APPENDIX B
Case Studies of the Implementation of the Direct Food and Color Additives Amendments to the Federal Food, Drug, and Cosmetic Act of 1938a

Technological And Social Factors That Have Affected Introduction Of New Direct Food Ingredients And Processes

Introduction

The companion paper by Lars Noah (1997) provides a well-documented history of the development and application of the legal and regulatory background of the present processes for the review and approval of new direct food ingredients and technologies into the marketplace.

The case histories that follow are intended to illustrate how that legal and regulatory structure has operated in actual practice over the past four decades. The purpose of these case studies is not to point out error or assess blame. Instead, the value that can be extracted from these histories is to learn what specific features of process and procedure have been effective in protecting public health while permitting innovation, and why they have been effective. We wish also to learn which features have not been effective, and why. Efforts to improve the overall structure will be more effective when it is possible to draw carefully and dispassionately from experience.

The cases summarized did not evolve in an atmosphere defined solely by statutory and regulatory boundaries. Rather, they have been heavily, and sometimes decisively, influenced by changes in available technology and in society in general. This brief introduction cannot examine these influences in detail. Simply listing them with minimal comment, however, begins to suggest the impact they have had on attitudes and lifestyles, and most importantly, on the approval processes that are the subject of this exercise.

Developments in Analytical Chemistry

Since 1958, one of the most significant developments in technology has been the dramatic increase, approximately six orders of magnitude, in the sensitivity of instrumental methods of chemical analysis. Discrimination and speed have increased almost as much, with a consequent, and comparable, reduction in cost per analysis. As a consequence, there is now far more information on the enormous complexity and variability of the trace constituents in the food supply (IFBC, 1990; NAS, 1996). The great majority of these constituents occur naturally. A small minority, but still very many, are present as a result of human activity. Analytical

a  

Prepared by Food Forum members and consultants, John Kirschman, Ph.D., Kirschman Associates, C.K. Gund, Ph.D., and Clyde Takeguchi, Ph.D., Phoenix Regulatory Associates, Ltd.



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 111
APPENDIX B Case Studies of the Implementation of the Direct Food and Color Additives Amendments to the Federal Food, Drug, and Cosmetic Act of 1938a Technological And Social Factors That Have Affected Introduction Of New Direct Food Ingredients And Processes Introduction The companion paper by Lars Noah (1997) provides a well-documented history of the development and application of the legal and regulatory background of the present processes for the review and approval of new direct food ingredients and technologies into the marketplace. The case histories that follow are intended to illustrate how that legal and regulatory structure has operated in actual practice over the past four decades. The purpose of these case studies is not to point out error or assess blame. Instead, the value that can be extracted from these histories is to learn what specific features of process and procedure have been effective in protecting public health while permitting innovation, and why they have been effective. We wish also to learn which features have not been effective, and why. Efforts to improve the overall structure will be more effective when it is possible to draw carefully and dispassionately from experience. The cases summarized did not evolve in an atmosphere defined solely by statutory and regulatory boundaries. Rather, they have been heavily, and sometimes decisively, influenced by changes in available technology and in society in general. This brief introduction cannot examine these influences in detail. Simply listing them with minimal comment, however, begins to suggest the impact they have had on attitudes and lifestyles, and most importantly, on the approval processes that are the subject of this exercise. Developments in Analytical Chemistry Since 1958, one of the most significant developments in technology has been the dramatic increase, approximately six orders of magnitude, in the sensitivity of instrumental methods of chemical analysis. Discrimination and speed have increased almost as much, with a consequent, and comparable, reduction in cost per analysis. As a consequence, there is now far more information on the enormous complexity and variability of the trace constituents in the food supply (IFBC, 1990; NAS, 1996). The great majority of these constituents occur naturally. A small minority, but still very many, are present as a result of human activity. Analytical a   Prepared by Food Forum members and consultants, John Kirschman, Ph.D., Kirschman Associates, C.K. Gund, Ph.D., and Clyde Takeguchi, Ph.D., Phoenix Regulatory Associates, Ltd.

OCR for page 111
methodology is now at the point where it is almost possible, with sufficient effort, to find traces of almost anything, in virtually everything. This awareness has complicated enormously the task of devising appropriately protective specifications for food-grade ingredients. Particularly for pesticide residues, it has also undermined the simplistic black-or-white, permit-or-prohibit, zero tolerance assumptions on which the Delaney clause was based. Developments in Toxicology In the past four decades the cost in constant dollars and complexity of toxicological tests, particularly the two-year chronic (carcinogenicity) bioassay, have increased several fold. A fundamental tenet of toxicology is dose/response—the higher the dose, the more frequent and severe will be the adverse effects. Humans are almost always exposed to comparatively very low doses—doses at which the chance of finding adverse effects in a reasonable number of animals would be negligible. In order to be likely to detect adverse effects, particularly carcinogenicity, dose levels have increased, often to the "maximum tolerated dose" (MTD). This greater assurance of finding adverse effects is purchased at the cost of possible, and sometimes demonstrated, lack of relevance to humans at the low doses humans normally ingest. Since 1958, the number of animals per test level has more than doubled and the number of tissues and organs typically examined has increased from 10 to 50. In addition to traditional endpoints such as carcinogenicity, neurological, behavioral, and immune-system effects are now being studied. However, the debate about how to measure, analogize, or transfer results in test animals to humans continues. In 1973, a quick, inexpensive screening test was developed by Ames and co-workers (Ames, 1973) with the aim of predicting carcinogenicity of a broad range of substances. This test measures the mutagenic potential of substances using, as the test organism, a genetically modified bacterial strain, Salmonella typhimurium. Mutagenesis is the ability of chemicals to cause changes in DNA, RNA, and other cellular macromolecules. This hoped for relationship between mutagenicity and carcinogenicity that underlies the Ames test was questioned when substances with no known carcinogenicity tested positively. Mutagenic potential, however, remains a serious concern even though, as yet, no heritable human disease has been traced to a chemical mutagen. The Ames test, and the later genetic toxicology tests it stimulated, have become a useful way of distinguishing those ("genetic") carcinogens that act through attack on DNA and RNA from "epigenetic" carcinogens acting by more indirect mechanisms. Obviously, these changes have vastly increased the volume and complexity of data that must be collected, recorded, analyzed, and evaluated as part of the approval process. The Emerging Understanding of Mechanisms Lagging greatly behind the advances in analytical chemistry, and the increasing power of toxicological testing, has been the development and acceptance of the sciences that are coming to be increasing useful in the design of, and the interpretation of, the results of analytical and toxicological testing. Among these interpretive sciences are: Comparative Metabolism—studying the similarity or differences between test animal and human metabolism of a substance; Structure /Activity Relationships (SAR)—the study of the effect of the chemical structure of a substance on its biological activity;

