification and so on and so
15 forth. That is step one, and we call that a detailed
16 review document.
17 These documents can be substantial,
18 usually 100 pages or more where it has a clear
19 comparison between all the systems.
20 As a next step, we would then propose a
21 harmonized system which is based on existing systems.
22 We had no intention to develop new systems. Although
23 some of the experts would really like to take that
24 opportunity and say, well, we've learned from the
25 past, let's develop things again, we try not to do so.
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
{202} 234-4433 WASHINGTON. D.C. 20005 (202) 234-4433
-------
207
l The reason for that is that we want to
2 avoid that after you have developed a harmonized
3 system that everybody has to classify everything that
4 had been classified before. So we would like to
5 minimize the changes in classification that would
6 occur once you have adopted the harmonized system,
7 which is very difficult in itself.
8 Step three is then the discussion that
9 would fall out at the various levels, the technical
10 level, policy levels, and so on.
11 Step four would be the endorsement, and
12 that is an official endorsements by the governments of
13 the OECD member countries. Since we have only 29
14 member countries, together producing still 85 percent
15 of all the chemicals in the world, we want to reach
16 out further.
17 That's why we have sought acceptance in
18 the U.N. system and the decision making system there
19 in the U.N. which we will do together with other
20 international organizations and especially those that
21 are also involved in the hazard communication. That
22 means in that center part which you see there, IOM
23 stands for the International Organization of -- well,
24 I forgot it myself for the moment. It's an
25 international group of organizations that, like U.N.
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
208
1 organizations and OECD and the Commission would all
2 work together and to bring that together to U.N., to
3 ECOSOC where it will be adopted in the near future.
4 Well, where are we today? That is the
5 sort of framework in which we work. We have
6 identified these endpoints that you see up there,
7 because classification systems, most of them but not
8 all, include all these endpoints in one way or
9 another.
10 You see, the ones that are listed at the
11 upper part are all endpoints that are considered, and
12 most of them are real, true endpoints. The last one,
13 chemical mixtures, is a different entity. We have
14 added that, because currently do there exist specific
15 classification systems for mixtures that are different
16 tihan for chemical substances.
17 There are also a couple of endpoints that
18 are not covered in any existing system, which are
19 neurotoxicity, unfortunately immunotoxicity, and then
20 a couple of smaller hazards. We call them smaller
21 hazards, because they are not really interesting for
22 all the member countries, like water activated
23 toxicity such as in contact with water would vaporize
24 and, by that, cause a hazard, aspiration hazards,
25 hazards of defatting agents, and so on and so forth.
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-1433 WASHINGTON. D.C. 20005 (202) 234-4433
-------
209
1 In order to deal with all these different
2 subjects, we need different groups of experts. We
3 cannot just put all the experts together and say,
4 well, let's talk about that, because we need different
5 experts for these different groups.
6 So we have established groups of experts
7 on acute toxicity, mutagenicity, repro-tox, all these
8 different things. These groups came together to
9 discuss their various endpoints and to reach consensus
10 or to fight about that.
11 In that last slide you saw one group on
12 chemical mixtures, and I will not go into the details
13 of the work that we did on substances, because it's
14 not so much of interest for this audience today. So
15 I will focus a bit more on how we dealt with mixtures.
16 One of the things that surprised me today,
17 being the last speaker -- we've heard many speakers,
18 and nobody really started to discuss what is a
19 mixture. Well, Debbie did it, to some extent.
20 Well, there are a lot of people that would
21 say, yes, that is a mixture. A mixture is now being
22 considered, at least as a working definition in OECD,
23 as a mixture of two or more chemicals that do not
24 react. Especially the latter addition is an important
25 one, because you can put chemicals together and they
NEAL R. GROSS
(202} 234-4433
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W
WASHINGTON, D.C. 20005
(202) 234-4433
-------
210
l react. Well, that can become a mixture after the
2 reaction process is done, then fine. Then you have a
3 mixture that will no longer react, and that is what
4 you work with.
5 Indeed we have spent meetings talking
6 about where does it stop. Is this table a mixture?
7 Is the dress a mixture? These are all part of that
8 definition. We had to stop that discussion and say,
9 well, we all sort of understand what a chemical
10 mixture is, and let's don't really bother about that.
