Archived Content

The content on this page is provided for reference purposes only. This content has not been altered or updated since it was archived.


FOOD ADVISORY COMMITTEE Meeting: Acrylamide Transcript of Proceedings - February 24, 2003

February 24 - 25, 2003

Transcript of Proceedings February 24, 2003

Transcript of Proceedings February 25, 2003


Sanford A. Miller, Ph.D., Chairman
Catherine DeRoever, Executive Secretary


Francis Fredrick Busta, Ph.D.
Annette Dickinson, Ph.D.
Johanna Dwyer, D. Sc., RD
Brandon Scholz


Jean Halloran
Ken Lee, Ph.D.
Harihara Mehendale, Ph.D.
Robert Russell, M.D.
Clifford W. Scherer, Ph.D.
J. Antonio Torres, Ph.D.


Robert Brown, Ph.D.
Tim Fennell, Ph.D.
Stephen S. Olin, Ph.D.
Sorell Schwartz, Ph.D.
David Zyzak, Ph.D.


Welcome and Introductions,
Dr. Sanford Miller

Conflict of Interest Statement,
Ms. Catherine DeRoever

Opening Comments,
Mr. Joseph Levitt
Dr. Lester Crawford

Development of the Acrylamide Action Plan,
Dr. Terry Troxell

Report of the Contaminants and Natural Toxicants Subcommittee (CNTS) Findings,
Dr. Henry Kim

Questions of Clarification

Revised Action Plan,
Dr. Terry Troxell

Questions of Clarification

Mechanisms of Formation,
Dr. David Zyzak

Questions of Clarification

Reduction Strategies,
Dr. Robert Brown

Questions of Clarification

Exposure Assessment,
Dr. Donna Robie

Questions of Clarification

Adduct Studies,
Dr. Tim Fennell

Questions of Clarification


Call to Order and Opening Remarks

DR. MILLER: Good morning. I am Sandy Miller and I am Chairman of the Food Advisory Committee. I want to take this opportunity to welcome the members of the committee and the subcommittee that are with us today, and our guests and staff members, to deal with the issues that are concerned with acrylamide and food safety.

Before we go any further, let me take this opportunity and ask each of the members of the committee to introduce themselves briefly and then we can go on to the next agenda item.

DR. MEHENDALE: I am Dr. Mehendale from the University of Louisiana and the Monroe School of Pharmacy.

DR. RUSSELL: Rob Russell. I am Director of the USDA Human Nutrition Center in Boston at Tufts.

DR. TORRES: Antonio Torres, Oregon State University Food Science.

DR. BUSTA: I am Frank Busta, Professor Emeritus at the University of Minnesota.

MS. DEROEVER: Catherine DeRoever, the executive secretary for the Food Advisory Committee, FDA.

DR. MILLER: I am Sandy Miller, as you all know.

DR. DWYER: I am Johanna Dwyer at Tufts New England Medical Center in Boston.

DR. DICKINSON: Annette Dickinson, Council For Responsible Nutrition.

DR. LEE: Ken Lee, Ohio State University Food Science and Technology.

MR. SCHOLZ: I am Brandon Scholz, Wisconsin Grocers Association.

DR. SCHERER: Cliff Scherer, Cornell University, with an interest in risk communication.

DR. MILLER: Thank you. The committee has been organized for the purposes of reviewing an Action Plan prepared by the staff in order to deal with the issues of acrylamide in food.

I am not going to provide the background to this because this will be done by some of our speakers in time, but I want to remind the committee that, as you will see, what the committee is being asked to do is to review the Action Plan, not to make a decision concerning the safety of acrylamide in food. As we continue our activities I want to focus on that issue rather than the ancillary issue, although there are questions concerning safety that are going to have to be resolved.

Let me also remind all of the speakers that we have a number of people to speak and to make sure that everybody gets their proper time, we are going to have to stick very close to the time schedule. What I will do, I will interrupt you when your time begins to run out and then as you continue to talk, continue to interrupt you and, if necessary, I will adjourn the committee until you sit down. But it is really important. Quite seriously, it is very important that people stick to their time.

I will also at periodic times ask the committee if they have questions and, hopefully, these will be questions for clarification rather than comment. We will come to comments and opinions at the end of the meeting.

Before we move on to our opening comments, Kathy DeRoever has some comments she needs to make concerning conflict of interest.

Conflict of Interest Statement

MS. DEROEVER: Good morning. As you have heard, I am Catherine DeRoever, executive secretary for the Food and Drug Advisory Committee. First, I would like to welcome all of you to the meeting, particularly the members of the committee, Dr. Miller, Dr. Busta, Dr. Dickinson, Dr. Dwyer and Mr. Scholz; the members who serve on our subcommittees who graciously agreed to be here today, Dr. Lee, Dr. Mehendale, Dr. Torres; and also our temporary voting members--Jean Halloran, as you can see, has not arrived yet but Ms. Halloran, Dr. Russell and Dr. Scherer have been appointed as temporary voting members for the purpose of this meeting by the authority granted to the Center Director, Mr. Joseph Levitt, in the Food Advisory Committee Charter.

Second, the following announcement addresses the issue of conflict of interest with respect to this meeting, and is made part of the record to preclude even the appearance of a conflict of interest. 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 has been screened for interests in the food industry. As a result of this review, in accordance with 18 USC, Section 208(b)(3), Dr. Busta, Dr. Dwyer and Dr. Miller 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 Freedom of Information Act request.

Our invited guest speakers have also been screened for conflicts of interest. Dr. Saunders, who I understand is being replaced by Dr. Brown, and Dr. Zyzak are employed by the regulated industry. No other conflicts have been reported.

I now turn the meeting back to Dr. Miller.

DR. MILLER: Thank you, Cathy. For opening comments, Mr. Joseph Levitt, Director of the Center for Food Safety and Applied Nutrition, will make some remarks.

Opening Comments

MR. LEVITT: Good morning. Let me again welcome you to our wonderful Washington, DC area in the middle of winter. I last saw this committee in the heat of summer. So, I guess, we have a way of playing these things around the weather. At least the subcommittee that Dr. Busta so ably chaired in December was in the middle of a snow storm so here we at least waited till some of the dig-out occurred.

Today's subject is acrylamide. We are at what I think of as phase two of this process. Phase one I would describe as discovery, where we all heard less than a year ago, last spring, of new findings reported from Sweden on acrylamide levels in food. That created a lot of activity that many of you know about and you will hear about in more detail. Phase two is what I call formulating a course of action. That is what we have been doing for the last six months or so. We presented a Draft Action Plan at a public meeting at the end of September, presented that to the Subcommittee of Contaminants and Natural Toxicants in December and are trying to finalize that today. Phase three, therefore, would be the implementation and finding solutions.

This has been such an important issue that actually phase three started before phase two was over. We are happy also to be sharing continued results of research that has been ongoing.

So, what you will hear today in summary is as follows: Number one, you will be hearing a summary of our plan, including the subcommittee's review and suggested improvements to that that have been incorporated. You will be hearing continued findings of research that FDA has conducted and, for the first time, our overall exposure assessments on acrylamide. You will be hearing a presentation from industry research later this afternoon, preliminary research on how to reduce the levels. That is really a major goal here. The results, though preliminary, look very encouraging from what we have heard, and we are pleased to be able to present that to you today.

So, in less than a year I think you will find, as all of us involved, that an awful lot has been done in less than a year though, clearly, a lot more needs to be done to see this important issue through to its conclusion.

As always, what we want from you is your best advice. How can we best crystallize, direct and target this plan to accomplishing what we need to do in the shortest time possible, recognizing that research does take time, that it does take work by a lot of people in order to try and solve a problem of this degree of complexity?

With that, it is my pleasure to make way for the FDA Deputy Commissioner, Dr. Lester Crawford. Thank you very much.

Opening Comments

DR. CRAWFORD: Thank you, Joe and thank you Dr. Miller. I would like to share a few thoughts with you with respect to the agency's current position on acrylamide, but before doing that I want to recall a couple of points of nostalgia to be sure that I run over my time, Dr. Miller.


One is that I used to be on the Food Advisory Committee back in the golden days when we successfully solved the problem of food biotechnology--


--and also somatotropin and then finally Ephedra. But we didn't have a chairman like Dr. Miller. That is for sure. Even earlier, about half a mile from here, my first stint at FDA was out in Beltsville, vaccinating pigs. So, I am happy to be here.


Somebody is out there vaccinating pigs right now.

First of all, let me just say that we are fully committed to developing better knowledge of acrylamide in foods and passing this information along to the public as quickly as possible. We have had a number of issuances on acrylamide, and we are about to have another as a result of this meeting.

At this point we simply don't know what the actual human health risk of acrylamide might be at the low levels found in food. We know it causes cancer and reproductive problems in animals in high doses and is a neurotoxin in humans at high doses. That is why the FDA created an Action Plan that Joe just talked about and that Terry Troxell will talk about a bit later to understand the risk that might be associated with acrylamide in foods, and to reduce levels of acrylamide in foods, including active involvement of this committee.

As a result, we are learning more about acrylamide levels in a broader range of foods than has been previously analyzed. We are also learning more about acrylamide forms in foods and steps that may help reduce acrylamide formation.

Based on our current understanding of the science, FDA continues to advise consumers to eat a balanced diet, choosing a variety of foods that are low in trans and saturated fat and rich in high fiber grains, fruits and vegetables.

The purpose of the meeting, as Joe has said, is to seek input from the committee to assist FDA in analyzing and finalizing the revised Draft Action Plan. The revised Action Plan includes greater detail and reflects new research activities and comments from the subcommittee which we received on December 4 and 5 of last year.

We are releasing 110 additional test results in the spirit of openness and transparency. The findings released today are generally similar to the preliminary results FDA released previously.

FDA exploratory survey findings on levels in foods have shown substantial variability among a wide variety of foods, as well as substantial variability within foods. The initial exposure assessment has found results similar to those conducted by organizations worldwide, such as the World Health Organization and the Food and Agriculture Organization.

The exposure assessment has found that many foods contribute to acrylamide exposure. No single food accounts for the majority of acrylamide exposure for the U.S. population. Some foods that have lower levels of acrylamide contribute appreciably to the overall exposure because they are commonly consumed. Researchers are seeking ways to get acrylamide levels down. Preliminary data are encouraging and we are convinced that there will be some techniques and some technologies developed that might help and we are eager to learn more about that from presentations that will be made at this meeting.

With that, I will conclude and will officially declare, from my position as Deputy Commissioner, that I was on time, Dr. Miller. Thank you.

DR. MILLER: Thank you. I am always happy to be present at rare events.


I wanted to remind Dr. Crawford that the function of this committee is to provide advice, not to make decisions on implementation, and if a problem still exists that is not because the committee didn't give good advice.


That brings me to the next issue. As I said before, the function of the committee is to review the Action Plan and to review the conclusions made by the Subcommittee on Contaminants and Natural Toxicants and to add or make recommendations concerning any changes in this plan.

There are three basic issues we have to think about as far as the Action Plan is concerned. One, as mentioned, is the toxicology. How strong is the data? What does the data tell us in terms of calculating risk? And, what else needs to be done in order to provide a stronger base for an agency Action Plan?

Secondly, exposure and, as Dr. Crawford mentioned, some additional data is being released today. Is that right? It is going to be released today, some additional exposure data?

DR. CRAWFORD: Yes, 110 additional analyses.

DR. MILLER: All right, I wanted to be sure I heard that right. Thirdly, as was pointed out, there are available technologies for reducing exposure to acrylamide and to determine whether or not these are important in order to reduce the potential risk from exposure from this material.

In order to accomplish these tasks we have a number of speakers who will talk about the different subjects, but I think the committee ought to focus on these issues, and we will come back to them at the end of the meeting when we have our general discussion concerning recommendations that we need to make to the agency.

The first speaker this morning to talk about the Action Plan is Dr. Terry Troxell, who is director of the Office of Plant and Dairy Foods and Beverages, CFSAN. Terry?

Development of the Acrylamide Action Plan 

[See presentation slides for Dr. Troxell]

DR. TROXELL: Good morning. We bring you another simple problem to solve, like biotech and Ephedra.

What I would like to do first thing is to go over the charge. You all have the charge and questions in your packets. I think we should focus on that for a minute so we orient the meeting. The charge is to evaluate the revised Action Plan as a tool for providing the scientific basis from which to assess the significance of acrylamide in foods and potential public health consequences.

The first question of the committee is does the revised Action Plan meet its intended goal of serving as a tool for providing a scientific basis to assess the significance of acrylamide in foods and its potential public health consequences?

The second question is new data on acrylamide levels exposure and potential interventions have become available in recent months. Does the Action Plan accommodate these new data? Please comment on the new data, including the exposure assessment of potential interventions.

The final question is FDA's consumer message stresses the importance of eating a balanced diet. Given the uncertainties associated with the current state of the scientific knowledge, FDA has concluded that there is insufficient data to revise this message. Please comment.


For this talk I want to go over the development of the plan and the content, FDA's overall goal and ongoing work on the Action Plan.


This all began back on April 24 of 2002 when the Swedish national food agency surprised the world with a report that acrylamide was in numerous foods, surprised the world partly because cooking process had been thoroughly studied and the mutagenicity, and so on, had been worked on for many years and this one was missed apparently because it wasn't positive in the Ames test.

Basically, what happened then was that many, many countries began developing methodology. The methodology was not initially available. We knew we had an LC/MS/MS method but we basically had to go from scratch, pretty near scratch, to develop a method. Then, as I said, we started thinking through the problem to try to clear up what in the world could be underlying the formation.

We posted our first version of our LC/MS/MS method on June 10th and we have updated it two times. It is either being updated on the web or was just recently updated, in the last couple of days. It is an excellent method of linear quantitation for 10 ppb.

Due to the intense interest worldwide in the subject, the WHO and FAO put on a consultation only two months after the announcement by the Swedish authorities. FDA sent three scientists to this meeting and the consultation concluded that acrylamide was a major concern.

We posted our original Action Plan on September 20th so it didn't have the benefit of the next meeting, which was our interagency round table for which FDA brought together various components of CDC, NIOSH, National Institute of Occupational Safety and Health, and the National Center for Environmental Health. Of course, our National Center for Tox. Research was there and NIEHS of NIH was also at the meeting to exchange information on what was going on with acrylamide research and pathways and work of interest to various groups. So, we began our process of trying to organize the work to deal with the problem. Finally, we had our public meeting on September 30th and that was the first time we presented the plan.


The next event was the JIFSAN/NCFST workshop on October 28th to 30th. JIFSAN is our Joint Institute for Food Safety and Applied Nutrition. It is a consortium with the University of Maryland. Our National Center for Food Safety and Technology is our consortium with the Illinois Institute of Technology. This workshop brought together researchers from many countries, four months after the original consultation, to thoroughly review the status of methods, mechanisms and formation for reduction of exposure, toxicology and risk communication. The workshop led to a list of priority research for each category. In addition to this, JIFSAN is operating the WHO/FAO acrylamide in food information network on the web to try to bring together research from around the world.

The next event was our subcommittee meeting in the snow storm of December 4th and 5th. At that time, of course, we presented the original Action Plan and got very valuable comments from the group.


The Action Plan outlines FDA's goals and planned activities on acrylamide in food over the next several years. The plan discusses intentions to work with other federal agencies and participate in international efforts. We saw from the very beginning that there was a large amount of work to do, and work that would cost a lot.

It was our belief that we could get the answers that we needed as quickly as possible by leveraging our efforts and as important was a means to coordinate work with researchers and communicate throughout the process in order to accelerate the solutions. For example, several labs discovered concurrently that asparagine and glucose through their reaction were the primary cause of acrylamide formation.

It would be useful if there was a way of limiting redundancy in order to optimize our efforts. However, the desire of academics and other researchers to publish gets in the way of that because they want to hold their work until the publication occurs. In any event, the Action Plan and coordination we are trying to foster, we believe, will move this effort in the right direction.


The original Action Plan included sections on testing foods, toxicology formation, methodologies, meetings and collaborative projects, consumer messages and regulatory options. The revised plan, summarized in the next talk, covers the same areas pretty much in more depth, with the addition of a couple of new sections.


The scientific review at FDA and WHO indicated cause for concern, as I indicated. Actually, although we note neurotoxicity here, the consultation did not expect neurotoxic effects from acrylamide levels present in foods. But discussions since then, for example at JIFSAN meetings, suggested that more work was needed to be done to characterize potential neural development of effects in chronic exposure.

Another factor that was involved in developing the Action Plan was that the consumption was in the tens of micrograms range. In contrast to that, food additives and water purification effects of exposures were 100 to 1000 times lower. Also, the scientific review, as I have said, indicated that there were quite a few gaps and a lot of work needed to be done.


The results were intense worldwide interest, as indicated by the desire to have the consultation very shortly. Our optimal desire, of course, would be to harness that intense interest and at least try to influence it toward a better coordination and less redundancy. Research indicates acrylamide is formed through traditional cooking practices. As I said, it appears to be the product of one of the standard cooking reactions that is essential to cooking foods at processors and in the home. The reaction produces desirable flavors and browning. Thus, acrylamide is formed in a wide number of foods and is going to be a challenging problem for chemists.


We developed an overall goal for this project, and that is, through scientific investigation and risk management decision-making, to prevent and/or reduce the risk of acrylamide in foods to the greatest extent feasible.


As far as ongoing work, we revised the plan based on the subcommittee's input as well as other developments along the way. We will present that plan today and will look forward to your assistance in finalizing the plan.


What is going on in the future here? Of course, our work on the projects is outlined in the plan. Then, there is the CCFAC meeting coming up shortly, March 17th. Of course, as we have discussed, there is intense national interest and we expect that the CCFAC will undertake new work at this meeting on acrylamide. As part of our effort to coordinate and share information on acrylamide, we are proposing a formal workshop on acrylamide in conjunction with the CCFAC meeting. The WHO and FAO have adopted that proposal and we will be conducting a workshop at the meeting. Finally, the WHO and FAO JECFA, the Joint Expert Committee on Food Additives, which also evaluates contaminants, will be looking at acrylamide sometime in 2004. Whether it is February or June is not clear but they will be evaluating, and that is the group that does the risk assessment that flows into the CCFAC committee which is the risk management group.

There are tens of meetings on acrylamide going on all over the world, as well as in the U.S., and we have just highlighted a few of those but there is just very intense interest and it is hard to know which ones to highlight and which ones not to highlight.


Anyway, in the next talks we will be betting a report from Dr. Kim on that subject and then I will come back and summarize the revised Action Plan. Thank you for your attention. Any questions?

DR. MILLER: We have some time if there are any questions of clarification. Johanna?

DR. DWYER: Thank you for an interesting introduction. I have one question on one of your early slides where you mentioned that the Swedish report came out in April, and then you developed a method which you released on June 20th. Why was the Swedish method not available?

DR. TROXELL: Again, they released a method but they were waiting to publish the work.

DR. DWYER: And they would not release it to the authorities?

DR. TROXELL: Right, they released a sketch of the methodology which helped accelerate our development of the method pretty quickly and observe it in an appropriate LC Mass Spec. method.

DR. MILLER: Any other questions? If not, thank you. As you know, the organization of the Food Advisory Committee consists of the committee itself and several subcommittees. This issue was referred to the Contaminants and Natural Toxicants Subcommittee, which you have heard about already. Dr. Henry Kim is the executive secretary and he will present the report of the committee. Henry?

Report of the Contaminants and Natural Toxicants Subcommittee (CNTS) Findings 

[See presentation slides for Dr. Kim]

DR. KIM: Thank you, Dr. Miller. As the executive secretary for the Contaminants and Natural Toxicants Subcommittee, I was asked by Dr. Busta, who is the chairman of that subcommittee, to present a report of the subcommittee meeting's findings and it is my pleasure to do so this morning.


For my presentation what I would like to talk about is briefly on the purpose of the subcommittee meeting, as well as the presentation that was made at this meeting, and then talk a little more in detail about the subcommittee's discussions and recommendations that were made.


The purpose of this subcommittee meeting, which was held on December 4th and 5th of 2002 was to present the major components of the FDA's Draft Action Plan and then to seek advice and recommendations from the subcommittee.


In presenting the FDA's Draft Action Plan, the subcommittee was 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 addition, the subcommittee was posed with three questions. That is, given what we know of acrylamide, that is, the toxicology, the occurrence, formation, exposure and risk, one, are the research steps appropriate to describe and address the public health significance of acrylamide in food? Two, are there gaps in the research plan or areas where emphasis should be increased? Three, are there priority research needs that should be addressed first?

