IG UNITED STATES DEPARTMENT OF AGRICULTURE

 

 

 

 

 

 

 

 

 

 

 

 

 

FOOD ADVISORY COMMITTEE

 

CONTAMINANTS AND NATURAL TOXICANTS SUBCOMMITTEE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Wednesday, December 4, 2002

 

8:34 a.m.

 

 

 

 

 

 

The Inn and Conference Center

Founders Ballroom

University of Maryland University College

3501 University Boulevard East

Adelphi, Maryland  20783

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

 

MEMBERS

 

Dr. Francis Busta, Chairman

Dr. Henry Kim, Executive Secretary

 

Dr. Alex Acholonu

Dr. Lawrence Fischer

Dr. Marion Fuller

Dr. George Gray

Dr. Lawrence Kuzminski

Dr. Ken Lee

 

TEMPORARY VOTING MEMBERS

 

Dr. Goulda Downer

Dr. Joseph Hotchkiss

Dr. Thomas Whitaker

C O N T E N T S

 

AGENDA ITEM     PAGE

 

Welcome and Introductions - Dr. Francis Busta,

Chairman, Contaminants and Natural Toxicants

Subcommittee (CNTS)     5

 

Conflict of Interest Statement - Dr. Henry Kim,

Executive Secretary, CNTS     8

 

Opening Remarks - Joseph A. Levitt, Director,

Center for Food Safety and Applied Nutrition

(CFSAN)     10

 

FDA Action Plan

 

 - Goals and Timetable - Dr. Terry C. Troxell,

   CFSAN, Director, Office of Plant and Dairy

   Foods and Beverages (OPDFB)     19

 

 - Charge and Questions to the Subcommittee -

   Dr. Terry C. Troxell     27

 

Scientific Overview of Acrylamide in Foods -

Dr. Bernard A. Schwetz, Senior Advisor for Science,

FDA     31

 

FDA Action Plan

 

 - Major Components of the Action Plan - Dr.

   Richard A. Canady, CFSAN, Division of Risk

   Assessment, OPDFB     53

 

FDA Action Plan

 

 - Toxicology - Dr. Richard A. Canady, CFSAN,

   Division of Risk Assessment OPDFB     83

 

FDA Action Plan

 

 - Analytical Methods/Occurrence - Dr. Steve

   Musser, CFSAN, Division of General Scientific

   Support, Office of Scientific Analysis and

   Support     131

C O N T E N T S (Continued)

 

AGENDA ITEM     PAGE

 

FDA Action Plan

 

 - Formation - Dr. Lauren S. Jackson, CFSAN,

   Division of Food Processing and Packaging,

   OPDFB     184

 

FDA Action Plan

 

 - Consumption/Exposure - Dr. Michael J. DiNovi,

   CFSAN, Division of Biotechnology and GRAS Notice

   Review, Office of Food Additive Safety     222

 

FDA Action Plan

 

 - Consumer Risk, Dr. David W. Acheson, CFSAN,

   Office of Science     236

 

Joint Institute for Food Safety and Applied

Nutrition (JIFSAN) Workshop, October 28-30,

2002 - Dr. David R. Lineback, Director, JIFSAN     258

 

Adjournment     311

P R O C E E D I N G S

     DR. BUSTA:  Can we bring this meeting to order, please?  Good morning, everyone, this morning again.  We're pleased that you are here in full force.  According to the weatherman, I'm glad we didn't start tomorrow because we might not be here in full force.  We have a long agenda, and we are going to do our best to stay with that agenda.

     I would like to remind each and every one of you--and I'll probably have to do this on a regular basis--to speak into the microphone, not only to be heard but also for the recording of your comments.

     Again, welcome to the Subcommittee, and we have a significant challenge that we will soon hear from Terry Troxell.

     I'd like to have each of the group here introduce themselves, and I'll start off with Larry Kuzminski and we'll just go around.

     DR. KUZMINSKI:  Thank you.  Larry Kuzminski from Roxbury, Massachusetts.  I've been retired from the food-processing industry for three, three and a half years.  The postings during that employment tenure were with Ocean Spray and with the Kellogg Company prior to that.

     DR. WHITAKER:  I'm Tom Whitaker.  I'm with the USDA, Agricultural Research Service, in Raleigh, North Carolina.  I'm an agricultural engineer, and my main emphasis of research has to do with detection of mycotoxins in agricultural products.

     DR. FISCHER:  I'm Larry Fischer from Michigan State University.  I direct the Institute of Environmental Toxicology at Michigan State.  I'm an environmental toxicologist.

     DR. FULLER:  I'm Marion Fuller.  I'm with the Florida Department of Agriculture and Consumer Services and director of food safety and also a toxicologist.

     DR. LEE:  I'm Ken Lee.  I chair the Food Science and Technology Department at the Ohio State University.

     DR. BUSTA:  I am Frank Busta.  I'm professor emeritus and head of the Department of Food Science and Nutrition at the University of Minnesota.

     DR. GRAY:  I'm George Gray.  I'm at the Harvard School of Public Health and the Harvard Center for Risk Analysis, and my background is in toxicology.

     DR. ACHOLONU:  My name is Alex Acholonu.  I'm a professor of biology at the Alcom State University in Mississippi, president of the Faculty Senate, and my forte is microbiology and epidemiology.

     DR. HOTCHKISS:  I'm Joe Hotchkiss.  I'm Chair of the Department of Food Science at Cornell University and also a food chemist and toxicologist.

     DR. DOWNER:  I'm Goulda Downer, president and CEO of Metroplex Health and Nutrition Services and adjunct professor at George Washington University.

     DR. BUSTA:  We have the FDA representatives on the back table, and each one of you, we will be seeing you in due order.

     If there is no problem, we'll move on and try and even get a little ahead of the agenda.  Henry Kim will now tell us about conflict of interest.

     DR. KIM:  Thank you, Dr. Busta, and good morning, everyone.  I am Henry Kim, the Executive Secretary for the Contaminants and Natural Toxicants Subcommittee of the Food Advisory Committee.  First, I would like to read the appointment of our temporary voting members for the record.  It reads:  By the authority granted under the Food Advisory Committee Charter of July 2002, I appoint Dr. Goulda Downer, Dr. Joseph Hotchkiss, and Dr. Thomas Whitaker as temporary voting members of the Contaminants and Natural Toxicants Subcommittee of the Food Advisory Committee for the December 4-5, 2002, meeting of acrylamide.  Signed, Joseph A. Levitt, Director, Center for Food Safety and Applied Nutrition.

     We would also like to note for the record that Dr. Lawrence Kuzminski is participating in this meeting as the acting industry representative and a non-voting member of the Subcommittee.

     Second, the following announcement addresses the issue of conflict of interest with respect to this meeting and is made a part of the record to preclude even the appearance of such at this meeting:  The issues to be discussed at this meeting are issues of broad applicability.  Unlike issues in which a particular sponsor's product is discussed, the matters at issue do not have a unique impact on any particular product or manufacturer but, rather, may have widespread implications with respect to foods and their manufacturers.

     To determine if any conflict of interest exists, the committee participants have been screened for interests in the food industry.  As a result of this review, in accordance with 18 U.S.C. Section 208(b)(3), Dr. Francis Busta, Dr. George Gray, and Dr. Joseph Hotchkiss have been granted a particular matter of general applicability waiver that permits them to participate fully in the matters at issue.  Copies of the 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.

     With respect to FDA's invited guest speaker, Dr. David Lineback reported that no conflict of interest exists.

     And now I will turn the meeting back to Dr. Busta.

     DR. BUSTA:  Thank you, Henry.

     We will proceed now with opening remarks from Joe Levitt, the director of CFSAN.

     MR. LEVITT:  Thank you very much, Mr. Chairman.  It's a pleasure for me to be here but, more importantly, it's a pleasure for me to welcome each of you, not just to this meeting but to serving on this Subcommittee, this being the first meeting of the Subcommittee on Contaminants and Natural Toxicants of our overall Food Advisory Committee.

     As you'll come to understand better with each meeting, we have restructured our Food Advisory Committee over the last year so that we have a general, what I think of as the parent umbrella committee, as well as six subcommittees.  The subcommittees tend to be, as permanent members, small and focused so that we can add and supplement as needed given a particular issue being raised, and you've already seen some of that today.

     We will be bringing to you a number of important scientific issues for your review, for your comment and advice, and, as always, we really need your best advice and will do our best to heed that, and we will strive to make this an integral part of our workings at CFSAN.

     So thank you for serving and coming to the D.C. area despite the snow warnings, but at least the transportation from the room to the meeting site is adequate in snow conditions, so that's good.

     Today's subject, of course, is an important one, the subject of acrylamide.  When you think back just a year ago, this was on nobody's agenda at all.  And just since last spring, think of the number of things that have occurred.  There was the announcement and publication of the Swedish findings last spring.  There was very quickly at WHO expert consultation with experts around the world, including very strong FDA participation.  There have been analytical methods developed.  There have been meetings of government scientists.  FDA has drafted an action plan.  We had a public meeting at the end of September, a JIFSAN research-oriented workshop in October, and this meeting that we're having today at the beginning of December.  So this has been an enormous amount of activity.

     Sometimes, when you have a lot of activity, I always ask the question:  Are we seeing just a lot of motion, or is there also movement with that motion?  And I think here we've really seen an enormous amount of movement in a very short period of time when you think of what we know now that we didn't know six or eight months ago.

     I think, first of all, everybody--there's general agreement that this is an issue that must be seriously addressed, and I know one person remarked before the meeting that these kinds of meetings are happening all over the world.  Everybody agrees that this is an issue that we need to address and that it is an important one.

     We have seen good, solid analytical methods developed and utilized that have confirmed the initial findings.  I think when everybody first heard the first official findings, they were perhaps a little skeptical.  But the methods have been developed; the findings have been confirmed.  And so that's an important first step.

     We know more about how acrylamide is formed.  We know that it wasn't put in there to begin with.  We know that it's formed as a natural part of the cooking process, and we're trying to understand more about exactly what causes that formation and what can be done about it.

     We've seen already a number of testing results, some that FDA released at our public meeting in September, additional results that you'll see released today, as well as research from other places.  We know and are finding more and more that acrylamide is found in some foods, but not in others.  We know that even where it is found, there's really enormous variability, even within the same type of product or the same brand of the product.  So we need a richer database.

     Therefore, we know a lot more testing is needed, and I would just offer one caution.  As you look at the data today, you will see brand names identified with the samples that FDA tested.  But we haven't really done enough testing to know with any measure of precision what to compare with between brands, to know which brand may be higher or lower than another.  That's simply part of the raw data of what we've collected.  So it's an area where really a lot more testing is needed.

     We're finding a lot more research results already.  You'll be hearing--and I'm sure you've already heard--of important research identifying asparagine as one of the key components of how acrylamide is formed.  You'll hear some important data today on time of cooking and how the time of cooking may well be related to how much is formed.  These, again, are important parts of the building blocks.  I think of this as a puzzle to be solved, and each piece of research is another piece of the puzzle.  But we still have a fair amount of that puzzle yet to be uncovered and discovered.  But when you think back on a few months, that's an awful lot.

     For consumers, there's still a lot of unknowns.  We know, for example, that acrylamide is an animal carcinogen at high doses, but we don't know if it's a human carcinogen at the lower doses found in food.  Again, we know there are other toxicities associated with acrylamide, but its applicability to humans is not clear.  That's why we need more research.

     At this time, until more is known, we do advise consumers to eat a balanced diet consisting of a variety of foods that are low in fat and rich in high-fiber grains, fruits, and vegetables.  Today, what we really want this committee to focus on is the ambitious research agenda that has been developed.  This research agenda is designed both to assess the risk in people, to understand that much better, but also to understand how acrylamide is formed and what steps can be utilized to drive those levels down.  That has got to be our goal.

     So as you listen to the presentations, we would ask you to think of three overriding questions to see if we really are on the right track:

     Number one, are we addressing the right scientific issues?

     Number two, are there any significant gaps in the research plan that is being presented to you?

     And, three, have we assigned the right priority to them?  Are we doing the right things first so we can put together this puzzle most effectively and most expeditiously?

     I want to take a moment and thank in advance a large number of FDA staff that you'll be hearing from today that have put a considerable amount of work in and will continue in the future, beginning with Dr. Bern Schwetz, the senior science advisor, who led the U.S. delegation to the WHO consultation and who will provide an excellent overview; Dr. Terry Troxell, who is director of our Office of Plant and Dairy Foods and Beverages, what we fondly refer to as "land food," and this looks like a land-food issue if we saw one; a number of scientists in Dr. Troxell's division and elsewhere:  Dr. Rick Canady, Dr. Lauren Jackson, Dr. Steve Musser, Dr. Mike DiNovi, as well as our new chief medical officer, Dr. David Acheson, who will all be making presentations to you today.

     I also want to thank in advance Dr. Dave Lineback, who is director of JIFSAN, for his leadership in the research workshop that I referenced and for his taking time later today to share findings from that workshop with the committee.

     Finally, I want to thank Henry Kim, our exec. sec. to the committee, who works hard to keep everything going.

     As a final note, while the main focus of these meetings is on the scientific issues--obviously that's why you're here--we also want to be sure that, for lack of a better phrase, you feel well taken care of in terms of the facilities, any other logistics.  So if you have issues with that, we welcome feedback.  We accept positive feedback, also, should that be suitable.  But as we have moved into the College Park area and we're learning how to have meetings in this area, one of the best places to stay, one of the best meeting facilities, how do we make these meetings, we want you to want to come back.  We promise we won't have snow every time.  But we do need honest feedback, and we will try to be as responsive as we can.

     With that, I agree with the Chair.  We have a long agenda.  We want to be sure we get through it.  And so I thank you again for your expertise, for your attention, and your advice on how we can keep this ball rolling and continue to fill in the pieces of the acrylamide puzzle.

     Thank you very much.

     DR.BUSTA:  Thank you, Joe.

     We will continue on.  The FDA Action Plan, two sessions, both the goals and timetable and the charge and questions to the Subcommittee, Dr. Terry Troxell.

     DR. TROXELL:  Good morning.  I also want to welcome you to the first meeting of the Subcommittee on Contaminants and Natural Toxicants.  I want to thank you for your hard work in this matter as well as many other issues we will bring to the Subcommittee in the future.

     We are bringing acrylamide, a very challenging issue, to you as the first issue for the Subcommittee to consider.  The presence of acrylamide in foods was considered a major concern by the WHO/FAO consultation in June 2002, and FDA agrees with that conclusion.

     Acrylamide was quite a surprise to food processors, scientists, and public health officials worldwide when, on April 24th, the Swedish National Food Administration announced a finding that significant amounts of acrylamide formed particularly in carbohydrate-rich foods cooked with high-temperature processes.  Formation occurred using traditional cooking processes, whether prepared commercially or in the home.

     What is surprising is that numerous food scientists who published hundreds of papers had previously studied the reactions during cooking processes, particularly in research into chemical-based flavor development, and acrylamide was not discovered during those investigations.  As you will hear during the presentations today, acrylamide formation is closely associated with the Maillard reaction that is responsible for flavor in browning of carbohydrate foods.  As such, acrylamide has likely been present in cooked foods for thousands of years.

     FDA needs to be able to assess the risk to consumers of acrylamide.  Ideally, we need to know the actual risk in order to make the best decisions to protect the public health.  However, there are substantial data gaps regarding the risk of acrylamide in the food supply.  For example, we do not know the bioavailability of acrylamide in foods versus water where it has been studied.  And we do not know the metabolism of acrylamide in the rat versus humans, particularly at low levels of exposure.  Thus, the uncertainty about risk to human is large.

     Ideally, FDA would want a range of risk management options and would want to understand the impact of applying various alternatives on acrylamide risk reduction, on other risks that might be introduced by the alternatives, and the feasibility of the alternatives.  While there has been a tremendous amount of research in many countries since the Swedish announcement and there has been some very good progress, particularly on the mechanism of formation, technological solutions to prevent or minimize formation of acrylamide are likely to be more difficult than elucidating the mechanism.

     Some risk management options for acrylamide reduction by telling consumers to not cook food excessively have the potential to expose consumers to additional risk from foodborne pathogens.  In addition, because of the wide occurrence or wide variation of acrylamide in foods shown in the initial results in our exploratory survey, it's clear that more sampling is needed to describe the distribution of acrylamide in foods.

     As a science-based regulatory agency, the FDA believed it needed to assure that sufficient science was developed to permit it to make sound risk management decisions for the public good.  Therefore, FDA developed its action plan for acrylamide in foods.  We're bringing it to you today to seek your expert opinion and comment on the scientific adequacy and completeness.  We'll come to the charge more specifically a little later.

     Next slide, please.

     The action plan is designed around our overall goal, which is through scientific investigation and risk management decisionmaking to prevent and/or reduce potential risk of acrylamide in foods to the greatest extent feasible.

     The action plan consists of a number of sub-goals which are intended to ensure that we have a strong research agenda to provide a sufficient scientific basis to develop risk management options and public health messages.  We'll just briefly run through the sub-goals as the speakers will address the research goals in depth.

     The sub-goals are:  develop rapid screening methods and validate confirmatory methods of analysis.  The second sub-goal is assess the dietary exposure of U.S. consumers to acrylamide by measuring acrylamide levels in foods.  This morning we have provided you with the results of the levels in foods we have analyzed through November 15th in our exploratory survey.

     Identify mechanisms responsible for the formation of acrylamide in foods and identify means to reduce acrylamide exposure.  The primary mechanism is now believed to be known.  The major effort in the future will be to use that knowledge to reduce formation.

     The fourth sub-goal is to assess the potential risks associated at acrylamide by evaluation of available information and expanding the research into acrylamide toxicology.

     The next two goals are part of our action plan but are not about the substance of the research that needs to be done for the most part.  The first of these is to develop and foster public-private partnerships to perform the scientific research.  There is intense interest around the world in this issue, and there is a large amount of work to do.  Therefore, we believe that the quickest and most efficient way to address this problem is to foster collaboration and coordination of all research to the extent practical.

     As you will hear this afternoon from Dr. Lineback, the Director of JIFSAN, which is the Joint Institute for Food Safety and Applied Nutrition, our consortium with the University of Maryland, JIFSAN has undertaken a major effort to foster that collaboration and coordination through the WHO/FAO Acrylamide in Food Network that JIFSAN operates.