11 So we left it aside, but we consider it all in the
12 sort of the general way that we in our various
13 legislatures and regulations consider a mixture.
14 The expert group, all in the
15 classification for mixtures, we have extended
16 expanded our group beyond OECD where you see countries
17 here like Brazil, because we felt that classification
18 of mixtures is especially important in developing
19 countries where a lot of the chemical -- not so much
20 the production of the chemicals, but the preparation
21 of mixtures is being done. Products are being made,
22 and people are exposed to large extent. We want to
23 have specific input from those countries as well.
24 You see, the group is more limited here.
25 We wanted to have a sort of regional representation
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON. D.C. 20005 (202) 234-4433
-------
211
1 from the southeast like Australia, Canada and Mexico,
2 then also to sort of represent developing countries;
3 two people from EU member states. There's a couple of
4 countries that alternate on that, and then we have the
5 U.S. I think I forgot to mention the U.S. on that.
6 The U.S. is on it definitely, the European Commission,
7 Brazil, BIAC, TUAC, and a couple of other people.
8 The branching of the tree continues,
9 because below that group there are again subgroups
10 that deal with the various endpoints that are of
11 importance for mixtures.
12 Of course, in principle, all the endpoints
13 that I listed before for substances are also being
14 considered for mixtures, but it was not considered
15 necessary to have separate groups of experts for each
16 of the endpoints, because we have dealt now with
17 substances, and I didn't mention that before. But
18 that is the past station. We have reached agreement
19 on that.
20 So using the harmonized classification
21 systems for a particular endpoint like sensitization
22 and dermal and eye irritation and corrosion, people
23 felt sufficiently confident as a bigger group to work
24 with that and try to harmonize that.
25 The ones that are listed here are the ones
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
212
1 that are considered difficult. Acute toxicity has
2 been and still is extremely difficult, how to deal
3 with that.
4 C&M stands for cancer and tnutagenicity,
5 and reproductive toxicity are also very difficult.
6 Environmental hazards are still in its infancy anyway,
7 the classification for chemical hazards to the
8 environment. Then we have a separate group that tries
9 to put this altogether into a proposal.
10 Again, I will not focus too much on all
11 these endpoints, because I can spend hours just
12 talking about that.
13 One thing that struck me is that within
14 the discussions or part of the discussions, one of the
15 biggest differences in all the discussions about
16 mixtures is that additivity of effects is considered
17 a very important aspect in Europe.
18 In U.S. it's not considered an important
19 aspect. I just learned this afternoon, at least in
20 EPA, in your guidelines that are coming out for
21 mixtures, additivity is considered. But as you
22 understand, in this international forum we deal, of
23 course, with EPA, but we also deal with FDA.
24 We deal with your Department of
25 Transportation. We deal with OSHA. We deal with
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
213
1 CPSC, and all the other agencies, they don't want to
2 be bothered about additivity. They have a
3 classification system for mixtures which is based on
4 the individual hazards, and they just apply sort of
5 cutoffs for each of those separate hazards and don't
6 add them up.
7 In Europe they have very complicated
8 systems for additivity, and they do not -- Well, they
9 do consider also synergism, but they do that in a sort
10 of a, let's say, expert judgment type of approach,
11 which is more or less on a case by case rather than
12 using formalized or other stuff, which they do for
13 additivity.
14 This is just to show you the number of
15 meetings that we have only on that particular group of
16 mixtures in order to reach consensus. The seventh
17 meeting is in parentheses, because we hope that next
18 month at the sixth meeting we will reach final
19 consensus.
20 You see also -- This is just to give you
21 an impression how much work is involved here, that
22 this is only the list of meetings of the expert group.
23 All the subgroups below that are not listed, and then
24 in addition we have more than 50 chemicals.
25 We've heard them already today, and
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON. D.C. 20005 (202) 234-4433
-------
214
1 teleconferences is one of the nightmares that I sort
2 of have, because I have teleconferences almost every
3 day, if not every day, on all kinds of different
4 subjects.
5 So where are we with mixtures today?
6 Well, first of all, we have the data review document.
7 This is available on the Web, and one of the slides
8 that will follow this gives you the address. If you
9 forget about it, it's just www.oecd.org.ehs for
10 Environmental Hazards Safety, but it comes up in one
11 of the slides.