In order to facilitate our responses from the subcommittee on the Action Plan, FDA representatives presented five major components of the Draft Action Plan, that is, the toxicology, the occurrence, formation, exposure and risk.


First, Dr. Canady talked about information that we already know in addition to data that we need, and what we are planning to do to obtain those additional data with regard to the five toxicology elements that are shown on this slide. That is toxicokinetics, animal carcinogens and human neurotoxicants, reproductive/development effects and safety and risk assessment.


For the occurrence component of the Draft Action Plan, Dr. Musser discussed the development and performance of the LC/MS/MS method for the quantitation of acrylamide in a wide variety of foods, and also presented results of the exploratory survey data that was collected through November 15th, 2002.


Then Dr. Jackson provided an extensive summary on the current state of knowledge about mechanisms, the precursors and factors affecting acrylamide formation, and identified additional research needs in this area, and then discussed two major elements regarding acrylamide formation in the Draft Action Plan, that is, understanding the conditions leading for acrylamide formation and developing methods to prevent or reduce acrylamide formation.


For the exposure component of the plan, Dr. DiNovi talked about the FDA approach for conducting exposure assessment using the occurrence and consumption data to estimate exposure, and then talked about the two exposure studies that were conducted by Sweden and FAO/WHO and indicated that the FDA was currently conducting an initial exposure assessment which is to be completed shortly.


Finally, Dr. Acheson talked about achieving a balance with respect to the importance of balanced diet, risk from exposure to acrylamide in food and potential dangers of inadequate cooking for addressing the complex issue of reducing risk to consumers from the presence of acrylamide in food.


Some of the other presentations that were made just briefly, Mr. Levitt provided an opening remark sort of to kick off the meeting, if you will. Dr. Schwetz talked about scientific overview of acrylamide in food, focusing mainly on the work that was done at the FAO/WHO consultation in June. Then, Dr. Lineback, the Director of Joint Institute of Food Safety and Applied Nutrition, provided an overview of the JIFSAN workshop on acrylamide that was conducted last October. Finally, public comments were made by a representatives from National Food Processors Association, SNF and the American Council on Science and Health.

That is what happened on the first day of the meeting so now I would like to move on to the second day of the meeting, which was devoted mainly to discussions by the subcommittee members about the various components of the Draft Action Plan, as well as responding to the three questions that were posed to them. So, I would like to first talk about the responses that were made by the subcommittee members with respect to the three questions and then highlight some of the major recommendations that were made on five components of the Draft Action Plan.


In response to question one, are the research steps appropriate to describe and address the public significance of acrylamide in food, the subcommittee generally supported the overall Draft Action Plan as well as the research plan that was outlined in the plan.


In response to question two, are there gaps in the research plan or areas where emphasis should be increased, the subcommittee felt that more detailed information was needed in the plan in discussing the risk assessment, human toxicology and epidemiology studies, as well as animal tox. studies and sampling and analytical variability and food consumption data by various population groups.


In response to question three, are there priority research needs that should be addressed first, the subcommittee generally felt that the priority as outlined in the draft plan, that is, the methodology, occurrence, formation, exposure and then risk--they felt that this outline was appropriate but there were suggestions made that perhaps an explicit statement about the priority in the plan may also be appropriate.

The subcommittee also recommended that the toxicology and risk assessment studies should move forward quickly because these types of studies take a long time to conduct, as well as focusing on developing a rapid and inexpensive method to analyze a wide variety of foods and to provide science-based risk communication messages.

Now I would like to move on to highlighting some of the major recommendations that were made by the subcommittee on various components of the plan.


In the area of toxicology the subcommittee recommended that more human studies should be conducted, such as physiological studies looking at absorption, metabolism, distribution and excretion, as well as toxicokentic studies of ingested acrylamide in humans. They also recommended that more detailed discussion about studies with respect to animal neurotoxicity and genotoxicity should be made; as well as animal bioassay studies, particularly the short-term studies; as well as looking at the dose-response relationships at lower levels, that is, between the no observed hazardous effect level and the levels at which there is tumor formation; and also looking at the mechanism of action in these bioassay studies.

Finally, they also suggested that the Action Plan should look at discussing biomarker-exposure relationship in smokers. They felt that data from those type of studies may be useful in conducting biomarker studies for acrylamide exposure from foods.


In the area of epidemiology the subcommittee recommended that a separate section on epidemiology should be included in the Draft Action Plan, talking more about the human epidemiology studies such as identifying the study populations with higher or lower exposures to acrylamide from various diets, as well as investigating epidemiology studies of workers exposed to acrylamide to determine whether those types of studies may be applicable to food exposure.


With regard to the exposure assessment component of the plan, the subcommittee recommended that the Action Plan should highlight the exposure assessment element of the plan more prominently by providing information about data bases that would be used to conduct the exposure assessment; as well as looking at improving the ability to blind the data that will be released to the public, such that that will facilitate more data sharing; as well as obtaining quality data, and to statistically determine when enough samples have been collected, such that the exposure assessment can be conducted; as well as obtaining inputs from the exposure assessors in the type of foods that should be sampled; and, finally, to use food consumption data by various population groups.


With respect to risk assessment, the subcommittee recommended that more information should be incorporated into the Draft Action Plan, particularly in discussing the importance of conducting a risk assessment, as well as the methods for conducting that risk assessment and on how to incorporate new data as they become available.


Finally, this is the last recommendation section within the area of risk communication. The subcommittee recommended that more emphasis should be made on the risk communication activities by FDA and to provide, as I mentioned previously, science-based information for dietary choices, as well as involving dietetic and nutrition associations in communication efforts, such as the American Dietetic Association and the American College of Nutrition and, finally, to disseminate consumer and cooking messages through extension services.


In summary, the subcommittee spoke positively about the FDA's Draft Action Plan and generally agreed with the approach and the planned research activities that were outlined in the plan, and then provided some very valuable recommendations with respect to toxicology, epidemiology, exposure and risk assessments and risk communication.

With that, I turn the meeting back to Dr. Miller.

DR. MILLER: Thank you, Henry. Are there any questions or comments?

Questions of Clarification

DR. BUSTA: I just want to compliment Henry on doing a very succinct presentation of our two-day meeting. There are other members who were at the subcommittee meeting if there are additions, but I think you summarized in a very succinct manner a lot of comments.

DR. KIM: Thank you.

DR. MILLER: Dr. Torres?

DR. TORRES: We heard a lot about variability within food, within food batches and then I see that the plan does include studies on the formation of acrylamide in foods. But when I look at the subcommittee recommendations there seems to be nothing in that area of formation and I am a little bit surprised about that. I don't see any specific recommendations about the formation kinetics of acrylamide. We see so much variability within food, between foods, between batches of food, however, I don't see any recommendations for further studies on formation.

DR. KIM: Yes, I did mention in the formation slides that Dr. Jackson presented formation information and did go into some of the additional research needed, such as looking at factors like pH, temperature, moisture content and those types of research activities. But I did not kind of highlight that in this presentation because of lack of time to present everything that was discussed at the meeting. There was a whole lot of discussion on all the major topics and I just wanted to focus on the major components of the plan.

DR. MILLER: I think the point is that this issue which is, of course, vitally important wasn't highlighted as one recommendation for the Action Plan in your presentation.

DR. KIM: Well, I can add that to the recommendation list.

DR. MILLER: That is the point that Dr. Torres was trying to make. It is a vital issue and is certainly deserving of the same emphasis, at least in my view, as other aspects such as toxicology.

DR. KIM: Yes, I would agree with that.

DR. MILLER: Do you want to say something, Terry? You look anxious.

DR. TROXELL: I always look that way. Our original Action Plan did highlight research on mechanism and formation. So, I think the subcommittee recognized that and I know that no recommendations came out on that, to change that. They just agreed that we should keep pushing on those aspects because they are vitally important.

DR. MILLER: I think the report ought to reflect that. It is an area that obviously got missed because in reading the report, it doesn't come across. Johanna?

DR. DWYER: Thanks for a good report. One thing that troubles me about this whole area is the database that one is using to get at the foods. Just a cursory reading of the report we were given this morning doesn't seem to highlight that for further attention. My concern is this, I know that on page two or three of the Action Plan you talk about FDA doing some analyses and then we were told that today more values will be released, but the question is how to capture all the values that are there, not just these but others. We went through this with methyl mercury where we found out that there is a lot more data out there than we had in the database. I wondered what efforts would be focused on all the different government agencies that may have data on this.

DR. KIM: I was told that this will be addressed in the revised Action Plan. There were discussions about data collection. I think with respect to FDA--we obtained data from surveys--I don't know, maybe Dr. Busta or Dr. Lee may have some additional thoughts.

DR. DWYER: I am talking about the presence of this compound or compounds in food, not the further step of taking those data and trying to get exposure estimates for food. I am talking about the database that is used to say the French fries have this much, the broccoli has that much.

DR. BUSTA: I thought Henry covered that under the exposure assessment in improving the ability to blind the data and facilitate the data sharing, and maybe that should be more explicit but when I see that I think that there a lot of data in the food industry that, if they were made blinded, would be made available but might not be offered freely if it said somebody's brand.

DR. LEE: Just to amplify on the subcommittee's recommendation, a lot of time was spent on the JIFSAN website and ways to share acrylamide data, and there was also a lot of discussion about the need to at least single-blind acrylamide levels in food so an industry that has a lot of data on acrylamide content of its products would not be in danger of being fingered as a source in the diet. So, I think that thought is there and, in all fairness, Henry gave a great summary but couldn't get into that level of detail in his presentation.

DR. MILLER: Other comments? This is not a clarification, Henry or Terry, but how much emphasis--excuse me, let me go back, I think it was you, Terry, who said something about acrylamide not being positive in the Ames test? Did I hear that correctly?

DR. TROXELL: That is my understanding, and when they did research on cooking carcinogens that form in foods, and so on, apparently that is probably the reason acrylamide was missed among the other chemicals that are formed.

DR. MILLER: But it is genotoxic?


DR. MILLER: That is what I thought. That is why I didn't understand why it didn't show any results the Ames test.

DR. TROXELL: If you want further clarification, Dr. Canady probably can provide some.

DR. MILLER: It is a matter of curiosity, but thank you. Any other comments? If not, we are actually ahead of time. I think what we are going to do is--Terry, why don't we move on?

MS. DEROEVER: We have a technical difficulty.

DR. MILLER: We have a technical difficulty? All right, we will take a break and return in about 30 minutes.

[Brief recess]

DR. MILLER: Our next speaker is Dr. Terry Troxell, who will describe the revised Action Plan. Terry? I assume these are revisions that were made in response to the recommendations of the subcommittee.

Revised Action Plan 

[See presentation slides for Dr. Troxell]

DR. TROXELL: Right. Thank you. You should have the revised Action Plan before you and also those 110 new data points should be in your packet. They are kind of at the bottom of the packet, under "exploratory survey."


I am going to go very briefly over the revised Action Plan that we updated based on the input from the subcommittee, as well as the public, and also all the things that have happened in the meantime. Like I said, we put out the first Action Plan before the interagency meeting that we had, also before the JIFSAN meeting, and so on. Again, what we are looking for today is we are seeking your input to assist us in finalizing this.


What I am going to do is summarize key changes; review the major goals of the plan; walk through the action sections; and then highlight the changes section by section.


The major change, of course, is that we reorganized the plan. It now has a more logical flow, at least from my viewpoint. We have added new sections on exposure assessment, epidemiology and risk assessment and this is in response to subcommittee recommendations. We added more details and updated the information. The plan is about double the size of the previous plan and this also was in response to the subcommittee.


There are seven sub-goals to our work. I already mentioned our overall goal before. The first sub-goal is to develop rapid and inexpensive screening methods and validate confirmatory methods of analysis. The second is to identify mechanisms responsible for the formation of acrylamide in foods and identify means to reduce acrylamide exposure. Again, we know at this point that asparagine and glucose with high heat are responsible for most of the formation of acrylamide. What we don't know at this point is how to quench that reaction.


The next goal is to assess the dietary exposure of U.S. consumers to acrylamide by measuring the levels in various foods and estimating the dietary exposure.


The next goal is a rather long one and I am not going to read this but basically what it says is that we are going to explore the toxicology and epidemiology, and we are going to use quantitative risk assessment to determine the risk and the uncertainty associated with those risk calculations.


The next goal is to develop and foster public/private partnerships to gather scientific and technological information and data for assessing the human risk. We believe this is really essential and at the core of getting this work done and also it will harness that interest around the world by all parties, academia, industry and the governments.


The next goal is to inform and educate consumers and processors about the potential risk associated with acrylamide throughout the assessment process and as knowledge is gained.


The next goal is to provide all the essential elements for risk analysis, that is, risk assessment, risk communication and risk management. That is kind of a sum-up goal because that kind of covers the waterfront of what we need to do on this process and pretty near in any project we have on food safety.


There are nine sections detailing the actions toward accomplishing these goals. They cover the methodologies, research on formation, measuring exposure, toxicology and health effects, epidemiology, risk assessment, meetings, inform and educate the public, and then further actions. I will now look at each of those sections.


As I said, our LC/MS/MS method was posted on the web back in June. We have recently updated it. Dr. Musser of CFSAN and his group have done an enormous job of developing the method and analyzing around 400 samples to date. This is in addition to the numerous analyses they have done in determining the characteristics of the method.

I am going to diverge a second to make a point. Acrylamide has been a real team effort here, at FDA. It is not only my office, Office of Plant and Dietary Foods and Beverages, that is involved, but the Office of Food Safety is doing the exposure assessment, the Office of Systems and Support, the office in which Dr. Musser has done the analyses, and there are others in CFSAN like the food safety staff and the Office of Science. In the FDA in a broader sense, our National Center for Tox. Research is doing an enormous amount of work on toxicology and our Office of Regulatory Affairs will be running total diet study samples. This gives a sense of the energy going into developing the science around this issue, and is being repeated in many countries by many different kinds of parties.

With respect to validation, while there are certain excellent methods for determining acrylamide, Dr. Musser's method is an excellent definitive method that we believe should go through the full AOAC validation process.

Then, there are screening alternatives. We want to explore, and are encouraging others to explore the screening methods. Dr. Diachenko's lab in CFSAN is presently looking at use of LC/UV as a screening method. We believe the equipment, the LC/UV detectors are available in many labs around the world and this could provide a simple means for people in many countries to get a handle on the screening levels.


So, what is new in methodologies? Our second update is being posted for our LC/MS/MS method. We are explicitly talking about AOAC validation rather than just validation and we are currently exploring LC/UV alternatives.


Research on formation, CFSAN's Division of Food Processing and Packaging is located in Chicago at the National Center for Food Safety and Technology. NCFST has the capability of doing pilot process research so we can take the bench-top research on reduction, whether done in academia or done in industry labs, and test it under pilot processing conditions. Currently, NCFST is planning research on the formation of acrylamide during home cooking of toast, French fries and other foods. This seems to be a gap that we don't believe others are pursuing. The laboratory out there has the capability of measuring surface temperatures of food and also looking at the degree of browning so we can actually bring some science to bear on what would be happening in the home with toasting and frying of foods.

Also, the interaction of academia and industry, again, we expect that reduction strategies will vary with many types of food in which acrylamide is formed. Therefore, it will take the efforts of the food science departments like Mike Pariza's Food Research Institute at the University of Wisconsin and the food industry to do the heavy lifting on research in this area. You will hear from Frito Lay and Procter and Gamble this afternoon on their progress on reduction strategies. Procter and Gamble will talk about the mechanism, Frito Lay will talk about reduction strategies.


So, what is new in the research on formation? NCFST is going to investigate scientifically home cooking.


Now to get at some of the questions we had this morning of measuring exposure, as I think I said earlier, Dr. Musser's group has done about 400 locally collected food samples thus far with this additional 110. These are the analyses we have used in our initial exposure assessment that you will be hearing about later, and we have handed out that data. I hope you each have a copy of that.

We are contracting through JIFSAN, our consortium with the University of Maryland, for 400 to 500 more foods to be analyzed, and these will be collected across the country, so we will get a national scope now, by our field forces.

In addition, we have begun testing total diet study foods. The total diet study look at foods as eaten and serves as a good indicator of exposure to contaminants--I mean, that study looks at contaminants and pesticides, and looks at nutritional minerals, and so on. So, it will be a means for us to understand what the kind of gross exposure is and then what our progress is on reduction.

As far as other testing goes, we plan to continue testing in subsequent years in part to fill in the distributions. You will see in our exposure assessment presentation that there are too few samples in many categories so we need to fill those in. As was discussed earlier, JIFSAN is compiling data from industry and academia which will fill gaps in distribution and help solve this data problem.

We also understand that Europe has compiled a large amount of data. The U.K. apparently has a database that may run up to 4000 samples but their processing conditions may be somewhat different than processing conditions in the U.S., therefore, they may not always be directly applicable to the U.S. situation. But we think it is really important for us to tap into the data that is out there from all parties to basically fill out the distributions for each of the food categories and understand the variability on exposure.


Then there is exposure assessment and I am going to pretty much skip over this part because Dr. Robie will be thoroughly discussing our methodology but, again, we will be periodically updating that exposure assessment as more data is compiled and, hopefully, we will be bringing to bear this blinded data that industry can make available to us.


So, what is new in exposure? Well, it is a new section and we put in a lot more details in the revised Action Plan on what we are going to do, and those details include information on how we are going to model the database we are going to use.


Now we move on to toxicology and health effects. Toxicology will provide information for risk assessment and consist of two general areas, animal toxicology and human toxicology.


Now I am going to pretty much summarize the next three or four slides with the following: There are basically four things we want out of this work. One, we need a better understanding of the carcinogenicity, neurotoxicity and germ cell toxicology in rodents.

Two, we really need to understand the bioavailability of acrylamide from food versus water in which most of the studies have been done.

Three, we need to understand the difference between high and low dose metabolism at high doses and, again some others may need to correct me because, as they always say, I am not a toxicologist, I am only a "Troxell-cologist." So, at high doses the p450 pathway produces glycidamide which is thought to be the penultimate carcinogen, but at low doses a smaller fraction of acrylamide may go through the p450 pathway. So, it is really important for us to understand the high versus the low dose differences.

Fourth, we need to understand the differences in metabolism and processing of acrylamide between animals and humans.

For carcinogenicity we need to understand the levels of adducts that form on ingestion of acrylamide. There is a very convenient biomarker adduct that is formed with hemoglobin. Once the relationship is established between acrylamide intake and adduct formation we will be able to use the adduct levels in red blood cells as a convenient means of understanding population exposure to acrylamide. Of course, that will integrate everything. That will integrate the acrylamide from smoking and occupational and all possible sources.

We don't really believe that this adduct with hemoglobin leads to adverse effects. In fact, it probably prevents acrylamide from ultimately reacting with the DNA. It is the DNA adducts after enough insults to the DNA that are thought to lead to the tumor production. So, we need to also understand the correlation between acrylamide intakes, the hemoglobin adducts to the DNA adducts, and we need these studies for rodents where we can study cancer production at the highest doses and we need these studies on humans who are exposed to very low levels from foods. Such studies should significantly reduce the uncertainty about the risk to humans but, so everybody is clear, we never are going to be totally precise in this. There are still going to be fair uncertainty bars but we can reduce those uncertainty bars by quite a bit.

The challenge is going to be to measure adducts in situations where there are low levels of exposure because, obviously, at low level exposure you are forming very few adducts and, therefore, you are having difficulty pulling those out in your analysis. Many of the studies that we list on the next several slides can contribute to these key needs in a relatively short time for toxicology. To me, a relatively short time for toxicology is one to two years. So, with that having been said, I think I have summarized the short-term studies at the National Center for Tox. Research on bioavailability and adducts.


Then, they are going to be doing a gold standard, you know, NTP bioassay. The subchronic and mechanistic components of that will be done fairly fast and will contribute to our understanding of the toxicokinetics and can be pumped into physiological-based pharmacokinetic modeling. But obviously definitive results on number of tumors that are produced in the rodents won't be available for many years because that takes a lifetime of study in the rodent.

The NCTR will be looking at neurodevelopmental effects and also there will be mechanistic research on germ cell toxicology and neurotoxicity by the National Institute for Environmental Health studies.


Then there will be the adduct studies that CDC is going to be party to, as I mentioned.


We are looking at using the NHANES study and measuring adducts to understand population exposure to acrylamide. As well, NIOSH is looking at this area and the acrylamide manufacturers are doing a variety of toxicokentic studies. So, we are trying to bring all parties to bear in the process.