OCR for page 111
Physiologically-Based Pharmacokinetics (PBPK)—detailed study of the rates, volumes, and capacities of a test animal's metabolic pathways at varying doses; and Molecular Biology—the internal biochemistry of the individual cell, with particular focus on the role of the nucleic acids in determining cellular response. These interpretive sciences have begun to contribute heavily to understanding the gross results of toxicological studies. They help to separate adverse effects relevant to humans from those that are not relevant, and genuine effects from spurious ones. The above sciences greatly improve the estimation of actual human health risks. Unfortunately, they involve a broad range of cutting-edge science not available in-house to any single organization—governmental, industrial, or academic. All organizations, therefore, require extensive outside expertise in order to be effective. Only in the past 10 to 15 years has it been widely accepted that cancer is a genetic disease, that is, due to changes in the genetic makeup of cells. These changes are usually triggered and heavily influenced by environmental and lifestyle factors. This is a key portion of our current understanding, far broader and more specific than in 1958, of the extent to which diets and lifestyles affect the risk of chronic diseases, including cancer, coronary heart disease, and stroke. This improved understanding of cancer is a direct outcome of molecular biology, in particular. Sound toxicology has always held that no single test or criterion can be decisive. All of the data must always be examined. These newer interpretive sciences, however, have resulted in a gradual, but major diminution in the role of the once controlling chronic study, and an insistence on additional supporting and interpretive data before applying the results of such studies. Changes in Risk and Safety Assessment The more obvious safety decisions on direct food ingredients (substances intentionally added) and constituents (substances naturally or unavoidably present) have been made simply and easily. Some substances are so clearly toxic that regulations specifically exclude them (dulcin, CFR Sec. 189.145). Others are so obviously safe because of lack of toxicity (sugar or salt) or because of trivial exposure (hydrogen cyanide, cucurbiticin E, and other natural toxicants), that regulations do not specifically deal with them and they are treated under the general provisions of the FD&C Act. In such cases, a formal process of risk assessment is neither needed or used. It is the large group of "in between" substances, too useful and not toxic enough to prohibit, too risky to leave totally uncontrolled, that require some level of risk assessment, and that may also require risk management. This can range from good manufacturing practices (GMPs) or labeling to tightly restricted uses. Until 30 years ago the only internationally recognized method of establishing a safe level of exposure (consumption) of a substance was to take the highest dose that produced no observed adverse effects in animals (the NOEL), and divide it by a suitable safety factory, often 100, to obtain the acceptable daily intake (ADI) which humans might safely consume. For most toxic endpoints, such as neurotoxicity or non-carcinogenic organ damage, this approach continues to be used and well serves human health. FDA commonly uses the concept, although seldom employs it explicitly, preferring instead to rely on "the weight of all of the evidence." In the last few decades, however, and particularly for substances that show carcinogenic effects in test animals, that approach has been replaced by "quantitative risk assessment" (QRA). Although it can be mathematically complex, QRA usually involves a series of conservative "default" assumptions. These default assumptions are made where the needed biological data simply do not yet exist. They typically deal with:

OCR for page 111
The estimated incidence of adverse effects that would be found at the levels to which humans are exposed, usually several orders of magnitude below the lowest animal dose (extrapolation); The assumption that humans will react like the most sensitive test animal (analogy); The maximum possible level of lifetime human exposure, because actual human exposures are often not well known. All of these assumptions are fraught with uncertainty. Depending on the particular assumptions used, extrapolation may grossly under- or overstate human risk. The default assumption is therefore conservative to avoid understatement. Because of metabolic differences, the test animal may be a poor model for human risk, and better understanding of mechanisms (see previous discussion) is intended to minimize this problem. However, QRA can seldom factor in human differences in susceptibility, except in the same way that the safety factor is used in the ADI. Exposure estimates range from the mean or 95th percentile, when available, to the maximum possible, or something in between. They, too, tend to be conservative. This process leads to what might reasonably be called a ''probable upper bound" of risk which is usually expressed as a lifetime risk of one-in-a-million or some other large number. Unfortunately, it is often carelessly and misleadingly called "the risk." QRA has some clear merit, particularly when default assumptions can be replaced by biological data. It also appears to deal more comfortably with those undesirable trace constituents chemical analysis indicates are present. But it also has the deceptive appeal of a "hard number" that often conceals the softness of the underlying data. We are never exposed simply to one chemical substance, natural or synthetic. We are constantly exposed to a shifting pool of complex mixtures of substances, most of which we encounter at very low levels. How to assess the safety of mixtures has become a matter of increasing interest, because testing them all, in every conceivable proportion, is obviously impossible. In addition, there is pressure from various interest groups to reduce the use of animals in toxicology testing. Substances that are metabolized by different routes are very likely not to be additive, in either load or risk. Where, because of exposure or toxicity, a formal risk assessment is appropriate, each substance should be evaluated independently, not lumped together. Absent special reason for concern, very low level exposures can be safely ignored. Substances that act by the same routes may constitute additive risks. However, if the level of exposure is many orders of magnitude below the levels at which effects have been observed in test animals or humans, the significance of possible synergistic effects is still likely to be negligible. This is not true as exposure to a component of a mixture approaches the ADI, or some other appropriately conservative interpretation of the available biological data. A useful discussion of this complex problem is found in 1997 report of the Presidential Commission on Risk Assessment and Risk Management. Changes in Agriculture In 1958, classical genetics had already demonstrated its power with hybrid corn, and the "green revolution" was about to begin. But the family farm, which produced enough food to feed 27 people, was already disappearing. Today, each farmer produces food to feed 128 people at home and abroad. Thus, farming techniques have changed dramatically.

OCR for page 111
Agricultural technology seemed always to favor larger operations with the resources to make use of new technologies including irrigation control, herbicides, pesticides, plant and animal hormones, and low-dose antibiotics. This trend added to the perceived distance, geographic, economic, and technologic; which separated consumers from the source of their food, and with this increasing distance inevitably came increasing uncertainty and discomfort. Environmental Concerns Rachael Carson's Silent Spring was published in 1962. Prescient in some respects, erroneous in others, its poetry and power mainstreamed the environmental movement, and promoted an often healthy, but poorly informed, skepticism of new technologies. Current awareness of the potential environmental consequences of industrial and governmental activities was largely lacking in 1958. Environmental impact assessments are a permanent feature of new ingredient and technology development. Consumer Activity Consumers as individuals and as organizations have contributed significantly to raising and debating public health and economic issues. Consumer pressure and publications were prominent in the passage of the 1906 and 1938 Food and Drug Acts and in activity leading to the Food Additives Amendment of 1958. Since then, they have been frequent, often vocal, but unevenly effective contributors to food safety issues. Part of this uneven effectiveness may be due to thinly spread resources and limited access to the expertise necessary for dealing with large volumes of data and complex issues. In addition, it may also be due to their continuing concern with social and economic issues, as well as the science. Social and economic issues, however important, lie outside the regulatory authority of FDA. Their perceived urgency may lead to an artificial and unproductive prolongation of the safety discussion as the only forum in which a new technology can be discussed and opposed or delayed. Changes in Food Marketing and Consumption Trends Consumer purchasing trends, visible in the marketplace, do not necessarily reflect the agendas of consumer organizations, and these purchasing trends often are contradictory. Fortythree percent of food dollars are now spent on meals eaten away from home and this trend continues. The percent of foods subjected to prior processing before reaching their point-of-sale continues to increase. Since 1958 there has been a generally, although unevenly, increasing number of new food products introduced each year. Most new food products fail, but there are some monumental successes. New food processes have been introduced that include extrusion cooking, pulsed light, aseptic packaging, membrane filtration, and supercritical carbon dioxide extraction. Food costs by families and individuals as a percent of disposable income have continued to decrease, and now stand at 11.2 percent. Nutrition has become a far more prominent and widespread concern, with a consequent interest in lower fat, lower calorie, reduced salt, and higher fiber foods and the ''food pyramid," even in elementary schools. And, somewhat contradictorily but simultaneously, "minimally processed," "organic," and "natural" foods, often at substantially increased prices, have moved from niches into supermarkets. These developments in the U.S. food supply have depended on new or modified ingredients and processes. For example, lower calorie and lower fat foods require new processes, such as microparticulated protein (SimplessTM), non-metabolizable fat replacements (Olestra), and non-