12 Here we compare the classification systems
13 that are existing today, and in addition to the North
14 American from Canada and U.S. and the European EU,
15 Sweden being a member of the European Union still has
16 a classification system that's slightly different from
17 the Commission.
18 As you may know, Sweden is one of the more
19 recent members of the European Union, and still is in
20 a period of transition, and now tries to get some of
21 their specific aspects into this harmonized system,
22 which they could not manage at the time they were
23 becoming a member in Europe. But we also see that
24 countries like Slovenia, Korea -- there are countries
25 in the world where you would not have expected that
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
215
1 they have detailed systems for a chemical mixtures
2 classification, and they have.
3 They have all been considered. We have
4 compared consumer products, transport, pesticides, new
5 chemicals, hazardous wastes, and the title of the
6 document you see there below.
7 The step two proposal that we have now,
8 the first one, and the second one is coming up in May,
9 has these chapters: General consideration, building
10 block approach -- and I wish I had more time to talk
11 about the building blocks to what really constitutes
12 harmonization, because harmonization is not
13 standardization -- definitions, and then these
14 endpoints are all part of that.
15 Here at the bottom you see target organ
16 toxicity, and that is the whole on neurotoxicity
17 today. It was considered by member countries
18 officially that there was no need to classify
19 substances specifically for neurotoxicity, no need to
20 classify mixtures specifically for neurotoxicity.
21 What was agreed is that we would consider
22 target organ as the home for that, and if we know that
23 the target organ is the brain or central nervous
24 system, we could mention that, but not have a special
25 system for it.
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
(202) 234-4433 WASHINGTON. D.C. 20005 (202) 234-4433
-------
216
1 Within mixtures we had this sort of
2 ranking. We classify --We have a system to classify
3 whether the data are available for the mixture as a
4 whole. You would say, well, that's the ideal
5 situation, which is not always true.
6 There are countries, especially in Europe,
7 that feel that with respect to cancer and mutagenicity
8 and even reproductive toxicity, data on the mixture as
9 a whole is not good data, because you have dilutions
10 of the substance that can cause that carcinogenicity
11 which will not show up in your testing when you test
12 the mixture.
13 So we have a system where you would deal
14 with when you have the data of the mixture as a whole.
15 Then we apply bridging principles where we have
16 additivity, synergism and that sort of thing included
17 in rules and regulations.
18 Then we classify based on where we have no
19 data on the mixture as a whole, but we have data on
20 all the components. You can imagine that we say --
21 easily say all the components that, we have quite an
22 extensive discussion, what is a component? How far do
23 you go? Do you go to the .0001 percent contaminant in
24 your mixture or do you have a cutoff? We did set a
25 cutoff of one percent now.
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON. D.C. 20005 (202) 234-4433
-------
217
1 Then we have classification based on when
2 data are available for some of your ingredients, but
3 not all. But this is just to show the complexity.
4 Finally, because I understand you're
5 interested in neurotoxicity, and that's what I said
6 where we have neurotoxicity data, we have separately
7 identified two systems, one for effects of the single
8 exposure and one for effects after repeated exposure.
9 These have gone back and forward many
10 times, because they said, well, they should be all
11 included into one system, both single and repeated
12 exposure. Then we didn't manage to reach a harmonized
13 system. So it was decided to separate them. Then we
14 had them back again.
15 So we went back and forth quite a while,
16 and we ended up having two different systems which
17 have two classes today, and the first class is based
18 on human data, basically, and the second on animal
19 data. That's one distinction.
20 Also, the separation is based on the
21 severity of the facts. The severity of effects, of
22 course, can be related to the testing where the test
23 methods for dose specific endpoints are considered
24 important.
25 The criteria -- We have expert judgment
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
(202) 234-*433 WASHINGTON. D.C. 20005 (202) 234-4433
-------
218
1 which is the leading thing. You can easily, well,
2 that makes it pretty vague. Expert judgment in one
3 country is very different from expert judgment in
4 another country. But in order to harmonize even
5 expert judgment, which is hard, we provide guidance of
6 effects that are considered to support classification,
7 effects that are not considered to support
8 classification, and we have added also guidance values
9 for cutoff of effects, which is currently used in
10 Europe mostly.
11 Just one that shows a little bit where we
12 have references to, neurotoxicity or at least it is
13 made clear that neurotoxicity is definitely included
14 there, like under criteria and like under effects
15 supporting classification, we have statements there,
16 significant functional change in central peripheral
17 nervous system, including central nervous system
18 depression and special senses.
