1

DEPARTMENT OF HEALTH AND HUMAN SERVICES

FOOD AND DRUG ADMINISTRATION

CENTER FOR FOOD SAFETY AND APPLIED NUTRITION

FOOD ADVISORY COMMITTEE MEETING

Advice on CFSAN'S Draft Report:

Approaches to Establish Thresholds for Major Food Allergens and for Gluten in Food

Friday, July 15, 2005

8:00A.M. to 9:45 A.M.

Greenbelt Marriott

6400 Ivy Lane

Grand Ballroom

Greenbelt, Maryland 20770

P A R T I C I P A N T S

FOOD ADVISORY COMMITTEE STANDING MEMBERS:

Richard A. Durst, Ph.D. - Acting Chairman

Jeffrey A. Barach, Ph.D. (Industry Representative)

Patrick S. Callery, Ph.D.

Dennis Gonsalves, Ph.D., M.S.

Jean M. Halloran (Consumer Representative)

Douglas C. Heimburger, M.D., M.S.

Margaret C. McBride, M.D.

Mark Nelson, Ph.D. (Industry Representative)

Carol I. Waslien Ghazaii, Ph.D., R.D.

TEMPORARY VOTING MEMBERS:

Petr Bocek, M.D., Ph.D.

Margaret Briley, Ph.D., R.D.

Erica Brittain, Ph.D.

Ciaran P. Kelly, M.D.

Soheila June Maleki, Ph.D.

David O. Oryang

Marc D. Silverstein, M.D.

Suzanne Teuber, M.D.

FOOD AND DRUG ADMINISTRATION:

Robert E. Brackett, Ph.D. - Director

Food and Drug Administration, CFSAN

Catherine Copp, J.D. - Senior Policy Advisor

Food and Drug Administration, CFSAN

Steven M. Gendel, Ph.D. - Senior Scientist

Food and Drug Administration

National Center for Food Safety and Technology

Rhonda Kane, M.S., R.D. - Consumer Officer

Food and Drug Administration, CFSAN

P A R T I C I P A N T S (Continued)

FOOD AND DRUG ADMINISTRATION STAFF:

Michael M. Landa, J.D.

Deputy Director for Regulatory Affairs

Food and Drug Administration, CFSAN

Stefano Luccioli, M.D. - Senior Medical Advisor

Food and Drug Administration, CFSAN

Marcia Moore, Food Advisory Committee, Executive Secretary

C O N T E N T S

PAGE

Call to Order and Welcome and Introductions 5

Richard Durst, Ph.D., Acting Chairman

Welcome 6

Robert E. Brackett, Ph.D., Director, CFSAN, FDA

Committee's Discussion and Response to 7

FDA's Charge and Questions

Closing Comments 85

Michael M. Landa, J.D., Deputy Director

Regulatory Affairs, CFSAN, FDA

Adjournment 87

P R O C E E D I N G S

Call to Order and Welcome and Introductions

CHAIRMAN DURST: I would like to convene

the final session of our committee meeting this

morning. Let me begin by again stating that there

are the conflict of interest statements over on the

side table for anyone who wants to avail themselves

of that information.

Also, rather than waiting for the very

last minute to thank the various people who made

this meeting possible, I just wanted to thank

Marcia and her staff for taking such good care of

us and providing all of the information we needed

to have a successful meeting.

I also want to thank the USDA Graduate

School for providing support and the Marriott Hotel

of course for providing very nice facilities to do

this work.

In addition, I would also like to take the

opportunity to introduce Dr. Robert Brackett, who

is the director of the Center for Food Safety &

Applied Nutrition, who was able to join us for a

couple of hours this morning.

Bob, if you want to say a few words?

Welcome

DR. BRACKETT: Thanks, Dick.

The only thing I wanted to say -- I know

you are all in a hurry to get things done and get

to your airplanes this morning, too -- I had hoped

to be here for more of the meeting this week, but I

have been stuck at other meetings. However, I do

want to let you know how supportive I am of the

advisory committee structure.

I served on this Committee a number of

years ago, and I know that it is a big time

commitment and there is a lot of studying to be

done, a lot of discussion, so FDA really does

appreciate your participation and your expertise.

This is something I think that has more

value added to it for us than we could have gotten

individually and breadth of knowledge, and so I do

thank you also.

I would also like to extend that I hope

that you will continue. I hope that you will tell

your colleagues, when they are asked to serve on

this Committee, that this is something that they

really do provide a great service to the country

and to the regulated industry that we deal with.

With that I will let you conclude your

meeting this morning, and thank you for being here.

CHAIRMAN DURST: Thank you, Bob.

Yes?

DR. TEUBER: I found out, in my rush to

get over here, one page is still on the printer at

the concierge desk. If there is a Marriott staff

person who could pick up the third page from the

printer, behind the concierge desk, that would be

fantastic. It is a printout of just a draft of our

summary here.

Committee's Discussion and Response to

FDA's Charge and Questions

CHAIRMAN DURST: Okay. Thanks, Suzanne.

I assume I don't have to read the charge

again, since we know what we are here for. We have

been dealing with this for two days now.

To remind you, we don't have to come up

with, certainly, a vote. We don't even have to

come up with a consensus. What we want to do is

provide the FDA with some guidance and

recommendations on the draft report that we have

been discussing the past two days.

I have asked several members of the

Committee to try to summarize the remarks or

comments and discussion that has gone on for the

past two days.

I have asked Marc and Suzanne to summarize

the food allergens part of our first day of

discussions and Ciaran to summarize the celiac

disease, the gluten portion.

After their presentations, we will open it

for discussion to make any comments agreeing,

disagreeing or just filling in some blanks that

they think are important.

Then, after that part, we will go back and

deal with the general questions that are on the

charge sheet. That I think should go fairly

quickly after we have agreed on some of the other

items that we are going to discuss.

Even though it is out of order, maybe we

will start with Dr. Kelly with the discussion of

gluten, since we are waiting for the page on the

food allergens.

Ciaran.

DR. KELLY: Sure. Ciaran Kelly here.

Should I read the questions, or just go straight to

the answers?

DR. TEUBER: Yes, please.

DR. KELLY: So the first question is

regarding gluten and celiac disease. Is there a

distinct subpopulation of individuals with celiac

disease that have an increased sensitivity to

gluten?

If so, for the safety-assessment-based

approach, is the proposed uncertainty factor for

intraspecies differences tenfold sufficient to

ensure that exposure levels will be below the level

of sensitivity for this highly sensitive

subpopulation? If this uncertainty factor tenfold

is not sufficient, what uncertainty factor should

be used?

Sensitivity to gluten does vary from one

individual to another at the level of clinical

symptoms. However, symptoms of celiac disease do

not parallel small intestinal mucosal injury as

assessed by small bowel biopsy histology, which is

the widely accepted quantitative method of

assessing gluten-induced injury in celiac disease.

There are insufficient available data to

state with any certainty or to what extent

individual variations influence the intestinal

mucosal changes of celiac disease in response to

specific levels of gluten exposure. Thus, it is

not possible currently to assign a reliable

uncertainty factor for intraspecies differences in

gluten sensitivity.

The Committee is uncertain as to whether

or not it is appropriate to apply an uncertainty

factor for intraspecies variation in the

immunological responses to gluten and celiac

disease that is based on the standards normally

used for toxicology studies.

The magnitude of the uncertainty factor

will also be influenced by the level of individual

variation observed in the studies used to determine

that threshold. The choice of an uncertainty

factor for a dietary gluten threshold will also be

influenced by the ability to measure the gluten

content of foods.

It is likely that the gluten threshold

together with a modest or moderate uncertainty

factor will lie close to the lower limits of

performance of the currently available assays and

this may, at least in the short-term, dictate the

measurable threshold.

CHAIRMAN DURST: Okay. Does anyone have

any comments or discussion on Ciaran's presentation

on that question?

(No response.)

CHAIRMAN DURST: I guess most people are

in agreement on that. Very good.

DR. KELLY: The second question: Is it

scientifically sound to use data from short-term

clinical studies that evaluate the effects of acute

gluten exposure to predict the effects of long-term

gluten exposure in gluten-sensitive individuals?

