Chapter 31: Food Toxicology

The typical Western diet contains hundreds of thousands of substances naturally present in food and many more, which are formed in situ when food is cooked or processed. Many of these substances affect the nutritional and esthetic qualities of food including appearance and organoleptic properties (ie, flavor, texture, or aroma) that determine whether or not we will even try the food or take a second bite. Whereas substances present in food may be nutritional and/or gratifying, they may not necessarily be “safe” in any amount or for any intended use. The Federal Food, Drug and Cosmetic (FD&C) Act gives the Federal government the authority to ensure that all foods involved in interstate commerce are safe. Congress, in writing the Act (and its subsequent amendments), understood that safety cannot be proved absolutely and indicated instead that the safety standard for substances added to food can be no more than a reasonable certainty of no harm. As will be pointed out in other sections of this chapter, the language of the FD&C Act effectively provides for practical and workable approaches to the assessment of safety for food, food ingredients, and food contaminants. Because food is highly complex, the legal framework provided by Congress for the regulation of food and substances in food was kept simple so that it would work.

The basic element of the framework is that food, which is defined as articles or components of articles used for food or drink for humans or animals, bears the presumption of safety (Sections 201[f] and 402[a][1] of the FD&C Act). This means that a steak or a potato is presumed to be safe unless it contains a poisonous or deleterious substance in an amount, which is shown to make it ordinarily injurious to health. In essence, this presumption of safety was born of necessity. If the hundreds of thousands of substances naturally present in food were subject to the same strictures and limitations that apply to added substances, virtually all food would be suspect and food shortages could easily result. To avoid such crises, Congress developed a safety standard that would not force regulatory authorities to ban common, traditional foods. In cases in which the substance is not naturally present in food but is a contaminant or added ingredient, the safety standard is quite different. This standard decrees a food to be adulterated if it contains any poisonous or deleterious substance that may render it injurious.

Thus, for additives and contaminants, Congress recognized that these substances are not as complex as food and should therefore meet a higher standard of safety. However, because neither the law nor Food and Drug Administration (FDA) or US Department of Agriculture (USDA) regulations explicitly define the term “safety” for substances added to food, scientists and their legal and regulatory counterparts have worked out operational definitions for the safety of such substances.

As with food, a practical and workable approach must be found for the contaminants of added ingredients, because all substances contain a myriad of contaminants at trace or even undetectable amounts with current technology. In this case, the approach involves setting specification limits on contaminants that are intended to exclude the possibility that the level present in an additive may render the food to which the substance is added, unsafe. It should be emphasized that specifications can serve their purpose of assuring suitable purity only if the manufacturing processes used are adequately controlled to assure consistency in the quality and purity of the product. The philosophy by which specifications are established for substances added to food embodies the belief that not all risks are worthy of regulatory concern and control (ie, the concept of de minimis).1 Implicit in this philosophy is the important unifying concept of threshold of regulation in food safety assessment (Flamm et al., 1994, 2002).

Food, as stated earlier, contains hundreds of thousands of substances, most of which have not been fully characterized or tested. The presumption that a food is safe is based on a history of common use and that the consumption of certain foods is deeply rooted in tradition. When the uncertainty about the risk of the added substance is small compared with the uncertainties attending food itself, the standard of “reasonable certainty of no harm” for the added substance has been satisfied. Thus, for food-like substances, the presumption is that the substance resembles food, is digested and metabolized as food, and consequently raises fewer toxicological and safety-related questions than do non-food–like substances. Moreover, when non-food–like substances are added in only very small or trace amounts, the low levels of exposure aid in demonstrating that the intended conditions of use of these substances are safe. These broad generalizations, however, do not suffice to exempt these food ingredients from the requirements of thorough safety evaluation.

Over the past two decades, there has been increasing interest on the part of consumers regarding the health-enhancing properties of foods and the components they contain. Substances such as phytosterols from vegetable oils and isoflavones from soy have been isolated and added to other foods at elevated levels to impart cholesterol-lowering properties. Such products have raised regulatory questions about whether these substances are functioning as drugs, and should they be regulated as such, or whether they should be viewed as new nutrients and allowed in foods, as are the historically recognized nutrients, vitamin C and iron. Some experts in nutritional science have concluded that the concept of nutrients should be expanded to include a growing number of desirable food constituents that produce quantifiable health benefits related to disease prevention (Sansalone, 1999). This isolation of, and fortification with, new food components will necessitate a thorough evaluation of safety at the intended level of intake and for the population at large (Mackey and Kotsonis, 2002).

Lastly, in many parts of the world, food safety (ie, food free of microbiological, chemical, and radionuclide contamination) and food security (ie, availability of food) represent by far the major food concerns faced by consumers. Thus, although our vigilance in assuring the safety of substances added to food (often with safety factors of 100 to a million) is not trivial in the context of our culture, we should not lose sight of other parts of the world where availability of potable water and sustaining caloric intake, represent daily challenges.

Uniqueness of Food Toxicology

The nature of food is responsible for the uniqueness of food toxicology. Food occupies a position of central importance in virtually all cultures and, although good agricultural practice has improved the quality of unprocessed food, most food cannot be commercially produced in a definable environment under strict quality controls. Food generally cannot meet the rigorous standards of chemical identity, purity, and good manufacturing practice met by most consumer products. The fact that food is harvested from the soil, the sea, inland waters, or is derived from land animals, which are subject to the unpredictable forces of nature, makes the constancy of raw (unprocessed) food unreliable. Therefore, it is clear that meat, milk, eggs and produce are held to a different standard than the ingredients of processed food, as a practical matter dictated by necessity.

Food also acquires uniqueness from its essential nutrients, which, such as Vitamin A, may be toxic at levels only 10-fold above those required to prevent deficiencies. The evaluation of food ingredients often must rely on reasoning unique to food science in the sense that such substances may be normal constituents of food or modified constituents of food as opposed to the types of substances ordinarily addressed in the fields of occupational, environmental, and medical toxicology. Assessing the safety of such substances, which are added to food for their technical effects, often focuses on digestion and metabolism occurring in the gastrointestinal (GI) tract. The reason for this focus is that in many cases, an ingested substance is not absorbed through the GI tract; only products of its digestion are absorbed, and these products may be identical to those derived from natural food.

Nature and Complexity of Food

Food is an exceedingly complex mixture of nutrient and non-nutrient substances, whether it is consumed as a raw agricultural product or as highly processed ready-to-eat foods such as a “Meal Ready to Eat” military ration (Table 31-1). The Western diet consists of items of caloric and noncaloric value, that is, carbohydrates supply 47% of caloric intake, fats supply 37%, and protein supplies 16% (all three of which would be considered “macronutrients”), whereas minerals and vitamins, the “micronutrients,” have no caloric value but are no less essential for life.

Nonnutrient substances are often characterized in the popular literature as being contributed by food processing, but nature provides the vast majority of nonnutrient constituents. For instance, even among “natural” (or minimally processed) foods, there are far more nonnutrient than nutrient constituents (Table 31-2). Many of these nonnutrient substances are vital for the growth and survival of the plant, including hormones and naturally occurring pesticides, estimated at approximately 10,000 by Gold et al. (1992). Some of these substances may be antinutrients such as lectins, saponins, trypsin, and/or chymotrypsin inhibitors in soybeans, phytates that may bind minerals (also present in soybeans) and antithiamines (in fish and plants), and there may also be frankly toxic constituents such as tomatine or cycasin. An idea of the large number of substances present in food is given in the database VCF Volatile Compounds in Food 12.3 (latest version) (VCF, 2010), in which approximately 7800 volatile substances are noted as occurring in one or more of the 485 different foods. However, this is only the tip of the iceberg, as the number of nonvolatile and/or yet unidentified natural chemicals in food vastly exceeds the number that has been identified to date.

Nonnutrient substances are also added to food as a result of processing, and in fact, 21 CFR 170.3(o) lists 32 categories of direct additives, of which, there are about 3000 individual substances. Approximately 2000 of these added ingredients are flavor ingredients, most of which already occur naturally in food and are non-nutritive (Burdock, 2002). Of the 2000 flavoring ingredients that may be added to food, approximately one-third are used at concentrations below 10 ppm (Hall and Oser, 1968), about the same concentration as is found naturally.

Importance of the Gastrointestinal Tract

FDA considers that humans consume approximately 1500 g of food and an equal amount of liquid per day; therefore, within a year, humans ingest on the average 548 kg of food (or about 1200 lbs). To respond to this load, it is easy to appreciate the fact that the gut is a large, complex, and dynamic organ with several layers of organization and a vast absorptive surface that has been estimated to be 250 m² or more (Hall, 2011). The GI transit time provides for adequate exposure of ingesta to a variety of processing conditions including, but not limited to, variable pH, digestive acids and enzymes (trypsin, chymotrypsin, etc, from the pancreas and carbohydrases, lipases, and proteases from the enterocytes), saponification agents (in bile), and a luxuriant bacterial flora (estimated to be 100 trillion organisms in adults), the latter of which provide a repertoire of metabolic capability not shared by the host (eg, fermentation of “nondigestible” sugars such as xylitol and sorbitol) (Drasar and Hill, 1974; Vickery et al., 2011). In addition, the enterocytes (intestinal epithelium) possess an extensive capacity for the metabolism of xenobiotics that may be second only to the liver, with a full complement of phase (type) I and phase (type) II reactions present. This enteric monooxygenase system is analogous to the liver, as both systems are located in the endoplasmic reticulum of cells, require NADPH and O₂ for maximum activity, are inhibited by many of the same substances, and are qualitatively similar in their response to enzyme induction (Hassing et al., 1989).

The constituents of food and other ingesta (eg, drugs, contaminants, inhaled pollutants dissolved in saliva and swallowed) are a physicochemically heterogeneous lot and, as a result, the intestine has evolved into a relatively impermeable membrane with specialized mechanisms of absorption that have developed, which allow only certain substances to gain access to the body from the intestinal lumen. The four primary mechanisms for absorption are passive (or simple) diffusion, facilitated diffusion/cotransport, active transport, and pinocytosis. With some exceptions, each of these mechanisms characteristically transfers a defined group of constituents from the lumen into the body, although some substances may use more than one mechanism, such as fructose (Table 31-3). As is noted in the table, xenobiotics and other substances may compete for passage into the body.

Aiding this absorption is the rich vascularization of the intestine, with a normal rate of blood flow in the portal vein of approximately 1.2 L/h/kg; however, after a meal, there is a 30% increase in blood flow through the splanchnic area (Concon, 1988). It follows then, that substances which affect blood flow, also tend to affect the rate of absorption of other materials; an example is alcohol, which tends to increase blood flow to the stomach and thus enhances its own absorption as well as other substances. Few stimuli tend to decrease flow to this area, with the possible exception of energetic muscular activity and hypovolemic shock.

Lymph circulation is important in the transfer of fats, large molecules (such as botulinum toxin), benzo[a]pyrene, 3-methylcholanthrene, and cis-dimethylaminostilbene (Chhabra and Eastin, 1984). Small- to medium-chain fatty acids are absorbed directly and do not require conversion to monoglycerides and do not rely on the triglyceride → micelle → chylomicron pathway (Hall, 2011). Lymph has a flow rate of about 1 to 2 mL/h/kg in humans, and a few factors are known to influence its flow including tripalmitin, which has been shown to double the flow and therefore double the absorption of materials such as p-aminosalicylic acid and tetracycline (Chhabra and Eastin, 1984). Another factor that lends importance to lymph is the fact that the lymph empties via the thoracic duct into the point of junction of the left internal jugular and subclavian veins, preventing “first-pass” metabolism by the liver, unlike substances transported by the blood.

Many food ingredients are modified proteins, carbohydrates, fats, or components of such substances. Thus, understanding the changes these substances undergo in the GI tract, their possible effect on the GI tract, and whether they are absorbed or affect the absorption of other substances is critical to an understanding of food toxicology and safety assessment. Some of the factors that may affect GI absorption and the rate of absorption are listed in Table 31-4.

Apart from its duties of absorption and metabolism, the GI tract is also the largest immunologic organ in the body and is constantly exposed to a large number of antigens in food (approximately 88 kg of protein annually) and commensal and ingested bacteria. One cell layer away from these antigens is the lamina propria of the GI tract, which contains the mucosal-associated lymphoid tissue, comprised of lymphocytes and antigen-presenting cells, as well as unique dendritic cells, which interact with dietary antigens and ultimately determine whether an antigen is tolerated or an immune response is launched (Brandtzaeg, 2010; Vickery et al., 2011).

The FD&C Act Provides for a Practicable Approach

As incredible as it may seem, there were few specifics regulating food until the Food Additives Amendment of 1958; the original Pure Food and Drug Act of 1906 only stipulated that food should not be “filthy, putrid, or unfit for consumption.” In the meantime, many other facets of commerce had seen significant regulatory oversight initiated: animal feed (~1902), the Meat Inspection Act (1906) and, drugs and cosmetics (1938). The 1938 amendment superseded the 1906 Act and initiated the important concept of requiring premarket approval of a product, but only for drugs, not food ingredients. In practice, the 1938 amendment was found to have less benefit than expected for consumers or manufacturers and created a financial burden on FDA. That is, on one hand, the manufacturer was largely unrestricted on the addition of substances to food. On the other hand, the burden of proof was on FDA to show the lack of safety of an ingredient (requiring resource demands for testing) and once FDA declared an ingredient injurious to health, the manufacturer was forced to prove that it was “harmless per se” (ie, harmless at all doses) or forego the use of the substance in any amount. The “harmless per se” concept ignores the basic rule of toxicology that the “dose makes the poison.”

By the early 1950s, it was obvious that the law in place was unworkable and a mechanism for decision making was needed to determine the safe use of foods, the hundreds of ingredients already in use in food, and finished (processed) foods derived from the ingredients. Congress understood that a requirement for premarket (regulatory) approval of all of the ingredients would create an unimaginable burden in terms of costs of testing and time to review each food additive petition and such a process made no sense, knowing that many of these substances, such as spices, had been in use for over a thousand years. At the same time, it was untenable to simply “grandfather” ingredients already in use because for some, there were no safety data or history of use upon which a decision of safety could be based. In the end, Congress decided that all ingredients added to food were food additives (which would be approved by FDA), unless they were generally recognized as safe (GRAS) (a decision left to experts inside or outside of FDA and not requiring FDA approval); this act had the effect of avoiding a log jam of food additive petitions that would likely not be resolved to this day. Congress discarded the safety per se concept and developed the Act, which, by itself and subsequent regulations, mitigated potential harm to the consumer through the creative use of specifications, process and manufacturing controls (Good Manufacturing Practice), action levels, tolerances, warning labels, and outright prohibitions.

The Application of Experience: Generally Recognized as Safe

The FD&C Act permits the addition of substances to food to accomplish a specific technical effect if use of the substance is determined to be GRAS. The Act does not require the FDA to make this determination, although it does not exclude the agency from making such decisions. The Act instead requires that scientific experts base a GRAS determination on the adequacy of safety, as shown through scientific procedures or through experience based on common use in food before January 1, 1958, under the intended conditions of use of the substance (FD&C Act, section 201[s]).

In addition to allowing GRAS substances to be added to food, the Act provides for a class of substances that are regulated food additives, which are defined as “any substances the intended use of which results in its becoming a component… of any food… if such substance is not generally recognized… to be safe.” Hence, a legal distinction is drawn between regulated food additives and GRAS substances. Regulated food additives must be approved and regulated for their intended conditions of use by the FDA under 21 CFR 172–179 before they can be marketed. Under section 409 of the Act, the requirements for data to support the safe use of a food additive are described in general terms. The requirements or recommended methods for establishing safe conditions of use for an additive are available in the form of guidelines issued by the FDA. These guidelines, referred to as the Redbook, provide substance and definition to the safety standard applicable to regulated food additives: “reasonable certainty of no harm under conditions of intended use” (Burdock and Carabin, 2004).

Use of Tolerances

If a food contains an unavoidable contaminant even with the use of current good manufacturing practice (cGMP), it may be declared unfit as food if the contaminant may render the food injurious to health. Thus, for a food itself to be declared unfit, it must be ordinarily injurious, while an unavoidable contaminant in food need only pose the risk of harm for the food to be found unfit, and subject to FDA action. The reason for the dichotomy is practicality. Congress recognized that if authority were granted to ban traditional foods for reasons that go beyond clear evidence of harm to health, the agency would be subject to pressure to ban certain foods.

Foods containing unavoidable contaminants are not automatically banned because such foods are subject to the provisions of section 406 of the FD&C Act, which indicates that the quantity of unavoidable contaminants in food may be limited by regulation for the protection of public health and that any quantity of a contaminant, which exceeds the fixed limit, shall be deemed unsafe. This authority has been used by the FDA to set limits on the quantity of unavoidable contaminants in food by regulation (tolerances) or by informal action levels, which do not have the force of law. Such action levels have been set for aflatoxins, fumonisins, and patulin (Table 31-5). Action levels have the advantage of offering greater flexibility than is provided by tolerances. Whether tolerances or action levels are applied to unavoidable contaminants of food, the FDA attempts to balance the health risk posed by unavoidable contaminants against the loss of a portion of the food supply. In contrast, contaminants in food that are avoidable by cGMP are deemed to be unsafe under section 406 if they are considered poisonous or deleterious. Under such circumstances, the food is typically declared adulterated and unfit for human consumption. The extent to which consumers who are already in possession of such food must be alerted depends on the health risk posed by the contaminated food. If there is a reasonable probability that the use of or exposure to such a food will cause serious adverse health consequences or death, the FDA will seek a Class I recall, which provides the maximum public warning, the greatest depth of recall, and the most follow-up. Classes II and III represent progressively less health risk and require less public warning, less depth of recall, and less follow-up (21 CFR 7.3).