     The last sub-goal is to inform and educate consumers and processors about the potential risks, provide options on how to reduce the risk as knowledge is gained.  This is primarily an outreach element, but we are not going to discuss our consumer message at this meeting.  We will be discussing the message with the Food Advisory Committee at our meeting in late February.

     In the next several slides, I have listed some of the events that have or will occur to give you a perspective of where we are in the sequence of events and process.

     On April 24th, Sweden announced the finding of acrylamide in foods.  At that time, no methods were available.  Therefore, we developed our own liquid chromatography MS/MS method, which we posted on June 20th on the website for general use by all researchers.

     The WHO/FAO expert consultation was held only two months after the announcement by Sweden and provided a sound initial review of the science and the research needs for acrylamide in food.

     On September 24th, we held a meeting of many federal agencies working on acrylamide that primarily focused on the toxicology issues.

     On September 30th, we presented this action plan to the public.

     October 28th, JIFSAN and our National Center for Food Safety and Technology held a meeting of scientific experts from around the world to update the status of the research and then determine the research needs.  We have provided you with a copy of the research needs developed by each work group at the meeting, as well as the Planning Committee's short-term priority research needs.  Dr. Lineback will address this meeting later today.

     We plan to take this issue, after considering your recommendations, to the full Advisory Committee in late February.  We're also planning to hold a meeting to determine the status of the science as of late summer, probably.

     Next slide.

     I have listed here a variety of scientific meetings, clearly not complete, as this subject will be discussed at many meetings in the next several years.  But this Society of Risk Analysis tox forum and germ cell mutagen risk assessment workshop and so on are just some of those.

     In the next slide, on the international front, in addition to individual country and EU activities, there are notable international community events.  The Codex Committee on Food Additives and Contaminants is responsible for international food standards for contaminants and toxicants.  We expect acrylamide will be taken up as new work to develop a position paper which will be used to determine if the code of practice for maximum levels are appropriate for food in international trade.

     Also, the WHO/FAO Joint Expert Committee on Food Additives, JECFA, will evaluate acrylamide.  JECFA provides risk assessment to the CCFAC.  JECFA's determinations are used by many countries in the world as the definitive safety and risk assessment of food additives and contaminants.  Acrylamide has been tentatively scheduled for the winter-spring of 2004.

     Now I would like to turn to the Subcommittee charge.  You should each have a copy of the charge and questions.

     The Subcommittee is being asked to evaluate whether the research steps outlined in FDA's Action Plan are scientifically adequate to describe and address the public health significance of acrylamide in food.  In order to support our overall goal, our intention is to build a strong research agenda that we are asking for your expert advice on.  We have the following questions for the committee:

     Based on what we know about acrylamide toxicology, occurrence, formation, exposure, and risk, are the research steps appropriate to describe and address the public health significance of acrylamide in foods?  We have listed three sub-categories of research steps in relation to the first question relating to occurrence, to exposure, formation, and toxicity.

     Our second question then is:  Are there gaps in the research plan or areas where emphasis should be increased?

     The third question relates to priorities.  Besides gaps and emphasis, we really need to understand if particular items are of higher priority for research to address.

     The next slide deals with the process.  This last question on priorities is particularly important in order to permit the agency to make timely decisions informed by sound science.  We view the process as an iterative one.  We have done the initial hazard assessment and determined that there is not enough information to develop risk management options.  We are asking you today to help us complete the data needs identification.

     In the meantime, as you will see, we have begun data development.  We will review the state of the science periodically and use risk assessment as appropriate to determine if there is another research component that needs to be added, or if we have enough information to support development of sound risk management alternatives.

     Next slide.

     We do not have a specific timeline for the periodic reviews, but these are among the time points when we could review the science for decisionmaking.  One of those time points would be when there is a breakthrough in formation research that leads to substantial reduction of acrylamide levels in foods; when key toxicology data needs are met; at the FDA-sponsored scientific symposium in the summer of 2003; at the time of the JECFA evaluation in the winter-spring of 2004; and also in connection with CCFAC position paper development and possibly development of code of practice.  That will run between 2004 and 2005 and possibly later.

     In summary, as you will see, much has been done in a very short amount of time, and there is a need to do a great deal more.  At this point in time, we are asking you, the Subcommittee, to evaluate the scientific foundation of our overall approach.  Also, as you will see, you have a very full day, and thank you very much for your attention and work on this.

     DR. BUSTA:  Thank you, Terry.

     Are there any questions of clarification for Terry from the Subcommittee?

     [No response.]

     DR. BUSTA:  Well, we're rolling along, very good timing.

     DR. LEE:  Don't worry.

     [Laughter.]

     DR. LEE:  Don't be too optimistic.

     DR. BUSTA:  Always get a little bit of a lead.

     Our next item on the agenda is the scientific overview of acrylamide in foods by Dr. Bernard Schwetz.

x      DR. SCHWETZ:  Good morning to all of you.  I will try not to carve into the lead that has been created.  But I also want to thank you for your willingness to help us with this important issue.

     What I want to do is provide a little bit more background information.  Terry referred to a lot that's happened in the last few months.  I want to go back to the kind of information that we reviewed at the WHO consultation to give you an idea of what the stage was like at that time as a basis for developing the plan and for some of the things that have happened in the last few months.

     It was just in April that the Swedes did report finding acrylamide in a wide range of food, particularly in carbohydrate-rich foods cooked at high temperatures.  The results were published in the Journal of Agricultural and Food Chemistry in July, and that started a lot of activities.

     The Swedish research was prompted by the observation that people without a known exposure to acrylamide had measurable levels of acrylamide adducts primarily to hemoglobin, and that triggered the question, of course:  Why would they have these adducts to hemoglobin when we didn't even know they were exposed to acrylamide?

     Well, subsequently, they did further observations in humans, but they also did a study in rats where they fried rat chow and were able then to measure acrylamide in that rat chow and were also able to measure the hemoglobin adducts in the rats.  So that gave further evidence to the fact that high-temperature treatment of feed or food could contribute to acrylamide and, therefore, measurable adducts on hemoglobin.

     Well, they proceeded then to measure acrylamide in a variety of foods and looked at the process of cooking as a means of contributing to the formation of acrylamide.  They certainly suggested that acrylamide in food was a significant component of the acrylamide exposure as we knew it, but there was also that concern that it may not account for all of the hemoglobin adducts that humans carry.  So there could well be exposures other than food, and the uncertainty of that, of course, led to the need to look harder at what was contributing to this load of hemoglobin adduct.

     I want to make the point that the formation of acrylamide in food--we have no reason to believe that this is something that just started in April.  It's very likely that that has been going on as long--we have no idea why it wouldn't have gone on before then.  So we have to assume that while the discovery occurred in April, and the report of it, this is something that's been going on for a long time.

     What is new is the observation that acrylamide might be present in food from traditional methods of cooking food, and that gave us a handle of what we needed to be looking at.

     In response to the Swedish findings, the WHO and FAO convened the expert consultation on the public health implications of acrylamide in Geneva, Standard, June 25-27 of this past summer.  There were three of us from the FDA who participated in this group of some 30 people who were convened to look at this broader issue, and I was one of those who represented the FDA.  So what I want to do is present you with some of the information that we looked at and the conclusions that were drawn from that consultation.

     As you might expect, this is not a simple issue.  The fact that acrylamide is present in food and what to do about it and how to assess the risk is a complicated issue.  And the consultation members were challenged to review and evaluate the new and existing data and the research on acrylamide relevant to toxicology, to epidemiology, to exposure assessment, to analytical methodology, the formation, and bioavailability of acrylamide from cooked food, but also to identify needs for further information and further studies to help understand this issue, as well, though, to develop and suggest possible interim advice for governments and for industry and for consumers relative to the presence of acrylamide in food.

     In order to do this, we had access to international documents that have been written, the published literature.  We had all of these documents in front of us in one form or another so that we could review in detail the information that various countries were looking at to analyze this problem.

     Some basic facts about acrylamide as we reviewed it.  Acrylamide is also something that has been on commerce for many decades.  It isn't a new use in any sense.  It has approved uses as grouting materials, and it has been used in water purification procedures.  Many of us who worked in laboratories are familiar with polyacrylamide gels in laboratory uses.  So there's been a lot of exposure other than food through the years, and, in fact, some of the known human neurotoxicant situations were from its use in grouting materials, the application of them.

     There clearly is a contribution to our body burden from cigarette smoke, and that's another factor that we have to take into account as we consider what would be consideration of a level that we would try to achieve of human exposure when you have another significant source of exposure like smoking.

     Another point is that acrylamide and its major metabolite, glycinamide, are both chemicals of concern because they adduct to DNA and to protein.  Glycinamide appears to be particularly important in creating adducts to DNA.

     Just a comment about the hemoglobin adducts.  While it's convenient to be able to go out and measure these hemoglobin adducts in people, we have to be careful that we treat these as measures of exposure, not necessarily measures of risk.  And we go through this a lot in toxicology because you can measure it.  You tend to draw conclusions and make decisions based on what you can measure.  But we have to be careful to remember that this is probably a good biomarker of exposure, but whether or not it's a good measure of risk for individual people or from certain types of exposure is something that we need to explore in greater detail.

     We talked in this consultation quite a bit about the formation and occurrence of acrylamide and the fact that it was found in certain foods that were processed at high temperatures.  Carbohydrates, proteins, amino acids such as asparagine, as well as lipids, the natural components of food, are all potential precursors to the formation of acrylamide.  The likelihood that there would be one mechanism of formation is very small.  It is more likely that there would be several mechanisms of formation, each contributing some amount of acrylamide.  Increased cooking time or temperature and low water content seem to be associated with an increased level of formation in the foods.  The acrylamide levels in food probably reflect formation as well as disappearance of acrylamide, not just the level of formation in food.

     What about methods of analysis?  It was important to get this group of people because of the new data that came out from Sweden and the methods that were being used, to get this group to weigh in on whether or not the analytical methods that were available that led to this new information coming out were methods that the group was comfortable with.  So we talked about the GC/MS and the LC/MS/MS methods that were available.  The consultation expressed considerable confidence in these methods, but recognized that they had not been fully validated, and that validation process continues today yet.  So while we have methods that people are using, there is a formal process of validation that we're still going through.  And as was mentioned by Terry, when the FDA had its LC/MS/MS method we were using internally, we put it on the Web so that everybody would know what method we were using.

     In addition, Dr. Musser's group has analyzed hundreds of foods, and that's part of the information that you will be seeing during this meeting.

     Regarding exposures, the consultation prepared an estimate of exposure to acrylamide from food, but we also noted that there were some pretty significant limitations in being able to narrow the confidence limits around what that exposure number might be.  It was based on a limited number of samples, a limited range of foods, and the observation that there could be a broad range of acrylamide levels in any given type of food.  So what you might find in one chip isn't necessarily representative of what chips in the next bag might be, the next batch.  So a lot of variability.

     But if you consider an estimate of a short-term exposure, it was something in the range of 0.8 microgram per kilogram per day for the average consumer, just under a microgram per kilogram per day.  Also recognizing that children probably have on a body weight basis a higher level of exposure than adults, and that becomes a point of consideration as we worry about what are the things that we should really be concerned about.

     Let's talk about the toxicity and the toxicology just for a minute.  Acrylamide is associated with neurotoxicity, with genotoxicity, carcinogenicity, and fertility effects in laboratory animals.  Regarding the neurotoxic and fertility effects, the consultation concluded that the amount of acrylamide in food was not expected to have neurotoxic and reproductive effects, and that was an important consideration, but obviously it isn't a conclusion that won't be challenged.  And as we develop more data, we will be looking to see whether or not, in fact, that holds up to be true.

     Remember that the estimate of human exposure is a microgram per kilogram per day.  The NOAEL for the neurotoxic effects in animals is about 500 micrograms per kilogram per day, a 500-fold margin.  The NOAEL for fertility effects is about 2,000 micrograms per kilogram per day.  So those are the comparisons that we were looking at.

     But I should remind you also that the one known human toxic effect of acrylamide is its neurotoxicity, and that's primarily from occupational exposures.

     The group talked quite a bit about adduct formation.  Both acrylamide and its metabolite, glycinamide, form adducts on proteins, and I talked about the hemoglobin adducts as a marker of bioexposure.  But acrylamide and glycinamide also form adducts on DNA, and it's in that context that we worry about its genotoxic potential and its carcinogenic potential.

     Acrylamide is genotoxic.  It induces heritable damage in germ cells and somatic cells, and high doses of acrylamide induce tumors at multiple sites in rats and mice.

     IARC, the International Agency for Research on Cancer, has classified acrylamide as a probable carcinogen to humans.  Epidemiologic studies are limited, but people exposed to acrylamide in the workplace have not shown evidence of a carcinogenic effect.  But the studies are limited because of the typical problems that you have with epidemiology studies:  limited number of people, limited number of replications of these kinds of observations.

     So at this point, we're faced with the knowledge that it is a rodent carcinogen, and no evidence of carcinogenicity in humans from the limited epidemiology studies that have been done.

     To put this in the context of other food carcinogens, food contains a number of other chemicals that are carcinogenic or mutagenic in laboratory systems.  Polycyclic aromatic hydrocarbons, heterocyclic aromatic amines, natural components of foods, also are known to be carcinogenic in laboratory models.

     One concern about acrylamide is that the level of human exposure of acrylamide in food might be higher, probably is higher than some of these other natural carcinogens that we find in food, and that's one of the reasons we're concerned about it.

     The consultation reached a conclusion that I think is shared by all of us that the exposure for such chemicals as a carcinogen in food should generally be as low as reasonably achievable.  So it isn't a matter of recognizing that it's there and not doing something about it.  We all share the goal of trying to reduce this.  But there was pressure during this meeting to come up with some quantitative estimate of risk, and we resisted that.  And as a group, we did not come up with any risk number because we all felt that the data did not support that at that time, and that, in fact, is something that will be attempted in the future as we have more data about exposure and the toxicity.

     Some specific recommendations that came out of the consultation.  In the area of toxicological research, recommendations that we need to know more about the metabolism of acrylamide.  We need to know more about acrylamide and glycinamide binding to DNA and proteins in relationship to the toxicologic profile.  We need to know more about the bioavailability of acrylamide from food.  We also need to know more about the non-food sources of exposure to acrylamide.  And also if there are opportunities for doing cancer epidemiology studies in humans, we need to be going there.  We need to do the best we can to find these populations.  There are populations that have measurable levels of hemoglobin adducts occupationally in the world.  The possibility that we might do some cancer epidemiology on those people.

     Further studies to define the carcinogenic and genotoxic potential of glycinamide.  There is information on acrylamide, primarily given in drinking water, but no studies on glycinamide by itself.

     Also, this issue of germ cell damage, the fact that there's a germ cell mutagen in food is of considerable concern to us, and that's one of the reasons Rick Canady and others have been exploring a workshop to get the best minds that we have together to look at that issue and try to decide what risks might be associated from that standpoint.

     From the question of acrylamide formation, the consultation recommended that there would be a systematic examination of processing conditions so that we would have a better handle on what conditions are associated with the formation and then to optimize the processing conditions to minimize the formation of acrylamide in the industrial processes, but as well as home.  We can't overlook the fact that we can add acrylamide to our food based on how we cook it in the kitchen at home.  It isn't just a matter of what the major manufacturers do out in the plant.

     The recommendation also was that we examine foods from different regions of the country, of the world, and different diets to find out how exposures to acrylamide might vary based on different types of diets that people have from the standpoint of the methods of analysis, the need for inter-laboratory validation of the analytical methods, and the need to have data from a broad range of food types.  The need to develop reference materials so that laboratories across the world could all be working with the same methods operating under the same conditions.  And also eventually to develop low-cost and simple and reproducible methods for routine monitoring of acrylamide in food.

     From the standpoint of exposure, again, to focus on--develop knowledge of food from different regions of the country and different diets, different regions of the world, and also continue to look at the use of biomarkers as an index of exposure and correlate that with what we know about its presence in food.

     Another recommendation was that there would be an international network for sharing data on acrylamide in food, and Terry made reference to the fact that Dr. Lineback volunteered that JIFSAN would be a place where that could happen, and that has moved forward so that the development of that database to collect information from industry, from academicians, from government agencies, that all that information would be made available in one database.  That is now happening at JIFSAN.

     As you hear about the FDA Action Plan the rest of this meeting, you will see that what is in the plan is consistent with a number of these recommendations.

     Regarding the last recommendation that the consultation looked at, what do we tell the public, what do we tell consumers, the recommendation was that food should not be cooked excessively but should be cooked thoroughly to destroy pathogens.  And the second piece was that general advice on healthy eating should be followed, to eat a balanced and varied diet with plenty of fruits and vegetables and moderate consumption of fried and fatty foods.  So this is not an acrylamide-unique message.  It's the kind of message that people have been giving for some time, but it is good that if we are going to try to minimize people's exposure to acrylamide, that is consistent with a healthy diet, as we have been talking about it in other contexts.

     Just a couple of points to leave you with, to remind you again of the fact that acrylamide in food is not new.  Our knowledge of its awareness is new, but it's not a new issue in terms of it being in food.  Acrylamide is a potential human carcinogen based on the laboratory data, but we don't know yet what risk it poses to humans regarding the carcinogenic risk.

     We certainly are doing a lot of planning.  There are a lot of meetings, as Terry showed.  The world has organized to look at this issue, and that was one of the good things about the consultation.  It brought all of us together very early on to identify how different groups might look at the data differently and to begin to talk about common methods of analysis and common interpretations of the data.

     So that's all I will say.  I'll be happy to answer any questions that you might have.

     DR. BUSTA:  Are there any questions for Dr. Schwetz?

     DR. HOTCHKISS:  Yes, Joe Hotchkiss.  You briefly touched on this, and it's a consideration that had come up in my own mind, and I wondered if particularly the consultation explored--to what depth the consultation explored the relationship between the known tumorigenic and genotoxic pyrolysis products in foods, the heterocyclic amines and the series you mentioned.  And acrylamide, I think you mentioned that acrylamide is likely to be higher exposure for food, and I think there probably is sufficient data to agree with that.

     On the other hand, the toxicology would suggest that many of these pyrolysis products are considerably more toxic, particularly more genotoxic and carcinogenic.  And I wonder to what depth that comparison was explored.