OCR for page 111
or low-calorie sweeteners (saccharin, NutrasweetÔ, Acesulfame K, etc.). Thus the list of substances and processes seeking market entry grows steadily. Congress has enacted laws to increase information and economic protection for consumers, including Fair Packaging and Labeling Act (FPLA) and the Nutrition Labeling and Education Act (NLEA). The Impact on FDA In summary, all of the following factors have added enormously to FDA's task load since 1958: the knowledge of the presence of trace constituents and contaminants in food; more sensitive, more complex toxicological testing, involving much more data and the need for far more interpretive expertise; the understanding of the means, the biological mechanisms, by which substances exert their toxic effects; the growing need for a broad range of external expertise; the need of assessing the environmental impact of every major action; the increasing globalization, diversity, and complexity of the food supply; increasing dependence on new ingredients and processes; and FPLA, NLEA, and the higher level of consumer interest in nutrition, in "healthy lifestyles," and in influencing regulatory decisions. By contrast, in recent years, the resources of the FDA have remained almost level in constant dollars. The overwhelming competing pressures on the federal budget strongly suggest that the future will be no brighter. FDA has responded to these pressures with a number of changes in the premarket evaluation of food additives, including the Prioritizing the Assessment of Food Additives (PAFA), the concept of the "threshold of regulation," "levels of concern" in the Redbook, ''fast-track" approvals for straightforward and well-supported petitions, and other measures discussed in this symposium. All of these measures are aimed at prioritizing tasks and sensible use of limited resources. That directly complements the purpose of this symposium. Provisions That Continue to Work Well after Six Decades Surprisingly, in the midst of all these changes, two provisions of the FD&C Act have continued to serve the public health well. The "does not ordinarily render it [the food] injurious to health" standard applies to foods themselves. If that standard were significantly more strict, there would be little left to eat, for nearly all foods contain natural toxicants capable of causing harm under improper circumstances and with relatively narrow margins of safety (NAS, 1996, 1973; IFBC, 1990; Hall, 1977). The standards, "may render it injurious," in the statute, and "reasonable certainty of no harm," from the legislative history of the 1958 Food Additives Amendment, are more stringent standards that are appropriate for those changes that are intentional and which are therefore under human choice and control.

OCR for page 111
Conclusion The thoughtful reader of the case studies that follow will note the pervasive impact of these technological, social, and economic factors, not only on the cases themselves, but on the lessons that can be extracted from them. Cyclamate Overview In 1969, the FDA banned cyclamate from use in food after studies suggested that it might be an animal carcinogen. The reintroduction of this food additive petition, by Abbott Laboratories, Inc. in 1973, placed FDA under tremendous scrutiny by all interested parties. The agency had to consider whether to reverse its previous decision while assuring the public that the statutory safety standard would be met. Cyclamate's previous regulatory history led to heightened sensitivities for both the agency and the petitioner resulting in the FDA Commissioner's direct involvement from the time of initial submission. Over 300 published and unpublished studies were submitted. Some studies were poorly designed but reported adverse effects, such as carcinogenic bladder tumors, testicular atrophy, and cardiotoxicity which caused major concerns to the regulatory scientific reviewers. When scientists from various disciplines review complex issues, there are often disagreements about final interpretations and when placed under tight review time frames, most regulatory scientists will exercise the worst-case scenario. There were several rounds of scientific review and debate about incomplete data and interpretation of data. Some issues were resolved while others remained as major scientific safety concerns. Lack of communication between the agency and the petitioner led to distrust between them. Another major point of departure from a typical review was the Commissioner's public announcements through talk papers at every decision point and the placement of the entire food additive petition minus trade secrets at the public docket for review. The agency moved very quickly in reviewing the enormous data package and found many deficiencies which led to a rejection letter within the first year. When the Commissioner found the petition inadequate a second time, he assured the public that FDA would not act alone in its scientific review but would ask an outside expert panel to review and provide recommendations. The National Cancer Institute (NCI), a sister agency, served as the expert panel. NCI played a major role in the government's decision to ban cyclamate in 1969. The introduction of the NCI's Temporary Committee's review delayed the final decision for nearly three years. Another departure from a typical review was the manner in which the final decision was reached. Although the FDA reviewing scientists concluded that the NCI's Temporary Committee had resolved the carcinogenic question and the FDA's management recommended limited approval, the Commissioner, after personally analyzing NCI's Temporary Committee report, concluded that cyclamate could not be approved for use in food. With the formal denial, Abbott requested a formal evidentiary hearing. This administrative procedure took another four years with vast resources utilized by both the agency and the petitioner. Although the Commissioner's final denial came as no surprise to some observers, there were many from the scientific community that felt a total miscarriage of regulatory decision-making had taken place.

OCR for page 111
Chronology 1961 FDA advise Abbott laboratories that sodium cyclamate was GRAS as a food substance. 1969 An independent laboratory conducting a two-year feeding study in rats, using the 10:1 ratio of cyclamate:saccharin, reported the presence of bladder tumors. NCI concurred with the finding of carcinogenic bladder tumors and cyclamate was removed from general purpose use in food. There was no evidence that cyclamate caused cancer in humans. 1973 On November 13, Abbott submitted a petition to gain approval for cyclamates as a new food additive. The petition provided all relevant data for cyclamate and the cyclamate/saccharin mixture for the twenty years that it had been used in the United States, plus new unpublished data on cyclamate and cyclohexylamine (CHA). 1973 On December 21, FDA sent Abbott a letter of filing. 1974 FDA toxicology review stated that no conclusions based on the reference articles submitted could be drawn, but concluded that cyclamate must be considered a weak carcinogen. The agency concluded that the mutagenicity data were sufficient. 1974 On September 5, FDA sent a letter to Abbott asking that the petition be withdrawn without prejudice for a future filing. The letter included a discussion of the inadequacies of the toxicology data and the various problems found in the chemistry review. 1974 On September 10, Abbott met with the FDA Commissioner and the agency's staff. The petitioner took exception to the request for additional studies 1974 On November 1, Abbott responding to the rejection letter, submitted additional data. They disagreed with nearly every issue raised by the agency and noted that 22 long-term studies in rats, mice, and hamsters had been negative for carcinogenic potential for cyclamate and CHA, while bladder tumors were observed in only two rat studies, neither of which had been designed to test for carcinogenic potential of cyclamate. These two studies were further complicated by the presence of bladder calculi and parasites and Abbott declared "the suggestion of carcinogenicity of cyclamate has no scientific validity." Abbott found no significant testicular changes based on two other studies in which researchers fed CHA to rats over 24 months and mice over 80 weeks and concluded that testicular changes probably emanated secondary to dietary inanition 1974 On November 13, FDA held a public meeting. Abbott provided scientific expert opinions on the inappropriate interpretation by the FDA of the submitted studies and indicated that data were available to demonstrate that cyclamate does not cause teratogenic effects in mice, rats, and rabbits. 1975 In March, FDA rescinded the request to withdraw the petition based on the evaluation of data submitted by Abbott on November 1, 1974, and the transcripts of the November 13, 1974 public meeting. 1975 On March 14, FDA requested that NCI convene a blue ribbon panel of oncologists to decide the carcinogenic issue. 1975 On November 17, Abbott submitted more data and concluded the following: NOEL of CHA in the diet was 5,000 ppm; testicular changes in rats treated for long duration (1828 months) with large doses of cyclamate (3-5%) probably resulted as a secondary response to nutritional imbalances; no testicular changes had been seen in monkeys treated with 500 mg/kg/day cyclamate for 5 or 7.5 years; no testicular effects observed in rats fed 10% cyclamate for 12 months; and the data indicated that an ADI for cyclamate of 3 g could be established. 1976 On March 8, the NCI Committee concluded that the evidence did not establish the carcinogenicity of cyclamate or CHA in experimental animals. No conclusion could be

