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

Wednesday, July 13, 2005

8:30 A.M. to 5:50 P.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:

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

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

Jenny Slaughter - Director

Food and Drug Administration Integrity and Ethics Staff

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

GUEST SPEAKERS:

Rene Crevel, Ph.D. - Senior Scientist

Unilever, Safety and Environmental Assurance Centre, United Kingdom

Susan Hefle, Ph.D. - Associate Professor and Co-Director

Food Allergy Research and Resource Program, University of Nebraska

Stefano Luccioli, M.D., - Senior Medical Advisor

CFSAN, FDA Assistant Professor, Georgetown University

Anne Munoz-Furlong

Director, Food Allergy & Anaphylaxis Network

Steve Taylor, Ph.D. - Maxcy Distinguished Professor & Director

Food Allergy Research and Resource Program, University of Nebraska

Robert Wood, M.D. - Professor

Johns Hopkins University School of Medicine

C O N T E N T S

PAGE

Call to Order, Welcome and Introductions

Charge to the Food Advisory Committee

Richard Durst, Ph.D., Acting Chairman 6

Conflict of Interest Statement & Other

Instructions

Jenny Slaughter, FDA 10

Welcome and Opening Statement

Michael M. Landa, Esq., CSAN, FDA 14

Use of Food allergen Thresholds

Catherine L. Copp, Esq., CFSAN, FDA 18

Introduction to Food Allergens

Robert Wood, M.D. 23

Patient Perspectives on Food Allergies

Anne Munoz-Furlong 72

Allergenicity: Analytical Methods

Susan Hefle, Ph.D. 99

Question and Answer Session 126

Oral Challenge Studies: Purpose, Design,

and Evaluation

Stefano Luccioli, M.D.

Senior Medical Advisor, CFSAN, FDA 133

Question and Answer Session 161

Threshold Modeling Approach

Rene Crevel, Ph.D. 170

Food Allergen Thresholds

Steve Taylor, Ph.D. 189

Overview of Approaches to Establishing

Thresholds: Allergens

Steven M. Gendel, Ph.D., FDA 218

C O N T E N T S (Continued)

PAGE

Public Comments:

Tracy Atagi 228

John Carroll 232

Diane Castiglione 242

Will Duensing 247

Martin J. Hahn, Esq. 250

Peggy Mockett 254

Kim Mudd 257

Kim Mulherin 260

Linda Webb 264

Jupiter Yeung 268

Committee Discussion:

Panel Discussion with Guest Speakers

Thresholds for Food Allergens - Questions 272

Adjournment

Richard Durst, Ph.D., Acting Chairman 414

P R O C E E D I N G S

CALL TO ORDER, WELCOME, AND INTRODUCTIONS

CHARGE TO THE FOOD ADVISORY COMMITTEE

CHAIRMAN DURST: I would like to call the

meeting to order.

Good morning. I am Dick Durst, professor

of chemistry in the Food Science and Technology

Department at Cornell University. I was asked to

chair this meeting over the next two and a half

days. I would like to make a few announcements

before we begin our meeting this morning.

I would appreciate it if everyone would

turn off their cell phones, unless they are

expecting a call of a super emergency nature. I

would also like to ask if the guest speakers could

make themselves available for the discussion this

afternoon, I would really appreciate it. We may

have some additional questions.

We have received a charge from the FDA to

give our evaluation of the draft report prepared by

the Threshold Working Group. I assume all of the

members have read that thoroughly. In my opinion,

I it was fascinating.

It was an excellent article and I commend

the Committee for coming up with it. It was very

educational. Not being an expert on food allergens

myself, it was extremely educational, and I was

able to follow it quite clearly.

Our charge is to evaluate this report to

determine whether the approaches that are presented

in there are the only ones or the better ones,

which of the ones that are in there might be the

most appropriate. This is the focus of our meeting

today, both on the food allergens and on gluten.

Let me also begin by asking the committee

members to introduce themselves. We will start

with Dr. Silverstein.

Marc, would you start it off?

DR. SILVERSTEIN: Good morning. My name

is Marc Silverstein, and I'm a general internist

and geriatrician at Baylor Health Care System in

Dallas.

DR. TEUBER: Good morning. My name is

Suzanne Teuber, I am an allergist at UC-Davis.

MR. ORYANG: Good morning. I am

David Oryang. I am a risk analyst and agricultural

engineer at the United States Department of

Agriculture, Animal and Plant Health Inspection

Service.

DR. KELLY: I am Ciaran Kelly, and I am a

gastroenterologist at the Harvard Medical School in

Boston.

DR. MALEKI: I am Soheila Maleki. I am a

scientist with the USDA.

DR. BRITTAIN: Erica Brittain, I am a

statistician at the National Institute of Allergy

and Infectious Disease.

DR. BRILEY: Margaret Briley, University

of Texas at Austin, nutritionist.

DR. BOCEK: Good morning. I am

Petr Bocek, medical officer in NIH's National

Institute of Allergy and Infectious Diseases.

MRS. MOORE: I am Marcia Moore. I am with

the FDA as the executive secretary of the Food

Advisory Committee.

DR. WASLIEN: I am Carol Waslien. I am a

nutritional epidemiologist at the University of

Hawaii.

DR. McBRIDE: I am Margaret McBride. I am

a child neurologist at Akron Children's Hospital.

DR. CALLERY: I am Patrick Callery, a

pharmaceutical scientist from West Virginia

University.

DR. GONSALVES: I am Dennis Gonsalves, a

scientist with USDA in Hawaii.

DR. HEIMBURGER: I am Doug Heimburger, a

physician and nutrition specialist at the

University of Alabama at Birmingham.

DR. BARACH: Jeff Barach with Food

Products Association, vice president for special

projects and regulatory affairs.

DR. NELSON: Mark Nelson with the Grocery

Manufacturers Association responsible for

regulatory and scientific policy.

MS. HALLORAN: Jean Halloran from the

Consumers Union where I am director of food policy

initiatives.

CHAIRMAN DURST: Thank you very much.

One other item is that we may have some of

our members leave early on Friday, depending on the

amount of time we can spend. What I propose is

that today and tomorrow that we anticipate having

to go perhaps till 6 o'clock so that we can be sure

that we have enough time for all of our

discussions.

Okay. Let me introduce our first speaker.

This will be Jenny Slaughter, director of Ethics

and Integrity Staff at the FDA, to describe the

"Conflict of Interest Statement" and other

instructions.

CONFLICT OF INTEREST STATEMENT

AND OTHER INSTRUCTIONS

MS. SLAUGHTER: Well, good morning and

welcome. The Food and Drug Administration is

convening today's meeting of the Food Advisory

Committee under the authority of the Federal

Advisory Committee Act of 1972.

With the exception of the industry

representatives, all members of the Committee are

special government employees or regular Federal

employees from other agencies subject to Federal

conflict of interest laws and regulations.

FDA has determined that members of this

Advisory Committee are in compliance with Federal

ethics and conflict of interest laws including, but

not limited to, 18 U.S.C. 208 and 21 U.S.C. 355 and

354.

Under 18 U.S.C., Section 208, applicable

to all government agencies, and 21 U.S.C. 355,

applicable to only FDA, Congress has authorized FDA

to grant waivers to special government employees

who have financial conflicts when it is determined

that the Agency's need for particular

interventional services outweighs the potential

conflict of interest.

Members who are special government

employees at today's meeting including special

government employees appointed as temporary voting

members, have been screened for potential financial

conflicts of interest of their own as well as those

of their spouse, minor child, and employer, which

are related to the discussions of today's and

tomorrow's and Friday's meeting regarding the "FDA

Draft Report: Approaches to Establish Thresholds

for Major Food Allergens and for Gluten in Foods."

These interests may include investments,

consulting, expert witness testimony, contracts,

grants, research and development agreements, public

speaking, writing, patents, royalties, and primary

employment.

In accordance with 18 U.S.C. 208(b)(3),

full waivers have been granted to the following

participants, Dr. Suzanne Teuber and Dr. Soheila

Maleki, please note that all of the interests in

the firms that could potentially be affected by the

Committee's decisions.

A copy of the written waiver statements

may be obtained by submitting a written request to

the Agency's Freedom of Information Office, Room

12A-30 of the Parklawn Building.

In addition, the following individuals are

participating as FDA's invited guest speakers,

July 13th: Dr. Rene Crevel, Dr. Susan Hefle,

Anne Mun [MLM1] oz-Furlong, Dr. Steve Taylor, and

Dr. Robert Wood.

The following individuals will be

participating as FDA invited guest speakers

tomorrow, July 14th: Dr. Pekka Collin,

Dr. Alessio Fasano, Dr. Donald Kasarda,

Dr. Cynthia Kupper, and Dr. Joseph Murray.

Lastly, I would like to report that

Dr. Jeffrey Barach and Dr. Mark Nelson are serving

as the industry representatives on the Committee at

today's meeting. They are acting on behalf of all

regulated industry.

Dr. Jeffrey Barach is employed by the

National Food Processors Association and

Dr. Mark Nelson is employed by the Grocery

Manufacturers of America.

A copy of this document will be placed on

the back table, if anybody wishes to take a look at

it. I thank you.

CHAIRMAN DURST: Thank you very much.

We will now go on to the welcome and opening

statement by Dr. Michael Landa, the deputy director

for Regulatory Affairs at CFSAN, the FDA.

Mike.

WELCOME AND OPENING STATEMENT

MR. LANDA: Thank you, Dr. Durst. You

will be pleased to learn that I don't have a

doctorate or an M.D. I'm just a plain, old J.D.

(General laughter.)

MR. LANDA: Thanks again. Good morning to

everyone. Welcome to the members of the committee,

to the guest speakers, to members of the public who

have joined us today, and to my fellow FDA

employees.

I would like to give a special thanks to

the Committee members for your willingness to take

time from busy schedules to help us with your

expertise for a meeting that will be several days

long. We are all here today, tomorrow and a fair

chunk of Friday.

Let me just add that Dr. Brackett had

hoped to be here this morning, but he wasn't able

to make it. I am hopeful that he will be here for

some portion of the meeting. He was called

downtown for a meeting this morning.

I am going to refer to a couple of points

on the food allergens, but the points I'm making

apply to celiac disease as well. It is just less

cumbersome to start with the food allergens. The

agenda has been making, I think, an opening

statement, of course I'm really not going to do

that.

There are just a few points I want to make

as you go into your inquiry today. The first is

virtually every FDA speaker makes at this kind of

proceeding which is what we do really is based on

science.

We talk about being a science-based

agency. It is the bedrock; it is the foundation.

In that context, I am going to paraphrase what may

be a rather obscure 19th century Senator, Karl

Shrews from Pennsylvania.

The paraphrase essentially is, Our science

correct or incorrect, when it is correct, help us

keep it correct; when it is incorrect, help us to

correct it. That is as much as anything else what

we want from you here in terms of your expertise in

the science.

If with respect to the threshold in the

Draft Report, we have gotten it right, we want to

know from you that we have gotten it right. We

want your help in keeping it right. If we have

gotten it wrong, we want your help in getting it

right. That includes, as you will hear, if we have

not considered an approach that we should have

considered, we want to know that from you.

The third point I will make is that

Americans suffer from food allergies, particularly

children. There is some evidence that the number

is increasing. If you add to that family members,

you really have tens of millions of folks who are

involved. At the moment their principle means of

protection really is exquisite attention to the

food label. That is their pathway to safety I

suppose.

We are hoping that eventually thresholds

will provide another path to safety. This is the

beginning of the inquiry into thresholds, that is,

the approaches that are outlined in the report. It

is the first step in a very important process.

The last point I will make is just that

this is as much as anything else for members of the

public, the docket is going to remain open until

about the middle of August.

If people have comments, based on what

they have heard today, for example, they should

feel free to submit those comments to the docket.

Again, it is until about, I don't remember the

precise date, but it is the middle of August.

In that connection, I should say we are

especially interested, as I think is always the

case, in data. In this case, data of the type

outlined in the report.

Thank you.

CHAIRMAN DURST: Thank you, Mike. Since

Mr. Landa didn't want me conferring a doctorate

degree on him, I will not do it with Catherine

Copp, who is the policy advisor at CFSAN, also the

FDA, who will discuss the use of food allergens

thresholds.

USE OF FOOD ALLERGENS THRESHOLDS

MS. COPP: I was hoping. Oh, well.

(General laughter.)

MS. COPP: Thank you, Dr. Durst.

Good morning. As you know, the focus of

this meeting today and tomorrow and the discussion

on Friday is the Draft Report of CFSAN's Threshold

Working Group: Approaches to Establish Thresholds

for Major Food Allergens and For Gluten in Food.

I have been asked to provide a context for

the Draft Report in terms of CFSAN's programmatic

efforts. This is one thing that if I were a real

doctor I could do. Lawyer's don't do this.

(Slide.)

MS. COPP: Last August, Congress enacted

the Food Allergen Labeling and Consumer Protection

Act, which we refer to in-house by the somewhat

awkward acronym "FALCPA."

This new law amends the Federal Food, Drug

and Cosmetic Act, the principle statute

administered by FDA by requiring that the label of

a food product that is or contains an ingredient

that bears or contains a major food allergen

declare the presence of the allergen as specified

in the law. In shorthand, the declaration is to be

in "consumer friendly" terms.

FALCPA defines a "major food allergen" as

one of the eight foods or food groups or a food

ingredient that contains protein derived from one

of these foods. Those are listed on the bottom of

this slide. By "food groups," I mean fish, tree

nuts and crustacean shellfish, which were

identified by Congress in the law.

(Slide.)

MS. COPP: The possible existence of

threshold levels for food allergens is an important

scientific issue, as Mr. Landa has pointed out,

associated with our implementation of FALCPA.

Although the law does not require FDA to

establish thresholds for any food allergen, there

are three possible ways, which are listed on this

slide, that such thresholds could be used to

implement the new law, these are: administering the

petition process provided for in FALCPA,

administering its notification process, and

addressing the issue or the occurrence of

cross-contact.

(Slide.)

MS. COPP: FALCPA provides two processes

by which an ingredient may be exempt from the

FALCPA labeling requirements, a petition process

and a notification process. I'm trying to read my

own slides (laughter). No, okay.

Under the petition process, an ingredient

may be exempt, if the petitioner demonstrates that

the ingredient does not cause an allergenic

response that poses a risk to human health.

Given this language for the petition

exemption standard, we believe it will be very

important for us to both understand food allergen

thresholds and to have a sound scientific framework

for evaluating the existence of such thresholds.

Under the notification process, an

ingredient may be exempt, if the notification

contains scientific evidence that demonstrates that

the ingredient does not contain allergenic protein,

or, if FDA has previously determined under the food

additive approval process that the food ingredient

does not cause an allergenic response that poses a

risk to human health.

(Slide.)

MS. COPP: Given this language for the

notification exemption standard, we also believe

that it will be very important for us to understand

food allergen thresholds and to have a sound

scientific framework for evaluating the existence

of such thresholds.

(Slide.)

Finally, the FALCPA directs FDA to prepare

and submit a report to Congress. This report will

focus principally on the issue of cross-contact of

foods with food allergens and is to describe the

types, current use of, and consumer preferences

with respect to so-called "advisory labeling."

Processed in a facility that also processes tree

nuts is an example of such labeling.

Cross-contact may occur during food

production when residues of an allergenic food are

present in the manufacturing environment and are

unintentionally incorporated into a food. Because

the food is not intended to contain the allergen,

it is not declared as an ingredient on the food's

label. In some cases, however, the potential

presence of the food allergen is declared by a

voluntary advisory statement.

We also believe that understanding food

allergen thresholds and developing a sound

scientific framework for evaluating the existence

of such thresholds may also be useful to us in

evaluating and addressing food allergen

cross-contact and the use of advisory labeling.

Thank you.

CHAIRMAN DURST: Thank you very much.

Does the Committee have any questions or

discussion of this presentation?

(No verbal response.)

CHAIRMAN DURST: If not, I think we will

proceed.

The next speaker is Dr. Robert Wood,

professor at Johns Hopkins University School of

Medicine, who will give us an introduction to food

allergens.

INTRODUCTION TO FOOD ALLERGENS

DR. WOOD: Thank you very much. It is a

pleasure to be here. What I was asked to do is to

provide an overview of food allergens and food

allergy leading into the discussion that is going

to go on over these next couple of days.

(Slide.)

DR. WOOD: The beginning of this, any talk

about food allergy really requires that we have

some common definition that we can all agree on.

This is something that is not as easy as it might

sound and often generates a lot of confusion. The

reality is that a lot of what is called food

allergy is really not food allergy and may fall

under more of a food intolerance category.

When we are talking about food allergy,

there are a couple of key ingredients. One of them

is that there is an immunologic component to the

reaction. The reaction is typically to the protein

component of the food as opposed to a food

intolerance that is more often related to the

carbohydrate component of the food. Importantly to

this meeting, exquisitely small amounts may cause a

reaction and that these reactions can be severe and

even life threatening.

(Slide.)

DR. WOOD: The pathophysiology of the

allergic response is sort of very schematically

diagramed here. What we are thinking about is a

process that begins with exposure and with most

allergy, probably all allergy, you have to have

some prior exposure to develop your sensitivity.

(Slide.)

DR. WOOD: There is a genetic

predisposition that makes some people particularly

more prone to develop allergy in general, whether

it be food allergy or respiratory allergy, than

others. There are some people who no

matter what, how, when and where they are exposed

they will never develop an allergy, and others who

with very trivial exposure may develop a severe

allergy.

If you are in this group who is

genetically predisposed, your immune system then

goes through a process we will refer to as

sensitization. Sensitization is most often

involving the production of IgE antibodies. We

will talk about this in a little bit more detail

about some different food allergy syndromes.

However, it is also important to note that

not every food allergy involves IgE and that there

may be differences in the types of reactions and

the doses of food required to induce a reaction in

those patients that have IgE versus

non-IgE-mediated food allergy.

Once you have become sensitized, then

reexposure to this food will lead to symptoms.

These symptoms may be abrupt, they may occur within

seconds of eating the food, or they may be very

low-grade and chronic. This is another concept

that we will come back and talk to a little bit.

With some patients it will be very easy to

determine a threshold, and in some patients it will

be virtually impossible to determine a threshold

because their symptoms will not appear in a

challenge test. They may take days or weeks of

chronic exposure and then develop very significant

disease based on that chronic exposure.

(Slide.)

DR. WOODS: The prevalence of food allergy

is substantial. The numbers that we would be most

comfortable with would be 5 to 7 percent of young

children; 2 to 3 percent of adolescents and adults;

at least 10 or 11 million Americans affected.

We do believe that the prevalence is

rising. We don't believe that this is specific to

food allergy. There has been a substantial rise in

asthma and other allergic diseases as well as food

allergy.

Now, the reason that these numbers change

between childhood and adolescence and adulthood is

because a large proportion of food allergy is

outgrown over the first five to seven years of

life.

(Slide.)

DR. WOOD: There is a long list of

potential food allergens out there. At least 200

foods have been identified and characterized as

truly food allergens, but there is a relatively

shorter list that are focused upon because they are

responsible for the vast majority of food allergy

that occurs.

The list on the left-hand side

representing what is most common in young children:

milk, egg, peanut, soy, wheat, and tree nuts.

Then, the list shifts a little bit as you get into

older children, adolescents and adults and is

dominated by peanuts, tree nuts, fish, and

shellfish.