     DR. SCHWETZ:  There was minimal discussion about that whole topic.  I provided this information on the other chemicals just to provide the context, that we do have a number of foodborne carcinogens, food carcinogens, but that was not a topic of discussion to talk about the relative risk or the relative toxicity of these various carcinogens in food at the consultation.

     DR. BUSTA:  Are there any other questions for clarification?  Dr. Fischer?

     DR. FISCHER:  Larry Fischer.  Bern, you mentioned that the estimate was mentioned of exposure around 0.8 micrograms per kilogram per day.  I'm wondering how that was done.  Why were you reluctant, in fact, to think about a risk but you were willing to make a huge guess at what a typical exposure might be?  Can you tell us about that number?

     DR. SCHWETZ:  Well, it was based on the known food consumption patterns in various populations and an estimate of what the Swedes and other groups had already analyzed of acrylamide in food.  So it clearly was an estimate, and I don't even known the confidence range around that estimate.  But it was based on known patterns of food consumption and what limited data we had on its presence in certain food types.  So it is an extrapolation to really come up with a number and say that it's 0.8.

     DR. FISCHER:  This was for our type of diet, right?  A Western diet or whatever you want to--

     DR. SCHWETZ:  Well, no, it would be in those parts of the world where there's good information about what people eat.  It took those various food patterns into account.  But we did resist going through a quantitative risk assessment to predict how many people would die from cancer.  So it's a different thing to take what data you have on what people eat and a limited number of data points on what's in the food.  But to extrapolate and try to predict how many people would die from their diets, we didn't want to go there.  And we still don't.  That's why we have a research plan.

     DR. FISCHER:  Yes, I understand.

     DR. BUSTA:  Alex?

     DR. ACHOLONU:  Please, did I hear you say that children have higher level of exposure to acrylamide than adults?  If you said that, why and how does that occur?

     DR. SCHWETZ:  The reality is that on a per kilogram body weight basis, food intake by a youngster is higher than it is in an adult.  So that by itself contributes to a higher--if the children are eating foods that have acrylamide in it, just based on the fact that they eat more food per unit of body weight means that they would have a higher level of exposure.

     The offsetting question is:  Do children eat food that has acrylamide in it?  And they probably don't as infants.  For the most part, they don't eat potato chips and French fries and some other things that might have a higher level of acrylamide.  But at some point as youngsters they begin to at a time when they still have a high level of food consumption per unit of body weight.

     DR. BUSTA:  Any other questions?

     [No response.]

     DR. BUSTA:  Thank you very much, Dr. Schwetz.

     We'll move on to the FDA Action Plan, major components of the action plan, Dr. Richard Canady.

x      DR. CANADY:  Good morning.  I'm a toxicologist with the Center for Food Safety and Applied Nutrition, and it's my pleasure to walk you through the major components of FDA's Action Plan for acrylamide in food.

     My basic goal in this presentation is to give you an overall framework for the presentation that I'm going to follow.  Most of what I'm going to talk about is going to be covered in detail within individual presentations later today.  But the purpose here right now is to give you an overall idea of the framework, the overall action plan that FDA has put forward.

     Next slide, please.

     This slide is to help orient you to my particular talk.  I'm going to focus on five subject areas, five of the major components of the action plan.  I'll talk first about testing foods, then move into mechanisms of formation for acrylamide, our action plan with regard to that; talk then about the toxicology part of the action plan; move into education; and then finally wrap up with how the action plan works through meetings and collaborations.

     Next slide, please.

     The first major component I'd like to talk about is testing foods, and within this component, the first hurdle that FDA needed to cross was developing a method to test for acrylamide in foods.

     Can you hear me okay now?  Is this working?  No?  Okay.

     [Pause.]

     DR. CANADY:  Methods development is usually a need-specific undertaking.  For example, we need specific methods to answer a specific decision need.  With regard to acrylamide in foods, our current testing need is to establish the scope of occurrence of acrylamide across foods, and then also to estimate exposure.  So within that particular need for a method for analysis, we need to have, for example, a good level of detection or an adequate level of detection, and also sufficient throughput to work through the various food samples that we need to address in order to understand scope of occurrence and then to understand exposure.

     As we move into other decision needs, the method of analysis may change.  For example, as we move into monitoring needs or into enforcement needs, we may need to develop additional methods.  But the main point here is that we've developed a method that is designed to address our current needs, and as we move forward with the action plan, we may need to move into additional methods development.

     Next slide, please.

     So the second aspect I want to talk about with regard to testing is testing for occurrence. In a minute, I'll talk about exposure, testing for exposure, but the point about talking about occurrence first is that we need basic information about the scope of occurrence before we can move into testing for design to understand exposure.

     Another way of thinking about is that we need to test for occurrence to understand if we're, for example, missing the big picture.  We need to understand all of the foods or at least have a fairly good idea of the range of foods in which acrylamide occurs before we can move to the next step of understanding how to estimate exposure.

     Another aspect of testing for occurrence is we need to also understand, for example, how many samples we would require in order to understand an estimate--or get an estimate of average for a given food or a given brand, a given exposure group, a given demographic group, for example.  So we need to have an understanding of occurrence and an understanding of variability of occurrence before we can move on to estimating exposure.

     There are three stages to determining scope.  First is that we need to confirm occurrence in U.S. foods, and this is something we've already accomplished, obviously.  Second is, again, we need to map occurrence across foods to inform both formation research.  Knowing that it occurs in one food and not in another food helps you understand something about the formation mechanisms, or at least the processes that lead to formation.

     Then after we've described the range of occurrence across foods and understand something about where it occurs and where it doesn't occur, we would then move into describing the variation within a food, for example, or within a class of foods.

     Next slide, please.

     As we develop sufficient understanding of the occurrence, the scope of occurrence, including its variability, the next stage again is that we move to exposure evaluation.  So the next stage of testing would be used to determine what we need to know in order to understand exposure.  There's two aspects that we would use this.  First is in risk characterization--this exposure information.  First is in risk characterization so that we can understand the decision context we're in with regard to regulatory options or doing something about acrylamide in food.  And the second is to monitor changes over time.  We want to be able to see, for example, if our actions or any circumstance, actually, is leading to a decrease in the exposure to acrylamide through food.  So we have an intention to look periodically or look continuously at acrylamide levels in food to understand exposure over time.

     Next slide, please.

     We'll essentially use all available data that we can to understand exposure, but there's going to be primarily several breakdowns of that exposure or that occurrence information.  We're going to use our exploratory survey data, obviously.  We're also going to use information from FDA's Total Diet Study.  And, again, I want to emphasize there's going to be more detailed presentations of this information later.

     So we'll use exploratory survey data.  We'll use Total Diet Study information.  We'll use other information as appropriate to understand the variability and the occurrence of acrylamide in foods.

     Given some of the information you've heard already this morning, we're also quite interested in understanding other sources of exposure or just generally understanding exposure to acrylamide from all sources of exposure.

     To accomplish that, one of the things that we're working towards is collaboration with the Centers for Disease Control, the National Center for Environmental Health down in Atlanta, and through that what we intend to do are two basic things:  one is look at NHANES, which is the National Human--I always forget this acronym because it's not exactly what the letters are, but it's Human Health and Nutrition Exposure Survey.  You all know it so I don't need to actually say it.

     We'll be working through NHANES, but then we'll also be working through specific focused population studies prior to NHANES because, as most of you know, if you know how to spell out the acronym, you also know that it takes a long time to get the information out of NHANES because it is a large population-based survey and requires a lot of front work.

     Next slide, please.

     I wanted to give you an idea of the level of sampling that we are currently undertaking and that we're considering for the future.  Within our initial exploratory survey, we originally thought of looking at around 600 samples, and we're in the middle of that exploratory survey.  We're going to, as needed, develop more sampling or plan to look at different foods, extend our sampling for specific foods and so on.

     We also heard recently that around 4,000 or so results are soon to become available.  Now, these results come from a variety of sources.  They may consider pre-processing results as well as post-processing results, pre-cooking and post-cooking results.  But that information should be useful, again, in helping us understand exposure.

     Again, we also will be looking at Total Diet Study samples over the next year, and then continuously after that as needed.  We plan on a basic level of sampling of around a thousand samples a year from the Total Diet Study.  And these are samples that come from across the country in a rather regimented way.

     Next slide, please.

     I want to stress that FDA's approach to sampling is necessarily an iterative process.  It's iterative in the sense that what we know today influences what we need to sample, and our decisions needs today also influence what we need to sample.  Again, seven months ago, we didn't even understand that acrylamide occurred in foods.  We had an initial understanding of some of the foods it occurred in after the Swedes published their results.  We used that information to decide how to sample for the first iteration.  That's our exploratory survey.  We also considered information like food chemistry, food processing techniques and so on, in order to understand where to sample.  And we considered information with regard to consumption rates, again, to understand where to sample.

     That first iterative process we're still in the middle of, but once we've collected that information, we will then understand how to go to the next step.  I just want to stress that this is, again, a process of understanding where we are, figuring out where we need to go.  It's an iterative process.

     Next slide, please.

     When you think of iterative processes, you need to have a little bit of an understanding of where you're going to stop the current iteration, and we have obviously some understanding of where we would like to stop.  Putting it into specific terms is a little bit hard to do.  It is, again, a point of understanding where you are and then figuring out where you need to go.

     With regard to the scope of occurrence across foods, one of the scope definition points is really just a sampling to a point of diminishing returns.  As we get to a point of understanding that we've looked across the different food categories adequately and we're not seeing any more new samples where we see detectable acrylamide, we'll have an understanding that we're coming to a definition endpoint.

     With regard to variability, the key issue here is being able to confidently estimate differences between foods or differences between brands or differences between sources and so on.  So that definition is going to have a little bit more of a statistical basis in the sense that we need to be able to confidently estimate the need.

     Next slide, please.  I don't want to get too far out of my notes here.

     With regard to formation, this is the second major component of FDA's plan that I'd like to talk about.  What we're attempting to accomplish here, again, is an understanding of how acrylamide is formed and then through that understanding hopefully develop methods to prevent or reduce formation of acrylamide.  One thing to keep in mind, though, is we want to have a reasonable assurance that our actions result in an overall reduction of potential risk, an overall reduction of net risk, so that we through our actions don't engender new risk, don't create new risk through changes to how foods are cooked, for example, through changes to nutritional content and so on.  This is something that Dr. Acheson is going to speak to in a little more detail later in today's presentations.

     Formation research has two sort of natural divisions or components to it.  One is process evaluation, how the foods are cooked, how they're prepared.  And the second is chemistry evaluation.  What are the actual chemical mechanisms through which acrylamide might be formed?

     Next slide, please.

     FDA's Action Plan with regard to formation uses leveraging to a fair degree in order to take advantage of the substantial interest and the substantial research that's out there right now with regard to acrylamide formation.  We're working through our National Center for Food Safety and Technology in Chicago, investigating mechanisms of formation, primarily through processing evaluation.  And Dr. Lauren Jackson will be talking about that later today.

     FDA is also working through the Joint Institute for Food Safety and Applied Nutrition as a way of leveraging out to a lot of the food chemistry knowledge and also food processing knowledge with regard to the formation of acrylamide in foods.  Both consortia provide conduits to and participation with academic institutions, other government bodies, and industry research.

     Next slide, please.

     The third major component of FDA's Action Plan is evaluation of toxicology, and I have the pleasure of helping you understand that particular component later on this morning--actually, the next talk.  Within this component, the overall goal is to examine the likelihood that adverse events or adverse health effects are caused by exposure to acrylamide through food.  There's three major sub-components here.  First is identifying the gaps in our understanding.  We need to understand what we don't know before we move forward with determining the overall risk.  We need to ask basic questions:  Do we have enough information?  Do we know all of the endpoints that we need to consider and so on?

     The second is prioritization of these data needs and these data gaps according to our decision needs, in this case in terms of our risk characterization needs.  We don't want to do research just for research's sake.  We want to do research that allows us to understand the risks better.

     And the third component is that FDA will sponsor, coordinate, and in other ways encourage research in order to accomplish these goals for toxicology.

     Next slide, please.

     The fourth major component of FDA's Action Plan is the very important aspect of education.  FDA intends to develop educational material to inform and educate both consumers and industry with regard to the initial risks, and as knowledge is gained, we would like to be able to provide options as to how to reduce the risk.  In providing this information and working with stakeholders--in providing this information, rather, we intend to work with stakeholders in developing the messages.  So it's not just going to be solely an FDA-directed enterprise.  We will work with stakeholders in developing these messages.

     Next slide, please.

     The fifth major component I'd like to focus on--and this is a component that will not be focused on in detail through other presentations, although you've already heard about this through a couple of the presentations, and there will be individual mentions of collaborations and so on in subsequent presentations.  But that's meetings and collaborations.  There's intense interest out there and intense participation at this point with regard to research for acrylamide in foods.  This intense interest has obvious positive aspects in the sense that there's a lot of new information coming out.  But it has a challenge in it as well, and that is that we have the potential of what in computing is called a massively parallel computing situation where what we have is a lot of information that could be coming from a lot of different sources.  The challenge to us is to try to channel that information and coordinate that information so we can get the most progress within the shortest amount of time.

     And so meetings and collaborations come to be a quite important component of FDA's overall plan, and there's four general categories that I would place before us.  First is data needs meetings.  We've had a fair number of those already.  We had four or five of those already, and where we try to understand what we know and what we need to know.

     The second quite important one that WHO/FAO came up with or had as one of the outputs of the meeting back in June was that we need a centralized location for understanding--or for sharing research projects and understanding about acrylamide in foods.  The WHO/FAO has developed the food--or, rather, JIFSAN working for WHO/FAO has developed a food acrylamide information network, and the intention of this is to allow sharing of information across organizations, again, to help encourage a massively parallel computing approach that we get the most done in the least amount of time.

     We also have interagency working groups, and this is within the U.S., obviously, where we are trying to coordinate--first of all, understand what research is going on among the different agencies, and then coordinate that research to the best advantage.  And, again, we have consortia that we're working through to leverage both through academic, industry, and other stakeholders with regard to research needs.

     Next slide, please.

     This is a long, detailed slide.  The message really is the amount of text on here, not each individual text.  You've seen this list already with Dr. Troxell's presentation.  The intention here is to show that there's a lot of meetings and collaborations going on where we are attempting to and making progress at coordinating and encouraging research.

     Next slide, please.

     So, to sum up, for food testing FDA has initiated an exploratory survey and is moving into exposure assessment as scope and variation become better understood.  For formation research, FDA, JIFSAN, and NCSFT are working with industry, academia, other national governments, of course, to look into how acrylamide is formed and try to develop an understanding of how it is that we can prevent or reduce, as feasible, formation of acrylamide in foods.

     In toxicology, we're seeking an improved understanding of the potential risks for acrylamide exposures through food, starting with an understanding of the data gaps, and then moving into sponsoring and coordinating research.  Again, the goal there is to help understand the risk, not to just do research for research's sake.

     For education, we will pass knowledge on to stakeholders as it's gained and determine messages with their input, with stakeholders' input.  And, again, given the worldwide scope of interest for this food acrylamide issue, I think Joe Levitt mentioned earlier that there's meetings like this going on all over the world right now.  That's been noted several times.  We have an opportunity to try to collaborate, try to coordinate that research to our best advantage.

     Thanks very much.

     DR. BUSTA:  Thank you.

     Are there questions for clarification for Dr. Canady?  Dr. Downer?

     DR. DOWNER:  Goulda Downer.  You talked about the FDA Total Diet Study, and I suspect I'll hear more about it from other presenters.  And you also mentioned the variability across foods and within food types.  Are you also looking at combination of foods since we tend to eat a combination of foods and not just one food or food group?

     DR. CANADY:  Yes, that's an important part of the exposure assessment, and it's one of the reasons why the estimates that were developed by WHO/FAO were a preliminary range of magnitude, order of magnitude estimates and so on.  Because you're right, for different subpopulations particularly, and even within the general population, the combination of foods, when you eat particular foods, what combination you eat them in, will have an important effect on the variability of intake, at least the daily and the temporary variability of intakes.  So, yes, that's a quite important part of the exposure assessment.

     DR. BUSTA:  Dr. Hotchkiss?

     DR. HOTCHKISS:  Yes, Joe Hotchkiss.  You've quite appropriately recognized the importance early on of exposure.  I wonder if you could just clarify something for me.  You mentioned NHANES and a couple other databases you said you've used, and maybe this is included in it, but the most powerful dietary assessment or exposure data that I know of is used, for example, by EPA for pesticide exposure.  It's available commercially in the private sector through TAS and other organizations, basically the same database, which is in part based on NHANES as well as the USDA surveys.

     As a matter of fact, I think that database is, at least in my personal experience, the most commonly used database on petitions submitted to FDA for food additives and stuff.

     DR. CANADY:  Right.

     DR. HOTCHKISS:  You didn't mention that one specifically, and I wondered if there was a reason for that.

     DR. CANADY:  No, there isn't a reason for that--well, the reason is that I didn't get into that detail.  The reason for bringing up the NHANES survey and working with CDC is to capture the total exposure or, rather, to use this advantage we have in a biomarker of exposure through adducts to hemoglobin to our advantage.  We will, of course, use the USDA's database and software packages that combine various consumption data in order to provide the exposure estimate.

     And, you know, Dr. DiNovi is going to talk about that in some detail.  And, again, I didn't mention it simply because I didn't get into the detail of the exposure assessment.  It's not something that we're not going to use this time for some reason or another.

     DR. BUSTA:  Are there other questions for clarification?

     [No response.]

     DR. BUSTA:  I have one on an iterative approach to testing.  What activity will be utilized to keep the consistencies of value of the earlier data as you develop new and different types of methods?

     DR. CANADY:  That's an important question because, as you first cast about, not knowing where all it's occurring, you may not design your sampling to match what you need to sample later on.  So the utility of the earlier data and the degree to which you could mix it in with later data is an important part of the overall sampling.  And that's one reason that for our initial approach we chose to try to survey across foods and try to understand where it occurred and not try to initially, for example, develop an exhaustive and statistically based sampling for each food, because we really didn't know enough about the variability to design that kind of sampling.

     But you're absolutely right.  Some of the information we have collected initially may turn out to be not as useful for the exposure assessment as data collected later where we have a better understanding of what is needed in terms of statistical design.

     DR. BUSTA:  Yes, George?