Pesticide Residues

A pesticide is defined under the Federal Insecticide, Fungicide, and Rodenticide Act (FIFRA), as any substance used to control or mitigate pests (such as insects, rodents, weeds, or fungi), or intended for use as a plant growth regulator, defoliant, or desiccant. In the United States, the regulation of pesticides is the responsibility of the Environmental Protection Agency (EPA) and is accomplished under both FIFRA and the FD&C Act. FIFRA governs the registration, sale, and use of all pesticides. It is illegal to use a pesticide unless it is specifically registered and labeled for the intended use. In order to obtain registration, an applicant must supply EPA with data on pesticide composition, mammalian and ecological toxicity, environmental fate, and potential human and environmental exposures.

A major part of the registration process for most pesticides involves the establishment of tolerances. The EPA must establish tolerances, or exemptions from tolerances, for all pesticides that may come into contact with food or feed, that is, those intended for use during the production, storage, transportation, or processing of food or feed crops, on livestock, or in food-handling establishments. The tolerances are intended to represent the highest expected residue levels from legal uses of the pesticide. All pesticide tolerances are now established under Section 408 of the FD&C Act. Prior to 1996, if pesticide residues in processed foods exceeded those in raw agricultural commodities, they were considered to be intentional food additives and were required to be assigned “Food Additive Tolerances” under Section 409 of the Act. If the pesticide chemical in question had been classified by the EPA as a human or animal carcinogen, the Delaney clause would be invoked and the Section 409 food additive tolerance(s) could be denied on that basis. However, the Food Quality Protection Act (FQPA) of 1996 revised Section 201(s) of the FD&C Act to exclude pesticides from the definition of food additive—even in the case of concentrated residues in processed fractions. Although an additional tolerance for a processed fraction is still required if the pesticide residue in the fraction exceeds the tolerance for the raw agricultural commodity, that tolerance is now established under Section 408 rather than Section 409. Consequently, although the Delaney clause has not been repealed from Section 409, and continues to apply to intentional food additives other than pesticides, it is no longer applicable for pesticides.

Prior to 1996, the EPA used 100× as a default safety factor when conducting risk assessments to ensure that the necessary food and feed tolerances would be safe for human health. However, FQPA also requires that an additional 10× safety factor “shall be applied for infants and children to take into account potential pre- and postnatal toxicity and completeness of the data with respect to exposure and toxicity to infants and children.” It further states, however, that “the Administrator may use a different margin of safety for the pesticide chemical residue only if, on the basis of reliable data, such margin will be safe for infants and children.”

Drugs Used in Food-producing Animals

The Federal Food, Drug, and Cosmetic Act (FFDCA) (section 201[g]) defines the term “drugs” as “articles intended for use in the diagnosis, cure, mitigation, treatment, or prevention of disease in man or other animals” and “articles (other than food) intended to affect the structure or any function of the body of man or other animals.” An animal drug “means any drug intended for use for animals other than man” [section 201(w) of the FD&C Act]. The Center for Veterinary Medicine, the arm of the FDA responsible for regulation of animal feed and drugs, considers that any substance added to animal feed used for growth promotion and increased food production is a drug. Determination of the potential human health hazards associated with animal drug residues is complicated by the metabolism of an animal drug, which results in residues of many potential metabolites.

The primary factors that must be considered in the evaluation of animal drugs are (1) consumption and absorption by the target animal, (2) metabolism of the drug by the target food animal, (3) excretion and tissue distribution of the drug and its metabolites in food animal products and tissues, (4) consumption of food animal products and tissues by humans, (5) potential absorption of the drug and its metabolites by humans, (6) potential metabolism of the drug and its metabolites by humans, and (7) potential excretion and tissue distribution in humans of the drug, its metabolites, and the secondary human metabolites derived from the drug and its metabolites. Thus, the pharmacokinetic and biotransformation characteristics of both the animal and the human must be considered in an assessment of the potential human health hazard of an animal drug.

For new animal drugs (ie, those not declared GRAS and effective), safety assessment is concerned primarily with residues that occur in animal food products (milk, cheese, etc) and edible tissues (muscle, liver, etc). Toxicity studies in the target species (chicken, cow, pig, etc) should provide data on metabolism and the nature of metabolites, along with data on the drug’s pharmacokinetics. During this phase, the parent drug and its metabolites are evaluated both qualitatively and quantitatively in the animal products of concern (eggs, milk, meat, etc). This may involve the development of sophisticated analytic methodologies. Once these data are obtained, it is necessary to undertake an assessment to determine potential human exposure to these compounds from the diet and other sources, pursuant to the establishment of a tolerance.

To comply with the Congressional intent regarding the use of animal drugs in food-producing animals as required in the no residue provision of the Delaney clause, the FDA began to build a system for conducting risk assessment of carcinogens in the early 1970s (FDA, 1977). In the course of developing a policy and/or regulatory definition for “no residue,” the FDA was compelled to address the issue of residues of metabolites of animal drugs known to induce cancer in humans or animals. As the number of metabolites may range into the hundreds, it became apparent that as a practical matter, not every metabolite could be tested with the same thoroughness as the parent animal drug. This forced the FDA to consider threshold assessment for the first time. Threshold assessment combines information on the structure and in vitro biological activity of a metabolite for the purpose of determining whether carcinogenicity testing is necessary (Flamm et al., 1994, 2002). If testing is necessary and if the substance is found to induce cancer, the FDA’s definition states that a lifetime risk of one in a million is equivalent to the meaning of “no residue” as intended by Congress.

Tolerance Setting

Foods are regarded as such because they are edible—they cannot be unpalatable or toxic—and foods must have nutritional, hedonic, or satietal value, otherwise there would be no point in consuming them. Therefore, FD&C Act presumes that traditionally consumed foods (produce such as apples or carrots and animal products, including meat, milk, and eggs) are safe if they are free of contaminants. The fact that foods contain many natural substances, some of which are toxic at a high concentration, is in itself an insufficient basis under the Act for declaring a food as being unfit for human consumption. In practice, a food containing a poisonous or deleterious substance may be considered adulterated (unfit for consumption), the exception being that if the substance is naturally present as a constituent of that food in a quantity that is not ordinarily injurious (§409 of the Act). The implementation of the use of this concept (a) allows the presence of solanaceous glycoalkaloids (SGAs) in potatoes, as the SGAs occur naturally in potatoes, (b) but only allows the SGAs to be present at a nontoxic level (defined as “not ordinarily injurious”), and (c) effectively prohibits the addition (or presence) of SGAs in foods where SGAs would not appear naturally (such as in wheat products, where the SGAs could only be present as a contaminant). A tolerance has considerable reach—for example, a potato might have an acceptable level of SGA at the point of harvesting, but if subject to mishandling during shipping, handling, or storage, the potato may generate additional SGA to a point in excess of the prescribed tolerance.

Food and Color Additives

Despite the initial successes with GRAS status following passage of the 1958 amendment, colors were designated additives in the Color Additives Amendment of 1960 and thereby not eligible for GRAS status, thus requiring a Color Additive Petition. As per the FD&C Act (§201[t]), “a color is a dye, pigment, or other substance… when added or applied to a food, drug, or cosmetic, or to the human body or any part thereof, is capable (alone or through reaction with other substance) of imparting color thereto.…” Blacks, whites, and intermediate grays also are included in this definition. A provision in the Act and in the regulation (21 CFR §70.3[f]) exempts the use of a substance, which although imparting color to the final product, it is not a color per se. For example, food ingredients such as cherries, green or red peppers, chocolate, and orange juice, which contribute their own natural color when mixed with other foods, are not regarded as color additives (such as chocolate to produce chocolate milk); but where a food substance such as beet juice is deliberately used as a color, as in pink lemonade, it is a color additive. Although a color additive has only one function, a food additive may have any one of 32 functionalities (Table 31-6).

There are two distinct types of color additives that have been approved for food use: those requiring certification by FDA chemists and those exempt from certification. Certification, which is based on chemical analysis, is required for each batch of most organic synthesized colors because they may contain impurities that may vary from batch to batch. Most certified colors approved for food use bear the prefix FD&C. Certification involves in-depth chemical analysis of major and trace components of each individual batch of color additives by FDA chemists and is required before any batch can be released for commercial use. Such color additives consist of aromatic amines or aromatic azo structures (FD&C Blue No. 1, Blue No. 2, Green No. 3, Red No. 3, Red No. 40, Yellow No. 5, and Yellow No. 6) that cannot be synthesized without a variety of impurities. Orange B and Citrus Red No. 2 are the only certified food colors that lack the FD&C designation (21 CFR 74 Subpart A). Two additional color designations reference the range of approved use: FD&C colors, which indicate use in foods, drugs, and cosmetics, and D&C colors, which indicate their use solely in drugs and cosmetics, not foods.

The estimated daily intakes (EDI) of certified color additives for the US population and the acceptable daily intakes (ADIs) determined from toxicology studies are in Table 31-7. The calculated values of the EDIs assume an even distribution of the respective additives for the entire population. A comparison of these levels indicates that the EDIs are well below the ADIs in both adults and children (FDA, 2011a, 2011b).

Despite the fact that aromatic amines are generally considered relatively toxic substances, the FD&C colors are notably nontoxic. The principal reason involves sulfonation of the aromatic amine or azo compound that constitutes a color additive. Such sulfonic acid groups are highly polar, which, combined with their high molecular weight, prevents them from being absorbed by the GI tract or entering cells. All the FD&C food colors have been extensively tested in all Concern Level (CL) III tests (Table 31-8) and have been found to be “remarkably” nontoxic (refer to section titled. Assignment of Concern Level and Required Testing).

Food colors that are exempt from certification typically have not been subjected to such extensive testing requirements. The exempt food colors are derived primarily from natural sources. Whereas synthetic food colors have received the majority of public, scientific, and regulatory attention, natural color agents are also an important class. Currently, 26 color additives have been given exemption from certification in 21 CFR 73. These agents consist of a variety of natural compounds generally obtained by various extraction and treatment technologies. Included in this group of colors are preparations such as dried algae meal, beet powder, grape skin extract, fruit juice, paprika, caramel, carrot oil, cochineal extract, ferrous gluconate, and iron oxide. A problem encountered in attempts to regulate these additives is the lack of a precise chemical definition of many of these preparations. With a few exceptions such as caramel, which is the most widely used color, the natural colors have not been heavily used. In part, this may be due to economic reasons, but these colors generally do not have the uniformity and intensity characteristic of the synthetic colors, therefore necessitating higher concentrations to obtain a specific color intensity. They also lack the chemical and color stability of the synthetic colors and have a tendency to fade with time.

Concerns about certified color additives (synthetic dyes) were raised in the 1970s by Dr. Benjamin Feingold, a pediatrician, who claimed that these additives were linked to behavior changes and hyperactivity in children and proposed a diet free of synthetic dyes. Recently, the FDA and FDA’s Food Advisory Committee confirmed earlier findings that there is not enough evidence to conclude that certified color additives contribute to hyperactivity in children in the general population (FDA, 20lla and 2011b).

The maximal intake of food colors is estimated to be ~53.5 mg/day, whereas the average intake per day is ~15 mg (Committee on Food Protection, 1971). Only about 10% of the food consumed in the United States contains food colors. The foods that utilize food colors in order of the quantity of color utilized are (1) beverages, (2) candy and confections, (3) dessert powders, (4) bakery goods, (5) sausages (casing only), (6) cereals, (7) ice cream, (8) snack foods, and (9) gravies, jams, jellies, and so forth (Committee on Food Protection, 1971).

The Importance of Labeling

The importance of labeling was first realized in its ability to protect consumers from economic fraud by requiring that the weight and exact contents of the product be stated; otherwise, the product was mislabeled. Later, it became obvious that labels could also serve a purpose in assuring the safety of the consumer by including safety warnings for particularly susceptible groups, including those exhibiting allergies or food intolerance.

Food allergies have a considerable impact on modern society. There is no known cure for food allergies and although accidental exposure is common, avoidance of the offending foods is the only successful noninterventional approach. Food allergy is the leading cause of anaphylaxis, a severe type of allergic reaction requiring hospitalization. It is estimated that 3% to 4% of adults and about 6% of young children in the United States suffer from food allergies, and food allergies account for 35% to 50% of all cases of anaphylaxis, the latter of which has increased from 21,000 per year in 1999 to 51,000 per year in 2008 (AAAAI, 2011). Food allergies are reported to cause approximately 150 to 200 fatalities annually (AAAAI, 2011).

Effective from January 1, 2006, as a result of the Food Allergen Labeling and Consumer Protection Act of 2004 (FALCPA), manufacturers are required to identify the presence of ingredients that contain protein derived from milk, eggs, fish, crustacean shellfish, tree nuts, peanuts, wheat, or soybeans. These eight major food allergens account for 90% of food allergic reactions. In addition, FALCPA labeling regulations require declaration of the specific type of tree nut (eg, almonds, pecans, or walnuts), the species of fish (eg, bass, flounder, or cod), and the kind of crustacean shellfish (eg, crab, lobster, or shrimp). FALCPA requires, with a few exceptions, the label on a food product (eg, conventional foods, dietary supplements, infant formula, and medical foods) that is or contains an ingredient (such as a spice, flavoring, coloring, or incidental additive) that includes a “major food allergen” (Carabin and Magnuson, 2006). Also, because the source of ingredients such as lecithin, flour, or whey may not be obvious to the consumer, the source must follow the name of the ingredient. For example, the above ingredients must be given on the labels as follows—lecithin (soy), flour (wheat) and whey (milk); alternatively, immediately after or next to the list of ingredients in a “contains” statement—for example “Contains wheat, milk and soy.”2

Labeling requirements for nonallergens include those for intolerance (eg, lactose or gluten intolerance) or, for example, the presence of phenylalanine for phenylketonuria patients, are especially important when these substances may be present in foods where their presence may not be expected. Label warnings also include those warning of a threshold for a laxative effect (eg, polydextrose, mannitol, sorbitol). The FDA has indicated that, at this time, it is not aware of any information that foods developed through genetic engineering differ as a class in any attribute from foods developed through conventional means that would warrant a special label (Thompson, 2000). The FDA allows companies to include on the label of a product any statement as long as the statement is truthful and not misleading.

Methods Used to Evaluate the Safety of Foods, Ingredients, and Contaminants

Safety Evaluation of Direct Food and Color Additives

The basic concept that forms the foundation for the safety evaluation of direct food and color additives is the recognition that the safety of any substance added to food must be established on the basis of the intended conditions of use in food. Factors that need to be taken into account include (1) the purpose for use of the substance, (2) the food to which the substance is added, (3) the concentration level used in the proposed foods, and (4) the population expected to consume the substance.

The evaluation of a new food additive is a complicated and expensive undertaking, especially when the additive will be widely used in many foods. Each additive can pose unique safety questions depending on its chemistry, stability in use, metabolism, safety study results, and estimated human exposure. Integral to a discussion of exposure is the concept of the “whole food additive.” This refers to the additive, the degradation or conversion products arising from the use of the additive in foods, and the impurities found in the manufactured additive itself (Kotsonis and Hjelle, 1996).

Exposure: The Estimated Daily Intake

As noted earlier, prior to 1958, the FDA held to the philosophy that food additives (and potential contaminants) should not be harmful at any level. This is impractical, as many substances critical to life, such as vitamin A or even water, are toxic at high doses. These examples underscore the fact that exposure level is a major factor in a safety evaluation, and is reflected in the FDA’s Toxicological Principles for the Safety Assessment of Direct Food Additives and Color Additives Used in Food (FDA, 1982a),3 a source used to determine what testing methods should be done to determine the safe use of a substance.

In food additive safety determinations, exposure is usually referred to as EDI and is based on the daily intake (I) of the food in which the substance will be used and the concentration (C) of the substance in the food. Many food ingredients are used in several different food categories, but as an example, if an additive is used only in breakfast cereals at a concentration that does not exceed 20 μg/g (ppm) and the mean daily intake of breakfast cereals is 175 g/person/day, the EDI is calculated at 3500 μg/person/day. As most food additives are used in many foods, the total exposure is the sum of the exposures from each of the food categories. The formula for exposure to food additive B is EDI_B = (C_Bf × I_f) + (C_Bg × I_g) + (C.…) where C_Bf and C_Bg are the concentration of B in food category f and g, respectively. I_f and I_g are the daily intake of food category f and g, respectively. Therefore, the EDI is the sum of the individual contributions of B in each of the food categories.

The same principles may be applied to the estimation of the consumption of residue from secondary direct additives (substances not intended to remain in a food after the technical effect has been accomplished, examples include processing agents such as solvents, sanitizers, and defoaming agents) and contaminants. Additional information on the exposure to direct food additives and contaminants has been made available by the agency’s Center for Food Safety and Applied Nutrition.4

Calculating the EDI raises several basic questions: (1) how does one determine the amount of a food additive that is added to each food category and (2) how are food categories determined? To determine the amount of food additive added to each food category, the amount or, if a range, the highest end of the range of use levels for the new substance is used. These food group maximums are not to be exceeded by a food manufacturer based on the cGMP regulation (cGMPs; 21 CFR 110), which requires a manufacturer not to add more of an additive than is reasonably required to achieve the specific technical effect of the food additive.

General food categories have been specified by the FDA (21CFR170.3(n)). A sample of the categories is presented in Table 31-9. These categories were derived from a survey of food additives conducted by the National Academy of Sciences/National Research Council and published in 1972 (NRC/NAS, 1979). This survey pioneered the use of categorizing foods, but changes in the consumption patterns of the US population, in addition to changes in the types of foods available, have necessitated the generation of additional, more current data.

More contemporary data on food intake has been calculated through the use of food consumption surveys (Table 31-10). Food consumption databases have specific characteristics and are based on particular assumptions. Methods commonly used by regulatory agencies, manufacturers, nutritionists, and general researchers for assessing food consumption by individuals include 24-hour dietary recalls, dietary records, food frequency records, and dietary history accounts (Matulka, 2005). For example, one database may be based on an individual’s food intake from the past 24 hours, whereas another may utilize dietary records taken over a three-day period of time, and yet another may cover average consumption over 14 days. Some databases may provide only general population consumption values, whereas others may provide a detailed breakout of particular subpopulations (eg, the elderly, women, teenagers).