What uncertainty factor is appropriate for

thresholds developed using available short-term

clinical studies in order to prevent adverse

effects associated with chronic effects.

Data from acute challenge studies that

examine intestinal mucosal changes in response to

brief exposure to gluten peptides of several hours

or days' duration are not widely accepted as a

valid method to determine a gluten threshold.

However, there is general acceptance in

the medical and scientific community of studies

that examine mucosal responses to several weeks or

months of exposure.

If threshold values are based on challenge

studies that examine in a quantitative fashion the

mucosal responses to several weeks or months of

gluten exposure, then the uncertainty factor needed

for chronic exposure will be minimal.

Additional valuable data are also

available from other countries, particularly in

Europe, that have many years of experience with

enacted threshold values. Those data may also

reduce concerns regarding the need for an increased

uncertainty factor based on prolonged duration of

gluten exposure.

Since a determination of threshold values

must be made in the context of incomplete and

evolving medical and scientific knowledge, the

Committee endorses the Working Group's finding that

any threshold value that may be set for gluten must

be continually reevaluated, and, if new information

warrants, be adjusted.

CHAIRMAN DURST: Erica.

DR. BRITTAIN: Okay. I guess I just want

to summarize. I have a slightly different

perspective on this. I don't think my bottom line

is really that different. But from a statistical

perspective, it is hard for me to know that a

three-month exposure study would tell me everything

about the cumulative, chronic exposure; so, I would

be a little more uncertain than some of the people

on the Committee were yesterday.

However, that concern is softened by the

fact that we do have the observational data that

seem to suggest a similar effect; so, I think my

bottom line is probably pretty similar to yours,

although maybe a little more uncertainty.

CHAIRMAN DURST: Thank you.

Anyone else?

David.

MR. ORYANG: Yes. David Oryang. I think

your concern is well-taken. I think the key thing

here is to remember that if it is clearly

documented or at least if evidence documents how we

came about determining that safety factor, that is

the key thing.

As long as it is documented and

transparent, then people will be able to comment on

it or provide better information to determine

better safety factors.

The clear thing is that there should be a

good documentation of how the safety factor came

about. I think that is the key so that experts and

others who review it can then give better comments

or suggestions on how to improve it, if it doesn't

seem right.

CHAIRMAN DURST: Okay. Thank you.

Any other comments?

(No response.)

CHAIRMAN DURST: I guess we can proceed.

Ciaran.

DR. KELLY: The third question: Are

current data sufficient to conclude that a portion

of celiac patients are or are not also susceptible

to gluten proteins naturally occurring in oats,

i.e., prolamines and glutelins, if not, what

additional data is needed to draw such a

conclusion?

Published data indicate that the majority

of individuals with celiac disease do not

demonstrate significant symptoms or signs in

response to oats.

A meta-analysis of these published studies

may serve to strengthen this conclusion. There are

a very small number of documented cases where

individuals with celiac disease showed an

immune-based response to oat proteins.

However, the low frequency of these

reports indicate that the overall approach to

setting a threshold for gluten should not be unduly

influenced by the relatively minor concern

regarding oat-sensitivity. Of greater concern is

the issue of cross-contact leading to low-level

contamination of foodstuffs with the known toxic

gluten proteins.

CHAIRMAN DURST: Thank you.

Comments?

(No response.)

CHAIRMAN DURST: Good. Thank you.

DR. MALEKI: Well, I have one.

CHAIRMAN DURST: Oh, I'm sorry.

DR. MALEKI: Soheila Maleki. Just one

comment. The other concern that I think wasn't

mentioned is that limiting the food choices of the

people that are celiacs is probably self-included.

I just wanted to mention that.

CHAIRMAN DURST: Okay.

DR. KELLY: Then, the fourth question:

Are all individuals with celiac disease equally at

risk for developing consequences -- for example,

cancer -- and increased mortality from the

long-term ingestion of gluten?

Are current data from clinical studies or

from individuals with celiac disease on a

gluten-restricted diet sufficient to estimate the

magnitude of any increased risk of mortality for

these individuals?

The outcomes of celiac disease vary

widely, from lifelong silent disease to fatal

malignancy. However, at this time the only

identified risk factor for bad outcomes, including

death from malignancy, is poor or absent compliance

with a gluten-free diet.

Prolonged, strict adherence to a

gluten-free diet clearly reduces the risk for

gastrointestinal symptoms and nutritional

deficiency states such as anemia and osteoporosis

and celiac disease.

The available data, though limited and

imperfect, indicate that prolonged, strict

adherence to a gluten-free diet also reduces the

risk for malignancy.

Thus, instituting measures that facilitate

compliance with a strictly gluten-free diet are the

only known approach to reduce the overall risks

associated with celiac disease.

Comments? Questions?

(No response.)

CHAIRMAN DURST: Good job, keep going.

(Laughter.)

DR. KELLY: Question five: Is

evidence of minimal intestinal pathological change,

for example, increased intraepithelial lymphocytes

following a gluten challenge, an appropriate

symptom upon which to base a LOAEL for long-term

consequences?

Are other biomarkers such as clinical

symptoms or more severe intestinal pathological

changes more accurate predictors of long-term

consequences?

Yes, the characteristic intestinal

pathological changes of celiac disease, for

example, reduced villus-to-crypt ratio and

increased intraepithelial lymphocyte counts

constitute the widely accepted gold standard for

celiac disease diagnosis.

These changes are also widely accepted as

the gold standard method for evaluating disease

activity following a gluten challenge. Other

disease markers such as symptoms, antigliadin

antibody, tissue transglutaminase or endomysial

antibody levels, or measures of mucosal

permeability are considered of secondary value in

quantifying disease activity.

CHAIRMAN DURST: Comments?

(No response.)

CHAIRMAN DURST: Very good, Ciaran. Thank

you very much.

I don't know, as far as what is allowed,

may I ask the Threshold Working Group if they have

any additional questions or clarifications they

need on those points?

(No response.)

CHAIRMAN DURST: Well, we are moving,

then.

Okay. Suzanne and Marc, are you ready to

present to your comments?

DR. TEUBER: Okay. Suzanne Teuber here.

For this discussion, it turned out as we went

through the questions that there were actually some

that we did not specifically address in the

Committee discussion, and so there will be

discussion that is needed this morning in order to

answer the charge

For the first question: are there distinct

subpopulations of highly sensitive individuals

within the allergic population for each of the

major food allergens?

We know that there are huge differences in

threshold doses and a continuum of reaction

severity upon ingestion from mild to

life-threatening for each of the major food

allergens.

However, it is not possible at this time

to identify "distinct" subpopulations of

individuals by clinical criteria, previous

frequency or severity of allergic reactions, or

threshold responses on a double-blind,

placebo-controlled, food challenge within

populations sensitive to specific allergens for

which thresholds or uncertainty factors can be

identified. That is number one.

Any thoughts on that one?

(No response.)

DR. TEUBER: Okay. Then, number two, I

will hand over to Marc here.

CHAIRMAN DURST: Erica.

DR. BRITTAIN: Yes, I just want to add one

comment. In terms of all of these questions about

the food allergies, from my statistical

perspective, again, to me the first step here of

when the Working Group actually is setting a

threshold is to define what the precise goal of the

threshold is; in terms of the sensitive population

or the overall allergic population, what risk level

is acceptable. I think that is the first step.

DR. SILVERSTEIN: Actually, these are

additional comments under one, so why don't I

continue with that.

22

DR. TEUBER: Okay.

DR. SILVERSTEIN: The next part of the

question of number one says: "If so, for the

safety-assessment-based approach, how to propose

uncertainty factors for intraspecies differences

10-fold, under severity of responses for this

sensitive population tenfold, sufficient to ensure

exposure levels will be below the level of

sensitivity for the highly sensitive populations?"

The uncertainty factor for sensitive

populations is unknown when considering food

allergy and immune response as compared to classic

safety assessment and toxicology.

The selection of an uncertainty factor for

allergens should be informed by the distribution of

the NOAELs and the LOAELs using measures of the

spread of data such as standard deviation,

interquartile range or ranges.

If reproducible, subjective responses in

patients with a history of life-threatening

anaphylaxis are included in setting LOAELs and

NOAELs, the uncertainty factor might be lower than

10.