Before I leave this area, again, there are many groups that are going to be expected to contribute to this work, and this work costs millions. We are trying to facilitate the research that needs to be done in all areas, but outside the agency it will be up to the researchers, their priorities, and how much they can invest to see how far they go. This will be particularly true for doing epidemiology studies which I will discuss a little later.


So, what is new in the tox. area? We have listed a series of new toxicology studies, new areas, neurotox. and germ cell tox. and industry working on toxicokinetics. We have talked about doing PBPK modeling and so on. Since the early report, we now know that the NTP is going to be doing a chronic bioassay.


Let's move on to the epidemiology area. Some of the work on adducts is viewed as epidemiology so it is kind of cross-cutting. Of course, we are going to be exploring the feasibility of prospective studies and also we will be using all the studies that are reported in the literature, whether they are going to be on worker exposure or food exposure.

Now I will go to a little description of my view of the epi. area here. There are basically two avenues of epi. studies in my view, the occupational and the food exposure. If I understand it right, occupational studies are most likely to be able to tell us that acrylamide at high levels can cause cancer in humans. I think occupational exposures may be the highest. Thus far they have not shown an effect. This may be due to the fact that not enough years of exposure have been studied.

The other avenue, of course, is intake from foods. Can epi. studies make a difference? Dr. Acheson will discuss the Mucci study, which was published several weeks ago, in his talk. I believe that is tomorrow.

This was a case control food intake study and no effect was seen. The basic problem, as I see it, is that we should have an epi. study that can detect a lifetime risk of 1/10,000 and we don't want to wait 50 years for the result. If the study can only detect 1/1000 risk then we are not any better off than what the rodent studies can give us. So, I have asked our experts if it would be possible to detect such low risk. I don't have the answer yet but the initial reaction is that it would be very hard, at best.

The kind of study we have been discussing at the FDA is to do a case controlled study within one of the ongoing prospective studies, like the Nurses Health Study. They have collected diets as well as blood samples for the study. Many pieces would need to come together in order to do such a study but, if they did come together, it could be done in one to two years.

But consider the complications. How accurate are the diets? Misclassification would reduce the power of the study. Further, because acrylamide is in so many foods, it will not be possible to find a population that is exposed versus not exposed. What we are talking about is groups that may vary in levels of exposure by maybe 10-20 microgram differences. Then, there is the further complication with food in that an array of chemicals make up the composition of any foods. So, if one sees an effect or doesn't see an effect, is it because of acrylamide or is it because of other chemicals that are also present in the foods that are being looked at which have carcinogenic and also anti-carcinogenic effects?

So, there are a lot of difficulties in looking at these food studies because of the confounding factors and many other factors that are also involved concurrently in causing various effects. So, in sum, we are exploring epidemiology but we need to carefully consider any study to determine if it will likely provide added value once it is completed.


So, what is new in epidemiology? Well, it is a new section. It explains how the epi. data could benefit FDA work on acrylamide, and a number of studies and collaborations are under consideration.


The next action section is risk assessment. Clearly, when we have adequate information we want to characterize potential risk, including the uncertainty analysis. We think that key data needs are bioavailability, biomarkers, metabolism and toxicokinetics.


We would revise the risks assessment when significant developments materially change the assessment.


So, what is new in risk assessment? We have given an explicit section to risk assessment and it outlines our goals, data needs and expected output of the risk assessment.


The next section is meetings which we will continue to participate in and convene meetings as appropriate. As I said earlier, we convened the interagency meeting and we participated in the JIFSAN meeting, and so on. I am not repeating the meetings here because I went over the key things that have been going on in the first presentation.


The next action section is to inform and educate the public. We are, and have been, committed to communicating with the public throughout this process. This is the third public meeting. So, that is one form of our communication and transparency on the acrylamide issue. Also, our website has a lot of information on our Action Plan, on levels and so on. Also, there has been an FDA Consumer article published in the January/February issue.


We have a message and do not believe that we have enough science at this point to change it. That message is eat a balanced diet; choose a variety of foods that are low in trans fat and saturated fat and rich in high fiber grains, fruits and vegetables. We may recruit diet and nutrition organizations to help spread that message and changes to the message.


What is new in this section? We have expanded it to indicate that we have a continued process of risk communication going on and participation of outside organizations.


The last section is further actions. One of our further actions, of course, is to develop and revise regulatory options as additional knowledge is gained on acrylamide in food. Another action is to encourage industry to adopt feasible, practical and safe processes that are successful at reducing acrylamide, as needed.


Another action is to develop and revise the consumer message about dietary choices and cooking, as additional knowledge is gained to assist consumers in making informed choices. Finally, any adjustments to messages would be made as dictated by the totality of the science.


In conclusion, the Action Plan outlines goals and planned actions. It was revised to reflect comments of the subcommittee and new information from many sources. It was reorganized, with new sections added and more details. Finally, we are here seeking your input to assist us to finalize the plan. Thank you for your attention.

Questions of Clarification

DR. MILLER: Thank you. Before we begin the questions, let me just introduce Jean Halloran. Comments or questions?

DR. RUSSELL: Yes, I was wondering if it is known which p450 enzymes are involved in creating the ultimate carcinogen. The reason I am asking that is that you might be able to target certain populations whose p450 enzymes, for example, are turned on or induced for a particular effect of this carcinogen because more carcinogen would have been formed. One example would be people who drink alcohol where certain p450 enzyme profiles are turned on, and if those enzymes are the same ones that are involved in forming the carcinogen you may be able to target more supopulations that may be at particular risk. So, the question is do we know that or is that a part of the puzzle also?

DR. TROXELL: I don't know that. Dr. Canady may be able to comment on that.

DR. CANADY: Rick Canady, FDA. 2E1 is the enzyme and it is pretty clearly implicated in the conversion of glycidamide and it is one of the things that we are looking at in designing studies.

DR. RUSSELL: Another comment is that you mentioned in the epidemiology, perhaps just as an example, that the Nurses Health Study might be a population to look at. My only caution there would be that they are not a particularly representative group because they have other health habits. They may be more health conscious, for example, than the normal population that you are dealing with. So, it may be that a longitudinal population that has been followed that is more representative, perhaps a Framingham study for example, might be a better population group to look at.

DR. TROXELL: Thank you.

DR. SCHERER: I am just wondering whether any consideration has been given, or discussion, in terms of expanding a bit the risk communication section. For example, it seems to me that the idea of communicating all of this to the public is certainly a sound one but it seems to me that maybe a part that is missing is more of the preparation for how to do that. This particular risk, it seems to me, is in some ways potentially, if the science bears it out, what might be described as the perfect risk, risk challenge. It has all of the elements of being one of the more difficult ones to communicate. So, the idea would be that to some extent there needs to be some work done on understanding how to change human behavior in this case because that seems to me to be the real challenge.

DR. TROXELL: I think we would agree with you that it is always a challenge to provide messages and education that consumers will respond to. That is the challenge of any message and in this case probably more so because consumers hear many negative messages and they are puzzled many times about which ones are really important to listen to. So, yes, a study of what would work and what won't work is important. You know, if we are going to go out with a targeted message, for example relating to cooking, we certainly want to make sure that what we do has scientific underpinnings like the toasting issue. We clearly don't want consumers to hear cook food less generally because then we would end up having more problems from pathogens. So, we would be reducing a risk from this a little bit but we could be creating an actual health hazard from people not cooking their hamburgers well enough.

DR. TORRES: When I look at where most of the studies will be done on formation, it seems that the focus is on the National Center of Food Safety. So, I have two questions. One is that most of the research at that center is fee-based access. So, if you pay a fee you have access to the research. So, the question is how accessible would that research be to any industry?

Second, you mentioned that the focus should be on home use study so where would the non-home use study be done?

DR. TROXELL: Actually, most of the FDA research will be at the National Center for Toxicology Research. They are undertaking the NTP bioassay and all the mechanistic work. As to the studies relating to reduction, the National Center will do some of that work, and has the capability to do some additional work on pilot process testing, and that work will be fully available to everybody as early as possible and generally is presented as early as possible.

Yes, I think the Center does some private work but our Division is not involved in that kind of work out there. That is work that is between the researchers at IIT and particular industry members. We will only do a portion of the formation research. We don't have the capacity to do a large portion. We really need to depend on other organizations, other governments, industry and so on to push the research on reduction forward. We also expect that is necessary because food is so complex we will probably need different strategies for different foods. While we can hope for a magic bullet to quench the reaction, that is something that I am not aware of anybody having a handle on at this point.

DR. TORRES: My question specifically is will we hear more details about the research done at the National Center for Food Safety. I agree with you that they do have the capability to do pilot studies and they have the equipment to simulate industrial processes, but I don't see that anywhere in this report.

DR. TROXELL: Well, part of our effort is to make information available as early as possible to help fuel the process of discovery. Obviously, the FDA is committed to that so we will make any discoveries available as soon as possible. Dr. Jackson, at the subcommittee meeting, presented on the mechanism and also showed some initial results on potato chip work. So, we are making our research available as soon as possible.

DR. MILLER: Johanna?

DR. DWYER: Thanks for a good presentation. Just a couple of points, the first is I was delighted to see that you are using HANES, or are hoping to use HANES, but the number of subjects above the 95th percentile will be very small, as you well know, because the sample size is only 5000. But the notion of using blood and connecting it with diet and other exposures is very appealing.

I was taken by Dr. Russell's comment about longitudinal studies and Framingham is one example of a place where other agencies in the public health service have invested considerable money in maintaining sample sizes. The other one is the Women's Health Initiative which, not surprisingly, eliminates men but is very large, about 35,000 people I think in the randomized study and I think the observational study is even larger. That, again, is heavily supported by the NIH. It is one of the largest studies in the country. It would seem like you would also want to explore that particular study when you look at feasibility of prospective studies. The advantage of that particular study is that the data and the assumptions that are used in the food frequencies are totally transparent because it is all federally funded, and basically the forms they use are things that were developed I think at NCI.

The second point is simply this issue again of getting all the data that are available on acrylamide levels in food into your database. It sounds to me like you are spending at least a million dollars, if not more, to try to do these assays of food within the agency. I haven't looked at the FDA budget lately but it strikes me that it is a lot of money, and if there are others who also have data on this we have to find ways--and food composition in general--to get this data into national databases for risk assessment.

DR. TROXELL: Thank you for those comments. Getting the data together on levels was a major discussion point at the JIFSAN meeting. It looks to me like they are going to be able to compile, hopefully, an enormous amount of data that is going to fill in the gaps for us substantially.

Clearly, we are going to have to look at individual foods to see what kind of data we have for those distributions and where we need to fill them out. We will be consulting with our statisticians to make sure that we have adequate information. But, as you know, you can do an exposure assessment with one data point and it is just how good is that exposure assessment. So, we ultimately want to understand the distributions from the major foods.

DR. MILLER: More comments? I have just a couple of points. Clearly, some of this work is already in progress. Can you give us some idea of what the timeline is for your priorities about what is going to be done first?

DR. TROXELL: I am sorry, I don't have a timeline on each of these studies. Much of the tox. work is either in the planning stage or has already begun. There is short-term work from which within a year or two we should be reaping benefits. Of course, the bioassay is going to be very long. We have just put in a proposal for the home cooking work for the National Center. It is going to take some time, I am sure, for JIFSAN to compile the database and so on but that work was begun a while ago to try to get the different players to submit some of their analyses.

We are--what?--ten months into this project. We started without a method and, obviously, the first thing we need is to be able to detect and to find out what we are dealing with so we are pushing forward as readily as we can. I don't know if that is enough. Dr. Canady may have some additional information on timelines.

DR. MILLER: Well, it just occurred to me that we are looking at an action plan that may already be in progress. Unless there is something seriously wrong, other than beating a dead horse, there isn't much you can do about changing it. The committee made a decision that they would like to see you modify some of the activities but if a study is already under way it would be difficult to change.

DR. TROXELL: Well, I think any recommendations coming out of the committee would be carefully examined to see if something should be modified in approach to a short-term study or other studies. Things are not that far down the road, I believe, that they cannot be altered and, I mean, we are here to receive this input. It is important to us, very important to us that we get it as right as possible at this point so we have what we think is a good framework, a good template for proceeding, not that we won't need to evolve that template in another year.

DR. MILLER: The other question I had concerned funding. How are you, or are you, attempting to get other groups, particularly university research groups, involved in some of this research in order to expand the research base?

DR. TROXELL: We have tried to do some extramural proposals. I don't know that we have been successful in getting those through. It is not clear to me how much we are going to be able to accomplish as, you know, the budgets are not growing at this point; they are decreasing. So we are hoping to tap into I guess other people's budgets.

DR. MILLER: That is what I meant, like NIHS.

DR. TROXELL: Right, we will work on that but, again, that isn't in our control. We can control the agency's dollars and where they go. We can work through the Department to try to get additional funds focused on this and, hopefully, we will be successful in that.

DR. MILLER: Have you had discussions with NIHS on this issue, or any of the NIH institutes?

DR. TROXELL: Well, they are definitely interested in looking at the neurotox. and the germ cell part of the work so that is in progress.

DR. MILLER: Dr. Lee, yes?

DR. LEE: Terry, since we are on this subject of "Troxell-cology" I would just like to ask perhaps philosophically what is the long-term outcome? Do you think we will ever get to a point with acrylamide that no more information is needed?


DR. TROXELL: Well, in the sense of good "Troxell-cology" as with all toxicology, we always need one more study.

DR. LEE: Right.

DR. TROXELL: If we can nail down those essential needs of bioavailability, the high dose/low dose and so on, that is going to really help us get a handle on what the potential human impact is. Again, trying to understand high dose/low dose is going to depend on some pretty sophisticated chemistry trying to measure adducts. This is science and, as you know, you never know exactly where science is going to take you and how far you are going to go with it, and we are going to push it as well as we can and certainly some of the best researchers in the country are measuring adducts and you will hear from one of them later, Dr. Fennell, on adducts and we will see how far they can go with this work.

DR. MILLER: Johanna?

DR. DWYER: I am concerned a little bit though about singing from the same hymn issue of all the government agencies working on this. I know you are trying very hard to coordinate the work, and so forth, but I am just concerned that we don't sit here in a year or two and hear three different agencies, all telling us different things about the exposure assessments they have done, and so forth. And, what kind of guarantee is there that this isn't going to happen? I am also concerned about Dr. Miller's question of the time frame. How do we know if the time frame has been revised if we don't know the time frame to start out with?

DR. TROXELL: Well, as far as the exposure assessment, the FDA is responsible for the exposure assessment I believe so we will be doing that. We are doing our best to coordinate the work and all I can say is we are working very hard at it. We had an interagency meeting. We are going to go back and have further meetings to try to ensure coordination. There have been meetings along the way on the neurotoxicity. Dr. Canady attended a meeting that had a session on acrylamide in November. We will be doing a meeting at some point this spring, I believe, on the germ cell toxicity, and so on. So, we are making every effort to coordinate the actions and we hope, you know, that it pulls together. I don't want to provide any absolute guarantees because obviously it is not all under the control of FDA but we will work our best to keep it coordinated and on track.

DR. MILLER: Have you made an attempt to develop some kind of a consortium of those people who are working in this area? Because if you just have meetings periodically there is always a delay in this thing but a consortium with different groups given the responsibility of coordinating information on different aspects of the problem might be a way of getting the data distributed much more rapidly.

DR. TROXELL: The closest thing to a consortium is the work of JIFSAN to coordinate the efforts.

DR. MILLER: Coordinate which efforts?

DR. TROXELL: They are basically coordinating all aspects and gathering the information together on all aspects internationally to make sure that all information is available to everyone. That is more of an information exchange kind of system. So, I guess I would have to say, no, we don't have an explicit consortium to work on this. The FDA, because this is a problem in FDA regulated foods, is basically taking the leadership to coordinate the efforts among the agencies and to do what we can to leverage the academia and industry. So, in a sense, we are the head of the consortium to explore acrylamide.

DR. MILLER: Yes, Dr. Mehendale?

DR. MEHENDALE: I was pleased to see that you mentioned the high dose/low dose bioavailability studies as being very critical. I was hoping that there was a timeline on those studies. But it seems like the results of those studies would help us to at least get the direction of the larger issues I think.

The second point I wanted to make is I didn't see anything in what you mentioned, if you are considering any special populations. A couple of them I can think of, You know, just last week we had an issue of Science devoted to obesity and I know lots of people are considering diet restriction or caloric restriction. One of the things that happens with caloric restriction is induction of Cyp 2E1. That is a special population or special physiological condition. Diabetes is also thought to be associated with some induction of Cyp 2E1, and there may be other special populations. I wonder if you have any comments on that aspect. I think considering these populations might also be helpful when you are looking at more sensitive populations and, therefore, it might be helpful in arriving at safe levels.

DR. TROXELL: Number one, I don't have any further information personally on timelines, other than to expect that the toxicology on the short-term studies, which can give us some insight on high/low dose bioavailability, are going to take one to two years. Do you have anything further, Rick?

DR. CANADY: With regard to the toxicology studies, we have already got some results. For example, work on adduct formation, DNA adduct formation has already been developed. Bioavailability and studies along those lines with regard to toxicokinetics, some of the initial findings I guess we will have within the next six months to a year. We have proposals that we are currently reviewing on the specific studies but some of the studies as you can imagine, for example bioavailability, are relatively short-term studies comparing area under the curve for different routes of exposures and that is something that can be done fairly quickly. So, those studies with regard to toxicokinetics we can expect to see, as Terry was saying, in the next year or two but probably some of the initial results will be in the next six months to a year. Again, some adduct studies are already available.

I am not sure how else to give a more specific deadline or specific timeline, other than to say we have proposals and we have specific information on modes of action we are evaluating within those timelines. Is there more detail you would like with regard to that? It is not like we can give you that the bioavailability studies will be done by June or--

DR. MILLER: I think the question was in terms of priorities, what is being done and what is going on. I think you have done that.

DR. CANADY: Good. With regard to the special populations, clearly in considering the prospective studies that is a very important determination, very important source of confounding or a way of designing the studies. Glutathione transferase, Cyp 2E1, polymorphism with regard to those inductions are things that are clearly important to consider when thinking about how you might go about doing those studies, and we are well aware of that.

With regard to specific animal studies, NIEHS has already initiated some studies with Cyp 2E1 knock-out mice to look at the conversion of acrylamide to glycidamide and looking at the mechanistic interpretation of that information with regard to neurotoxicity, germ cell toxicity and also DNA adduct interactions. NCTR and NHANES are working together to develop that line of mechanistic research. Specifically, NCTR will do the DNA adduct work for that set of experiments.

The NHANES studies, as you know, sampled some specific populations but they weren't designed to specifically look at acrylamide issues so maybe others can comment more specifically with regard to how NHANES would be used to look at specific populations. You would have information that would be relevant to intake but whether it is specifically relevant to caloric restriction, probably not. Whether it is specifically relevant to foods that are high in acrylamide, that is something we need to evaluate still. But the bloods have been set aside to do specific adduct work, the bloods that were taken earlier this year, in fact, for NHANES so that information would be worked into the study design to the degree it could.

Results of separate studies being considered by CDC and CEH are intended to look at sort of add-on acrylamide dosing or foods that have acrylamide and through that we could evaluate some of the questions that you are talking about.

DR. MILLER: I think this issue of stress populations, particularly for something as ubiquitous as this is, would seem to me important enough to mention as part of the Action Plan. I am not exactly sure how it would be done but it seems to me it needs to be considered, and more general stress situations where we know there are metabolic changes and inductions that occur in response to stress.

DR. TROXELL: See, it makes a difference when you have a real toxicologist instead of a "Troxell-cologist" to provide the answers!

DR. MILLER: Any other comments or questions? Yes?

DR. RUSSELL: Jumping ahead to risk communication because I don't see any other time to really comment on this, I think it is a good idea to have the dietetic nutrition associations in concert with you in communication efforts, but you will have to cast the net pretty wide for these nutritional organizations. It is not that you can just pick one or two or them. There is a whole group of them that have different kinds of missions, if you will, but they have very important audiences and very important journals, a couple of them, even though the societies are pretty small. So, you need to really cast the net pretty wide there. You have mentioned one very big one, the American Dietetic Association, and one very small one but you need to go much beyond that.

DR. TROXELL: Thank you.

DR. MILLER: Yes, Johanna?

DR. DWYER: Specifically, I would suggest that you go to the American Society for Clinical Nutrition, the American Society for Nutrition Sciences and certainly IFT. I believe Dr. Russell is currently the president of the ASCN.