19 These sort of syntheses are added
20 scattered around in the document, and not only for the
21 endpoint of neurotoxicity but also for many of the
22 other endpoints.
23 A few minutes on the test guidelines
24 development in OECD. That is sort of the foundation
25 of the work on classification, because you need test
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE. N.W.
(202; 234^433 WASHINGTON. D.C. 20005 (202) 234-4433
-------
219
1 data. So you need test guidelines to do that.
2 These are based on what we call the MAD
3 decision or the Mad Decision, which tells you that if
4 data are generated according to OECD guidelines and in
5 compliance with OECD GOP, practices shall be accepted
6 in all member countries, which is today the case.
7 That means that, despite the fact that
8 certain member countries, including the U.S. and
9 Japan, for that matter, still have their domestic
10 guidelines and even have different guidelines between
11 agencies or within an agency, when a test is conducted
12 according to an OECD guideline and then that OECD
13 guideline covers the data required -- which, of
14 course, that's a prerequisite -- this will be accepted
15 by all member countries.
16 Just to give you a clue about how big that
17 program is, we have currently -- Well, in 1981 we
18 published our first set of 51 guidelines. Since then,
19 the 11 addenda have been published with 90 new and
20 updated guidelines altogether, and we have now a
21 second edition of the work which was published in
22 1.993, and it's now today available as hard copy or CD-
23 Rom, and also online in our OECD bookshop.
24 This is the site where, if you go there,
25 you'll find all the information on testing and
NEAL R. GROSS
(202) 234-4433
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
WASHINGTON, D.C. 20005
(202) 234-4433
-------
220
1 assessments and also on the classification. The
2 publication are, unfortunately, a priced publication,
3 but recently I've managed with the OECD management
4 that they can also be bought now separately as
5 guidelines, individually as guidelines rather than
6 buying the whole set.
7 We distinguish various sort of groups that
8 involve physical chemical properties, biotic systems
9 facts, degradation, health effects, and special
10 activities. You see a number of projects that are
11 listed that we have today in our portfolio. We have
12 one project on acute toxicity, for instance, which is
13 a very large one. We have smaller ones. So
14 altogether you have a bit of an idea how many projects
15 we have.
16 Yo see that some projects are led by
17 member countries. They take the leads. They organize
18 the work. They have their experts providing first
19 proposals for draft guidelines, the guidance
20 documents. They could be member countries, could even
21 be the industry.
22 Who do we involve in the work here? We
23 involve not only the member countries, but we have
24 partner organizations, the European Chemicals Bureau,
25 ICH, which has been responsible for the harmonization
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
221
1 of pharmaceutical guidelines; IOMC, mentioned earlier,
2 the combination of international organizations; IZO
3 standards; industry organizations; and then we have
4 input from academia, government and industry. All
5 these things come together finally in a proposal at
6 the Secretariat where we try to deal with that.
7 I will now use only two more minutes, if
8 you don't mind. I will skip the procedural part, how
9 we develop. I will just show you one example of
10 neurotoxicity, just to see how much time it sometimes
11 takes to indeed develop a guideline.
12 It started in '87, our Guideline 424. For
13 those who know, that guideline started in '87 as a
14 combined proposal from the U.S. and the Netherlands
15 for a neurotoxicity test. After formatting, language
16 changes, and so on and so forth, in '88 we can
17 circulate that for review.
18 Well, you see, we had a follow-up meeting
19 here in Washington, an expert meeting in OECD where we
20 discussed the guideline, including the comments that
21 were received. After review by international experts,
22 and just besides, national experts means we have a
23 database of about 6,000 experts that are all copied on
24 guidelines, and that means that we do receive a whole
25 lot of responses and comments, and in the Secretariat
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON. D.C. 20005 (202) 234-M33
-------
222
1 we try to deal with that, put it together and revise
2 guidelines accordingly.