In safety assessments, one must consider other sources of consumption for the intended use of the food additive, such as whether it is already used in food for another purpose, is used in nonfood products (eg, toothpaste, lipstick, drugs, or dietary supplements), or the additive occurs naturally in foods. In summary, to estimate human consumption of a particular food substance, it is necessary to know (1) the levels of the substance in food, (2) the daily intake of each food containing the substance, (3) the distribution of food intake within the population, and (4) the potential exposure to the substance from nonfood sources (Tennant, 2002).

Before a food additive is approved, evidence is required by regulatory agencies that indicate the additive is safe for its intended use(s) and that the EDI for the additive is less than its ADI (Butchko and Kotsonis, 1996). Regulatory agencies may impose restrictions on certain uses of food additives if the EDI exceeds the ADI, or restricts future approvals for new categories of use. Chronic, long-term rodent toxicity studies are usually used in determining the ADI. These studies are used to determine the no-observed-adverse-effect level (NOAEL) for the additive. To provide an adequate level of safety from animal to human extrapolation, a 100-fold safety factor is usually employed to account for species differences and the interindividual variation among humans, to determine the ADI for a food additive. This factor provides a reasonable certainity in estimating safe doses in humans from animal studies (Lehman and Fitzhugh, 1954).5

Assignment of Concern Level and Required Testing

Structure-activity relationships are now the basis for developing many therapeutic drugs, pesticides, and food additives. These relationships are put to good use in the Toxicological Principles for the Safety Assessment of Direct Food Additives and Color Additives Used in Foods (FDA, 1982a), which describes a qualitative “decision tree” that assigns categories to substances on the basis of the structural and functional groups in the molecule. Additives with functional groups with a high order of toxicity are assigned to category C, those of unknown or intermediate toxicity are assigned to category B, and those with a low potential for toxicity are assigned to category A. For example, a simple saturated hydrocarbon alcohol such as pentanol would be assigned to category A. Similarly, a substance containing an α,β-unsaturated carbonyl function, epoxide, thiazole, or imidazole group would be assigned to category C. Thus, based on structure assignment and calculated exposure, the CLs are assigned (Table 31-11). For example, 0.03 ppm of a substance in Structure Category B would be assigned CL II. In contrast, the same dose (ie, 0.03 ppm) of a substance in Structure Category A would be assigned to the lesser CL I.

Once the CL is established, a specific test battery is mandated (Table 31-8). The tests for CL III are the most demanding and provide the greatest breadth for the determination of adverse biological effects, including effects on reproduction. The tests are comprehensive enough to detect nearly all types of observable toxicity, including malignant and benign tumors, preneoplastic lesions, and other forms of chronic toxicity. The tests for CL II are of intermediate breadth. These tests are designed to detect the most toxic phenomena other than late-developing histopathological changes. The short-term (genotoxicity) tests are intended to identify substances for which chronic testing becomes critical. The CL I test battery is the least broad, as is appropriate for the level of hazard which substances in this category may pose. However, if untoward effects are noted, additional assessment becomes necessary. Studies of the absorption, distribution, metabolism, and elimination characteristics of a test substance are recommended before the initiation of toxicity studies longer than 90 days’ duration. Of particular importance for many proposed food ingredients are data on their processing and metabolism in the GI tract.

Unique to food additive carcinogenicity testing is the controversial use of protocols that include an in utero phase. Under such protocols, parents of test animals are exposed to the test substance for four weeks before mating and throughout mating, gestation, and lactation. Most countries and international bodies do not subscribe to the combining of an in utero phase with a rat carcinogenicity study, as this presents a series of logistical and operational problems and substantially increases the cost of conducting a rat carcinogenicity study. The FDA began requesting in utero studies of the food industry in the early 1970s, when it was determined from lifetime feeding studies that the artificial sweetener saccharin produced bladder tumors in male rats when in utero exposure was included. Subsequently, the FDA required the food, drug, and cosmetic color industries to conduct lifetime carcinogenicity feeding studies of 18 color additives in rats using an in utero exposure phase.

Genetic toxicity tests are done to test chemicals for potential carcinogenicity and to assess whether a chemical may induce heritable genetic damage. Currently, genetic toxicity assays can be divided into three major groups: (1) forward and reverse mutation assays (eg, point mutations, deletions), (2) clastogenicity assays detecting structural and numerical changes in chromosomes (eg, chromosome aberrations, micronuclei), and (3) assays that identify DNA damage (eg, DNA strand breaks, unscheduled DNA synthesis).

Because the correlation between carcinogens and mutagens has proved to be less than desirable, as has been demonstrated by false-positive and false-negative findings when carcinogens and noncarcinogens have been examined in genetic toxicity tests, it is recommended that several tests be selected from a battery of tests. It should be kept in mind that as the number of tests employed increases, the possibility of false-negative results increases as well. Consequently, the National Toxicology Program (NTP) has advised that the vast majority of carcinogens are detected by both the Salmonella assay and rodent micronucleus tests.6

Safety Determination of Indirect Food Additives

Indirect food additives are food additives that are not added directly to food, but enter food by migrating from surfaces that contact food. These surfaces may be from packaging material (eg, cans, paper, plastic) or the coating of packaging materials or surfaces used in processing, holding, or transporting food. As defined in the Act, a food contact substance is “any substance intended for use as a component of materials used in manufacturing, packing, packaging, transporting, or holding food if such use is not intended to have a technical effect in such food.”

Essential to demonstrating the safety of an indirect additive are extraction studies with food-simulating solvents. The FDA recommends the use of three food-simulating solvents—10% ethanol, 50% ethanol, and corn oil or a synthetic triglyceride—for aqueous and acidic, alcoholic, and fatty foods, respectively (FDA, 2002a). The conditions of extraction depend in part on the intended conditions of use. Extraction studies are used to assess the level or quantity of a substance that might migrate and become a component of food, leading to consumer exposure.

To convert extraction data from packaging material into anticipated consumer exposure, the FDA has determined the fraction of the US diet which comes into contact with different classes of material: glass, metal (coated and uncoated), paper (coated and uncoated), and polymers. For each class, the FDA has assigned a “consumption factor” (CF), which is the fraction of the total diet that comes into contact with an individual class of material.

The fraction of individual food types (aqueous, acidic, alcoholic, fatty) for which such packaging material is used is referred to as the food-type-distribution factor (f_T). To calculate cumulative estimated daily intake (CEDI), the following equation is used:

CEDI = CF × [(f_Taqueous + f_Tacidic) × (ppm of migrating substance in 10% ethanol) + (f_Talcoholic × ppm of migrating substance in 50% ethanol) + (f_Tfatty × ppm of migrating substance in corn oil)] × 3 kg/person/day

= mg/person/day7.

For additives with virtually no migration (<0.5 ppb), in which the CEDIs correspond to 1.5 μg/person/day, no safety studies are recommended. Migration levels, as determined by extraction studies, that are greater than 0.5 to 50 ppb (150 μg/person/day), in vitro genotoxicity tests should include bacterial mutagenicity and cytogenetic evaluation of chromosomal damage using mammalian cells or an in vitro mouse lymphoma assay. Where there is significant migration, that is, 50 ppb to 1 ppm (3 mg/person/day), genetic toxicity tests should be conducted and the substance should be further evaluated by two subchronic oral toxicity studies (one in a rodent species and one in a nonrodent species). The studies should provide an adequate basis for determining an ADI for the indirect additive or a constituent in the indicated range of CEDIs. In addition, the results of these studies will help determine whether longer-term or specialized safety tests (eg, metabolism studies, teratogenicity studies, reproductive toxicity studies, neurotoxicity studies, and immunotoxicity studies) should be conducted to assess the safety of these substances. For cumulative exposure greater than 1 ppm, FDA recommends submission of a food additive petition (FDA, 2002a).

Safety Requirements for GRAS Substances

In spite of the fact that the FD&C Act and the relevant regulations scrupulously avoid defining food except in a functional sense—“food means articles used for food or drink for man or other animals… [and includes] chewing gum, and articles used for components of any such article”—it regards foods as GRAS when they are added to other food, for example, green beans in vegetable soup (Kokoski et al., 1990; Burdock and Carabin, 2004). It also regards a number of food ingredients as GRAS, and these ingredients are listed under 21 CFR 182, 184, and 186. However, the language used acknowledges that there are substances the FDA considers to be GRAS, which are not listed. This accomplishes two things: (1) It leaves the door open for additional nonlisted substances to be affirmed as GRAS by the agency and (2) reinforces the concept that substances can be deemed GRAS whether or not they are listed by the FDA or on a publicly available list. A list of substances regarded as GRAS is given in Table 31-12. It is important to reemphasize that GRAS substances, though used like food additives, are not food additives. Although the distinction may seem to be one of semantics, it allows GRAS substances to be exempt from the premarket clearance requirements of a food additive petition, an opportunity to demonstrate safety through a history of safe use, and an exemption from the Delaney carcinogens clause because that clause of the Act pertains only to food additives.8

Although the courts have ruled that GRAS substances must be supported by the same quantity and quality of safety data that support food additives, this does not mean that the data must be identical in nature and character to that for a food additive. For uses of substances to be eligible for classification as GRAS, there must be common knowledge throughout the scientific community about the safety of substances directly or indirectly added to food (21 CFR 170.30); this is termed the “common knowledge standard” by FDA.9 The studies relied on for concluding that a given use of a substance is GRAS must be published in the scientific literature. Such studies are unlikely to be conducted in accordance with FDA-recommended protocols, as these studies often are done for reasons unrelated to FDA approval and the data are thus not identical in nature and character to those for a food additive.

GRAS status also can be based on experience with common use in food before January 1, 1958,10 which further distinguishes GRAS data requirements from those demanded of food additives. However, in 1974, the FDA promulgated a regulation (21CFR170.3(f)) defining the term “common use in food” for purposes of section 201(s) of the FD&C Act as follows: “‘[c]ommon use in food’ means a substantial history of consumption of a substance by a significant number of consumers in the United States11.” This requirement for use in the United States was challenged in court and held invalid on appeal.12 In its decision, the court indicated that in the legislative history of the Act, discussion centered on a prolonged history of safe use, rather than a geographical restriction. Although the geographical limitation is no longer mentioned in the regulation, documentation of safe use in whatever domain relied upon must be clearly established. The FDA has made it clear, that although an ingredient may have an extensive history of use prior to 1958, this does not place it beyond regulatory reach, as new data generated must be taken into account—new data may trump history of use.13

FDA does have the last word as indicated in 21 CFR §170.38. The Commissioner, on his own initiative or on the petition of any interested person, may issue a notice in the Federal Register proposing to determine that a substance is not GRAS and is possibly an unsafe food additive subject to section §409 of the Act and subject to recall. Anyone has 60 days to respond to this notice, FDA will review the responses and act on its determination (ie, affirmation of the GRAS status or possible recall).

Lastly, because a GRAS determination does not have to be reported to FDA, the GRAS process should not be thought of as a refuge for those substances with a weak or “cherry-picked” data package that might not survive the scrutiny of a GRAS subjected to the voluntary notification procedure or those GRAS determinations approved by unqualified “experts” (FD&C §201(s); Federal Register 75: 81536–81543, December 28, 2010). Placing a substance in commerce that has not met the appropriate requirements can have consequences; under the Food Safety Modernization Act, signed into law January 4, 2011, new powers were conferred to FDA regarding adulterants in the food supply, including mandatory recall and substantive civil and criminal penalties.

Establishing Safe Conditions of Use for New Foods, Macroingredients and New Technologies

New and novel foods or ingredients and new technologies present new challenges and may require innovative methods for determining safety. For example, with each new additive, it has been traditional (and rooted in a regulation such as 21 CFR 170.22) to establish an ADI, which is usually based on 1/100 of the NOEL (or NOAEL) established in animal testing. This works well for additives projected to be consumed at a level of 1.5 g/day or less (which is equal to or less than 25 mg/kg), for this extrapolates at a 100-fold safety factor to consumption by a rat at a level of 2500 mg/kg/day (about 5% of the rat’s diet). The problem arises when a new food or macroingredient substitute becomes a substantive part of the diet (estimated to constitute as much as 15%–20%). For example, a macroingredient substitute or food projected to be consumed at a level of just 5% of the diet (150 g/day) would require the test animal (rat) to consume 250 g/kg/day, or slightly more than the 25% of the rat’s body weight. To administer test substance in the diet at such high levels is an untenable test requirement; at such high levels in the feed, dietary (ie, nutrient) displacement occurs and the investigator would establish an effect level only for malnutrition in the test animal, not for the toxicity of the macroingredient. Further, for some essential nutrients, such as vitamins A and D and iron, doses 100 times the nutritional use level would be toxic (Kokoski et al., 1990). The answer therefore lies in the careful interpretation of toxicological data and the conduct, where appropriate, of special studies to assess drug interactions, nutrient interactions changes in gut flora, changes in gut activity, and the like (Munro, 1990; Borzelleca, 1992a,b). Also, it may be appropriate to consider what effect, if any, macroingredients may have on individuals with compromised digestive tracts, those dependent on laxatives, and those on high fiber diets, etc.

The regulatory approval of a new food additive is generally based on traditional toxicology studies. The rationale is that data from such studies will adequately predict adverse effects that could occur in humans. However, such studies, especially for novel foods, may not be adequate. Therefore, although human studies are not generally required for food additives, in the case of novel foods, human studies are likely essential in evaluating their safety (Stargel et al., 1996).

Another useful tool in ensuring the safety of a food additive is monitoring it after its approval, or postmarketing surveillance. With widespread use of a food additive, monitoring for consumption can determine whether actual consumption exceeds the EDI and monitoring for anecdotal complaints may identify adverse health effects that escaped detection in earlier studies. This could be especially important for novel foods when traditional toxicology studies are not done at large multiples of the EDI (Butchko et al., 1994, 1996). Also, mandated reporting to the Reportable Food Registry (mandated reporting of any reasonable probability that an article of human food or animal food/feed [including pet food] will cause serious adverse health consequences or death to humans or animals), although meant to detect adulteration, may also serve to provide information of a critical nature. Thus, the combination of traditional toxicity studies, special animal and human studies, and possibly postmarketing surveillance will ensure the safety of consumers and provide evidence to justify a safety factor different from 100.

Transgenic Plant (and New Plant Varieties) Policy

Crops have been genetically modified using conventional breeding methods for more than a hundred years to produce new plant varieties with improved characteristics. Methods such as wide crosses of distantly related species that normally would not interbreed and mutagenesis of developing seeds using radiation or chemical mutagens have been successfully employed to produce genetic variation for selection of improved plant varieties. Over the past 10 to 15 years, scientists have employed biotechnology to add one or more specific genes into crops like soybean, corn, cotton, and canola to improve pest and disease management, resulting in agronomic, economic, environmental, health, and social benefits for farmers (Brooker and Barfoot, 2005; James, 2006, 2011). For example, a significant portion of the corn and cotton crops planted in the United States contains genes derived from the common soil bacterium Bacillus thuringiensis (Bt) that produce insect control proteins. Bt insect control proteins are toxic to certain caterpillar insect pests that attack corn and cotton plants but have no apparent effect on nontarget organisms (EPA, 1988; McClintock et al., 1995; Mendelsohn et al., 2003; OECD, 2007). By enabling the corn and cotton plants to protect themselves from this insect pest, the use of this product can reduce the need for and use of conventional insecticides (Gianessi and Carpenter, 1999; James, 2011).

Irrespective of the breeding method used to produce a new plant variety, tests must be done to ensure that the levels of nutrients or toxins in the plants have not changed and that the food is still safe to consume. For food/feed from biotechnology-derived crops, compositional analyses are done to ensure that the levels of key nutrients or toxins are comparable to food from conventional varieties. This is also done for a few conventionally bred crops where levels of important toxins such as glycoalkaloids in potatoes and erucic acid in rapeseed oil have been monitored. The International Life Sciences Institutes (ILSI, 2006; Alba et al., 2010) supports a large crop composition database that provides information on the natural variability in composition for conventional corn, soybean, and cotton crops. This database provides a reference for comparing the nutrient composition of new crop varieties (Drury et al., 2008; Lundry et al., 2008; Harrigan et al., 2010; Herman et al., 2010). Animal feeding studies are also done with biotechnology-derived crops fed over several weeks to months to a variety of farm animal species to ensure that the performance (feed efficiency, milk production, etc) is comparable to that of conventional controls (Flachowsky et al., 2005, 2007). Food safety studies have also been done with various biotechnology-derived crops to ensure that there are no treatment-related adverse findings (EFSA, 2008). The European Food Safety Authority (EFSA, 2008) reviewed the published animal feeding studies that have been done and concluded that in the majority of the studies there was no evidence of adverse effects. Clearly, new proteins produced in plant varieties must be nontoxic and not have the characteristics of proteins known to cause allergies. Thus, the proteins produced in genetically modified crops are evaluated for toxicity and allergenicity (Sjoblad et al., 1992; Betz et al., 2000; Pariza and Johnson, 2001, Metcalfe et al., 1996; Astwood et al., 2003; Delaney et al., 2008; Hammond and Cockburn, 2008; Goodman et al., 2008; Thomas et al., 2009; Codex, 2009; Parrott et al., 2010; EFSA, 2008, 2011a). The DNA that is introduced into genetically modified plants to direct the production of such new proteins has been determined to be GRAS (FDA, 1992).

The safety of new plant varieties (transgenic plants, genetically modified plants) is regulated primarily under the FDA’s postmarket authority [section 402(a)(1) of the FD&C Act]. This section, previously applied to occurrences of unsafe levels of toxicants in food, is now applied to new plant varieties whose composition has been altered by an added substance. The new policy that has been applied to plants containing substances that are GRAS [Federal Register 57(104): 22984–23005]. The Federal Register notice (May 29, 1992) indicates that “[i]n most cases, the substances expected to become components of food as a result of genetic modification of a plant will be the same as or substantially similar to substances commonly found in food, such as proteins, fats and oils, and carbohydrates.” The notice also indicates the responsibility of the FDA to exercise the premarket review process when the “objective characteristics of the substance raise questions of safety.” In regard to substances within the new variety that are not similar to substances commonly found in food, a food additive petition may have to be filed.