DR. TROXELL: Thank you.

DR. MILLER: Have you made a commitment, Robert?

DR. RUSSELL: Well, that journal reaches a large audience.


DR. SCHERER: I wonder whether there has been any thought or discussion about the potential problems that the agency faces in terms of release of information? I certainly support the idea of open communication but given the nature of the wide range of studies that are being done, it is likely, it seems to me, that at least one of those studies at some point will have implications for serious health concerns, whether that is eventually supported or not but the possibility exists. So, the challenge, it seems to me, for the agency is how do you respond to that information that is likely to be picked up by the media and, as we all know, perhaps be exaggerated? But the idea of trying to put it in perspective of the wide range of studies that are going on, whether there has been any thought given to that process?

DR. TROXELL: Well, for every one of our meetings we do a lot of thinking about how appropriate the message is about acrylamide, and we will clearly bring a lot of energy and information to bear from our people who do consumer messages, and so on, in the chain who understand risk communication we bring that kind of to bear.

I think also, as you suggested, it is important for us to think about how we are going to communicate what would reach consumers, and I think that is a valuable comment.

DR. MILLER: Actually, that is an issue that really needs special emphasis because I have been trying to think if there were any other materials that I could remember that are so ubiquitous in diets everywhere in the world so any culture that produces a dough baked product is going to be exposed to some level of this material and how do you deal with that? That is going to be a big problem, it seems to me, as a research issue and it needs to be emphasized. I think maybe it requires going back and looking at it again to reemphasize it again.

DR. TROXELL: Thank you. I think these are valuable points.

DR. MILLER: Any other comments or questions? We are caught between a rock and a hard place in terms of timing. We are substantially ahead of time and I was thinking of going on to our next speaker.

DR. TORRES: Maybe I could ask one more question.

DR. MILLER: Please, do.

DR. TORRES: On the subject of cooperation with different groups, one thing that really caught my attention was the message that the U.K. had done analysis on 4000 food items. That seems to be a lot of work. Could you explain to me why there is a difference between 100 in the U.S. and thousands in the U.K.?

DR. TROXELL: Well, I don't think the Central Science Lab of the U.K. has done 4000. They have compiled 4000 from industry and around Europe, as well as their own analyses. It is just the level of progress we are at in different areas.

DR. TORRES: Thanks.

DR. MILLER: Terry, there is another question that I meant to ask before, given the different methods--well, let me ask you this question, how many different methods are being used by the different laboratories?

DR. TROXELL: That is a good thing to dwell on for a minute because even with the most elegant methods, like the LC/MS/MS method, it is pretty easy to mess up and get erroneous results. But there is LC/MS/MS; there is GC/MS; there is GC after bromination. Well, if you brominate your extract and then run it through GC with electron capture detection you get pretty good results. We are trying to look at this LC/UV as more of a screening method. I mean, we are looking for something that people can use as quick and dirty to screen their products, to help fuel their research, for developing countries to be able to use, and so on, and also maybe as a quick screen for us although we will probably end up as an agency, when we get into the screening mode, using something like the GC/MS.

Surprisingly, when you look at the cost, and they did this at the JIFSAN meeting, the costs are--what?--$200 or something per sample for the LC/MS/MS and you are still talking about $100 or so per sample for some of the GC/MS. So, this is very expensive work and you can imagine that developing countries wouldn't have the capability to be doing much of this work.

Of course, some countries have quite different foods, different fried products and so on that can reach the high temperatures. So, for the world to get its arms around the distribution of acrylamide in foods is an interesting point, and also for us to understand if the analyses that are being put into the database are valid is also another interesting question. We are going to have to be very careful, based on the information they provide, to understand if the results are good. I mean, there are many methods that give good results but, again, you have to perform the analyses correctly and it is very easy, initially particularly, to have problems. Therefore, people are circulating proficiency samples to try to make sure that the results that other labs are getting are going to be accurate. That is one of the efforts that is going on, to get proficiency samples in different types of foods to help improve the credibility of the results.

DR. MILLER: Looking at the developing world, is FAO doing anything to try to collect some samples?

DR. TROXELL: Not that I am aware. As I said, we have proposed an informal workshop and FAO/WHO were behind getting that together. Of course, FAO and WHO were behind the consultation and also will be working on gathering information for the next meeting.

DR. MILLER: Someone has to do those samples.

DR. TROXELL: Right, and if you look at the JIFSAN info. net you will find, I think, there are samples in there from South Africa and I think there might be some samples from Egypt. People are sending data. That is not the JIFSAN info. net; it is the WHO/FAO information network that JIFSAN is operating for them. Anyway, these data are coming in; the databases are growing and there are criteria for methods. There are many fields to fill out--what is your methodology, and so on--so that in the end, hopefully, we will be able to understand what these results look like and how valid they might be in different foods.

DR. MILLER: I imagine a lot of these populations that consume flat breads might get a little higher exposure.


DR. MILLER: Johanna?

DR. DWYER: Back to the issue of the quality ratings of the values that are coming in, I don't have any knowledge of this particular compound but certainly with flavonoid in food, where there are a lot of advantages for industry and others to try and find out what is in there because there are possibly positive health effects, there are quality rating systems already in place that can be used and adapted for determining whether a value is adequate or not adequate. I believe Dr. Beecher and others at the Ag. Research Service, right over in Beltsville, developed those methods some time ago. I wonder whether you are using those methods, adapted for acrylamide, to screen your values so that the database is as inclusive as possible, at the same time meeting quality criteria for analysis, sampling and so forth.

DR. TROXELL: Clearly there are performance criteria one can apply to analytical methods, and the GEMS database at WHO has many data fields which list out what is your method, what is your limit of quantitation and how you establish it, and so on. I believe an awful lot of information is requested of submitters so probably it is more than adequate and, in fact, if anything, there is so much information requested that it kind of inhibits people from submitting data because they have to put so much energy into providing the data. So, yes, I think we should be able to distinguish between the good data and the questionable data. Clearly, when it is questionable we won't use it.

DR. MILLER: Any other comments or questions? If not, we are going to adjourn for lunch and return at 12:45 for the afternoon session. Thank you.

[Whereupon, at 11:15 a.m., the proceedings were recessed, to resume at 12:45 p.m.]


Mechanisms of Formation 

[See presentation slides for Dr. Zyzak]

DR. ZYZAK: April 24, 2002, this is a shocking date to many people in the food industry as Stockholm University, in connection with the Swedish NFA, revealed acrylamide presence in a variety of foods. As many organizations began to develop analytical techniques with a capability of analysis of acrylamide in foods it came out that the data released at this time was accurate.


We, at P&G, did our own product survey and here is a sample list of that. As you can see, we also found acrylamide present in a variety of foods such as toasted bread products, roasted asparagus, corn chips and potato chips.


What is acrylamide? Acrylamide is a conjugating e-amine molecule. It has a high boiling point and it is also very hydrophilic or water loving. I think this is probably why it took a while for us to detect it in the food system. Typical analytical techniques in the food industry involve head space which is dependent on the boiling point or extraction of organic solvents and in this case acrylamide would tend to stay more in the aqueous phase.


Now that people have accepted the fact that acrylamide is present in foods, the next issue is how is it formed. Some initial mechanisms that were proposed from the Food Research Institute were based on equivalence formed in the frying process so there are lipid oxidation products which could be precursors, and some of these have similar structures to acrylamide. You can see acrylamide here and this is acrylic acid. The only difference is we need an amide bond here and acrolein, another lipid oxidation product.

However, this formation of the amide bond is not a very favorable reaction in typical food conditions and subsequent research in this area has shown that this is not occurring under typical cooking conditions. Again, the trouble came in the formation of this amide bond.

Well, if you look at the amino acids in the food system, there are already a couple that have amide bonds present in the side chain. One of these is asparagine and the other is glutamine. Now, amino acids during typical cooking conditions undergo a process of decarboxylation followed by deamination. If you look at this asparagine side chain, it looks very similar to acrylamide. What is known as the structured aldehyde of the amino acids would form this aldehyde here, and there is the possibility that there is a side reaction going on where we can form acrylamide from asparagine. You will see, as I show further on, that this is a small part of the reaction. Glutamine here just has an extra methylene group but there is the possibility that under typical cooking of food where you get decarboxylation followed by deamination it might form some type of rearrangement of products which could lead to a conceivable small portion of acrylamide formation.


To address this we developed a model system. In this case we tried to make it like a potato chip. We took potato starch and water. We were able to heat this kind of like a dough sheet, and to this system we could add a variety of amino acids, reducing sugars and a variety of other ingredients which could include inhibitors. After this we could take it to a frying process and then measure acrylamide in the finished product. This could also be baked and we showed in our studies that baking can result in the formation of acrylamide.

So, the elegance of this model system--we have to make sure that our system is inert itself. Here we looked at a potato starch system and we went through the frying process and got less than 50 ppb acrylamide. To our potato starch we started adding reducing sugars such as dextrose. We take our potato starch, we add in asparagine alone and we start to detect acrylamide formation. However, the combination of dextrose and asparagine gives us significant amounts of acrylamide formation here.

We also looked at other amino acids, such a alanine, aspartic acid, lysine, threonine and glutamine. We can actually detect a small level in glutamine, 156 ppb versus asparagine with 9000. So, about one percent the level is formed in glutamine compared to asparagine. However, arginine, cysteine, all these other amino acids do not form detectable levels of acrylamide. You can see from this that we kind of felt that asparagine is really the source of acrylamide.


How does this relate to food systems? Well, what about amino acid composition of potatoes? We looked at that and approximately 50 percent of the amino acids in potatoes are in the free state, which means it is not incorporated into the protein. Out of that, asparagine is roughly half of the free amino acid content. So, it is conceivable that potatoes have such a high level of free asparagine that that could be the source of acrylamide formation in potato products.


Other reactions were carried out with using free amino acids such as asparagine. We wanted to find out whether protein-bound asparagine could also participate in the formation of acrylamide. So, as an analog, you can purchase N-acetyl asparagine where the alpha-amine group is tied up to this bond, here, mimicking a protein analog. We reacted this with dextrose to see if it formed acrylamide. The results were that no acrylamide formation was observed. So, from our understanding, we felt that all we need to be concerned with is free asparagine because that is what is the precursor to acrylamide formation during heat in the food system.


We know that asparagine is required for homolysis and dextrose is also required. So, we looked at a dose-response curve from dextrose. If you look at a potato, to get a rough estimate, asparagine is actually 1.25 percent in a potato. The dextrose or reducing sugars is about 0.5 percent in a fresh product when harvested, but as product is harvested, usually in the late summer early fall, it may go into storage and be stored for quite a period of time and the level of reducing sugars will actually increase in potatoes. So, you can see that as you increase your level of free reducing sugars you actually increase your level of acrylamide. So, we know that you need not only asparagine but a level of reducing sugars in potatoes.


Are there other carbonyl sources that can form acrylamide? Some recent work speculated that the formation of acrylamide from asparagine, the structured degradation reaction--structured degradation reaction is implicitly explained, actually a di-carbonyl such as, in this case, glyoxal reacting with the amino acid causes the reaction to proceed. We also showed that glyceraldehyde, 2-deoxyglucose and ribose are also efficient at forming acrylamide in food systems.

People who are familiar with the Maillard reaction understand that the typical browning reaction involves first a reaction of a carbonyl amino acid. If you use a molecule such as 2-deoxyglucose where it is C2 here, you do not have a hydroxyl group. This prevents the molecule from undergoing the rearrangement. So, this actually lets us know that all we need is to Schiff base the formation for the formation of acrylamide. This is also verified by reactions we did, lipid aldehyde such as decanel, and Dr. Adam Bakowsky at Health Canada also published about octynal, another lipid aldehyde that can react with asparagine to form acrylamide.

However, people may ask is lipid oxidation contributing to acrylamide formation? I would think not because if you look in the food system, typically the reducing sugars are probably on the order of about one to two magnitudes higher than the lipid aldehydes. So, I think what we need to be concerned with is level of reducing sugars.


To prove that, initially we are thinking that the asparagine actually going into acrylamide is the side chain here. Just to make sure we can prove that this is going on, you can purchase isotopes which can be incorporated and enriched in either 15nitrogen or 13carbon. So, we carried out experiments just to confirm that this is where the source of carbon nitrogen is coming from in asparagine.


For the initial experiment we used amide-labeled asparagine so it is 15N here. Reacting with dextrose, we should form acrylamide where we have a 15nitrogen at the amide bond. Acrylamide has a molecular weight of 71. We are going to be monitoring and any unlabeled acrylamide would show a mass at 72. However, since we are incorporating 15N here, we think we should be detecting this at a mass of 73. And, this is what we see. We do not detect any unlabeled acrylamide. Really 97 percent of the total acrylamide response is at the monolabel, suggesting that this amide nitrogen is being incorporated into the acrylamide.


We did further studies where we labeled the alpha-amine nitrogen. Again, this is the source of the carbon nitrogen so when we add dextrose to form acrylamide we should get a mass at 72 here so any detectable levels of acrylamide were the unlabeled acrylamide, again suggesting that this nitrogen is not being incorporated into the acrylamide formation.


Next was to verify where the carbons were coming from. So, we purchased a uniformly labeled asparagine where all the nitrogen and carbons are labeled. In this case we got this nitrogen label and these three carbons and we should see an increase of four mass units and we should be detecting acrylamide at a mass of 76. Indeed, from our analysis all we could detect was the acrylamide at 76. You see in this chromatogram that we also monitored at 75, 74, 73 and 72 for any other molecules. So, from these experiments we concluded that this side chain of asparagine is what is being incorporated into the acrylamide.


From these studies we were able to form this following mechanism of acrylamide formation. The alpha amine group of asparagine here is a mucophilic attack on the carbonyl source, forming glycocyamine. As you are driving away the water, you get the formation of Schiff base. So, actually this process is favored under reduction of water in your cooking system. After this, as heat is applied we get decarboxylation that forms as intermediate which rapidly degrades to form acrylamide, os hydrolyzed to form beta alanine amide, and beta alanine amide itself can undergo elimination of acrylamide. We showed that we can heat this under typical frying conditions and it will be able to decompose even in the absence of sugars to form acrylamide.

So, this is kind of a proposed mechanism. How can we prove that? Well, utilizing LC/M mass we are able to do that. What I will show you on the next slide is where we are going to be monitoring our carbonyl source. In this case we use dextrose so we can monitor that at a molecular weight of 180. We are going to be monitoring asparagine. And, as we heat a product out we are going to be looking at the formation of Schiff base, the beta alanine amide and the acrylamide to prove our mechanism.


Here are the first monitoring intermediates in acrylamide formation. In this case, we use just regular asparagine reacted with dextrose. You can see that at our initial time, zero seconds, we get a response for dextrose and a response for asparagine. At intermediate time, 180 seconds, we actually start to see our first intermediate Schiff base being formed. As we heat this on to 270 seconds we have actually depleted all our source of asparagine, dextrose. The Schiff base is gone. We get an extra intermediate beta alanine amide and also acrylamide. So, we can monitor these intermediates in this reaction system.

Just to confirm that they were there we actually used isotope labeled 13C and 15N molecules and we can see the corresponding shift to mass units. In this case we are monitoring asparagine at 133, with these all being labeled, and incorporation of six mass unit difference. We monitor at 139 and we can see the increase so the acrylamide has gone from 72 to 76. The beta alanine amide has gone from 89 to 94. So, we have confirmed that these intermediates are actually formed during the reaction process.


Next is understanding acrylamide formation in food products. All these studies I have been showing you right now are a model system so we need to prove in a real food system that asparagine is the source of acrylamide. Questions arising--is asparagine the only precursor to acrylamide in heated foods? What about other potential sources of acrylamide, methionine, glutamine, cysteine or acrolein? These have been postulated by people to maybe provide a minor amount of acrylamide. In our model system, we think we have disproved this fact and have shown that they are not sources.

But another way to do this, we decided to do selective removal of asparagine from a real food product with the enzyme asparaginase to address these questions because we felt like if we could have asparaginase in this real food system degrade all the asparagine and look at acrylamide formation we would show that that reduces acrylamide formation and asparagine is the source of acrylamide in a real food product.


Asparaginase, this enzyme, will hydrolyze the amide bond of asparagine utilizing water and will form aspartic acid. If you remember our initial model system that I showed you, we analyzed aspartic acid's ability to form acrylamide and it formed undetectable levels. So, we feel like if we do convert asparagine to aspartic acid it should result in reduction of acrylamide formation.


So, here is our real food system. We took washed, Russet bake potatoes purchased from the local grocery store, boiled for one hour, and then we blended the flesh on a one to three ratio with distilled water. We have two plots here, one as a control and the other one we did with enzyme asparagine-treated, carried out for 45 minutes at room temperature. Then we microwaved this at two minute intervals for a total of ten minutes. This is a highly cooked product to maximize acrylamide formation. Both the control and asparaginase treated products were dry and brown after the step.

To make sure that the enzyme was working correctly we analyzed for change in asparagine and aspartic acid. This is our control sample and, as mentioned earlier, free asparagine is high in potatoes so you have a nice peak response here for asparagine and a smaller response for aspartic acid. In our asparagine-treated sample you can see where the asparagine has been depleted. We depleted about 88 percent here and the aspartic acid is subsequently increased. So, we know that in our system here the reaction was carrying out the way we expected.

Next was to monitor for acrylamide. We can see in our control product we have 20,000 ppb acrylamide. Asparagine is treated down to 164 so we actually got greater than 99 percent reduction in acrylamide by using asparaginase. So, we feel that this experiment is able to prove that asparagine is the mechanism for acrylamide formation in a real heated food system, in this case being the potato.

How does this relate to other foods? In our studies we looked at the yield of acrylamide from asparagine. We deduced that the yield was less than 0.5 percent. Dr. Adam Bakowsky at Health Canada also showed in his work that the yield was about 0.1 percent. So, we have taken these numbers and looked at a variety of food products. If you look at the amount of free asparagine in the starting food products and you correspond to the yield of about 0.1 to 0.5 percent, I think that will compensate for all the level of acrylamide that has been detected out there. So, we feel that asparagine is the source of acrylamide formation in all food products.


Acrylamide precursors are ready to intervene. We know that asparagine is important for this formation and also reducing sugars. Typically in food systems reducing sugars are glucose and fructose. Some foods will contain sucrose and in the cooking process will undergo hydrolysis to form glucose and fructose.

How is this affected in potatoes? Well, the level of asparagine and reducing sugars actually varies by the source of potato. I think there are many people out there in potato processing areas who are monitoring asparagine in a variety of potatoes and also looking at reducing sugars, and we know that in storage conditions, as potatoes are stored for periods of time, the level of reducing sugars will increase. If you look at a product in the early fall, it will probably have a low level of acrylamide but as people start to use more potatoes that have been stored for a longer period of time the acrylamide will potentially increase.


In conclusion, asparagine is the major source of acrylamide formation in foods. Carbonyl source typically in food systems is going to be reducing sugars as required in the reaction. Oil oxidation products and starch do not appear to be significant factors in acrylamide formation.

One thing I forgot to mention when I was showing you our model system where we took our potato starch and added amino acids and fried it, we also looked at fresh oil versus an oxidized, aged oil to see if acrylamide formation was affected and there was no difference. So, we were able to conclude that oil quality such as oxidized oil did not significantly affect the level of acrylamide formation.

So, that is the conclusion of my talk. Thank you.

Questions of Clarification

DR. MILLER: Thank you. Comments or questions?

DR. BUSTA: You were generating acrylamide in a microwave in a water system which wouldn't be over 100 C. Right?


DR. BUSTA: I thought we required a higher temperature than that to generate acrylamide.

DR. ZYZAK: I mentioned that in the microwave system it was dry and brown, and there are results out there that you can form acrylamide during microwave conditions. What we found is a big factor in the level of acrylamide formation is the moisture of the product. So, if you do microwave something and you still have a higher moisture content, it is probably okay. It is when you get down to low moisture content that you drive that reaction. As I showed the mechanism, as we remove water from our Schiff base you get the decarboxylation step. So, I think finer moisture content is a critical factor in acrylamide formation in food products.

DR. BUSTA: Are you saying temperature is not?

DR. ZYZAK: Temperature is. I think it is a combination of both. You need to have a low moisture environment to get that Schiff base and you need to get heat involved to get the decarboxylation step going on. You do need both of those going on, but we do see that at 100 degree C you can form acrylamide.

DR. MILLER: Johanna?

DR. DWYER: I think I am right that ascorbic acid is a reducing sugar and is present in some foods.