The reason that this list changes from

childhood to adulthood is because four of these

most common food allergens in your children --

milk, egg, soy, and wheat -- are typically

outgrown.

Eighty to 90 percent of children will

outgrow those food allergens and not carry them

into adolescence or adulthood, whereas the peanuts,

tree nuts, fish and shellfish are significantly

more difficult to outgrow, less commonly outgrown,

and tend to persist into adulthood and actually

through the patient's entire lifespan.

(Slide.)

DR. WOOD: Now, the signs and symptoms of

food allergy are highly varied. They may be

chronic and low grade as I mentioned, they may be

acute and life threatening. What I want to run

through in the next couple of minutes are just some

examples of allergic reactions that will point out

a number of things about not only the kinds of

reactions, but the exquisitely small amounts of

food that induce these reactions we are going to

show you, and the sort of day-to-day issues that

patients with food allergy are facing.

(Slide.)

DR. WOOD: The first couple of patients I

am going to show you have urticaria or hives. This

is a total body hive reaction that this boy is

experiencing, a patient I have known since he was

an infant.

He is school age at this point. This

reaction occurred when he was in the grade school

cafeteria, was being teased about this food

allergy, another child blew a straw full of milk

across the table into his face, and he had this

really significant reaction.

(Slide.)

DR. WOOD: This baby here was identified

with milk allergy in the first few weeks of life.

There are some children who don't show up with food

allergy until they are two or three or four years

old, while there are others who are really

demonstrating food allergy in the first days of

life.

This was a baby who was so allergic that

he would react very acutely if his mother, who was

breast feeding him, ingested any milk protein. She

was on a very strict avoidance diet after we

identified his milk allergy, but on the occasion of

her birthday ate a piece of cheesecake, breastfed

him an hour and a half later, and he had this acute

hive reaction.

(Slide.)

DR. WOOD: Now, when we are thinking about

urticaria or hives, there are patients that may

have chronic urticaria. Food allergy is rarely a

cause of chronic urticaria.

However, when someone shows up with an

acute episode of hives, the chance that it is food

allergy becomes higher. Again, we are looking a

relatively short list of foods that are most

commonly implicated: peanut, nuts, eggs, milk,

fish, and shellfish.

Importantly, these reactions are usually

very quick in their onset. Ninety percent of them

or thereabouts will have an onset within 30

minutes; at least half of them, within 5 minutes;

and virtually all of them, within 2 hours.

When a patient has this type of reaction,

it is often very easy to identify the culprit food

because of the abrupt association of the ingestion

of that food with the onset of these hives.

Then, in more severe episodes, there may

be swelling or angioedema or associated

gastrointestinal or respiratory symptoms. That is

moving into more of a systemic reaction that we

would refer to as "anaphylaxis."

(Slide.)

DR. WOOD: Now, this is a patient here who

is having an anaphylactic reaction. When you look

at her back here, it looks just like hives. When

you see her front, though, she is having swelling

and breathing difficulty.

(Slide.)

DR. WOOD: This is a patient who was

having a reaction in the midst of a food challenge

-- not in the midst of it, after her first tiny

dose of egg protein, she went into this very

severe, anaphylactic reaction.

(Slide.)

DR. WOOD: This boy here is someone who is

having a dramatic episode of swelling. His

reaction occurred. Most patients, we should say,

who are having severe reactions know about their

food allergy and are making efforts to avoid it.

He was shellfish allergic -- he is

shellfish allergic. He was making efforts to avoid

shellfish, and he had been reaction-free for

several years.

Then, on another birthday occasion, he ate

chicken in a restaurant and the chicken had been

fried in the same oil as shrimp had been fried.

With that cross-contact, this severe reaction.

(Slide.)

DR. WOOD: Anaphylactic reactions are

defined as a systemic allergic reaction,

involvement of multiple organ systems. These have

an abrupt onset typically. They are related to IgE

antibodies.

You can identify these by doing a skin

test or a blood test looking for IgE. The

manifestations are not always severe. There is an

impression that all anaphylaxis is

life-threatening. Some episodes are relatively

mild, but others progress rapidly to

life-threatening or fatal reactions.

We think that there are at least 150

deaths in the United States each year due to fatal

food-induced anaphylaxis. That number is probably

a substantial underestimation, but we would be very

comfortable saying that it is well identified of

100 to 150 deaths per year.

There are different types of reactions:

some are single phase and some have two phases,

where a patient may look better and then two or

three or four hours later have an even more severe

reaction than they had initially, some of those

lead to the worst outcomes.

(Slide.)

DR. WOOD: This is a patient with one of

the more chronic forms of food allergies, the

patient with severe itching due to his eczema. In

Eczema, a food allergy is often underappreciated

because there is not an obvious cause and effect.

This is one where it is more of a

low-grade, chronic reaction. Hence, this is much

harder for a patient or a family member to identify

that, yes, he ate this food and he is more itchy

now, rather it is really more of a low-grade

reaction where you don't see these direct

relationships between ingestion of the food and the

outcome being their eczema or atopic dermatitis.

It is also a condition where food allergy

is underappreciated by physicians and where

patients may be treated with a variety of different

creams and lotions and only later on find out that

it was really a food allergy that was driving the

eczema.

Overall, 40 to 50 percent of patients with

severe atopic dermatitis and 20 or 25 percent with

less severe cases have an underlying food allergy.

The same list of foods: egg allergy being

most common, followed by milk, peanuts, soy, wheat,

and fish. These six foods account for the vast

majority of food sensitivities seen in atopic

dermatitis.

From our standpoint, it makes it

relatively easy to screen patients and find which

of them are allergic by testing for a relatively

short list of foods.

(Slide.)

DR. WOOD: Now, the last category that I

want to mention is something that we will lump

together as gastrointestinal food hypersensitivity.

There are a variety of conditions that fall under

this umbrella.

There are some that are in the immediate

hypersensitivity category. This would be part,

say, of an anaphylactic reaction where someone ate

food, broke out in hives, had vomiting, diarrhea,

abdominal pain, or other gastrointestinal symptoms.

There is another condition called "oral

allergy syndrome" where patients have reactions

that are confined to their mouth or throat or lips,

particularly related to fresh fruits and

vegetables.

There is another group of conditions that

are lumped under a category of eosinophilic

disorders of the GI tract. There is a specific

condition, eosinophilic esophagitis, where only the

esophagus is involved. As most people in the

audience know, the eosinophil is a type of white

blood cell that is most affiliated with allergic

reactions.

If you take someone who is having a bad

hay fever day outside today and look at their nasal

secretions, their nasal secretions will be loaded

with eosinophils. If you take someone that is

having difficult asthma, their bronchial mucosa

will be loaded with eosinophils.

By the same token, if you have allergic

eosinophilic esophagitis, the lining of your

esophagus is loaded with eosinophils. It may be

isolated to the stomach, it may be more diffuse

where we would call it "allergic eosinophilic

gastroenteritis." This is somebody who may have

disease anywhere in their GI tract, and oftentimes

very diffusely.

There are some other conditions,

enterocolitis syndrome and dietary protein

proctitis, that are much more common in very young

babies.

The importance of presenting these

different syndromes here is that some of these

syndromes are IgE mediated and some of them are not

IgE mediated, some of them are very acute and some

of them are very chronic.

It turns out that those syndromes that are

more chronic and low-grade that don't present with

any acute symptoms, don't present with any clear

cause and effect of eating the food and having

increased gastrointestinal symptoms are going to

be, potentially, the most difficult for this

Committee to grasp. That is because these patients

are often reacting to remarkably small exposures.

I will come back at the end to sort of

give a couple of examples of the dilemma that kind

of patient is going to present to us as we really

try to figure out what is safe and what is not

safe.

It also turns out in the same vein that

the non-IgE conditions in general are probably

going to be most difficult to deal with, both

because they often don't have the acute IgE-type

symptoms, and because they are predominantly

mediated by a different part of your immune system

that can recognize even smaller degrees of these

food proteins that identifying thresholds are going

to be much more difficult.

(Slide.)

DR. WOOD: Now, when we are trying to

approach a patient with a food allergy, one of the

real difficulties is making an accurate diagnosis.

The diagnosis, as in most everything we do, begins

with a history, talking about the foods they

suspect are causing problems, whether we think the

symptoms are consistent with food allergy, whether

this is something that may not be food allergy at

all, or whether it may be a food intolerance rather

than an allergy. We are going to be interested in

the timing of the symptoms and the reproducibility

of reactions.

It turns out that when you do a very

careful history, most of the time it is wrong. It

will be correct in the acute reactions, where you

have a patient who comes in and says, "I fed him

scrambled eggs for the first time last week, and he

had hives all over."

"She took her first bite of peanut butter,

and developed hives within 2 minutes."

It is very likely that the history will be

born out when you do further testing. However,

when you look at the bulk of patients with food

allergies, many of them will have these more

chronic conditions like eczema or the

gastrointestinal disorders. When you are looking

at those patients, you will only verify the history

when you do further testing about a third of the

time.

(Slide.)

DR. WOOD: The next set of tests we do

after taking a history would typically either be

skin testing or serologic testing. A RAST test,

"radioallergosorbent test," is the most common

serologic test that is used.

These tests have some value and they also

have some problems. The problems they have is that

there is a relatively high rate of false-positive

tests. They do not have a terribly good positive

predictive accuracy.

They are generally accurate when they are

negative. Although, they will only be active when

they are negative when you are convinced this

patient has an IgE-mediated condition, because both

of these tests rely on the presence of IgE

antibodies to identify the specific food allergy.

An example would be if a patient develops

hives or anaphylaxis, which typically are

IgE-mediated, and they suspect that it is a certain

food. If you get a positive test back, it is very

likely that they have that allergy. If you get a

negative test back, then you need to keep looking.

It was not likely that food that caused that

reaction.

However, if you have a patient with

something like the allergic eosinophilic

gastroenteritis where there may not always be IgE

antibodies, you cannot stop with a negative test

and say, "We've proven you don't have food

allergy." That is something that happens all the

time, but it is often going to lead to a

misdiagnosis and mismanagement of that patient.

The bottom line is that we need to

carefully interpret our tests in the context of the

overall clinical picture, and that we need to rely

on oral challenge tests as the more accurate tests,

so that we will say that they are not completely

definitive. They are more definitive but not

completely definitive.

Again, they are going to be less

definitive in the patients that have more delayed

type reactions or more chronic conditions where

they won't react in that four-hour observation

period of your food challenge.

(Slide.)

DR. WOOD: You are going to hear more

about food challenges this afternoon, but I will

just mention a couple of issues here in terms of

the way that they can be done. They can be broken

down as open challenges where both the patient and

the person administering the challenge knows what

is being given.

A single-blind challenge is where the

patient is blinded but the person administering the

challenge knows the food that is being

administered, whereas a double-blind,

placebo-controlled challenge is regarded as the

most accurate test because it eliminates the bias

that may occur on the part of both the patient, who

may be feeling a great deal of anxiety about this

food challenge, or on the part of the observer, who

may have their own biases about this patient's

allergy and might overinterpret or underinterpret

symptoms.

We would say that these are going to be

the most accurate tests for the diagnosis of food

allergy. We would use them, if the history and lab

results don't provide a clear diagnosis. That is

often the case, again, when we have both a history

that may not be accurate and laboratory tests that

may not be completely accurate.

Then, we also do them very commonly to

determine when an allergy has been outgrown. This

would be a patient who has been known to be

allergic to a food, and we would be monitoring them

with some regularity in determining at some point

that it is worth trying to retry that food.

We would typically do it in a controlled

setting, just because even in some patients you

don't expect to react at all there may be

significant reaction. Consequently, we have to do

these with considerable caution.

(Slide.)

DR. WOOD: I think I pretty much mentioned

this.

(Slide.)

DR. WOOD: Now, they asked me to mention,

briefly, a study that we published last year

looking at the risk of oral food challenges. What

we have presented in this paper were results on

almost 600 challenges, 253 of which were failed

challenges. The patients reacted in the challenge,

so that is where we can look at the risk. The

other 57 percent, the patients had no symptoms, so

it was a risk-free challenge once they might have

gotten over the anxiety of being there.

We collected a lot of information on

demographics, other atopic disease, symptoms during

challenges, treatment needed, doses at which

reactions occurred. Even though there is a lot

said about safety of food challenges, there has

been very little published before this paper on

what really occurs.

Now, I'm going to say this again a couple

of times looking at the data, but I will say it up

front here, that these results are not

representative of the general population of food

allergy.

These patients that are being challenged

in this either had an unclear diagnosis, so it

wasn't a dramatic kind of situation, or they were

thought to have potentially outgrown their allergy

and were being challenged to potentially prove that

their allergy was gone.

We are really looking at very low-risk

population, and it is not representative of the

whole population of food allergy patients that are

out there. Again, I will say this a couple more

times looking at the specific data.

(Slide.)

DR. WOOD: Now, whenever we are doing this

sort of analysis, we try to break things into

categories. One of the tough categories to decide

is how do you rate reactions. You will see in the

literature some different definitions that have

been used.

We chose to create our own for a series of

studies that we were doing, and talked about mild

reactions that were skin and/or oral symptoms only.

Oral symptoms is just at itching or they will often

have an obvious hive-like reaction in their mouth

or pharynx when they are having one of these

localized reactions.

A "moderate reaction" was described as

upper respiratory and or GI symptoms only or any

three systems. When we are talking about systems,

we broke that into: skin, GI, upper respiratory,

lower respiratory and cardiovascular.

Then, severe reactions were those that

were that were potentially life threatening, where

they have lower respiratory and/or cardiovascular

symptoms or any four systems were involved.

(Slide.)

DR. WOOD: When we broke things down into

these different systems which were involved in

which challenges, you will see here that when we

look at this column on the right here, which is the

total in this paper we reported on milk, egg,

peanut, soy and wheat.

The greatest number of failed challenges

was to milk, 90; 56 to egg; 71 to peanut; 21 to

soy; 15 to wheat; for a total of 253. You will see

that skin manifestations were most common, 78

percent.

This is actually similar to what we have

seen and what is in the literature in terms of

reactions that happen out in the real world.

Eighty percent of food reactions, 80 percent of

anaphylactic reactions involve the skin, but about

20 percent do not.

Oral symptoms occurred in about a quarter,

upper respiratory in a quarter, lower respiratory

in about a third, GI in 43 percent. We,

thankfully, had no cardiovascular reactions in this

population.

Now, why would that be the case? It would

be for two reasons. The biggest reason is that

cardiovascular reactions are not that common in

children.

The cardiovascular system of a child is

really sturdy enough to put up with the insult of

an allergic reaction without necessarily becoming

involved. Cardiovascular reactions are much more

common in adults, and this population was entirely

childhood.

The other reason that we might have seen

the absence of cardiovascular reactions would be

that we were dealing with a relatively low-risk

population.

When we break it down into those three

severity classifications -- mild, moderate and

severe -- you will see that the numbers are

relatively similar for each food. When we look at

the total category, they broke pretty close to a

third in mild, a third in moderate, and a third in

severe.

When you look across the specific foods,

the most important point that came out of this is

that you can't say that one type of food allergy in

this kind of setting is more dangerous than

another.

It turned out that the greatest number of

severe reactions occurred with egg challenges.

This was important information we thought to get

out to get out to people doing challenges.

A lot of allergists will say, "I'm going

refer you, Dr. Wood, all of my peanut challenges.

I'm not touching a peanut challenge because they

are really dangerous. However, I will do egg and

milk challenges out in my office any time."

The message there is that really all of

these foods have a potential to have severe

reactions and need to be done in a setting where

you are really equipped to deal with that potential

for a severe reaction.

(Slide.)

DR. WOOD: When we looked at the RAST test

score or the median IgE level for these different

challenge results, we found that there was really

no strong association between their IgE level and

the reaction severity.

Now, this is an example of where this

population is not a good one to look at for this

data. The reason is that we were essentially only

challenging people that had relatively or very low

levels.

We were not challenging people with very

high levels where they were extremely likely to

fail the challenge. There is no reason in most

instances to prove that they are allergic. When

you know with, say, 99 percent certainty that they

are allergic, we would not put that patient through

a challenge.

Consequently, if you went out in the real

world where the RAST test levels range anywhere

from zero to 100, you would typically see

escalating reaction severity with levels that are

higher. We have that data for peanut allergy where

the group of patients that had levels at 100 did

have more severe reactions when they had accidental

exposures.

(Slide.)

DR. WOOD: Then, I think the last thing to

present from this study is whether reaction

severity was correlated or related to the percent

of food ingested in these challenges. It turns

out, if anything, it is inversely correlated. The

more severe reactions, and none of these were

statistically significantly, but if you look at the

general trends, you will see here that the more

severe reactions occurred with milk and eggs.

As you can see, the severe reaction for

milk is 15 percent and 30 percent for eggs. When

you look at the total group here, 50 percent, 45

percent and 30 percent.

(Slide.)

DR. WOOD: What is the reason this

happens? Does this make any sense at all? Do you

have your more severe reactions with smaller

exposures? The reason we think it happens is

because it is just identifying the more reactive

patients.

It is picking out those that even though

our test scores said that they are not so allergic

that they should do this, it is picking out those

that react more abruptly and have more severe

symptoms early in the challenge just because they

were higher risk patients.

Now, we have come up in our studies about

some decision making about when we would do food

challenges. This is purely for clinical purposes.

These are for those reasons of when we are trying

to decide if they are truly allergic or when we

think that the food allergy might have been

outgrown.

What we would say is that we would do food

challenges based on their history of reactions. If

they have reacted recently, we wouldn't feel the

need to do a food challenge.

We would base it on their laboratory

testing, the skin testing and the RAST testing.

Then he would base it on the importance of the food

to the diet. There are some foods that are

obviously much more important to the diet.

A family may never care whether that child

ever eats a pea again the rest of their life. They

may elect to never have a pea challenge done, but

they may be jumping to do a milk or what challenge

at the first opportunity, because milk or wheat

back in the diet would make such a dramatic

difference in their day-to-day life.

Then, we have come up with some

recommendations based on RAST testing of when we

would recommend doing challenges. These cutoffs

for milk, egg and peanut are all where we found a

greater than 50 percent chance of passing the

challenge, if you have levels below that range.

For other foods, it has been harder to determine

cutoffs, and we would challenge at higher levels

for things like wheat and soy.

(Slide.)

DR. WOOD: Just to go through an algorithm

of how we approach diagnosis, then, because it does

impact on the discussions that are going to happen

here, we would first take our history.

Based on the history, we would make some

distinction whether we think this is consistent

with an IgE type reaction or whether we think that

it is consistent with a non-IgE type reaction.

If it is IgE-mediated in all likelihood,

then a skin test or a RAST test will help identify

whether that food that was suspected to cause a

reaction probably did or probably didn't.

If the test is negative, because the

negative predictive accuracy is so high, we would

feel that you could stop worrying about that food

at that time. If the skin test is positive,

because there are false-positive tests that occur,

we need to do something more.

We might do a trial on an elimination

diet; we might do a food challenge in one order or

the other; and based on all of that information, we

would arrive on the specific elimination diet

recommended for that patient.

If it falls into a non-IgE category, the

situation is much more difficult because we can't

rely on a simple screening test to weed out those

patients.

They are going to need some combination of

challenges -- endoscopy, if it is a

gastrointestinal symptom; elimination diets,

rechallenges, maybe a reendoscopy -- so there is a

much more difficult plan on this side of the screen

to sort out those patients.