The Federal Register notice offers points of consideration for the safety assessment of new plant varieties (Table 31-13). Accompanying these points of consideration are a decision flowchart and advice that the FDA be consulted on certain findings, for example, transference of allergens from one plant to another, a change in the concentration or bioavailability of nutrients, and the introduction of a new macroingredient.

In the United States, new plant varieties are regulated not only by the FDA, but also by the EPA and USDA. The FDA is responsible for the safety and labeling of foods and feeds derived from crops, irrespective of the method used to produce the new plant variety. The EPA is responsible for assuring the safety of pesticides; thus, in the example cited above whereby a pesticide is produced in a new plant variety, this product would also fall under EPA’s jurisdiction. The USDA’s Animal and Plant Health Inspection Service has responsibility for the environmental safety of field-testing and commercial planting of new plant varieties.

The developer of a biotechnology-derived crop variety must obtain registration from not only the country of origin but from importing countries as well (OECD, 1993). A variety of European/Global Scientific authorities (WHO, 1995; FAO, 1996; OECD, 1997; Codex, 2009; EFSA 2008, 2011a) have provided guidance on the safety assessment process for food and feed derived from biotechnology-derived crops. The process considers two main categories of potential risks: those related to the properties and function of the introduced protein(s), and those related to the whole food crop since insertion of the introduced gene(s) into the plant genome theoretically could cause unintended environmental effects. As in conventional crop breeding, agronomic studies carried out under diverse environmental conditions are used to screen for varieties that exhibit unintended changes so they can be eliminated from development. A review of research projects covering a period of more than 25 years done by the European Commission conclude that “biotechnology, and in particular GMO’s, are not per se more risky than for example conventional plant breeding technologies” (EU, 2010).

Food Macroingredients

The safety assessment of food ingredients consumed in large amounts presents some unique issues. Such ingredients include bulking agents (such as sorbitol, xylitol, or modified starches) to provide mass to the final product. For example, if one teaspoon of sugar (~4 g) is required to sweeten a cup of coffee, it would take only about 20 mg of aspartame to have the same effect—an amount too small for the consumer to measure accurately, so the familiar blue table-top packet will include aspartame and many times that amount in bulking agent. Bulking agents can be used on a larger scale, such as modified starches, synthetic starch (polydextrose), or indigestible fiber, to provide bulk to a meal, employing non-nutritive bulk to promote satiety.

Potential issues associated with macroingredients are (1) physical—a primary mechanical effect resulting from consumption of large amounts of any, especially a nondigestible, material that may result in mechanical blockage of the GI tract (such as konjac gum) or, the reverse, as demonstrated by certain sugar alcohols or polydextrose, which, because of their inability to undergo digestion, may result in osmotic diarrhea at high doses; (2) physical—a secondary effect of preventing absorption of other substances such as calcium and zinc loss because of the presence of phytic acid in certain plant fiber resulting in the formation of insoluble phosphates or the potential for loss of fat-soluble vitamins consequent to the use of large amounts of fat mimetics; (3) primary and secondary nutritional displacement: (a) primary—frank nutritional loss resulting from prolonged use of a nutrient substitute, and (b) secondary loss—the potential loss of micronutrients present in the substituted food and/or potential secondary (possibly undefined) benefits of consuming the displaced food; (4) possibly a toxic effect, but more likely a nontoxic metabolic or physiological effect, may occur as the result of exceeding the capacity of the user to metabolize or excrete the macroingredient, whereas the system in place was adequate for the displaced ingredient at its historic level of consumption; and, (5) toxicity—the amounts consumed by humans may exceed the proportionate amounts that could be given to test animals, which may have contained toxicants whose effects were not realized in the test animals, but may be seen in humans because of the increased consumption—minor processing impurities (eg, antinutrients such as trypsin inhibitors) or contaminants (eg, metals, reaction products such as furan or acrylamide, mycotoxins) may assume greater-than-usual importance (IPCS, 1987; IOM, 1999).

Nanotechnology

Nanotechnology as applied to food science is a truly transformational technology, such as genetic engineering of plants and Nicolas Appert’s invention of airtight food preservation were for previous generations. What makes nanoparticles (nanoengineered materials [NEM]) so unique is that conventional-sized particles (greater than 100 nm) obey conventional Newtonian physics, in which the individual atoms of a molecule or element act as a sum of all the parts. In contrast, NEM tend to obey the laws of quantum physics because their surface-to-mass ratio is so much greater—and with individual atoms closer to the surface, these individual atoms exercise much more influence on the physicochemical properties of the substance, often conferring physical properties much different than a conventional-sized particle (EPA, 2007; NNI, 2011). Potential differences of NEM as compared to their historical-sized counterparts include particle charge, melting point, solubility, surface tension, and many other characteristics that are normally thought of as immutable. In short, at the nanodimension, many new properties and thus new applications are possible for substances whose properties were previously thought to be known and fully exploited (Burdock, 2011).

There is no universal agreement on what constitutes a NEM, although the EFSA has defined a nanomaterial as a material that meets at least one of the following criteria: (1) consists of particles with one or more external dimensions in the size range of 1 to 100 nm for more than 1% of their number within the size distribution; (2) has internal or surface structures in one or more dimensions in the size range of 1 to 100 nm; and/or (3) has a specific surface area by volume greater than 60 m²/cm³, excluding materials consisting of particles with a size lower than 1 nm (EFSA, 2011b). Although FDA has declined to specifically define nanomaterial, in some recent (June, 2011) draft guidelines, it does offer two points industry might consider (1) “[w]hether an engineered material or end product has at least one dimension in the nanoscale range (approximately 1–100 nm),” and (2) “[w]hether an engineered material or end product exhibits properties or phenomena, including physical or chemical properties or biological effects, that are attributable to its dimension(s), even if these dimensions fall outside the nanoscale range, up to 1 μm.”14

The GI tract is a particularly formidable digestive and protective organ with the stomach at a pH approaching 1.0, vigorous pancreatic enzymes, bile salts, and the luxuriant bacterial flora. The objective of this rugged environment is primarily for destruction of ingested pathogens and reduction of foods to absorbable particles. The GI tract is an organ for selective absorption and is the first barrier to substances that should not be absorbed (eg, large molecules or charged molecules such as aluminum lakes of colors) or those substances that serve their function best by not being absorbed (eg, fiber). In this rugged environment, nanotechnology can provide a protective barrier (encapsulation technology) to allow labile substances to sustain their integrity until absorbed in the appropriate area of the intestine. Nanotechnology can also enhance absorption; for example, under normal circumstances, less than 20% of a dose of vitamin E is absorbed; however, the resistance of the intestine may be overcome and absorption enhanced by the ability of nanosized particles to migrate through cells (transposition) or gain entry between enterocytes or through the use of encapsulation technology (Oberdörster, et al., 2005; Kwok, 2011).

The question then arises as to what impact nanotechnology might have on this balance of selective absorption, as many of the principles of absorption (as well as distribution, metabolism, and excretion) may be affected (EPA, 2007). For example, because the intestinal barrier may act as a safety mechanism by slow absorption of a substance, what effects might be seen when the rate and extent of absorption become equivalent to that of an intravenous injection? Likewise, the absence of teratological effects of many substances may be due to their inability to cross the placental barrier, or the lack of central nervous system (CNS) toxicity the result of their inability to cross the blood–brain barrier. Further, for a nanoparticle whose design allows easy entry, as the result of, for example, its size or coating, what effect will this attribute have on the ability of the target species to eventually excrete the particle (Burdock and Williams, 2008). Lastly, a striking observation regarding effects of such particles on health is their ability to generate toxic effects at the site of initial deposition as well as significant systemic toxic responses (EPA, 2004).

The above is not to imply that nano-sized particles have never been with us, as homogenization of milk has always resulted in some nano-sized micelles and a milling or grinding process is going to produce some particles <100 nm. No specific toxicity has been reported as the result of ingestion of nano-sized particles in food, but this may be because of the low level of exposure to the incidental presence of nano-sized particles or there may be observed effects of previously unknown origin and the linkage to nanoparticles has not been made. The physicochemical characteristics of NEM distinguish themselves from the parent substance in a non-NEM format; thus, regulatory agencies are in the process of promulgating guidelines for testing.

In January of 2011, the EFSA produced draft guidelines calling for toxicity testing of NEM, with special reference to in vitro genotoxicity testing (especially if a reactive oxygen species might be generated), ADME studies (likely to be the most affected by nanoparticles and metabolism and excretion parameters are important indicators of biopersistence and bioaccumulation), and a 90-day repeated dose study in rodents. Secondary in the strategy for NEM vetting are in vitro digestion studies, reproduction/developmental studies, and possibly chronic toxicity/carcinogenicity testing. Further, even prior to hazard identification (toxicity testing) is the mandate to characterize the NEM, which includes, but is not limited to, chemical composition, particle size (at least two methods and including size range and number distribution), physical form and morphology, particle and mass concentration, surface chemistry, surface charge, redox potential, dissolution/solubility, viscosity, density, pH, chemical reactivity/catalytic activity, photocatalytic activity, and others as appropriate (EFSA, 2011b). Importantly, as emphasized in the EFSA document, “If not indicated otherwise by consideration of the data, the conventional default uncertainty factors of 10 for inter- and 10 for intraspecies differences should be applied, as currently there are no indications for a need to modify these factors” (EFSA, 2011b).

FDA has not yet promulgated specific guidelines for testing, as FDA prefers to regulate on a product-by-product basis and does not regulate a technology per se, as this would tend to stifle innovation. However, as noted in its June, 2011 “Points to Consider…,” FDA emphasizes examining particle behavior and characteristics (even if outside traditional 100-nm boundaries) “… because properties and phenomena of materials at the nanoscale enable applications that can affect safety, effectiveness, performance, quality, and, where applicable, public health impact of FDA-regulated products.” FDA provides several examples to illustrate the changes that may be brought about by the use of nanotechnology, “… dimension-dependent properties or phenomena may be used for functional effects such as increased bioavailability, decreased dosage, or increased potency of a drug product, decreased toxicity of a drug product, better detection of pathogens, enhanced protection offered by improved food packaging materials, or improved delivery of a functional ingredient or a nutrient in food.” FDA indicates these attributes may be brought about by changes typical of NEM, that is, “[T]he properties and phenomena may be due to altered chemical, biological, or magnetic properties, altered electrical or optical activity, increased structural integrity, or other unique characteristics of nanoscale materials not normally observed in their larger counterparts.” In conclusion, FDA makes it very plain that conversion of an approved product (GRAS or food additive) to the nanoscale or producing changes to the material that may elicit the type of behavior seen at the nanoscale may well render the material unsafe. That is, the FDA statement to this effect is, “these changes may raise questions about the safety, effectiveness, performance, quality, or public health impact of the products.”

Functional Foods

Functional foods are those foods (or food ingredients) whose contribution to the structure and function of the body exceeds their nutritive value. Functional foods include vegetable oil sterol esters and phytostanol esters (block cholesterol absorption), olestra (a medium for frying foods, but not absorbed by the GI tract and therefore does not contribute to caloric intake), fiber (promotes growth of beneficial bacteria, which lowers colonic pH, displaces pathogens, etc), resveratrol (antioxidant), decosahexaenoic acid and eicosapentaenoic acid (DHA/EPA) (essential fatty acids that play a role in reduction of risk of cardiovascular disease), equol (ameliorates symptoms of menopause without concomitant estrogen stimulation), and dimethyl glycine (antioxidant, improves oxygen utilization). Although these foods (or food ingredients) are consumed as such or added to food, they lack a regulatory niche and have been referred to by various names including “bioactive foods” (Rulis, 2005) or more commonly, “nutraceuticals.” These are foods or food ingredients, not drugs, as long as the claims for them do not indicate they are “intended for use in the diagnosis, cure, mitigation, treatment, or prevention of disease in man or other animals” (the definition of a drug, FFDCA §201(g)(1)(B)). FDA has held that claims for any beneficial effects for these ingredients (as food ingredients) must be based on their nutritive value; however, these same ingredients, when sold as a dietary supplement, may include claims for the beneficial effects, irrespective of nutritive value as long as the claim is truthful and not misleading.

Fiber

Dietary fiber belongs to the broad category of carbohydrates. Fiber can be classified into soluble (eg, gums, pectins in fruits, peas, and beans), insoluble (eg, cellulose, whole grain foods, tomato skins), or mixed (eg, bran), and fermentable (eg, pectins, inulins, oligosaccharides) or nonfermentable (eg, cellulose, lignans) (Carabin and Flamm, 1999). Fiber sources are now also divided according to those found naturally in plants (dietary fiber—cellulose, pectin gums, etc) and isolated or synthetic fibers (functional fiber—resistant starch, pectin, and commercially produced fibers such as PGX—a molecule produced from konjac, alginate, and xanthan gums) (Carabin et al., 2009).

Insoluble, nonfermentable fibers are known for their bulking effect that decreases transit time and increases fecal mass. The extent of fermentation of fiber depends on its physical and chemical structure. Fermentable dietary fiber is acted upon by colonic anaerobic bacteria leading to the production of lactic acid, short-chain fatty acids (acetate, propionate, and butyrate), and gases (H₂, CO₂, and CH₄) (Roberfroid, 1993). Modulating effects of fiber considered beneficial for overall health are: (a) increased water-holding capacity of the stool, (b) increase in stool volume and decreased transit time, (c) increased vitamin and mineral absorption, (d) as the product of fermentation, an increase in the colonic bifidobacteria and lactobacilli count (lowering of pH and subsequent decrease in the pathogenic types of bacteria), (e) fermentation also gives rise to absorption of certain organic compounds (eg, short chain fatty acids that promote immune function), and (f) histological and functional changes in the intestinal epithelium (Roberfroid, 1993, 2007; Carabin and Flamm, 1999; Greger, 1999; Scholz–Ahrens and Schrezenmeir, 2007).

In general, dietary fiber has not been shown to exhibit toxicity upon repeated dosing, to be genotoxic or carcinogenic and in general, to adversely affect glycemic control, lipid metabolism, or colonic flora (Carabin and Flamm, 1999). Fiber can enhance uptake of minerals, if phytic acid (hexakisphosphate, a plant storage form of phosphate) is not present. The only potential adverse effect is the inability of the consumer to tolerate the ingested fiber (which may result in GI discomfort, including gas, or diarrhea), the extent of which is generally related to the average degree of polymerization (DP) of the fiber. DP for various fermentable products includes inulin (DP = 2–60, DP_avg = 12) and oligofructose (DP = 2–8, DP_avg = 4) (Roberfroid, 2007); in general, the lower the average DP, the quicker the fermentation time and the increased possibility of intestinal discomfort. Therefore, clinical “tolerance studies” are conducted to determine the reaction of test subjects de novo or following an acclimation period (Carabin et al., 2010).

DMG

Dimethylglycine (N,N-dimethylglycine or DMG) is present in a broad range of foods including meats (especially liver), grains, legumes, and other edible plants (Balch and Balch, 1997). A necessary and endogenous component of human, animal, and plant metabolism, DMG is an intermediary metabolite in cellular choline and betaine metabolism and is involved in a variety of biological roles, including a major role in the contribution of glycine for synthesis of glutathione, an endogenous antioxidant (Friesen et al., 2007). Additional to and independent of its role in glutathione synthesis, the scavenging potential of orally administered DMG for free radicals has also been demonstrated in rats with ulcers; also, DMG neutralized free radicals and inhibited lipid peroxidation to reduce oxidative stress under conditions of depleted glutathione concentrations (Hariganesh and Prathiba, 2000).

Although DMG has been available since the 1970s as a dietary supplement for the improvement of athletic performance (Tonda and Hart, 1992), its efficacy as a performance enhancer in humans has not been confirmed (Harpaz et al., 1985; Bishop et al., 1987). DMG may, however, act to enhance the immune system (Reap and Lawson, 1990) and has been suggested to assist in the maintenance of aged mitochondria (Pak and Jeong, 2010).

In animal feed (eg, swine and poultry feeds), DMG acts as an external emulsification agent and improves the digestibility of crude fat, protein, and nitrogen-free components, which makes these fractions more available for nutritional purposes (Cools et al., 2010; Kalmar et al., 2010). Emulsifiers have been used for years in swine feeds to improve nutrient digestibility and, due to improved nutrient utilization, to decrease feed costs (Jones et al., 1992; Dierick and Decuypere, 2004). As an added benefit for commercial broilers, the metabolic and nonenzymatic antioxidant properties of DMG also improve oxygen utilization and reduce oxidative stress. Improved oxygen utilization compensates for the inadequate cardiovascular system that makes modern broiler strains susceptible to the development of pulmonary hypertension and ascites, which in turn can result in high (>25%) mortality and financial loss (Kalmar et al., 2010). The reduction of oxidative stress, a known physiological factor in the pathogenesis of ascites (Bottje and Wideman, 1995), attenuates the progression of pulmonary hypertension. DMG has also been shown to increase oxygen uptake and retard the formation of lactic acid in rats, rabbits, and horses under stressed conditions (Meduski et al., 1980; Levine et al., 1982) and to reduce recovery periods for equine and canine athletes (Gannon and Kendall, 1982; Rose et al., 1989).

Safety Requirements for Dietary Supplements

Following a nearly 70-year history of questions concerning the regulatory classification of dietary supplements (during which time dietary supplements had been variously classified as drugs or unapproved food additives), on October 15, 1994, President Clinton signed into law the Dietary Supplement Health and Education Act. Relevant to this discussion on safety, the amendment to the Act is significant in several respects: (1) a dietary ingredient must have its origin as a constituent of food or a metabolite thereof (see FFDCA§201(ff) of the act for a complete list); (2) there is a different standard of safety for dietary supplements than for food ingredients, that is, a “reasonable expectation of no harm” in contrast to that for food ingredients, “a reasonable certainty of no harm”; (3) a genuine grandfathering provision allowing history of use to count for all or part of the determination of safety with no notification required for (a) a dietary ingredient in use before October 15, 1994, or (b) if an equivalent amount had already been approved for use in food, or (c) if an equivalent amount is present in a food that has been consumed for at least 25 years; (4) a requirement for notification of the FDA for any “new” dietary ingredient and that the notification should describe the rationale used to determine the “reasonable expectation of no harm”; and, (5) the burden of proof of lack of safety is the responsibility of FDA.