     DR. GRAY:  George Gray.  I'd actually like to follow up on that question a little bit.  It's thinking about the huge potential task that is out there to identify and to characterize the variation in the levels of acrylamide that might be in foods.  And one of the things that struck me in your presentation is one of the things I think FDA wants to think about is matching the data that are available on consumption to the sampling data.

     For example, you mentioned something about taking the time to characterize the distribution, the variability of acrylamide across brands.  It seems to me that doesn't make much sense given that our data are on consumption of foods, not brands of foods.  So sort of thinking backwards from the information you have to get to a place where you can design your sampling strategy--this is probably going to be talked about in the exposure assessment section.  But think backwards from what you have so that you don't get more detailed information in one area than you have to match up within another.

     DR. CANADY:  Yes, and this relates obviously to the earlier question as well.  It speaks to the need within our current exploratory survey to address two goals.  One is to inform the formation research, and the other is to help us understand where to sample for exposure.

     So, for example, sampling across brands was important in the sense that it could have provided information with regard to processing or ingredients and so on with regard to formation research.  But you're absolutely right.  When we go to exposure assessment, that cross-producer or cross-food type information will need to be generated as it applies to the exposure assessment, not as it applies to differences between foods.

     DR. GRAY:  And you probably need fewer samples to characterize that variability.

     DR. CANADY:  Yes.  So, for example, you would focus on getting a weighted average based on consumption to some degree, or you would use information in a way that informed you on that aspect, right.  But, again, we had a dual need initially with our exploratory survey.  One was to try to scope out what we needed to understand for exposure, but then we also at the same time needed to understand where it occurred to understand formation--to help inform formation, rather.

     DR. BUSTA:  Other questions for clarification?  Dr. Fischer?

     DR. FISCHER:  Larry Fischer.  I had this thought, and I'm not sure you can answer the question.  But I imagine the food industry will shortly, if they're not already, be working very hard to learn how to produce food that is low in acrylamide, and certainly there are going to be products advertised with the acrylamide content.

     I'm wondering whether the Food and Drug Administration has any way to get information from the industry on how they are lowering the acrylamide levels in their products, which would give very good clues as to how we can solve the problem.  So, in other words, I think we ought to work--we said we're going to work on how we can lower the levels in food, but is there a way when industry is working on this that this information can be made public so that it can be used at home?  Because there are two ways to produce acrylamide:  either buy it or produce it at home.

     DR. BUSTA:  Can you answer that one?

     DR. CANADY:  Well, probably not, but I'll make an attempt.

     [Laughter.]

     DR. CANADY:  This is why when we talk about education, we talk about both consumers and processors, processing industry.  We recognize the need to share information across both groups, and we recognize also that this is a traditional--a product of traditional cooking, so that education across consumers is important.  And I notice somebody is walking up behind me.

     DR. TROXELL:  I would just add that the International Acrylamide Food Network, Information Network, is designed to provide, to facilitate the sharing of information from all stakeholders, including industry.  And, you know, from my discussions with the food processors, they're extremely interested in working very hard to find ways to reduce levels, and they're committed to releasing that information as soon as they have something that's concrete so that everybody can share in that achievement.

     DR. BUSTA:  That was Dr. Troxell.

     Dr. Kuzminski?

     DR. KUZMINSKI:  Thank you.  Larry Kuzminski.  I would like to come at the same question that Dr. Fischer asked, but from a different angle, perhaps.  I think it was partially answered by Dr. Troxell's comments, but the question that I had in the same direction was:  Can you give us some more detail about the nature of the consortia that you are establishing with your various partners in the research on the formation of acrylamide?

     DR. CANADY:  There's two speakers that are going to speak to that in a little more detail.  One is Dr. Lauren Jackson, who is at the National Center--at FDA within the National Center for Food Safety and Toxicology in Chicago.  The other is Dr. Dave Lineback who's going to speak later on today.  So a specific discussion of how the consortia work I think is maybe better served through those detailed discussions.

     DR. BUSTA:  Dr. Acholonu?

     DR. ACHOLONU:  Yes.  You spoke about the need for the FDA to develop education material on acrylamide.  What do you think is the appropriate time for doing this?  At this stage or later on when we get more information on it?

     Then the next question is:  Under the rubric of education, you seem to be advising that people eat balanced diets.  Will eating a balanced diet reduce the intake of acrylamide?  Why did you give that as an advice, to eat a balance diet?

     DR. CANADY:  Does anybody else want to handle this?

     [No response.]

     DR. CANADY:  Well, with regard to the question of when we provide educational information, we provide it when--we would provide it when we have enough information to help consumers understand the risk.  One of the things we've discovered recently, for example, is that there's a tremendous amount of variation across products, within the product class.  This is something that Dr. Musser is going to talk about in some detail.

     That helps us understand that processing--or cooking, traditional cooking, is quite important in the formation of acrylamide.  There may be other factors involved.  The point is we don't want to go out too soon with information that provides partial or perhaps misleading information in response to the situation.  So it's important to go out with information that has an intended effect of reducing the exposure to acrylamide, and prior to providing that education, we need to feel fairly confident that we have an understanding of the process.

     We need to provide educational information as we understand it enough to understand the effects of the information, and Terry is going to give a little more in response to that.  Maybe you can answer the second one, too.

     DR. TROXELL:  I was just going to say Dr. Acheson is going to talk about the multiple factors involved to be considered on this issue later on.  So we'll be getting more of that kind of information then.

     DR. BUSTA:  Joe, if this is urgent?

     DR. HOTCHKISS:  A response.

     DR. BUSTA:  Okay.  Go ahead.

     DR. HOTCHKISS:  Yes, Joe Hotchkiss.  I just want to recommend, being a veteran of mycotoxins and nitrosamines and a whole variety of things, and the IQs and all of this, that my recommendation, this is the time to go to the consumer.  I think the consumer is less concerned that you know all the information but, rather, the consumer's concern that you're concerned and you're taking action.  And that's what I think the consumer would like to know posthaste.

     DR. CANADY:  Thank you very much.

     DR. BUSTA:  That sounds like he's starting his comments for tomorrow.

     DR. CANADY:  Right.

     [Laughter.]

     DR. BUSTA:  With that, thank you very much.  We will take a 15-minute break, and we used up our time already.  We will start promptly at 10:20.

     [Recess.]

     DR. BUSTA:  Let us reconvene, please.  We now have the opportunity to hear from Dr. Canady again.  Am I pronouncing your name correctly?

x      DR. CANADY:  Yes, you are.

     DR. BUSTA:  The FDA Action Plan toxicology.

     DR. CANADY:  First I want to make sure everyone can hear me.  Lauren, Dr. Jackson, can you hear me?  You can hear me.  Could I have somebody running the slides?

     [Pause.]

     DR. BUSTA:  The technology event should really have occurred under analytical methods rather than toxicology.

     DR. GRAY:  Nothing ever goes wrong under toxicology.

     DR. BUSTA:  Right.

     DR. CANADY:  Although we're bringing a lot of advanced technology to bear on toxicology.

     DR. CANADY:  Well, you know me already.  Let's got to the next slide.

     My goal in this talk is to go through the toxicology component of FDA's Action Plan.  The overall goal of the toxicology component is to examine the likelihood that adverse health effects are caused by acrylamide-containing food, by exposure through acrylamide-containing foods.  I guess I don't need this in my face.  Next slide.

     Before I get into the details of the toxicology component of FDA's plan, I'd like to provide a little context.  And the context, as Dr. Schwetz and Joe Levitt referred to earlier this morning, is that this is not a new contaminant; this is not a new chemical under consideration.  There has been nearly 30 years, or perhaps more, of toxicology research with regard to acrylamide, and there have been a number of individual safety risk assessments that have been done over the years for different routes of exposure, different kinds of exposure to acrylamide.  And some of those are listed up here with regard to past assessments.

     What has changed--and you've heard this already, but I'll say it again.  What has changed is how we're exposed to acrylamide.  We're exposed through food.  We thought we were exposed through occupational exposures and, you know, to some degree through water exposure previously.  But we know we're exposed through food now.

     The second thing is that the duration of exposure and the length--or, rather, the length of exposure and the consistency of exposure is different than we thought it was before.  So we get exposure through a wide variety of foods, and we get exposure probably most of our life, if not all of our life, to acrylamide.  This is a new understanding that we have that influences our need for toxicology information and how we develop that information.

     Next slide, please.

     I'm going to talk a little bit about data gap evaluations.  Again, you've seen some of this information before.  The point I want to make is that data gap evaluations have gone on.  I'm going to talk about some of the data gaps and some of the data needs that we've identified.  The data needs that I will talk about should not be construed as the essential data needs, rather, as critical to progress with regard to describing risk characterization.  However, I'm giving them as a way of helping you understand where the major areas of uncertainty are with regard to risk characterization through toxicology research.

     Next slide, please.

     Here's the outline of what I'm going to present.  I'm going to start off with a discussion of toxicokinetics.  Then I'm going to go into three areas of toxicology:  first, carcinogenicity; second, neurotoxicology; and then, third, reproductive and developmental effects.

     Information that's not on this slide is that for each of the three in the middle, there should be a phrase "at high dose" injected.  Our understanding with regard to these three endpoints or these three areas of toxicology--as is fairly often the case for risk assessment, but certainly true for acrylamide--comes from information from high-dose exposures, and it's more so with some of the endpoints than others.

     Then the last point I want to talk about, the last aspect that I want to cover under the rubric of toxicology, is safety risk assessment, what's been done, where we are within the risk assessment paradigm at this point for acrylamide.

     Next slide, please.

     One very important aspect of the toxicology for acrylamide, as is the case for every time we look at toxicology, is the toxicokinetics.  Easy for me to say.  What we do know is that it's absorbed orally.  We don't know the degree of absorption from food, and the reason we don't know the degree of absorption from food is that most of our understanding for toxicokinetics and for toxicology comes from exposures through drinking water.  We do know that acrylamide is distributed widely throughout the body, and it's a fairly uniform distribution.  There are some areas that have higher levels than others at different developmental stages, for example, and also within organs in the body.  But, generally, it has a fairly uniform distribution.

     Metabolism has been studied and is to some degree fairly well understood for doses above 1,000 micrograms per kilogram a day.  And throughout this presentation, I'm going to keep the units equivalent so that whenever I give you a number with regard to a risk value, it's going to be in this unit of micrograms per kilogram of body weight per day.

     We know that acrylamide undergoes a saturable metabolism to glycinamide through a specific cytochrome P450, CYP2E1.  That conversion to glycinamide, again, is saturable, and it's something we have some fair amount of understanding about.  It's also conjugated through glutathione.  So right away, for those of you who understand susceptibility and so on, understanding polymorphisms for 2E1 and understanding polymorphisms for glutathione S transferase are fairly important things in understanding susceptibility variation across populations.

     We do know that acrylamide elimination occurs rather rapidly.  Acrylamide elimination occurs rather rapidly on the order of hours to days.  The thing to keep in mind with regard to this, though, is that it also forms stable adducts, with protein particularly and with DNA.  Those adducts can stay around for the life of the protein; hemoglobin, for example, through the life of the red blood cells.  So adducts remain in the body following exposure to acrylamide.

     Another important aspect--now, I'll get into this more with regard to neurotoxicity--is that the duration of exposure affects the level that's effective for neurotoxicity.  A fairly well-studied phenomenon, and we have a fair amount of understanding with regard to it.

     Next slide, please.

     Toxicokinetics data needs are really some of the most critical since this kind of information is going to allow us to bridge the gap between high-dose exposures that occur in animals and the much lower doses that occur through food exposures.  So it's one thing that we really want to have an understanding of as it applies to all the other areas of toxicology.

     Some of the data needs, as Dr. Schwetz mentioned this morning, include bioavailability of acrylamide through food.  We need to understand just generally dose response with regard to toxicity, disposition in the body, binding to various macromolecules.  In DNA and in hemoglobin, adduct relationships we understand, in other words, some more detail about the mechanisms of some of the toxicities and how that relates to external measures of exposure or to biomarkers of exposure.  And then we need to understand obviously toxicokinetics within humans to the degree that we can and relate that to toxicokinetics and animal data that we have some abundance of.

     Next slide, please.

     Our plans for toxicokinetic research include these three major--or, rather, plans for toxicokinetic research, and again, I want to stress that there's a lot of interest out there with regard to making progress in understanding the toxicity of acrylamide through foods so that we're drawing from a lot of different areas for the information.

     The FDA's National Center for Toxicology Research, as I'll talk about in some detail later, has nominated acrylamide and its metabolite glycinamide to a National Toxicology Program.  As part of the studies that are being developed for that long-term study, NCTR is looking into protein adduct relationships and DNA adduct relationships as they inform both exposure and toxicity.  NCTR is also going to be looking at bioavailability.  Again, it's a key component of our understanding of--or, rather, our bridging of the animal data to the food data.  Again, the animal data come largely from drinking water exposures, and, of course, we're exposed through food.

     CDC's National Center for Environmental Health is also, as I mentioned earlier, getting involved in terms of understanding the relationship between biomarkers of exposure and then potentially biomarkers of effect.

     Again, within the context of ongoing research, the acrylamide monomer industry has been interested in the toxicology of acrylamide, you know, as you might expect, for quite a long time.  So there's information, ongoing research with regard to toxicokinetics for acrylamide through that large body of research.

     Next slide, please.

     I want to talk about cancer, what we know about cancer with regard to exposures to acrylamide.  It's pretty clear that acrylamide causes cancer in animals.  There's two bioassays that provide information, two long-term bioassays that provide information, and a number of individual mechanistic studies that provide information.  What we don't know is whether acrylamide causes cancer in humans at the very low doses we see through foods.  The information with regard to animal bioassays is pretty clear.  It shows that it does occur.  Information with regard to epidemiology doesn't give us information one way or the other, doesn't give us stronger weight of evidence or weaker weight of evidence for carcinogenicity in humans.  And this is largely due to the power issues with regard to those studies.  There's really only a couple of studies that have looked for this endpoint, and they did not have enough power.

     Can you hear me with the competing noise?  Okay.  What do I need to do?  I should close the door.  Actually, it doesn't close.  I'll just try to get this closer to my--is that better?  I feel like I'm blowing away the front audience and not reaching the back.

     So, again, with regard to carcinogenicity, we have information from animals, but the information from epidemiology doesn't help us out at this point.  However, I do want to stress that every health body and government organization that has looked at the weight of evidence for cancer for acrylamide comes to the conclusion that it's a major concern.

     Next slide, please.

     I want to point out two areas of data needs for carcinogenicity research.  The first has to do with new studies.  As I mentioned, FDA has nominated acrylamide and its metabolite glycinamide to the National Toxicology Program.  That nomination is being decided.  We expect it to go forward.

     We also need, in addition to that, animal bioassay information which would bring to bear advances in our understanding of biomarkers and potentially bring to bear techniques that could allow us to understand low-dose extrapolations better, biomarkers of effect, biomarkers of exposure, and so on.

     In addition to that, if possible, epidemiology information would, of course, be very useful.  It's not likely we're going to find populations that are occupationally highly exposed to acrylamide at this point, at least within the U.S.  So it's not clear whether we can move forward with regard to epidemiology at this point.  But if those populations could be identified in some way, that information would obviously be useful.

     There's two areas of research with regard to existing information that may help us make progress in the shorter term.  As I mentioned, there are existing bioassays out there, animal bioassays out there.  Looking back over that information using more recent diagnostic criteria may help us understand a little bit more about the dose response.  And, also, because in those bioassays thyroid tumors were identified and we have a fairly well-developed process for understanding thyroid tumors, evaluation of that endpoint through IARC protocols or through recent EPA protocols would be a way of understanding that endpoint a little better and applying that information to the human situation.

     Next slide, please.

     Our study plans, again, include nomination to the NTP.  I mentioned this a few times.  I won't go into it in any detail.  FDA, Center for Food Safety and Applied Nutrition, is going to participate fully within the development of those plans.

     Next slide, please.

     The mechanistic studies, again, are an important part of that study because they allow us, again, to extrapolate down to the low doses we expect through food.  As with all bioassays, we're going to have to use fairly high doses, obviously, within the shorter lifetime of the animals to understand the tumor response.  Understanding the toxicokinetics running from the milligram per kilogram dose range to the microgram per kilogram dose range is obviously essential to making decisions for food exposures.  And we have tools through the biomarkers--we've already made progress with regard to DNA adducts, for example.  We have tools that could allow us to bridge that gap and may help us make less uncertain determinations of the risk at low dose.

     Next slide, please.

     Neurotoxicity.  As Dr. Schwetz mentioned earlier today, neurotoxicity is a known effect of acrylamide exposures.  It's known for exposures through occupational settings.  Dr. Schwetz mentioned grout applications.  There's also been other industrial applications where this endpoint has been studied.  We don't know whether acrylamide neurotoxicity happens through food exposures.

     The exposures that caused it in humans were quite a bit higher than those that we see through food.  This is one of the reasons that the WHO/FAO consultation came to the conclusion that focusing on carcinogenicity and germ cell toxicity was an appropriate conclusion at that time.

     Neurotoxicity has been studied widely, not just in humans.  It's been studied across various animal models.  So we have a fair amount of information.  Again, at very high doses with regard to the toxicity, neurotoxicity in animals.

     We know that cumulative dose is important.  We know that if you are exposed for a longer period of time, the amount of acrylamide needed is less.  It's not a strictly additive relationship.  In fact, there's been a fairly detailed study that's looked at this relationship, this Haber's Law relationship.

     One thing we also have a developing understanding of is that there are effects to younger animals versus older animals.  We do not have information that allows us to say that neuro-developmental effects can be ruled out for acrylamide.  And we have information that suggests that younger animals are more susceptible than older animals.

     Next slide, please.

     So based on that information--and I'm going to go into the safety assessment and talk about what studies are available to understand neurotoxicity for acrylamide.  And when we talk about that, I'll show you that the study that we used for safety assessments is a 90-day study, not a chronic study.  Because of that availability of information, one important piece of data that we need to uncover is further understanding of the relationship between dose and duration.  And, again, because we don't have a clear understanding of neuro-developmental effects for acrylamide, and given our newer understanding of the continuous exposure to acrylamide, we need to have an improved weight of evidence for neuro-developmental effects.

     Next slide, please.

     Plans for neurotoxicity within FDA are still at the early developmental stages, the early planning stages.  We're considering including neurotoxicity endpoints in the NTP study.  We're also considering other ways of developing neurotoxicity, and particularly neuro-developmental information.