OCR for page 111
  made on potential carcinogenicity in humans due to the short post-exposure observation time and the insensitivity of epidemiological studies to detect relatively small changes in cancer incidence and other factors. 1976 FDA final toxicology memorandum offered several additional conclusions as follows: (1) the NCI Committee appeared to have resolved the carcinogenesis issue; (2) the ADI could range from about 200-300 mg/day for the average 60 kg adult, but should be proportionally less for children; (3) with this ADI, general use of cyclamate could not be approved; (4) the tablet and drop form for home use should be the item of choice for the individual who needed artificial sweeteners; (5) specific label instruction would be helpful because the ADI could be exceeded by some individuals; and (6) use in soft drinks should be prohibited because excessive use by the young would soon exceed the ADI. 1976 The agency informally recommended limited use of cyclamate for consideration by the Commissioner. 1976 On May 11, because of concerns raised by the NCI Committee, the FDA denied the Abbott petition and requested withdrawal without prejudice to a future filing. 1976 In a June 16 letter, Abbott decided not to withdraw the petition and stated that the Chairman of the NCI Committee believed that the conclusions in the agency's letter were not consistent with the NCI report. The petitioner offered to submit additional data and to conduct an experiment to demonstrate that the proper NOEL of CHA would be 5,000 ppm rather than 2,000 ppm. This compromise was offered if the Commissioner modified his position and allowed cyclamate to return to the marketplace. 1976 On October 4, FDA announced the formal denial of the petition to re-market the artificial sweetener. 1976 On November 3, Abbott requested a formal evidentiary hearing. 1977 The formal hearing with oral testimony and cross-examination took place. 1978 On August 4, the administrative law judge concluded that cyclamate has not been shown to be safe, to not cause cancer in humans or animals, and to not be a mutagen. In addition, if the carcinogenicity and mutagenicity questions are resolved, the record would not support a finding that the ADI is 5 mg/kg/day or less and does not establish probable consumption patterns of cyclamate to the extent necessary to establish safe conditions of use. 1979 On June 29, the FDA published a Notice of Interlocutory Decision that reopened the proceeding to develop the record more fully for the identified areas of concern. 1979 At an October 22 hearing, testimony concerning statistical and biological significance was presented by both the FDA and Abbott. 1980 On February 4, the administrative law judge concluded that the evidence suggests that cyclamate may be a carcinogen, but that the record falls short of establishing that cyclamate is a carcinogen. However, Abbott has failed to establish that cyclamate is not a carcinogen. 1980 On September 16, the FDA published the Commissioner's Final Decision denying the food additive petition for cyclamate. Comments During these above seven years, many congressional and consumer inquiries were received—some mandating that FDA place cyclamate back in the marketplace and some adamant that this non-nutritive sweetener was unsafe and lauding the agency for its sound

OCR for page 111
decision-making in banning it. All letters were answered, thereby depleting resources that could have been used to review scientific documents and develop appropriate regulatory policy for this matter. Although this case study ends with the Commissioner's decision to uphold the denial of cyclamate for use in food, the process is still ongoing. Another petition for cyclamate's use is pending before FDA. In addition, other major government regulatory bodies (European Union, Canada, Australia) have reassessed the regulatory status of cyclamate and allowed it back into the marketplace, with some countries acting within a few years after the initial U.S. ban. Lessons Learned 1.   Communication has to be mutual between the petitioner and agency. 2.   When a chemical's history results in notoriety, the agency can decide on new procedures to manage the approval process. 3.   Petitioners and the FDA both need to follow the rules. 4.   Petitioners need to understand the basic petition requirements before submitting a formal petition. 5.   When homework on scientific issues presented in the petition is not adequate, the petitioner is the loser. 6.   A petitioner of a controversial substance needs to recognize the in-depth scrutiny and resolve issues satisfactorily at the first encounter. 7.   Incomplete submissions lead to extensive questions and resource utilization by the agency and ultimately the petitioner. 8.   Identical data can be interpreted differently by competent scientists leading to lost time, resources, and difficult decision-making, because the regulatory process typically feels compelled to support the most conservative interpretation and outcome. 9.   Scientific positions change as new information become available (e.g., the percent cyclamate converted to cyclohexylamine) illustrating that communicating findings and documenting with data facilitates resolution. 10.   In-house conflicts on scientific issues result in lost time. 11.   Careful selection of outside expert panels and educating their members on the regulatory safety evaluation process is mandatory for assuring relevant and sound recommendations. 12.   Rushing the process does not necessarily result in a sound decision. 13.   Decisions involving a controversial subject will always require an unusually long time for closure. 14.   The Commissioner is the senior manager and final decision maker of FDA and he/she should carefully consider the final recommendations of the agency's scientists together with those from outside experts and other regulatory bodies when making a controversial decision. Irradiated Poultry Overview Work on the use of radiation to reduce or eliminate microbiological and insect contamination of food, inhibit sprouting, and for other purposes, began shortly after World War II. This was the first new food preservation technique to be developed since the invention of canning by Appert early in the 19th century. It was therefore, the first major food process to be tested for safety by