A troubling finding is that in 1994, the FDA estimated that 4000 dietary supplements were on the market. The agency now says there are more than 55,000 products on the market, but in the interval of October of 1994 to July of 2011, the agency has received only 700 new dietary-ingredient notifications (filings).15 Of those filings, approximately 70% have been rejected by FDA as being incomplete (inadequate identification of the product, errors in directions for use, inadequate evidence of safety, etc). This high rejection rate fuelled demands for guidelines from the agency on what should be included in a notification and how the safety requirements might be met. The agency ultimately responded with draft guidelines in July, 2011.16

Unlike the testing guidelines for food ingredients, wherein functional groups and exposure dictate the CL and resulting testing regimen, testing guidelines for dietary supplements rely heavily on proposed exposure of the material as a supplement versus historical exposure (length of exposure and quantity of exposure) of that material (Table 31-14).

Assessment of Carcinogens

Carcinogenicity as a Special Problem

Congress provided the FDA with wide latitude in assessing safety and assuring a safe food supply with one exception. That exception is a provision of the FD&C Act known as the Delaney clause, which prohibits the approval of regulated food additives “found to induce cancer when ingested by man or animals” [sections 409(c)(3)(A), 706(b)(5)(B), and 512(d)(1)(H)].

The Delaney prohibition applies only to the approval of food additives, color additives, and animal drugs; it does not apply to unavoidable contaminants or GRAS substances or ingredients sanctioned by the FDA or USDA before 1958. The clause also does not apply to constituents that are present in food or color additives or animal drugs, provided that the level of such contaminants can be demonstrated to be safe and the whole additive, including its contaminants, is not found to induce cancer in humans or animals. This interpretation of the Delaney clause was set forth by the FDA in its so-called “constituent policy” published on April 2, 1982, as an Advanced Notice of Proposed Rulemaking. The policy mandates the development and use of animal carcinogenicity data and probabilistic risk assessment to establish a safe level for the contaminant in the additive under its intended conditions of use.

The constituent policy and, as will be discussed later, the implementation of the so-called diethylstilbestrol (DES) proviso for animal drugs under the Delaney clause have forced the FDA to develop a means for establishing safe levels for carcinogenic substances. The DES proviso allows the addition of carcinogenic animal drugs to animal feed if they leave no residue in edible tissue as determined by an approved analytic procedure. To do this, the FDA has turned to the use of probabilistic risk assessment in which tumor data in animals are mathematically extrapolated to an upper bound risk in humans exposed to particular use levels of the additive. The FDA takes the position that considering the many conservative assumptions inherent in the procedure, an upper bound lifetime risk of one cancer in a million individuals is the biological equivalent of zero.

Much controversy surrounds the use of risk assessment procedures, in part because estimates of risk are highly dependent on the many assumptions that must be made. The common practice of testing at a maximum tolerated dose (MTD) (Williams and Weisburger, 1991) raises the question of appropriateness to human exposure. Do high test doses cause physiological changes unlike those from human exposure? The basic assumption in quantitative risk assessment (QRA) is that the dose–response curve is linear beneath the lowest observable effect and may result in the calculation of relatively high risks even at doses that are much lower than the lowest dose that produces cancer in experimental animals. QRA is more a process than a science; many steps in the process are based on assumptions, not proven scientific facts. If only the most conservative assumptions are made throughout the process, many will represent overestimates of human risk by 10- or 100-fold, leading to a combined overestimate of perhaps a million-fold or more. Table 31-15 provides some rough estimates of potential ranges of uncertainty that might lead to large overestimates (Flamm and Lorentzen, 1988).

Historically, the FDA has employed a high threshold for establishing that a food or color additive has been found to induce cancer when ingested by humans or animals. If these additives are found to induce cancer, they cannot be approved for foods or colors, no matter how small the estimated risk. In the end, very few substances have been disapproved or banned because of the Delaney clause. Two indirect food additives (Flectol H and mercaptoimidazole) that migrate from packaging material were banned. Among direct additives, safrole, cinnamyl anthranilate, thiourea, and diethylpyrocarbonate were banned because of the Delaney clause, and diethylpyrocarbonate because it forms urethane.

A number of substances (eg, butylated hydroxyanisole [BHA], xylitol, methylene chloride, sorbitol, trichloroethylene, nitrilotriacetic acid [NTA], diethylhexyl phthalate, melamine, formaldehyde, bentonite) listed in the Code of Federal Regulation as food additives are also listed as carcinogens by NTP, the International Agency for Research on Cancer (IARC), or the state of California (under the Safe Drinking Water and Toxic Enforcement Act of 1986, also known as Proposition 65). How is this possible, and on what basis do these food additive listings continue?

Despite the fact that tests and conditions exist under which each of these substances will produce cancer in animals, the FDA has found it possible to continue listing these substances as food additives. The reasoning applied in almost every case is based on secondary carcinogenesis. The one exception is formaldehyde, which is carcinogenic only on inhalation, and there are compelling reasons to believe that inhalation is not an appropriate test in this case (Flamm and Frankos, 1985). Therefore, formaldehyde is not treated as a carcinogen prohibited by the Delaney clause.

For BHA, which induces forestomach cancer, the concept has been advanced that its carcinogenicity is attributable primarily to a cycle of irritation and restorative hyperplasia (Clayson et al., 1986). For xylitol, a sugar alcohol, an increase in bladder tumors and adrenal pheochromocytomas are considered secondary to calcium imbalance resulting from the indigestibility of sugar alcohols and their fermentation in the lower GI tract. Sorbitol, another sugar alcohol, behaves in a similar manner. For NTA, the argument is secondary carcinogenesis, and although specific explanations vary, the mechanism involving zinc imbalance has considerable scientific support.

Thus, the FDA has generally interpreted the phrase “found to induce cancer when ingested by man or animals” as excluding cancers that arise through many secondary means. Therefore, to be a carcinogen under the Delaney clause, a food or color additive must be demonstrated to induce cancer by primary means when ingested by humans or animals or to induce cancer by other routes of administration that are found to be appropriate. This is interpreted to mean that the findings of cancer must be clearly reproducible and that the cancers found are not secondary to nutritional, hormonal, or physiological imbalances. This position allows the agency to argue that changing the level of protein or fat in the diet does not induce cancer but simply modulates tumor incidence (Kritschevsky, 1994).

Biological Versus Statistical Significance

Much can be learned about the proper means of assessing carcinogenicity data by studying large databases for substances that have been tested for carcinogenicity many times. The artificial sweetener cyclamate is an example. The existence of more than a dozen studies on cyclamate and the testing of multiple hypotheses at dozens of different organ and tissue sites in all these studies led to the awareness that the overall false-positive error rate could be inflated if individual findings were viewed out of context (FDA, 1984). Therefore, very careful attention must be paid to the totality of the evidence.

The possibility of false-negative error is always of concern because of the need to protect public health. However, it should be recognized that any attempt to prove absolutely that a substance is not carcinogenic is futile. Therefore, an unrelenting effort to minimize false-negative errors can produce an unacceptably high probability of a false positive. Further, demanding certainty (ie, a zero or implicitly an extremely low probability of false-negative error) has negative consequences for an accurate decision-making process. This is the case because it severely limits the ability to discriminate between carcinogens and noncarcinogens on the basis of bioassays (FDA, 1984).

In addition to the false-positive/false-negative trap, which is a statistical matter, there are many potential biological traps. The test substance, typically administered at high MTDs, may affect one or more of the many biological processes known to modulate tumor incidence at a specific organ site without causing an induction of tumors at that or any other site. Nutritional imbalances such as choline deficiency are known to lead to a high incidence of liver cancer in rats and mice. Simple milk sugar (lactose) is known to increase the incidence of Leydig cell tumors in rats. Caloric intake has been shown to be a significant modifying factor in carcinogenesis. Impairment of immune surveillance by a specific or nonspecific means (stress) affecting immune responsiveness and hormonal imbalance can result in higher incidences of tumors at specific organ sites. Hormonal imbalance, which can be caused by hormonally active agents (eg, estradiol) or by other substances that act indirectly, such as vitamin D, may result in an increased tumor incidence. Chronic cell injury and restorative hyperplasia resulting from treatment with lemon flavor (d-limonene) probably are responsible for renal tumor development in male rats by mechanisms that are of questionable relevance to humans (Flamm and Lehman–McKeeman, 1991). In these examples, the increases in tumor incidence at specific organ sites probably are secondary to significant changes in normal physiological balance and homeostasis. Moreover, the increases in tumor incidence, and hence the increases in the risk of cancer, probably would not occur except at toxic doses (Ames and Gold, 1997).

To preserve the ability of a bioassay to discriminate between carcinogens and noncarcinogens, the possibility of false-positive or false-negative results and the possibility of secondary effects must be considered. To be meaningful, evaluations must be based on the weight of evidence. Particular attention must be given to the many factors that are used in deciding whether tumor incidences are biologically as well as statistically significant. These factors include (1) the historical rate of the tumor in question (is it a rare tumor, or does it occur frequently in the controls?); (2) the survival histories of dosed and test animals (did dosed animals survive long enough to be considered “at risk” and what effect did chemical toxicity and reduced survival have in the interpretation of the data?); (3) the patterns of tumor incidence (was the response dose-related?); (4) the biological meaningfulness of the effect (was it experimentally consistent with the evidence from related studies and did it occur in a target organ?); (5) the reproducibility of the effect with other doses, sexes, or species; (6) evidence of hyperplasia, metaplasia, or other signs of an ongoing carcinogenic process (is the effect supported by a pattern of related non-neoplastic lesions, particularly at lower doses?); (7) evidence of tumor multiplicity or progression; and (8) the strength of the evidence of an increased tumor incidence (what is the magnitude of the P value, for pairwise comparison and for trend?).

A good discussion of the use of these factors by scientists in deciding whether a substance induces cancer in animals is contained in the notice of a final rule permanently listing FD&C Yellow No. 6 (51 Federal Register 41765–41783, 1988). An elevation of tumor incidence in rats was identified at two organ and/or tissue sites: (1) medullary tumors of the adrenal glands in female rats only and (2) renal cortical tumors in female rats only. Scientists at the FDA concluded that the increase in medullary tumors of the adrenal glands in female rats did not suffice to establish that FD&C Yellow No. 6 is a carcinogen. The basis for the decision was (1) a lack of dose response, (2) the likelihood of false positives, (3) the lack of precancerous lesions, (4) morphological similarity of adrenal medullary lesions in treated and control rats, (5) an unaffected latency period, (6) a lack of effect in male rats, and (7) a comparison with other studies in which there was no association between exposure to FD&C Yellow No. 6 and the occurrence of adrenal medullary tumors.

A similar judgment was made with respect to the cortical renal lesions in female rats, which were not found to provide a basis for concluding that FD&C Yellow No. 6 can induce cancer of the kidneys. The main reasons leading to this conclusion were (1) the relatively common occurrence of proliferative renal lesions in aged male control rats (28 months or older), (2) the lack of renal tumors in treated males despite their usually greater sensitivity to renal carcinogens, (3) the lack of malignant tumors indicating no progression of adenomas to a malignant state, (4) the lack of a decreased latency period compared with controls, (5) the coincidence of renal proliferative lesions and chronic renal disease, (6) the lack of genotoxicity, and (7) a lack of corroborative evidence from other studies that suggests a treatment-related carcinogenic effect of FD&C Yellow No. 6 on the kidney. Both these examples emphasize the importance of considering all the evidence in attempting to decide the significance of any subset of data.

As essential elements, vitamins, sugars, and calories by themselves can increase tumor incidence in test animals; the mechanism by which tumors arise as the result of exposure to food or food ingredients is critically important to assessing the relevance of the finding to the safety of the substance under its intended conditions of use in food. In an earlier review, McClain (2002) discussed mechanistic consideration in the regulation and classification of chemical carcinogens. More recently, Cohen and Arnold (2011) reviewed chemical carcinogenesis and included discussions on mechanism and testing of carcinogens as well as on mode of action and human relevance.

Carcinogenic Contaminants

The Delaney clause, which prohibits the addition of carcinogens to food, could ban many food additives and color additives if strictly interpreted to include contaminants of additives within its definition. Clearly, this was not Congress’s intent, and just as clearly, the FDA needed to develop a common sense policy for addressing the problem that all substances, including food and color additives, may contain carcinogenic contaminants at some trace level.

Toward this end, the agency argued (FDA, 1982b) that banning food and color additives simply because they have been found or are known to contain a trace level of a known carcinogen does not make sense because all substances may contain carcinogenic contaminants. The agency asserted in its constituent policy that the mere fact an additive contains a contaminant known to be carcinogenic should not automatically lead the agency to ban that food additive but should instead cause the agency to consider the health risks it poses based on its level of contamination and the conditions of its use (FDA, 1977).

Up to 25% of respondents in a survey indicated they had experienced an adverse reaction to a food at some point in their lives (Chapman et al., 2006). Assuming these respondents were not exposed to a frank toxin and, that in fact, they consumed foods not reported harmful to others, these adverse reactions could be divided into allergies, idiosyncratic reactions (often confined to a specific demographic exhibiting a genetic anomaly), an anaphylactoid reaction, a food-drug interaction or, a metabolic (or normal physiological reaction) as the result of a wilful action of the consumer (such as overconsumption or inadequate processing of the food) (Table 31-16).

The categories in this section are not exhaustive, nor all inclusive, and one substance may produce a number of effects. For example, an adverse reaction to wheat can be exhibited in a variety of ways: (1) acute IgE-medicated reactions, (2) local inhalational reactions (eg, Baker asthma), (3) systemic reactions that occur when wheat is ingested following exercise and, (4) cell-mediated reactions in atopic dermatitis and celiac disease (Chapman et al., 2006).

Food Allergy

Food hypersensitivity (allergy) refers to a reaction involving an immune-mediated response. Such a response is generally IgE-mediated, although IgG4- and cell-mediated immunity may also play a role in some instances (Fukutomi et al., 1994). What generally distinguishes food allergy from other reactions is the involvement of immunoglobulins, basophils, or mast cells (the latter being a source of mediating substances including histamine and bradykinin for immediate reactions and prostaglandins and leukotrienes for slower-developing reactions) and a need for a prior exposure to the allergen or a cross-reactive allergen. An allergic reaction may be manifested by one or more of the symptoms listed in Table 31-17. The list of foods known to provoke allergies is long and is probably limited only by what people are willing to eat, although milk, egg, peanut, tree nut, fish, and shellfish account for more than 50% of all food allergies (Schneider–Chafen et al., 2010). Although cutaneous reactions and anaphylaxis (a severe allergic reaction, resulting in a drop of blood pressure, and may be fatal) are the most common symptoms associated with food allergy, the body is replete with a repertoire of responses that are rarely confined to only a few foods.

Eosinophilic esophagitis (EoE) is a newly recognized chronic disease that can be associated with food allergies. EoE is characterized by inflammation and accumulation of eosinophils in the esophagus. Symptoms include nausea, vomiting, and abdominal pain after eating and can resemble acid reflux. In severe cases, difficulty in swallowing solid food or solid food sticking in the esophagus may be manifested; in infants, this disease may be associated with a failure to thrive (NIAID, 2011).

A curious type of food allergy, the so-called exercise-induced food allergy (or exercise-induced anaphylaxis or exercise-induced asthma), is apparently provoked by exercise, which has been immediately preceded or followed by the ingestion of food. Chapman et al. (2006) have divided the patients into two subsets, one of which may develop a reaction in temporal proximity to any type of food and the second subset within temporal proximity to certain foods including, but not limited to, shell fish, cephalopods, peach, grapes, tomato, wheat, buckwheat, celery, dairy products, chicken, matsutake mushrooms, and “solid” food (Taylor et al., 1989; Taylor and Helfe, 2002; Chapman et al., 2006). The exact mechanism is unknown, but it may involve enhanced mast cell responsiveness to physical stimuli and/or diminished metabolism of histamine, similar to red wine allergy (Taylor and Scanlan, 1989).

Non-IgE–mediated immunological reactions to foods include (1) food-induced enterocolitis and colitis, (2) malabsorption syndromes (such as celiac disease), (3) cow’s milk-induced syndromes, and (4) dermatitis herpetiformis (which, despite the name, has nothing to do with the herpes virus, but an immunological reaction to dietary gluten with IgA involvement (Chapman et al., 2006)). Food intolerance in patients with chronic fatigue syndrome may not have much of an allergic component, but has been shown to be a somatization trait of patients with depressive symptoms and anxiety disorders (Manu et al., 1993).

Chemistry of Food Allergens

Food allergens are generally glycoproteins with molecular weights ranging from 10 kDa to 70 kDa (Chapman et al., 2006). Although almost all foods contain one or more proteins, a few foods are associated more with allergic reactions than are others. For example, anaphylaxis to peanuts is more common than is anaphylaxis to other legumes (eg, peas, soybeans). Similarly, although allergies may occur from bony fishes, there is no basis for cross-reactivity to other types of seafood (eg, molluscs and crustaceans), although dual (and independent) sensitivities may exist. Interestingly, patients who are allergic to milk usually can tolerate beef and inhaled cattle dander, and patients allergic to eggs usually can tolerate ingestion of chicken and feather-derived particles (Anderson and Sogn, 1984), although in the “bird-egg” syndrome, patients can be allergic to bird feathers, egg yolk, egg white, or any combination of the three (DeBlay et al., 1994; Szepfalusi et al., 1994),

Some of the allergenic components of common food allergens are listed in Table 31-18. Although food avoidance is usually the best means of protection, it is not always possible because (1) the content of some prepared foods may be unknown (eg, the presence of eggs or cottonseed oil); (2) there is the possibility of contamination of food from unsuspected sources (eg, Penicillium in cheeses or meat, Candida albicans) (Dayan, 1993; Dorion et al., 1994), and cow’s milk antigens in the breast milk of mothers who have consumed cow’s milk (Halken et al., 1993)); (3) the presence of an allergen in a previously unknown place (the insertion of Brazil nut DNA into soybeans and subsequent appearance of the allergic 2S protein in soybean products (Nordlee et al., 1996)); and (4) there is a lack of knowledge about the phylogenetic relationships between food sources (eg, legumes include peas, soybeans, and peanuts).