     There is quite a lot of ongoing academic research with regard to neurotoxicity, and that's something we'll obviously draw from.  And, hopefully, given this new interest in exposures through food, we'll get more information with regard to low-dose effects and developmental effects through academics.

     Next slide, please.

     A somewhat less certain but still concerning area for acrylamide toxicity is the developmental reproductive effects.  Dr. Schwetz alluded to or talked about, rather, in his presentation germ cell toxicity.  This is a rather unusual finding for chemicals.  It's not something that we run across every day.  So finding that acrylamide has germ cell toxicity is something that we need to consider carefully.

     The evidence that leads us to the conclusion that acrylamide can cause germ cell toxicity comes from very, very high-dose exposures, and this is higher than the neurotoxicity doses that were used, and clearly higher than the studies in the bioassay that showed carcinogenicity.  We're talking about 40 milligrams per kilogram exposures and through intraperitoneal injection for that matter, too.  So this information, while it is concerning in its unusual nature, still needs to be taken within the context of very, very high doses that have been used to show this relationship.

     Other developmental effects are that there are clear reduced litter sizes, probably related to dominant lethal effects, but those have been shown.  But I do want to point out that there have been no structural malformations that have been shown in the developmental studies that have been done.  This is not something that would come across as a teratogen through those studies.

     Next slide, please.

     Reproductive and developmental data needs, again, this is to try to give you a context for what general data needs or general areas of uncertainty or areas of information need are, not to provide specific crucial data needs for development of decisions.  But we need epidemiology for germ cell toxicity.  This information could be derived through existing cohorts, perhaps, and it's something we're exploring through NIOSH and other opportunities.

     Bringing advanced techniques with regard to genotoxic effects is something that was mentioned at the WHO/FAO consultation and at the JIFSAN consultation or JIFSAN workshop.

     We need particularly a better understanding of the mutations generated and the dose response for mutations that are caused in germ cells.  Again, because they're very high-dose exposures and typical routes are not routes that are directly related to food exposures, we need information with regard to low-dose exposures, with regard to adducts, with regard to dose-response relationships, obviously.  So mechanistic studies are quite important for this because of the conversion of acrylamide to glycinamide through P450.  We don't have a clear mechanistic understanding of whether it's glycinamide or acrylamide that causes the germ cell mutations.  Having that understanding would help us obviously understand better how to do low-dose extrapolation.

     Next slide, please.

     Plans for reproductive developmental effects study.  Again, FDA, we're in the early stages of determining how we might work this into, for example, the NTP study or in terms of developing other research programs for this endpoint.  One of the data needs that I didn't go into in the previous slide was that because the finding of germ cell toxicity is a relatively unusual event, we don't have straightforward--we don't have methods for evaluating the risks or making decisions for risks that are routinely used.  So one key aspect of understanding this endpoint is trying to figure out how we would understand the risk to low-dose exposures, how we would extrapolate risks, how we would make decisions using risk characterization for this endpoint.

     So for that reason, we have planned to develop a workshop for germ cell mutagenicity, risk assessment, where we can help push that area of science forward a little better and help us understand how we might make decisions for this endpoint.

     NIOSH has a protocol that they put together already for worker studies.  They're collaborating with NIEHS to try to incorporate some reproductive endpoints within that study.  NIEHS is also looking at some mechanistic things.  They're looking at these knockout mice that don't have the enzyme that allows the conversion from acrylamide to glycinamide.  Looking at the effects in those mice will help us understand the mechanism, obviously, and help us understand low-dose extrapolation a little better.

     Finally, NTP's Center for Evaluation of Risks to Human Reproduction is considering doing a detailed review of the reproductive effects of acrylamide.

     Next slide, please.

     Safety risk assessment.  I'm going to talk about safety risk assessment in several slides, several following slides, and the idea is to give you an understanding of where we are in the process and also give you an understanding of the data that underlies that evaluation.

     Both FDA and EPA have used the same study, Burek et al., 1980., to develop safety risk comparison values.  Safety risk comparison values are used as an initial part of the overall risk assessment process.  If you do a safety risk evaluation and you find that the doses that you're exposed to are lower than the safety risk evaluation number, you have an early understanding that there's not a need for further evaluation, that you've done enough evaluation.  If, on the other hand, the value is higher, the exposure value is higher than your safety risk comparison number, then you have an indication that you need to do further analysis.  We're at that stage with regard to acrylamide.

     For the EPA and FDA safety risk comparison values, we both use the Burek, et al., 1980 study, which is a 90-day study in rats, again.  The lowest effect level was 1,000 micrograms per kilogram, and the no-observed-effect level was 200.  Only three animals were used to develop that dose response per dose value, and they saw recovery at 144 days within that group of animals that showed the effects of the lowest effective dose.

     Both EPA and FDA in this initial determination of safety risk evaluation numbers used thousand-fold uncertainty factors, and I should mention that EPA is currently in the process of re-evaluating their IRIS (?) file for acrylamide.

     Next slide, please.

     As Dr. Schwetz mentioned, WHO/FAO did do safety risk evaluation to some degree, or at least they considered the evidence for toxicity and the evidence for exposure and made initial evaluations.  Their judgment at that time was that non-cancer effects were unlikely at doses from food.  Dr. Schwetz mentioned a NOAEL of 500 microgram per kilogram per day.  The Burek, et al. study has a NOAEL of 200 micrograms per kilogram a day.  Different data sets were used in the evaluations, and I can go into that in some more detail if you'd like to in questions, but the point is that the effective dose remains at around 1,000 micrograms per kilogram a day, and the no-effective dose is something that is driven largely by the dose studies in various studies--the dose choices in various studies.

     DR. KUZMINSKI:  Excuse me.  I just want to clarify your first bullet that's a double negative, non-cancer effects judged unlikely, so does that mean cancer effects are judged likely?

     DR. CANADY:  No.  This is a peculiarity of our way of dividing the toxicology world.  We divide into cancer and non-cancer.

     DR. KUZMINSKI:  Okay.

     DR. CANADY:  So another way of saying this is that effects other than cancer were judged unlikely, and then the next part of the slide talks about what the WHO/FAO came to with their other conclusions for cancer.

     As Dr. Schwetz mentioned, again, there was no consensus on quantification of cancer risk within that consultation.  But they did come to a qualitative determination.  We compared carcinogenic potency of acrylamide using sort of a two-step process.  We talked about relative potency with regard to benzopyrene and heterocyclic aromatic amines, for example.  There's already been some discussion about this, and we can go into that in some more detail if you'd like.

     Then they also used the aspect of intake level to give a relative sense of the cancer--the level of concern with regard to cancer for acrylamide.

     Next slide.

     And then their conclusions were--rather, our conclusions were that there's a major concern for acrylamide based on relative cancer potency and uncertainty regarding the germ cell mutagenicity finding.  So they focused on both of the endpoints of cancer and germ cell mutagenicity.

     Next slide, please.

     I want to go through the safety risk assessment values in a little more detail.  There's three groups of information here.  The first group of information is the dose response information, which I've already gone through in some detail.  The second is the exposure, the order of magnitude exposure estimate that WHO/FAO came up with, and that is something around a microgram per kilogram of body weight per day.  And then the derivation of FDA's acceptable daily intake, which used a thousand-fold uncertainty factor based on the 200 microgram per kilogram per day no-effect level to come up with a value of 0.2 micrograms per kilogram per day.

     I'm giving you this information really just to try to orient you to where we are, again, in the process of risk assessment.  If you do the comparison of dose to the comparison of safety risk evaluation number, you come to the conclusion that risk analysis is needed to make decisions.

     I think I'll leave further discussion of this to questions, which I'm anticipating there may be some questions.

     Can I go to the last slide, please?

     In summary, acrylamide causes effects in animals, and those effects are seen at quite a lot higher doses than we see through food.  It also causes effects in humans, and, again, those have been seen at quite a lot higher doses than we see in humans.  However, even though exposures that have been shown to cause effect are quite a lot higher than those we see in food exposures, safety risk assessment leads us to the conclusion that we need to do further risk analysis.

     There are substantial gaps in our knowledge about the toxicology and even at this point the exposures to acrylamide.  So we need to make further progress on that.

     Data gap analyses by a variety of expert panels have led us to the conclusions that I've gone through in some detail earlier.  But they basically give us an understanding that we need more information, and there are specific kinds of information that would be more helpful than others.

     Finally, FDA has initiated a program to try to develop some of this information, and I'd be happy to talk about that in more detail.

     Thanks very much.

     DR. BUSTA:  Thank you very much.

     Questions of clarification?  Dr. Fischer?

     DR. FISCHER:  I have a question about the use of the cancer bioassay and regulation of acrylamide in drinking water.  Does the EPA regulate the drinking water level of acrylamide based upon the cancer bioassay or do they use non-cancer effects?

     DR. CANADY:  I feel a little uncomfortable talking about EPA programs.  My understanding of it is that it's a technology-based guidance level for water.  It's based on the available technology and the demonstration of its ability to keep acrylamide out of the water.  Maybe--I don't know.  Henry, do you want to respond to this?

     DR. KIM:  I don't have that information offhand.

     DR. CANADY:  Yes, and, again, responding to EPA programs I think--I don't feel comfortable actually responding to that at this point.

     DR. FISCHER:  Well, I went to IRIS, you know, and tried to decide from IRIS what they were doing, whether they were regulating on a non-cancer or cancer effect, and couldn't really spend enough time to get it out of there.  I wasn't clear from looking at the IRIS thing.  So I guess from your answer it sounds to me like that the FDA doesn't want to start at that point, that is, using the drinking water standard procedure for calculating risks to acrylamide exposure.

     DR. CANADY:  Well, one of the initial data needs before using that information is an understanding of the bioavailability through food.  We have an understanding of the bioavailability through water.  Before we would move to using information like that, one of the first steps, obviously, is understanding that there is a difference.  Also the pervasiveness of exposure and the continuousness of exposure through food is a different aspect with regard to those exposures compared to the drinking water exposure, and that information would need to be taken into account, too.

     DR. BUSTA:  Dr. Hotchkiss?

     DR. HOTCHKISS:  I had almost exactly the same question as Dr. Fischer, but with a different acronym.  Acrylamide, as we've discussed the range of carcinogens that occur in foods, some through a result of processing, acrylamide is quite unusual in that there is also a significant occupational exposure and a history of that exposure.  I would presume that OSHA, like EPA, regulates acrylamide exposure in the environment of the workplace.  I think you briefly mentioned something about NIOSH in there, but I wonder what history is to be learned, at least as a starting point, similar to EPA in water, from OSHA's regulations and on the basis for which they've made those regulations.  And I understand your answer is going to be that that's inhalation compared to foods.

     DR. CANADY:  Right.

     DR. HOTCHKISS:  And we don't know about foods.

     On the other hand, I heard earlier that we made this kind of thumbnail sketch about exposure from foods, and you were comfortable with the working group, I believe it was 0.8 from foods, thumbnail sketch kind of thing.  It seems to me that--I would presume that OSHA has put some effort into their regulatory limits as has EPA, and I wonder if those have been taken into consideration.

     DR. CANADY:  Well, the easiest answer to that is that we intend to take all information into account as we develop our risk assessment.  OSHA does have a value for workplace exposures to acrylamide; I believe it's a TLV and I believe it's 0.3 milligrams per meter cubed.  If there's anybody that--what is it?

     DR. FRIEDMAN:  0.3.

     DR. CANADY:  It's 0.3.  And that's the TLV?

     DR. FRIEDMAN:  That's BEL.

     DR. CANADY:  BEL.

     [Inaudible comments of microphone.]

     DR. CANADY:  Sorry, Mr. Chairman.

     DR. BUSTA:  Yes, if you could--

     DR. FRIEDMAN:  I'm Marvin Friedman.

     DR. BUSTA:  Could you give us the answer/

     DR. FRIEDMAN:  For the issue of--

     DR. BUSTA:  Would the speaker--

     DR. FRIEDMAN:  I'm Marvin Friedman.  I'm from the chemical--

     DR. CANADY:  I'm sorry, Mr. Chairman.  I've opened this meeting up, and I didn't intend to.

     DR. BUSTA:  Yes, Dr. Friedman, would you go to the--if it's all right with the speaker, come to a microphone so we can record your comments, please?

     DR. FRIEDMAN:  First, I want to apologize.  I just sort of answered the question and didn't mean to use your time.

     With reference to OSHA, OSHA's permissible exposure limit is 0.3 milligrams per cubic meter, five days a week over eight-hour shifts.  You can calculate what that means in terms of individuals' exposure.

     We've run the math on that, and at 2.3 years of working at that, it's equivalent to a lifetime's worth of eating whatever the--what is it?--18 micrograms per every day?  That's the way it works out.

     With reference to EPA, EPA's regulation of drinking water, EPA has two levels of regulation, what they call an MCLG and an MCL, the contaminant level and the contaminant level goal.  All carcinogens have a contaminant level goal of zero.  So, you know, it doesn't--risk assessment doesn't play in.

     As far as the non-risk number, acrylamide gets into drinking water from polymers, and the EPA regulates the amount of acrylamide in the polymer.  And, therefore, that will turn around and--because the amount of polymer added to drinking water is regulated, it will regulate what's added to drinking water.  And that's done virtually on feasibility.  The number comes out to be vanishingly small because polymer is used at a ppm level.  But that's how EPA does it.

     My apologies.  I really didn't mean to get into this.

     DR. CANADY:  No, that's fine.  Thanks.

     DR. BUSTA:  It's for information, and we've got the numbers down so we can be sure that you're correct by checking them later.

     DR. CANADY:  Yes.  But, again, what we intend to do is use all available information, and prior analyses will, of course, be important in helping us understanding the approach that we might take.

     DR. BUSTA:  Dr. Gray, please?

     DR. GRAY:  I'd like to clarify something that just wasn't clear to me when we talked about all the studies that are ongoing, and it's frankly because I share Dr. Schwetz's concern about the use of biomarkers and also know that biomarker information is going to be coming out for public consumption in the not too distant future.  That is, are we being careful to get our dose-response information in with the biomarkers?  For example, in the NTP bioassay, I could imagine that we could do biomarker measurements both in--we could do blood measurements, we could do DNA measurements, we could do target tumor measurements, in order to somehow be able to put in context, even if it's only in the animal studies, the dose-response information back to biomarkers so that we actually have dose response for the endpoints we're looking at with biomarkers.

     DR. CANADY:  Yes, thanks for that question.  There was some text that I skipped over in my talk that went directly to that, and the point is that NCTR in their nomination to the NTP intended to and will include mechanistic studies that will help do that extrapolation from external dose to hopefully markers of effect.  So there's a range of understanding of dose and effect that extends through our action plan, from CDC's evaluation of external exposure and biomarkers of exposure, through to NCTR's understanding of DNA adducts, biomarkers of exposures that inform us with regard to toxicokinetics, and then hopefully the relation of one of those markers in some quantitative, to the degree possible, way to the effects that are seen, as you say through tumor incidence or markers in tumors.

     So the intention is to bring to bear this very powerful set of tools of understanding dose through external, through biomarkers of exposure to hemoglobin, biomarkers of exposure/effect, if you would, to DNA, adducts to DNA, adducts to other significant nuclear proteins, for example.  So, yes, the understanding of mechanism and how that informs low-dose extrapolation is one of the main reasons for going forward with a bioassay.  You can bring to bear all that information.

     DR. GRAY:  Right, because I'm just concerned that right at the level of as we do these tests we get that information if we can.  For example, I could imagine--it may not be true, but I could imagine, for example, that these biomarkers we're measuring are actually sinks.  They're taking some out.  And if we look at target tissues or something, there may not be anything there.

     DR. CANADY:  Absolutely.

     DR. GRAY:  We need to understand the way in which these work.

     DR. CANADY:  So a key component--

     DR. GRAY:  And get that by linking it very tightly to the endpoints that we think we care about.

     DR. CANADY:  Right.  So another way of phrasing that is we need to understand the dose response of the relationship between hemoglobin biomarkers and biomarkers of effect or measures of effects.

     DR. GRAY:  Measures of effect.

     DR. CANADY:  Right.  Either one.  One may be a good correlate of the other.

     DR. BUSTA:  Dr. Downer?

     DR. DOWNER:  With respect to safety risk assessment, given the fact that research is ongoing and that we still lack concrete knowledge of bioavailability, metabolism, for example, of acrylamide, how did the FDA quantify this marker of 0.2 micrograms per kilogram body weight for the daily acceptable intake?

     DR. CANADY:  These are fairly--

     DR. DOWNER:  This is going back to Dr. Fischer's question of the 0.8.

     DR. CANADY:  These are fairly standardized approaches for safety assessment.  And, again, they're ways of understanding--of gaining an early understanding of the toxicology information that allows you to compare it at an early stage in risk assessment to what you understand about exposure.  And we fully understand that our estimates of exposure are early and just that, they're estimates.

     In the same sense, we understand that our estimates of toxicology or the dose response are at an early stage.  And so comparing them at this point in the risk assessment gives us a sense of sort of order of magnitude comparisons.  It's not an exact comparison that allows us to say that this effect will occur.  All the regulatory agencies that use these technologies--or use these techniques, rather, to estimate safety risk agree that the excedence of these safety risk values does not mean that effects would occur.  So the derivation--to get to your question a little more directly--of these safety risk values, the ADI in the case of FDA, is a relatively routine process that allows us to take an early look across a lot of different compounds at what the relative risk is, and also allows us, again, at an early stage to compare that to what we think may be occurring through whatever exposure route it's occurring.

     But I just want to stress that it's an early look.

     DR. DOWNER:  It's essentially a proxy indicator, then.

     DR. CANADY:  I'm not sure what you mean by proxy.

     DR. DOWNER:  If the levels are this, then possibly this is what we're looking at.

     DR. CANADY:  Yeah, but it's--I understand what you're saying.  It's an early decision point.

     DR. DOWNER:  Right.

     DR. CANADY:  And it's a permissive decision point.  If you find that the exposures are much less than your toxicity value at that early stage, then you, depending on your uncertainty, have the option of saying there is no need to go further with analysis.

     DR. BUSTA:  Dr. Lee?