(Slide.)

DR. WOOD: Now, I'm going to finish here

with a couple of conclusions and present a couple

of dilemmas. The conclusions are that food allergy

is very common. This is a remarkably worthwhile

initiative that is going on here, and that right

now avoidance is the only treatment plan.

We really hope in the next 5 or 10 years

that there are going to be other treatments for

food allergy. It may be enough so that even if

they don't cure the disease, that they will elevate

the threshold to a point that we don't even need to

have these meetings, that small exposures won't

even be relevant. We are not even close to their

yet, so avoidance is the only option.

Strict avoidance is essential to prevent

reactions obviously, but we also think that in many

patients it also helps to promote the outgrowing

process.

Here is where we may have very different

thresholds. We may have a threshold that this

child, say, with milk allergy -- they know for a

fact that they can eat this bread that has whey as

the tenth ingredient and never have a symptom.

They are perfectly fine with it.

What we have found that getting that bread

on a regular basis may keep their immune system

more revved up to maintain the allergy so this

thing that is way below their threshold for

reacting acutely may still drive the immune system

to maintain the allergy and prevent them from

outgrowing the allergy.

The next conclusion is that food

challenges are a useful means to diagnose food

allergy and a useful means to determine threshold

doses. There are going to be some limitations of

challenges, and one of them is that as opposed to

the study that I presented that Dr. Perry did with

me, you have to include in a threshold type study

the most allergic patients.

Doing the kind of patients that we are

studying on the lower end of the spectrum has

nothing to do with thresholds. It is irrelevant

data. You can't go to my study and say, "This

looks like a threshold because we are not including

in those kinds of studies those highly allergic

patients."

The greater dilemma, and this one is

solvable, there are plenty of real allergic

patients out there. They won't necessarily want to

undergo these studies, because it is not a pleasant

thing to have allergic reactions, but that part is

potentially solvable.

The more difficult thing is a

determination of the threshold doses that I

mentioned for the chronic allergic conditions,

especially those that are not IgE mediated probably

isn't possible.

To give a couple of examples, if we take,

say, milk allergy, the most common food allergy of

all, and we are talking about an infant who is on a

formula, there are a bunch of different options we

could have. Some of them can have soy, but some of

them are also allergic to soy.

Some would go on to a formula like

Alimentum or Nutramigen, which is a formula where

the milk protein has hydrolyzed to a small enough

fragment that in 98 or 99 percent of kids with milk

allergy. It completely solves the problem. They

don't react at all to that level or that type of

protein that remains in that formula.

That other 2 percent, though, may react

severely to that. They are typically the patients

with the gastrointestinal disease. They are

typically very sick; they are typically not

growing; they are typically malnourished.

They are a group of patients who aren't at

risk for the acute dangerous reactions, but they

may be at very high risk for chronic disease from

their food allergy.

Those patients will typically respond

dramatically to a formula that is based in a single

amino acids as a protein source, and that is a

formula like Neocate and Elecare.

Now, when you take that population, and

this is what I deal with every day, there is going

to be a group of them -- and that is probably even

less than 1 or 2 percent, it is probably only 1 out

of 500 -- who still react to the Neocate. They can

react severely to it.

We know that because of their

gastrointestinal biopsies, their biopsies that are

taken from their esophagus or stomach or intestinal

tract still show evidence of severe allergy.

What we think those patients are reacting

to would be either the absolutely trivial amounts

of, say, soy protein that is in the soy lecithin,

that is the eighteenth ingredient in Neocate, or

the trivial, trivial amounts of protein that may be

left in the safflower oil that is used as a fat

component of Neocate.

When we switched those patients off of

Neocate we can prove, and we have 15 patients now

who we have proven, that taking them off Neocate

resolved their food allergy. In this supposedly

non-allergenic formula, they were still reacting.

Now, whether the direction this Committee

needs to focus on is this very unusual patient or

not is sort of a separate debate all together, but

it is safe to say that there are going to be

patients out there who break all rules. No matter

what rules are established, there will be patients

who completely break them and make all of our lives

difficult from that standpoint.

I would be delighted to take any questions

from the Committee or otherwise. Thank you for

your attention.

CHAIRMAN DURST: Thank you, Dr. Wood.

Are there questions for discussion?

Suzanne.

QUESTION-AND-ANSWER SESSION

DR. TEUBER: This is Suzanne Teuber. I

had a question about your patients with the Neocate

sensitivity in terms of what the company reported

for the soy lecithin, did they have any values that

you could report back as to a chronic ingestion

threshold?

DR. WOOD: No. I mean, most of these kids

it is most likely the soy lecithin. SHS doesn't

have that data on the protein content of their soy

lecithin. They say it is zero. These kids when

they were switched to Neocate One Plus, which has

no soy lecithin, their disease went away. We have

to assume that there was enough there to drive that

process.

CHAIRMAN DURST: Yes.

MS. HALLORAN: Jean Halloran. Could you

say something about the process about growing

allergies? How does that work? What actually

happens?

DR. WOOD: Well, that is a very good

question. There are a number of things that we

don't understand too well. However, what we think

is that in the majority of patients we think that

outgrowing is most related to the immune system

gradually forgetting about that concern that it

earlier had.

That is where we think that strict

avoidance is likely to promote the outgrowing

process, and with a prolonged period of strict

avoidance for many of these foods, the immune

system has a memory that isn't long enough to

maintain the allergy and that it will gradually

wane and then full tolerance will be accomplished.

There are probably lots of other mechanisms going

on immunologically that are not well understood.

The other question with this that we have

no great explanations for, lots of theories but no

great explanations, is why you can take a food

allergy like milk, which in early infancy can be

every bit as severe as a peanut allergy, and have

most kids outgrow that allergy, while very few kids

outgrow the peanut allergies. There is something

very different about the immunologic memory of one

food allergen versus another.

CHAIRMAN DURST: Yes.

DR. KELLY: Ciaran Kelly. I wanted to

come back to the issue of challenging individuals

with severe allergies as a method for determining a

threshold. I would like to hear your comments as

regards the feasibility and safety and whether that

would be ethical to perform? I guess my concern is

that once the threshold is crossed, whatever that

threshold might be, isn't there a potential for

severe allergic reaction?

DR. KELLY: Yes. Absolutely. There have

been threshold studies done for the biggie, peanut,

with very allergic people so it is doable. Now,

what we can say about this is that these studies

won't be done in children. It is not going to

happen.

That automatically limits your population

of people, because when you go out and try to find

your group of milk-allergic adults to do these

studies on, you are limited.

Now, they do tend to be more severe

reactors. From that standpoint, you have some

patients out there, but there is no IRB that is

going to let us do this in children. There has to

be demonstrated benefit to do a study with risk.

The safety element is one that we are

comfortable with, recognizing that you need to have

emergency management available to you because there

will be people that have bad reactions.

The safety that is built into that is

starting with exquisitely small doses and working

up very gradually and aborting the challenge

whenever you see your first symptom.

That may lead you to end some challenges

prematurely. You may end up with a false

threshold, but you are obligated to stop when you

have objective signs that patient is reacting.

The ethics beyond that to me is that if it

is an adult patient who is willing to consent to

that process, I have no problem with the ethics of

doing it and have no fear that I will ever lose a

patient to a food challenge.

CHAIRMAN DURST: Yes.

DR. BRITTAIN: This is Erica Brittain.

Since you can't study children in that way, do you

know how this threshold might be different in

children, if you've got the threshold for adults?

CHAIRMAN DURST: No, we don't know that.

That data is, to my knowledge, not available in a

large enough sample to have any validity

whatsoever. It is a superb question. The argument

is going to be and will always be these children

are much more reactive than the adults for most of

these foods.

For peanut allergy it is going to be the

simplest, because allergy tends to persist. We

think that people usually hit their peak level of

severity as an adolescent or young adult, so that

would be fairly easy to solve.

However, when you look at the others like

milk and egg and soy and wheat, you are by and

large going to have the highest level of reactivity

in your first couple of years of life.

When we think about those allergies, we

usually think of growing into the allergy for one

or two or three years where they are becoming more

and more allergic, and then they are becoming less

and less allergic over the next one or two or three

or four or five years as they outgrow the allergy.

It is a moving target at all points, but the most

severe reactivity is likely to be early on.

CHAIRMAN DURST: Dr. Wood, I have a

question -- this is Dick Durst -- just points of

clarification. On your slides where you indicated

"wheat," now this is the IgE-mediated type allergy

as opposed to our discussion tomorrow on celiac

disease?

DR. WOOD: Yes, these results are entirely

IgE.

CHAIRMAN DURST: Okay. Do other grains

cause the IgE type reaction as the wheat?

DR. WOOD: Yes, our study there, about 600

challenges, came out of about 3,000 food challenges

that we have done. There were five most common

foods that I had enough data to make some

conclusions that we were comfortable with. All of

the grains cause allergic reactions.

It turns out that wheat and rye are very

cross reactive from an IgE-mediated allergy

standpoint, and that most patients allergic to

wheat are also allergic to rye; it turns out that

about half are allergic to barley; and 10 to 20

percent are allergic to oat. Beyond those grains,

all of the other grains and grain substitutes are

clearly capable of causing allergy in select

patients.

CHAIRMAN DURST: Thank you. One other

question as far as clarification at least for my

mind. One of your slides with the food challenge

decision making had the units in caps "KU/L." I

don't know if you defined that? I was curious.

DR. WOOD: Yes. It stands for "kilo unit"

of IgE in a specific assay that Pharmacia has

developed called an immunoCAP RAST. It all goes

back to this one technology that is thought to be

the most accurate quantitative measure of specific

IgE, and the results are represented in that kilo

unit of IgE, the specific IgE antibody per liter of

serum.

CHAIRMAN DURST: Thank you.

There is another question?

DR. KELLY: I have one other question.

Dr. Wood, you made a very important comment about

the potential for continued subclinical exposure to

allergens perpetuating an allergic response. How

well accepted and how well documented is that, or

is that largely a clinical impression?

DR. WOOD: Very well accepted, very poorly

documented. It is widely accepted. There is very

poor information to support it. There are only a

couple of studies. The problem we have is we tried

to do the study, and we were turned down because it

is so widely accepted that to go to the IRB and

propose to them that we are going to take this

group of kids with milk allergy and keep them on

low-dose milk and take this group and have them

strictly avoid it was turned down.

Now, there is some work being done that

has identified instead of looking at the IgE

against milk globally, it has turned out that if

you have IgE against certain portions of the milk

molecule it may be more predictive of a longer-term

allergy, and if you have it toward others, other

epitopes, it may be more predictive of an allergy

that is easier to lose.

We think that it may be feasible to focus

on that population that has a very good chance of

losing their allergy, even if we make a mistake, to

be able to do this study. It is doable, but the

outcome is about 10 years down.

CHAIRMAN DURST: Marc.

DR. SILVERSTEIN: I have had some

experience --

CHAIRMAN DURST: Identify yourself.

DR. SILVERSTEIN: Marc Silverstein, Baylor

Health Care System in Dallas. I have had some

experience in studying the epidemiology of asthma

and anaphylaxis. In both of those conditions, your

findings are very much dependent upon your

diagnostic criteria.

In clinical medicine, we have diagnostic

criteria. You have described the criteria for food

allergy, which would involve components of:

history, physical exam, laboratory tests, food

challenge, and response to clinical management with

elimination diets.

Are there standardized criteria that you

would see moving the diagnostic criteria that you

would use from clinical practice to investigation

and publication in peer review literature and/or

perhaps the policy in making regulatory decisions?

I am interested in, Is there a set of

standardized criteria that professional

organizations or clinicians would use for

investigation or for recommending policy? I

understand there is some recent work on definitions

and standards for anaphylaxis?

DR. WOOD: The definitions for

IgE-mediated food allergy are pretty clear and it

is pretty well accepted that it is if you have a

history that is consistent, you have a positive

allergy test, and you either fail a challenge test

or pass a challenge with a dose that is generally

accepted to indicate full tolerance. It is fairly

straightforward and well accepted in the peer

review literature.

It is much more difficult on the group of

patients with, say, eosinophilic gastroenteritis

where they don't necessarily have IgE. You require

a histologic diagnosis to identify the condition,

and then figuring out whether they have food

allergy driving the process exclusively, partially

or not at all is a much more difficult process.

It is doable, but you have to eliminate

foods, rebiopsy, reintroduce foods, and rebiopsy.

There are studies that have done that, but it is so

much more difficult to do that there is much less

of an acceptance of an absolute diagnostic

criteria, much, much less.

It is being looked at. This is a form of

allergy that is clearly either happening much more

often or being identified much more often or both,

so that the potential is there, but it is much

further away from a definition that is well agreed

upon.

CHAIRMAN DURST: Yes.

DR. BRITTAIN: This is Erica Brittain. I

have a clarification question on the food

challenge. How is the placebo control implemented?

DR. WOOD: I think you are going to hear a

lot more about food challenges this afternoon, but

the idea, and it is going to vary depending on the

age of the patient and what they can do, but the

idea that it needs to be well disguised and

obviously safe from the perspective of that

patient's allergen --

(Simultaneous discussion.)

DR. BRITTAIN: But --

DR. WOOD: Go ahead.

DR. BRITTAIN: I'm sorry. Is it by a

dose? Is a particular dose placebo, or does a

patient get all placebo?

DR. WOOD: Yes. I'm sorry I

misunderstood. The normal way the challenge is

done is to have a separate challenge for the

placebo and for the actual food being studied. The

usual way it is done is that the patient would come

in and have a day doing a placebo challenge and

come in and have a day doing the food challenge.

Challenges can be done in a matter of a

couple of hours in some situations, but to do

highly allergic people in a placebo-controlled

manner would usually take 8 or 10 hours for each

day.

CHAIRMAN DURST: All right. Seeing no

further hands in the air, I think we will thank

Dr. Wood. We are right on schedule. Thanks again.

Our next speaker will be

Anne Munoz-Furlong, who is director of the

Food Allergy and Anaphylaxis Network, who will

discuss patient perspectives on food allergies.

PATIENT PERSPECTIVES ON FOOD ALLERGIES

MS. MUNOZ-FURLONG: Thank you. I would

like to thank the organizers of the meeting for the

opportunity to be here.

(Slide.)

MS. MUNOZ-FURLONG: What I would like to

do is in that time that I have been allotted is

give you a sense of who this food allergic consumer

is; the food allergen labeling from their

perspective; and then, most importantly, their way

of looking at threshold levels for food allergens.

(Slide.)

MS. MUNOZ-FURLONG: By way of background,

the Food Allergy & Anaphylaxis Network or "FAAN" is

a non-profit organization. We were established in

1991 and have 27,000 members, almost 28,000

members. Eighty percent of these people come to us

from physician referrals, so we know we are talking

about IgE-mediated responses when we are looking at

our membership.

Our mission has four points: to increase

public awareness, provide advocacy and education,

and advance research on behalf of those with food

allergy.

(Slide.)

MS. MUNOZ-FURLONG: Now, as Dr. Wood said,

food allergy is believed to affect about 11 million

Americans or 4 percent of the population; fish and

shellfish allergy, 2.3 percent or 6.5 million;

individuals in peanut and tree nut, 3 million.

Consequently, between these four foods we

are talking about almost 10 million Americans.

These are the four foods, as was presented earlier,

that are lifetime allergies and also are believed

to cause the majority of the severe or fatal

reactions in this country.

The other point I want to make here is

that although we are talking about 11 million

patients, our data shows us over and over again

that most of these patients have families who

follow their restricted diet. The impact is

actually many times greater than the number of

patients.

(Slide.)

MS. MUNOZ-FURLONG: When we look at

shellfish allergy, this is looking at data that we

published about a year ago now. Te prevalence of

shellfish, we found about 2 percent of the

population or 6 million Americans.

The key foods responsible for the majority

of these reactions in rank order are: shrimp, crab,

lobster, and clam. For fish allergy, .4 percent of

the population: salmon, tuna, catfish, and cod

being the primary fish that cause reactions.

However, if you look at these a different

way, these foods, especially shrimp or salmon, are

available on almost every menu that you are going

to look at in a restaurant or food service

establishment. Therefore, the risk for these

individuals is constant.

(Slide.)

MS. MUNOZ-FURLONG: Talking about tree

nuts, and these most of you already know, are not

peanuts; they are different. Most people with a

peanut allergy avoid tree nuts as a precaution but

not because they are allergic to them. About

20 percent of the 20 peanut allergic population is

allergic to tree nuts as well.

When we are talking about tree nuts, it

affects about 1.5 million Americans. Again,

looking at data from our patient registry of 5,000

patients, we find that walnut, cashew, almond and

pecan are the leading cause of tree-nut-allergic

reactions in this country.

(Slide.)

MS. MUNOZ-FURLONG: What does it mean to

have food allergies? It is vigilant label reading.

You have got to read labels not just for food

ingredients but anything coming into the home.

Bath products can have tree nuts, milk or eggs in

them, for example.

Pet food, if you have ever looked at the

ingredient statement on a pet food, it can have

almost every single one of the major eight

allergens.

That is something you have to worry about,

especially if you have a toddler who will pick up

food from the floor or anyplace else they can get

it. Also, medications have been known to have

allergens in them, particularly milk.

It is not just a question of label reading

for food; it is for anything. Trace amounts can

cause an allergic reaction, and that has been

proven over and over again.

Just one bite can cause a reaction.

Therefore, we can't tell by looking at someone how

allergic they are going to be or what their

tolerance will be to that food.

Currently, as Dr. Woods said, the only

cure now is a dose of epinephrine, if the patient

has a history of severe reaction. The onus is on

the patient or the family to read the label and

avoid the allergen and then be quickly prepared to

handle an allergic reaction, if they have made a

mistake or accidentally ingested the food to which

they are allergic.

(Slide.)

MS. MUNOZ-FURLONG: Because there is no

cure, decisions about any part of the person's life

are centered around food allergy. This is what

makes food allergy so stressful on the family and

on the patients.

Whereas with other allergies you have

seasonal components and you might have an easy

spring but fall is the bad season or if you are

allergic to cats or dogs you can avoid those, with

a food allergy every decision every single day is

affected by your food allergy.

Food shopping can take two to three to

hours just from reading labels. Cooking, if the

family is bringing the allergen into the home, they

then have to prepare two meals, the

non-allergen-containing meal and then the

allergen-containing meal, and take precautions to

avoid cross-contact.

Decisions about dining out and socializing

are made based on not a food preference, but is the

food safe.

"Can the manager be trusted to give us

accurate information?"

"Can the person we are visiting be trusted

not to slip some of the allergen into the food?"

Then, the decision is made to move forward

based on the answers to those questions.

Even what school or childcare the

individual will be sending their food allergic

child to are going to first be centered on food

safety from a food allergy perspective.

Vacation and travel where you and I might

decide whether we want to go someplace warm or go

skiing in the winter, these families have to think

first about food.

"Can we ship food there?"

"Is there a safe place?"

"Can we rent a room with a kitchenette and

make some of the meals so that we can maintain some

level of safety?"

Even family relationships, there is always

somebody in the family that does not believe the

food allergy is real, and so decisions are made

about whether they can visit that individual or

not.

(Slide.)

MS. MUNOZ-FURLONG: As a result of all of

this, it has a tremendous impact on quality of

life. We published a study several years ago

looking at the impact of food allergy on quality of

life.