Cross-reactivity has not been well studied, but can have severe consequences nonetheless. The latex-fruit syndrome is an example of a sensitivity of latex, which cross-reacts with the proteins in fruit, the most commonly cited of which are banana, avocado, and kiwi. Seed storage proteins (particularly the 2S albumin family) appear to be the common denominator for allergy to sesame, mustard, sunflower, and cottonseed (Chapman et al., 2006).

Demographics of Food Allergy and Intolerance

Although children appear to be the most susceptible to food allergy, with adverse reactions occurring in 4% to 6% of infants, the incidence appears to taper off with maturation of the digestive tract, with only 1% to 2% of young children (4–15 years) susceptible (Fuglsang et al., 1993). The increase in the number of adults exhibiting food allergy may be due in part to an expanded food universe, that is, an increased willingness to try different foods. In one study, allergies among young children were most commonly to milk and eggs, whereas allergies that developed later in life tended to be to fruit and vegetables (Kivity et al., 1994).

Familial relationships also play a role. Schrander et al. (1993) noted that among infants with cow’s milk protein intolerance, 65% had a positive family history for atopy (first- or second-degree relatives) compared with 35% of healthy controls.

Food Idiosyncrasy

Food idiosyncrasies are generally defined as quantitatively abnormal responses to a food substance or additive; this reaction differs from the physiological effect, and although it may resemble hypersensitivity, it does not involve immune mechanisms. Food idiosyncratic reactions include those that occur in specific groups of individuals who may be genetically predisposed. Examples of such reactions and the foods that probably are responsible are given in Table 31-19.

Lactose Intolerance

Probably the most common idiosyncratic reaction is lactose intolerance, a deficiency of the lactase enzyme needed for the metabolism of the lactose in cow’s milk. A lack of this enzyme results in fermentation of lactose to lactic acid and an osmotic effect on the bowel, with resultant symptoms of malabsorption and diarrhea. Lactose intolerance is lowest in northern Europe in 3% to 8% of the population; it reaches 70% in southern Italy and Turkey and nearly 100% in Southeast Asia (Anderson and Sogn, 1984; Gudmand-Hoyer, 1994). Lactose intolerance is prominent among blacks with an incidence of 27% in black children ages 12 to 24 months, which may increase to 33% by age six years (Juambeltz et al., 1993).

Favism (Glucose-6-phosphate dehydrogenase Deficiency)

Favism is the adverse reaction by certain glucose-6-phosphate dehydrogenase (G6PD)–deficient individuals to certain foods, resulting in a type of debilitating hemolytic anemia that can be fatal. There are nearly 400 different types of G6PD deficiencies (G6PDD) affecting nearly 400 million people, but classically described favism has a distinct geographical demographic (Mediterranean countries, including Italy, Greece, Spain, Portugal, and Turkey). Favism may well be the oldest enzymopathy, as it was recognized even by the early Greek mathematician Pythagoras, who is said to have warned his students against eating fava beans (Luzzatto et al., 2001).

Consumption of fava beans (broad beans) results in the release of the pyrimidine aglycones divicine and isouramil, which in combination with ascorbic acid and other possible cofactors ultimately interfere with G6PD in its task to regenerate reduced glutathione. Because red blood cells are exposed to relatively high concentrations of oxidants, the lack of glutathione activity leads to oxidative cross-linking of erythrocyte membrane proteins and consequent distortion of the cell and oxygen-carrying capacity is impaired. Ca⁺²-ATPase activity may be reduced and intracellular proteolytic enzymes may be activated, resulting in hemolysis. The end-stage process, hemolysis, often results within 2 to 24 hours following consumption of the beans. An episode may also be initiated by inhalation of pollen from the fava plant flower. Symptoms include fever, jaundice, dark red urine, headache, and increased heart rate. Affected individuals become weak and experience pain in the back and abdomen.

G6PDD is an X-linked recessive disease (mapped to Xq28 and close to the genes controlling hemophilia A and color blindness), with many variants and some unknown cofactors, such that not all individuals react the same to an identical amount of fava beans, if there is any reaction at all. Indeed, the beans themselves have variable amounts of divicine and isouramil, with the greatest amount in the springtime. In addition to this sensitivity to fava beans, individuals also may not tolerate certain drugs including, but not limited to, primaquine-type antimalarials, sulfonamides and sulfones, naldixic acid, methylene blue, and some chemicals such as naphthalene (mothballs) and trinitrotoluene. If there is an upside to G6PDD, it may be its association with resistance to malaria (Luzzattto et al., 2001; Anonymous, 2002; Brunton et al. 2006; Beutler, 2008; Khoo, 2010).

Asian Flush Syndrome

Many Asians are known to metabolize alcohol poorly and following adequate alcohol intake, exhibit facial flushing, nausea, dizziness, tachycardia, sweating, etc. The intolerance is the inability to effectively metabolize acetaldehyde to acetate (by NAD⁺-dependent ALDH2), the second step in ethanol detoxification, leading to a build-up of acetaldehyde.

There is a superfamily of aldehyde dehydrogenases (ALDH) and eight are found in human DNA, but the most studied are ALDH1 (a cytosolic form) and ALDH2 (a mitochondrial form). A variant of ALDH2, found in approximately 50% of Asians, upon sequencing, was found to be a single glutamate-to-lysine substitution at position 487, and although the enzyme was active, the K_m for NAD⁺ was so elevated the enzyme was functionally inactive. This acetaldehyde buildup is functionally similar to that produced chemically by disulfiram (tetraethylthiuram disulfide or Antabuse), which inhibits the enzyme, although the cytosolic form is more sensitive to disulfiram than the mitochondrial form. Asian flush syndrome has been reported in Japanese, Chinese, and Koreans and quantitative estimations in Japanese have shown 40% lack of ALDH2 activity. ALDH2 deficiency is a dominant negative trait, with homozygotes having essentially no ALDH2 activity, but heterozygotes having reduced but detectable activity. The lack of ALDH2 in a large portion of Asians may, in part, account for a comparatively low rate of alcoholism among Asians.

Alcohol dehydrogenase is also important in alcohol metabolism and σ-alcohol dehydrogenase of the stomach was missing in approximately 30% of those biopsied (Crabb et al., 2004). Although this missing enzyme led to an increased level of alcohol reaching the blood stream, it is unlikely to produce a meaningful contribution to Asian flush syndrome (Steinmetz, et al., 1997; Crabb et al., 2004).

Anaphylactoid Reactions

Anaphylactoid reactions are historically thought of as reactions mimicking anaphylaxis through direct application of histamine, the primary mediator of anaphylactic reactions. Anaphylactoid reactions occur following ingestion of scombroid fish (eg, tuna, mackerel, bonito) that have high levels of histidine in their flesh, as well as some nonscombroid fish such as mahi mahi and bluefish. The fish flesh has been acted upon by microorganisms whose growth involves production of histidine decarboxylase to produce histamine, the essential ingredient for the anaphylactoid reaction (Table 31-20) (Clark et al., 1999). Because histamine is heat-stable, cooking has relatively little effect on the outcome. Symptoms of scombroid poisoning include skin flushing, pruritus, throbbing headache, dizziness, nausea, vomiting, abdominal cramps, and diarrhea (Codori and Marinopoulos, 2009). Characteristics distinguishing scombrotoxicosis from enteric pathogens are the relatively sudden onset (10–30 minutes) and the skin flushing of the head, neck and upper torso.

Scombrotoxicosis was reported to be mimicked by the direct ingestion of 90 mg of histamine in unspoiled fish (Van Gelderen et al., 1992), but according to Taylor (1986), the effect of simply ingesting histamine does not produce the equivalent effect. Instead, Taylor claims that histamine ingested with spoiled fish appears to be much more toxic than is histamine ingested in an aqueous solution as a result of the presence of histamine potentiators in fish flesh. The apparent mechanism of potentiation involves the inhibition of intestinal histamine-metabolizing enzymes (diamine oxidase), which causes increased histamine uptake. Melnik et al. (1997) proposed that anaphylactoid responses may be the sum of several mechanisms: (1) an increased intake of biogenic amines (including histamine) with food, (2) an increased synthesis by the intestinal flora, (3) a diminished catabolism of biogenic amines by the intestinal mucosa, and (4) an increased release of endogenous histamine from mast cells and basophils by histamine-releasing food. Scombrotoxicosis in the absence of high histamine levels (less than the FDA action level for tuna of 50 mg histamine/10-g fish) was reported by Gessner et al. (1996). Ijomah et al. (1991) claimed that dietary histamine is not a major determinant of scombrotoxicosis because potency is not positively correlated with the dose and volunteers tend to fall into susceptible and nonsusceptible subgroups. Ijomah et al. (1991) suggested that endogenous histamine released by mast cells plays a significant role in the etiology of scombrotoxicosis, whereas the role of dietary histamine is minor. An exception to this endogenous histamine theory was described by Morrow et al. (1991), who found the expected increase in urinary histamine in scombroid-poisoned individuals but did not find an increase in urinary 9α, 11β-dihydroxy-15-oxo-2,3,18, 19-tetranorprost-5-ene-1, 20-dioic acid, the principal metabolite of prostaglandin D₂, a mast cell secretory product; thus, no mast cell involvement was indicated.

Food–Drug Interactions

Once known as pharmacological food reactions or as “false food allergies” (Moneret–Vautrin, 1987), these adverse reactions were thought to be exaggerated responses to pharmacological agents in food and possibly due to receptor sensitization. However, the majority of food and drug interactions are actually the result of food-induced changes in drug bioavailability or metabolism (ie, pharmacokinetic interactions), although some are the result of pharmacodynamic interactions. For drugs with a narrow therapeutic index and the need for dose titration, even small changes in the dose–response effects can have great consequences (Schmidt and Dalhoff, 2002). The potential to alter therapeutic effect can be great and in recognition of the role that food plays, test meals are now given to determine their effect on drug therapeutic effect (FDA, 2002b).

Pharmacokinetic effects on absorption (eg, gastric pH, gastric emptying, lymphatic flow) were described earlier. Examples of foods affecting a number of cytochromes, Phase II enzymes, and transporters are provided in Table 31-21; the effects on CYP3A4 and P-glycoprotein may be the clinically most important (Harris et al., 2003). Other dietary ingredients that may produce an effect on the overall pharmacokinetics of drugs would include substances that change the pH of urine or simply the presence of fiber in the intestine. Examples of pharmacodynamic interactions might include the effect of unsaturated fatty acids in the diet on anticoagulants or membrane potentials of the membranes in which they become incorporated, or high potassium intake from potassium-rich foods and the risk of hyperkalemia during therapy with angiotensin-converting enzyme inhibitors or spironolactone (Schmidt and Dalholff, 2002). Other pharmacodynamic interactions might include phytoestrogens and other estrogen-stimulating substances during treatment for hormonally sensitive cancers (eg, breast and prostate) and caffeine or other stimulatory methylxanthines from coffee, chocolate, and soft drinks flavored with guarana, during treatment for hypertension.

Metabolic Food Reactions

Metabolic or physiologic food reactions are distinct from other categories of adverse reactions in that the foods are more or less commonly eaten and demonstrate toxic effects only when eaten to excess or improperly processed (Table 31-22). The susceptible population exists as a result of its own behavior, that is, the “voluntary” consumption of food as a result of a limited food supply or an abnormal craving for a specific food.

Excess Consumption of a Normally Nontoxic Food

Glycyrrhizism—Licorice and Hypokalemia

A “pica” or abnormal craving was reported by Bannister et al., (1977), who noted hypokalemia leading to cardiac arrest in a 58-year-old woman who had been eating about 1.8 kg of licorice per week. In “glycyrrhizism,” or licorice intoxication, glycyrrhizic acid is the active component, with an effect resembling that of aldosterone, which suppresses the renin-angiotensin-aldosterone axis, resulting in the loss of potassium. Clinically, hypokalemia with alkalosis, cardiac arrhythmias, muscular symptoms together with sodium retention and edema, and severe hypertension are observed. The syndrome may develop at a level of 100 g licorice/day but gradually abates upon withdrawal of the licorice (Isbrucker and Burdock, 2006).

Isothiocyanates, Thiocyanates, and Goiter

Isothiocyanates are present in a number of foods, especially cruciferous vegetables, including mustard and horseradish (as allyl isothiocyanate), which provide the “bite” associated with these foods, and in watercress (as methyl isothiocyanate), which confers a slight zaniness to the taste. In mustard seed, the glycoside, sinigrin, is acted upon by myrosin in the presence of water and when the seed is injured, liberating the (allyl)-isothiocyanate, a potent antimicrobial (especially antifungal). Other members of the Brassica family, including broccoli, kale, and cabbage, release a thiocyanate ion. Once ingested, both the iso- and thiocyanates bind iodine in the body, preventing its organification, leading to diffuse hyperplastic (iodine-deficient) goiter. Although the degree to which I⁻ is bound by the thiocyanates is not comparable to say, perchlorate anion (ClO⁻₄), nevertheless, in areas of low iodine and high Brassica consumption, pathology could result (Capen et al., 2002; Farwell and Braverman, 2006). Ermans et al. (1972) indicate that chronic consumption of thiocyanate may play a role in endemic cretinism. Paradoxically, excess iodine may also cause goiter (ie, iodine-excess goiter). Excess iodine appears primarily to block the release of T3 and T4 from thyroglobulin and interferes with peroxidation of 2I⁻ to I₂ and disrupts the conversion of monoiodothyronine to diiodothyronine (Capen et al., 2002). The FDA is aware of the possibility of iodine toxicity and has placed limits on iodine in kelp, the products of which have extensive use in food (21CFR172.365).

Nutmeg, Myristicin, and Hallucinations

Myristicin is a naturally occurring insecticide and acaricide that is found in nutmeg and mace (Myristica spp) at concentrations of 1.3% and 2.7%, respectively (Hallstrom and Thuvander, 1997). It is also present in black pepper, carrot, celery, parsley, and dill (Deshpande, 2002). It is estimated that the average total intake of myristicin from dietary sources is on the order of a few mg per person per day (Hallstrom and Thuvander, 1997). Myristicin is a weak inhibitor of monoamine oxidase and is structurally related to mescaline. At a dose level of 6 to 7 mg/kg body weight (bw), it may cause psychotropic effects in man such as increased alertness and a feeling of irresponsibility, freedom, and euphoria. Unpleasant symptoms, such as nausea, tremor, tachycardia, anxiety, and fear have also been reported in humans ingesting this dose. Although the metabolism of myristicin resembles that of safrole, there is no evidence to suggest that myristicin is carcinogenic (Hallstrom and Thuvander, 1997).

At the concentrations normally present in spices or food, the likelihood of toxicity arising from myristicin is low. However, ingestion of greater than 5 g of nutmeg (corresponding to 1–2 mg/kg bw myristicin) has produced toxicological symptoms in humans that are similar to alcohol intoxication. Because the myristicin content of nutmeg is approximately 1% to 3%, it is likely that other components of nutmeg in addition to myristicin contribute to nutmeg toxicity (Hallstrom and Thuvander, 1997; Dolan et al., 2010).

Improperly Prepared Food

Cassava and Konzo

Konzo is a progressive and irreversible spastic paraparesis or tetraparesis resulting from the inadequate processing of the root and leaves of cassava (Manihot esculenta), a hardy plant found in sub-Saharan Africa. Recent outbreaks have occurred in the Democratic Republic of Congo, Central African Republic, Tanzania, and Mozambique. For many people, cassava is a survival food and outbreaks are associated with an agricultural crisis or in periods of food insecurity. Where cassava is routinely eaten, the outbreaks are naturally associated during the cassava harvest period, when the fruit is most plentiful and would constitute a major portion of the diet. The incidence is highest in children and young women and is often associated with populations with decreased intake of sulfur-containing amino acids, and in an iodine-deficient population there is an exacerbation of cretinism. The disease is the result of the cyanogenic glucosides, primarily linamarin (α-hydroxyisobutyronitrile-β-d-glucopyranoside), which is converted to cyanohydrin, cyanide, and cyanate (OCN). Victims have increased serum and urinary thiocyanate concentrations (Cliff et al., 2011; Kassa et al., 2011). Although konzo is often the end product of chronic consumption, in lower and/or less frequent doses, responses include acute intoxication (dizziness, headache, nausea, vomiting, etc). In chronic doses insufficient to cause typical konzo, responses include ataxic neuropathy, loss of sensorium in hands, blindness, deafness, and weakness (Bradbury et al., 2011).

Avoidance of the toxic effects of linamarin is relatively straightforward, but requires a process that takes time and the availability of potable water. The vegetable should be ground to a flour, placed in a flat pan or tray, and sufficient water added to essentially double the volume. The container is placed in the sun for two hours or the shade for five hours. This technique activates the luminase, which produces cyanohydrins, which break down to cyanide. Evaporation of the water allows the cyanide to be released harmlessly into the air and the cyanide will be reduced approximately three- to sixfold, at which time the flour may be made into the traditional porridge (Bradbury et al., 2011).

Cyanogenic glycosides are present in animal feed (flax, Trifolium repens L, or white clover and sorghum species) and are a concern because of the large amounts animals may ingest. Poisoning in dogs has been reported following ingestion of macadamia nuts, which contain the cyanogenic glucoside, proteacin (Swenson et al., 1989; Hansen et al., 2000). Clinical signs include weakness, depression, vomiting, ataxia, and tremor, among others. Clinical signs were reported within 12 hours in most cases and most resolved within less than 24 hours; the average amount of macadamia nuts ingested to produce symptoms was 11.7 g/kg bw (Hansen et al., 2000). Cyanogenic glycosides are also present in the pits/seeds of members of the genus, Prunus (cherry, peach, and apricot pits and apple seeds), but these are rarely eaten (Barceloux, 2008a).