OCR for page 111
the modern methods of analytical chemistry, microbiology, and toxicology. It also resulted, for the first time, in foods which could not be tested by conventional toxicological methods because test animals cannot be fed enough of the test food to provide conventional margins of safety. Attempts to do so cause nutritional and toxicological problems in the test animal unrelated to the irradiation. Much of the early testing of irradiated foods, done in the 1950s and 1960s, did not fully recognize these problems, was poorly designed, and gave uninterpretable results that can be attributed to the irradiation conditions, palatability of the irradiated food, and the effect of feeding animals a diet that was nutritionally compromised. Thus, there was little useful background for successful rulemaking on which later petitions could draw. The rulemaking process for the irradiated poultry regulation is unique because it involves two petitions, one from a contract radiation company, Radiation Technology, Inc. (RTI), and the second, from USDA's Food Safety and Inspection Service. RTI submitted safety studies generated in foreign countries by other sponsors that were originally submitted as data for the safety of food irradiation in those countries. FDA requested additional studies on the effect of irradiation on C. botulinum and competitive vegetative bacteria. In addition, FDA had specific questions about some of the safety studies submitted by RTI. However, RTI could not respond adequately to specific questions about the studies because it was not the designer or sponsor of the some of the studies. Chronology 1978 RTI submitted petition describing a process to irradiate chilled poultry at a dose range of 3-7 kGy. 1978 FDA stated need for data on rates of growth of C. botulinum in poultry at good refrigeration and higher temperatures. 1979 FDA established the Bureau of Foods Irradiated Foods Committee (BFIFC) to focus on how the safety of irradiated foods can be scientifically evaluated, applying scientific principles and rationale. Thus, much of the insight needed to draw the conclusions FDA eventually reached had not been developed when the petition was submitted. 1979 The FDA review of toxicology data resulted in a request for additional data from the animal feeding studies by the Central Institute for Food and Nutrition Research, TNO Laboratories, The Netherlands, and Bio Research Laboratories. 1981 Joint FAO/IAEA/WHO Expert Committee on the Wholesomeness of Irradiated Food (JECIF) concluded that the irradiation of any food commodity up to the overall average dose of 10 kGy presents no toxicological hazard. 1981 BFIFC recommended that (1) food irradiated at doses below 1 kGy (100 krad) was wholesome and safe, (2) food (e.g., dried spices) that comprise only a small fraction of the diet and irradiated at doses up to 50 kGy (5 Mrad) is safe, and (3) food irradiated at doses exceeding 1 kGy be subject to toxicological testing consisting of a battery of four short-term mutagenicity tests and two 90-day feeding studies. 1983 RTI changed maximum permitted dose to 3 kGy based on a report that demonstrated that C. botulinum type E does not pose a problem for poultry irradiated at a maximum dose of 3 kGy and under adverse conditions.

OCR for page 111
1975 to 1997 TCE continues to be listed by FDA for use as a solvent in food processing, including coffee decaffeination. Dichloromethane (DCM, Methylene chloride) 1975 Although FDA did not move to de-list the food uses of TCE, the affected industries felt that with the cloud over the substance, they needed alternatives. An intensive search began for a technically adequate and commercially feasible alternative decaffeination solvent. While DCM had not undergone long-term carcinogenicity testing comparable to the NCI bioassay, there were chronic feeding studies on coffee products decaffeinated with DCM, showing no adverse effects. Also, SAR analysis suggested that DCM did not form an epoxide intermediate similar to TCE, which was suspected to mediate TCE-induced cancer effects. Based on this evidence, coffee manufacturers that had been using TCE switched to using DCM, and the NCA immediately began discussions with NCI and FDA expressing its commitment to supporting the very best research program possible for evaluating the safety of DCM for use as a decaffeination solvent. 1976 NCA began a research and testing program on DCM. In the course of that effort and related activities by individual firms, General Foods spent $300,000 for metabolism studies for use in PBPK evaluation and for designing chronic drinking water studies in rats and mice. 1978 NTP's chronic gavage studies on DCM in rats began at Gulf South Laboratories. 1979 Through the Interagency Regulatory Liaison Group, FDA scientists agreed to use the MTD approach to dose selection despite their expressed reservations over its use for testing this substance. NCA began its chronic rat and mouse drinking water studies on DCM at Hazleton Laboratories. 1980 The NCA unprecedented "dancing mice," not DCM or dose-related. This phenomenon required urgent assembly of a wide variety of specialists. This observation was attributed to the single housing of the B6C3F1 mice, never before seen at NCI, NTP, or animal suppliers, where mice are group housed. 1982 NCA studies on DCM reported no adverse effects. 1984 The Nutrition Foundation sponsored a Food Solvents Workshop. Dr. Robert Squire, formerly head of the NCI Bioassay Program, stated: "NCA's testing program for DCM is an example of how a material should be tested. In 8 years at NCI, this is the first chemical tested by industry we discussed with industry representatives." 1984 Chronic studies under contract from NTP on DCM, 1,1,1-trichloroethane, and methyl chloroform, were so seriously flawed that the toxicology unit of Gulf South Laboratories was shut down. 1986 Test results showed orally administered DCM not to be carcinogenic at high, but less than MTD doses, in either rats or mice. 1986 The NTP reported finding elevated liver and lung tumors in mice, but not rats, exposed to DCM in inhalation at the MTD. 1996 Research continues. Twenty years of intense and costly research, including the first use of PBPK modeling of a food-use chemical has shown the species specificity for liver and lung tumors in the mouse to be a direct consequence of the high activity and specific localization of a glutathione S-transferase theta enzyme in the mice. In the absence of such high or localized enzyme activity in other species, the metabolites of the glutathione S-transferase pathway are too unstable to interact with DNA. Thus, DNA interactions were not detectable in short-term mammalian mutagenicity assays in rats invivo or in hamster and human hepatocytes exposed to toxic dose levels of DCM in-vitro.

OCR for page 111
  Circumstances have not been identified under which the effects seen in mice could occur in other species, including humans. 1975 to 1997 DCM continues to be listed by FDA as a solvent for certain uses in food processing, including coffee decaffeination (CFR 173.255, 172.560, and 73.1) Comments The finding of tumors in the NCI inhalation studies, while not directly relevant to oral consumption, continued to cast a cloud over TCE and was responsible for much of the later research to remove this compound. While industries often work cooperatively on research to solve common problems, they typically are highly competitive in marketing. (See Lesson Learned No. 3.) In the late 1970s and early 1980s, FDA based its decision to continue the listings of TCE and DCM on the de minimis principle. The court rejected this principle, but only for food colors, in the case of Public Citizen v. Young, 831 F.2nd 1108 [D.C. Cir. 1987]. On the basis of that decision however, and the results of the NCI inhalation studies on DCM, FDA terminated cosmetic uses of DCM, but stated that it would take some time to assess the impact of that decision on the food uses, for which listing continues (FR 54, 124, 27328 ff.). Lessons Learned 1.   Pace-setting chronic toxicology is extremely expensive, twice that for following standard protocols. But it can lead to some success for the test material itself and help advance the science generally. Nevertheless, few industrial research funds are allocated purely for the purpose of advancing the sciences. 2.   Once a report of carcinogenic effect is attributed to a food-related chemical, the only economically reasonable thing an industry can justify doing is to stop using the material, unless defense of the material is the last resort for remaining in business. Having a safety cloud over one's product brands, over a 5 to 10-year span required to scientifically resolve a safety issue, is very costly. 3.   In the case of both TCE and DCM, the industries reacted far more precipitously than did FDA, partly out of justified fear of customer and consumer reaction, and partly, by finding a better alternative first, to acquire a competitive advantage. This contributed heavily to the rush to move from TCE to DCM, and from DCM to other solvents. 4.   The then existing policies and practices of the NTP bioassay program, and the resulting political and public relations climate, made it extremely difficult to obtain support for the practice of good in-depth science even after professional consensus (e.g., The report of the Task Force of Past Presidents of the SOT, 1982). Fortunately, support was ultimately found. 5.   Once FDA makes the decision to approve a substance, it feels that it has acquired some "ownership" that makes it difficult and embarrassing to change its mind. It understandably does not like to find that later data have cast doubt on an earlier approval. This tends to make FDA extremely cautious about new approvals, particularly if they involve highly novel substances or technologies. Conversely, it also, as in this case, can make the agency reluctant to be pushed into a premature decision to restrict or ban the substance.