What we found is that families who have a

food-allergic child score lower on their perception

of whether their child has good health or not, the

emotional health and family activities than the

general population.

Certainly, they scored lower or worse than

families who are looking at or dealing with other

chronic diseases such as diabetes, juvenile,

rheumatoid arthritis and attention deficit

disorder, for example.

We also looked at some of the other

influences. If the individual has a food allergy

and asthma or atopic dermatitis, that further

lowers their score for the quality of life.

If a family has a child with two or more

food allergies, that group scored much lower in 9

out of 12 scales compared to those who only have

one or two food allergies that they are dealing

with.

When we look at our patient population at

FAAN, we see that it is not uncommon for our

members to report a child with a milk, egg and

peanut allergy simultaneously. You can imagine

eliminating those three foods and how it compares

to the impact on the quality of life for the entire

family.

(Slide.)

MS. MUNOZ-FURLONG: This is how, again

looking at the same data, you can see here in blue

is "General health" perception. Food allergy lower

than the normal for asthma, attention deficit

disorder and some of these other symptom scores.

Now, in talking about label reading, which

is really the cornerstone of managing a food

allergy. Here is what goes on.

(Slide.)

MS. MUNOZ-FURLONG: The person with a food

allergy is told by the physician, as you heard

earlier from Dr. Wood, "You have an allergy, avoid

the food." Zero tolerance. They must live in a

black-and-white world. If you are allergic, you

don't eat that product.

If the allergen is listed on the label or

the label says "Contains allergen," they are not

going to eat that product because they are trying

to avoid a reaction. As a result, they expect

ingredient labels to be consistent and, most of

all, reliable because this is what they are basing

the decision about food on. It will affect their

health and safety.

When they see the same product with

different ingredient statements, it makes them very

confused and frustrated and sometimes very nervous

because they, again, are looking for consistency in

labeling.

What we are already seeing with some of

the companies complying with FALCPA regulations is

that there are products on the market that are

pre-FALCPA and FALCPA compliant with different

ingredient information regarding allergens.

Already we are getting calls from our members.

"Which one of these labels is correct?"

"What if I hadn't picked up that second

label? How would I have known?"

This is what we are heading into as we

start to change these labels.

(Slide.)

MS. MUNOZ-FURLONG: The challenge for

food-allergic individuals is that the patients are

told to strictly avoid the allergen, there is zero

tolerance or be prepared to handle an allergic

reaction. Once a reaction begins, we don't know

how severe that is going to be.

They are not aware that there are

scientific names to foods when they are newly

diagnosed. This is something FAAN spends a lot of

time doing. It will get better as FALCPA is

implemented because labels will have simple

ingredient terms on them.

We have to remember it is not just the

patient or the patient's family reading the label,

but it is the teacher, the scout leader, the

friends and family members. The impact for any

labeling decisions are going to be quite broad.

(Slide.)

MS. MUNOZ-FURLONG: Allergens can appear

in unexpected places. This is just one slide of a

number of examples that we have for "Common Foods

in Unexpected Places." Every one of these examples

have caused an allergic reaction to one of our

members, because they were not expecting to find

the allergen.

Just to give you an example, if you have a

milk allergy, you would not have expected that

barbecue-flavored potato crisps might have milk in

them, and you might not have read that label, or

that canned tuna might have soy in it. Therefore,

it is not as easy as avoid the food, you've got to

be looking for unexpected sources.

(Slide.)

MS. MUN [MLM2] OZ-FURLONG: We can see this

reflected in a study that was published in 2002 by

Joshi, et al. They took some food-allergic

individuals, gave them products that were on the

market, and asked them to read the label for the

food they were trying to avoid.

You can see here that families avoiding

milk, only 7 percent were able to accurately

identify milk on the labels that were presented to

them; for soy, they did a little better at 22

percent; but peanut, only 54 percent got the label

reading correct, and most of this was because of

confusion about allergen labeling information.

(Slide.)

MS. MUNOZ-FURLONG: The problem with

allergen labeling information, there are no

guidelines or standards for use. This is

completely voluntary. As a result, every company

has their own decision tree and algorithm and

wording for what terms they will use and under what

conditions.

This makes it very difficult for us to

educate consumers and the others who are reading

labels on their behalf and telling them what to do

and what these mean.

The proliferation of "may contain"

labeling has really caused us some problems. Just

to give you a sense of what is going on, we had one

volunteer go out in the Northern Virginia area to

one grocery store and look at products from

cookies, crackers, candy and bakery. We were

trying to follow the model of a previous FDA study.

She came back with 28 different versions

of "may contain" statements. From the consumer's

perspective, what does that mean? Can they be

trusted, or should we ignore them?

(Slide.)

MS. MUNOZ-FURLONG: The current

environment because of this, there are some

physicians that advise their patients to ignore

precautionary labeling, because it is everywhere

and there wouldn't be any food for them to eat.

There are others who tell them, "Heed the

warning and avoid those foods."

Then, there are some companies who tell

the consumers, "It is on the package only because

our legal counsel has advised us to put this on

there."

Then, there are others that say, "You have

to trust that wording and not go near the product."

How does a consumer determine which is

which?

We are also seeing advisory statements for

peanut allergy only. The way the consumer

interprets these statements is that they are

shortcuts to label reading.

If they see "contains peanuts" or "may

contain peanut," they may not read the rest of the

ingredient declaration if they are looking for milk

or soy, because they think that the company

understands food allergy and would have listed all

of the allergens on there.

As a result of all of this, consumers are

confused and frustrated. Particularly what is

going on as their food choices are further

minimized is that there is risk taking behavior by

parents of kids with food allergies who decide,

seemingly randomly to us, that some companies can

be trusted and others not, so they will ignore "may

contain" on the companies they trust.

Then, the teenagers, our highest-risk

population for a severe reaction, want to be like

everyone else are reporting that they are ignoring

"may contain" statements, because it is on so many

foods they have eaten the food and not had a

reaction, so they don't really believe that these

are true.

(Slide.)

MS. MUNOZ-FURLONG: This is one of the

labeling studies that we conducted with our FAAN

members during a spring meeting a year or two ago.

We asked a question. They were supposed to answer,

"I would never purchase a product that says it

contains" whatever the "allergen" is. You can see

that almost 100 percent of them would avoid a

contain statement.

However, as you go from very specific to

black-and-white to vague "packaged in a facility

that also produces," say, peanuts or nuts or

whatever the allergen might be, only 74 percent

would avoid purchasing that product.

Consequently, 25 percent of the allergic

consumers are going to purchase products where they

don't really understand the precautionary labeling.

If the company is putting this on here because of

some risk, we've got a miscommunication or a

communication gap going on.

(Slide.)

MS. MUNOZ-FURLONG: Let's talk about

thresholds, then. Again, from the consumer's

perspective, their physicians advise, as you heard

from Dr. Wood, is strict avoidance or a reaction

may occur and you will not outgrow this allergen.

They are very motivated to try to strictly avoid

that food.

When we talk about thresholds to our

members, and these tend to be the most motivated

and well-educated of the food allergy population,

this is what we consistently get back. They

believe that threshold levels may put their

children at risk because their child is so

allergic.

They also wonder whether the threshold

levels, the whole discussion is based on the

industry or the government trying to figure out a

way not to have to clean or label for allergens.

Again, they are wary that this might be a loophole

that is trying to be directed at them.

(Slide.)

MS. MUNOZ-FURLONG: The catch 22 here,

from where we are at FAAN, is that we understand

that if we label for all allergens at all levels it

will further restrict diets. If we further

restrict the diet, we are going to increase

frustration which will yield risk taking.

It is going to undermine the integrity of

the ingredient label. As I showed already with

"may contain," we are already seeing that. They

believe "contains." However, if we put "contains"

on everything and they eat it and don't have a

reaction, we are going to diminish the validity of

that statement.

If we undermine the integrity of the

ingredient label this will potentially lead to more

allergic reactions as they take more risk, which is

going to increase the number of doctor visits;

hospital visits; and, potentially, fatalities.

(Slide.)

MS. MUNOZ-FURLONG: Here is an example of

what can go on and what we see as what we may all

be facing. This is a report that came to us from

one of our members who had a soy-allergic child who

had safely eaten soy lecithin in the past. Most of

our members, although we tell them to read the

ingredient declaration on products every time they

purchase them, become brand dependent and stop

reading the ingredient label. That is exactly what

happened here.

This was a product that the child had

safely eaten in the past. The mother did not read

the label, gave it to the child, he started eating

it. She then started reading the label and saw

that it now says "contains soy." She got very

nervous and screamed that it contained soy and

asked the child to spit the food out.

Immediately, he started having itching,

leading to hives, and a feeling of impending doom.

The mother gave him medication and thought she was

having a full-blown reaction.

The question we have to ask ourselves, Was

this a reaction, or was it a panic attack? She

called the manufacturer and was told that the

"contains soy" is because it contains soy lecithin.

Therefore, the ingredients hadn't really changed

from the product that they had safely eaten before.

From our perspective, we do not want to

see consumers or their families subjected to this

kind of fear. Because what you don't realize is

that once this reaction is taken care of, it takes

a long time for the family to trust again. We do

have reports of children developing eating

disorders and just being very cautious about being

around other people once they have had a reaction.

(Slide.)

MS. MUNOZ-FURLONG: From the consumer's

perspective, if we are looking at developing a

threshold level, and as I said there are pros and

cons to both sides of this issue, the key here is

we have got to do a good job of education. We have

got to educate physicians and registered dieticians

so that they can counsel patients accurately.

As you saw, we have done no training for

"may contain." We have got some doctors that say,

"Just ignore it." We can't afford to do that with

threshold levels.

We also have to educate patients and their

families and assure them that the food is still

safe and that they can trust the information on the

label. We also have to do outreach to the food

industry so that they can answer the queries from

food-allergic consumers in a way that will give

them confidence instead of make them nervous or

suspicious about whether they can trust the

information on the label.

(Slide.)

MS. MUNOZ-FURLONG: In summary,

food-allergic consumers want as many food choices

as safely possible. This is really why we are here

and why we are seeing some of this behavior with

advisory statements.

They want to open the diet. The children

want to be like everyone else, and they want the

least amount of restrictions, but they need to be

safe.

The consumer needs to understand the

information on the ingredient statement. They need

most of all to trust that that information is

reliable and it is going to be consistent from one

product to the other. They also need a minimal

number of precautionary allergen statements and a

guideline so that they understand what these

statements mean and what they should do as a result

when they see these on products.

(Slide.)

MS. MUNOZ-FURLONG: In conclusion, the

current labeling and manufacturing practices

present enormous challenges to food-allergic

consumers. As Dr. Wood said, the number of these

patients is increasing.

To give you an example, we conducted a

prevalence study of peanut and tree nut allergy in

1997, repeated that same study in 2002, and found

that in that five-year period the number of

children with peanut allergy had doubled. We don't

know how it is continuing to trend, but reports are

that it is still increasing.

(Slide.)

MS. MUNOZ-FURLONG: The bottom line is

above all we must protect the integrity of the

ingredient information. Because from the

food-allergic consumer's perspective, they depend

on this information to avoid an allergic reaction

and, most of all, to maintain their health and

safety. We already have data showing that food

allergy impacts the quality of life. We don't want

to further diminish their quality of life.

With that, I will end here and open for

questions.

CHAIRMAN DURST: Thank you.

Does the Committee have any questions?

Yes.

MS. HALLORAN: I mean, obviously a person

can survive without ever having to buy any packaged

food. I am wondering in terms of the kinds of

things you were talking about -- teenager's

preferences, the needs of a busy mother, et cetera

-- are there particular categories of food that are

prepared and packaged that are most sort of

important and essential in our modern life? I

mean, would it be bread or breakfast cereal or--?

MS. MUNOZ-FURLONG: If they ate

vegetables, they would be fine. How many kids want

to eat vegetables?

(General laughter.)

MS. MUNOZ-FURLONG: I think it really goes

back to quality of life. Children want to be like

everyone else, and they will do everything they can

to fit that mold.

I have a daughter that was diagnosed with

milk allergy and egg allergy when she was an

infant. I will tell you that I did everything I

could to make sure that she felt like her friends.

It is not just the patient or the child,

it is also the family wanting to not have their

child isolated or feel stigmatized because of the

allergy.

If everyone else is having breakfast in a

box, that is what these kids want. What we want is

to make sure that those labels are accurate, if the

family makes that decision.

Granted, there are some families that are

very cautious and will only make food from home,

make it from scratch. However, as the child gets

older and is out with friends, that is just not

doable.

MS. HALLORAN: Are there any particular

categories of foods?

MS. MUNOZ-FURLONG: No. As you saw in

that slide, "Common Foods In Unexpected Places," we

are seeing allergens everywhere. We have just got

to make sure that all of the labels are correct and

can be trusted.

CHAIRMAN DURST: Yes.

DR. KELLY: Ciaran Kelly. A question for

you from your perspective and the perspective of

the people you represent, the patients with food

allergies.

I understand that you are frustrated and

find it very difficult to work with the current

system of many different types of wording. Would

it be better for you to have a two-level system,

"does not contain" and "may contain traces of" --

or even three levels, "contains" and "may contain

traces of" and "does not contain"? Would that be

acceptable?

MS. MUNOZ-FURLONG: Well, I will start

from the back end of your question. If you poll

our members or just the general consumers, they all

want "does not contain" labeling.

I would caution to you because of the

reports I've seen. This is very widely used in the

U.K., our colleagues in the U.K. have reported,

recalls to products that say "does not contain

peanuts" when they do contain peanuts undeclared.

From the way the consumer is going to

behave if they see "does not contain," they may not

read that ingredient declaration because that is

the guarantee they have been waiting for.

I am not in favor of "does not contain."

I am in favor of let's have them read the

ingredient declaration and know that they can trust

if it doesn't have peanuts in that ingredient

statement, the product should be safe for them.

When we start to see different allergen

statements, we want to make sure that those can be

trusted. When we are talking about "does not

contain," that is an implied endorsement or

guarantee, which makes me very worried. If the

company makes a mistake and that is on the label in

error, we could have someone pay for it by having a

reaction.

Now, if we have two levels, "contains" and

"may contain," as along as we know what that means

and that all companies are following this

guideline, that makes it much easier. Right now,

you can go poll 12 companies and they each do

different things.

CHAIRMAN DURST: I think we need to move

on.

Thank you.

Our next speaker will be Susan Hefle,

associate professor and co-director of the Food

Allergy Research and Resource Program at the

University of Nebraska, who will be speaking on

"Allergenicity: Analytical Methods."

Dr. Hefle?

ALLERGENICITY: ANALYTICAL METHODS

DR. HEFLE: Thank you, Chairman Durst.

Good morning. I am going to discuss the

basic analytical methods for allergens. The model

used is the ELISA-based model which has lateral

flow. This model has been used for several years

now. We will discuss this more later.

Our second bullet, the most successful

kids do use polyclonal antibodies but occasionally

a kit uses monoclonal antibodies directed against a

single protein. Usually, the antibodies are

directed against a crude extract of an allergenic

food not the specific proteins themselves. It is

not necessary to really measure the allergen.

The industry just cares if any peanut is

there, not if one particular protein from a peanut

is there. "Ara h 1" is a particular peanut

allergen. The industry just wants to know if any

100

peanut or whichever peanut is there.

A lot of times a lot of the successful

kids use a much more kind of crude approach to

detecting peanut rather than specifically horning

on the allergens themselves.

There is a challenge, though, in that

different standards are used in the different kids,

depending on the manufacturer, and also different

antibodies are used in the different kids depending

on the manufacturer. It is not like a standardized

approach across the board, necessarily.

(Slide.)

DR. HEFLE: The detection limits range

from around 0.1 to 2.5 parts per million for the

quantitative methods. There are also quality

methods; however, if we are talking about threshold

levels, we need to talk about quantitation here.

Using a method that has a very low

detection limit has certain challenges. Every kit

has the ability to have a low detection limit. Ten

years ago, when I started developing kits,

Steve Taylor and I sat around and thought about

101

what the detection limit should be based on our

years of experience in dealing with consumer

reactions and things like that.

We set a certain level with the kits we

developed; we picked 2.5. It seems to have gone

very well over the last seven or eight years since

these kits have been on the market. Some of the

other companies have a little bit lower range of

detection limit, and that seems to work okay, too.

However, if you go way too late, I mean,

they can all push these kits really, really, really

low. The problem is, Is there clinical relevance

at that point?

If there is no clinical relevance,

companies may be chasing molecules around their

processing plant. They will have all of this

positive data at a low level, and they won't know

what it means. We like to call this "paralysis by

analysis."

We want the data to be relevant. We want

the data to be useful. If the industry goes back

in and says, "I want to fix this, but what if I get

102

all of these positive results at a low level?"

Detection limits have to be kept in mind. They

should be tied to threshold levels, whatever we

decide the threshold levels should be.

It adversely affects the quality of life

for food-allergic consumers, if you use detection

limits that are really low or push those detection

limits without a good clinical basis. Because of

the industry reaction in the form of increased use

of "may contain" label.

When they did paralysis by analysis and

they get positive results, maybe they throw a lot

of "may contain" labeling on that product that they

are worried about and so they are going to put that

on there. That decreases the number of foods that

allergic individuals can eat.

The current detection limits that are set

that the industry uses right now have worked very

well for seven years in protecting the

food-allergic consumer.

I don't think at this point there is any

need to change them right now. But, again, as

103

science comes in and we know more about threshold

levels, there might be an adjustment here or there.

We just finished an egg threshold study.

Contrary to what Robert Wood said, you can do

threshold studies in kids, because we did this in

30 egg-allergic children. That is the only kind of

people we could find to have egg allergy are kids.

When we crunch those numbers and look at

that data, if the threshold is low enough that we

need to adjust the kids for egg out there, the

manufacturers have all said they would be willing

to do that based on the science.

(Slide.)

DR. HEFLE: Many companies are testing for

allergen residues. What they are primarily testing

is not-finished product, but they are using it to

verify sanitation procedures. They have been using

them for as long as they have been on the market.

Certainly with the new law coming up,

there are a lot more using them than used to use

them. In general in the U.S., companies are

incorporating testing using these test kits. As

104

the test kits get faster and easier to use, it is

easier for them to use them.

Again, the ELISA or lateral flow, which is

kind of like a dipstick method are the preferred

methods. Some do the test in house. They really

like it if they can do that because they can fix

things right away.

However, if you don't have in-house

capabilities, they will send it out to a contractor

lab or if they want third-party verification, they

will do that.

Most companies, as I said, are not testing

finished product. They are testing to validate

sanitation methods or doing environmental swabbing

to try to find where the problem is before they get

to the final product. They want to fix the problem

before they get there and figure out if their

sanitation is accurate before they get to the final

product.

Some testing of finished product on

certain occasions though is done, especially when

you can have the product under full control. They

105

don't usually want to release something that they

have tested and they find out there is a problem

and they have to call it back from the marketplace

later and perhaps put consumers at risk.

There are tests that are based on DNA

detection, and they are called "PCR." We don't

advocate these for allergenic residue detection

because it doesn't prove the presence of the

protein. You need the protein to have the allergic

reaction. It just says that there is DNA from that

particular allergenic food there.