Cycads (and Lytico-Bodig Disease)

Following WW II and the occupation of Guam and neighboring islands by the US, Army physicians noted the native Chamorro people as having an especially high rate of amyotrophic lateral sclerosis (ALS or Lou Gehrig disease or specifically; in this case, Western Pacific amyotrophic lateral sclerosis or lytico-bodig disease). Lytico-bodig disease was often seen in combination with the tremorous symptoms of Parkinson’s disease or, a dementia similar to Alzheimer’s disease and sometimes, a combination of all three of these diseases. A combination of the latter two diseases has been referred to as parkinsonism–dementia (PD) complex. ALS is characterized by progressive weakness of the extremities, muscle spasticity, fasiculations, and followed by atrophy. The PD complex is characterized by bradykinesia, tremor, decreased blink reflex, muscle rigidity, etc, with eventual flexion contractures and progressive dementia (Matsumoto, 1985; Barceloux, 2009a; Cox and Bradley, 2009).

The source of the disease(s) is exposure to cycad seed kernel and to a somewhat lesser extent, consumption of the flying fox (Pteropus mariannus), a mammal known to consume large amounts of cycad seed. The cycad responsible for intoxication is Cycas micronesica, not Cycas circinalis, as was previously thought. The cycad was a traditional food source for the Chamorro until the 1960s. What is in dispute is the actual causative agent and its origin. The cycad is known to contain the neurotoxins β-N-methylamino-l-alanine (BMAA) and methylazoxymethanol β-d-glucoside (cycasin). There is evidence that BMAA may be the product of symbiotic cyanobacteria, instead of the plant itself.

Among people who have used cycads for food, the method of detoxification is remarkably similar. The seeds and stems are cut into small pieces and soaked in water for several days and then are dried and ground into flour. The effectiveness of leaching the toxins from the bits of flesh is most directly dependent on the size of the pieces, the duration of soaking, and the number of water changes (Matsumoto, 1985; Barceloux, 2009a; Cox and Bradley, 2009).

The Grass Pea and Neurolathyrism

Another starvation- and poverty-influenced disease is neurolathyrism, clinically indistinguishable from konzo produced by cassava, but attributed to the excessive intake of the grass pea (Lathyrus sativus L) and the causative agent, β-N-oxalyl-l-α-β-diaminopropionic acid (OADP). The grass pea is a hardy legume that is tolerant to drought, water logging, and salinity and is fairly insect-resistant. It may be among the first to blossom following a flood and silting over of fields. Its growth is now mainly confined to the Indian Subcontinent, Ethiopia and to a lesser extent in Europe, Australia, Asia, and North Africa. The grass pea is often part of a diet for people with no other choices and, despite is widespread reputation as a food with toxic consequences, it is sometimes grown as rotation crop because it fixes nitrogen and is fairly protein- and carbohydrate-rich. As a consequence of its ease of cultivation and nutrients, grass pea flour is often mixed with chickpea. Contributing factors to the toxic effects experienced by the victims, may also include low dietary sulfur amino acids and trace metal deficiencies, but clearly, the largest cohort of victims are those that have the greatest exposure, including hungry children supplementing meager meals with the wild-growing peas or those with the greatest caloric demand and, thus, often the family breadwinner. Detoxification methods include presoaking the pea seeds in cold or hot (50°C) water or cooking in boiling water (alkaline or acidic solutions), roasting at high temperature, or fermentation—all processes result in measureable (but variable) diminution of the OADP, but never total elimination of the toxin (Jiao et al., 2011; Kumar et al., 2011).

OADP is a nonprotein amino acid and comes in an α- and β-form (30:70 ratio at room temperature) with the former less toxic than the latter and heating can convert some β-form to the less-toxic α-form. Concentration of the toxin in the grass pea is dependent on the particular strain and a host of nutrient and environmental variables and even symbiotic relationships with the nitrogen-fixing bacteria on the roots. The toxin is subtle and hyperexcites the neuron causing spastic movements of the legs and, eventually, paralysis, with muscular rigidity and weakness, although with no loss in sensitivity. The disease is characterized by degeneration of pyramidal-tract neurons in the spinal cord and in the area of the cortex controlling the legs. The cascade of events leading to pathology involves excitotoxicty and oxidative stress involving Ca⁺⁺ homeostasis, and may be a model for ALS (Kumar et al., 2011; Van Moorhem et al, 2011).

Ostrich Fern Fronds (Fiddleheads)

Fiddleheads (crosiers) are the unfurled fronds of the ostrich fern (Matteuccia struthiopteris) and are a seasonal delicacy harvested commercially in the Northeastern United States and in coastal provinces of Canada. Called fiddleheads because their spiral appearance resembles the scroll configuration of a stringed instrument or the crosier of a Bishop’s staff.

The ostrich fern was a spring vegetable for American Indians of eastern North America and became part of the regular diet of settlers to New Brunswick in the late 1700s. Until recently, it was consumed primarily in the Maritime Provinces of Canada and in the Northeastern United States. The ferns are available commercially either canned or frozen, but since the early 1980s, farmers’ markets and supermarket chains have sold fresh ferns in season. None of the fiddlehead ferns of eastern and central North America previously have been reported to be poisonous. Although some ferns may be carcinogenic, the ostrich fern has been considered to be safe to eat either raw or cooked. However, in May 1994, outbreaks of food poisoning were associated with eating raw or lightly cooked fiddlehead ferns in New York and eastern Canada. In one incident, approximately 60% of restaurant patrons consuming raw or minimally processed ferns (eg, light sautéing) experienced nausea, vomiting, abdominal cramps, and/or diarrhea within hours (onset can range from 30 minutes to 12 hours and the illness generally lasts two hours) (Health Canada, 2010). Those consuming ferns subjected to more rigorous processing (eg, boiling for at least six minutes) did not experience symptoms. The authors speculated the ferns contained a heat-labile toxin and recommended that ferns be boiled for 10 minutes prior to eating (MMWR, 1994).

Pre- and Postharvest Changes

Cucumbers and Cucurbitacins

Members of the Cucurbitacea family (zucchini, cucumbers, pumpkins, squash, melons, and gourds) produce cucurbitacins (oxygenated tetracyclic terpenes) that act as movement arresters and compulsive feeding stimulants for Diabriticine beetles (corn rootworms and cucumber beetles). Because cucurbitacins act as feeding stimulants, they are added to insecticidal baits to increase efficacy (Martin et al., 2002). To most nondiabrotic herbivores, such as humans, cucurbitacin is perceived as extremely bitter, even at nanogram amounts (Subbiah, 1999).

Under normal circumstances, cucurbitacins are produced at low enough concentrations the bitterness is not detectable by humans. However, in response to stresses such as high temperatures, drought, low soil fertility, and low soil pH, concentrations of cucurbitacins may increase and cause the fruits to have a bitter taste (Feather, 2010). Occasional cases of stomach cramps and diarrhea have occurred in people ingesting bitter zucchini. Twenty-two cases of human poisoning from ingestion of as little as 3 g of bitter zucchini were reported in Australia from 1981 to 1982, and in Alabama and California in 1984. The cultivar implicated in the Australia poisonings was “Blackjack” (Burfield, 2008).

Solanine and Chaconine

The humble potato, Solanum tuberosum, may produce toxic steroidal glycoalkaloids if exposed to light in the field or during storage or otherwise stressed by mechanical damage or improper storage or sprouting. Previously known only as “solanine” or now, (total) SGAs, the major players are α-solanine and α-chaconine and are normally present in amounts of <5 mg/100 g of tuber fresh weight, although a normal toxin load in potatoes of 20 to 100 mg/kg tuber is considered safe. The toxins are not affected by normal baking, frying, or microwave cooking temperatures, but will undergo significant degradation at 210°C (410°F) for 10 minutes, where the SGA are reduced 40% (Barceloux, 2008b). SGAs inhibit cholinesterases and can alter the effects of neuromuscular blocking drugs and anesthetics (Sorensen, 2002). Low doses of these glycoalkaloids can produce GI upset with diarrhea, vomiting, and severe abdominal pain. At higher doses, neurological symptoms are evident with drowsiness and apathy, confusion, weakness, and vision disturbances, followed by loss of consciousness and sometimes, death in the absence of supportive treatment.

Rodents do not absorb solanines as well as humans and the LD₅₀ is >1000 mg/kg in mice and 590 mg/kg in the rat, values that are 300 to 500 times more than the toxic dose of 2 mg/kg in humans and, in which, the estimated lethal dose may be as little as 3 to 6 mg/kg. There was thought to be an association of potato (and SGA) consumption with neural tube defects in children, although prospective studies have not borne out this hypothesis, nor has this effect been seen in animal testing (ICPS, 1993).

In the 1960s, a new variety of potato was produced from a cross of a Delta Gold with a wild type from Peru. The new variety called “Lenape” was tested as a “new potato” for roasting with meats and vegetables. Upon eating the potato, however, the grower was sickened and upon analysis, the potato was found to have high levels of SGAs. It was later determined that a product of the breeding process, the gene for glycoalkaloids, was stimulated to produce a high level of SGA. Immediately, all the seed potato growers were contacted and stocks were recalled or destroyed (Fedoroff and Brown, 2004).

Furocoumarins in Parsnips, Celery, Earl Grey Tea, and Perfume

Furocoumarins are phytoalexins (natural pesticides) produced by the plant in defense against predators including viruses, bacteria, fungi, insects, and animals and, in the latter, have been shown to be phototoxic and photomutagenic. Furocoumarins are present in cold-pressed oils (but not distilled oil) of citrus fruit (oranges [esp, bergamot], lemons, and limes) and are extensively used as fragrance ingredients and as flavor ingredients, including use in Earl Grey tea and halva (a sesame paste confection). Although concentrations of furocoumarins tend to be high in cold-pressed citrus oil, exposure to humans is fairly low (citrus flavor for soft drinks and, importantly, margarita mix, are made from the distilled oils). Concentration of furocoumarins in vegetables (the family Apiacae/Umbelliferae—carrots, parsley, celery, and parsnips) is low, but the risk of adverse effects is greater because mistreatment of the vegetables can result in increased furocoumarin levels and a relatively large amount of the vegetable could be consumed in a single day.

Postharvest treatment of parsnips can greatly influence the concentration of furocoumarins. Ceska et al. (1986) reported that older “spoiled” and diseased parsnips freely available in grocery stores may contain furocoumarin concentrations 2500% higher than fresh parsnips. Microbial infection of parsnip roots can result in a dramatic increase in furocoumarin levels. Furocoumarin concentrations (the sum of five furocoumarins: angelicin, isopimpinellin, 5-methoxypsoralen (MOP), 8-MOP, and psoralen) in freshly harvested parsnips are generally lower than 2.5 mg/kg and do not increase after storage at −18°C for up to 50 days. In contrast, storage of whole parsnips (but not cubes or homogenate) at 4°C resulted in a marked biphasic increase of furocoumarin concentrations (to approximately 40 mg/kg) after seven or 38 days of storage. A dramatic increase in furocoumarin concentrations (up to 566 mg/kg) was observed when whole parsnips were kept at room temperature over 53 days, resulting in a visible microbial (mold) presence (Ostertag et al., 2002).

In celery, infection with fungal pathogens has been shown to produce trimethylpsoralen (which is absent from plants that are not infected) and increased concentrations of 8-MOP. The resulting “pink rot” has caused repeated outbreaks of photophytodermatitis in commercial celery handlers (Zobel and Brown, 1991). Fungal infection also has been shown to stimulate a 155-fold increase in furocoumarin production in carrots (Wagstaff 1991).

Americans may well consume 1.3 mg furocoumarins/day (0.02–0.023 mg/kg/day), although consumption of 200 g of infected parsnips could result in an exposure of 100 mg. A therapeutic dose (treatment of psoriasis) is 0.5 to 0.6 mg 8-MOP/kg bw and the lowest phototoxic dose known is 0.23 mg/kg bw/day. Furocoumarins are phototoxic in combination with UVA exposure. Furocoumarins are also known to be cytotoxic and mutagenic as the result of furocoumarins intercalating between base pairs of DNA, which when exposed to UVA form covalent photoadducts. Furocoumarins have shown carcinogenicity in rats (37.4 mg/kg bw/day × 2 years) and hepatotoxicity in dogs (48 mg/kg) and induce vomiting in monkeys at 6 mg/kg bw/day (SKLM, 2006; Raquet and Schrenk, 2009; Dolan et al., 2010; Guth et al., 2011).

Heavy Metals

There are 92 natural elements; approximately 22 are known to be essential nutrients of the mammalian body and are referred to as micronutrients (Concon, 1988). Among the micronutrients are iron, zinc, copper, manganese, molybdenum, selenium, iodine, cobalt, and even aluminum and arsenic. However, among the 92 elements, lead, cadmium, and mercury are familiar as contaminants (or at least have more specifications setting their limits in food ingredients). The prevalence of these elements as contaminants is not due so much to their ubiquity in nature but rather to their use by humans.

Lead

Although the toxicity of lead is well known, lead may also be an essential trace mineral. A lead deficiency induced by feeding rats <50 ppb lead (vs 1000 ppb in controls) over one or more generations produced effects on the hematopoietic system, decreased iron stores in the liver and spleen, and caused decreased growth (Kirchgessner and Reichmayer-Lais, 1981), but apparently not as a result of an effect on iron absorption. Although the toxic effects of lead are discussed elsewhere in this text, it is important to note that its effects are profound (especially in children) and appear to be long-lasting because mechanisms for excretion appear to be inadequate in comparison to those for uptake (Linder, 1991). Foods may become contaminated with lead if they are grown, stored, or processed under conditions that could introduce larger amounts of lead into the food, such as when a root crop is grown in soil that has been contaminated from the past use of leaded pesticides.

Over the years, recognition of the serious nature of lead poisoning in children has caused the World Health Organization (WHO) and FDA to adjust the recommended tolerable total lead intake from all sources of not more than 100 μg/day for infants up to six months old and not more than 150 μg/day for children from six months to two years of age to the considerably lower range of 6 to 18 μg/day as a provisional tolerable range for lead intake in a 10-kg child. The FDA is recommending that lead levels in candy products likely to be consumed frequently by small children do not exceed 0.1 ppm because such levels are achievable under good manufacturing practices and would not pose a significant risk to small children for adverse effects. This recommended maximum level is consistent with the FDA’s longstanding goal of reducing lead levels in the food supply (FDA, 2005).

Initiatives to reduce the level of lead in foods, such as the move to eliminate lead-soldered seams in soldered food cans that was begun in the 1970s and efforts to eliminate leachable lead from ceramic ware glazes, have resulted in a steady decline in dietary lead intake (Shank and Carson, 1992). However, lead remains in the diet as the result of still extant lead water pipe (EPA, 2011) and has triggered recent recalls of containers (lead in stainless steel brandy flasks) and food (lead in turmeric spice and, ironically, in a bubble gum called “Toxic Waste”) (FDA, 2011c, d, and e).

Cadmium

Cadmium is a relatively rare commodity in nature and usually is associated with shale and sedimentary deposits. It is often found in association with zinc ores and in lesser amounts in fossil fuel. Although rare in nature, it is a nearly ubiquitous element in American society because of its industrial uses in plating, paint pigments, plastics, and textiles. Exposure to humans often occurs through secondary routes as a result of dumping at smelters and refining plants, disintegration of automobile tires (which contain cadmium-laden rubber), subsequent seepage into the soil and groundwater, and inhalation of combustion products of cadmium-containing materials. The estimated yearly release of cadmium from automobile tires alone ranges from 5.2 to 6.0 metric tons (Davis, 1970; Lagerwerff and Sprecht, 1971). However, in the absence of an overt environmental contamination, food remains the primary source for cadmium uptake for nonsmokers (EFSA, 2009).

Cadmium, like mercury, can form alkyl compounds, but unlike mercury, the alkyl derivatives are relatively unstable and consumption almost always involves the inorganic salt. Of two historical incidents of cadmium poisoning, one involved the use of cadmium-plated containers to hold acidic fruit slushes before freezing. Up to 13 to 15 ppm cadmium was found in the frozen confection, 300 ppm in lemonade, and 450 ppm in raspberry gelatin. Several deaths resulted. A more recent incident of a chronic poisoning involved the dumping of mining wastes into rice paddies in Japan. Middle-aged women who were deficient in calcium and had multiple pregnancies seemed to be the most susceptible. Symptoms included hypercalciuria; extreme bone pain from osteomalacia; lumbago; pain in the back, shoulders, and joints; a waddling gait; frequent fractures; proteinuria; and glycosuria. The disease was called itai-itai (ouch-ouch disease) as a result of the pain with walking. The victims had a reported intake of 1000 μg/day, approximately 200 times the normal intake of unexposed populations (Yamagata and Shigematsu, 1970). Cadmium exposure has also been associated with cancer of the breast, lung, large intestine, and urinary bladder (Newberne, 1987).

Cadmium absorption is greatest with pulmonary exposure, but it is relatively low (3%–5%) after dietary exposure to humans, although absorption is effected by several factors including nutritional status (eg, low body iron), number of pregnancies, and general health. Cadmium is retained in the kidney and liver in the body, with a very long biological half-life ranging from 10 to 30 years. Cadmium is primarily toxic to the proximal tubular cells of the kidney and tends to accumulate in the cortex of the kidney, eventually leading to renal failure. Bone demineralization occurs with prolonged exposure, as illustrated in the Japanese exposure, which may be secondary to renal dysfunction (EFSA, 2009).

JECFA (2005) investigated cadmium consumption for several countries including the United States and found that the greatest impacts of maximum (allowable) levels (ML) were seen with stem/root vegetables, other vegetables, and molluscs (41%, 68%, and 42%, respectively, when the lowest MLs were used), which represented an insignificant change over a previous intake assessment. Therefore, the current total intake of cadmium was only 40% to 60% of the Provisional Tolerable Weekly Intake (PTWI) of 7 μg/kg bw/week established several years ago, and a slight variation of 1% to 6% due to the use of the proposed new maximum limits is of no significance in terms of risk to human health (JECFA, 2005).