     DR. LEE:  Dr. Canady, I appreciate your two very comprehensive presentations, but speaking of estimates, I want to kind of get your perspective.  Certainly there's a bit of guesswork in food risk.  We've got a guesstimate from CDC of about 5,000 deaths per year from foodborne illness.  We've got a guesstimate of about 540,000 deaths per year from cancer from NCI.  In your perspective, can you guess as to how many cancers, if any, we might see per year in the United States as a result of acrylamide exposure?

     DR. CANADY:  No.

     DR. LEE:  Do you think that is something that we can eventually work towards?  Is that something is immeasurable?

     DR. CANADY:  Maybe other members of the committee would like to respond to this as well, but within this risk assessment paradigm, what we're talking about for risks from a chemical are relative risks.  They are not actuarial risks.  We are not talking about estimating the number of people that will die.  We're talking about comparing, for example, the relative risk of one chemical that causes cancer to another chemical that--one chemical that may cause cancer to another chemical that may cause cancer.  There's a way of prioritizing our efforts and making decisions about what to do.  It should in no way be taken to mean that we know how many cancers will be caught.

     DR. LEE:  Well, maybe there's a different way I can put this.  If you think we could like magically, say within a year's time, eliminate all acrylamide in food, would that have an effect on cancer incidence?

     DR. CANADY:  I think this is an opportunity to take advantage of understanding of potential risk and potential health effects to see whether we can remove this potential risk from the food supply.  And that's really the best way to approach that question.

     DR. LEE:  Thank you.

     DR. BUSTA:  Dr. Hotchkiss?

     DR. HOTCHKISS:  I was a little surprised.  We were provided--and I know something exists which I believe is in-house at FDA--the Quantitative Risk Assessment Committee reviewed the then current knowledge of acrylamide in 1998, and then calculated some specific unit risks in that document, and--which I was surprised you didn't mention.  Then let me refresh everybody's memory.  I'm sure you all stayed up last night reading this document.

     The conclusion was that the Risk Assessment Committee found the dose-response curves plotted for the statistically significant tumors to be quite flat, which could indicate that acrylamide is not a particularly potent carcinogen under the conditions of the assay in the Fisher rat.  And I'm just wondering--my impression is that your impression is that this risk assessment done by FDA four or five years ago was not addressed by you, and I just wonder whether there was a particular reason.  Did you feel that the information was not sufficient given what we now are beginning to learn about acrylamide or what?

     DR. CANADY:  I did not go into that detail.  As Dr. Troxell mentioned earlier, we agree with the conclusions of WHO/FAO that this is a major concern, at least partly on the basis of the cancer response.  But providing a quantitative estimate of the relative risk at this point, we have not gone to that level of analysis, partly because we need a better understanding of the exposures, but also because we see an opportunity to understand the dose response better.

     We're talking about a contaminant that's fairly widely distributed in the food supply.  We need to make decisions in a careful way.  The dose response assessment that you're talking about was very useful for making the food contact material decisions that were made at that time.  In that case, using that slope or using that dose-response information, we were able to come to the conclusion that the food contact uses that were proposed were fine.  They would not provide a risk that was above our criteria, de minimis criteria.

     The question becomes:  Can that dose-response information, which was adequate for that use, be considered adequate for the current situation?  And that's a question we're addressing right now.  So I'm sorry if I didn't get into it in detail, but, you know, I don't mean to--what I mean to do is say that we are considering all dose-response information and considering, again, opportunities to improve our understanding of low-dose extrapolation using biomarkers of exposure and using biomarkers of--using actually effect measures to get to a better understanding of the low-dose risk.

     DR. HOTCHKISS:  Let me understand.  It wasn't a matter of detail.  I didn't hear you mention it anywhere.

     DR. CANADY:  Right.

     DR. HOTCHKISS:  So that's a little bit--

     DR. CANADY:  I didn't.

     DR. HOTCHKISS:  We're interpreting that a little bit different.  Let me also try to clarify my understanding that this has nothing to do with exposure, whether it's from food contact or anything.  This is a risk assessment, a unit of risk calculated based on the then available toxicology data that will not change given that the exposure changes.  It's really unrelated to exposure.  Am I right?

     DR. CANADY:  That's right.  The dose-response analysis we could do using the existing information.

     DR. HOTCHKISS:  We might generate better dose-response data that will change this, but we're not going to--that's not going to be related to dose.  And so I just wonder, this is--

     DR. CANADY:  Right.  No, I' sorry.

     DR. HOTCHKISS:  --the most current thinking on risk assessment from the agency, and I just wondered whether you found inadequacies in this or you're just not considering it or moving forward from it or dismissing it.

     DR. CANADY:  No, we're moving forward and using that information as well as other information we've collected since then, sure.

     DR. BUSTA:  Dr. Gray?

     DR. GRAY:  Just quickly, I think toxicokinetic data could make a big difference, because the absorption, of course, the amount of material that gets in the body can be important to dose response.  So there are things that could change, but that was a very thorough evaluation that I'm glad you brought to our attention.

     The second thing is just a very quick question on this same dose-response issue.  Has anyone calculated under different potential dose-response models the upper-bound risk that would be consistent with the epidemiologic studies at this point that find nothing so that we can even know if this is--how sharp a line to the epi studies draw for us in knowing--

     DR. CANADY:  Are you talking about power calculations and so on?

     DR. GRAY:  Not just power calculations, but then what is the--yeah, it's based on power calculations.  What's the sort of maximum relative risk that could be there and could not have been detected under certain assumptions about the shape of the dose response?

     DR. CANADY:  During several of the data needs meetings that we've had, this question has come up, and our understanding is or my understanding from those discussions was that the power of the studies is not informative as to what an upper-bound risk might be.

     So if I can understand your question a little better, what I think you're saying is:  Can we derive an upper-bound risk using the epidemiology data?

     DR. GRAY:  No.  What I'm saying is the epi studies that find nothing--they find nothing, and it may be that they're zero, or it may be that they were not powerful enough to detect a very small increase.  And you can calculate what that very small increase could be, which would just say is that something that's even in the ballpark of where we are now.  Is it helping us understand that these risks could be lower than the animal data might be telling us?

     DR. CANADY:  And the epi studies were on the order of several thousand people and the exposed were quite a bit fewer than that.  So I guess the way to approach this is the slopes that have been discussed, both FDA's and EPA's, are not ruled out by the epi data.  That's about as direct as I can be.

     DR. BUSTA:  Unless there is an urgent, burning question for clarification, we will move on.  Thank you very much.

     DR. CANADY:  Sure.

     DR. BUSTA:  The next presentation of the FDA Action Plan is on analytical methods/occurrence by Dr. Steve Musser.

x      DR. MUSSER:  Good morning.  It's my group's responsibility to develop the analytical assay for analyzing acrylamide in foods as well as conducting our exploratory survey of acrylamide levels in U.S. foods.  It makes me very happy to present this data.  We've done what I consider to be an enormous amount of work in a very short period of time.  We generated a lot of data.  And we still have a long way to go.  We're probably about halfway through our goal of 600 different types of foods, but we have done this in a very, very rapid time frame.

     May I have the next slide?

     I'm going to talk about two topics today.  The first part of the presentation will deal with the analytical methodology itself, why we chose that particular analytical methodology, how we validated it, why we believe it's accurate, why we think this particular methodology gives us good results, and over a broad range of foods.  And then the final part of the presentation will deal with the actual levels that we're finding in our exploratory survey in various food groups.

     In a recent meeting sponsored by JIFSAN in Chicago, there was an analytical working group, and at that time we went through a number of methods that were used for detecting acrylamide in foods.  Of them, there were four methods that came out kind of as most often used or most frequently used by people measuring acrylamide in foods.  They are gas chromatography/mass spectrometry, GC/MS, either derivatizing the acrylamide or looking at it directly as an underivatized compound; liquid chromatography/mass spectrometry; and liquid chromatography/tandem mass spectrometry.

     In all cases, the specificity was fairly high, and the limit of quantitation was well within the lower levels of acrylamide that we're finding in foods.  I'll also talk a little bit more about these methods because I think it's important when we compare data from different groups that we have some confidence that independence of the method that was used, we are still getting reliable results from other laboratories.

     We chose to use the liquid chromatography/tandem mass spectrometry method, LC/MS/MS.  The reason we did this when the Swedish administration first announced their findings, we knew their methodology was based on this particular type of methodology, and we wanted to confirm their results as accurately and introduce as few variables as possible in confirming those initial results.  Subsequently, we developed our own methodology in the absence of having any other thing to work with, and that method that we're consistently using now is LC/MS/MS.

     Just a couple of method highlights, and these highlights are issues that have been brought to me either through some of the meetings, public meetings that we've had or personal phone calls, questions about the method, or interest in why we chose a particular way of doing the analysis.

     First of all, sample size, we used a 1-gram sample size.  We homogenized a serving, so, for example, a single bag of chips is about an ounce or 28 grams of material, so we would homogenize that serving.  Or if we were looking at cereal, again, we'd have a box of cereal and the serving size is an ounce.  So we'd take about 28 grams, grind it up, homogenize it, and then take a 1-gram subsample of that material and do the analysis.

     Stable isotope.  We used a stable isotope-labeled internal standard, carbon 13.  We have three carbon 13s replaced the natural carbon 12s that would be found in acrylamide.  This is a stable isotope dilution method.  We use this because our internal standard is identical to native acrylamide with the exception that it differs in mass by 3 atomic mass units.  So if we're getting quantitative recovery from the matrix, then the internal standard such as this controls for all other recovery and losses that might occur through processing, because we're simply looking at a ratio of the internal standard to the native acrylamide.

     We extract with water.  A number of groups look at hot water, various organic solvents.  We found that there is no difference using hot water, cold water, and that our recoveries are just as good by shaking it a few minutes with cold water, though there is a volume effect, and you have to have a sufficient volume, about a 10:1 ratio, between the particular matrix you're looking at and the amount of water you use for the extraction.

     We use a two-step solid base clean-up.  When we initially came out with our methodology and published it on the Web in June, we had a single-step clean-up.  We found that there were a number of particular food matrices where we had to look at much lower levels, and we needed an additional clean-up step to get rid of some interferences in some matrix to look at approximately the 10 ppb level.

     Chromatography, we control the column temperature.  That gives us very reproducible retention times, so, for example, we're in kind of a cold room today.  If it were in the summertime, that temperature might vary by five degrees and retention time would shift.  All of this goes to we want very precise, reproducible, accurate results that we can rely on from day to day to day to day.  It's all part of the validation.

     We've looked at about three dozen different kinds of HPLC columns, evaluated them all for their performance.  We found three that give us acceptable results.  We're currently using one of them, and I'll talk to you a little bit about that next.

     Our method of detection, as I said, is liquid chromatograph/tandem mass spectrometry.  We use Electrospray as the ionization method.  Other groups use atmospheric pressure chemical ionization as their ionization method.  It depends on the instrument manufacturer.  Our particularly manufacturer gives us better results with Electrospray.  Other groups get better results, lower limits of detection with APCI.  It's really not a dependent variable that makes a significant amount of difference in the actual analysis.

     Next slide.

     We were interested in using a smaller subsample, 1 gram of subsample, primarily to reduce our waste stream.  We knew that we were going to be running many, many hundreds of samples, and if we were looking at using a 10-gram sample size, that meant 100 mLs of waste, larger sample vials, larger containers.  Right now we've done about 700 individual analyses, probably a little more.  That equates to almost 700 liters of waste that we would have been generating.  We wanted to try and minimize that while still maintaining accurate levels and accurate measurements.

     What I'm showing here are some results where we've looked at four different matrices--bread crumbs, cereal, coffee, and potato chips--where we've taken 1-gram subsamples, and this is of a giant homogenate where we've taken kilogram amounts, homogenized it, done a number of analyses on that, and then taken 1-gram and 4-gram subsamples of those particular matrices.  And what we find is that quantitatively the results are identical whether we use 1-gram or 4-gram subsamples.

     We're also going to expand this to 10 grams just to make sure that we're not missing anything obvious, but the preliminary results indicate that there is really no difference as long as the sample is well homogenized between taking a 1-gram subsample and a 4-gram subsample.

     I mentioned about the two-step solid phase extraction.  This particular slide--in the top slide where we have only one solid phase extraction step, this is for a food that has about a 20 ppb level of acrylamide down closing in our own quantitation limit.  This is the acrylamide peak right here, and actually, it's a bit obscured by this giant contaminant that's in the matrix.  Using the second SPE step, we've been able to completely eliminate that, and we get a much cleaner, tighter, well-defined chromatographic peak that gives us better precision in our measurements.

     Next slide, please.

     This is just an example of the kind of--what the peaks look like on the different types of columns that we found acceptable.  Currently we're using this Phenomenex-Hydro RP.  It's a high-carbon-loaded C-18 column.  It gives us very nice peak shape, and for the analytical chemists in the group, sensitivity, of course, is related to how much--how fast and narrow the peak you can get into the detector.  The top one gives us that best peak shape, and we've just replaced this particular column after doing about 600 analyses, individual analyses.  So we know that it performs well over a wide variety of matrices and a large number of samples.

     Quality control, of course, is an important factor in doing these types of analyses where we're looking at a broad range of food products.  The most important issue for us is recovery of the internal standard.  So if we put the internal standard in, because we're basically putting acrylamide into our sample, if we don't get acrylamide back or we have very small peaks for the stable isotope-labeled internal standard, we know we've got a problem with that particular matrix or that we made a mistake during the extraction of the clean-up.  And so recovery of the internal standard and the peak shape for the internal standard is a critical factor for one of our quality control points.

     We do a duplicate analysis of every food sample.  That means not just a re-analysis of the extract, but we actually take another sample of that food and take it through the entire sample process.  And what that does for us is allow us to compare--if we had some odd problem that happened to occur where one of the samples was particularly high, we like to see them both very close together.

     Re-analysis.  We also do periodically re-analysis of samples that have been analyzed previously to see whether they fall close to those samples that we've already looked at.  We look at the performance of standards every day before doing analyses.  And we also look at the signal-to-noise and ion abundance ratios.  So in the tandem mass spectrometry, we're producing not only the parent molecular ion species but fragments, and we look at the relative ratios to each other to make sure that they are within the expected values, and that gives us just another way of determining the specificity and accuracy of the measurement.

     Next slide.

     Some more on method performance.  Linearity, of course, is important.  We use a range from eight parts per billion to 3,200 parts per billion.  That's the limits of our standard curve that we've developed.  More importantly, the response factor is very close to unity across from the very low levels to the very high levels.  And that's a very good indication of the performance of the method, at least with standards.

     Recovery.  This is actually a fairly difficult point to get at.  We used the method of standard additions to calculate recovery.  We feel that that's the most accurate means of determining recovery percentages.  We've done it for a large number--about ten different matrices, as well as proficiency testing samples.  And in all cases, it's generally greater than 90 percent recovery, and usually a little better than that, around 95 percent recovery, for the matrices where we've actually done the standard additions experiments.

     We've done some preliminary experiments with spiking the matrix with known amounts and get very similar recoveries, although we feel the method of standard additions gives us a better estimate of recovery.

     Precision.  These are well-defined means of determining precision.  Again, if we look at our four different matrices, we find that our RSDs are all less than 5 percent for all of those analyses, and that continues to be consistent throughout the range of samples that we've looked at.

     Next slide.

     But I think the most important aspect of any analytical method is its accuracy, and accuracy is one of those things that's very difficult to get at with food matrices.  And the way that we've approached this is to look at proficiency testing where we've had enough laboratories participating with a well-defined sample and how close we come to the mean of those determinations.

     And so we've participated in two rounds of testing--one sponsored by the National Food Processors Association, in which five laboratories used LC/MS/MS; and another by the Food Analysis Performance Assessment Scheme, FAPAS, which is an internationally known proficiency testing organization under the CSL of the U.K., Chemical Science Laboratories in the United Kingdom, and they do proficiency testing throughout the world.

     What's interesting about the FAPAS results are that there were 37 people who actually returned results, and of them, 32 of them received satisfactory scores, which means they were within what would be expected to be the correct range of the correct result.  That means there's a lot of labs out there doing analyses and a lot of them are getting at least very consistent results.

     Also of interest was that there were about an equal number of laboratories using GC/MS versus LC/MS, and that those--there was no statistical bias in terms of reporting numbers for GC/MS or LC/MS methodology.  I think that's important as we develop these databases of people looking at different products that we have some assurance that we can compare those results and know that they're relatively in the correct ballpark.

     In our case, the assigned value from this particular proficiency testing scheme was 1,213 micrograms per kilogram, and we reported 1,264.  That's within our 5 percent standard deviation expected results.

     The next slide?

     That brings me to conclude the discussion of the method portion of the talk.  We're in the process of preparing a final report for a single lab method validation.  We've been a little slow in doing this, and it's been complicated by a number of factors.  This is a fairly--typically, when we would do a method validation, it would be for a single component in a single matrix.  What we're doing now is a single component in multiple matrices.  So we're working at do we have enough validation data to cover the range of samples where we might expect a problem, and then within those individual samples, do we have enough information on those particular samples to complete our validation study?  And we're really at the point where we believe we've done that, and we're compiling that information.  It's voluminous now.  But I think that that work will be done.  We do have a valid method for analysis of foods.

     We're also in the process of preparing proficiency testing samples.  They're going to be four to six different types of matrices that we can send out to laboratories who are interested in doing these analyses, either for us as a contract service or other FDA field laboratories that might be doing analyses where we want to have some confidence in their analytical results and everyone is doing the same--reporting numbers on the same particular matrix.  Those materials have been prepared and are now ready to be sent out.

     The next process for this method will be a three-lab peer verification of the method where we use another two laboratories, including our own, to verify the range of results that we get.  This is a fairly standard procedure and, if necessary--and we haven't reached a conclusion on this--doing a full collaborative study of the method.  There's good and bad points to doing a full collaborative study, but we still haven't decided to proceed that way yet.

     Another item that we're exploring and we're hoping we can get other people interested in exploring are alternative testing methodologies.  LC/MS is good, it's fast, it's rugged.  It gives extremely accurate results.  But it's very expensive.  Contract laboratories are charging anywhere from $200 to $250 per sample analysis.  So if you get one sample run in duplicate, you could be looking at a $500 bill.  So we're looking at, you know, are there other methods that can be used that are not as expensive but give results which could be used for process control or just analysis of--a broader range of analysis--getting more analyses done by more people and yet still getting accurate results without having to do LC/MS or GC/MS.