DR. ZYZAK: Correct.

DR. DWYER: Is that a significant factor? For instance, my ancestors ate a lot of potatoes and I want to know if I have gene damage.


DR. ZYZAK: I think you were talking about ascorbic acid and four carbon sugars can participate in this reaction, but I think if you look at the level of reducing sugars, such as the glucose and fructose, they are about half a percent and can range up to two percent in a potato. So, that far outweighs the level of ascorbic acid in there that is enough to facilitate the reaction.


DR. MEHENDALE: I was wondering if you have tried any carbonyl blocking mechanisms in your reactions.

DR. ZYZAK: Yes, the typical anti-browning reason are sulfites. All this would be simple if we could add sulfites to solve the problem but it didn't work and you can only add a pretty low percentage of sulfites in a food product, like if you buy dehydrated potato products I think it is less than a percent or something like that, the level of sulfite you can add in there. Since you already have a couple of percent of reducing sugars we didn't see any benefit to adding sulfites.

However, we also looked at another amino acid like lysine. We added lysine in there to block the carbonyl source, dextrose in this case, that was ineffective. However, if you add the amino acid cysteine, you can actually decrease the level of acrylamide formation. The question is, is the cysteine reacting with the dextrose tied to the carbonyl, or is it reacting with the acrylamide once the acrylamide forms? It is actually a later part of the reaction so it is complexing with the acrylamide.

DR. MEHENDALE: I have a follow-up question. You know, in some old literature a gentleman by the name of Serami has done a lot of work on di-cosylated end products for aging.


DR. MEHENDALE: It seems to be that it was carbonyl groups of dextrose that are involved.


DR. MEHENDALE: So, it seems to me like there may be some potential for either blocking or reacting the carbonyl groups with other things.

DR. ZYZAK: Yes, you bring up Dr. Anthony Serami and I did my graduate studies with Dr. John Baines, who were kind of competing with each other so I am very familiar with his work. We also did studies by adding one protein and other things in our model reactions. Maybe we can add a protein source that will either react or just tie up the reducing sugars. We weren't very successful at that.

We did try adding a protein source to our model system and didn't seem to have a significant increase. But, again, in that case we were using something off the shelf like a relatively inexpensive source. I mean, you can go out and buy some yeast products which may have a high content of glutathiamine. It is a very pricy product. So, if you want to try to reduce the level of acrylamides, you could probably incorporate a source of protein which may have a high concentration of thiol groups in there which are known to be very active with acrylamide once it is formed. People have done studies looking at amino acids. Amine groups will react with acrylamide but not very readily where there are high thiols.

DR. MEHENDALE: So, how long can we keep potatoes?


DR. ZYZAK: I think many people in the industry are also looking at that, you know, how is acrylamide going to be affected as later in the season we are using older potatoes? I think we are all looking at that now and I think many people are addressing that so it will come up in the future I believe.

DR. MILLER: It depends on how you cook the potato.

DR. ZYZAK: Sure, yes.

DR. MILLER: Any other comments? Yes?

DR. TORRES: Are there any other food systems where there would be a lot of free asparagine?

DR. ZYZAK: Yes, asparagus has a high level of free asparagine I believe. I believe at the subcommittee meeting Dr. Lauren Jackson showed some data that the level of free asparagine is high in almonds. I think it is high in legumes, beans, bean products. Asparagine actually is used by plants as a source of nitrogen storage system. So, most plants utilize asparagine to store the nitrogen for further use as energy or convert it into protein. So, I think we are kind of stuck with this because, you know, that is the way plants are going to grow so they are going to use nitrogen as a fixation source until we can use some biotech and utilize some other source of nitrogen.

DR. RUSSELL: I was just wondering is there much of a difference between white potato and sweet potato in the acrylamide formed under similar conditions?

DR. ZYZAK: You know, I don't have that data. We haven't done that experiment but I think definitely there is activity in that. We, ourselves, know that the variety of potato will affect the level of acrylamide because different varieties of potatoes will have a varying factor of asparagine in them and reduced sugars so we are also looking at that.

DR. MILLER: It seems to me that most root vegetables cooked at high enough temperatures should be excellent sources of acrylamide.

DR. ZYZAK: Yes. We actually talked with kind of a potato professor in industry just to understand more about how it is using asparagine so we can get potatoes with a lower source of asparagine. There doesn't seem to be a lot of information out there about that. You know, I was specifically told that roots are different from tubers so I don't know all the botanical aspects of that but it can be different whether it is a potato versus a carrot.

DR. MILLER: Well, if it stores asparagine as a nitrogen source then, depending on how it is cooked, it will have a high level of acrylamide.

DR. ZYZAK: Exactly, yes. I think any source of product out there that has free asparagine, if you cook it under conditions where you are going to drive off the moisture and heat it up, you are going to get acrylamide formation.

DR. MILLER: Certainly the big concern would be legumes as well.


DR. BUSTA: Is this information readily available to anyone who wants it now?

DR. ZYZAK: Which information? What I just presented? I think it is going to be up on the website so anybody can download it.

DR. BUSTA: How about before this?

DR. ZYZAK: Yes, actually the JIFSAN--you know, you have heard the struggle between people whether you are in academia and there is a need to publish--I think Procter & Gamble is a great company to work with. Actually, at the JIFSAN meeting back at the end of October I presented the mechanism and I told people we identified these intermediates and I informed people we used the enzyme to confirm that. At AIOC we showed mechanism formation, which was in late September. So, once we felt confident and we knew this was the mechanism we have been trying to be forthcoming to the industry and the academic people, releasing the information.

DR. MILLER: Terry, do you have a comment you want to make?

DR. TROXELL: Thank you. If you look at the spectrum of foods in which we find acrylamide, you are tracking foods that contain enough asparagine and glucose to form acrylamide so we are talking about wheat products, corn products. So, it is not just tubers and so on.

DR. MILLER: No, no, that is the point I am trying to make.

DR. TROXELL: Exactly. Might I ask a question of the speaker?

DR. MILLER: Not a good idea! Johanna?

DR. DWYER: I was just wondering, I think I followed your chemistry but I wasn't sure about instant mashed potatoes. Would those be high because of the extrusion product?

DR. ZYZAK: You may have small levels.

DR. DWYER: I am talking about the instantizing process.

DR. ZYZAK: Yes. In industry I think most of the mashed potatoes you buy from the shelf actually have sulfites in there but there is still a level of acrylamide. We are even monitoring our flakes, our starting material, and during the flaking process of potatoes they go through cooking and they are mashed and they undergo a spray-drying process and, yes, there is actually a small amount of acrylamide. I think it is probably around 100 ppb but you also have some precursors there too which are formed, such as the Schiff base. So, during the formation of these dried potato products you do have a small amount of acrylamide and probably some precursors.

DR. MILLER: Depends on how they are cooked again.


DR. MILLER: Other questions or comments? Thank you.

DR. ZYZAK: Thank you.

DR. MILLER: Our next speaker is Dr. Robert Brown, substituting for Dr. Steve Saunders, from Frito Lay.

Reduction Strategies 

[See presentation slides for Dr. Brown]

DR. BROWN: I feel very fortunate to be here today. There have been some very good presentations this morning and it has been nice. I don't have a handout. We will have to print one off. Anyhow, I feel fortunate to have heard the particulars this morning. There is some interesting science going on and it is amazing how far we have moved forward in a short period of time on this. FDA and other groups have really moved forward quickly.


I also feel very fortunate to be standing in here for Dr. Steve Saunders because Steve is not only my mentor and my colleague but he is also a good friend of mine. Unfortunately, Steve was not able to be here today due to unforeseen circumstances and I know that he wishes he could be here to be making this presentation today, and he wanted me to convey to all of you his sincerest regret for not being able to be here in person to make this presentation. I am a nutritionist and I was attending a nutrition meeting in town, and he asked me if I could step in and present this information for him. I am sure I can't really substitute for Steve but I am going to give it a shot. That said, I want to present the slides that Steve sent to me and the notes that he provided for me.


That was an excellent presentation from Procter & Gamble today, very compelling information on the formation of the acrylamide passing through a Maillard reaction product and that is what we have here so I can just skip past this.


If we look at the first intermediate of the reaction between asparagine and glucose or reducing sugar, we see this intermediate. When the typical Maillard reaction product is formed that typically has an energy activation level of about 25050 kilo calories per mole. These products that are formed are the typical browning colors and the flavor compounds that are formed in the typical Maillard reaction product.

As David mentioned, there is a second pathway, a minor pathway proceeding through the Schiff base and going to decarboxylation and beta elimination and proceeding to acrylamide. We have done some work in our laboratory on a model system similar to what P&G has done, and we have estimated that the energy of activation of acrylamide formation is on the order of 70 kilo calories per mole. So, you see, it takes more heat energy to form this compound and it is more of a minor pathway as compared to that going to the Maillard reaction products.


Clearly, our first insight then is that in a chemical pathway leading to acrylamide is a low yield pathway with a higher activation energy. This will be demonstrated in a couple of slides I have coming up to show a difference in concentration between reactants and the products in this reaction.


If we look at a summary of the data on acrylamide values in food, you are all very familiar with this data but I want to make a couple of points about the different concentrations of acrylamide in foods. First of all, if you look at the foods on this table you will see that there is a wide variety of foods that contain acrylamide, and across these different foods there is a huge range of concentrations of acrylamide found in these food products.

Additionally, even across and within a category of food there is a very, very wide range of acrylamide formation. You will see in some foods that the range of the acrylamide can be as much as two orders of magnitude. So that is quite a bit.

The third thing is that undoubtedly we are going to uncover more food products that are going to contain acrylamide, and I think the data that was presented by Procter & Gamble makes it clear that we probably can find those foods quickly by determining the concentration of asparagine in those foods and looking at potentially the cooking process, and then looking at the level of reducing sugar in those foods.


This slide looks at those different foods that were on that list. We did a food consumption survey and we looked at this information to look at all foods that contain acrylamide to get an idea of what the impact on the American diet would be. If we look at this list, this list shows that approximately 38 percent of total calories consumed in the American diet are foods that contain acrylamide. As you look down the list you see that, of course, many of the nutrients at about that same level also are coming from foods which contain acrylamide. The variation in micronutrients is dependent on the type of food, some of which is fortified. For instance, many of the bread products are fortified with iron, and such. So, that level of iron would be higher than you would expect as a percentage of the calories.


Clearly, this insight tells us that the acrylamide question really is affecting a large fraction of the American food supply and is something that we have to be concerned to understand.


When we began our thinking at Frito Lay about means to look at the issue of acrylamide formation and how one might begin the investigation of reduction of acrylamide in food we organized our thinking in these three areas: One could look at removing the reactants, either the glucose or the asparagine; one could look at disrupting the reaction to form acrylamide by a number of means; or one could look at removal of acrylamide once it is formed in food products.

Finally, at the end of the talk today I just want to talk a little bit about the significance of the study of exposure to acrylamide, which should be an interesting endeavor that we are looking forward to.


If we start with looking at ways to reduce the reactants, first of all, this data is coming out of Dr. Mottram's lab at the University of Reading in the U.K., and if we look at this, this was mentioned just a few minutes ago, the difference in acrylamide that is potentially found in different potato products and, in this case, what we see is that if we look at the baking potatoes as compared to the King Edwards potatoes we will notice that in the raw state we find essentially no acrylamide. In both potatoes, when they are boiled, in other words the heat stays below 100 degrees C, we see no formation of acrylamide. But once the products are fried we see the formation of acrylamide. In this case we see a ten-fold difference in the formation of acrylamide between the typical baking potatoes and the King Edwards potatoes.

Clearly, what we would anticipate from the information we just saw from Procter & Gamble is that the level of reactants is probably different between these two, whether it be more reducing sugars in King Edward's potatoes, though those were not measured in this particular study, but that would be a good assumption that we might make.

DR. MILLER: Just a matter of clarifying something, which is the potato generally used to make chips?

DR. BROWN: There is a variety of potatoes used in all industries. Frito Lay uses a proprietary potato and I am not sure of the exact variety name of that one. We call it the Frito Lay variety and it is called a chipping potato, and they tend to be lower in reducing sugars but, as was mentioned, as potatoes are stored over time the levels of reducing sugars rise.


In this slide if we look at the asparagine in various crops, as was mentioned just a couple of minutes ago, in fact the level of asparagine does vary quite significantly across crops. As you look at this, there are many crops and food products that have quite a high level of asparagine. In this case, you see in some of the products that there is a huge range of asparagine level in potatoes, looking at 0.5 to 10 milligrams/gram. In other products such as asparagus there is a huge range in the level of asparagine. Then, if you look at wheat the range can be 100-fold difference between low level wheat and high level wheat. So, what we have is a system that has a lot of complexities in it and you can't look at one solution that is going to work for everything.

Just to mention too, this table was compiled by the Food Research Institute and they have a list that goes quite a bit longer than this, but just to know that we have a very complex problem involving the entire food supply.


In our laboratory we did a model system and we looked at the substrate concentration and applied the reactants, both glucose and asparagine, and looked at the change in concentration with the formation of acrylamide. What we see here is a second order reaction where the maximum level of acrylamide is formed when both the substrates are in fairly equal concentration. Whenever one of the substrates is at a reduced concentration you see that there is quite a large reduction in the level of acrylamide formation. This is an interesting insight for us because at this point we were able to understand that you need to look at your specific food product and make the determination which is the reactant that is highest in that food product and that is the one that is probably going to be the one reactant that you want to go after in that particular food product. The equation on the bottom of the slide describes the fit of the surface plot.


Then, the insight is that the reaction is a second order reaction and that the concentrations of the two reactants need to be in fairly equal concentrations to get maximum acrylamide formation. The reaction becomes very limiting for that reactant that is at lower concentration. So, in the case of the example that Procter & Gamble brought up when the reducing sugars are at much lower concentration than the asparagine in potatoes the reducing sugar is, in fact, the rate limiting step on the formation of acrylamide.


The next area that we want to look at is what is the possibility of disrupting the reaction of acrylamide formation. If we go back to the data from the U.K. and we look at the bottom there and we look at what is the effect of cooking on acrylamide formation, if we look at the data on the bottom there, you look at frozen frying potatoes, French fries, and you look at a cooked product having approximately 3500 ppb acrylamide, then overcooking the product increases the level of acrylamide by four-fold. This demonstrates very clearly the time-temperature relationship to the formation of acrylamide and is something that gives us insight into some of the things that we may begin looking at to reduce the formation of acrylamide.


In our laboratory we wanted to validate our model system and this is similar to data that was done by Mottram et al., in their lab in Reading. In Reading, they showed that there was an inflection point of formation of acrylamide at 120 degrees C.

I would be interested in the data that was presented on microwave cooking, looking at the moisture level of the products that were done in the microwave, what the real temperature was at the surface of the product in formation of acrylamide and looking at the browning reaction. If there was a browning reaction that was actually taking place in the microwave, I think that the surface temperature may have been higher than 100 degrees, but that may not be the case. I don't know if they had measurements of the surface temperature but that would be interesting to know.

But in our model system in the laboratory, if we look at this in the model system we have no formation of acrylamide at temperatures under 110 degrees C. The reaction really starts going as you go above 120 degrees C and then is exponential in the increase in rate of acrylamide formation. I think this data is important because, as Terry mentioned earlier today, the study to look at surface temperatures and being able to probe, temperature at the surface is going to be very critical. And, the information from Procter & Gamble regarding moisture content--in most food systems, as you know, if the moisture content is high you cannot drive temperatures beyond 100 degrees C. As you drive off the moisture, that is when your surface temperatures can actually reach above 100 degrees C and that is where you will see the formation of acrylamide beginning.


We plotted a kinetic model of the formation of acrylamide over temperature in Kelvin here, and if you look at the formation of acrylamide in this kinetic plot what you see is that the inflection point appears to be around 120 degrees C. From that point the rate of acrylamide formation is very rapid. The rate on the bottom of the chart, on the right-hand side there, is approximately 175 degrees C and that is the typical baking temperature for most food products. As you can see by that temperature, at 175 degrees that is well above the temperature to drive the formation of acrylamide to a maximal rate.


The insight, therefore, is that acrylamide formation is extremely temperature dependent and occurs well below temperatures needed in typical baking or frying operations. It is probably not possible to cook products without any formation of acrylamide. The surface temperature studies that will be undertaken I think will be very interesting, looking at whether one could modify the final surface temperature of products by alternate cooking methods.


We have also investigated the pH dependency of acrylamide formation in foods and we found that at pH under five acrylamide formation is severely inhibited. Even at the pH of six there is some significant inhibition of acrylamide formation and as you get towards neutral pH you see that the acrylamide formation is maximized at a pH of around seven.


A very interesting idea in the whole disruption of acrylamide formation is can you come up with an inhibitor of acrylamide formation similar to the case of vitamin C added to reaction to inhibit the formation of nitrosamines. It is a very exciting idea to think that we could come up with something like that, but in this case the whole Maillard reaction process is one of amino acids forming the color and flavor compounds that we expect in our cooked foods and in this case we are trying to inhibit a single amino acid reaction rather than the whole cascade of free amino acids that would react.

However, one other point on that is that the inhibitor would have to be food safe and it would have to be approved for use in foods. I think we are all hoping that a similar simple story can be developed, and we have seen published reports and we have heard from Procter & Gamble today about some of the potential ingredients that could be added to foods to inhibit this reaction. Rosemary flavonoids have been reported to inhibit the reaction of asparagine going to acrylamide but we haven't seen data on that.

There was, of course, the interesting JIFSAN conference that we just heard about, and the use of the amino acid cysteine to inhibit this reaction. It would be interesting to look at what levels of cysteine could inhibit the formation of acrylamide and how could it be added to foods. Obviously the wide number of foods we have would require different mechanisms to incorporate cysteine into a surface of a product that would try to inhibit this reaction because there would be no need to have the inhibitor throughout a product if, in fact, the formation of the acrylamide was occurring only on the surface of the product where the browning is taking place.

In our model system we have also studied other inhibitors that might be functional in this area, and we have looked at divalent and trivalent cations and found that they also inhibit the formation of acrylamide. However, it takes a large amount of the divalent or trivalent cation to have this inhibition come into place. It takes approximately one equivalent of cation per mole of reactant to inhibit the acrylamide formation. We are not sure how these could be practically applied to foods. It may not be successful in all types of food because, again, of trying to get the cation in the area to inhibit the reaction.


If we turn to the kinetic model of Wedzicha and Mottram we have a pretty interesting array of areas that we might look at to try to develop a mechanism to inhibit the formation of acrylamide. As we see here, if glucose is in fact in a particular reaction system, the rate limiting step, there would be means to force the reaction of glucose, to deplete the glucose by other reactants that might react with glucose such as other free amino acids that would compete against asparagine for the formation of this reaction. Also, pH, temperature and time variables are other potential things that we might look at. If we can look in the cooking process at some way to control the final surface temperature of the product as another means of looking to inhibit this rate of formation. As we learn more about these reactions and these kinetic constants out of the lab, we hope to learn much more where the best place to attack this issue will be.


Finally, I want to talk just a couple of minutes about our attempts to actually remove acrylamide after formation in food products. In this case we have tried a couple of things. The supercritical CO-2 is very effective in removing acrylamide. Of course, it removes everything and completely destroys the product.


UV light is something that we had high hopes for to be an effective use, a fairly simple technique to cause the acrylamide to polymerize and effectively eliminate the hazard. In this case we take ground product. We expose it to all levels of light. I think we came up with this, remembering back when we used to make polyacrylamide gels and we thought this was something that, you know, was really going to work. We exposed to product to wavelengths of light ranging from UV to red and essentially we have found no effect at any wavelength that we have tried to date. Potentially the level of acrylamide is too low in food products for this to be effective.


Finally, I am going to talk just a little bit about the toxicology of acrylamide and its presence in the food supply.


On this first slide is data that was also developed for us looking at the food consumption survey and mirroring that to the data on acrylamide content of foods consumed in the American diet. If we look at the total here and then if we look at the red line on the bottom, which is essentially drawn on the X axis, that red line on the X axis would be the standard risk assessment line that would be developed under standard techniques of developing risk assessment for average daily exposure. As you look at that line you can see that it is orders of magnitude, at least two orders of magnitude lower than the exposure of acrylamide in the U.S. food supply. You can see that no one food is going to be the bulk of that. Assuming that we can eliminate the content of all potato products, fried potatoes, other kind of cooked potatoes, mashed potatoes and potato chips and we remove the acrylamide from all those products, we are still in the range of two orders of magnitude too high with the total acrylamide concentration in the diet using standard risk assessment techniques to form this line.