The selection of thresholds for allergens

should be informed by evidence of the thresholds of

NOAELs and LOAELs. However, as we mentioned, bias

due to exclusion of the most sensitive individuals

who have experienced life-threatening allergic

reactions, anaphylaxis, require caution in using

currently available data.

All currently available published and

unpublished data should be specifically assessed

for potential selection, referral bias, and other

factors that influence the individuals who are

actually studied.

There is also uncertainty due to variation

between individuals in a population and uncertainty

due to variation within individuals over time.

There are inadequate prospective studies performed

with the goal of seeing if the objective response

thresholds have changed in patients with persistent

food allergy, except in those who are expected to

have developed a tolerance.

A highly sensitive individual might have a

lower or higher LOAEL compared to baseline

depending on such factors as: the season of year;

theoretically, increased histamine release

potential based on activity of conditions such as

allergic rhinitis and asthma, which might be

seasonal; status of an atopic dermatitis; the time

of day; stability of the patient's underlying

asthma; ingestion of other factors such as alcohol,

exercise, pre- or post-ingestion, matrix effects of

food, processing the food, progression of the

degree of their allergy based on IgE target,

epitope diversification, antibody increases, and

other variables which are known to individuals in

the field.

The next part of the question: If these

uncertainty factors are not sufficient, what

uncertainty factors should be used for the

safety-assessment-based approach?

A concrete number was not offered by the

Committee. The Committee noted that the

uncertainty levels of tenfold or a hundredfold had

been used in biomedical toxicology. IgE-mediated

allergic reactions essentially are amplifiers.

They amplify reactions to minute amounts of

allergens.

So, the application of uncertainty factors

to thresholds on the double-blind,

placebo-controlled, food challenge may not be

sufficiently large to handle this variation of

amplification of an allergic response.

DR. TEUBER: That was the entirety of

number one.

CHAIRMAN DURST: I beg your pardon?

DR. TEUBER: It was the entirety of number

one; it is put out in sections.

CHAIRMAN DURST: It is open for

discussion.

Erica.

DR. BRITTAIN: I agree with all that. I

just want to add one point that as an alternative

to the uncertainty factors another strategy is the

modeling approach that we heard the speaker talk

about. I think that is a really promising approach

to assessing risk.

However, even with that, you would need to

make sure that you have data that represents the

entire target population. I don't know that really

is available at this point.

CHAIRMAN DURST: Any other discussion?

Mark.

DR. NELSON: Yes, this is Mark Nelson. I

agree with your synopsis as well, and I think you

have captured a lot of hours of struggle, but one

thing I wanted to clarify. I do agree that the

uncertainty factors should be based on the range,

the largest range possible, of the sensitive

individuals. But I heard, and correct me if I'm

wrong, that some of the studies did in fact include

some extremely sensitive individuals.

DR. TEUBER: Yes, some did. And then we

also had the situation where in some studies the

extremely sensitive people challenges were stopped

at subjective which -- well, that was in some of

the hazelnut study.

But for the Hourihane study, for instance,

they did go on. It was a twenty-fold difference in

one patient, a fifty-fold difference in the other

between a subjective response and an objective

response.

DR. NELSON: Right.

DR. TEUBER: We definitely want things to

be based on objective, when possible, but I am

concerned that in the threshold studies that are

going to be done by the consensus protocol there is

still room for a physician or a patient to decide

to stop.

I mean, they have the ability and informed

consent procedures to stop at any time, and so if

they are recruiting the most sensitive, we may have

folks who back out before an objective response or

where the physician decides, "Ah, you know, they're

complaining of throat swelling, and I can't see

anything, but I'm hesitant to go on."

There, that data of the subjective I think

should be used. It doesn't carry the weight of an

objective NOAEL but certainly could be used to help

estimate an uncertainty factor.

DR. NELSON: Yes. I don't mean at all to

imply that people should be forced to participate

in these studies or continue on, if subjectively

they have lost comfort. What I wanted to point out

was that I understood that the database as it

exists now, there are some studies that do have

very sensitive individuals in them.

DR. SILVERSTEIN: I think this is one of

the most difficult -- this is a discussion point

not a summary point -- issues is assessing

potential bias. There are of course in

randomized-controlled trials a careful focus on

eligibility an exclusion criteria.

In observational studies and in studies of

diagnostic test assessment, the eligibility and

exclusion criteria may or may not be as explicitly

stated.

In any case, when a study does have

well-stated eligibility and exclusion criteria, you

often don't get a description of those people who

are referred or screened and not studied. Because

they are not studied, you usually have less

information.

Because this is a serious life-threatening

condition, because these studies were done at

distinguished academic centers by individuals who

have distinct experience, to appear in such a study

often you need to be referred; and so the referral

bias, we are often not able to assess it.

The strongest studies are those that we

would call "population-based studies." Those would

be the things you would want to look for.

Oftentimes, that is not stated or it is only

implicit in understanding that this study, as study

subjects, the study subjects are often referred to

as "the population," because we are making

inferences about similar subjects. But the study

subjects, because of referral, weren't truly

representative of the population of allergic

individuals.

That is the concern, and that is the

challenge the FDA will have in evaluating this

literature, but it needs to be looked at for all

available literature.

CHAIRMAN DURST: Dick Durst. I would also

like to remind the Working Group that when they set

these thresholds they also have to be cognizant of

the analytical methods.

When you put on an uncertainty factor or a

safety factor onto these thresholds, you have to be

tempered by the knowledge of what analytical

methods can do as far as verifying and making sure

that foods comply to these thresholds. At the

present state of the art I think there are some

problems in this respect, so that this has to be

taken into consideration.

DR. TEUBER: Suzanne Teuber. When you go

through the back of the binder and look at the

different foods and the sensitivity and actually

for the objective, quantitative measurements of

what can be done, if an uncertainty factor is

applied that is too large, you will be below those

levels. You end up then with the analytical method

as the method of choice for some of these.

Additionally, if you consider, the Working

Group should consider, that the serving size may be

subject to discussion when you are determining how

many parts per million may be acceptable.

I guess I really need to retract that. It

is just that based on clinical experience now so

many patients go by that first subjective response

in the mouth of having some tingling or some

itching and have been able to stop -- I'm sure if

Anne Munoz-Furlong were here she has thousands of

stories of this; I have hundreds and hundreds.

So, the serving size where a person may

notice a subjective response may be much smaller

than the 100-gram serving size that may be used to

calculate how many parts per million are going to

be acceptable before a NOAEL with an appropriate

uncertainty factor applied is reached.

CHAIRMAN DURST: David.

MR. ORYANG: Yes, David Oryang. Just

adding, I think, yes, if a safety factor is derived

in a science-based way, even if the sensitivity is

much greater than the sensitivity of any of the

test methods.

I think it should be transparently

communicated, and then the decision will be made,

hopefully, at FDA as to how to resolve that, maybe

through labeling and saying, well, it is clear that

we don't have the methods to be able to detect that

level, those parts per million that individuals are

sensitive to.

I think the products could be labeled

appropriately so that those at risk can make the

choice whether or not to go ahead and take on that

risk, but I don't think we can temper the science

based on available methods, necessarily.

They are two different things: there are

the safety issues and then how do we apply what we

know. If there is no method of applying or

detecting that level of sensitivity, then it needs

to be transparently presented, I think, as opposed

to altering the safety factor so that it is within

a range of detectability.

CHAIRMAN DURST: Erica has a comment.

DR. BRITTAIN: Yes. This is touching on

something that I tried to say yesterday, and I

think I said it badly. Let me take one more chance

to try to say it again. At some point the FDA will

be establishing a threshold for each allergen. I

hope it will be a really safe and really

conservative threshold.

Presumably, it will be above the level of

detection. It means there will be this gray zone.

Some products will fall in the gray zones between

the level of detection and the threshold that was

set.

I am wondering if there would be any value

in being able to provide that information to the

consumer that this product say something like

"contains peanuts but below the allergenic level."

So, that mom who was here on Wednesday who

said, "I don't want this threshold based on a

statistical estimate, I want it based on fact," if

she does not want to take any chance at all, she

can see, "Oh, it's in that gray zone, and I don't

want to take any chance."