It is not practical at all for in-plant

use. You can't put one of these machines next to a

processing line. It is very expensive and requires

a lot of segregation and things. It is meant more

for a regulatory agency or a big corporate lab who

has this ability. It does not prove the absence or

presence of the protein or the allergen. It is

just an indirect kind of a marker.

There are ATP tests out there. This is a

test that is commonly used in the industry for

sanitation assessment. Some companies would like

106

to use this to detect allergens, specifically the

ATP does not detect protein also. Right now, it is

not knowing that these correlate well with the more

specific protein-based tests.

(Slide.)

DR. HEFLE: Three peanut, like ELISA test

kits, have been performance tested by FDA through

AOAC-RI. Those companies with those tests are

Neogen, R-Biopharm, Tepnel.

Five peanut ELISA kits have been studied

in one JRC interlab trial. This is the European

Union's group in Belgium that does these sorts of

things, and they put these three tests plus two

more through a validation trial. They are

currently doing another validation trial on the two

peanut lateral flow devices. They are not finished

with that yet, but they are only doing one matrix

not several matrices. They are just testing it in

cookies right now.

(Slide.)

DR. HEFLE: FDA works with AOAC and has

said they plan more validation studies with other

107

test kits, and that has been the case for more than

a couple of years now with no apparent progress on

this front, though.

The U.S. food industry and other

regulatory agencies -- for example, the Canadian

regulatory agency, the JRC -- has moved way ahead

of FDA/AOAC at this point. The industry is not

running validation trials themselves, but they run

in-house validation things like that.

However, there are regulatory agencies who

have said, "Well, we're going to move ahead. We

can't wait for AOAC anymore. We have to get these

things done in validated interlab trials." There

are several trials that are planned right now

internationally to, hopefully, get some of these

things "validated" in the next few years.

The U.S. industry has been testing for

about seven years now, since the first peanut tests

came out. They have increased the amount of

testing each year and, I've got to say, have spent

millions of dollars once they've gotten test

results to change equipment, to make modifications,

108

for allergens specifically.

Before about 10 years ago, we didn't have

any tests at all to do this. Since the tests have

become implemented, they have used them to make

changes in how they manufacture food.

Health Canada/CFIA has a Compendium of

Food Allergen Methodologies. They crunch through a

lot more of these kind of in-house validations that

they do so that they can use them for their

purposes. There is a Web site for that. They use

both commercial and their own in-house methods.

(Slide.)

DR. HEFLE: Validation of kits, there are

more JRC trials coming out of the EU more likely.

We know of several that are planned, and other

groups have them planned, too. Other groups are

planning more interlab trials, some with kind of

"modeled" foods.

A lot of these tests are done where you

spike peanut into something else. It is not really

like a manufactured product, so it doesn't really

mimic the manufacturing process.

109

A "model food" is actually where the

allergen is manufactured into the matrix, so that

it more appropriately represents what would happen

in the food industry.

Therefore, those are king of challenging

to make. You can't just make them in your back

yard or in your home kitchen. You need to make it

on an industrial level, so it can be quite an

undertaking.

(Slide.)

DR. HEFLE: Kit companies do much more

extensive validation than ever will be done by any

regulatory agency or academic center. It is

usually that the are in the process of selling

kits, and they don't necessarily share the data

like they should. I have been encouraging all of

them to go ahead and publish all of this great data

they have, and it would be a lot easier for all of

us to evaluate how good their kits really are. So

far, they still want to sell kits and not spend

time writing papers.

However, they do have liability issues.

110

Their kits have to work. They have liability

issues. They have reputation issues if the kits

don't work, so it behooves them to do their own

validations before they put a product on the

market.

(Slide.)

DR. HEFLE: Reference materials are solely

lacking for allergens. It would be really nice if

we had a bunch of reference materials we could do

all of these interlab validations with.

However, we are having a problem finding

the appropriate reference materials. There are not

many available, and they are really needed. NIST

is one source of reference materials.

Unfortunately, the NIST standards that are

available were not made for allergen testing, were

not designed for that and often do not represent

the type of allergenic materials used in the food

industry.

A case in point was the standard that was

used in the AOAC-RI-FDA study. It was peanut

butter made by a major manufacturer. It is fine

111

for things like aphlatoxin determination and other

things. Unfortunately, the varieties are not known

with certainty, because the manufacturer wouldn't

tell FDA about every little peanut that might be in

there. They wouldn't divulge it. I'm referencing

NIST not FDA, I'm sorry.

Different peanut varieties have different

responses in the kits. It is imperative to know

exactly what is one of these standards.

Unfortunately, there aren't a whole lot of other

standards around the world around to do that.

There are other sources of materials that

could be used as reference materials, but we have

to come to a worldwide decision on what is the

appropriate criteria for considering something in

the reference material. Is something the JRC makes

in Europe representative of something we use in the

United States?

There are several of these materials

available, and we could begin to talk about going

through some interlab trials with some of these, if

they met certain criteria.

112

(Slide.)

DR. HEFLE: Processing can have a huge

effect on extraction and kit performance. Most

kits are not validated using these model foods, so

we have to do some more of this stuff;

international call for more use of modeled foods.

The old method of spiking, which is where

you put a peanut extract into some of the matrix

and mix it together and see how it performs. This,

again, does not truly represent what happens in the

food industry.

However, the spiking does provide some

useful information, but the manufacturing of these

model foods gives the best information about how a

kit will work.

Model foods have to be made on a pilot,

plant or industrial size scale. If you make this

in your backyard or your kitchen, then it doesn't

really appropriate what a model food is in the

industry, either.

If you make many cookies in a home-size

oven or a Suzy Homemaker or Easy-Bake Oven, it is

113

not going to be the same thing as what Keebler or

what Pepperidge do on a huge scale.

The results of these are not practical or

useful for the food industry. Let's make some real

model foods. They are involving for assessing how

a kit is going to work with a specific commodity,

how efficient the extraction method is under

industrial conditions.

It is becoming more and more important to

use these types of standards in assessing the kit's

performance for certain commodities and processing.

I think spiking is pass.

I get yelled at in my professional

society, AOAC, because spiking is the way of the

food chemists. However, we have to do some spiking

and look at things, but we have to make these model

foods and do that sort of assessment also.

(Slide.)

DR. HEFLE: The extraction method, is it

sufficient? We've got to think about it. Is it

sufficient? Is the recovery good? Can we trust

the results?

114

Some foods are challenging. There are

tannins and polyphenols in dark chocolate that bind

protein. It is a famous matrix, one of the most

difficult matrices to do with allergen

determination.

High fat levels can hide the allergen in

other types of ingredients. If the product is

hydrolyzed, you cannot analyze hydrolyzed or

fermented ingredients in these test kits. They

were never designed for this. When you start

chopping up the proteins, the ELISA signals go

away. The methods are meant to detect intact

proteins and not peptides.

Processing, if you burn stuff, it is going

to be less detectible; it is less soluble. That is

a factor. Now, most companies don't burn their

food, but sometimes they want to detect burned

foods on band ovens or something they can't readily

clean. These are challenges to kit performance,

too.

(Slide.)

DR. HEFLE: Most kits for most allergens

115

have good reactivity with processed forms of the

allergenic food in my experiences over the last 15

years, and that is just my experience.

The use of polyclonal antibodies and crude

extracts and making antibodies against processed

forms are recipes for successful kits. There are

several on the market today that do very well.

Monoclonals are okay if they use a

heat-resistant epitope in making the monoclonals.

They can accommodate the processing changes that

occur.

Some of the egg residue kits have some

issues in this regard. The industry has been able

to adjust and adapt. Many survey the raw material

instead. Instead of worrying about the processed

egg, they will just do the raw egg and handle it

that way, or use a kit that has antibodies against

raw and processed egg, to get around that

particular issue.

Matrix effects, my lab has used all of the

ELISA-based test kits available on the market in

our own validations and tests. It is kind of my

116

hobby so I like to do this. The matrix effects are

usually not a problem for most of the test kits out

there, for the vast majority.

Kit companies have added extraction

additives to their extraction buffers to assist.

When it was recognized dark chocolate was a

problem, they added some secret extraction

additives to help you pull the protein out of dark

chocolate easier.

Model foods, though, again are going to be

of great use in assessing the true extraction

performance of a kit. Again, I can't stress enough

we need to make more of these.

In cross-reactivity issues, even though

most methods do use polyclonal antibodies, which

those of you who know something about polyclonal

antibodies could say, "Boy, there could be a lot of

cross-reactivity problems with them."

We really don't see this happening. The

kit companies really couldn't sell any kits if

their peanut kit cross reacted with everything

else, too. Therefore, we don't usually see these

117

problems in that they have looked at that before

they have launched it, so we don't see the

cross-reactivity. I am not saying that there isn't

one that is going to crop up sometime.

(Slide.)

DR. HEFLE: Again, we've got a problem

with hydrolyzed proteins, hydrolyzed vegetable

proteins, hydrolyzed soy proteins. You can't

really detect them.

The industry would love to do this, to

chase them through the facility and see if they

have cleaned up afterwards because we know there

can be some residual allergenicity in hydrolyzed

protein preparations.

However, the ELISAs are pretty much

rendered useless when trying to analyze for

hydrolyzed protein. It is not what they are

designed to do. The company has had to make a

decision, "What is most of our market?" It is not

chasing hydrolyzed proteins, but it is chasing the

intact proteins. We have to balance the kits to go

towards that, so you can't use it for this.

118

Unfortunately, a negative result in an

ELISA in this case does not mean that there is no

allergenic residue left. You have to ascertain

residual allergenicity via a different method using

human allergic IgE in something like a Western blot

or a RAST analysis.

Another related area is the analysis of

fermented ingredients: gums, Lactobacillus

cultures, starter cultures. Once they start eating

at the substrate, the proteins are partially

hydrolyzed and the ELISAs won't detect them

anymore. You need to use an IgE-based method to

just ascertain the true allergenicity.

Companies don't tell contract labs the

nature of their samples. They just say, "Here is

Sample X." They are not going to tell them it is

hydrolyzed, so we have some challenges.

I try to communicate with the contract

labs and say, "Be sure you ask the question. Just

don't give them a negative result, because it

couldn't be truly negative maybe from an allergenic

standpoint." I think this is the minority of the

119

samples out there.

(Slide.)

DR. HEFLE: My lab performs testing for

food-allergic consumers, their physicians, their

lawyers when they call for free when they report a

reaction to a food. We work with some members of

the Food Allergy & Anaphylaxis Network when they

have a problem.

If there is an analysis I can do, I will

try to help a food-allergic consumer identify what

happened with that particular food, if they have

managed to keep it in the height of the moment.

In 10 years of doing this, we have only

seen "large" -- now notice I say "large" with a

quotation around it, I don't want to make a lot of

judgments on that right now -- amounts of

undeclared allergenic food causing reactions.

(Slide.)

DR. HEFLE: We cannot currently do

immediate monitoring in the food industry, though.

The technology doesn't exist. It is getting

120

better. These lateral flow devices can sometimes

get down to 5 minutes now. I think in the future

they will be able to make a more immediate

response.

Right now, a lot of them are 30 minutes

long. If you are swabbing things and waiting

around for 30 minutes to see if the result is

positive and then having to go back and clean

again, it is pretty impractical for the food

industry to do.

Sanitation and verification is the most

practical, not the test and release kind of thing.

My dad is a fisherman, so I like to the catch and

release and test and release kind of analogy.

We do not have tests for some of the

allergens, and fish is a notable example. You

cannot test for the hydrolyzed or the fermented

allergen sources using these types of methods.

Some types of cross-contact are not

homogenous or 100 percent cleaning is not possible

due to the nature of the product. Food equipment

was never historically designed for allergen clean.

121

Sometimes these facilities are quite old,

and there is no room. There is no room to bring in

different equipment. They have to try to redesign

as they can, but they can't get completely rid of

hangup areas.

You cannot take enough samples to

practically test, to be a hundred percent sure all

of the time. That is impossible. If I get a

statistician in to tell me how many samples I would

need, the industry would just spend the whole day

testing rather than trying to make food product.

In some of these cases, precautionary

labeling is justified due to the nature of the

product and the process in FARRP's opinion. For

example, dark chocolate and milk chocolate on the

same line is one example where we think

precautionary labeling is justified. That doesn't

mean we think precautionary labeling is justified

in every case.

(Slide.)

DR. HEFLE: This is a study that we

recently completed and published in 2004 of some

122

incidents from milk allergic consumer complaints.

These were the casein levels we found in those

particular products. They range from 5,000 on up

to 44,000 parts per million in things that were

supposed to be free of milk or labeled even

"dairy-free" or "kosher," quite high numbers of

parts per million.

They also asked me to talk a little bit

about highly refined oils. What does HRO mean? In

FARRP's opinion, "highly refined oil" means

neutralized, bleached and deodorized or refined

bleached and deodorized.

The definition of what "refined oil" is,

is kind of debated a lot right now in terms of

FALCPA, opinions based on scientific review of oil

challenges with oils in the literature and what we

feel refined oil should be.

The available quantitative methods, there

are methods used in the literature including ELISA

and other methods that reports the levels of

protein in highly refined oils. None of these,

though, have been validated in interlab trials or

123

other types of validation for protein and oil

determination to date.

Somebody will run something and they will

report it, and they will do a certain number of

samples, but no one has looked at whether that is

an appropriate method across the board for

detecting this.

There is a question as to whether a small

amount of protein in the HRO is completely

extracted in aqueous buffer. "Aqueous buffer" is

something that people often use to do these sorts

of biochemical tests. It means trying to partition

the proteins from the oil into an aqueous buffer.

If they really like oil, they might not

all come over. They might want to stay in the oil.

The question is, Does this capture the true protein

content of the oil or whether some of the more

hydrophobic proteins stay in the oil fraction, and,

therefore, do not get extracted and therefore

determined?

My lab uses an amino acid determination

based on Edman degradation, but we also use aqueous

124

extraction. We try to maximize that aqueous

extraction.

We use heat; we use a large amount of

buffer; and we concentrate the sample. However, I

cannot guarantee that I'm pulling all of the

protein out of that highly refined oil when I

measure that.

We report the results as relative and not

a complete picture of the possible protein count

out of HR oil. I still think you are capturing

most of the protein that is there, but I just can't

sit up here and say we are covering a hundred

percent of it.

(Slide.)

DR. HEFLE: The protein levels of HRO are

reported in the literature, and there are lots of

different reports and levels. The caveats again:

The use of aqueous buffers in the determination,

how good if they use an immuno-chemical-based

method is the epitope recognition of the antibody?

Does it really recognize those soy proteins at that

level of processing?

125

Relating "total" nitrogen, sometimes they

use the total nitrogen amount to what the protein

is. Well, total nitrogen can be free and running

around in the protein and not associated with --

free and running around in the oil and not

associated with the protein. Consequently, it may

be an overestimate actually of the protein amount.

Limitations of certain types of methods

like dye binding. "Dye binding" is a method that

will bind to certain proteins preferentially and

not bind to others as well. When you use a

dye-binding method, is it really representative of

everything that is in there? You can't absolutely

tell.

The protein levels reported in the

literature are usually a few milligrams per

kilogram, which are a few parts per million. You

will see some widely ranging estimates, though,

from different investigators. A lot of times I

question their methods sometimes or their ability

to reproduce that particular result.

I think that is the end of my

126

presentation, and I thank you very much for your

attention.

CHAIRMAN DURST: Thank you.

Committee, do you have any questions or

comments?

QUESTION-AND-ANSWER SESSION

DR. MALEKI: Soheila.

DR. HEFLE: Soheila.

DR. MALEKI: Yes, Soheila Maleki. I was

wondering, just based on your experience and you

have been around a lot of industry, if there is any

kind of correlation or if there are any standards

between what the companies use to label "may

contain" versus "contains"? Do they use the same

2.5 parts per million that the kits provide as a

may contain or a not contain and so forth?

DR. HEFLE: They don't really use the

analytical results to make a definite decision

about that. Usually, the companies make a decision

to put precautionary labeling on through a certain

stringent set of criteria. It is something they

have tried to clean up, and they are still having

127

issues.

They have intermittent contamination.

They would never allow something that consistently

had a significant amount of allergen in it to be

called a "may contain." They would try to clean up

more, if it is not supposed to be there.

They don't set a level like that. They

use the analytical results to help them determine

whether that is justified or not. It has to be

potentially hazardous, intermittent, hard to clean.

Those sorts of things are taken into consideration

much more than just the simple analytical result.

DR. MALEKI: Thank you.

CHAIRMAN DURST: Yes.

DR. NELSON: Mark Nelson. I just wanted

to follow up to that in response to Soheila. In

2001, the food industry, a group of associations

representing their members did put together

guidelines on labeling.

The preference is obviously and clearly

the requirement is to label the ingredient in the

presence of an allergen when it is directly added

128

to the food. In the situation where there is a

potential for cross contact, we did establish some

guidelines before companies should use "may

contain" labeling because of the concerns we have

heard about before.

One of those key guidelines was to make

sure that we could not avoid it even after applying

good manufacturing practices: appropriate cleaning,

appropriate separation, and so on and so forth.

DR. MALEKI: I see. Depending on how much

you detected, it didn't matter, if you detected, it

went to "may contain," if it was on the line or --

well, if it contains it was directly added to the

product? I'm trying to make sure I understand that

correctly.

DR. NELSON: Yes, I think it is more to

Sue's point that we aren't necessarily measuring

the finished food so much. It is not a catch and

release situation.

DR. MALEKI: I see.

DR. NELSON: It is understanding your

system; what ingredients are going into the food;

129

what other products might be made on that line;

validating your cleaning processes between

products; scheduling products, depending on the

ingredients that they contain; the sequence in

which you might make the product and so on. There

are a lot of things that go into it.

CHAIRMAN DURST: Anything else?

Yes.

DR. CALLERY: Pat Callery. It appears

that the allergens themselves are not that well

defined, especially when you can find in actuality

generated new allergens by treating food in a

certain way. I am wondering how you address the

analytical problem of false negatives and false

positives?

DR. HEFLE: For a lot of foods the

allergens are indeed known, and there are very rare

cases where you make new allergens through

processing. That is an extreme case in the

literature, I think.

However, false positives and false

negatives are evaluated at the company level first

130

by testing tens of thousands of food commodities

and looking for potential issues. Also, I kind of

poke around myself and see if there is anything

that I can challenge the kits with.

In my experience, the false positive/false

negative rate for most of these methods is very

low. I can't give you a number. I can't tell you

how good that is, because I haven't done a

systematic study.

However, I think that the use of these

interlab trials with model foods will help us look

at some of those issues a little bit more, but I

don't have a good sense of how much false positive

and negative is out there.

I just know in our experience, and we use

these every day, we don't have a lot of issues.

When the occasional issue crops up, and we call the

manufacturer. We usually work through it pretty

easily.

They do tell the manufacturers to validate

or run their own in-house validations before they

truly test the results. The manufacturers do tell

131

the manufacturers to do that, so, theoretically,

they should hopefully find some of these things.

However, every method has a chance of a false

positive or a false negative.

DR. CALLERY: I'm not sure how you do that

without standard materials.

DR. HEFLE: I'm sorry?

DR. CALLERY: I don't know how you do any

of that without standard materials to validate

them.

DR. HEFLE: Some of the manufacturers will

give you a standard to work with, either the

standards from the kit or a recognized standard or

perhaps one of the NIST standards, which is what we

are all defaulting to because we have nothing else.