This is where I think the research on toxicology, looking at the low concentration versus high concentration p450, 2E1 metabolism of acrylamide is going to be very important to understand the difference in low and high concentration of acrylamide in the diet.


Steve gave me some food for thought here. Everybody is probably not that hungry right now but I didn't have any lunch so I still have some room for food. The whole idea of carcinogen in food is not a new one to us, and is something we have dealt with before if we look at cooked meats and such. We also have the NAS report and the Ames/Gold as other references to talk about that.

Also, as we evaluate acrylamide we have to understand that humans have been cooking foods for millennia and we have been exposed to acrylamide for all those years. So, we need to understand what is going on and what is happening or not happening with acrylamide exposure low dose versus high dose. We need to understand the toxicology there because it is going to be very important to us.

As Terry said this morning, there are going to be no quick fixes of this issue because acrylamide is going to be so widespread in the entire food supply. We are going to have to look at different bullets. Instead of one magic bullet for all foods, there are going to have to be different bullets for different food products as we move forward to look for means to address this issue.


As we begin doing feasibility analyses of what kind of intervention we can have in the food supply and what we are going to do, we really need to understand the whole kinetics and the removal of substrates from food products. We need to understand each individual food, determining whether in this case the asparagine or the reducing sugar would be the rate limiting step for the formation of acrylamide.

Also, if we look at low temperature intervention what would be required in the development of new cooking techniques that could potentially reduce the level of acrylamide in foods. For some foods it will be impossible to develop these low temperature techniques. In other foods it may be possible to use two-step cooking where you heat at higher temperatures before moisture levels are driven down and then lower temperatures as the food product is drying out and temperatures on the surface can increase.

Again, there are no magic bullets here. We have to look at individual potential solutions to each of the problems, and there is absolutely no precedent to the kind of ordered magnitude of intervention we have to have in the food supply from the processing to the cooking to the growing of foods in our food system. It is dependent on what we find in the toxicology and how far we have to drive the acrylamide level down.


Then some final thoughts here, the issue affects a large portion of the food supply. Lowering acrylamide clearly in one food is not going to do much to lower the overall level of acrylamide exposure in the population. We need to understand the toxicology here so that we know what our target is in this case.

Clearly, as was mentioned this morning, foods cooked at home and foods cooked at restaurants are going to be a significant source of acrylamide so foods that are processed by the manufacturer are taken into the home and then baked out, for instance as apple pie. Those foods are going to have a large potential exposure to acrylamide so we need to look at methods that would also address those.

If we look at what does the future look like, given the magnitude of the change in the food supply that could be represented by this, we need to really understand the nature of the low dose hazard of acrylamide to humans, and we need to really look at the impact of any proposed interventions and the consequences, if there are any unintended consequences to the public health. In going forward, as we begin to study the toxicology of acrylamide, we need to be simultaneously looking for all the interventions that are possible to be driving the acrylamide level down so that we are working on both ends of the spectrum, to both lower acrylamide concentration in the food supply and to understand the real significance and the real health effects of acrylamide levels in the food supply and on human health. Thanks very much.

Questions of Clarification

DR. MILLER: Thank you. Comments or questions?

DR. DWYER: Thanks, Bob, for a very, very interesting talk. I just have a question that is sort of silly but I was wondering does free asparagine have a taste in food so if you took it out it would taste different?

DR. BROWN: I don't know. I am not a flavor chemist so I am really not too sure. Thanks very much.

DR. MILLER: Thank you. We are ready to take a break. We are a little early again. I will tell you what, why don't we go ahead with the exposure assessment, Dr. Robie? We will go ahead and have that and then we will take a break after she finishes.

Exposure Assessment 

[See presentation slides for Dr. Robie]

DR. ROBIE: I am going to be referring to the handout you have in your package. I am going to be presenting the exposure assessment for acrylamide for FDA as prepared by Dr. Michael DiNovi and myself.


I am going to start out the presentation by going through some history, then I am going to move right into our exposure estimates and the model that we used, the assumptions that we made and the future work on it, and then the results and the interpretation.


You have already seen this a bunch of times. Less than a year ago a group of Swedish scientists reported the occurrence of acrylamide in food, and in their report they included a preliminary exposure estimate. This was based on acrylamide data in about 100 food samples. They split that into eight food categories. They may some assumptions about the foods for which they didn't have acrylamide data, and they calculated a preliminary exposure estimate of about 0.7 mcg/kg body weight per day for a 60 kg individual.


Then, about two months later, two months after the Swedish scientists publicized their findings, there was an FAO/WHO consultation held to discuss the issue of acrylamide exposure through the consumption of food. They also performed some exposure estimates. They used the same residue data as the Swedish scientists used in conjunction with some food consumption data from several national food consumption databases. They used a couple of different approaches to exposure estimates but the bottom line they reported was a range of about 0.3 to about 0.8 mcg/kg body weight per day, which is in agreement with the Swedish scientists.


That is all the history I am going to bore you with. Now I am going to go on to how we approached calculating the exposure to acrylamide. Really we approached it in the same way that we would approach the exposure to any additive or contaminant or naturally occurring substance in food, following this general equation.

Basically, we need information on the concentration of the substance that we are interested in, in the food. Then, we need information on the food, and we need to know how frequently it is consumed, and we need to know how much is consumed and when it is consumed. We get the frequency and portion size from the food consumption databases and the concentration data for acrylamide we get from the laboratory. This expression, this multiplication takes place for every food, each individual food, the concentration of a food, times the intake of the food. This is summed over all of the foods that would contain the substance to get estimated daily intake of the substance for an individual. Then, this information is summed over individuals to get an EDI for the population. Again, this is the way we do it for every food additive, everything in food. So, there is nothing different there.


Food consumption surveys--I mentioned that we are going to be getting a lot of our data from them. We use three food consumption surveys. Let me explain them, the differences between them and why we use three.

The first two that we use are CSFII surveys. We use a three-day consumption survey and a two-day, and each has about 20,000 participants. We just wanted to see the comparison between these two. This is going to help us evaluate the robustness of our model. Of course, in making this comparison we have to have knowledge as to how the number of days of the surveys is going to affect the outcome, the final answer that we are going to get. So, that is what I have in this last bullet. I was just going to leave this up here and then it was explained to me last week that this might not be intuitive so I am going to spend a couple of minutes explaining what this means.

The percent eaters, the percentage eaters value that we get from our food consumption survey is the percentage of eaters that report eating or consuming food on any one day of the survey. So, the longer the survey, the better chance we have to get higher percentages. We capture more of the eaters the longer the survey duration is.

The intake for the eaters is considered overestimated for a shorter survey because the intake for the eaters as reported from the survey is reported as an average over the days of the survey. So, the more days of the survey, the lower the intake for the eaters, the mean intake for the eaters.


We also use an MRCA 14-day survey. Again, we had the data; we wanted to make the comparison and test the robustness of our model. After saying what I just said about how the difference in the number of days is going to affect what we expect to see from the survey, I need to point out that this is a 14-day survey but it is 14 days of reported frequency. So, the percent eaters is going to be higher. The participants in the survey, every time that they consume the food, they report it as a consumption occasion. So, we are going to get percent eaters but they don't report the amount of food. The amount of food we get from the USDA Nationwide Food Consumption Survey so the frequency data from MRCA is linked to this food consumption survey and this is a three-day survey.

So, I am going to talk about the differences between the CSFII surveys and the MRCA surveys because, again, we are going to be comparing them and we have to have knowledge as to what the differences are between them to make valid comparisons. There is a lot of difference in the food groupings between the two surveys. The CSFII food coding system allows us to be a lot more specific about the foods that we choose. The MRCA survey has big, broad categories. For example, for the CSFII survey we were able to separate crisp breads from crackers and other salty snacks, whereas for MRCA we weren't able to do that and they were all grouped together. When we get to the data and you see the tables you will see how the categories are changed slightly.

Another difference in the surveys is the time periods over which they were carried out. For the MRCA we are talking about early to mid-'80s, and the two-day CSFII survey was carried in the mid to late '90s. Certainly, food consumption patterns and habits have changed over that time. Whether or not it is significant enough to affect the model is debatable but it is a difference that merits mention.


I am not going to spend a lot of time talking about this because our office is not actually doing this. This is work that is being performed by Dr. Clark Kerrington with the Office of Plants, Dairy Foods and Beverages Risk Assessment Division. He is looking at taking the two-day CSFII data, our most recent data, and expanding it to longer than two days. We really want to model product exposure and using a two-day, three-day or even a 14-day survey may not be that appropriate. I mean, it is data that we have so that is what we are using right now but Clark is going to be working on adjusting this two-day CSFII survey data and that is work that is in progress right now.


So, we are back to our exposure equation. This is basically the same equation that we had a few slides ago. The food frequency and portion size are combined here from the information that we get from the food consumption survey. Again, we multiply that by the concentration of the substance that is in each individual food. This is summed over food and summed over individuals. This is typically done at the mean so we arrive at a point estimate. The result we get is one point. This approach is useful for substances only in a few foods or when the EDI and the ADI to TDI in this case are very different from each other. In the Frito Lay presentation we saw that the ADI and TDI are not that different from each other so these two things don't really fit for acrylamide. Acrylamide is in a lot of foods so we decided to use probabilistic modeling.

So, instead of just using the means and getting point estimates we are able to use the entire distributions for the food consumption and also for concentration data. I am going to talk about this in a lot more detail in the next few slides.


Probabilistic modeling is an iterative process and for each iteration, as I said, we use the whole food consumption distribution so each iteration is a random sampling of the food consumption distribution for an individual food. In a similar way, the computer will randomly sample the distribution for the acrylamide level that we have so we are not just going to be using an average acrylamide value that we have for each food. We are going to actually use distribution and then we are going to again apply the percentage of eaters to the result we get from multiplying the food consumption to the acrylamide level.


I mentioned random sampling. I am going to show an illustration of what that is. This is known as Monte Carlo sampling. To be strictly correct, Dr. DiNovi and I used a variation of Monte Carlo sampling called Latin hypercube sampling. But for purposes of what we are talking about today, which is illustrating how the food consumption distribution is sampled, this sampling is certainly appropriate.

What we have is the cumulative probability distribution for a given food from zero to 100 percent. So, from zero to one the computer generates a random number for each iteration. It generates a random number from zero to one. So, in this case, this is where the computer is generating a random number for this iteration. This random number is used to sample the food consumption distribution. This is the generated value for food consumption that we got for this iteration. The acrylamide concentration distribution is sampled in a similar way. Again, those two numbers for each iteration are multiplied together to arrive at the acrylamide intake that we will get from consuming that much of this particular food.


For this expression I am going to be using AA for acrylamide throughout this presentation. I believe this is probably the first time you have seen this. This is just the same exposure equation that we have seen, just reworked a little bit. The food amount, the information that we get from the food consumption surveys is here. We are going to view this as one iteration. This is what is happening for each iteration for each food. The acrylamide level that is randomly sampled from the acrylamide concentration distribution is put there. Then, these two things are multiplied together to give us the amount of acrylamide that this iteration or virtual consumer will be exposed to by eating this amount of the food containing this amount of acrylamide.

Now, I have been saying that we are applying the percent eaters to this, and the way that we do that is also in each iteration and either a zero or a one is multiplied by the amount of acrylamide that we have decided we get from eating that food in this expression, here. One is in proportion to the percent eaters. Let's say a certain food is reported as having 80 percent eaters, for 80 out of 100 iterations this number is going to be one, and this food will be considered to be eaten by this iteration, this virtual consumer. The other 20 times this number will be zero and there will be no contribution to the total acrylamide for this iteration or this consumer from this particular food.

Again, this expression is for a food for an iteration. The values are summed over all the foods and then we have combined all of the iterations to arrive at distributions for acrylamide intake, which I am going to be showing later on in the presentation.


This is what we will refer to as the first page of the handout, virtual consumer number one. I just want to illustrate the point that we showed on the last slide a little bit better by actually showing you the output that we see on the computer screen when we run an iteration.

I just have the first seven rows reproduced up here. You have the whole table. The food consumption of the eater is the result that we get from sampling the food consumption distribution for this iteration for each individual food and then the acrylamide concentration, again, is sampled from the acrylamide level distribution for this particular food. These are multiplied together and, again, they are multiplied by either zero or one depending on whether or not this food is considered to be eaten by this consumer for this iteration. So, we either have a zero here or we have a number for acrylamide contribution to this particular eater. These are all summed then to arrive at the value of 0.49 for this particular eater or this particular iteration.

We have provided you a couple of additional virtual consumers in your packet. I think there are three more virtual consumers just to give you an idea and make sure everybody understands the point of what we are doing here and the results that we are getting.


We did 25,000 iterations. I am going to explain some of the assumptions that we made in the model. The first one is that there has been no accounting for correlations between food choices, either positive or negative. Examples of that could be if you look at virtual consumer number four, virtual consumer number four is shown as consuming both oven browned and restaurant French fries on the same day. All of this is entirely possible as an example of a negative correlation. A positive correlation could possibly be between peanut butter and bread or coffee and toast. These are things that we haven't included in the model. It is not that it can't be done; it can be done but historical knowledge has shown us that really it won't make that much difference and won't have that much of an effect on the bottom line result that we are going to get from running the model.

Another assumption that we made, or what we wanted to make you aware of at least, is that we have used the food consumption distributions only from zero to the 99th percentile. The 100th percentile we saw what we considered to be some pretty irrationally high values. I know that is a strong word but 13 liters of coffee for one person in one day not only seems like a lot of coffee but it is also 11 liters higher than what is reported at the 99th percentile. So, we didn't want these numbers, which we considered irrationally high and possibly reporting errors or calculation errors of some kind, to interfere with our model. Another example of that is cookies, 620 grams of coolies per day was reported at the 100th percentile by the CSFII two-day survey and the recommended portion of cookies is 30 grams, and also at the 99th percentile this value of cookies was 130 grams. So, you know, it is a five times difference between the 99th and 100th percentile. So, we just went ahead and took off the 100th percentiles and used from zero to 99.

DR. BUSTA: Why did you use zero?

DR. ROBIE: It is possible that people consumed zero. I see what you are saying, to just cut off both extremes of the distribution. There is no zero percentile? Good point.


Some of the limitations that we see in the model include the surveys and the laboratory data which are two inputs for getting information from the surveys about the food consumption. We have already talked about these. We have talked about the duration of the surveys. Two or three days, even 14 days to model product intake is not an ideal situation. And, the food classifications, I have already talked about, especially for the MRCA. They have very broad groupings. It would be better if we could separate the groups as much as possible.

A primary limitation of the model is laboratory data. You have heard people talk about the laboratory data all day and I am going to do it again. Some of the samples that we have of food types are represented by fewer than five data points from our laboratory and we see that as a limitation. Most of these are the samples that are lower consumption and don't contain a lot of acrylamide. We have tried to focus on the ones that we think are going to have a big impact on the total overall intake for the population.

Another model limitation that we see is variability in acrylamide levels in different foods. Again, you have already heard people talking about this all day today. What we do see is that as you go down this list we see greater variability. There is a significant amount of lot-to-lot variability, even more brand-to-brand variability. Different products can be included in the same food category. I don't want to pick on potato chips here but we have a lot of potato chip data so we have definitely seen this. We have seen differences in lot-to-lot, brand-to-brand and product-to-product. If you have a baked potato chip versus a fried potato chip, they both have acrylamide. The differences between the two are pretty drastic.

Then, there is always the problem with the foods prepared at home and, again, you have heard about this. We do have data on toast and we have data on oven browned French fries but these are samples that were prepared in our laboratory following package instructions and I don't really think we have probably captured the variability that we are going to see in people making it at home to their color and taste preference.


Just one more slide and then we will go to data. We wanted to make you aware that we did apply some factors to some of the food types, typically the foods that are consumed as liquids. The analytical technique that we are using makes it a lot easier to test these as solids than as liquids as they are consumed but, of course, they are reported in food consumption surveys by the survey participants as liquids as they drink them or eat them.

So, for the last three on this list, instant coffee, dry soup and dry cocoa powder, the calculating effect was pretty straightforward in that we are taking the amount of the dry powder that we are going to then add a known amount of water to and dilute and consume the entire amount. So, whatever is in the package, you are consuming it all.

For ground coffee it is not so straightforward because when you make coffee you don't eat the grounds. We know how much acrylamide is in ground coffee. What we need to know is how much acrylamide is in coffee as consumed so we can compare it with the food consumption data that we have. We use a value of 24 for this. This has been experimentally derived by Dr. Musser's lab. They measured the acrylamide in the ground coffee. They made coffee from it and then they measured the acrylamide in the coffee as consumed. They are still working on fine-tuning this number, if it is 25 or 23, but we are right there with the preliminary value that they have given us, I am sure.


As I said, we used three food consumption surveys and data from those and we did it for two populations. So, really you are going to have six slides for these. Instead of tabulating all 30 foods for all three surveys for both populations, I am going to go ahead and just show two slides just to show you all the foods that we are considering, and then we are going to go on to the survey-specific information. Actually, these are results from the CSFII two-day survey for the two years and older population. The data that you have here is the mean population for acrylamide intake for this food. This is restaurant fried French fries; this is oven browned French fries. These are sorted in order of their contribution for acrylamide, the contribution that they are going to have on the total acrylamide intake for the population. And, we have cumulative percentiles too.

So, restaurant fried French fries will contribute 15 percent acrylamide to the total population. Oven browned French fries will contribute another 13 percent. So, the cumulative percentile there is 27 percent, and so on down the slide to soft bread. We pick it back up again at corn snacks and then down to soup mix. Here are the last, cocoa to breaded fish and then donuts to multi-drinks, with multi-drinks having the least contribution to the total population acrylamide intake of 0.37 mcg/kg body weight per day. I know that these look like they have no contribution. I decided to only go out to three decimal places on the slide but they do contribute, just not very much.


Now we are going to look at data specific to the populations in food surveys. Let me refer you to page six of the handout. The way that these are ordered is just from oldest data to most recent data. There is no other significance to the order that I will be showing you.

These are the results of the MRCA 14-day survey two years old and older population. I have up here the foods contributing five percent or more individually to the total acrylamide population of 0.48. These are eight foods. You have the rest in your handouts. For each food we have the percent eaters, the mean food consumption for the eaters, the acrylamide concentration for that food. If we multiply the mean food consumption of the eaters by the mean acrylamide concentration for each food we get an eaters only acrylamide intake, which we then weight using the percent eaters to get a mean population acrylamide intake for that food. Then we sum all these to arrive at a value of 0.48 mcg/kg body weight per day as the mean acrylamide intake for the population using this survey data.


This is the same table for the CSFII three-day survey. Again, the foods that we have chosen to tabulate on the slide are the ones that contribute five percent or more individually to the total mean acrylamide intake of the population which, for this survey, is about 0.32. It is a little bit less than what we saw for the 14-day survey but, again, over 14 days we are going to have a lot greater percent eaters. That is why this value is lower here than for the MRCA 14-day survey.

The other thing of significance to look at on this slide--actually, I was going to show you the percent eaters difference. Look at the percent eaters for potato chips here. It is 18 percent for this survey and it is 76 percent for MRCA 14-day survey. So, the difference in the percent eaters can be very significant for some foods.

The other thing to note on this slide as compared to the slide for the MRCA data is that these top foods that each contribute five percent or more are the same for both surveys. The order is changed a little bit but the top contributors are the same for both surveys.


Then I have the same slide, which is the next page in your handout, for the CSFII two-day survey. The bottom line here is 0.37 and we saw 0.32 for the three-day survey. It is a little bit higher, which is expected, but everything is pretty much in the same ball park. Again, the top eight foods are the same. The order may be a little bit different but the top contributors are the same no matter which survey you look at, which shows us that our model is very robust.

Another thing I just want to point out is that all these values are very consistent with each other that we have gotten from these three surveys, and also they are consistent with previous exposure estimates which I talked about on the first couple of slides of this presentation.


We are going to go in the same order of surveys for two- to five-year old population. That is on the next page of the handout. The first things that are probably pretty obvious to you are that we no longer have eight foods; it is seven foods because the coffee dropped off. Two- to five-year olds aren't drinking a lot of coffee, which is good. Each of these foods, these tops seven foods, are again contributing five percent or more individually to the mean acrylamide intake for this population and it is the same seven foods as we saw on all the other slides, except for coffee.