3 In ' 92 we had another consultation meeting
4 where we -- neurotoxicity, the relationship between
5 that and the guidelines that we have for systemic
6 toxicity was discussed, and in 1994 we were finally
7 able to circulate a new proposal.
8 Another meeting was needed in Ottawa to
9 further discuss comments, and some of you in this
10 audience have been present at these meetings. In '95
11 we revised it again. In '96 we were finally ready to
12 bring it to the policy level, and that means we were
13 not really done there, because after the experts
14 finally agreed that this is technically a good
15 guidance or a good guideline, then policy people are
16 considering it.
17 They look at economical, social impacts
18 and welfare sort of things pressuring countries, and
19 finally it was adopted in 1997. So that took exactly
20 ten years.
21 It's one of those examples -- I think I
22 stop with my overhead. It's one of those examples
23 that I use when I say, well, if it takes more than ten
24 years, then this is sort of a continuous process,
25 because after about seven, eight years of discussion
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON, D.C. 20005 (202) 234-4433
-------
223
1 among experts, you see a new generation of experts
2 coming in, and that is very serious,- because a new
3 generation has different views, disagreements -- they
4 are elderly people, and they come up with new ideas.
5 You start over all again in the
6 discussion. So we feel that we have experience at
7 this. If it takes more than ten years, you better
8 stop it and start all over again, rather than
9 continue.
10 So we were happy that we could finalize
11 the neurotox in ten years. I hope that developmental
12 neurotox which we started just a few years ago in '96,
13 so to speak, and I have another series of slides on
14 that but because of time constraints, I will not show
15 them to you -- that we will be able to manage to
16 finalize that guideline within a year's time from now.
17 I think I'll stop here, Debbie. Thank you
18 very much.
19 (Applause.)
20 QUESTION: This may put you on the spot a
21 little bit. In October the Netherlands is having a
22 meeting to work on mixtures guidance or some mixtures
23 approaches, I guess, for their country.
24 I would love to ask them a question about
25 how that fits into what you are doing. How can I ask
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON. D.C. 20005 (202) 234-4433
-------
224
1 that question?
2 DR. KOETER: Well, you can ask that
3 question probably to address it to me, although I
4 could not give you that answer right away.
5 Unfortunately, member countries do not always share
6 information about their national events with the
7 Secretariat.
8 I think -- I'm not sure whether that
9 meeting that you refer to really deals with
10 classification of chemical mixtures in the same way
11 that I was talking about that. If that is the case,
12 then, well, at least the Netherlands should inform the
13 Secretariat. So if you wish, I can ask them and send
14 that information to you. I do not have it here at
15 hand.
16 MR. BLUOIN: Thank you very much, Dr.
17 Koeter.
18 In closing, I represent the Technical
19 Training Committee from OPPT. I want to thank all of
20 you for coming to this, those of you that have made it
21 through the day. The crowd has been diminished.
22 Special thanks to our speakers who
23 presented us with a worldwide view of this fairly
24 complex thing.
25 Special thanks goes to Debbie Norris, who
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
(202) 234-4433 WASHINGTON. D.C. 20005 (202) 234-4433
-------
225
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
had this bright idea about a year and a half ago.
(Applause.)
Thank you very much, Deb.
Last but not least, thanks to all the
people that helped put it all together, everybody from
people that did the nametags, the people that are
sitting out there, to Bev Sjoblad for hand holding us
with the money and stuff like that. So thank you all.
There's cookies and stuff out there. I
guess we have decided to not have the panel. So if we
have some discussion, we can have it out there over
cookies.
(Whereupon, the foregoing matter went off
the record at 4:07 p.m.)
(202) 234-4433
NEAL R. GROSS
COURT REPORTERS AND TRANSCRIBERS
1323 RHODE ISLAND AVENUE, N.W.
WASHINGTON, D.C. 20005
(202) 234-4433
-------
226
CERTIFICATE
This is to certify that the foregoing transcript in the
matter of: Symposium
Before:
Date:
Place:
U.S. Environmental Protection Agency
April 27, 2000
Arlington, VA
represents the full and complete proceedings of the
aforementioned matter, as reported and reduced to
typewriting.
------- |