     Interestingly, GC/MS contract laboratories are charging about the same amount for LC/MS, so there doesn't seem to be a breakdown there.

     And, finally, we're using this method to look at foods.  We're looking at foods pretty much every day.  We've got a continual shopping basket coming into the laboratory for analysis.

     The next slide?

     Okay.  I'd like to move now into what our actual results are and concentrate a little bit on exactly what it is that we found, or didn't find in some cases.  I'd like to point out that this is an exploratory survey, that this is by no means a representative sampling.  And I think what you're going to see from some of the data that I present later that there is a considerable amount of variation in the foods, and that even though we've looked at perhaps 300 different food samples, there is so much variation that this cannot be considered representative in any case.

     We chose the foods based on a number of different parameters.  Basically we started out looking at, okay, people in Europe, different laboratories reported finding this in fried foods.  So we picked a number of fried foods, potato-based foods.  We also then expanded it to crackers and other foods where other laboratories had initially reported finding acrylamide.

     Also, unpublished findings, contract laboratories and other academic laboratories would called us and say, hey, we found it in food matrix X, did you find it in food matrix X?  And so we would add that to our list of foods--much like Dr. Canady said, a very iterative process to food sampling.

     Then we also did try and take a more concerted effort to choosing some samples by group.  For example, we looked at infant formula to see if there were significant levels of acrylamide in infant formula; or by total, so, for example, in cereals we looked at cereals with very high sales that accounted for large market shares.  Again, the sampling is not representative, and many foods only have one sample.

     We'd also like to constantly expand our number of foods analyzed, so if there are things which we have--you have our survey results now.  If there are food commodities that you think we should be looking at, we would like to know about that so we can include them in our survey.

     Next slide.

     So where are we and what are we doing?  Since we first published or talked about some of our results in September, we've about doubled the number of analyses that we've done.  So far, about 300 different food samples analyzed, nearly 700 separate analyses, and these analyses don't include any of the analyses we do for standard curves, proficiency testing, method validation.  These are just analyses for foods in the survey.  More than 35 different food types, and by food types I mean baby foods, coffees, breads, vegetables, seasonings, those types of food categories.

     Next slide, please.

     The data is not all bad.  We don't find acrylamide in everything.  For example, potatoes, one of those sources for acrylamide in French fries, if you look at uncooked potatoes, we don't find any acrylamide in them.  The same thing for frozen vegetables or vegetable protein.  Uncooked, we find no levels of acrylamide.  Also with meat, fish, and chicken, raw or cooked, we don't find any significant levels of acrylamide in those particular samples, and infant formula, no significant level of acrylamide found in any of the 12 samples of infant formula we surveyed.

     Next slide.

     This next chart--and there's a lot of data in some of these charts, so if I'm going too fast or you have questions, please slow me down because there's a lot of information here.

     This is an example of some of the variability that we found to occur within certain food types, and one of the more interesting aspects of this is if you were to look at cocoa, for example, and you only sampled a couple of cocoas, you might think that, oh, well, there's not much acrylamide in any of them.  But what we're finding is we'll go along and analyze some samples, and as we add more to our data set, we'll find some of them that have very high levels.

     I should also point out that coffee in this case, the coffee samples are not for brewed coffee.  That's for the raw coffees.  So that would be we take the ground coffee out of the container, look at that.  This is not going out to a coffee vendor and taking the already brewed coffee and analyzing it.  This is the raw, unbrewed variety.

     If I can have the next slide?

     DR. FISCHER:  You haven't looked at any brewed coffee?

     DR. MUSSER:  We have not looked at any brewed coffee.

     DR. FISCHER:  Or cocoa?

     DR. MUSSER:  Or cocoa.

     DR. GRAY:  But you know water extraction's a pretty good way to get it out.

     DR. MUSSER:  But we--yeah, I mean, we have--I've talked with a number of laboratories who have done these analyses, and if you simply multiplied, you know, a six-ounce cup of coffee plus what we find, you'd find that those results are--it's almost 100 percent extraction.  And it's also an acidic environment, so it would be expected to be fairly stable.

     This is just some more example data.  I couldn't fit it all in one slide.  This is just some more examples.

     The fish and the chicken, if you remember from the previous slide when we said there's very low levels in there, we looked at things like chicken nuggets that were fried and fish sticks that were baked.  And we did find levels in those particular products, detectable levels of acrylamide.  And so we took it one step further, and we peeled off the breading and we looked at the individual meat that was in there and we looked at the breading coating.  And all of the acrylamide could be accounted for in the coating that was on these products.  And so when we say fish or fish sticks or chicken nuggets or chicken tenders, those are--the acrylamide is really in the outside of those products and not in the meat.  We don't find significant levels in meat.

     A couple of points about this particular slide.  Soups--these are dried instant soups.  We go along and we analyze six or seven different instant soups, find very low levels.  Then we find one that consistently gives us very high levels of acrylamide.

     Likewise, in French fries that are baked by the consumer and French fries that are obtained from fast-food restaurants, we find tremendous variability in the amount of acrylamide that's in product.  Not only do we find variability within individual manufacturers, but within different locations and the way they're prepared.  And this really--this data, as we've been developing it, prompted us to do a number of different experiments and look at a number of different food products, because it was starting to become clear to me and to a lot of the other people looking at this data that we really had an incomplete snapshot of what was happening in foods in the U.S., in particular with acrylamide levels in those foods.

     Can I have the next slide?

     The first thing we did was we said let's go out one weekend, and we all kind of live around the Washington, D.C., metro area, and, you know, go to your favorite fast-food place or go to some fast-food places, pick up some French fries.  So what we ended up with was a large number of French fries from several different fast-food restaurants, where we looked at the individual levels in those particular French fries from those particular locations.

     What we found was that, yes, there's a lot of variation.  If you look just here where we have five samples from McDonald's, they span from about 200 to almost 600, the amount of acrylamide that we found in those particular--parts per billion of acrylamide in those particular French fries.

     But if you look at the--what's beginning to develop here is that independent of restaurant, the median levels are about the same.  So if you had only sampled, for example, one low sample from Wendy's and one higher sample from, say, Fuddruckers, you might be led to believe that one particular company was giving you higher levels of acrylamide than another company.  But what you find when you do the sampling is that, at least with French fries and fast food, those median amounts are all about the same no matter where you would end up, that there is a detectable amount but that the mean levels are all approximately going to be right in a central area between 200 and 600 ppb.

     Can I have the next slide?

     This was a very interesting experiment and one where we have a lot of information and don't exactly know how to interpret it.  We went out and got four samples of big bags of potato chips with six different dates codes on them.  Now, the manufacturer was kind enough to provide us with the lots of potatoes that were used, where they were manufactured, and at what time.  And so in this case, the potatoes were all harvested and then immediately processed, so there was no storage time.  They were all run in the same facility and actually all run on the same line.

     So what we're looking at are not the--in most cases, we're displaying the mean values, but these are the actual individual analyses, so it looks like there's a lot more than there really was.

     So on November 5th, this is what we saw grinding up the entire bag of potato chips and then looking at individual amounts within those potato chips.  So we saw some variability bag to bag, and a lot of variability, you know, lot to lot.  And the different colors are not different colors to help you see, but the color codes are the codes for the variety of potato that was used.

     And so here all of the purple ones are a particular different variety that came from a different farm, but in one case, we find fairly high levels, and in another cases, you know, half of that value.  So the same variety produced on the same line producing levels that are of great variation.

     Also, because these are fresh potatoes, it doesn't take into account any of the storage factors.  You know, potatoes are really kind of living organisms, and they are constantly using starch in their storage process.  So we don't know what effect taking the potatoes and storing them for a while prior to the actually cooking of them, whether that will increase or decrease levels of acrylamide.

     Next slide, please.

     We did some more stuff.  When we got French fries, we didn't just get them frozen, you know, you go into the supermarket and you get a bag of frozen French fries.  Well, analyzing them as frozen material and looking at the acrylamide levels in there is probably not representative of really what you see in the final product, which would be baked.  And we baked them according to the manufacturer's instructions, so if they said--most of them were 450 degrees for anywhere from 8 to 20 minutes.  So we would take whatever the manufacturer said and cook them in the same oven, and what we found was if you just look at them--and some of them are precooked a little bit, so they're fried, maybe pre-fried before they're frozen.  So there are some significant levels of acrylamide that are present in the frozen product before it even gets baked.

     But when you bake it, the acrylamide levels start to change dramatically and cover a very wide range.  You know, these are fairly nicely grouped below 200 parts per billion, but we go all the way from 200 to almost 1,300 parts per billion when we bake them in an oven.  Of course, this prompted us to do more experiments.

     Next slide, please.

     In this experiment, we said:  Well, what do consumers really prefer?  Because the analysts, when they were cooking these frozen French fries, said, well, you know, this isn't how we would eat them.  In other words, we cook them, they're hot.  You know, we've cooked them for 10 minutes, they're hot, but that's not really how we would eat them.  So I thought about this for a little bit, and I thought, well, maybe what we ought to be doing is looking at how people like to have their French fries cooked or baked.  And so we took about 12 pounds of French fries from one particular manufacturer, frozen French fries, and in this case, the starting amount of acrylamide was very low.  So I should note that, unlike in the previous slide where we had a fairly high initial level--or higher levels of initial acrylamide in the French fries, in this case we have very low levels.

     So when we cooked them for the recommended time, which was 15 minutes at 450 degrees, we didn't see very much acrylamide.  But no one in the group of the six people we had cook them really wanted to eat the French fries in that particular condition.  They wanted them to be cooked a little longer, maybe a little browner.  And so what you see is that when people cook them the way they had a preference--and this is, of course, by no means a representative sampling.  Six people in my laboratory doesn't constitute a representative sampling.  But you do see that those levels do go up greater than what might be found just cooking them according to the instructions.

     So that adds another variable and dynamic to our analyses.  In other words, if we're taking measurements just by cooking them according to what the instructions say, is that really giving us a representative level for what might be encountered in consumers' kitchens?

     If I can have the next slide?

     This slide really, I think, is very demonstrative for what we see and why acrylamide levels tend to correlate with the amount of browning that occurs.  If you take a French fry that hasn't been cooked at all, we don't find any acrylamide in it.  We've looked at it.  I should also point out that we've microwaved them and we don't find any acrylamide.

     If you cook it for 15 minutes at 450 degrees, it's a nice color but it's kind of limpy and doesn't have any browning effect, and you get a very small amount of acrylamide.  If, however, you cook them until it starts to brown a little bit, 30 minutes--and I should also point out that when we cooked them according to the people in my laboratory's preference, it was about 25 minutes.  So the median was about 22 minutes of cooking, so about 7 to 10 minutes longer than what was recommended by the manufacturer.  And so we saw levels, you know, approaching this amount at 30 minutes, and this brown French fry kind of represents what people like to see, a slightly harder coating, it wasn't a real soft coating, it wasn't real limp.  And even after 45 minutes of cooking, where we have kind of astronomical levels compared to what we had found in typical products, these French fries were considered just as tasty and desirable even though they had very, very high levels of acrylamide in them.

     So if I could have the final slide?

     What we have here is lots and lots and lots of data.  But certain trends are starting to develop, and there is a lot of information to be gained from this initial survey.  But, first of all, the analytical method that we've developed we believe is reliable and comparable to other--gives us data that's comparable to other laboratories and other agencies producing information on acrylamide levels in foods.

     Also, the results clearly indicate that there is considerable variability in acrylamide levels both in commercially produced and in consumer-produced products; that there are a number of factors which contribute to the variability, and small numbers of sampling.  So if we were to have gone out and just gotten one sample, and maybe even five or ten, it's not probably representative of what the median level is going to be for that particular product.

     So that can be very misleading, and the amount of variability observed in certain food types and what we've seen with the consumer preparation indicates that there may be a way of dramatically reducing acrylamide levels.

     Those are the conclusions that I'd like to make from this presentation.  Thank you.

     DR. BUSTA:  Thank you very much.

     Are there questions for clarification?  Dr. Gray?

     DR. GRAY:  Thank you.  A quick question.  Can you just confirm for me that it looks--I just want to make sure that I get it through my thick skull.  The acrylamide that you're measuring is acrylamide monomer.

     DR. MUSSER:  Yes.

     DR. GRAY:  It's not bound and you're bringing it out with water, so you're probably not taking it off of things.  This is acrylamide that's floating around--

     DR. MUSSER:  This is free acrylamide, yes.  And it would be very unlikely that if you had polymeric acrylamide that you would go back the other way.

     DR. GRAY:  Right.

     DR. MUSSER:  It's pretty much a one-way street for the polymerization.  You bind it or you form a polymer.  You don't go back the other way.

     DR. GRAY:  And the same with binding to proteins?

     DR. MUSSER:  Yes.  If it was bound covalently to proteins, it probably wouldn't come off.

     DR. BUSTA:  Dr. Hotchkiss?

     DR. HOTCHKISS:  First a comment.  I've always considered analytical chemistry to be the strength of FDA, and you've done nothing to dispel that belief.

     Just a quick question.  Are the data, the numbers that you present for individual samples corrected for either your overall recovery or individual heavy isotope recoveries?  Or are they as found?

     DR. MUSSER:  They are as found.  It would be--if we had to do a standardization for every single one of these 300 samples, we would never be done.

     DR. HOTCHKISS:  Well, let me say, as I understand it, each sample was spiked with the internal standard.

     DR. MUSSER:  That's correct.

     DR. HOTCHKISS:  So you know what the recovery is of that internal standard for each individual sample, right?

     DR. MUSSER:  Yeah, really, we're taking advantage of the stable isotopes because as long as we can assure that we are extracting most of the acrylamide out of the sample into the solution and that that's representative of what's in the sample, the stable isotope will correct for all the other problems that may occur during the work-up and clean-up.  I mean, that's just the fundamental basis of stable isotope label standards.  And we're fairly certain--we've got only one matrix where we know we don't have good--we're not recovering internal standard from in all the ones we've looked at.  And we kind of use that as our basis for is it working or isn't it working.

     And so we can get a signal for the internal standard.  You can pretty much be assured that you can quantitate the level of acrylamide that was in the particular product.

     DR. HOTCHKISS:  I understand that, but now I'm a little more confused.  For example, on a given sample, you get 95 percent recovery of the internal standard, which I think was the number you--

     DR. MUSSER:  Yes.

     DR. HOTCHKISS:  And you find the level of acrylamide in that product that is not labeled.  There are two numbers you could report:  what you found, or you could take what you found and divide it by .95, assuming that the--

     DR. MUSSER:  Yes.

     DR. HOTCHKISS:  --found level was recovered at the same level as the internal standard.  Now, which way did you go?

     DR. MUSSER:  We do not calculate a recovery into that.  We calculate everything based on a unity of recovery.

     DR. BUSTA:  Could I pursue that a little farther?  Where do you put the spike in the French fries that are baked in your hands?

     DR. MUSSER:  Okay.  We take the French fries that are baked in our hands, and this is typical for every sample that we would do.  We homogenize the sample.

     DR. BUSTA:  This is after--

     DR. MUSSER:  After baking.

     DR. BUSTA:  You have at no time put them in the raw sample--

     DR. MUSSER:  No.

     DR. BUSTA:  --and then baked them?

     DR. MUSSER:  No.  No.  I don't know how I would do that, but we could try.

     DR. BUSTA:  Dr. Fischer?

     DR. FISCHER:  Let me get back to this a little bit, because I had the same question.  So acrylamide, in fact, is two things:  chemically reactive, and it's somewhat volatile, or is volatile.

     DR. MUSSER:  Yes.

     DR. FISCHER:  So there's a chance of it being lost from the product while being cooked as well, right?

     DR. MUSSER:  That's correct.

     DR. FISCHER:  And I understand from your answer just now that you don't know how much is really lost.

     DR. MUSSER:  We don't know how much is lost.

     DR. FISCHER:  Now, for the chemically reactive part, is it possible that acrylamide in a product is bound to protein or other substances in the product and, as such, is not extracted in that form; but, in fact, if you were to eat the product, it's possible, I suppose, that it might be released?

     DR. MUSSER:  Yes.

     DR. FISCHER:  So do you have any idea about the bound acrylamide in the product?

     DR. MUSSER:  We're pretty certain from a number of the matrices we've looked at--and if you look at coffee and potato chips and French fries and bread crumbs, where we have a fairly representative range of levels and matrices.  If the acrylamide is available free, we're extracting it pretty much quantitatively.  We can't comment on is there a reversible reaction that's occurring.  I can't even think of one that might be occurring where you've got some kind of way of binding it to a protein, as you suggest, or in an oil, which might be more likely, where it's encapsulated and so you would have a capsule formed, mycel, for example, where the water couldn't penetrate the mycel and so you weren't extracting it.  I mean, that's certainly a possibility and one that we don't want to rule out.  I can't--I don't know.

     DR. FISCHER:  I would suggest that if you use some isotope, radioisotope-labeled asparagine, maybe, or whatever, to get bound acrylamide in the samples and see whether you can get it out, it might be--

     DR. MUSSER:  We thought about doing those experiments, and the problem with doing those experiments is that if the acrylamide reacts with a protein and forms a covalent modification where it is then not available, in other words, when you make an acrylamide adduct with a protein, it's a one-way street for acrylamide.  It doesn't come off, unless you chop it up in acid, and then you still have acrylamide plus whatever amino acid it's bound to.  So--

     DR. FISCHER:  That happens in your stomach.

     DR. MUSSER:  Well, it could, yes.  And so you have this radioactive label now that's bound to acrylamide.  It's not available.  It's not really even acrylamide anymore.  It's just an adduct.  And so if you did that radioactive experiment, you're not going to come up with really the free available acrylamide.  You might come up with how much is bound, but you don't know whether it's actually bioavailable as acrylamide.

     Now, you could also do some other experiments where you hydrolyze the protein and look at the bound acrylamide.  But it's probably not reactive at that point.

     DR. ACHOLONU:  Alex Acholonu.  You said that--

     DR. MUSSER:  Wait a minute.  Dr. Canady--excuse me.  I think Dr. Canady--

     DR. CANADY:  One of the reasons we're doing bioavailability measurements is to look at this exact question.  We would look at the incurred amount at the--naturally incurred amount through cooking, and measure what's found in the blood so we could measure--we can see how much we extract using the analysis procedure, see how much we get in the bioavailability measures, and get to the question you're asking.  That's another way to get to it in addition to what Steve's talking about.