Another thing that I am sure you have noticed is the fact that the mean acrylamide intake for the population is about twice what it was for the two years old and older population. This is expected when we are talking about data on a kilogram body weight basis. Children tend to eat about half of what adults eat but they weigh about a fourth of what adults weigh. So, the factor of two is an expected result for what we see here.


Then, for the CSFII three-day survey data for the two- to five-year old population, again the coffee is gone and the top seven foods are the sane. The order is moved around a little bit. Again, the value that we have here is lower than we had for the 14-day. Again, we are capturing fewer eaters, and it is also about twice what we saw for the CSFII three-day survey for the two years old and older population.


One more table of data, the CSFII two-day survey, and we see the same trends as we saw for the two years old and older population. Again, this mean acrylamide intake for the population is a little bit higher than we saw for the three-day survey but with a shorter survey that is expected. Again, the top seven foods are the same.


I have shown you tables of data, lots of tables of data. Now I am going to show you a lot of distributions to go along with those tables of data. On the data tables that you saw we showed you mean acrylamide concentrations and mean food consumption and a mean result but we really don't get out just a mean; we get distribution of acrylamide intakes. This is for the MRCA 14-day survey two years old and older population. Acrylamide intake is on the X axis in units of mcg/kg body weight per day. The mean, again, is about 4.8 mcg/kg body weight per day. This occurs about the 70, 75th percentile and that is the case for all of the survey results, all the distributions that I am going to show.

DR. MILLER: Just to clarify, we are talking here about population means; we are not talking about eaters?

DR. ROBIE: Right, we are talking about population means. I am sorry if I misspoke.

Another thing of note, the 99th percentile we have shown here is 0.91 mcg/kg body weight per day, just about twice the mean for the population, which is an expected result.


This is the same looking type distribution for the acrylamide intake from the CSFII three-day survey data for the two years old and older population.


For the CSFII two-day survey for the two years old and older population.


Then we will go on to the two- to five-year old population. One difference of note here is that we have expanded the scale to six. We cut the scale off at three for the two years old and older population. We have already discussed how we expect to see a doubling in the population mean intake when we are talking about two- to five-year old population when comparing to a two years old and older population on a kilogram body weight basis.


Here is the acrylamide intake distribution for CSFII three-day survey two- to five-year old population.


And the two-day survey. So, we haven't seen anything unexpected. Again, all the surveys agree with each other and the previous exposure estimates, and that is all I am going to say about that.


The table that you have in your handout to go along with this slide I believe is on page five. What we have done here, just to round out the whole picture and put things in perspective, is to show you how much acrylamide an eater would get from consuming one recommended portion of any of these foods. We are applying the mean acrylamide concentration. We are multiplying these by recommended portion sizes. These are not portion sizes that we got from any survey; these are portion sizes from 21 CFR 101. This is the labeling section of CFR, the food labeling section. These are the portion sizes that you are going to see on labels. These are recommended portion sizes.

What I have tabulated here are the top eight foods that you kept seeing consistently in the previous tabulated data. You have the full list in front of you, and these are also in alphabetical order.

I guess the thing I want to point out here first of all is that this is wrong in the copies of the slides. This is 3.2 and the actual number is 2.0. It is correct on page five of your handouts. Something else to note is that certain foods that we saw having large contributions appreciably contributing to the total acrylamide intake for the population, notably breakfast cereal and soft bread, there is really not a lot of acrylamide per portion for these foods. These are frequently consumed foods and highly consumed foods.


We also ran some "what if" scenarios. Probabilistic modeling lends itself very well to carrying out these types of scenarios. What we are talking about here is looking at the effect of a chosen mitigation measure on the population acrylamide mean, final result. So, we have chosen some foods and food groups and set the acrylamide level of these foods and food groups to be zero and we ran the model. I am going to show you the results for several food groups.

It is important to note, however, that the foods are still included in the model. We are just assuming that the acrylamide can be removed from the foods. To remove the food from the model and consider it not being in the diet anymore we would have to consider what it would be replaced by and the implications behind that. So, that is not anything that we did here. We just set the levels to zero in the foods.


These are the results for several different food groups for the CSFII two-day survey two years old and older population. Remember, the population mean result for the acrylamide intake was 0.37. We have assumed zero concentration of acrylamide in French fries, both types of French fries, both the restaurant fried and oven browned French fries, and recalculated the mean to be 0.26 mcg/kg body weight per day.

We also ran it assuming that acrylamide would be removed from snack foods, and in snack foods we included potato chips, corn snacks, popcorn and pretzels. The mean was reduced to 0.31 mcg/kg body weight per day.

We did the same thing for breakfast cereal and coffee. The bottom line here, and Terry mentioned this in the morning, is that no one food is contributing the majority of the acrylamide to the total population of acrylamide intake that we are seeing in any of the surveys.


For our future work I have already mentioned the modeling of longer-term food consumption that Clark Kerrington is working on for us to more accurately model chronic intake. We will continue to run "what if" scenarios based on the technological capabilities in industry. Also, we are going to analyze the sensitivity analysis of our model. Sensitivity analysis will allow us to determine the sensitive inputs for the model which, in turn, allows us to identify important uncertainties and that is going to help direct our future efforts in our exposure estimate.


In summary, we have seen that the mean population acrylamide intakes that we got from three survey are consistent with previous exposure estimates. The greatest contributors to the mean population acrylamide intake, the top eight or the top seven for the children, is the same for all surveys. We see that some of the foods that have lower acrylamide levels do contribute appreciably to the overall mean population but, again, this is because they are commonly consumed foods. On the "what if" slide I just showed no one food or food group accounts for the majority of the mean population acrylamide intake.

Thank you very much for your attention.

Questions of Clarification

DR. MILLER: Thank you, Donna. Comments or questions? Yes?

DR. BUSTA: If you ran the "what if" scenario on the two- to five-year olds that had a mean of 1.0, knowing that the coffee wouldn't be there, would it be a similar kind of reduction or would it be more dramatic? These range somewhere between 35 percent reduction and less than that.

DR. ROBIE: Let me find the data.

DR. BUSTA: I can look for it. Down to 0.84, so go down from 1.0 to 0.84. You are talking about removing French fries?

DR. BUSTA: Breakfast cereal.

DR. ROBIE: Breakfast cereal.

DR. MILLER: Johanna?

DR. DWYER: Thanks for a very good presentation. I am sort of hung up on one thing, and that is this business of using food consumption surveys of different lengths of time, two, three 14 days, and the issue of a person reporting they didn't eat it because they didn't eat it on that day when what you are really trying to model is chronic consumption. It could be that they do eat it but they don't eat it the day that you observe them or they report on.

It is my understanding that the group at Iowa State has been doing some statistics, as they always do, and have developed a method that may help to adjust distributions in that respect. Now, I understand that, particularly when you get up to the 90 percentiles, they can be up to the 95th percentiles. It is also my understanding that a group at the National Cancer Institute, Dr. Dowd and his colleagues, are working on something called propensity to consume that basically is just--I hope he is not here--a guess as to how often you consume something that you are telling the observer, and that has now been approved for use in NHANES. Could you tell me in your modeling if you are taking that into account, adjusting in any way for the artifact because of those issues?

DR. ROBIE: The longer-term food consumption modeling that we are talking about that another scientist in our office is working on, I believe that is about the same as the IOC. I don't understand the intricacies and I didn't talk about it too much about it today, but this is something that has been done before and been published. I believe this is along the same lines as what you are talking about. This is just taking a little bit more time. We went ahead and did what we could with the two- and three-day surveys to see what results we could get. Of course, we are not stopping there. We are considering this preliminary. But I believe that what we are doing with these data here, these two-day data to expand over 365-day period is similar or the same as what you are talking about.

DR. DWYER: Well, if you are eventually going to be using the NHANES survey for looking at hemoglobins and all these other things and trying to relate it to consumption it might be worth talking to the people at NCHS to find out if, on the instrument they have, they have 120 foods I think where they are asking for propensity to consume, if those eight foods that you care a lot about from the standpoint of acrylamide are included in that list. If they are not, perhaps it is worth talking to them and seeing if there is a way to get them.

DR. ROBIE: Thank you.

DR. MILLER: If you did the same analysis you did on the population of eaters only, would the top seven foods change any or would the distribution change any?

DR. ROBIE: Well, we can look at one of the tables because on your tables we do also have a column of eaters only. I am just going to put up one of the tables for anybody who doesn't have the handouts. If we were to sort on the eaters only column, certainly the foods that you are seeing right now are not going to be the same ones if you sort if by that column. We have looked at that. The breakfast cereal is where it is because the percent eaters is high, especially for the 14-day surveys, 37 percent. So, I guess the answer to the question is yes, they would sort differently.

DR. MILLER: It just seemed to me that there were enough differences between the two, just quickly glancing at these curves, that you would get a different pattern of what the top seven or eight might be for people who are eaters and also, of course, the total intake is going to be considerably higher or at least would be significantly higher.

DR. ROBIE: For eaters only, yes, that is true.

DR. MILLER: It would be interesting to know how much greater the intake is for eaters only.

DR. ROBIE: Well, it is not really appropriate to add the eaters only values. If each of these categories was 100 percent eaters then that would be appropriate, or if we knew that they were the same eaters. That is why we have added for the mean population.

DR. MILLER: Right, I understand. Anybody else?

DR. DWYER: I have one more.

DR. MILLER: Yes, please.

DR. DWYER: I guess the first thing I would ask if I were coming off the street is the Swedes came up with an estimate of about 0.7 and then FAO came out with an estimate of 0.3 to 0.8. You are coming out with estimates that are toward the lower end. Is this because you are from a fast food nation or is it because of some defect in the model, or is our model better and were those early estimates simply imprecisely high?

DR. ROBIE: Well, I'd like to think our model is better but I can't necessarily say. I can say for the Swedish data, again, this is a very limited data set, only 100 food samples for only eight food categories. They made a lot of assumptions and predictions for the foods that they didn't have data for, but they assumed that they had acrylamide, things like meats and vegetables and fruits. They went ahead and assumed that they had the same average level of acrylamide as the food they did test. I believe that is why their value is so high.

As for the results from the FAO/WHO, it is hard to know reading the report. Again, they used the same residue data so we are talking about a limited data set to start with. It is not really clear from reading this report what they have done about the foods that haven't been tested at this point, if they also made assumptions or if they assumed those to be zero and ran the model. Again, there are several different food consumption databases so it is possible that the value is higher for Sweden and lower for another country and that is how they came up with this range. Their lowest number in the range is 0.3; our lowest number is about 0.3, and we go from about 0.3 to about 0.5.

DR. MILLER: Dr. Torres?

DR. TORRES: What is the body weight difference between the U.S. population and the Swedish population? I imagine it must be significant.

DR. ROBIE: I can't say I know the answer to that question.

DR. MILLER: It really isn't that much different. Anybody else?

DR. TORRES: One last question, if you were to run the computer model a couple of times what would be your position about the value obtained? Is it going to be pretty reproducible? The question I want to ask you is how confident are you of the numbers that you are getting?

DR. ROBIE: We are very confident in the numbers. The number of iterations we have chosen is 25,000 and Dr. DiNovi and I ran it once just to see how long it took to converge. It was 5000 iterations and we have run it several times and get the same results.

DR. MILLER: We will take a 20-minute break and be back here at three o'clock.

[Brief recess]

DR. MILLER: Given the nature of the product and how widely is used, why don't we see pizza on this list?

DR. ROBIE: Somebody else asked me the pizza question. We don't have laboratory data on pizza. We have some data on Boboli pizza crusts where they detected above the limit of detection. I can't remember the actual residue levels. Forgive me, I don't have the data right in front of me and there are quite a few data points. But I know that Steve Musser's laboratory has tested the Boboli pizza crust. I don't think we have tested any like take-out pizza from any pizza chain or anything like that.

DR. MILLER: Well, what kind of data did they get?

DR. ROBIE: For the Boboli pizza?


DR. ROBIE: I think it was non-detected. It might have been detected but under the limit of quantitation for the method. We deemed that the values weren't high enough to really have enough data. We are talking about a Monte Carlo method so we don't want to add something that has just a few levels that non-detects or are under the limit of quantitation.

DR. MILLER: But you have large numbers of eaters.

DR. ROBIE: Well, we would assume we have large number of eaters. I can't necessarily say that without looking at the survey data. You would assume I guess that there would be a large number of eaters but I don't want to say that unequivocally without looking at the data.

DR. MEHENDALE: To follow up, you know, crust is one aspect and also cheese and the crust, of course, is relatively low in moisture and so on.

DR. BUSTA: In one table it said 33, the crust. It is really low.

DR. MILLER: That is interesting. Given that it is basically a flat bread, you would expect to find a high concentration. Thank you.

DR. ROBIE: While I am up here, I just want to point out an error on the last two tables of the handout that you have. It says two-plus population and it should be two- to five-years population. That is one of the hazards of cutting and pasting. I apologize for that error.

DR. MILLER: Thank you for your patience. Dr. Tim Fennell, RTI International, will talk about adduct studies.

Adduct Studies 

DR. FENNELL: Thank you very much. I am very pleased to be here. I would like to thank the organizers for inviting me, and I would like to thank you all for your time.


What I am going to talk about, what I was actually asked to talk about are adducts of acrylamide, and metabolism is one of those things that is inextricably thrown in there so I am going to talk about metabolism also.

I am going to review some of the general concepts of metabolism and pharmacokinetics, hemoglobin adducts and DNA adducts, and then get into a little bit about the history of metabolism of acrylamide, hemoglobin adducts and DNA adducts, and some of the recent studies that we have been doing and where we are going.


I would like to acknowledge a number of my collaborators. I used to be at CIIT and while I was there I worked extensively with my wife, Dr. Susan Sumner. I would like to particularly call attention to a couple of people. One is Rodney Snyder who has moved to RTI with me, and Burham Ghanayem at NIEHS who has collaborated with me on Cyp 2E1 null mice. I would also like to call attention to the various sources of support I have had from CIIT and from the acrylamide industry, and in particular most recently SNF who is currently funding some of my studies.


Going back just to give you a little bit of history, I started work on acrylamide in 1989. I had never been to a meeting on acrylamide until last year.


It just shows you how much interest has been generated since April of last year. Since August of last year I have been to five specific meetings on acrylamide.


That is what it is like to be in vogue.

Acrylamide is a reactive chemical. It undergoes Michael additions. As we have heard, it is very reactive with sulfinyl groups. It also reacts with amino groups. It is extremely reactive with proteins and reacts very slowly with DNA.


I am going to be talking about a number of different kinds of labeled acrylamide and I just wanted to go through and review for everybody what I am talking about. For unlabeled acrylamide, we usually use that for pharmacokinetic studies for measurement of chemicals. For chemical measurements with things like mass. spec. For radio-labeled acrylamide we use 2,3, 14C acrylamide. The labels are in the vinyl carbons. Usually we have a small percentage labeled used for metabolism disposition, pharmacokinetics, adducts--you name it, we can do it.

The limitation here is how hot you can make this, how high a degree of radioactivity. It has a tendency to polymerize. So, that is always one of the big concerns. You can't make it tremendously high.

What we have done a lot of studies with is uniformly labeled 1,2,3 13C, 3C acrylamide which is essentially 100 percent labeled at each site. We use this for metabolism disposition studies and also for adducts. We developed at CIT a method for analyzing metabolites by taking this material, giving it to animals, collecting the urine and looking at metabolism in the urine using 13C NMR spectroscopy and we can use that to find and characterize specific metabolites even if we don't know they are there until we go looking for them and we can measure them.


One of the other things we need to talk about when it comes to metabolism is glycidamide. This is the epoxide metabolite of acrylamide. It is formed by oxidation. It is a reactive epoxide. It reacts with protein and it also reacts with DNA. This is the big concern from the standpoint of carcinogenesis. This reacts with DNA and can cause mutations.


So, when we are dealing with metabolism and pharmacokinetics, generally most chemicals undergo metabolism to things that are more water soluble and less toxic. While most of the metabolites are unreactive and heavily excreted, some are more reactive. The problem with reactive chemicals or metabolites is that they can react with micromolecules, with glutathione, and they can disrupt all kinds of cellular processes that can lead to toxicity or carcinogenicity.

When you have metabolism by more than one route, and we actually do have that with acrylamide, you may have one route that will give you the reactive metabolites and the other that will give you stable metabolites. Then, you can have those reactive metabolites undergo further metabolism so that life gets complicated.

What we really want to know is the balance between the various metabolic processes, the relative rates, and these can be an important determinant in toxicity and they can differ between species and between high and low doses.


So, when it comes to risk assessment, we want to know about metabolism and adduct formation so we can understand relationships between exposure and internal dose. Here, internal dose is something that pharmacokineticists think of as area under the curve in blood. It is the amount integrated over time at a particular site.

For dose response--we have already heard about dose response and linearity. Do we have linearity in range of effects? Can we compare our internal dose measures with effects that are generated in bioassays? Do we have differences between species? Can we use measures of dose for reactive chemicals or metabolites to improve other studies such as epidemiology studies, interpretation of bioassays? Can we start looking at susceptible subpopulations?


Really what we are interested in here is how much acrylamide gets in the body, how much is metabolized, what is our area under the curve in blood and the tissues? Do we have adducts in proteins, and do we have adducts in DNA? And, the same thing for glycidamide. And, what is the balance between the various processes?


To go back to a little history, one of the more comprehensive studies on pharmacokinetics was conducted by Miller et al., in 1982. They administered radioactive acrylamide to rats. I just draw your attention to the very last bullet. The half-life of acrylamide, they found, was about two hours and about 12 percent of the radioactivity remained associated with red blood cells. That is something that is actually specific to the rat. They have a very high level of binding in red blood cells and it is because of the presence of a highly reactive cysteine in rat hemoglobin that does not occur in mouse or in human red cells.


I have already alluded to this study that was conducted in 1992. This was the first study that I was involved in with acrylamide, looking at urinary metabolites following administration of this 1,2,3 14C acrylamide in rats and mice. We used a dose of 50 mg/kg in male rats and mice and we were able to characterize five urinary metabolites including four derived from glycidamide. Interestingly, we actually found glycidamide itself in the urine. We find that there was a big species difference, with rats metabolizing most of the acrylamide directly by glutathione conjugation and about 33 percent by glycidamide, whereas with mice it was the opposite, much more was oxidized by glycidamide than was conjugated directly with glutathione.


On this slide I just have a summary of the various metabolic pathways. The acrylamide undergoes direct conjugation with glutathione to form this mercapturic acid that we see in urine. This is the epoxide glycidamide and this is actually detected in urine and we see it is ring opened form, the dihydroxy form, and we also see two glutathione conjugates at each of the epoxide carbons to form also mercapturic acids.


Now I am going to switch to hemoglobin adducts. These are one of the great ways of detecting exposure long after the exposure may have happened. So, what do we know about hemoglobin adducts and how do they behave? Well, they are proportional to area under the curve for reactive chemicals or metabolites. You can get a lot of hemoglobin rapidly from a blood sample so it is easy to get and easy to measure. It has a number of reactive amino acid residues that can react with chemicals and their metabolites, cysteine, histidine, the internal valine and carboxyl groups. The internal valine is particularly useful in performing adducts because it is the same in both the alpha and beta chain of hemoglobin and it is the same in rats, mice and humans. So, that gives you a lot of simplification. You don't need to have different assays for different species.

The red cell has a long lifetime in circulation of about 120 days, and it is removed with zero order kinetics. That also is a big advantage. It is not a first order process so it is very predictable and we can really nicely model what is going on with adduct accumulation. What happens is if you have continuous exposure over the life span of the red cell, in humans, if you have an adduct formed per day of exposure of X, you would expect that the life span of the red cell, if you had continuous exposure, is that your adduct level would reach 60 times X. So, that is very useful for long lived exposures and long experiments but it makes it very difficult to actually do a short-term study because you need to have a fairly high level of exposure if you are trying to do a calibration study.

With repeated exposures the adduct accumulate and reach a steady state when the duration of exposure exceeds the red cell life span. So, it is good for things like dietary exposure, smoking, exposures in the workplace, those kind of endpoints.


This is just a very crude sketch of what happens if you have a reactive chemical in the blood. This is the concentration with time and, as your chemical would go up and down maybe in one or two exposures on consecutive days the adduct concentration would increase and would reach plateau when your chemical has disappeared from circulation, and then would increase again on the second exposure. So, that is sort of an illustration of the integration. You are looking at accumulation over time and the dose or the area under the curve equals the adduct concentration that you measure divided by the reaction rate constant for basically a second order reaction, a rate constant that you can determine chemically in vitro. So, that is one way that we can actually go from an adduct measurement to an indication of what the area under the curve was and can give us a way of tying back into pharmacokinetics and pharmacokinetic modeling.