OCR for page 111
d-Limonene Overview d-Limonene is a monoterpene hydrocarbon found in more than 30 species of fruits and vegetables, especially in citrus oils, and particularly in oil of orange. In addition to its natural occurrence, d-limonene is widely used as a flavor additive in foods and beverages including a variety of juices and non-alcoholic beverages, baked goods, gelatins, puddings, and chewing gums. Daily U.S. consumption is estimated to be approximately 0.27 mg/kg, but depending on citrus juice consumption, exposure can be as high as 1.2 mg/kg/day for young children. In addition to exposure to d-limonene in foods, this monoterpene is also widely used as a fragrance in perfumes and a variety of household products and as an industrial solvent and degreaser. Recently, its use as a cancer chemopreventive and chemotherapeutic agent has been reported. Chronology 1960 FDA recognized d-limonene to be GRAS as a synthetic flavoring substance. 1965 FEMA expert panel determined d-limonene to be GRAS. 1976 FEMA expert panel reviewed all available data on d-limonene and affirmed the previous GRAS status. 1987 Published data demonstrated that d-limonene caused a male rat specific nephrotoxicity mediated by the accumulation of α2u-globulin. 1989 Researchers found that renal toxicity seen with d-limonene treatment was caused by the binding of d-limonene epoxide metabolite to α2u-globulin. This prevented its degradation and caused it to accumulate in renal phagolysosomes only in male rats. 1990 NTP reported evidence of carcinogenic activity of d-limonene in male F344 rats, as indicated by increased incidence of renal tubular cell adenomas and adenocarcinomas. No evidence of carcinogenic activity in female F344 rats or in male or female B6C3F1 mice. There was no evidence of mutagenic activity of d-limonene. 1990 d-limonene given GRAS status by the expert panel of FEMA. They reviewed the data and were convinced that the evidence supported a unique response in the male rat that was not predictive of a similar risk for humans. 1991 EPA Risk Assessment Forum published criteria document for chemicals that caused male rat specific renal tubular tumors. EPA concluded that data would not be used if male rat tumors arose from process involving α2u-globulin. d-limonene was one of the standards from which these criteria were developed. 1992 International Agency for Research on Cancer (IARC) reviewed data on d-limonene and classified it as a Group 3 carcinogen (not classifiable as to its carcinogenicity in humans). 1993 JECFA reviewed d-limonene and decided that it did not represent any hazard to human health. A "not specified" ADI was published by this group. Lessons Learned 1.   There are chemicals that may be found to be carcinogenic in laboratory animals but for which the response is not relevant for humans. In the case of d-limonene, this qualitative difference is based solely on the presence of a unique protein, α2u-globulin. As such, the human risk assessment should be based on this qualitative difference, and no quantitative risk analysis can or should be developed.

OCR for page 111
2.   The mechanism developed to explain the carcinogenic activity of d-limonene has been substantiated across a broad class of compounds, and these data support further the unique male rat specific response associated with the presence of α2u-globulin. However, a significant research effort was invested to deduce the biochemical and cellular changes that linked the acute nephropathy to renal cancer and which established the lack of human relevance of these events. The level of detailed mechanistic information provided for d-limonene is likely to be the standard by which other efforts to deviate from default assumptions for human risk assessment will be judged. In this manner, it will likely take many years to convince regulatory bodies of a unique mechanism of action. Regulatory agencies were generally patient and restrained during the development of the essential data for d-limonene, and such judgment is needed in all cases. 3.   The major vehicle for human exposure to d-limonene is natural fruits. Further, the consumption of natural fruits that are high in d-limonene is associated with reduced rates of human cancer. Clearly, this knowledge had considerable impact on the patience and restrained action of the regulatory agencies. Benzyl Acetate Overview Benzyl acetate is used primarily as a component of perfumes for soaps and as a flavoring ingredient. As a flavoring ingredient, it is found in a variety of baked goods, soft and hard candy, various beverages, frozen dairy, and chewing gum products. Benzyl acetate is also a naturally occurring component of traditional foods, such as tomatoes, apples, mushrooms, and strawberries. The Consumption Ratio (CR), which compares the average intake of added flavoring materials to the quantities consumed as components of traditional foods, indicates that benzyl acetate is consumed about as much as a flavor ingredient as it is as a constituent of traditional foods. Its reported consumption volume of 22,500 kg in 1970 drew the interest of the NTP in testing and evaluating the toxicity of benzyl acetate. Chronology 1964 FDA approved benzyl acetate as a flavoring ingredient in foods. 1965 FEMA expert panel judged benzyl acetate to be GRAS. 1980 FEMA expert panel evaluated the available data and affirmed the previous GRAS status. Pre-1986 NTP conducted toxicity and carcinogenicity investigations of benzyl acetate (>99% pure). Positive trends for several types of neoplasms, none of which were statistically significant or dose-related, were noted. These positive trends included acinar-cell adenomas of the exocrine pancreas in F344/N male rats; neoplasms of the preputial gland; hepatocellular adenomas in mice of each sex; and squamous cell papillomas or carcinomas of the forestomach in male mice. No evidence of carcinogenicity was found for female F344/N rats. 1982 Reviewers of benzyl acetate NTP draft technical report were in agreement with the studies' conclusions, with only minor modifications. One reviewer believed that the known metabolites of benzyl acetate were non-mutagenic and likely not carcinogenic. 1983 Draft NTP technical report again reviewed with increased concern for the quantitative and qualitative limitations of the study. The use of corn oil was a