DR. CALLERY: I think you mentioned that

one kit, they have some secret materials that they

put into the kit to help extract protein. This

seems inconsistent with being able to validate a

method if you don't even know what the test

material, how it was made and what the scope of the

antibodies are that are made.

132

DR. HEFLE: Well, the extraction additive

is not a reference material. The extraction

additive is just an aid in extraction. Usually,

the companies will tell you what it is. It is

usually non-fat dry milk or soy protein. It is

secret, but it is not that secret.

It is just an additional protein in the

mix that helps pull the proteins out of oily

matrices or hard to extract matrices. The

companies know this, and they share that with

customers. However, these sorts of extraction

additives aren't really the reference materials or

the standards used in the kit.

CHAIRMAN DURST: Sue, will you be around

for discussion this afternoon?

DR. HEFLE: Yes, I will.

CHAIRMAN DURST: I think we will hold

further questions until that time because we are

running a little bit late.

DR. HEFLE: Okay.

CHAIRMAN DURST: I would like to take the

recess now. We will take a 10-minute break and

133

reconvene at 10:45.

Thank you.

(Thereupon, from 10:30 a.m. to 10:40 a.m.,

there was a pause in the proceedings.)

CHAIRMAN DURST: We will start with our

next speaker, who is Dr. Stefano Luccioli, who is a

senior medical advisor to CFSAN, FDA. He is also

assistant professor at Georgetown University. He

will be speaking on "Oral Challenge Studies:

Purpose, Design and Evaluation."

ORAL CHALLENGE STUDIES:

PURPOSE, DESIGN AND EVALUATION

DR. LUCCIOLI: Thank you, Dr. Durst.

Good morning. Today, I really want to not

talk to you as an FDA medical officer, but as an

allergist who has experience in performing and

evaluating oral challenge studies.

(Slide.)

DR. LUCCIOLI: The goals of my talk today

are basically just to give you a basic overview of

oral challenge studies, the purpose, why they are

done, the design and conduct, and also spend a

134

little time on evaluation and interpretation of

data, especially with regard to sensitivity of

subjects as well as clinical response and severity

and maybe present some data gaps that may be of

interest while you deliberate on thresholds

(Slide.)

DR. LUCCIOLI: The purpose of challenge

studies are manifold, but the primary reason is to

diagnose allergy, food allergy. The gold standard,

as we have already heard, is the double-blind,

placebo-controlled, food challenge.

As we have heard, also people outgrow

their allergies. They are done also to evaluate

tolerance where those individuals have outgrown

their allergies. They have also been done to

evaluate specific ingredients that are allergens in

specific populations. For instance, there have

been some studies on highly refined oils in

peanut-allergic populations.

However, in recent years, there has been a

lot of emphasis on using oral challenge studies to

determine minimal eliciting doses. This has

135

important implications potentially to determine

sensitivities of individuals within a population,

but also potentially some therapeutic opportunities

in that, as Dr. Wood had mentioned, there is a

feeling that maybe if we can't cure food allergy,

maybe we can raise people's sensitivity levels so

that they may not react to very low trace amounts

of food.

For reasons that you are all here today,

also for establishing threshold challenges, they

may be able to provide you data on low-effect

levels and no-effect levels.

A problem in this field is that there are

insufficient animal models which are commonly used

to evaluate toxicologic ingredients and also

scattered data about case reports where there is

not a lot of information about exact doses that

cause reactions.

Very few studies are done or have been

done. One study was reported by Dr. Wood on

evaluating reaction severity, and we don't have any

current biomarkers to predict severity. This is an

136

important, I think, factor when we are looking at

evaluating minimal eliciting doses.

(Slide.)

DR. LUCCIOLI: I'm just presenting this

slide, but I'm not really going to go into it, to

just give you an overview that oral challenge

studies are somewhat different to traditional tox

models that are used to determine potential

thresholds or acceptable doses. I will, hopefully,

be able to highlight some of these issues in my

talk and present, as I said, some data gaps.

(Slide.)

DR. LUCCIOLI: When you are designing oral

challenge studies, obviously the selection of

subjects is an important factor. Usually, you have

populations of adults, children or infants just to

keep the statistics in order. Most studies involve

both men and women as well as are from foreign

countries and most high ethnicities.

The selection of subjects is basically

geared to what the purpose of the study is for,

whether you want to diagnose individuals with an

137

equivocal IgE or clinical history; evaluate

evidence of outgrowth of tolerance, as we have

mentioned; and also potentially to evaluate

co-existent allergies, for instance, milk-allergic

individuals who may have soy, especially in the

infant population and also for evaluating specific

ingredients, in this case how to evaluate infant

formula.

Obviously, for specific ingredients, you

may want to pick particular populations for that.

In fact, most infant studies are done to evaluate

infant formulas, and the majority of studies are in

adults.

Another important factor is that there is

a notable exclusion of individuals from these

studies. As Dr. Wood had alluded to, there are

individuals who have a cutoff level of their IgE

where above this level they have a 95 percent or

more risk of already failing the challenge. The

challenge is basically useless. You already have

the information, and you tell those individuals to

avoid the food.

138

However, these individuals may represent a

fairly sensitive population. Now with IRBs as they

currently stand, it is very difficult to get these

individuals tested in studies.

Also, classically individuals who have had

anaphylaxis or very severe reactions which were

fairly convincing for the actual food are excluded

from the studies, because another rule of thumb is

do no harm.

Consequently, you don't really want to

test people who could have potentially severe

reactions when you have already had a high clinical

index that they are allergic.

Of course, there are a lot of people who

self-exclude themselves from studies who may be

part of a sensitive population. I also mentioned

here unstable asthma because in any study you don't

want to test individuals who are unstable to begin

with.

Individuals with asthma tend to have more

severe reactions and are probably the group most

representative of fatal reactions. By not

139

including these individuals, you may be missing not

only sensitive individuals but individuals who are

potentially very severe responders.

(Slide.)

DR. LUCCIOLI: With regards to test

materials, there is a variety of test materials

that can be used. Various preparations, if you

just look at peanut, you can have peanut flour,

ground peanut, peanut butter.

There is evidence that the processing

method of these various preparations may affect the

allergenicity profile of proteins within these

foods.

You may have some individuals who are more

sensitive to peanut flour versus peanut butter.

The importance, too, with choosing the material is

that for logistic purposes you want to have it for

an increased time, if you are going to be doing

challenges over multiple months or time points.

A preferred method for these types of

ingredients are dried ingredients. You get into a

problem where dried milk or spray-dried egg are not

140

very commonly ingested ingredients in the

population. It is more common, I mean, the raw or

cooked egg or milk, liquid milk. Therefore, these

are factors that need to be assessed.

Also, fresh versus processed foods, some

individuals are more likely to react to the fresh

food versus the processed as well as raw versus

cooked. These are issues that need to be

considered when you choose a food for a particular

challenge.

Then, the dose units are different within

these challenges. Some studies report milligram

for food; others milligram for protein of food;

and, very rarely, milligram per kilogram which

would be fairly ideal if we wanted to evaluate

potential differences between adults and smaller

adults, infants.

(Slide.)

DR. LUCCIOLI: Obviously, people who

partake in these studies are people who think they

have an allergy; may have had a fairly significant

reaction; and are, understandably, under a lot of

141

stress and are afraid.

Blinding is an important fact, since there

is unfortunately a high incidence of the "nocebo

effect," which is actually the opposite of placebo,

people reacting to a substance that they think is

going to harm them.

In blinding it is important to mask the

food, because you don't want the subject to know

what they are eating. Factors that are used are

called "vehicles" in one sense, and they are

basically other types of foods that are thick that

can hide the taste and smell and texture and that

are also pleasant tasting, you hope.

However, when you are thinking about doing

a challenge study over a few time limits, obviously

you don't want to give some of these vehicles too

much of this, too many milkshakes -- you have to

make sure that the individual is not milk allergic

-- but also they may cause some GI effects or other

things independent of what the actual food that you

are studying would have.

In some cases, they don't always mask the

142

taste. Therefore, some researchers have preferred

to use capsules, since this basically bypasses the

taste issue.

However, using a capsule is difficult,

especially if you are going in higher doses of

food, it is hard to put a serving of some food into

a capsule. I think people would know when they see

a big capsule that there is more food in that.

Also, an important factor is that you may

delay the absorption of that food putting it in a

capsule, and also you bypass the oral cavity which

may be a primary target organ for the initial

allergic response. You may have not only a delayed

response but potentially a less severe response.

I won't talk about the protocol, I think

that was basically well-mentioned by Dr. Wood, but

also a question about placebos. There are some

studies that use placebos within the challenge.

They use a dose and then the next is a placebo.

You know, it is a very complicated process

where you usually need some other people that blind

those to both the researcher and the subject, but

143

they are used as well. However, I think the

preferred method and the easier method is to do a

separate placebo day.

(Slide.)

DR. LUCCIOLI: Now this is just a

schematic of an example of a dose protocol. I

think the important factor is this is an escalation

study of divided doses. One of the important

things, too, is you don't want to be there all day,

and you don't want the patient, too, to be there

all day.

To be able to determine a dose of food and

get up to the final dose, which is usually a

serving of the food, which is like 10 grams of

solid or 60 grams of wet food is what you want to

achieve.

If there is no response at that dose,

there is a good likelihood that the challenge is

negative. However, in many cases you still want to

have the patient come back and do an open

challenge.

Now, with choosing the starting dose, this

144

varies among studies. In many diagnostic studies,

because of this issue about not wanting to be

there, you choose a dose that is roughly half of

the dose that caused the reaction.

Now, I don't know how a lot of people

figure that out, but that is what has classically

been used as the starting dose. Even within a

study, these doses shifts. This dose usually comes

out to be in the milligram range.

Now, more recent studies that have

actually been targeted to study minimal eliciting

doses, have started in doses in the even microgram

range. However, there are a variety of studies

when you are looking at evaluating studies for

eliciting doses.

Also, in this protocol, it is important to

know the time interval differences. Usually, also

that is tailored to the patient when their symptoms

first occurred. Most allergic reactions occur

within 15 to 30 minutes, so that is usually the

time gap, but some other reactions may be a little

bit more delayed.

145

As Dr. Wood discussed, there are

individuals who have delayed reactions as well.

Unfortunately, it is just not logistical to do a

study and wait for these people's reactions to

occur, because they might not occur that day; they

may occur on a separate day.

In this model that I use, I just use a

twofold dose incrementation, but also this could

vary. Some studies go up to even tenfold, so this

could affect also the starting dose and

interpretation of doses in the dose response.

Now, you go and you do the challenge. If

it is negative, it is negative, or you stop it

after the first objective symptom occurs. Some

studies will also record the subjective symptoms,

but that is not always the case, because the

objective symptom is the symptom that denotes a

positive allergic response.

When you record the dose, you can either

record it as the 4X, which is the discrete dose

recorded or the 7X, which would be the cumulative

dose adding the X, 2X or 4X.

146

Just to put this also into some

perspective in terms of safety assessment, when we

are talking about LOAELs and NOAELs, the 4X would

be the low-effect level for this study. If there

are doses before that, at least for this individual

you can say that this dose did not cause a response

and could be considered a no-effect level.

(Slide.)

DR. LUCCIOLI: Some other issues are don't

do this at home. People can have a very severe

reaction. These studies are done in a clinic or an

office where there is emergency equipment and

personnel. It is not a challenge that is done out

in the open. It is in an experimental setting, so

that can also affect the interpretation or results.

Medications, too, most studies now have

people stop the medicines, but with some earlier

studies this was not a factor. Antihistamines and

other things, if people are on these drugs, may

block the early responses so that can factor in.

Fasting, too, most people fast before the

study, but in some studies this was not necessarily

147

explained. If you have a full meal right before

the challenge, this could affect, potentially

affect, absorption of the allergen and therefore

affect the interpretation of the study.

The clinical history or reactivity, too,

is important. Dr. Wood talked about oral allergy

syndrome, but he did not mention about exercise.

There are some individuals who eat a food and have

no problem. However, if they eat the food and

exercise, they have a problem.

Some studies actually test the individual

and then put them on a treadmill and have them

exercise to see if you can elicit the reaction. I

mean, this is very rare, but that is something that

also can be done in terms of the oral challenge

setting.

(Slide.)

DR. LUCCIOLI: Statistical endpoints, I

think these are fairly straightforward for most

challenge studies. You want to just know what

percentage of individuals will react or not react

to the challenge, or in cases where you are

148

studying reaction severity which ones will have a

mild versus a severe reaction.

If you assume that all of these

individuals in the study are part of the sensitive

population or general population, you can maybe

make some assumptions about that and decide a

percentage that will or will not react to a

specific food concentration.

Also, there an importance in this is also

when you are designing a study, you may want to try

to achieve a certain number of individuals to give

you confidence levels for the incidence of allergic

reactions.

In this example, this is a table that

shows over here (pointing) the number of

individuals that need to be tested to give you a

confidence level that the incidence will be less

than this.

For instance, if we were to design a study

with 66 people, that would give us 99 percent

confidence that 1 in 10 would potentially react, so

90 percent would not react. Also, you could use if

149

66 is more than 59, you could also say, well, 95

percent confidence that 95 percent, 1 in 20, will

not react.

Twenty-nine has been usually seen as a

magic number for infant formulas. If 29 patients

do not react, if the infant with milk allergy does

not react to a cow's milk infant formula, that is a

basis for hypoallergenicity.

(Slide.)

DR. LUCCIOLI: I will spend the rest of my

talk on evaluation and interpretation of challenge

study data. Basically, a general interpretation as

we just talked about the statistics, many of these

studies are done in a very small population of

patients, therefore you cannot make a very general

assumption for the general population.

(Slide.)

DR. LUCCIOLI: Because some of these

studies do test the same food, there is a tendency

to group these studies together to try to get the

power higher and then potentially make some

assumptions.

150

The problem with this is that I think it

is important to note that all of the studies that

are currently available are not standardized. I

think that was a question asked just a little

earlier.

This is not standardized data. They are

not standardized to dose. Starting dose or

blinding or testing could also be a factor and also

interpretation of clinical symptoms, which I will

address a little later.

Another issue here is that all sensitive

populations, are they included. If you have

information only on adults, is that going to

predict what harm it will be to infants.

Again, in terms of statistical power, if

you get individuals who are not reactive, if you

are looking at total numbers to say "This is how

many people did not react to this dose," well, what

about people who didn't react to the challenge at

all? Should they be included in the final analysis

of individuals, or should only the ones who react

to the challenge be part of that analysis?

151

What about foreign study data. For

instance, China has a very low prevalence of peanut

allergy, presumably because peanuts there are

boiled or fried versus in this country they are dry

roasted. If you have all of this data in the

United States about peanut allergy, could that be

transferred to data in China?

(Slide.)

DR. LUCCIOLI: I just should mention, too,

that with standardization it is important to note

that there have been some very nice reviews on

actually proposed protocols, standardized

protocols, for food challenges which have been

published in the last year or so. However, to my

knowledge, there have been no studies that have

used this protocol at least for a major food

allergen for evaluation.

Another general interpretation is that

this is an experimental exposure. It is not real

life. There could be false negatives. Individuals

who have had a negative food challenge go out and

have an open challenge and react. It is not always

152

a definitive assessment of allergy. Also, I think

it is difficult to predict reactions to future

exposures. I will try to talk about that as we

come up.

(Slide.)

DR. LUCCIOLI: Subject sensitivity, this

is I think an important issue to consider when

looking at evaluating food ingredients. The

genetic heterogeneity of individuals, there are

multiple allergens in food.

People can be sensitized specifically to

certain allergens within that food. If you cook

the food in a certain way or process it, you may

affect their allergenicity positively or

negatively. This may be what is apparent when they

do studies and you see this enormous gap in

responders.

You have almost a millionfold gap between

the high responders or I should say the least

sensitive who respond to low doses and the most

sensitive to who respond to high doses.

There is also this potential link with

153

severity, as Dr. Wood study has suggested and some

others, that some studies suggest that the

individuals most sensitive to low doses appear to

have the most severe reactions. Are we talking

about a specific subpopulation of individuals here

who are not only sensitive but severe? Also, there

is a sensitivity issue between foods and between

food products.

Another important aspect is that the

individual sensitivities may vary over time.

Allergies can progress and individuals with food

allergies develop asthma later in life. This

asthma, therefore, makes their reactions a little

bit more severe.

Telling somebody right now that they

reacted at a certain dose and that it is okay to

ingest doses before that may not be relevant a year

or five years from now.

(Slide.)

DR. LUCCIOLI: This just is a hypothetical

dose curve adapted from Jonathan Hourihane, who has

done some nice research in this area, basically

154

just to show you how severity and sensitivity may

factor in. I don't really want to spend time on

that.

(Slide.)

DR. LUCCIOLI: Evaluation of clinical

responses, this is where interpretation of

eliciting doses is important with regards to

subjective versus objective symptoms as well as

reaction severity in the dose response.

This table summarizes some of the

reactions that you can see from an allergic

response. Basically, they are divided into

subjective versus objective. "Subjective" means

that they are reported by the individual or the

subject, and "objective" are responses that are

actually visible or observed by the observer.

These reactions are reported in this

manner. As I said, it is when objective symptoms

occur, that is when the study is felt to represent

a positive reaction and stopped.

(Slide.)

DR. LUCCIOLI: To just show you some of

155

these reactions, not only is there a wide range in

reactions, but there are some fairly milder

reactions, hives. You down here to shock and this

is anaphylaxis. Wheezing and syncope are very

close to systemic reaction and potential

anaphylaxis. Consequently, even within an

objective response, you may have a severe

anaphylactic response.

There are also some subjective reactions

that may be somewhat severe: throat tightness,

dizziness, sense of impending doom. I haven't had

the pleasure, fortunately, to experience a patient

with this, but I hear it is fairly dramatic. They

have this sense of impending doom and go rapidly

into anaphylaxis. It is very, very serious. It

doesn't take much for a subjective reaction to go

to something severe.

Also, there are some reactions that kind

of are in between the line of what is subjective,

what is objective: fussiness behavior, abdominal

pain. In adults, that could be suggestive of a

nocebo effect. However, in infants, infants don't

156

mess around. This is their symptom, so these could

be positive responses for infants.

At the same time, you could have skin

flushing or shortness of breath leading to

increased respiratory rate, which could be an

objective sign. However, many times this could be

due to also a nocebo effect. Whether these are

actual positive reactions is hard to determine.

There is some clinical interpretation differences

that can occur here.

(Slide.)

DR. LUCCIOLI: Subjective versus objective

symptoms -- as I told you, the measurable indicator

of allergic response is the objective symptom. It

has got many different endpoints, and the

interpretation may vary. This could also be true

for the subjective reactions.

Many times, subjective reactions do occur

as part of a nocebo effect. However, there are

some that are potentially indicative of an allergic

reaction.

How should these be factored into the

157

assessment? Many times they are not recorded in

the study, so we don't know if there are earlier

reactions to the objective dose, which may

represent an earlier adverse event level.

(Slide.)

DR. LUCCIOLI: Some other eliciting dose

considerations, the starting dose is important. If

the response occurs at this dose, you cannot

determine the no-effect level. Obviously, there is

no dose below that that doesn't cause an effect,

but is this starting dose the low-effect level?

Could you have given a dose a little lower and they

could have still reacted?

With dose increments, some are twofold and

some are tenfold. Using tenfold, you may miss some

increment in between that there could have been a

reaction, even maybe a fivefold difference.