Or, if someone has had experience in the

past with reactions in that gray zone, then they

may think they are the rare individual that cannot

tolerate that level. I just wanted to throw that

out one more time.

CHAIRMAN DURST: Okay. Carol and then

Soheila.

DR. WASLIEN: Well, one of the problems I

see with the detection method is that we don't even

know what the allergen is in a good number of

cases, so having a good detection method may be

detecting the wrong thing. It is that kind of

problem with detection methods, too, that lead to

the uncertainty factor.

DR. TEUBER: Suzanne Teuber. I actually

disagree with that. Because if you have a method

that is just aiming to detect the food, the

proteins in the food that are allergens are going

to track along with that measurement. Just as the

measurements for gluten, the glutelin fraction

will track along with the gliadin fraction is

measured.

If all of our tests so far that are

measuring the food given are based on total

protein, it all seems to track together in a

proportionate way. I think that is okay.

DR. WASLIEN: You don't see a case where

protein would be separated and only one of the

protein fractions would be included in the food and

therefore safe?

DR. TEUBER: Well, yes, where you have

casein used or alpha-lactoglobulin or whey as

separate fractions.

DR. WASLIEN: Yes.

DR. TEUBER: There you do have a situation

where the challenges that are done to determine the

NOAELs and LOAELs have been with the whole protein.

Actually, yes, I do see that as a potential

problem, but the actual challenge dose that would

elicit -- actually, yes, that is a good point, to

think about the separation there.

DR. WASLIEN: Yes, particularly with milk

protein.

DR. TEUBER: Yes.

CHAIRMAN DURST: Okay. Soheila and then

Petr.

DR. MALEKI: Soheila Maleki. One comment

for Erica. Well, currently the gray area exists

that is in "may contain" labeling. You are asking

for something that already exists that the consumer

is asking to take away. They want a more

definitive response.

Second, there are analytical methods that

can actually go down to measuring one molecule that

is completely insignificant. Like I said, on this

tablecloth here somebody can detect peanut or

wheat, if they wanted to. You can go down to a

molecular level, and washing won't get rid of it,

that these people will not react to it.

There is a limit where allergic people

will not react, and we do have the detection

methods. What happens is we don't have the

threshold data on the individuals.

Still, the consumer is asking for us to

make some kind of decision on the best data

available. I think that is something important to

think about for the consumer, because that is why

they are frustrated.

CHAIRMAN DURST: Petr.

DR. BOCEK: Petr Bocek. Soheila pretty

much said what I wanted to say because that would

apply to every product to assess their methods.

One would be saying, okay, this product doesn't

have, let's say, peanut at the allergenic

threshold, yet it contains.

Here you go, you are restricting the

consumer from basically anything. If you take the

PCR, you are going to detect peanut everywhere. So

I think it is absolutely impossible to apply that

method. It has to be the allergenic threshold

only. That is what the consumer wants, and that is

going to make it clear. I absolutely agree with

that. I don't think it is practical.

CHAIRMAN DURST: All right.

Mark.

DR. NELSON: Mark Nelson. Just to add to

that from the practical standpoint of actually

manufacturers labeling their products, we want to

communicate clearly to the consumer. We don't want

to add anymore gray to it, to the situation at all.

If a threshold can be established where

the great majority of allergic individuals for a

particular allergen can be benefited by having

that information, great, but we don't want to add

to the gray.

CHAIRMAN DURST: Thank you.

Any further discussion on this question?

(No response.)

CHAIRMAN DURST: That was good. The next

question.

DR. TEUBER: Number two: “Is the initial

objective response seen in a clinical challenge

study always an adverse effect that poses risk to

human health?,” is the first part of the question.

We said there was discussion of no, it is

just uncomfortable, but then we kind of wrapped it

up. We were making comments at the end, so here is

what we ended up writing.

"Yes, if there is an objective response to

a food in a double-blind, placebo-controlled study

performed in a patient with IgE-mediated food

allergy, this is an adverse effect that poses risk,

albeit usually low.

"The findings from an objective response

in a double-blind, placebo-controlled, food

challenge is sufficient for physicians to make

recommendations that patients avoid specific foods

and change lifestyle to avoid risk of

life-threatening allergic responses. This is

sufficient to conclude that objective responses are

associated with allergic reactions that pose risk

to human health."

Any comment on that part of the questions?

DR. MALEKI: Just one quick comment.

CHAIRMAN DURST: Soheila.

DR. MALEKI: Soheila Maleki, sorry. One

quick comment that, yes, the subject of milk came

up and I think when Petr and I were making

comments on that we were thinking instead of just

"to human health," we were thinking

"life-threatening." That is why we said no. I

agree, I believe that we can forward your response.

DR. TEUBER: Okay. Is it scientifically

sound to use this response to determine a LOAEL in

the absence of a NOAEL? We said no, reactions to

the first dose, because that was implied in the

question, mean that the LOAEL could be just a trace

lower or conceivably a thousand-fold lower. Such

data are not useful in the decision-making process.

Then, the next part of the question: For

the safety-assessment-based approach, is the

proposed uncertainty factor of tenfold sufficient

and appropriate to use in the absence of a NOAEL?

We said no, such data should not be used at all.

Then, more on that question: If a

clinical challenge study reports a subjective

response of a lower dose than the dose that caused

an objective response, should that observation be

taken into account when determining the appropriate

uncertainty factor?

Again, we have extremely limited data on

subjective responses and the relationship to

objective at this point. We said yes, if using a

subjective response as the LOAEL, the uncertainty

factor would be lower.

If using the objective response but

subjective responses were also recorded, the

uncertainty factor -- and this is a point that we

can discuss more today -- should probably extend to

cover the dose at which the subjective response

occurred and likely a bit further to account for

the individual variation.

I might like to stop right there, because

we did not actually specifically discuss that. As

we were coming up with this, we wrote that. Again,

our whole point is that we want the LOAELs to be

based on objective data.

However, if you have subjective data as

well, which the consensus protocol for threshold

studies they are now going to be recording this

subjective data, this might be very useful in

judging what these uncertainty factors should be.

I had put here that the factor should actually

extend a bit below that. That was without any

discussion yesterday.

How do you all feel about that? Again,

this is just our recommendation.

CHAIRMAN DURST: Carol.

DR. WASLIEN: Hi, Carol Waslien. Since we

don't know the individual factors that would

influence the subjective or objective reactions, I

assume that we have to include that.

I would hope that we would accumulate data

that says this is the kind of range within an

individual that you might see, because that makes

it exceedingly low in a sense or the lower limit.

Over time, I would think that kind of data should

be accumulated in trials.

CHAIRMAN DURST: Petr.

DR. BOCEK: Well, we didn't discuss it

originally, but I absolutely agree with that. In

practice, imagine you give a patient in a

double-blind challenge study 100 micrograms of

peanut, and they say, "I'm itching all over. My

mouth is tingling. I don't feel well."

You go on and they develop hives only at

100 milligrams. There is absolutely no way I will

say that 10 milligram is the norm, because they had

subjective symptoms which to me are significant at

100 micrograms. I certainly agree with that

approach.

CHAIRMAN DURST: Marc.

DR. SILVERSTEIN: I would like to make a

comment and an observation. In our understanding

the drugs, we have had tremendous benefits by

improved methodology that have been developed by

investigators, epidemiologists, and statisticians

in response to regulatory requirements that were

developed and industry was focused on because of a

need to develop drugs to be marketed.

It seems to me that in the setting of

thresholds for allergens there is an opportunity

here to specify a set of potential biases, a set of

potential confounding factors that the leading

investigators in the allergic diseases would use in

establishing these consensus protocols for

double-blind, placebo-controlled, food challenges.

If, for example, biases such as referral

bias, selection bias, disease-spectrum bias,

verification bias were judged to be important

factors.

A set of standards for performing these

studies could include reporting this information.

Journal editors have also been influential in

improving the quality of studies by saying, "As a

characteristic for publication, we would like you

to meet these criteria."

If there were a set of confounding factors

-- time of day, season, exercise, concurrent

medications -- that are thought to be important by

clinicians, by allergists, if those factors were

specified as factors that should be reported in the

collection of data, then as we go to making

judgments about policy and regulations, we would

have a more uniform and higher quality data to base

those regulations.