Again to review just a little bit of the history, the first measurements were done with cysteine adducts in the rat. There has been some indication of non-linearity, first in a study by Bailey and then in two studies by Calleman. These are probably the more authoritative ones. They measured adducts from acrylamide and glycidamide produced by reaction with cysteine. They detected the adduct from glycidamide as well as acrylamide. Then they did a very elaborate compartmental pharmacokinetic model and it ended up with a description of a saturable metabolic process, oxidation of acrylamide to glycidamide. what it implies is that there is a greater risk per unit exposure of low doses if glycidamide is the metabolite that generates adverse effects.


This just shows acrylamide internal dose that was calculated by Calleman. This is the administered dose in milligrams per kilogram body weight. You can see this is a straight line extrapolated from these points, and then this is the actual curve of the model with individual data points of each dose. So, that is for acrylamide.


For glycidamide it shows this evidence of saturation where, as you increase the dose up to 100 mg/kg, the amount of glycidamide produced is falling off.


Then, the last slide from this paper shows that they calculated the percent of acrylamide that would be metabolized through glycidamide at 15 mg/kg, which was actually the concentration we used or the dose that we used in our studies, and find about 30 percent or so metabolized through glycidamide in the rat. We got pretty good agreement. As you go down to lower doses the percentage that gets metabolized through glycidamide appears to go up.


Now I would like to switch gears to valine adducts. I am sure this is a name that is familiar to you all. Before Margarita was working on acrylamide in food she did a lot of work on hemoglobin adducts and actually developed one of the major methods that we use for acrylamide adduct analysis. That is when valine is involved in the adduct formation. It is the internal residue in both the alpha and beta chains, as I indicated before.

She developed a modified Edman degradation that selectively cleaves the N-terminus with its adduct, using a variant of the classic Edman degradation for peptide sequencing. This has been used widely with gas chromatography with negative chemical ionization mass spectrometry, and more recently with GS-MS. It is, however, a real pain to do this assay. It is tremendously labor intensive, although as assays go it is pretty good.


This method was applied in a number of studies. One was by Bergman et al. They looked at analysis of acrylamide and glycidamide adducts in China. Just to go through these really quickly, they looked at acrylamide and glycidamide adducts in 41 workers exposed to acrylamide. They only looked at glycidamide adducts in six samples and they find that there were about 30-100 percent of the acrylamide adduct levels. This methodology though left something to be desired. It was a completely different method for the AAVal or acrylamide valine versus the glycidamide valine.

There were other measurements reported by Emma Bergmark on acrylamide valine in smokers and non-smokers in laboratory workers. Then Perez et al. developed a method for acrylamide valine and glycidamide valine that was applied to workers in a production plant, and they recorded glycidamide valine levels that were about 3-12 percent of the acrylamide valine levels. These are probably a reasonably good measurement.

Then we have the Hagmar study, in 2001, where acrylamide valine was measured in 210 tunnel workers. This is the one that precipitated a lot of the measurements in food. In the unexposed workers it measured from 0.02 to 0.07 nanomoles/gram for acrylamide valine and in the exposed workers it was found in considerably elevated levels.


One of the things that I have been doing over the last few years has been to work on a number of different aspects. These are metabolic data on acrylamide and adduct data. One of these has been how is the internal dose related to exposure with different routes of exposure and different species. Is the GSH conjugation to oxidation ratio altered by exposure route and by dose?

We have looked at comparing route of exposure, dermal and inhalation and IP, and we have looked at comparing inhalation exposure in rats and mice, and looked at Cyp 2E1 metabolism and its role in acrylamide metabolism.


This just shows sort of a summary plot of glutathione conjugation versus oxidation. The first bar is the glutathione conjugation. The second bar is all of the metabolites of glycidamide. The last one is glycidamide itself in urine. When we look at dermal in the rat, it actually has the lowest amount of glutathione conjugation, the highest amount of oxidation, and compare that with inhalation IP and gavage and we see about 70 percent here at the peak. Whereas, in mice when we look at inhalation and gavage, we have a much greater percent of oxidation in the mouse compared with the rat.


We looked at metabolism of acrylamide via glutathione conjugation, oxidation to glycidamide and further metabolites of glycidamide in wild type mice and Cyp 2E1 null mice. We administered acrylamide at a dose of 50 mg/kg. In the wild type mice we find basically what we have seen before. About 50 percent or so was metabolized by glutathione conjugation, the remainder by glycidamide and glycidamide further metabolites. Whereas, in the 2E1 null we saw absolutely nothing derived from glycidamide or glycidamide itself. So, it argues that Cyp 2E1, at least in the mice, is the major or the only form of p450 that catalyzes the reaction and there doesn't appear to be any other oxidation pathway.


In my lab recently we have been working on a new method for adduct analysis, basically using the Edman degradation. We have adapted this for using LC/MS/MS for analysis. We get a higher throughput, a greater sensitivity with a smaller amount of globin and we can distinguish adducts from natural abundance acrylamide and 13C-labeled acrylamide. We have just recently published this, actually this month.


The method essentially works on taking our acrylamide valine hemoglobin, adding Edmund reagent and we selectively cleave our adduct and we form this kind of an adduct derivative, where this part, here, is derived from the acrylamide and this part, here, is derived from the valine, and this part, here, is derived from our derivatizing agent. We add an internal standard and we monitor by LC/MS/MS. We have three different ion reactions that we monitor. One is for our analyte; the second is for our 13C-labeled analyte where we have acrylamide administered to animals. The third one is our internal standard which is derived from labeling the valine with 13C.


We have a similar kind of procedure for analyzing glycidamide valine. This basically is the adduct. I won't bore you with the details here.


Our chromatograms look something like this. This is acrylamide valine. I think it is in the mouse. Here is our natural abundance channel, our 13C channel where we have administered 13C-labeled acrylamide, and this is our internal standard. From this we can understand our curves, we can quantitate and we can make measurements.


This is what we get with glycidamide valine. We actually have two isomers of our analyte and here is our internal standard.


We have basically looked at the same kind of studies as I mentioned with our metabolism studies so I won't run through all this.


We have looked at acrylamide valine and glycidamide valine in rats administered acrylamide. We have used a number of different routes and exposure scenarios. What I wanted to draw your attention to, since we are talking about food, primarily gavage as a route of exposure. We have done two studies here, at 50 mg/kg and at 3 mg/kg. These are actually exposed to 13C so the channel to really focus on is this one, AAVal and GAVal, and we see high levels at 50 mg/kg and much lower levels at 3 mg/kg.

The important thing is that the ratio of glycidamide valine to acrylamide valine changes in going from high dose to low dose. So, we do seem to have a dose-response difference in the percent going to acrylamide valine versus glycidamide valine, and that fits with the Calleman data.


I am just going to bypass this.


To look at inhalation exposure we used rats and mice actually during the same exposure event. Again, I am going to call your attention to these two sets of data and just look at the numbers. These got the same exposure and we have similar amounts of acrylamide valine from the acrylamide in blood, whereas in the glycidamide valine we have very considerably higher levels in the mouse than in the rat. Our ratio here is completely different. It is about 1.0 for the rat and about 3.5 for the mouse. So, we have big species differences and mice seem to make a whole lot more glycidamide valine than do rats.


So, what kind of range of adduct concentrations have we seen and what kind of ranges have been reported? There have been a number of studies by Bergmark, by Hagmar and by Perez. The background is sort of in general agreement in humans from these studies. Some are in the 20-70 range. In smokers that may be elevated. In acrylamide exposed workers we can have fairly high levels that have been reported and that vary from 300-34,000 in the Bergmark study and in the 3-12 percent range in the Perez study for glycidamide valine compared to acrylamide valine.

In rats, what is in the literature, apart from what I summarized on cysteine adducts, we have a background and in rats administered a fried diet there is an increase from about 20 up to about--well, an increase of the order of I guess somewhere around seven-fold. So, there has been reported an increase in rats fed a fried diet.


So I would like to conclude from some of our published work that we have seen route differences in internal dose in metabolism to glycidamide. We see that dermal administration results in a lower percentage of absorption. I didn't really go into that but I think dermal is one of the significant worker place exposure rates. We do see a species difference in metabolism to glycidamide.


We can readily measure acrylamide valine and glycidamide valine by LC/MS/MS. We do see a background for both of our adducts. The ratio of acrylamide valine to glycidamide valine depends on dose, rate of exposure and species. We do see a higher amount of glycidamide valine in mice compared to rats. Although I didn't really go into it, we do see a good correlation between metabolism data and hemoglobin adduct data.


I think one of the things we are all concerned about is where are the data gaps. I have my opinions on where these are. One of the big ones is when are we concerned about glycidamide and when are we concerned about acrylamide, and how metabolism fits into mode of action when you are dealing with risk assessment. How is acrylamide taken up and metabolized in people and what is the relationship between exposure and hemoglobin adduct levels in humans. Then, a final point is how good are the data on DNA adducts.


We have a number of studies in progress and I just wanted to summarize those very quickly.


The first one is a human study that has been funded by SNF. We are actually just doing the measurement of the adduct concentrations. That is to evaluate uptake of acrylamide, metabolism to glycidamide and to calibrate hemoglobin adducts with the known exposure and to compare uptake on dermal and oral administration. This was an exposure of sterile male volunteers to three different dose levels of 13C-labeled acrylamide, a dermal exposure to three times three doses of acrylamide, and then collection of urine for analysis of metabolites by 13C NMR, collection of blood for analysis of hemoglobin adducts before and after administration. We are almost at the conclusion of this study and we hope to have this finished probably by the end of next month.


There has been one paper published on acrylamide DNA adducts, and that is this paper by Segerback. They had synthesized an adduct, characterized an adduct standard from the reaction of glycidamide with guanine and then administered 14C-labeled acrylamide to rats and mice. They had identified DNA adduct based on comigration of 14C with adduct standard on HPLC. They quantitated the adducts.

These kind of approaches are limited by the specific activity of the acrylamide you can prepare and we do have more sophisticated methods for this. We have been working on one. I know that at NCTR there has been work going on, on this also and I think they are further ahead than we are. But I think it is something that needs to be done and we need to have a better understanding of is this, for example, the only DNA adduct and how is it repaired. There are quite a number of issues of how one deals with an adduct if it is formed.


I just wanted to finish up with this slide which is basically a summary of some of the points from the FAO/WHO consultation recommendations. At the time when these came out we were actually actively involved in pursuing some of these. One is evaluating and calibrating biomarkers of exposure. We need data on absorption, metabolism and distribution and excretion in humans by the oral route. We need information on glycidamide and binding to the DNA as a marker of toxicity and carcinogenicity. We need dose-response characteristics of acrylamide and glycidamide and the relationship between adducts with hemoglobin and adducts with DNA in different organs.

With that, I would like to stop and thank you for your attention.

Questions of Clarification

DR. MILLER: Comments or questions? Yes, Frank?

DR. BUSTA: I your human study that you are just concluding, the dose of 0.5--

DR. FENNELL: That was the lowest dose.

DR. BUSTA: Was it given orally?


DR. BUSTA: I am just guessing, I tried to calculate that, that is about 100 and 200 times above what you might get with a high concentration in food?

DR. FENNELL: Yes, I think we are talking about ball park. There are a couple of things that we were interested in, in this study. One is that when we initially developed our exposure protocol the driver was not acrylamide in food. The driver was acrylamide in the workplace. These are not, you know, outrageous levels that could be achieved in the workplace. I think that is issue number one.

The other is that when you come to measuring DNA adducts or hemoglobin adducts, or any of these things that have a background, and you are accumulating a background from the dietary exposure it is difficult to actually see a significant increment above that background without going to a higher dose than you would see ordinarily for example in food. So, in order to see, for example, a two-fold or just a doubling of a background in hemoglobin adducts you would probably need to administer something like 60 times the daily dose to see an increase statistically, and the levels that we operate for determining background that is quite a challenge. So, one needs to actually get to much higher levels in order to make that happen.

One thing that would did was to use 13C-labeled acrylamide so we are actually looking at a different species of acrylamide adduct and glycidamide adduct than we see in our indigenous background but even then we were concerned that it was going to be too big a challenge for us to go much lower than 0.5.

DR. MILLER: How do you deal with that when you have a substance that appears to be metabolized differently at high concentrations than low concentrations?

DR. FENNELL: I think the range that we are in--well, one of the things that we did want to find out was over that small concentration range did we see a difference. Certainly, what has been shown in the rat has been 0.5 to 100 mg/kg. So, in the range of 0.5 to 3.0 there is probably not going to be as wide a difference. You know, that big difference has been in the rat for the two dose points that showed the maximum or that showed the best evidence of saturation were at 50 mg/kg and 100 mg/kg. So, it is quite possible we won't be anywhere close to that at this point.

DR. MILLER: You said you calculated it was about 60 times?

DR. BUSTA: Well, if you have 1000 ppb in a food, I calculated 1 mg/kg of food and feeding at 0.5, for an average person that is 35 mg that you have to feed. So, you would have to have 35 kg of French fries to get that average load. Maybe my numbers are a little off.

DR. FENNELL: No, I think your numbers are about right. I think this is one of the issues. You know, this study is not going to be the answer to what happens to acrylamide in food but I think it is going to go a long way to providing a calibration for hemoglobin adducts which in the end will give us some idea of what is happening in food.

DR. BUSTA: Where is the background coming from?

DR. FENNELL: The background is coming presumably from food. One of the things that we have had difficulty with is as you get down to the low end of the calibration curve your matrix blank always has a measurement, there is always something in that, so how do you deal with that if you are going to use the appropriate matrix? We are still wrestling with some of those issues on our assay at the moment for dealing with our pre-exposure measurements. We clearly do see adducts. You know, it is quite easy to find them. I guess the question becomes when you do a calibration curve how you deal with them, and we are wrestling with some of those issues at the moment.

DR. BUSTA: I would think that there are groups of people that are consuming very, very low levels of acrylamide in the kinds of foods they are eating. So, you should be able to run an assay on their blood and find the minimum amount of adducts as a result of their consumption of that food. But you say you still get a baseline.

DR. FENNELL: We haven't done a huge number but we get a baseline even when we set up with rat hemoglobin where we are dealing with rats on defined diets, none of which we believe are fried but at least may be heated, and we do see a background. At the time we first observed it, it was more of an extreme annoyance and we thought, you know, when are we ever going to get a decent batch with a low background but now we know how to interpret this and it is more of an interesting observation.

DR. MILLER: Other comments?

DR. RUSSELL: With regard to the consultation recommendations that are up here now on the data on absorption, distribution and excretion in humans by the oral route, were those recommendations both for acrylamide in the food matrix as well as pure acrylamide? I guess another way of asking is not so much about the recommendations but is the work envisioned to go on in both realms, both looking at the pure acrylamide and in a food matrix?

DR. FENNELL: I wasn't at the consultation and I know there is at least one person in the room that was so.

DR. MILLER: Go ahead.

DR. CANADY: Yes, our intention was just for food but information from other routes would be informative to interpreting existing tox. studies.

DR. MILLER: Just one question, was there any suggestion made about the matrix itself, what the matrix should consist of? If everyone uses a different diet you may end up with different results.

DR. CANADY: Right. No, that level of detail was not gone into. It was more just the observation that we don't have information for absorption of acrylamide through the food matrix and so we need information that can help us interpret how we receive acrylamide through food versus what the animal studies tell us. That was a simple observation.

DR. MILLER: I have to admit that I am just uneasy about the apparent lack sometimes of agreement on fundamental standards. For example, if you are going to argue that there is a possibility that absorption or metabolism is going to have some impact on the outcome of the experiment, it would also be probably clear that different matrices are going to cause different effects, or at least it is possible. So, it seems to me there has to be some venue where people can agree on what that matrix is going to be. That equally applies to the analysis too. I mean, if everybody is using different analyses you have to show it doesn't make any difference, the differences between the different analytical methods, or you are going to have to agree on which method you are going to use.

DR. DWYER: I am just curious. It seems to me there are a lot of feeding studies around the country where people are fed defined diets or some kind of diet, kept samples of what they are fed or at least records of what they were fed, where they have blood samples. Have people looked at the similarities or has there been any work done to get an estimated dose and then looking at the hemoglobin adducts and so forth? Rather than starting from scratch and doing new studies, it would seem like this would be far less expensive and might give some first approximations.

DR. FENNELL: That is a good point. I think we are at a level of complexity in analysis that is much, much higher than acrylamide analysis in food when we are dealing with hemoglobin adducts, and the cost reflects that difference.

The way that we have collected samples in general has involved a fairly extensive sample preparation right at the point where the samples are collected. We wash the red blood cells is isotonic saline and then we store them frozen. We do that in order to decrease the amount of albumin binding or serum protein contamination. Once we have the red blood cell frozen it is pretty stable and we can do a lot of things with it at a later time.

So, I think if there are banked samples that are stored as red blood cells we think that we can analyze them. We are not quite there yet in having a definitive answer. You know, we can take an unwashed red cell where somebody has spun off the plasma or serum and see what we can get out of that. I think that is one important question. It has been raised to us. We have actually done one study and we think it was successful but we would like to do it a few more times before we make any definitive statements on it.

DR. MILLER: If you are going to use NHANES samples you are going to have to determine whether the storage conditions allow you to develop any worthwhile conclusions.

DR. FENNELL: To address just one point about NHANES, the procedure that is planned by CDC for NHANES is different from the one that I presented. I attended a meeting at CDC to discuss that and I think CDC's plan is that we will have some standards that can be passed around the labs to make sure that we are all in agreement as to what the measurement is, but they are planning to set up a new method which will involve HPLC mass spec. with tryptic digests and whether they can do that on banked samples or not I don't know.

DR. MILLER: If what you say is true, the sample storage might make a difference in the results.

DR. FENNELL: I think the way the sample is stored can make a difference depending on whether you want to store it to go prospectively or whether you are looking at retrospectively stored bank samples and I think they are looking at it from the standpoint of going forward from here on.

DR. MILLER: I see.

DR. DWYER: Would you say that again? Why are they developing another method if you already have a method?

DR. FENNELL: I am not sure I can give you an answer for that.

DR. DWYER: Aren't you all being funded by the Department of Health and Human Services?

DR. FENNELL: I am not. I am funded entirely by either by grant or contract and I am not funded by CDC at all.

DR. CANADY: One of the reasons the CDC has given for developing a new methodology is that the same methodology, if used for acrylamide and glycidamide adducts, could also then be used for other adducts to other chemicals. So, they are trying to develop a more generalized approach that they can use for other chemicals that they would asses through NHANES.

Having said that, they are in the process of determining what the method will be, whether it works and how well it correlates with other methods. They haven't decided on the method they will use, rather, they are developing a method that is appropriate. So, it is still in a method development stage although they are taking another approach, as Dr. Fennell indicated. They are trying to take another approach.

DR. DWYER: What are they doing with the blood they are collecting each day while they are waiting since the survey is in the field right now?

DR. CANADY: Right. The red blood cells are being stored for this purpose. There are samples being set aside of red blood cells for future analysis including acrylamide adducts and glycidamide adducts.

DR. MILLER: That is bothersome, frankly. I can understand why they would want to develop a more multi-purpose analysis given the things they have to deal with, but this is an issue that needs some resolution, and why wait until they have a multi-purpose analysis when it might be easier just to produce a more specific assay to begin with?

Any other comments or questions? Since there is no one registered for public comment, which is required by the rules of the committee, then we have come to the end of our first day's activities and, unless there is some further question or comment from the committee, I am going to adjourn the committee until tomorrow morning at 8:30.

[Whereupon, at 4:00 p.m., the proceedings were adjourned, to resume on Tuesday, February 25, 2003 at 8:30 a.m.]

Presentations available on request.

Contact FDA

Outreach and Information Center

10 AM- 4 PM EST
Closed Thurs 12:30PM - 1:30PM EST


Inquiries: Submit Your Question


Center for Food Safety and Applied Nutrition
Food and Drug Administration

5001 Campus Drive
College Park, MD 20740

Industry and Consumer Assistance

Page Last Updated: 09/15/2015
Note: If you need help accessing information in different file formats, see Instructions for Downloading Viewers and Players.
Language Assistance Available: Español | 繁體中文 | Tiếng Việt | 한국어 | Tagalog | Русский | العربية | Kreyòl Ayisyen | Français | Polski | Português | Italiano | Deutsch | 日本語 | فارسی | English