Also, time intervals between doses, as

Dr. Wood has explained, some doses are delayed.

However, time intervals, if you don't give enough

time, you might not know when a subjective response

has become a subjective response or so forth. This

158

could also affect interpretation of these eliciting

doses.

Of course, discrete versus cumulative

dose, some studies report just a discrete dose;

some the cumulative; some both, which is better.

However, how do these factor into a true exposure

assessment or prediction?

(Slide.)

DR. LUCCIOLI: I just want to just show

this, a few more slides, just to kind of put this

into perspective here, give you a mechanistic view

that allergy is a unique toxicologic response.

When you get food that gets challenged, it

causes a massive release of mediators and

cytokines. This is an amplification system that

the immune system uses to protect itself.

Now, in many cases, this response occurs

locally and may not amount to very much, but in

some cases this amplification can involve other

organs and spread systemically very rapidly.

(Slide.)

DR. LUCCIOLI: What has been observed is

159

that the severity of an allergic response is on a

continuum. You can have subjective responses at

some point, objective anaphylaxis, and potentially

death in worse cases.

A few points to note is that this is not a

fixed response. The early objective system may

rapidly progress to something worse. Also, the

degree of amplification, this is not always

predictable or reproducible, so symptoms may not

always be reproducible on subsequent rechallenge.

(Slide.)

DR. LUCCIOLI: To end, with the reaction

severity, most studies only report the actual

symptom. You don't know where this symptom is in

the continuum of severity many times. Those few

that do report the symptoms, they report them as

mild, moderate, and severe.

You have to interpret the researchers, I

guess, response to this, how they interpret it; so,

there is some interpretation. Also, when you have

severe response, like in Dr. Wood's study, in some

cases a third of individuals reacted and had mild

160

reactions, a third had moderate, and a third had

severe. How do you factor in those severe

responses when you determine uncertainty or other

issues?

Also, potentiating mitigating factors are

important: anxiety/stress, medications, and so

forth. These can either potentiate the reaction or

stop it.

Then, the challenge stops after the first

response. A lot of times we don't have the luxury

of knowing how far or how many more doses would

have caused a more severe response. Having that

information is important when you are wanting to

make some risk assessment decisions. Again, it is

a dose distribution, not a dose response.

(Slide.)

DR. LUCCIOLI: In conclusion, the oral

food challenge does provide data on clinical

sensitivity to minimal eliciting doses and also

reaction severity to the initial dose. However,

challenge data currently available for

interpretation is not standardized among studies.

161

The current data pool may not include

extremely sensitive populations with regards to

severity. Challenges have a proven value as a

diagnostic tool but less value in predicting

reaction severity to future exposures.

Thank you, and I will be glad to answer

some questions.

CHAIRMAN DURST: Thank you very much.

Are there any questions from the

Committee?

We will start here.

QUESTION-AND-ANSWER SESSION

DR. BRITTAIN: Yes. I have a comment on

the sample size table that you showed us. I am not

sure that is incorporating the statistical power

education we need to have more than these

individuals. Are you familiar with what I'm

talking about like the 29 there?

DR. LUCCIOLI: Yes.

DR. BRITTAIN: I'm wondering if you get

zero out of 29, then your confidence interval

excludes --

162

DR. LUCCIOLI: Well, what that 29 is, that

is usually a number that is targeted to challenge a

number of study subjects. If you show that 29

individuals with the specific allergy do not

respond to that ingredient, that gives you 95

percent confidence that 90 percent will not react.

DR. BRITTAIN: I guess what I'm saying is

that means if you observed 29, you get the desired

confidence interval. However, if you were planning

a study and you wanted statistical power to be a

certain amount, you would need to have a bigger

study.

DR. LUCCIOLI: Sure.

DR. BRITTAIN: You couldn't assume that

nobody would react.

DR. LUCCIOLI: Yes. Yes, I mean, you saw

that to be totally assured you would have to test

quite a few people.

DR. BRITTAIN: I do have another question.

You mentioned the placebos again, if someone does

have a reaction with a placebo, how is that

interpreted in terms of if they also have a

163

reaction?

DR. LUCCIOLI: Well, yes, many times you

don't know, so then you unmask the study and then

you find out that they reacted to the placebo.

Now, technically, some studies will rechallenge

that patient again. They will have them come back

just to say, "Well, maybe" -- sometimes people do

react.

The difficulty is when they react to the

active dose, to a real challenge, and to the

placebo. If the placebo is too close to the

active, it may be that by the time you gave the

placebo, they are still having the active reaction.

Basically, if they are rechallenged and

show again, they are excluded from the analysis.

Now, that is what should happen. Unfortunately,

you never get that information a lot of times from

these challenges.

CHAIRMAN DURST: Suzanne.

DR. TEUBER: One of the aspects that we

are all very concerned about is which threshold to

use and when it may cause a subjective reaction.

164

Actually, oral itching is a very important

subjective reaction that you didn't have on your

table up there in this presentation.

DR. LUCCIOLI: Okay.

DR. TEUBER: However, if that is

reproducible with two active challenges and not

seen with two placebos, which I think Dr. Taylor

may address a little bit later, but some of the

studies that Dr. Wensing and Bindslev-Jensen and

Dr. Hefle have been doing, they have been looking

at that. All of these have been followed by

objective reactions at higher doses.

DR. LUCCIOLI: Yes.

DR. TEUBER: I would really like people to

comment on that because this may be a much safer

way to approach obtaining thresholds to get these

extremely sensitive populations, if we can use

reproducible subjective data knowing, too, that

there are those other factors that may affect it.

DR. LUCCIOLI: Sure.

DR. TEUBER: For instance, in these

threshold studies that are being designed

165

specifically for thresholds, people with unstable

asthma would still be excluded. I am curious if

anybody knows anything about how unstable asthma

would affect the threshold for a LOAEL that is

seen? Is it a lawful difference? I mean, is there

any anecdotal experience with how that might

change? We want safety here.

DR. LUCCIOLI: Obviously, a speaker that

is coming after me would have some information on

that, but some information from Jonathan Hourihane

would suggest -- and I think some European studies

actually do test some severe patients. Now, I

don't think that any of these patients are

unstable.

I think that they are all excluding

patients who have unstable asthma, but with asthma

in general they haven't found that these

individuals have a lower minimal eliciting dose

than other individuals. However, when they do get

a reaction, they can have a much more severe

reaction.

The assumption, though, is that because of

166

the fatalities and other things that when their

asthma becomes unstable their sensitivity could

change and become more severe very quickly.

CHAIRMAN DURST: Marc.

DR. SILVERSTEIN: Marc Silverstein. I

have two comments that deal with sort of

clarification of terminology and one comment that I

think deals with a more difficult issue. I thought

this was a wonderfully helpful and concise summary

of a variety of complex factors.

In terms of the two clarifications, in

clinical medicine from the first days of medical

school we are taught the difference between

"symptoms," which are subjective, and "signs,"

which are objective.

Some of us from the clinical side who will

be reading the report will think that subjective

symptom is redundant and objective symptom is an

oxymoron.

To help the wide dissemination of the

report and presentation, I would like to suggest

that we in our thinking we may say "subjective

167

finding such as symptoms of the disease" and

"objective signs" is the sense that I use that.

DR. LUCCIOLI: Sure.

DR. SILVERSTEIN: I think it is helpful

because there will be a variety of readers of the

report who may not appreciate that on the clinical

side there is a clarification about that.

The second clarification had to do with

the incidence versus prevalence in the sample size

table. What we are talking about is the proportion

of subjects being tested to the proportion of

individuals in a population, so that sample size

table or the table we have is the expected number

of sample you would need. You label it "incidence"

but it is really a "prevalence" of a condition in a

population.

DR. LUCCIOLI: Okay.

DR. SILVERSTEIN: I believe that what you

are getting at is the sample size so that the lower

confidence interval is that 10 percent or 1 percent

rather than the sample size necessary to show that

two populations differ in proportion or the sample

168

size to show how tight you are around a rate of

zero, which would be a different population.

DR. BRITTAIN: Can I respond to that?

DR. SILVERSTEIN: Sure.

DR. BRITTAIN: I don't think when you are

designing a study you want to think of it in terms

of statistical power, which would be greater than

the sample sizes.

DR. SILVERSTEIN: The third comment I

have, which is substantive and I think we may need

to address this later in greater detail, has to do

with sources of error. There are two sort of

classes of errors that we made in our inferences.

One types of error an epidemiologist or a

clinical epidemiologist may say is biased, one of

the most common sorts of types of errors we could

make would be making inferences in the presence of

certain biases. The most common of which would be

selection bias.

Of course, the selection of individuals

who are referred to a physician, who are referred

to an allergist, who are selected for an oral

169

challenge, food challenge, study would potentially

lead to erroneous inferences if there were

non-representative selection.

That is something that in reading the

literature and making decisions about inferences

for studies or for policy I think we need to be

aware of up front, so that is an important class of

errors that we need to be alerted to.

The second would be an epidemiologist

would talk about confounding. In your example of a

study subject who has asthma, whether it is stable

or unstable and how that is defined, asthma might

be considered an extraneous factor that affects the

relationship between the allergen and the response

to the test. We could use the framework of

thinking of it as a confounding benefit.

Factors such as bias and factors such as

confounding, I think, are useful as we make

decisions about the report and the evidence for

that. I would like for us to be alert to that as

we think about the presentations.

DR. LUCCIOLI: Thank you for the

170

clarification.

CHAIRMAN DURST: Are there any other

Committee comments?

(No verbal response.)

CHAIRMAN DURST: If not, thank you very

much.

Our next speaker is Dr. Rene Crevel,

senior scientist at Unilever, Safety and

Environmental Assurance Centre in the United

Kingdom. He will be speaking on the "Threshold

Modeling Approach."

THRESHOLD MODELING APPROACH

DR. CREVEL: Well, first of all, thank you

for inviting me to share some thoughts on the work

we have been doing on modeling thresholds.

(Slide.)

DR. CREVEL: You have asked me to talk

about the following, to look at different modeling

approaches including what is named the

"hyperallergen." This is the model, and the

Bindslev-Jensen, et al., allergen model; talk about

the data requirements and underlying assumptions

171

behind them; and then say something about

interpreting the results of applying these models.

(Slide.)

DR. CREVEL: Now, just to take a step back

and think about why we are doing this, we've got a

challenge in allergen risk management as far as

industry certainly is concerned.

We want to protect allergic consumers of

course, but we also are aware that protecting them

by certain measures of risk management such as we

have heard about this morning like precautionary

advisory labeling does actually affect their

quality of life.

We want to minimize the effects on their

quality of life, the adverse effects on their

quality of life. We ultimately also want to

maintain economic operation of food manufacturing,

because if that doesn't happen, then that will

affect the quality of life of a considerable number

of individuals and people throughout society as

well. It is an important point to bear in mind.

(Slide.)

172

DR. CREVEL: How can we meet the

challenge? Well, first of all, of course we could

label where the allergen is present, and that is

fine. You have legislation over here now in the

U.S. for that; we have legislation in Europe; and

many other regions and nations also have

legislation.

(Slide.)

DR. CREVEL: Or, we can ensure that the

residual allergen content of product is low enough

to be harmless. I put in brackets here (to the

vast majority of allergic consumers), because we

have heard here this morning some instances of

people reacting to extremely low amounts.

I think it is questionable, and I think

the Committee must address that particular

question, whether those people can be protected by

whatever we can do in the food industry. We need

to think about what alternative measures may need

to be done, whether they can ever eat the sort of

foods we can produce.

(Slide.)

173

DR. CREVEL: How can we determine what is

harmless to an allergic consumer? Well, we have

several sources of data, which I have listed on

this particular table.

We can look at case reports from the

literature, and those we have heard. We have heard

about people reacting to very low amounts.

Unfortunately, the usefulness in risk assessment,

actually an allergen risk assessment in my view is

actually quite limited.

They do establish the hazard. Yes, they

tell you that a certain amount will affect some

individuals, will provoke a reaction in some

individuals. They don't tell you in how many

individuals that will happen.

We can use control challenge studies. In

fact, those of course provide the bedrock of what

is needed, the information needed in allergen risk

assessment because the population can actually be

quite accurately describe.

You can describe them in terms of the

symptoms they have experienced, the allergological

174

history, or the medical history as appropriate --

all of the demographics that you can think about.

Finally, dose distribution modeling, which

I am going to spend obviously some time on, also is

very useful in allergen risk assessment. But of

course it relies on the experimental clinical data

which is generated in control challenge studies.

It cannot operate in a vacuum. We do not have

enough of those sorts of data at the moment.

(Slide.)

DR. CREVEL: I have been asked to say

something about the hypoallergenicity approach. As

I understand it, it is an unofficial standard for

designating infant formula as hypoallergenic.

The original reference I found goes back

to 1991, although I think the American Academy of

Pediatrics has actually updated or at least issued

the guidance more recently, I think, in 2000 or

something of that sort.

The statistics of this approach are based

on the binomial theorem, quite simply. This shows

that, for instance, if you have a study with 29

175

participants, as we have already heard, and you

observe no reactions, then you can be 95 percent

confident that only 90 percent of the population

from which these people have been taken will not

react.

You can also extend that a bit, so if you

observed one reaction and you added people to the

study, then you would 46 for the same degree of

confidence.

If you wanted 95 percent confidence then

fewer than 99 percent would not react, then you

have got those other numbers which already become

very challenging, pardon the pun, for a challenge

study both in terms of recruiting the people do it,

to participate, and the economic cost of actually

doing it.

(Slide.)

DR. CREVEL: However, those are very

useful data when they are generated, but protecting

90 or 95 percent or even 99 percent of the allergic

population is not sufficient as far as we are

concerned as an objective for the food industry.

176

What we asked ourselves is, How can we

improve this? There are several ways. We could

try to look at the conventional toxicological

approach and apply a safety factor to the lowest

observed adverse effect level or the no observed

adverse effect level as the case may be, if you've

got the no observed adverse effect level.

However, that particular safety factor, I

would say, would be arbitrary because we don't

actually know enough about interindividual

differences within the allergic population to apply

a science-based factor, I think. The level of

protection still is undefined. You do not know how

many people you are protecting by applying that

particular safety factor.

What about modeling of those distribution

of minimum eliciting doses? Well, that can

actually define the level of protection for

individual allergen level. You can actually use a

safety factor there. You can use something which

is like a lower 95 percent confidence interval

instead of using the figure itself.

177

(Slide.)

DR. CREVEL: Does modeling actually work?

We asked ourselves could we fit a curve to the

distribution of minimum eliciting doses that are

generated by challenge studies, and could that

curve be useful to predict the number of reactions

likely to occur as a result of exposure to a

specified amount of inadvertently present allergen

in the food?

What I have to say, of course, is we are

not so much concerned about "declared allergen,"

people who are allergic can avoid that, but what

our concern is about is that which is present by

cross-contact, mainly inadvertently.

(Slide.)

DR. CREVEL: This just gives a very quick

model curve. It is just used to illustrate some of

the points, right, okay. From this particular

curve, okay, we have got the data points

schematically indicated like this. The dose on

this (indicating) particular axis an the proportion

of the study population reacting here. Obviously,

178

it goes up to 100 percent.

Then, you have these particular points,

which I have named "ED

50" here. This would be the

dose expected to provoke a reaction in 50 percent

of the study population or 10 percent. This

particular one is an extrapolation, one way of

extrapolating, which one could use.

(Slide.)

DR. CREVEL: What is the impact of the

choice of model on the predicted minimum eliciting

doses? I should go back a bit actually and say

something else.

We collaborated actually with

Dr. Bindslev-Jensen of Denmark in the initial

development of the model. At that particular point

we used the lognormal distribution. Having the

papers published and so on, after that we decided

to go back and look at a few more parameters and

try to refine this particular approach.

Good clinical data were available for egg,

milk and peanut. We fitted the data using the

following statistical distributions and calculated

179

ED10s, the dose which would be predicted to cause

responses in 10 percent of the population; and

ED1s, in 1 percent of the population for each of

those. We used the following linear extrapolation

from lowest observed adverse effect level to zero

dose, which I showed you that was the red line; the

lognormal model, which was the original one in the

2002 paper; the Weibull model; and the loglogistic

model.

(Slide.)

DR. CREVEL: What I want to do is to

illustrate how these variously fit using the

different distributions. This is using real data

actually from the study by Wensing, et al., in 2002

on roasted peanuts. You have got the data points

here. That is still a normal fit, which is the

original one we used.

You can fit loglogistic pretty well as

well and the Weibull as well. You can even fit a

linear -- you can even correlate these points

linearly as well.

Although I haven't got the parameter fits

180

here, I can tell you that they are all pretty good

for all of those. Basically, the fit which you use

doesn't actually tell you which is the most

appropriate one for the particular distribution.

(Slide.)

DR. CREVEL: This is illustrated in the

differences between ED10s and ED1s between studies

and models. Now, this is just using data on peanut

actually from -- in this particular case, we are

just comparing the number of studies on peanuts

including studies performed by Bock and May in the

1970s and the later study by Wensing.

What is quite clear actually is for ED10s,

which are still within the experimental zone, what

I call the "experimental zone" in one of the

previous slides.

The data actually drives what the

predictions are. I mean, there is not a lot of

difference between the ED10s, even though it is log

scale, I know, even in this particular case.

(Slide.)

DR. CREVEL: When you move away and go

181

outside of the experimental zone to the ED1 and

even further actually, the case is even stronger

beyond that, the actual choice of model starts to

drive the predictive responses. That is an

important point to bear in mind.

This is summarized on this slide for the

ED10s in the experimental zone, that the

differences between studies is greater than between

models. In order to address that, the best way of

doing that is to focus on standardizing protocols

and having consistent patient selection criteria

for the studies which you wish to undertake.

For the ED1 values, the differences

between models are much larger as I showed and

increase of course as we move further away from the

experimental zone.

What this actually illustrates is that you

need to validate the particular approach. You need

to validate whatever model you have chosen and

adjust parameters in accordance with that. What

I'm going to talk about is actually how we might go

about doing that, what sort of data is needed.

182

(Slide.)

DR. CREVEL: Now, there are a certain

number of assumptions underlying the values

generated by the model. We are talking here about

undeclared allergen, and that is quite important in

relation to the one of these particular points.

First of all, we assume that the

participants in a controlled challenge study are a

representative sample of the whole allergic

population. That is a very important assumption,

and one which actually is sometimes shall we say

overlooked.

We have heard a lot about whether people

are included or not included in particular studies.

I would tend to argue actually that the population

used in challenge studies, because of the way they

are selected, is actually shall we say at the more

severe, more sensitive end of the allergic

population, basically because there are people who

actually normally are referred to tertiary care

centers.

There are people whose allergies are

183

actually troubling them. They are not people who

might just get a small rash and just ignore it or

ignore the particular food that caused it. There

are people who actually need to manage their

allergy and they need some serious advice in doing

so.

The second point is actually in terms of

validating the model the allergic people actually

eat the same foods as the non-allergics. In this

particular case, it is quite important because if

they are already avoiding them, the number of

reactions that you will be able to enumerate in

epidemiologic studies will not be the correct one.

The distribution of allergic reactivity

study at the population level, now we've heard

thresholds for individuals. Minimal eliciting

doses for individuals do vary. However, what we

are saying here actually is that overall it will be

studied in these particular challenge studies.