If we think back about the way in which

our knowledge of clinical trials has benefited by

the need to have well-designed Phase I, Phase II,

Phase III clinical trials, I think we are at an

analogous point here.

These factors that we have, these

potential sources of error and the potential

incomplete data, I think could be used by those on

the cutting-edge in doing these studies or

designing these studies, so that as these studies

are conducted within one, two, three years we would

have a body of data to make better judgments about

the setting of thresholds. I think there is a very

important opportunity here to influence the type of

data that will be available in a couple of years.

CHAIRMAN DURST: Suzanne.

DR. TEUBER: A note to file, this would be

an excellent RFA. Again, these studies are

extremely expensive, and so the only ones that are

underway right now or being planned are those being

sponsored by industry, graciously, to help

determine these thresholds. This is an excellent

opportunity for us, as a Committee, to have in our

minutes the need for more funding for this.

CHAIRMAN DURST: Jean.

MS. HALLORAN: It does seem like,

following up on this, one of the critical questions

is the relationship of subjective responses to

objective responses. I don't know whether it has

been studied so far how well these correlate or

whether that is something that needs further study.

However, if in placebo-controlled,

double-blind studies a subjective response is a

very good indicator of a subsequent objective

response, then that would lend more validity to

using the subjective responses as a factor in

determining a threshold.

I was wondering if anybody knows whether

that kind of correlation has been done up to this

point; and if not, whether perhaps we might want to

recommend research in this area?

DR. TEUBER: Suzanne Teuber. That

actually is being incorporated into the current

consensus protocol for threshold studies, that

subjective responses will be recorded carefully,

and then the goal is to proceed to an objective.

This will allow us to have data on how those are

related in a wider range of patients.

Because right now, there are only a few

reports of proceeding on to an objective response

after initially having a subjective one, I should

also note, a subjective response verified by repeat

challenge with negative placebo. It will come.

CHAIRMAN DURST: Any further discussion on

question two?

DR. TEUBER: Well, actually we have more

on question two. That was just one little

subsection there.

(Laughter.)

DR. TEUBER: Suzanne Teuber, I should say.

CHAIRMAN DURST: I thought we went through

these.

DR. TEUBER: Oh, actually, no, you're

right. That was basically it, but we have one

little bit more that we wanted to add to this

discussion of number two. We wanted to note again

this recruitment problem.

Of note, recruitment of the highly

sensitive subpopulations to threshold studies may

be enhanced by recording subjective reactions that

are reproducible to the active dose but negative to

the placebo, two challenges of each, with an option

of stopping at that dose.

In threshold studies, highly sensitive

patients may or may not be willing to proceed to an

objective response or the physician may not be

comfortable proceeding.

There is acknowledged controversy that is

appropriate about the applicability of LOAELs that

are subjective. Objective responses are preferred,

with the concern that it be demonstrated in studies

that extremely sensitive subjects have been willing

to participate in, otherwise an uncertainty factor

greater than 10 may be needed because we just don't

have enough data.

Again, we wanted to note the previous

comments that Mark had made about using a range of

threshold values in determining the uncertainty

factor. We actually raised these points in our

discussion here again.

Then, proceeding on to number three: In

the absence of specific data that would allow

thresholds to be established for each of the major

food allergens, is it scientifically sound to use

the threshold established for a single food

allergen -- for example, peanuts -- as the

threshold for all major food allergens?

We really did not discuss this much

further. I think it could use a little bit more

discussion. I said no, it appears from the

available data that soy thresholds may be higher.

Such labeling would then restrict diets

unnecessarily as well as pose hardships to

industry.

We really didn't discuss the fact that for

other allergens we don't have much data or adequate

data, and it might be reasonable to use the most

stringent one until other data are available.

Any thoughts on this?

CHAIRMAN DURST: Erica.

DR. BRITTAIN: I mean, that sounds good,

just as long as you really are totally competent

about what is the most stringent factors.

CHAIRMAN DURST: Carol.

DR. WASLIEN: Well, I think there are a

good number of tree nuts, for example, that we have

limited data on. Hazelnuts, yes; but other tree

nuts, we don't have the data on. Other allergens,

the other 200 that aren't part of the 7, we have to

use something in their place.

Perhaps, until those ranges become more

clearly established, we are safer at using a peanut

allergen that is the most likely to show a response

for those foods until they are proven otherwise,

sort of guilty until proven innocent almost.

CHAIRMAN DURST: Marc.

DR. SILVERSTEIN: Marc Silverstein. It

seems to me that a parent making a decision about

food exposure for a child, a physician making a

decision to recommend diet for a patient and an

agency making a policy recommendation for consumers

and industry, all are faced with difficult

decisions.

The decision threshold and the potential

decision you might make might be weighed not only

by the likelihood of making a correct or incorrect

decision, but the consequences of making a correct

or incorrect decision.

Obviously, a parent making a decision for

a child, a physician making a decision for a

patient, and an agency making a decision for a

population and industry all have different

thresholds in terms of the value of a

false-positive or a true-positive recommendation to

avoid or not avoid, or to change diet or not change

diet, or to label or not label a specific level.

I do think that we are, in some sense,

changing our perspective. We are putting on our

hats or our roles as parents and individuals. We

are putting on our judgments as clinicians or

health care providers, and we are putting on our

roles as policymakers.

I actually think we should be cautious in

making judgments about how sensitive or specific

our thresholds would be for a decision about our

children, our decisions about our parents, and our

decisions about our public population.

I am not sure that I would want -- in

fact, let me phrase it positively -- I would not be

comfortable making a decision to be very

conservative or very liberal, if you will, high or

low, highly sensitive, or specific when I shift my

domain from that what I would do for my child to

that what I might do for my patient or that what I

might do for an agency's decision.

I do have some sense that I would rather

than say we should be conservative and extrapolate

from what we know about this other class of antigen

I might say that right now we have insufficient

data and cannot make a recommendation.

CHAIRMAN DURST: Dick Durst. Would it be

safe to say that for those allergens for which we

have sufficient data we would set a realistic

threshold based on that; and for those that the

data is currently insufficient, that we would use

the threshold of the most sensitive?

DR. SILVERSTEIN: That would be safe to do

that, but I feel that I could make some judgments

with regard to individual patients. I would be

cautious about making such a recommendation on a

policy basis for that. The consumer might and the

industry might want guidance. We may have

insufficient data to be able to provide that

guidance.

CHAIRMAN DURST: Soheila and then Jean.

DR. MALEKI: Soheila Maleki. Well, I

mean, I agree with you, Marc. I understand your

thought, but, again, going back to what the

consumer wants, the consumer wants us to err on the

side of caution.

I mean, of course that is something that

you don't want, to lump everybody in. I a hundred

percent agree with you. However, when you hear

from the consumers, they would prefer that you err

on the side of caution even in the absence of data,

which is what it essentially is asking.

In the absence of data to pick the most

sensitive food and set a threshold, I think is more

comfortable to the consumer or probably would make

them feel better than to say nothing on the label

at all.

DR. HEIMBURGER: Or, to leave it

ambiguous. Doug Heimburger. Or, to leave it

ambiguous, to say "It may contain" or whatever.

CHAIRMAN DURST: Jean.

MS. HALLORAN: Yes. I think we should

keep in mind that we are talking about a threshold

for labeling only. This is not a threshold for

excluding the product or taking it off the market

or anything like that. Particularly, the problem

comes with what is the threshold for requiring

somebody to say, a company to say that "This

product contains soy"?

I think to err slightly, perhaps, in the

direction of a lower threshold is the appropriate

course here until it can be shown that certain very

low levels of soy do not pose any hazard to a

person with allergies.

Because for most people it is not

relevant; it will not be of interest at all. What

we are trying to do here is to try to provide

information to consumers with a very special

concern.

CHAIRMAN DURST: Mark.

DR. NELSON: Yes, Mark Nelson. Earlier,

we were talking about gray areas. I think if we

went to establishing a threshold based on the most

sensitive or the most problematic allergen we would

be completely in the black area.