     DR. ACHOLONU:  You said that the acrylamide level tends to correlate with the amount of browning.  Is the factor here the color, the browning, or the intensity of heat applied in the preparation of the food?

     DR. MUSSER:  Could you repeat that again?

     DR. ACHOLONU:  Yes.  You talked about the fact that acrylamide level tends to correlate with the amount of browning in the cooking, and you showed that in one of your slides.  Now, what I'm trying to find out is:  Is it the color of the food when it is fried or the amount of heat applied that determines the level of acrylamide in the food or in, for instance, the French fries?  Browning--and if it is that, supposing it gets charred, gets black, what would be the amount of the acrylamide in the French fries?

     DR. MUSSER:  I would imagine pretty substantial as it got black.  But to get to your specific question, Dr. Jackson in the next presentation is going to talk about those manufacturing processes which might lead to differences in the acrylamide level, and she'll be able to tell you about temperatures and the length of time and talk to you a little bit more specifically.  Her presentation will deal with your question pretty much exclusively.

     DR. ACHOLONU:  Okay.

     DR. BUSTA:  Dr. Lee?

     DR. LEE:  I was just wondering, there certainly is a need for inexpensive and accurate screening methods, and there are several laboratories working on it.  Do you have any predictions as to when these might come online?  Say a year from now will we have a less than $250 per sample analysis?

     DR. MUSSER:  I don't know that it will ever get really cheap.  The expense here is primarily labor.  It's fairly labor-intensive just to get the sample to the point that you can inject it and analyze it.  But it should be much less, you know, per sample.  And one of the results we saw from the FAPAS test was that one of the participants had used an electronic capture detector, GC/ECD, and that gave results which were as good as the GC/MS results for the FAPAS sample.  So, I mean, that's certainly a possibility of one of the tests.  We're looking at some UV derivatization methods that might work as well.  But I don't--there's not enough data on the LC side to predict that.  Certainly it looks like ECD is already to the point that it could be used.

     DR. BUSTA:  Dr. Downer?

     DR. DOWNER:  Thanks for a very informative presentation.  I noticed that all the foods from the food guide pyramid were represented in your research with the exception of fruits.  Is there any reason why that's been excluded?

     DR. MUSSER:  No.  We did look at some fruits, and they were baby food fruits, like applesauce, and we didn't find any, any significant levels of acrylamide in fruits.

     DR. DOWNER:  So it's your thinking that perhaps foods for the general public's consumption--you remember now that we're recommending that we eat a wide variety of foods, including fruits and vegetables.  Are you going to be doing just regular fruit that adults consume?

     DR. MUSSER:  Well, keep in mind that raw fruits and raw vegetables in our study didn't show any acrylamide levels.

     DR. DOWNER:  Are you going to include cooked fruits?

     DR. MUSSER:  I can't--we don't have enough data to comment on that.

     DR. BUSTA:  Dr. Whitaker--oh, just a second.

     DR. CANADY:  One of the ways to address that particular question is through the Total Diet Study which looks at a representative sample of foods across a lot of different categories.

     DR. DOWNER:  Thank you.

     DR. CANADY:  And we intend to do that.

     DR. BUSTA:  Dr. Whitaker?

     DR. WHITAKER:  Steve, there's a lot of good data there, and it's hard to comprehend it all.  But I noticed in one of your slides on method performance precision, it seemed to--the variability seemed to increase with concentration that you were measuring.  Have you looked at that?

     DR. MUSSER:  Yes.  The variability--well, actually, the variability stays about the same.  It's about a 10 percent standard deviation.  What you're seeing is the amount that's varying.  So, for example, if you had a sample that has 1,000 ppb in it and you added 10 percent, then you would be plus or minus 100 ppb.  If you have a 100 ppb sample and you're plus or minus 10 percent, it's going to be, you know, 110 versus 90.  So it's an apparent change, but the actual standard deviation that's observed for the analysis is the same.

     DR. WHITAKER:  The coefficient of variation stays about the same.

     DR. MUSSER:  Stays about the same, yes.

     DR. WHITAKER:  But the standard deviation actually increases with concentration.  Have you looked at that?

     DR. MUSSER:  I'll have to look at that more carefully.  There's too much data there.

     DR. WHITAKER:  Yes, there is.  And we see that in mycotoxin work, that the variability as a function of the standard deviation does increase while the CB tends to stay constant.  So--

     DR. MUSSER:  I'll go back and plot that.

     DR. WHITAKER:  You may have information  here to go back and describe that.

     DR. MUSSER:  I'm sure we have enough.  It's just a matter of going back and massaging it.

     DR. WHITAKER:  Yes.  One more question.  Have you had the time to look at sample-to-sample variability within a lot or a consignment?  I mean, if you have a lot, you have a thousand bags of potato chips, have you had a chance to look at the variability from bag to bag or from sample to sample?

     DR. MUSSER:  Only minimally.  In that study I described on the different varieties of potato chips that came from the same facility, they were four bags from the same date code or same lot, produced on the same manufacturing line.  So we took those four different bags and ground up each one individually and then analyzed them, and there is, you know, a substantial amount of variability in those bags of potato chips.

     DR. WHITAKER:  Yeah, I would think that understanding that variability would be crucial in trying to get an accurate estimate of the acrylamide in that lot.

     DR. MUSSER:  I think that's very important yes.

     DR. BUSTA:  Dr. Kuzminski?

     DR. KUZMINSKI:  A couple of questions here.  Is it the agency's intent--or maybe this is not a practical suggestion--to recommend a standard method for the analysis of acrylamide?

     DR. MUSSER:  I don't know that we've ever done that, and I don't think that that would be appropriate in this particular case, especially given the fact that so many other laboratories have produced equivalent results with vastly different methods.

     DR. KUZMINSKI:  Another question.  Just on--a couple of the slides that you show had data which represented phenomena that you indicated you might to know what it meant, like the variability of levels given various code dates on potato chips, I believe, as an example.  I recognize that it's early in the process.  This phenomenon or this occurrence has bubbled up in the last several months very, very quickly.  But is it the intention of the agency to establish at least a procedure or a mechanism whereby when you--and I'm sure there will be other data which will reveal phenomena you will not understand also, in addition to this one.

     My question is:  Will the agency set up a procedure or mechanism whereby you consult with other experts in the field, be it in academia or in industry, that could help explain these phenomena to you?

     DR. MUSSER:  I can't speak for what the agency intends to do.  Perhaps, Dr. Troxell, you'd like to address that.

     DR. TROXELL:  Well, we have two consortia we're working through, for one, and the National Center for Food Safety and Technology is looking at the formation of the mechanism kind of questions and will be interacting with a lot of different stakeholders, and we certainly are going to collaborate not only through them but through JIFSAN, with people all over the world, to identify, you know, the reasons for the variables that are involved, and to try to go beyond just the levels but to also correlate particular levels to, you know, pH and temperature and water conditions and other chemical factors in foods that may affect levels, as people develop the knowledge to understand what causes the formation.

     I think Dr. Jackson's talk will be very informative to that this afternoon on what we've already learned about the mechanism and what factors are important and what needs to be explored in the future.

     DR. BUSTA:  Thank you.

     DR. ACHOLONU:  Last question.  Will roasting of food or baking affect the level of acrylamide in the food, baking or maybe grilling or barbecuing, any of those?  Are they out of danger?

     DR. MUSSER:  There's some limited availability of that data.  At the AOAC meeting in September, Procter & Gamble presented some data where they roasted asparagus, which is very high in asparagine, and they found in that particular sample it was the highest of all the acrylamide-containing samples they looked at.  So, yes, you can produce acrylamide levels by roasting or grilling.

     DR. ACHOLONU:  And baking?

     DR. MUSSER:  And baking.  Yeah, in fact, in the French fry data where we've done that as a consumer baking them, that was all baked.  That wasn't fried.  All that data is from baked French fries, baked frozen French fries.

     DR. ACHOLONU:  If that is the case, why did we--why do you emphasize the question of frying and not including baking and roasting in the write-ups?  Why is the frying of food so identified with acrylamide?

     DR. MUSSER:  Oh, that's--

     DR. ACHOLONU:  Why are all the readings we have not talking so much about roasted food, baked food, barbecued food, grilled food?

     DR. MUSSER:  That's probably related to how the initial results came out of Sweden where they only looked at primarily fried foods, and so there was a real interest in just looking at fried foods.  But recent data--and probably why you're not seeing it is that only recently have people expanded the kind of way they looked at foods to include--plus understanding the mechanism--into foods that are baked and roasted and not just fried foods.  So it's probably just a matter of this happened very recently and so it's taking a little bit of time for people to catch up with the initial results and expand their surveys to include other samples.

     DR. BUSTA:  I would like to sincerely thank all the FDA presenters for the morning, the very succinct presentations, comprehensive, on time, and the panel for their thoughtful questions.

     We will recess for lunch and reconvene at 1:15.

     [Luncheon recess.]

A F T E R N O O N  S E S S I O N

     DR. BUSTA:  I'd like to reconvene the Subcommittee.  Our first individual on this afternoon on FDA Action Plan on Formation is Dr. Lauren Jackson.

*     DR. JACKSON:  I'd like to welcome you back, and unlike the previous speakers, I'm from the FDA National Center for Food Safety & Technology, which is located in a suburb of Chicago.  For those of you who don't know what the NCFST is, it's a consortium between the FDA, the food industry and academic members, and the main focus of the research that we do at the NCFST is to look at the effects of food processing and packaging on the chemical and microbial safety of foods.

     The objectives of this presentation are to summarize what's known about acrylamide formation, the mechanisms, the precursors and factors that affect acrylamide formation, identify the research gaps, and then, finally, and most importantly, discuss the FDA Action Plan on acrylamide, with respect to formation, and which, to summarize, is to understand the conditions that lead to acrylamide formation in food for the purpose of developing methods to reduce or prevent acrylamide formation.

     What is known about acrylamide formation.  How is it formed in food, what are the precursors, the mechanisms, and what factors affect acrylamide formation, including food processing and food composition.

     Almost immediately after the Swedish group announced the presence of acrylamide in thermally processed high-carbohydrate foods, there was speculation as to where acrylamide was coming from.  What I'm going to talk about here in the next several slides are the precursors and pathways that have been discussed as potential ways of forming acrylamide.

     Listed here are four different pathways or mechanisms.  The first three that I have listed here are the acrolein, the acrylic acids, the amino acid alone have been pretty much disproved as the major pathway for acrylamide formation.

     And the fourth that I've listed there, the amino acid reducing sugar, Maillard browning, Strecker degradation has been, pretty much there's been a consensus that this is the major pathway in food.

     Fairly early on the acrolein pathway was speculated to be the major pathway in food.  Mainly, if you look at the structure of acrolein, it's very similar to acrylamide, and it's thermal degradation product of oil, so it's formed in frying oils if they've been heated excessively.  So there was a reason to believe that the reason why we find acrylamide in fried foods may be because of acrolein as a precursor.

     Acrolein is not only formed in thermally abused oils, it's also formed in thermally degraded starch, sugars, amino acids and proteins.  As I mentioned, this pathway has been pretty much disproved by several research studies--well, one major one by the one group at HealthCanada, Becalski and his group out there, where they heated acrolein in combination with ammonium carbonate, which is a source of ammonia, and they didn't form acrylamide.

     There have also been isotope-labeling studies that's pretty much disproved that this is a viable pathway.

     Similarly, the acrylic acid pathway was speculated to be a pathway as well.  Again, it's because acrylic acid is very similar structurally to acrylamide.  Again, it's a thermal degradation product of alpha and beta alanine, different diacids in foods and amino acids, and again it's pretty much been disproved as a major acrylamide mechanism or precursor.

     Amino acids alone, when heated, can form acrylamide.  Dr. Richard Stadler and his colleagues at Nestle Research in Switzerland did some work pyrolyzing individual amino acids and found that acrylamide does, indeed, form when heating Asparagine and Methionine.  However, the yields from these two amino acids are fairly low, and the relevance in most foods, such as potatoes and grains, this pathway is not believed to be a viable pathway in high-moisture foods such as potatoes and grains.

     However, the relevance in other foods such as coffee, which is roasted at a high temperature  under pyrolysis conditions, this might be a viable pathway and this needs to be verified.

     In September/October, there were four individual reports that linked acrylamide formation to amino acids and reducing sugars precursors, and Maillard and are Strecker degradation pathways.  There are two major mechanisms that this can occur.  What I'm going to do is, in the next couple of slides, summarize what has been presented at the acrylamide workshop in Chicago that occurred last month, two months ago.

     What are the Maillard and Strecker degradation reactions?  They are a reaction between amino acids and reducing sugars or other sources of carbonyls, and they are key reactions in the formation of color and flavor in food.  They are key to the flavor and color of french fries, baked products, coffee, cocoa.

     The reasons why people suspected that this mechanism or these mechanisms might occur is that potatoes and grain products are fairly high, they have fairly high amounts of free amino acids and are fairly rich sources of carbohydrates, and some anecdotal evidence was supplied by some of the work done by the Swedes, where they found a higher amount of acrylamides in foods, as the foods tended to brown.

     So which of the amino acids are acrylamide precursors?  The identity of the amino acid precursors was determined in model systems consisting of amino acids and glucose in an aqueous environment.  This table that I show here was taken from Mottram's paper in Nature.  He's from the University of Reading, I believe, in the U.K., and what he did was he heated up different amino acids, asparagine glycine, cysteine and methionine, glutamine, aspartic acid, glucose in an aqueous buffer at 185 degrees for 20 minutes.  This was in a closed system.

     And he found that asparagine was the major precursor as you can see from the numbers you have here.  You can see here, glycine, cysteine, methionine didn't form any acrylamide.  Small amounts of acrylamide were formed from glutamine and aspartic acid, and I'll talk about what the implications of those data are in the next couple of slides.

     In a similar study, Sanders, Zyzak and their group at Procter & Gamble did a similar type of study, where they looked at acrylamide formation, instead of in aqueous model system, they looked at a model potato chip.  So what they did was they formed potato chips from potato starch, glucose and a variety of different amino acids, and they've pretty much confirmed what Mottram found, is that asparagine is the major amino acid precursor, and again they found small amounts of acrylamide formed from glutamine.

     The conclusive proof that asparagine is, in fact, the amino acid precursor was, again, from Procter & Gamble's group.  They did some really nice work looking at isotope labeling of asparagine.  The left part of the slide here shows asparagine that was labeled at different parts of the molecule with different stable isotopes, nitrogen 15 or carbon 13, and what they did was they reacted these labeled asparagine with glucose, heated it up, and then measured the master charge ratio of the resulting acrylamide.

     What they found was all of the nitrogen and the carbon from acrylamide was derived from asparagine, and the amid part of the asparagine was actually incorporated as the nitrogen that's in the acrylamide.

     Now that we know about the precursors, how are they formed?  There are two mechanisms that have been proposed.  They're slightly different.  They both involve Maillard and/or Strecker degradation reactions.

     The first one is a combination of Maillard and Strecker degradation reactions, and basically what it is is asparagine undergoes Strecker degradation in the presence of dicarbonyls, which are formed as a consequence of Maillard browning.

     Mottram did some nice work looking at or verifying that this mechanism actually does occur by heating up dicarbonyl, butanedione he used, and he heated it up with asparagine, the top two.  This is where he introduced there, he did those two up, need then actually measured and found acrylamide formation which showed that this pathway is actually viable.

     There's a different mechanism that was proposed by Stadler and his group at Nestle, and this was a little bit different.  Basically, he found that acrylamide could form from the N-glycosides, which are early Maillard browning products.  What he did was he took N-glycosides of asparagine.  The first two are N-glycosides of asparagine, glucose and fructose and glycosides of asparagine.

     He took he an N-glycoside of glutamine and an N-glycoside of methionine as well, and he heated those N-glycosides up.  I believe he did it at 180 degrees for 30 minutes in the dry state, and then he measured acrylamide formation from those N-glycosides.  What he found was the asparagine N-glycosides formed substantial amounts of acrylamide.  Whereas, the glutamine, and methionine and glycosides form very small amounts, trace amounts, small amounts.

     As I alluded to, there are, other than asparagine, there are other amino acids that potentially could form acrylamide.  The formation of acrylamide from glutamine and aspartic acid has been disputed mainly because the model system work that was done with these pure amino acids, they actually were not pure.  The glutamine that was used in these experiments had trade amounts of asparagine, so that might account for the formation of acrylamide from the glutamine.

     Aspartic acid actually is contaminated with trace amounts of cysteine, which again might contribute to acrylamide formation.  Cysteine has been suspected to be--I didn't show any data on this--but it might be another possible amino acid.  It's not believed to be a major precursor.

     Methionine, there is a consensus that methionine actually is a precursor, although as I showed from the data before earlier, that the yield of acrylamide from methionine is far less than from asparagine.  Then, again, this might for acrylamide via N-glycoside formation by the two mechanisms that I've shown before--Maillard, Strecker or the N-glycoside formation.

     We know that these pathways occur in model systems.  We know they've been verified.  How do we know these actually do occur in food?  Some of the work, one of the pieces of information that the asparagine sugar reaction is valid in food comes from data we have on the free asparagine levels in some of the foods that we analyze for acrylamide.

     In this table, I've listed types of food:  potatoes, wheat flour, rye flour, asparagus, almonds, coffee and meat.  I've also listed the percent of asparagine as a percent of the total free amino acids, as well as the amount of this should be free asparagine on a wet weight basis of the food.

     The right is levels of acrylamide in the food after frying, baking or roasting.  This is not an absolute amount there.  This is a sliding range. You can find zero to four stars, depending on how you process the food.

     Basically, what this table shows is that foods such as the top six foods there that have high free asparagine potentially can form moderate to high amounts of acrylamide.  Whereas, meat, which has very low levels of free asparagine, you can cook it, roast it and form, depending on the frosting conditions, again, little or no acrylamide formation.

     This is in direct evidence.  Direct evidence was provided by Zyzak and his group at Procter & Gamble on potatoes, where they treated a mashed potato, and they treated it with asparaginase, which is an enzyme that cleaves off the amid group of asparagine.  It depletes the asparagine levels.

     And then he fried the potato, and then he found that acrylamide levels decreased to almost, well, 99 percent--by 99 percent.  So this is direct evid