OCR for page 111
  confounding variable that left the causation of pancreatic acinar-cell adenomas in male F344/N rats unclear. Additionally, the non-random mortality of female mice due to intercurrent disease was excessive, prohibiting a proper analysis of the association between incidence of liver adenomas and benzyl acetate. Thirdly, the incidence of liver adenomas in the vehicle control group was unusually low, and the incidence of liver adenomas in the benzyl acetate group was comparable with those found in historical corn oil gavage control groups. 1984 Metabolism studies show ring-labeled 14C-benzyl acetate to be rapidly excreted in adult male F344 rats and B6C3F1 mice, with no detectable tissue retention, or diminution of clearance, even at high doses in two-year NTP studies. NTP short-term genotoxicity studies found no statistically significant mutagenic activity in test objects exposed to benzyl acetate. The audit confirmed the forestomach as a target organ for carcinogenicity in the mouse. Pre-1993 NTP further studied benzyl acetate using the dosed feed route. No evidence of carcinogenic activity was shown in either male or female F344/N rats, or in male or female B6C3F1 mice at highest doses given (600 mg/kg/day). NTP suspected rats may have tolerated higher doses. Comments Peer Review Panel objections to NTP's 2-year study protocol of benzyl acetate have emphasized two major study concerns: (1) that B6C3F1 mice are a poor model for human carcinogenicity, and (2) corn oil gavage may be a significant confounding factor in the study of carcinogenicity. The high tumor incidence observed in rodent controls of NTP carcinogenicity studies raised concern within the scientific community, and led to independent investigations. A study considering the use of historical control data in the evaluation of tumor incidences for carcinogenicity studies observed liver nodules, adenomas, or carcinomas in 31 percent of untreated control male B6C3F1 mice. Further studies have noted that hepatocellular hyperplasia may occur secondary to necrosis or a degenerative process in the liver, and that high incidence of hepatocellular neoplasms in control male B6C3F1 mice should elicit some concern. Inflammation, necrosis/ulceration, and hyperplasia of the forestomach squamous mucosa have been observed in many long-term studies utilizing corn oil gavage, making it difficult to interpret potential carcinogenic responses. Other concerns for the NTP benzyl acetate study have been raised. After careful review of the NTP draft report on its bioassay of benzyl acetate, the FEMA and Fragrance Manufacturers Association (FMA) expert panels noted several troubling issues. A high degree of infection noted in mice of both sexes; poor handling of the untreated control groups; a lack of significant tumor incidence between the benzyl acetate treated groups and historical controls; and poor statistical analysis led the FEMA and FMA expert panels to disagree with the NTP finding of carcinogenicity. It was attention to points raised by independent review committees and investigations that led the NTP expert panel to push for a second 2-year NTP benzyl acetate study using a microencapsulation feed protocol. Lessons Learned 1.   Failure to address the limitations of standard protocols, even if they are widely utilized, limits the significance of such studies, and can create undue public concern. NTP's continued use of corn oil gavage and B6C3F1 mice in its two-year rodent studies, without proper regard for

OCR for page 111
their potential as confounding factors, raises concerns about study results and the clarity of interpretation. Such factors, without appropriate evaluation, can impugn the safety of a substance that presents no significant hazard whatever under its conditions of use. 2.   The NTP MTD/1/2 MTD bioassay was never originally intended to be any more than a screen for serious toxicity, and was not intended for, nor should it serve as, an appropriate source of data for risk analysis. 3.   Poor data from a defective study are a waste of time and resources. 4.   The expert review committee members should not only be experts in carcinogenesis, but must also be able to properly interpret information for safety assessments. This point is well illustrated by the case history of benzyl acetate. Prior concerns for NTP's testing protocols provided the foundation for issues raised during the Peer Review Panel's critique of the NTP 1986 publication, and the subsequent retesting of benzyl acetate by NTP. The results from this 1993 publication proved to be consistent with the scientific literature as a whole, supporting the 1964 FDA and the FEMA GRAS approvals of benzyl acetate as a synthetic flavor ingredient. 5.   The FEMA expert panel's 37 years of activity demonstrate how to properly use independent expert opinion. The depth of experience provided by panel members in the most relevant fields, the appropriate balance of system and flexibility in procedures, and total independence in reaching conclusions permits quick decisions on simple, negligible-risk matters. 6.   The FDA wisely exercised restraint in not responding precipitously to the defective first study by NTP. ISO-Amyl Acetate Overview Iso-amyl acetate occurs in nature and has been reported in cider, rum, and malt whiskey. The Consumption Ratio (CR), which compares the average intake of added flavoring materials to the quantities consumed as components of traditional foods, indicates that iso-amyl acetate is consumed about equally from both the two categories. Studies show that iso-amyl acetate is rapidly absorbed and metabolized. Chronology 1965 FEMA expert panel determined iso-amyl acetate to be GRAS. 1975 FEMA expert panel affirmed the previous GRAS status for iso-amyl acetate. 1994 FEMA expert panel reaffirmed iso-amyl acetate as GRAS based upon its facile hydrolysis and oxidative detoxification of its alcohol and acetic acid, its very low level of flavor use, the safety factor calculated from results of subchronic studies for iso-amyl alcohol and iso-amyl iso-valerate, and its very low acute oral toxicity. 1997 Midwest Research Institute conducted an in vitro study using whole rat blood and showed that the half-life of iso-amyl acetate is very short in vivo; 4 minutes in blood and 2 minutes in plasma. Comments Iso-amyl acetate has not been the subject of a significant amount of scientific research or toxicological testing. It was therefore necessary to draw upon experienced professional judgment in determining the potential risk iso-amyl acetate poses to the population. In conducting a safety

OCR for page 111
assessment, the FEMA expert panel deemed iso-amyl acetate GRAS based upon studies addressing structurally similar compounds, its low level of flavor use, and its rapid metabolism and excretion. Lessons Learned 1.   The case of iso-amyl acetate illustrates how important it is to have a contextual understanding of related scientific literature in conducting an assessment of the safety-in-use of a substance as a flavor ingredient. 2.   The FEMA expert panel's 37 years of activity demonstrate how to properly use independent expert opinion. The depth of experience provided by Panel members in the most relevant fields, the appropriate balance of system and flexibility in procedures, and total independence in reaching conclusions permits quick decisions on simple, negligible-risk matters. More complex problems may require many months of scrutiny, and can be appropriately addressed by the expert panel because of its depth of experience and flexibility. There can be little doubt that this is a cost-effective, timely method of providing highly expert decisions on safety-in-use. Furfural Overview Virtually ubiquitous in nature, furfural is a naturally occurring component of many fruits and vegetables. The CR, which compares the average intake of added flavoring materials to the quantities consumed as unavoidable components of traditional foods, indicates that furfural is almost entirely found in traditional foods. In addition, the formation of furfural during the thermal decomposition of carbohydrates, makes it a component of many processed food, products such as baked goods, meat products, and alcoholic and non-alcoholic beverages. NTP included furfural in its testing program based upon its widespread natural occurrence in food and historical data indicating exposure to high concentrations may be hepatoxic. Chronology 1960 FEMA expert panel judged furfural to be GRAS. 1975 FEMA expert panel affirmed the previous GRAS status of furfural for use as a flavor ingredient. Pre-1990 NTP conducted toxicity and carcinogenicity investigations of furfural (>99% pure). There was evidence of carcinogenic activity for male F344/N rats based upon the occurrence of uncommon cholangiocarcinomas and bile duct dysplasia. Additionally, there appeared some evidence of carcinogenic activity in male and female B6C3F1 mice based on increased incidences of heptocellular adenomas. No evidence of carcinogenicity was identified in female F344/N rats at the highest dose level. 1989 Reviewers of furfural NTP draft technical report found general agreement on the findings of the study with respect to male rats, and male and female mice. 1993 JECFA determined furfural was not an appropriate flavor addition, and could not be allocated an ADI. The NTP studies, taken with the perceived ''relatively high"