Echoing some of Jean's comments, I think

if we are to establish a threshold based on the

most problematic allergen, I can't imagine there

would be too many companies that would go through

the process of reformulating a product for soy to

meet that lower, tight threshold knowing full well

that there are data that exist that we are getting

close to better information for a higher threshold

for soy, and then going back and reformulating

again to meet that. In effect, we would be

postponing providing useful information to a good

portion of the allergic population, if we were to

take that tack.

CHAIRMAN DURST: Jeff.

DR. BARACH: Jeff Barach. I would

certainly agree with Mark's comments. One thing I

would say, though, is that we really from my

observation have fairly good data on at least four

of the major eight, and that is comforting to me.

I wish we had more, but that seems to be what was

presented to us.

If we think about what could happen to

those other four and we use, say, the lowest level

for the allergen of highest activity, that bothers

me a little bit because of what Mark said.

I think what we should recognize, though,

is that we do have sort of a default position.

Unfortunately, we have a zero tolerance now for

those allergens, so those products would be

labeled.

It is not like they wouldn't be labeled;

it is just that they don't have a threshold. If

there is any there, any detectable, then it would

have to be labeled. There is a default. We don't

have to in my mind assign a threshold for

everything at this point and still protect the

consumer.

CHAIRMAN DURST: Further discussion?

Ciaran.

DR. KELLY: So is the default threshold an

analytical de facto?

DR. BARACH: I would say it is more

ingredient-based than analytical.

CHAIRMAN DURST: Petr.

DR. BOCEK: Petr Bocek. When you say

"ingredient-based," so that goes back to "may

contain," or what does it mean?

DR. BARACH: It goes back to the system

that we are currently using.

DR. BOCEK: Okay. That is the system we

are trying to change.

DR. WASLIEN: Don't you mean that you have

to list all of the ingredients of a food, so it is

not the "may contain"? Is it the list of

ingredients? Isn't that what you are referring to?

It doesn't mean he is saying we will stick with the

"may contain" labeling option. It is if soybean

lecithin is added to a food, it is on the list of

ingredients.

DR. BARACH: That's right. The "may

contain" part covers the possibility of

adventitious presence or a contamination during the

manufacturing as well. We have the list of

ingredients, and then we have the "may contain" for

small amounts that may enter the product.

CHAIRMAN DURST: That was Carol and Jeff

responding.

Now, Mark.

DR. NELSON: It also covers not only

additives and potential cross-contact, but it also

includes the processing aids which are

intentionally used but really serve no function in

the finished product, but there may be trace

amounts of it in the product. At this point the

law requires us to do it, to label those as well.

DR. TEUBER: Suzanne Teuber. Actually,

for those processing aids, many of them will fall

under the petitioner notification process.

DR. NELSON: They could.

CHAIRMAN DURST: Any further discussion?

(No response.)

CHAIRMAN DURST: No? We will move on.

DR. TEUBER: Suzanne Teuber going on here.

The next part of that question was actually: if so,

which food or foods could serve this function; if

not, is there a more appropriate method to be used?

I think people discussed that here.

The question is, though, do we have a

consensus on that for Dr. Durst to write up a

statement? I'm not sure that we do. Do you feel

that we do? Because basically this might or might

not be an appropriate way to proceed. There were

concerns raised for and against. I don't think we

had really consensus.

CHAIRMAN DURST: Yes. Well, as I say, I

don't think we have to reach a consensus as long as

we can provide some guidance to the FDA as far as

directions for them to go.

I could ask Steve at this point if he

wants any further clarification on that point.

DR. GENDEL: No, you are correct, it is

not necessary to reach a consensus but simply

stating the range of opinions and the basis for

those opinions.

DR. TEUBER: All right.

CHAIRMAN DURST: Okay. Thank you.

DR. TEUBER: Suzanne Teuber. Continuing,

number four, the draft report discusses the

available data on the levels of protein present in

highly refined oils, that is, oil that is

hot-solvent extracted, refined, bleached and

deodorized.

Is there any physiologic reason -- for

example, food matrix effect denaturation of protein

-- why the protein levels in highly refined oils

could not be used as the basis for establishing a

threshold for other allergenic foods? Are there

any other limitations that should be considered in

applying this approach to the eight allergenic

foods?

With this there was complete consensus

that the levels in oils did not apply. The reasons

that were raised included the fact that we have

extremely poor measurement of proteins in oils. It

is very unclear as to their validity.

Secondly, the points raised about

denaturation, changing of epitopes and whether the

proteins in oils actually reflect what folks really

act to.

Then, third, the matrix effect was felt to

be extremely important and has been backed up by

studies showing that fat can affect the threshold

for response.

In addition -- let's see was there yet

another, this is where the printer didn't work on

that -- we had the configuration changes, we had

the measurement problems, and then the matrix. I

believe those were the three that we had covered.

That one we were in complete consensus agreement

on.

CHAIRMAN DURST: Has anyone changed his or

her mind on that?

(No response.)

CHAIRMAN DURST: Well, I guess we are

still in consensus. Okay, that takes care of those

specific questions. Again, I will refer back to

Steve, if he has anything that he would like

further discussed on the food allergen part?

DR. GENDEL: I don't believe so.

CHAIRMAN DURST: Okay. Then, we can move

on to the general questions on the first page of

our charge. I think some of these should be able

to go fairly quickly, since we have laid all of the

groundwork now for it. The first one I will read

the questions, and then we can discuss.

"In addition to the four approaches

identified by FDA for establishing thresholds

(i.e., analytical methods-based, safety

assessment-based, risk assessment-based and

statutorily-derived) are there other approaches

that FDA should consider? If so, please describe

and explain why FDA should consider them."

As I recall, there really weren't many

other options.

Erica.

DR. BRITTAIN: I just have a really brief

comment. It is not really another method, but just

that to me the safety assessment-based and the

risk-assessment based are sort of part of a

continuum.

I don't see them as, necessarily,

completely distinct in that I would like to see

more statistical principles brought into the

safety-assessment-based approach.

CHAIRMAN DURST: Okay. Soheila.

DR. MALEKI: Soheila Maleki. I think I

just brought up one method about celiac disease

measurements, and that is, possibility for looking

at T-cell activation, which is more likely to catch

that amplification of response than analytical

methods might. It was just a suggestion, if

anybody else has any ideas.

CHAIRMAN DURST: David.

MR. ORYANG: Yes. I would just like to

echo that. Yes, the risk-assessment based and the

safety-assessment-based methods are really distinct

and separate.

However, I think more statistic[s] could

be brought into the safety assessment-based such as

using distributions within the safety factor or the

uncertainty factor, or, for that matter,

incorporating uncertainty into the threshold by

using a distribution, and then using Monte Carlo

simulation to come up with a result, which is a

distribution, and then maybe looking at the 95th

percentile or whatever values come up from that

modeling to set what the ultimate threshold should

be.

It is really just adding some aspects of

risk-assessment into, some methods of risk

assessment into those factors within the

safety-assessment methodology, because there are

two dimensions of uncertainty.

If you look at the studies that were done,

from what I could see there is, first of all,

uncertainty about how closely or how well we

capture symptoms or signs. They observe signs.

When do the observed signs occur? Should we use

the subjective or objective? Because there is a

range there as to when the sensitivity is

different.

Then, the second part of it is you take

this study population, does that study population

truly represent the overall population? Can you

then take what you have observed here in this study

group and project it to the population?

We can't say a hundred percent of the

study population represents the other population,

so there is uncertainty there. There are really

two dimensions of uncertainty as well as, of

course, the model that we are using.

There is uncertainty based on the method.

There are multiple dimensions of uncertainty.

Unless we incorporate some statistic[s] into this,

I don't think we are doing a good job. I really

suggesting adding some statistic[s] to some of the

factors.

CHAIRMAN DURST: Good.

Carol.

DR. WASLIEN: Carol Waslien. I also think

I'm not sure if population-based studies of groups

exist that are living on gluten-free diets and are

included in the risk and safety kinds of categories

and studies, because they are not clinical trials.

I think there is some value to be added to

looking at population-based evaluations to increase

your ability to include highly sensitive, to

increase your ability to ask or collect more data

on range of responses.

Population-based studies such as that,

although they are crude and messy, would give you a

better idea of the kinds of ranges of responses

that you would see in the free-living population as

opposed to a sample, challenge trial kind of

clinical study.

CHAIRMAN DURST: Well, Marc and then

David.