Finally, the responses to a given dose of

allergen are similar in the clinic to those

experienced outside. We are doing some work

184

actually with Jonathan Hourihane, in fact, to try

to quantitate the differences that may exist

because, in fact, we are very aware that particular

assumption probably does not hold entirely.

(Slide.)

DR. CREVEL: Okay. What data do we

require for validation and application of a

modeling approach? We want to arrive at the risk

assessment. We have the hazard characterization.

I would put it to you that actually what we are

doing by the modeling approach is actually

characterizing the hazard. We are establishing how

many people are likely to respond to a particular

amount, and we use all of these. These particular

factors all influence it.

However, we also need to know the number

of allergic consumers. That is quite important in

terms of prioritizing allergen management and so

on. Effectively, what the legislation does is also

to acknowledge that particular fact.

The legislation either here, in Europe or

anywhere else does not protect everybody because of

185

course it only specifies a certain number of major

common allergies rather than all of the 200 or so

foods that may provoke allergic reactions.

We also need to know what the exposure is.

We need to know what residual allergen levels are

in the foods, residual allergen levels that are not

declared. Finally, we also need to know what the

number of reactions is overall in the community.

Taking all of that together, we can

actually validate the model. Using those sorts of

data, we can also apply it properly.

(Slide.)

DR. CREVEL: To summarize, I think the

modeling approach complements clinical studies and

it certainly compliments clinical studies to

establish minimal eliciting doses. Of course, it

relies on the data generated in those studies.

I think the advantage compared to just

using the data as such is it actually permits more

complete use of those data using the whole dose

distribution rather than just one particular point,

say, the lowest observed adverse effect level or

186

the no observed adverse effect level.

It also, I think, makes the whole process

of risk management more transparent, I guess you

would say, allowing a more informed discussion of

risk management objectives by all stakeholders.

That is very important I think.

In order to agree on objectives, I think

people need to know how or need to see the process

by which they are reached. However, and this is a

big proviso, it does require validation before it

can be fully operational.

We are doing work at the moment to see how

we can address that. Some of the data actually I

should say will contribute to this particular

assessment will be generated by some European

projects which are currently running, but of course

it will take a few years to get there.

That was my last slide. Thank you.

CHAIRMAN DURST: All right. It is open

for discussion.

Yes.

DR. BRITTAIN: That was a really

187

interesting talk. There was one aspect of it that

I'm a little --

DR. CREVEL: I'm sorry? I can't hear you.

DR. BRITTAIN: There is one aspect of it

that I'm a little confused by, and that was in one

of your last slides with all the graphics about the

needing to know the number of allergic consumers.

If you are trying to find the dose at

which the risk of a reaction, given you are

allergic, which is what I thought we were trying to

do, why do you need to know the number of allergic

consumers?

DR. CREVEL: Well, you need to know the

prevalence of the condition within the population.

In fact, perhaps the confusion is there isn't,

because I mentioned validation as well as

prediction in this particular context actually.

For validation, you certainly need to know

how many reactions are occurring in order to see

whether the model actually predicts the numbers of

reaction which you are actually observing.

I mean, this is a big data gap at the

188

moment. I mean, I don't think either in the U.S.

and certainly not in Europe do we have data on

actually the number of reactions that do occur.

Certainly, we do not have any information on the

total number of severe reactions or less severe

reactions.

DR. BRITTAIN: You mean the number of

reactions that occur across a population as opposed

to your study?

DR. CREVEL: Yes. No, across a

population, sorry. Sorry, that was in the

population, sorry, yes.

CHAIRMAN DURST: Any further discussion or

questions?

(No verbal response.)

CHAIRMAN DURST: If not, thank you very

much, Dr. Crevel.

Our final speaker for this morning's

session is Dr. Steve Taylor. He is the Maxcy

distinguished professor and director of the Food

Allergy Research and Resource Program at the

University of Nebraska, who will discuss Food

189

Allergen Thresholds.

FOOD ALLERGEN THRESHOLDS

DR. TAYLOR: Well, I would like to thank

the Food and Drug Administration for giving me the

opportunity to make a presentation to this panel.

There are advantages and disadvantages to being the

last speaker of the morning. Much of what you are

going to see on my slides may just be a reemphasis

of some things that have already been said.

I think I got a rather difficult topic,

also by being the last one on the agenda, because

I'm supposed to talk about uncertainty factors,

what are uncertainty factors and how are they

derived and what is the underlying scientific

rationale for such a factor. I only wish I thought

I knew the definitive answers to all of those

questions.

(Slide.)

DR. TAYLOR: I think the National Academy

of Sciences outlined risk assessment approaches a

number of years ago, and I always like to start

with this slide, even though I'm not going to

190

discuss all of these different points, because I

think that the same assessment can be used for food

allergens as is used for pesticide residues and

food additives and other things. This is a very

robust risk assessment approach.

(Slide.)

DR. TAYLOR: I am only going to focus on a

few things on this slide today, and one is

dose/response evaluation. I have been thinking

about this issue for probably 30 years.

This is one of the earliest slides that I

created. At that point in time we didn't know very

much, and I would argue we only know a little bit

more now than we knew when I wrote this slide a

long time ago.

Trace amounts can elicit reactions. I

would argue that the severity of the response is

directly related to the dose. The higher the dose,

the more severe the response.

I would agree that individuals can have

different responses on different days to the same

191

dose. However, I don't think those responses are

as dramatically different, or at least I would say

that is an unproven point regarding some of the

things that have been said this morning.

There are a lot of assumptions that are

made in this field, and I think as a panel you need

to identify all of the assumptions and question

them.

Stefano Luccioli made a good point, that

individuals vary widely in their degree of

sensitivity in these controlled challenge studies a

millionfold. I completely agree with that. That

is kind of amazing in itself.

The big question is, How much is too much?

The food industry has been focusing on trying to

get an answer to this question for a long time for

some of the reasons that Dr. Crevel just pointed

out.

(Slide.)

DR. TAYLOR: I think there is another part

that we haven't heard quite enough about, and Rene

kind of pointed it out in his presentation. It is

192

the exposure assessment piece of the equation.

How frequently are food products

contaminated with potentially hazardous levels of

unlabeled allergens, and how frequently do

food-allergic consumers suffer reactions? We

really don't know that part very well. Only

recently, as Dr. Hefle pointed out, do we have the

methodology necessary to determine with any degree

of confidence how frequently food products might be

contaminated and at what levels.

(Slide.)

DR. TAYLOR: Gil Houben from TNO [The

Netherlands Organization] prepared

this slide, and I always like to steal good slides

from speakers that I invite to be on programs. I

think this kind of pictorially describes the

situation that exists.

We have food products in the marketplace

that contained for one reason or another some level

of undeclared allergen. This may be from

cross-contact, this may be from use of ingredients

derived from commonly allergenic foods that are

processing aids and historically haven't been

193

labeled in most countries.

Then, we have individual thresholds for

clinical response that varied by a millionfold as

Dr. Luccioli pointed out. There is an intersection

here between products that have enough undeclared

allergens that at least the most sensitive

individuals have some probability of reacting to

those.

If I was going to draw this slide myself,

I would lengthen the tail of this curve because we

know from analytical studies that there are

products in the marketplace that are quite

hazardous for these individuals containing

comparatively higher levels of allergens that

provoke severe reactions. Dr. Hefle showed some of

those data today.

(Slide.)

DR. TAYLOR: I wanted to say just a little

bit about the different kinds of clues that we can

have for determining allergen thresholds. Stefano

already pointed this out, too.

Probably the best data we have comes from

194

double-blind, placebo-controlled food challenges or

clinical threshold experiments using double-blind,

placebo-controlled food challenges and

immunotherapy trials that also use challenge data.

(Slide.)

DR. TAYLOR: I actually don't think that

allergen cross-contact episodes turn out to be very

useful in determining thresholds, and I wanted to

emphasize that point, because there is a lot of

anecdotal material in the clinical literature about

these cross-contact episodes.

A lot of them are deficient, because the

analytical methods used to detect the residues in

those studies were probably not as accurate as the

methods that Dr. Hefle described in her

presentation, the methods that we have had for the

last few years. There is often a lot of lacking

information in the investigation of these studies.

(Slide.)

DR. TAYLOR: As I pointed out, this

question of how much is too much has intrigued out

group for a long time in the food allergy research

195

and resource program.

I want to point out that we are funded by

the food industry. We have more than 40-member

companies scattered around the world. We began to

focus on the threshold question in earnest in the

mid-1990s and beyond.

(Slide.)

DR. TAYLOR: We have held a series of

threshold conferences. The first one was held in

1999. I was asked to say a little bit about these,

and it is really hard to summarize it in 15 minutes

or less.

I will point out the fact that the results

of the First Threshold Conference have largely been

published in the peer reviewed, scientific

literature.

The question we asked at the First

Threshold Conference is we invited a number of

clinicians from around the world to come to South

Carolina, because we thought that perhaps they had

information on low-dose challenge trials.

When you hear studies of the kind that

196

Dr. Wood reported this morning, recognize that most

diagnostic challenges start at 400 to 500

milligrams.

No wonder some people have severe

reactions at those dose levels, because those are

quite high in my opinion. We were interested in

clinicians who sometimes, because of the patient's

history, started the challenge at a much lower

level.

(Slide.)

DR. TAYLOR: What did we find out? We

found out that there was considerable data on

low-dose challenges for peanut, egg and milk in

particular and more scattered data for some of the

other foods.

The data were really hard to evaluate

because of the lack of standardized protocols. I

will come back to that in a little bit. The lowest

provoking dose -- we had 306 patients for peanut,

281 patients for egg, and 299 for milk. These

physicians brought this data to this conference.

(Slide.)

197

DR. TAYLOR: The lowest provoking dose for

peanut was about 1 milligram of peanut, which is

.25 milligrams of peanut protein. However, I have

to tell you that Dr. Hefle and I spent an entire

weekend in the conference room trying to figure out

what the doses were in these challenge trials,

because the physicians don't calculate that,

particularly carefully in some cases.

Our personal favorite is the physician

that used a drop of peanut butter as his lowest

dose. We had him send us his dropper bottle and we

tried to figure out how much that actually was.

These data look really finite when you show them

this way, but there is a lot of glorified

guesswork. I just want you to understand that.

(Slide.)

DR. TAYLOR: We determined that minimal

eliciting doses or threshold doses do exist for

commonly allergenic foods, that the threshold doses

are finite, measurable and above zero.

However, it was really difficult to reach

consensus, and we didn't reach consensus. We had

198

about 20 clinicians at this conference, and we did

not reach consensus on what threshold doses should

be.

In fact, for most of them this was their

first introduction to this concept. We had to

teach them what NOAELs and LOAELs were. They make

risk assessments every day but not these kind.

(Slide.)

DR. TAYLOR: We also found that reactions

occur to hidden or undeclared allergens in foods.

No big surprise there. However, severe reactions

to undeclared allergens tended to occur at higher

dose levels.

We also determined that at least in these

populations with these low-challenge doses that

low- or very low-dose exposures, LOAELs, result in

mild reversible symptoms.

(Slide.)

DR. TAYLOR: The Second Threshold

Conference was held in 2002 and was geared to

address the biggest concern we had from the first

one, and that was a lack of a consensus protocol.

199

(Slide.)

DR. TAYLOR: I don't have time to describe

the consensus protocol other than to indicate that

it has been published; it does exist; and there are

ongoing low-dose challenge trials underway around

the world using this protocol or slight variations

of it.

As Dr. Luccioli pointed out, most of those

haven't been published yet because it takes a year

to two years to do these studies to find the number

of subjects to enroll in these studies.

(Slide.)

DR. TAYLOR: We did have the Third

Threshold Conference where we tried to determine

what you do with the data once you collect it.

(Slide.)

DR. TAYLOR: I won't go into that very

much, because much of it relates around the

modeling stuff that Rene Crevel already described.

Because the binomial approaches are just plain

difficult, because it is very difficult to identify

even 29 soybean-allergic individuals in the world

200

to do a challenge trial. Believe me, we've been

there, and we know how hard it is.

It is easier to do peanut trials than

perhaps others. It is hard to do milk and egg

because young children outgrow their allergies, so

you've got to be concerned that the child, the

patient, still has the allergy that you are looking

for.

(Slide.)

DR. TAYLOR: There were a number of

advantages to modeling. I think Rene pointed those

out. I will just make the point that the consensus

of the group was that you could do modeling. Of

course, you've got to figure out which model you

are going to use.

Maybe we haven't validated them yet so we

don't exactly know; however, using this lower

confidence interval as the threshold might be a

reasonable approach to consider.

(Slide.)

DR. TAYLOR: Well, classical risk

assessment involves determining the NOAEL for a

201

food additive or a pesticide residue or something

like that and dividing it by 100.

Classically, tenfold is for extrapolation

from animals to humans, and tenfold is for

intraindividual variation. Consequently, what

uncertainty factor should we use?

(Slide.)

DR. TAYLOR: For allergens, since you have

human subjects that can be used, the ideal thing

would be to determine the no observed adverse

effect level for specific allergenic foods among a

human population that is allergic to that food, and

then apply an uncertainty factor to get your

threshold dose.

(Slide.)

DR. TAYLOR: To do that with any degree of

confidence, you have to challenge a fairly large

number of allergic individuals. You would have to

identify the NOAEL for each patient.

You would probably also have to identify

the LOAEL for each patient to prove the person is

still allergic to the food that is under

202

consideration.

It would be good to determine the

variation between individuals in NOAELs because it

is probably a millionfold. A standardized protocol

would be handy so that you didn't have uncertainty

about the differences in protocols.

(Slide.)

DR. TAYLOR: There is no animal to human

extrapolation needed for food allergy

considerations because we have human data. We have

already selected a sensitive subpopulation of the

human population.

The question arises, Did we include the

most sensitive individual? I think that is an

important consideration for this panel. We have

heard several speakers say, "Well, maybe we have

not."

My argument is that perhaps in terms of

representing the whole allergic population to a

particular food we have actually excluded the other

end of the dose distribution curve, and we actually

have included a number of people from the most

203

sensitive subpopulation.

(Slide.)

DR. TAYLOR: I want to point to this study

again. People have interpreted this study as a

publication involving the dose distribution for the

whole food-allergic group allergic to peanuts,

eggs, and milk. It is not that; it is a study of

the most sensitive individuals in clinical

population.

(Slide.)

DR. TAYLOR: Well, how much data is out

there? Is there enough data to make your

decisions? I think there can be if you can wrestle

with the uncertainty factors and the differences in

protocols from one study to another.

I just went through what I think is the

most relevant literature. Some of these are in

your "FDA Report," which contains a big table at

the back that somebody very laboriously put

together. I think they actually found most of the

relevant studies.

I congratulate them for that, because that

204

is not particularly easy to do. I went through

those studies and added up the number of patients

that are in each one of these studies that were

subjected to double-blind, placebo-controlled food

challenges and for which a published LOAEL exists.

Now, there are lots of differences in

protocols, so there are uncertainty factors with

how to plug this data into one of Rene's curves.

What you can see is there are lots of subjects.

This is for peanut. Note, I put an asterisk by our

2002 paper, and that is because that is not

original data. Some of those patients may also

appear in some of the other studies. We got

concerned about whether to count them twice.

(Slide.)

DR. TAYLOR: This is for egg.

(Slide.)

DR. TAYLOR: This is for milk.

(Slide.)

DR. TAYLOR: We have got a lot of data

points. What are the uncertainty factors? Well,

you've got adults, adolescents, children, infants.

205

Many of the studies have been done on pediatric

populations; fewer studies have been done on

adults. You can do challenge trials on both of

those. A lot of the diagnostic challenge trials

are done on infants, but they are not done in

threshold study types of experiments.

(Slide.)

DR. TAYLOR: You've got the problem with

the nature of the challenge material and the

allergen content of that challenge material. This

is again from our 2002 study from Threshold 1.

You can see the number of different

materials: ground peanut, peanut flour, peanut

butter, egg white, dried egg white, whole egg,

dried whole egg, and raw versus cooked for most all

of those. Then, you've got whole milk, non-fat dry

milk and even infant formula as the milk challenge

materials.

In many of these cases, the physicians

didn't determine the protein content of the

challenge materials, so you've got to make

glorified guesses. There are uncertainties about

206

the challenge materials.

(Slide.)

DR. TAYLOR: I would argue that studies

should be compared using protein content. This

failure to provide that data makes the evaluations

really difficult. If the protein content of the

challenge material was not determined or cannot be

determined using reliable data in the literature,

then the study probably has to be rejected from

consideration by groups like this.

There are well-characterized challenge

materials like non-fat dry milk, dried egg white

and soy flour that I think you can assume what the

protein level is based on standardized industry

data. Thresholds should be established in terms

that can be related to analytical methods like

milligrams of food or milligrams of food protein.

(Slide.)

DR. TAYLOR: There are also issues related

to blinding that Stefano already talked about.

Some clinicians use labial challenges. They put a

drop of the food on the patient's tongue or lip.

207

That is often used for young infants.

I think that is particularly difficult to

interpret what the dose was. However,

diagnostically it is good procedure, but otherwise

it is kind of difficult to figure out what was

going on. Then, there is the choice of dosages

used for the challenges.

Probably the biggest uncertainty is this

issue of the fact that most of the publications

were done for diagnostic purposes, and so when you

look at the published literature you get the LOAEL

and not the NOAEL. I actually think a lot of the

NOAELs are clinically available; they are just not

published.

There is more uncertainty in using a LOAEL

rather than a NOAEL to established threshold doses;

there is patient selection criteria; exclusion of

people on probably both ends of the curve; and

there is variability in individual threshold doses.

Diagnostic challenges tend to report only

the LOAELs; the NOAELs in some cases may not be

recorded. As Dr. Wood pointed out to us in his

208

study, in many cases the patient reacted to the

lowest dose administered.

However, if it was 400 or 500 milligrams,

that is a pretty sizeable dose. I think there is a

lot of uncertainty if you use that as your LOAEL to

try to try and figure out what the threshold dose

might be. You are much better off to focus on

lower dose challenges where there is less

uncertainty even if you have to use the LOAEL.

How far above the NOAEL is the LOAEL; and

if using a LOAEL, how big should the UF be? I

think that is the question that they wanted me to

try to answer.

(Slide.)

DR. TAYLOR: I got bold and I did try to

answer it. If the LOAEL is based on subjective

symptoms, "My mouth itches," then I don't think we

have to be very concerned about uncertainty

factors, because in the limited experiences that

exist in the literature there is a ten- to a

hundredfold variations between the doses that begin

to provoke subjective symptoms and the doses that

209

tend to provoke objective signs. I learned

something today. I'll try to say "signs."

Now then, if you have a LOAEL based on

objective reactions, what uncertainty factors

should you use? Well, then I think you would need

to have very careful expert analysis of the

clinical study you are looking at.

I could argue that if you looked at one of

these clinical threshold trials that have been done

using microgram and low-milligram dose level, that

you could use a very low uncertainty factor.

However, if you are going to rely on

publications like Perry, et al., from 2004 where

you only have data on the reactions that occur at

the lowest diagnostic challenge dose and it is 400

or 500 milligrams, then you've got a much bigger

uncertainty factor because you could be a long ways

above the NOAEL.

I don't know what you would do in this

particular case, so I put a question mark by it. I

actually think those kinds of studies are not very

helpful.