OCR for page 111
  concentrations of furfural found in some foods, caused JECFA to "consider the direct addition of furfural as a flavor to be inappropriate." 1995 The International Agency for Research on Cancer (IARC) concluded that there was inadequate evidence in humans and limited evidence in animals for the carcinogenicity of furfural. IARC found furfural to be a component of 150 foods, including a wide range of fruits and vegetables. 1996 FEMA expert panel conducted a comprehensive review of the scientific literature and determined that furfural be reaffirmed as GRAS based upon its rapid absorption and metabolism in in vivo systems, its ubiquity in nature, a lack of evidence indicating any risk to human health under conditions of use as a flavor ingredient, and a lack of evidence implicating furfural as a carcinogen. Its only significant finding for carcinogenicity was two-year NTP bioassays, showing increased incidence of hepatocellular adenomas and carcinomas in the high-dose group of male mice. Review of these findings determined that the observed carcinogenicity was secondary to pronounced hepatotoxicity. Comments Despite the pervasive presence of furfural at low concentrations in the native food supply, concerns have been raised about its toxicological characteristics, particularly with respect to carcinogenicity. Low molecular weight aldehydes like furfural may undergo oxidation or condensation reactions associated with the aldehyde function either in digestive fluids prior to absorption, or in body fluids prior to entering the cell. Reactivity of these aldehyde groups has been demonstrated to produce toxic effects including the induction of tumors when administered under non-physiological conditions at high dose levels. Lessons Learned 1.   The testing and review history of furfural is a strong example of how results from chronic, high-dose testing can be given inappropriate weight. The 1990 NTP study publication on furfural noted some evidence of carcinogenic activity, which proved to be a strong factor in JECFA's failure to allocate furfural an ADI. In contrast, IARC and FEMA conducted critical reviews of the literature, concluding furfural to be of no carcinogenic risk to humans. The FEMA expert panel's GRAS conclusion, and subsequent affirmation and reaffirmation were based upon a contextual understanding of the wide range of studies conducted with furfural. Institutions that conduct primary research with regulatory implications must hold themselves to a high protocol standard, just as regulatory agencies that utilize these studies must be able to place referred studies into context with the questions of human risk. Therefore, regulatory bodies must critically evaluate the relevance of high-dose chronic studies to the safety evaluation of substances ubiquitous in the food supply at far lower levels of exposure. 2.   The FEMA expert panel's 37 years of activity demonstrate how to properly use independent expert opinion. The depth of experience provided by panel members in the most relevant fields, the appropriate balance of system and flexibility in procedures, and total independence in reaching conclusions permits quick decisions on simple, negligible-risk matters. More complex problems, such as those posed by furfural, require many months of scrutiny, and can be appropriately addressed by the expert panel because of its depth of experience and flexibility. There can be little doubt that this is a cost-effective, timely method of providing highly expert decisions on safety-in-use.

OCR for page 111
Pulsed Light Overview FDA was first contacted about the possible use of pulsed light as a surface antimicrobial treatment for food in the spring of 1993. PurePulse (then Foodco Corp) had developed a technology that used short, high-intensity, broad spectral bandwidth pulses from specialized flashlamps to destroy a variety of microorganisms (bacteria, yeast, molds, etc.) commonly found on the surfaces of food. Although the flashlamp technology had been used in other applications, it had not previously been used for food treatment. The company sought FDA guidance on the regulatory steps that would be necessary before the novel technology could be commercialized. Since the law categorizes sources of radiation used to treat food as a ''food additive" unless they are GRAS, submission of a food additive petition, and its subsequent approval by FDA, were necessary. Because a source of radiation is not actually added to food, nor does it come in contact with food, a petition for pulsed light would need to contain different information than a petition for a more typical food additive (direct or indirect). FDA worked with the petitioner to identify the type and amount of data and supporting information that would be submitted. Because of the potential public health impact of the technology, FDA further committed to an up-front "review" of sections of the petition, while still in draft, by the Consumer Safety Officer responsible for coordinating the technical review of the petition. Sections of the petition that provided the basis for critical quantitative arguments received detailed review, as did the environmental assessment. Chronology 1994 In March, pulsed light petition filed by PurePulse. 1994 In May, FDA review of the photochemical data and extensive quantitative arguments completed. 1994 FDA initial scientific review of the petition completed in August. 1995 In January, FDA questions in the area of microbiology, specifically, the area of competitive microbial populations, sent to petitioner. 1995 In March, petitioner sent FDA information needed to fill in the gaps in the petition. 1995 FDA completed review of March information in June. 1995 In June, Federal Register document that would grant approval was drafted and received scientific clearance as well as clearance by OPA management. 1996 Higher level clearance of Federal Register document, including review and clearance by the Office of General Counsel, complete in May. 1996 In August, final rule published. Lessons Learned 1.   Early consultation and communication between FDA and the petitioner is necessary on a novel food technology. 2.   FDA and the petitioner worked to reach a mutual understanding of the types and amount of data and information that would be needed to establish safety. The goal was to ensure a "good" petition. 3.   FDA committed priority resources to a "good" petition that had potential for positive public heath impact: an investment of agency resources in a limited review of draft petition materials, and a focused "team approach" in review once the complete petition was filed.

OCR for page 111
4.   Constraints included many urgent competing demands not within the agency's control. These involved serious outbreaks of food-borne illness that required immediate response from the agency and constituted a serious drain on personnel resources needed to complete review of the petition. In addition, many urgent competing demands on FDA and other agency staff and officials, including General Counsel, resulted from increased congressional interest in the food additive premarket approval program and in other premarket approval programs throughout the agency. Chymosin Overview In 1988 and 1989, petitions were filed independently by three separate companies seeking FDA's affirmation of the GRAS status for chymosin, the active enzyme component of rennet, used in cheese production. Because these chymosin preparations were produced through modern biotechnological means, all three petitioners awaited final FDA affirmation before marketing their products because of the potential for consumer rejection in the marketplace. FDA affirmation of GRAS status was critical for public acceptance of these newly developed food ingredients. The information provided herein is based on the record and on personal communications with FDA personnel who were involved in the review of the petition filed by Pfizer Inc., which was submitted in 1988 and affirmed in 1990 (a total of 25 months). The two other companies filed their GRAS petitions in May 1989 and October 1989, and were affirmed in 33 and 43 months, respectively. Chronology 1987 Pfizer had informal consultations with FDA to identify and elaborate issues of concern. FDA realized that although biotechnology presented new questions related to safety, those questions could be answered with the available scientific procedures, including those provided by biotechnological procedures. The inherent safety of the final chymosin product was confirmed from feeding studies in rats and dogs, and from a genotoxicity test battery. 1988 Pfizer submitted both a Food Additive Petition (FAP) and GRAS affirmation petition for chymosin. The FDA wanted to affirm the GRAS status of chymosin because a) traditional rennet was already GRAS affirmed, and b) it desired a review process open to public scrutiny during the petition review process. Pfizer submitted an FAP to ensure that the FDA would abide by the 180-day statutory deadline for FAP approval. 1989 FDA scientific review completed within 15 months of filing (circa July 1989). 1990 FDA affirmed chymosin as GRAS, 25 months after submission of the petition. Comment Although the scientific aspects of the review had been comprehensively elaborated, some effort was needed to establish a record that supported general recognition and to educate FDA management regarding the biotechnological aspects of the petition. It is important to keep in mind that the FDA had to answer policy questions subsequent to the scientific review period, yet

OCR for page 111
most policy questions were dealt with rather expediently (i.e., within a few months) because the scientific review conducted was definitive and conclusive to affirm GRAS status. Lessons Learned 1.   Good data and joint agency/petitioner planning permit reasonably prompt and effective decisions, especially on noncontroversial topics. Before the petition was submitted, the FDA met with the petitioner on an informal basis to identify issues and to develop a framework to answer questions to determine the safety of the product.