DR. SILVERSTEIN: I don't have a

suggestion for another approach. However, within

the approach, I was struck by the choice of a 10

percent proportion, a population that would respond

to a milk allergy in an identification of a

hypoallergenic formula to derive a buff sample size

of about 29 subjects using a conventional

95-percent confidence interval to identify a

proportion or a rate of 10 percent as a basis for

what has become an accepted sample size for these

sorts of studies.

Indeed, it is difficult to do the studies.

They are expensive; they are risky; and it takes

time to recruit subjects. These conditions perhaps

are not very common; so, referral in the collection

of an adequate number of patients means that, by

and large, investigators are performing studies of

about that size.

We have a collection of literature which

is maybe sufficient for one purpose but may not be

sufficient to make estimates of thresholds.

Estimates of thresholds are not estimates of a

certainty about an incidence rate of 10 percent,

but it is a measurement of a quantity around which

we can decide the amount of precision needed, and

then derive sample size estimates.

Now, indeed, when those sample size

estimates begin to be large, we have to face the

challenge of how do we do those studies. Do they

require multicenter studies? Do they require

longer periods and more effort to recruit? Do they

require larger budgets?

These are real factors. In the real

world, of course, we must live with the available

data and the available resources as we collect

information. I do think there is an important

distinction between the sample needed to estimate

with fairly reasonable or conventional confidence a

rate of 10 percent and the sample needed to make

estimates of thresholds.

Depending on how precise your estimates

could be, your samples might be about that size or

they might be much larger. There should be more

conscious decision making in setting the size of

those samples as well as thinking about the

thresholds themselves.

CHAIRMAN DURST: David and then Erica.

MR. ORYANG: Yes, David Oryang. Further

about some of the methodologies, a lot of times if

the resources are available, time being one

resource, and the need is there to really go do a

more in-depth analysis, then the risk-assessment-

based approach, looking at dose response and

exposures, would be the method to go, but the data

is not currently available.

I will just suggest that sometimes if one

does take the risk-assessment-based approach, you

can still do it, but there would be a lot of

uncertainty in a lot of the parameters.

To the extent that they can be

transparently presented, maybe it is still viable.

I haven't really seen the evidence that we have

enough on some of the dose responses and exposures.

I know that in this report there is enough

stated in there, and it is suggested that the

risk-assessment-based approach be the preferred

approach for the allergens but not for the glutens.

Maybe that is also because of the acute

nature of the allergens. I mean, the consequence

of not looking further into it seems to be more

acute than in the gluten case.

I concur with FDA in some of these

recommendations. I would just say that if the

resources are available and more information is

available, the risk-assessment-based approach is

the better approach.

CHAIRMAN DURST: Erica.

DR. BRITTAIN: Yes, I really want to echo

what Marc said. This magical sample size of 29

that we kept hearing I really think is not quite,

even if you want to just exclude a 10 percent rate,

which doesn't seem like really enough for the

threshold setting, it is not set up the way I think

people would normally want to set things up to have

appropriate statistical power. That is one

relatively minor point.

It all goes back to what I said originally

that I think you need to set a precise goal of how

much risk is acceptable, and that will drive the

sample size. If that sample size is way beyond

what can be done, then I think you have to go to

the modeling approach with the right data.

CHAIRMAN DURST: Further discussion?

Ciaran.

DR. KELLY: Ciaran Kelly. I just wanted

to reinforce the comment that has already been made

of the value of including population experiences

within the safety assessment.

Unfortunately, not only is there a wealth

of experience, but there are also a number of

publications that document the efficacy in a

population of certain levels of gluten exposure.

I think that if one were to consider the

more conservative thresholds that are currently in

use, there would be the likelihood that those

thresholds would be dangerous as opposed to safe

would seem to be very low.

CHAIRMAN DURST: Mark.

DR. NELSON: Yes, Mark Nelson. I wanted

to come back to comments that Marc and Erica made,

but maybe emphasize it a little differently. I

fully agree with basing this all on science and

having statistical rigor, but we also have to make

sure we are not having to prove we're the enemy or

the good.

We may need to adjust the studies for different

allergens, for example, depending on the severity

of the response, and so on.

We shouldn't forget that the sample size

used to support the hypoallergenic infant formula

was also for a population whose sole source of

nutrition was the infant formula that they were

taking in. I mean, safety was clearly considered

in that situation, and, again, it serves the

greater portion of that population.

CHAIRMAN DURST: Any further discussion?

(No response.)

CHAIRMAN DURST: We are ready to move on

to the second question. I will read it again:

"Are FDA's criteria for selecting and

evaluating the available data appropriate? If not,

should any of the criteria be modified or deleted?

Please describe any changes you would like to see

and why. Are there additional criteria FDA should

consider?"

Who would like to start that off?

DR. BRITTAIN: I think we have already

covered all of this. I don't know if there is

anything else?

CHAIRMAN DURST: Yes, I don't know if

there are any more.

Marc.

DR. SILVERSTEIN: Mark Silverstein. There

was a very useful document prepared and funded by

the Agency for Healthcare Research and Quality. I

mentioned it yesterday, and it was "Methods for

Evaluating the Strength of Evidence." It was I

think prepared by one of the evidence-based

practice centers under contract from the Agency for

Healthcare Research and Quality. It has been

published, and it is available on their Web site.

It basically summarizes a set of methodologic

criteria that you would use in doing, evaluating,

grain- or mice-controlled trials, cohort studies,

studies or diagnostic tests.

In looking through the list of the

criteria that are in the FDA report, almost all of

the criteria are mentioned, maybe not in exactly

the same format as in that publication. It seems

to me, that since it is our taxpayer dollars at

work and other agencies have already developed

this, it would be useful to look at it.

I do believe the diagnostic test section,

which would be relevant for some of the food

challenges, has a different way of phrasing some of

the focus on the selection of the patients.

I do think it would be a useful resource

both to cite and perhaps to look at to see that the

categories suggested by that thoughtful review are

well covered in all of the categories.

CHAIRMAN DURST: Okay. Ciaran and then

Erica.

DR. KELLY: Ciaran Kelly. Yes, I have

just one comment, and it has to do with the

criteria for evaluating analytic methods. To my

mind, transparency of the method, an adequate

description of the specific methodology, would be

important so that potential biases or weaknesses

that are inherent within the methodology could be

examined in addition to the other criteria that you

list.

CHAIRMAN DURST: Erica.

DR. BRITTAIN: I just have a very brief

comment. In terms of beefing up the summary of

studies in the appendix, maybe along the lines that

Marc suggested, one thing that is not in there is

number of patients in each study. That seems like

a critical omission.

CHAIRMAN DURST: Anything else?

Jean.

MS. HALLORAN: Jean Halloran. I was

intrigued by the suggestion from Dr. Taylor who

mentioned that clinicians have data on NOAELs and

LOAELs for their patients in their records. This

is certainly not data which is a

placebo-controlled, double-blind study; it is not

peer reviewed; and it is not many other things.

Given that there seems to be such a dirth

of data, especially for certain allergens, I am

wondering whether any effort could be made or

whether there would be any value to FDA asking for

such data just to somehow get a vague idea of what

is going on in those areas, whether it is possible

to do that, or whether once data is submitted, it

gets a life of its own and it is more trouble than

it's worth?

CHAIRMAN DURST: Okay. That actually

leads us into the third question.

DR. BRITTAIN: Well, I mean, it relates to

this, which is that my concern about that is that

if those studies were done almost exclusively on

patients for whom they were trying to confirm a

diagnosis, they would be really biased.

Even if there is a large number of them,

if the information came from it is really biased

and is wrong, then it doesn't really help that you

have a lot of data.

CHAIRMAN DURST: Soheila.

DR. MALEKI: Soheila Maleki. Just one

comment. The studies that they used for a

challenge, and I think Suzanne can confirm that, is

the doses that are used are often much higher than

threshold doses. Actually, if you use that data,

you would be targeting people that have really high

threshold doses, if they didn't react at that

level.

CHAIRMAN DURST: Margaret.

DR. McBRIDE: I think, though, that those

concerns about that kind of data could be handled

by only including subjects who responded.

Obviously, you don't want subjects that didn't

respond, because you don't even know if they are

sensitive. There are ways still of looking at that

data and possibly having something.

CHAIRMAN DURST: Suzanne.

DR. TEUBER: Suzanne Teuber. That data,

it actually would be very useful to have if we are

trying to start characterizing subpopulations of

patients. Because we know that the vast majority

of patients where there is likely this NOAEL data

that just wasn't published are patients who are

children with atopic dermatitis.

That is a distinct subgroup with usually

not as severe a reaction, and so it might be very

useful to have that, because we still need more

data. We can help characterize more, then.

CHAIRMAN DURST: Since we are dealing with

question three, let me just read it into the

record.

(Laughter.)

CHAIRMAN DURST: "Recognizing that some of

the key studies (i.e., challenge studies) are

ongoing, what if any use of preliminary data that

have not been peer reviewed for establishing

thresholds is appropriate?"

Now we can continue.

David, did you have your hand up?

MR. ORYANG: Yes, I did have my hand up.

Just briefly, about the data. I don't see very

much put in there about expert opinion and clear

methods of being able to elicit some of this kind

of information that is kept in records, and so

forth, from a panel of experts and clearly

documented in a way that can be presented

scientifically in the document to be included as

part of the record in how the decision came to be.

I think it would be useful if there was a

formal process for getting that kind of information

from doctors and other people who can be considered

experts on those specific areas, elicit that

information formally and document it, and then show

a clear method of how you came from the broad

opinion to the narrowed result or determination

that was made.

I think that is also a good process of

getting that kind of unpublished or informal

information which can be looked at as a knowledge,

which is important in decision making.

CHAIRMAN DURST: Soheila.

DR. MALEKI: Soheila Maleki. Again, I

mean, of course any data that any scientist is

beautiful and wonderful to get. But, again, I

don't think that in establishing thresholds, which

is what we are looking at, that the challenge

studies would be.

Again, it is wonderful, yes, if we were

going to divide people into subpopulations. Bottom

line, we are looking at the most severe reactions,

which I don't think will be included in that data

because of the high doses of the challenge, or

often higher doses.

DR. BRITTAIN: And the populations that

would undergo those challenges.

DR. MALEKI: Yes. Essentially for this,

yes. I mean, of course it is wonderful to have the

data. What scientist or doctor or agency doesn't

want more data? But bottom line that is --

CHAIRMAN DURST: Suzanne.

DR. TEUBER: Suzanne Teuber. Just to

address Mr. Oryang's comments, actually the First

Threshold Conference that was held, the paper by

Steve Taylor and all, that was bringing together

people who had unpublished data about thresholds,

and from there it has gone on to the current

consensus protocol.

The data that was referred to by Dr.

Taylor was something that was brought to the table,

but again it was this population that was more

atopic dermatitis. I still think it would be

interesting to have it. It would cost money to get

it, to pay somebody.

However, it is true it would not give

information directly related to the

safety-assessment threshold information we want.

It is just population data that is of interest.

CHAIRMAN DURST: Mark.

DR. NELSON: Mark Nelson. I guess I would

echo Jean's comment and a lot of other people, to

try to get more data. What I understood Steve

Taylor to describe, the specific example he gave,

was Hugh Sampson's work.

This is the extra data from a

peer-reviewed study, which strikes me, hearing the

clinicians speak and others, that this would be

very useful for helping specify the uncertainty

factors we have talked about and also help us maybe

with some of these subjective responses. I don't

know, I haven't seen the data.

CHAIRMAN DURST: Jeff.

DR. BARACH: Jeff Barach. As we heard,

other countries are struggling with the issues of

allergens also. I would just make note that the

European Food Safety Authority and the European

Commission have put together a Directive

2005/26/EC.

I would encourage the FDA to contact the

European Food Safety Authority to find out if they

have any data that would be of interest to our

group as they work on the issue.

CHAIRMAN DURST: Soheila.

DR. MALEKI: Soheila Maleki. I think that

is a really good point, and I also think Health

Canada, but I think they have already talked to

them about it. That is a very good point.

CHAIRMAN DURST: Margaret.

DR. McBRIDE: I understand the concern

about the difference, the potential difference, in

dosing for challenge tests versus threshold

testing. We have already established that if there

is a LOAEL in the absence of a NOAEL, that is not

very useful.

On the other hand, if within that data

there are individuals that prove to be sensitive to

whatever is being test and the NOAEL is available

as well as the LOAEL, that is clearly relevant

data.

CHAIRMAN DURST: Petr.

DR. BOCEK: Petr Bocek. Just a quick

comment to that. Yes, it is very valuable, but it

has to be done with the right population. If that

was a patient where it was a diagnostic challenge,

a question of biology, it is not as valuable as

somebody who has a clear-cut clinical presentation

of anaphylaxis to a food allergen. Yes, that is

the right person to look for a threshold.

CHAIRMAN DURST: Any further discussion?

Ciaran.

DR. KELLY: Ciaran Kelly. I just wanted

to perhaps more directly address the question about

non-peer-reviewed experimentation. Clearly, the

presentation that Dr. Fasano gave to us yesterday

is not peer reviewed and published, but,

nonetheless, it is directly pertinent and relevant

to the question at hand and also is performed by a

well-recognized, expert group of investigators.

Clearly, that is a double-blind,

randomized-controlled trial. Clearly, those data

are highly relevant. Although it is likely that

they will be published within the next year, that

publication process is sometimes very hard to

predict, and it could be much longer than a year.

I would feel personally, and I don't know if the

rest of the Committee agrees, that those data are

highly relevant.

CHAIRMAN DURST: Any further discussion?

(No response.)

CHAIRMAN DURST: Before we wrap things up,

I would also like to ask does anyone have any

specific comments on anything in the report,

anything that jumped out at them that needs some

modification or correction?

Yes, Ciaran.

DR. KELLY: Ciaran Kelly. I do have some

minor suggestions, but I don't want to take up the

Committee's time with it. What is the mechanism?

Should I submit some --

MRS. MOORE: He has asked the question, so

go ahead. Go ahead.

CHAIRMAN DURST: I mean, is it proper just

to provide that to the Working Group?

DR. KELLY: I don't want to waste the

Committee's time with very, very minor things.

MRS. MOORE: Okay. You can send it to me.

DR. KELLY: I can provide that in the form

of a memo.

MRS. MOORE: Okay.

CHAIRMAN DURST: Again, I want to refer

back to Steve. Any specific questions that you

might have before we wind things up here?

DR. GENDEL: (Off microphone.) Yes, I

just want to remind everyone that there is a

mechanism for submitting information through the

docket. You can access that through the FDA

homepage. Any information that is relevant can be

put in there.

MRS. MOORE: Steve, start from the top

because you weren't heard from the beginning of

your statement.

DR. GENDEL: Okay. This is Steve Gendel.

The response was that any information can be

submitted through the docket, which is available.

Any relevant information can be submitted through

the docket available through the FDA homepage.

CHAIRMAN DURST: Okay. I think we are at

the point now that I can ask Mr. Landa to make some

closing comments.

Closing Comments

MR. LANDA: Thank you, Dr. Durst. I will

be very brief. First, I just want to reiterate Dr.

Brackett's thanks to all of the members of the

Committee for lending us your expertise for these

several days. We rely heavily on experts from

outside as well as within the Agency. Of course,

this is one of the principle ways in which we

obtain outside expertise.

The second point, just in case not

everyone heard it, the point Steve Gendel made.

The docket will remain open for another several

weeks until the middle of August. Anyone who has

comments on the report please get them to us

through the docket.

We will consider any comments from the

public as well as the results and the comments,

sort of consensus statements, from this meeting in

making any changes to the draft report and in

taking our next steps after the draft report is

finalized.

The last thing is I would just like to

reiterate the thanks noted earlier to Marcia Moore

and her colleagues for putting on the meeting.

Thank you.

MRS. MOORE: Thank you.

CHAIRMAN DURST: Thank you.

I believe Marcia said the transcript of

this meeting would be available at the end of

August?

MRS. MOORE: Yes.

CHAIRMAN DURST: Without any further

comments, I will declare this meeting adjourned.

Thank you very much everyone.

(Whereupon, at 9:37 a.m., the meeting was

adjourned.)