Mercury

Exposure to elemental mercury is relatively rare, although it was once responsible for an occupational disease of hat manufacturers, as elemental mercury was used for the curing of animal pelts. Inhalation of the mercury fumes led to mental deterioration, subsequently named “mad hatter syndrome.” The effects of mercury were more pronounced following the application of mercurial ointments for treatment of syphilis and gave rise to the pun “a night with Venus is followed by a lifetime with Mercury” (Waldron, 1983).

Of interest to food toxicology is the methyl derivative, methyl mercury, formed by bacterial action in an aquatic environment from anthropogenic and natural sources of elemental mercury. Anthropogenic sources include burning of coal (which contains mercury), chloralkali process, and other sources of elemental mercury discharge into aquatic environments. In the case of Minamata, Japan, there was a direct discharge of methyl mercury into the environment. Methyl mercury exposure may cause neurological paresthesias, ataxia, dysarthria, hearing defects, and death. Developmental delays have been documented in children borne of mothers exposed to methyl mercury (Carrington and Bolger, 1992). Other than direct exposure to methyl mercury, exposure usually comes about as the result of methyl mercury becoming incorporated into the food chain. Near the peak of the food chain, methyl mercury becomes concentrated in fish including, bonito (Sarda spp), halibut (Hippoglossus spp), mackerel (Scomberomorus spp), marlin (Makaira spp), shark (all species), swordfish (Xiphias gladius), and bluefin tuna (Thunnus spp). The selection of these species for monitoring and tolerance setting was based on historical data on levels of methyl mercury found in fish consumed in the United States. The selection was also based on an FDA action level of 1.0 ppm in the edible portion of fish (FDA, 2001). However, the allowable level of mercury depends on whether the mercury was “added,” that is, did the presence of mercury arise from an anthropogenic source (ie, was the fish caught in an area known for mercury discharge or was it naturally present in the environment?) (Hutt et al., 2007).

Halogenated Hydrocarbons (Polychlorinated and Polybrominated Hydrocarbons)

Polychlorinated Hydrocarbons

Halogenated hydrocarbons have been with us for some time, and given their stability in water and resistance to oxidation, ultraviolet light, microbial degradation, and other sources of natural destruction, halogenated organics will persist in the environment for some time to come, albeit in minute amounts. However, with the introduction of chlorinated hydrocarbons as pesticides in the 1930s, diseases associated with an insect vector such as malaria were nearly eliminated. In the industrialized world, chlorinated organics brought the promise of nearly universal solvents, and their extraordinary resistance to degradation made them suitable for use as heat-transfer agents and carbonless copy paper, among other uses. As a result of their facile nature, their resulting wide-range uses, and resistance to degradation (and ease of detection), chlorinated hydrocarbons have been found in a wide variety of foods (Table 31-23).

There have been only a few incidents of mass poisonings via food, two of which involved cooking oil contaminated with polychlorinated organics. The first became known as yusho, or rice oil disease, from rice oil contamination originally thought to be confined to polychlorinated biphenyls (PCBs), but has since been revised to include polychlorinated quarterphenyls and polychlorinated dibenzofurans (PCDFs) (Kanagawa et al., 2008). The poisoning occurred in 1968 in Japan and affected approximately 2000 individuals. The most vulnerable were newborns of poisoned mothers. The liver and skin were the most severely affected. Symptoms included dark brown pigmentation of nails; acne-like eruptions; increased eye discharge; visual disturbances; pigmentation of the skin, lips, and gingiva; swelling of the upper eyelids; hyperemia of the conjunctiva; enlargement and elevation of hair follicles; itching; increased sweating of the palms; hyperkeratotic plaques on the soles and palms; and generalized malaise. Recovery requires several years (Anderson and Sogn, 1984; Guo et al., 2003), although more recent authors claim that recovery is gradual even at the 37-year mark, as some patients exhibit symptoms specific to the disease including, but not limited to, hyperglyceridemia, pulmonary disorders, and intractable headache, which were added to the initial diagnostic criteria (Kanagawa et al., 2008). The second incident occurred in 1979 in Yucheng, Taiwan, which also involved PCB-contaminated cooking oil and exposed a similar number of people as the earlier incident in Japan (Guo et al., 2003). The total intakes of PCBs and PCDFs for Yusho and Yucheng victims were estimated to be 633 and 3.4 mg and 973 and 3.84 mg, respectively (Masuda, 1985).

Polychlorinated biphenyls are not formed naturally and were manufactured for use as insulators in electrical machinery, although manufacturing was banned in the United States in 1977. Despite the persistence of chlorinated hydrocarbons in our environment and their reported ubiquity, their low degree of demonstrable toxicity at ambient levels indicates a relatively low risk to humans. Ames et al. (1987) described a method for interpreting the differing potencies of carcinogens and human exposures: the percentage human exposure dose/rodent potency dose (HERP). Using this method, they claimed that the hazard from trichloroethylene-contaminated water in Silicon Valley or Woburn, Massachusetts, or the daily dietary intake from dichlorodiphenyltrichloroethane (DDT) (or its product, dichlorodiphenyldichloroethylene [DDE]) at a HERP of 0.0003% to 0.004% is considerably less than the hazard presented by the consumption of symphytine in a single cup of comfrey herb tea (0.03%) or the hazard presented by aflatoxin in a peanut butter sandwich (0.03%). The FDA’s authority to set tolerances has been used only once in establishing levels for PCBs (21 CFR 109.15 and 109.30).

Polybrominated Hydrocarbons

Polybrominated biphenyls (PBBs) and polybrominated biphenyl ethers (PDBEs) (the latter are also called brominated diphenyl oxides) were used as fire retardants in, among other things, household furniture, airplane seat upholstery, and children’s pajamas, all of which had been identified as sources of preventable disfiguring burns and tragic deaths. Uses also extended to electrical equipment, paint, and plastics, and PBBs have been manufactured globally by several companies.

Appropriately, one of the names under which PPBs were sold was “FireMaster.” Tragedy struck in Michigan in 1973, when FireMaster was placed in bags labeled “NutriMaster” (a magnesium oxide animal feed supplement). Several farms received feed contaminated by FireMaster and feed consumption and milk production dropped by as much as 50%, but the causative agent remained a mystery. As the result of a persistent dairy farmer and more than one serendipitous event (eg, a chromatograph left on by accident, allowing PBBs to elute after a considerable period of time), PBBs were identified as the causative agent. It was found that approximately 650 lbs of PBB had entered the food chain through milk and other dairy products, as well as beef, pork, sheep, poultry, and egg products (Fries, 1985; MDCH, 2011). Over 500 Michigan farms were quarantined, and approximately 30,000 cattle, 4500 swine, 1500 sheep, and 1.5 million chickens were destroyed, along with over 800 tons of animal feed, 18,000 pounds of cheese, 2500 pounds of butter, 5 million eggs, and 34,000 pounds of dried milk products (MDCH, 2011). The incident was eventually publicized in 1981 as a television movie entitled “Bitter Harvest.” PBB-exposed Michigan residents had a variety of complaints including nausea, abdominal pain, loss of appetite, joint pain, fatigue, and weakness, and although a cause–effect relationship could not be established for these symptoms, there is a fairly strong case for development of acne from exposure to PBBs.

Like PCBs, many distinct isomers are possible of PBB and PBDE, but the three main commercial products are pentabromodiphenyl oxide or ether, octabromodiphenyl oxide or ether, and decabromodiphenyl oxide or ether. Since the Michigan incident, the production of PBBs has declined, but with a concomitant increase in demand for PBDEs. Worldwide demand for PBDEs in 2001 was estimated at 70,000 metric tons. These substances are chemically rugged entities and are very resistant to acids, bases, heat, light, reduction, and oxidation—all advantages for their intended use, but disadvantages when discharged into the environment. Interestingly, PBDEs are reported to occur naturally and are produced by Vibrio spp, in association with sponges of the Dysidea genus. Conversely, some environmental remediation may occur with two strains of Pseudomonas isolated from runoff areas of the Michigan manufacturer.

Absorption of PBDEs is limited with some 80% to 90% of a dose eliminated in the feces unchanged; degree of bromination is inversely proportional to absorption, but once absorbed, PBDEs are persistent and bioaccumulating. PBDEs in rat subchronic studies produced effects in liver, kidney, and thyroid in males and females. Liver enlargement, associated with increased metabolism, was reported at doses of 1 to 10 mg/kg bw. DecaBDE at 2.5% and 5% of the diet to mice and rats was carcinogenic. PBDEs are not genotoxic, but are reported to have reproductive effects in rats and rabbits. Data for humans were reported as being incomplete. Because of the limited amount of data, the JECFA committee could not allocate a Provisional Tolerable Weekly Intake (PTWI) (deBoer et al., 2000; JECFA, 2005).

Nitrosmaines, Nitrosamides, and N-Nitroso Substances

Nitrogenous compounds such as amines, amides, guanidines, and ureas can react with oxides of nitrogen (NO_x) to form N-nitroso compounds (NOCs) (Hotchkiss et al., 1992). The NOCs may be divided into two classes: the nitrosamines, which are N-nitroso derivatives of secondary amines, and nitrosamides, which are N-nitroso derivatives of substituted ureas, amides, carbamates, guanidines, and similar compounds (Mirvish, 1975).

Nitrosamines are stable compounds, whereas many nitrosamides have half-lives on the order of minutes, particularly at pH >>6.5. Both classes have members that are potent animal carcinogens, but by different mechanisms. In general, the biological activity of an NOC is thought to be related to alkylation of genetic macromolecules. N-nitrosamines are metabolically activated by hydroxylation at an α-carbon. The resulting hydroxyalkyl moiety is eliminated as an aldehyde, and an unstable primary nitrosamine is formed. The nitrosamine tautomerizes to a diazonium hydroxide and ultimately to a carbonium ion. Nitrosamides spontaneously decompose to a carbonium ion at physiological pH by a similar mechanism (Hotchkiss et al., 1992). This is consistent with in vitro laboratory findings because nitrosamines require S9 for activity and nitrosamides are mutagenic de novo.

NOCs originate from two sources: environmental formation and endogenous formation (Table 31-24). Environmental sources have declined over the last several years but still include foods (eg, nitrate-cured meats) and beverages (eg, malt beverages), cosmetics, occupational exposure, and rubber products (Hotchkiss, 1989). NOCs formed in vivo may actually constitute the greatest exposure and are formed from nitrosation of amines and amides in several areas, including the stomach, where the most favorable conditions exist (pH 2–4), although consumption of H₂-receptor blockers or antacids decreases the formation of NOCs.

Environmentally, nitrite is formed from nitrate or ammonium ions by certain microorganisms in soil, water, and sewage. In vivo, nitrite is formed from nitrate by microorganisms in the mouth and stomach, followed by nitrosation of secondary amines and amides in the diet. Sources of nitrate and nitrite in the diet are given in Table 31-25. Many sources of nitrate are also sources of vitamin C. Another possibly significant source of nitrate is well water; although the levels are generally in the range of 21 μM, levels of 1600 μM (100 mg/L) have been reported (Hotchkiss et al., 1992). However, on the average, Western diets contain 1 to 2 mmol nitrate/person/day (Hotchkiss et al., 1992). Nitrosation reactions can be inhibited by preferential, competitive neutralization of nitrite with naturally occurring and synthetic materials such as vitamin C, vitamin E, sulfamate, antioxidants such as butylated hydroxytoluene, BHA, and gallic acid, and even amino acids or proteins (Hotchkiss, 1989; Hotchkiss et al., 1992).

N-nitrosoproline is the most common nitrosoamine present in humans and is excreted virtually unchanged in the urine. The basal rate of urinary excretion of nitrosoproline, which is claimed to be noncarcinogenic, is 2 to 7 g/day in subjects on a low-nitrate diet (Oshima and Bartsch, 1981).

Epidemiological studies have not provided compelling evidence of a causal association between nitrate exposure and human cancer nor has a causal link been shown between NOCs, preformed in the diet or endogenously synthesized, and the incidence of human cancer (Gangolli, 1999). The cancer risk of nitrite, nitrate, and processed meats is weak and inconclusive, and the risk must be weighed against the health benefits of restoring NO homeostasis via dietary nitrite and nitrate. These benefits should be considered before arriving at any new regulatory or public health guidelines (Milkowski et al., 2010).

Food-borne Molds and Mycotoxins

Molds have served humans for centuries in the production of foods (eg, ripening cheese) and have provided various fungal metabolites with important medicinal uses; they also may produce secondary metabolites with the potential to produce severe adverse health effects, including behavioral changes (Cousins et al., 2005). It is possible that ergot mycotoxins may have exerted a major role in restricting population expansion and only the reduced dependency on rye cereal as the staple food in the sixteenth and seventeenth centuries, arising from the introduction of wheat and potatoes, allowed the steady upward movement in population growth in Europe (CAST, 2003; IFST, 2006a).

Mycotoxins are secondary fungal metabolites (ie, not essential for survival of the mold) secreted into the microenvironment around the mold. Mycotoxins represent a diverse group of chemicals that can occur in a variety of plants used as food, including commodities such as cereal grains (barley, corn, rye, wheat), coffee, dairy products, fruits, nuts, peanuts, and spices. A few mycotoxins also can occur in animal products derived from animals that consume contaminated feeds (eg, milk). However, because commodities are eaten in the greatest amounts, the mycotoxins present in these foods represent the greatest risk (Cousins et al., 2005).

The current interest in mycotoxicosis was generated by a series of reports in 1960–1963 that associated the death of turkeys in England (so-called turkey X disease) and ducklings in Uganda with the consumption of peanut meal feeds containing mold products produced by Aspergillus flavus (Stoloff, 1977). The additional discovery of aflatoxin metabolites (eg, aflatoxin M₁ in milk) led to more intensive studies of mycotoxins and to the identification of a variety of these compounds associated with adverse human health effects, both retrospectively and prospectively.

Moldy foods are consumed throughout the world during times of famine, as a matter of taste, and through ignorance of their adverse health effects. Epidemiological studies designed to ascertain the acute or chronic effects of such consumption are few. Data from animal studies indicate that the consumption of food contaminated with mycotoxins has the potential to contribute to a variety of human diseases (Miller, 1991). Reports of acute intoxications are few; however, prolonged exposure to small quantities of mycotoxin may lead to more insidious effects including growth retardation, birth defects, impaired immunity, decreased disease resistance, and tumor formation in humans and decreased production in farm animals (CAST, 2003).

With some exceptions, molds can be divided into two main groups: “field fungi” and “storage fungi.” The former group contains species that proliferate in and under field conditions and do not multiply readily once grain is in storage. Field fungi may be superseded and overrun by storage fungi if conditions of moisture and oxygen allow.

Importantly, the presence of a toxigenic mold does not guarantee the presence of a mycotoxin, which is elaborated only under certain conditions. Further, more than one mold can produce the same mycotoxin (eg, both Aspergillus flavus and several Penicillium species produce the mycotoxin cyclopiazonic acid) (El-Banna et al., 1987; Truckness et al., 1987). Also, more than one mycotoxin may be present in an intoxication; that is, as in the outbreak of turkey X disease, there is speculation that aflatoxin and cyclopiazonic acid both exerted an effect, but the profound effects of aflatoxin would have overshadowed those of cyclopiazonic acid (Miller, 1989). Although there are many different mycotoxins and subgroups (Table 31-26), this discussion will be confined largely to four of the more important: aflatoxins, trichothecenes, fumonisins, and ochratoxin A.

Organic foods, produced without the use of insecticides and fungicides, may be more susceptible to mycotoxin contamination than foods produced using conventional agricultural practices. For example, higher levels of ochratoxin were found in organic cereal when compared to nonorganic cereal and cereal products in Spain and Portugal (Juan et al., 2008). The UK Food Standards Agency found several organic maize meal products highly contaminated with fumonisin mycotoxins, whereas conventionally produced maize meal products analyzed concurrently had levels below recommended limits (UK Food Standards Agency, 2003). Because European agriculture faces a growing demand by consumers for organic produce, the European Union has established a project called “safe organic vegetables and vegetable products by reducing risk factors and sources of fungal contaminants throughout the production chain” focusing on organic carrots and reduction of alternaria toxins (EU, 2002).

Aflatoxins

Among the various mycotoxins, the aflatoxins have been the subject of the most intensive research because of the potent hepatocarcinogenicity and toxicity of aflatoxin B₁ in rats. Epidemiological studies conducted in Africa and Asia suggest that it is a human hepatocarcinogen, and various other reports have implicated the aflatoxins in incidences of human toxicity (Krishnamachari et al., 1975; Peers et al., 1976; Kensler et al., 2011).

Generally, aflatoxins occur in susceptible crops as mixtures of aflatoxins B₁, B₂, G₁, and G₂, with only aflatoxins B₁ and G₁ demonstrating carcinogenicity. A carcinogenic hydroxylated metabolite of aflatoxin B₁ (termed aflatoxin M₁) can occur in the milk from dairy cows that consume contaminated feed. Aflatoxins may occur in a number of susceptible commodities and products derived from them, including edible nuts (peanuts, pistachios, almonds, walnuts, pecans, and Brazil nuts), oil seeds (cottonseed and copra), and grains (corn, grain sorghum, and millet) (Stoloff, 1977). In tropical regions, aflatoxin can be produced in unrefrigerated prepared foods. The two major sources of aflatoxin contamination of commodities are field contamination, especially during times of drought and other stresses, which allow insect damage that opens the plant to mold attack, and inadequate storage conditions. Since the discovery of their potential threat to human health, progress has been made in decreasing the level of aflatoxins in specific commodities in developed countries. For example, in the United States and Western European countries, control measures include ensuring adequate storage conditions and careful monitoring of susceptible commodities for aflatoxin level and the banning of lots that exceed the action level for aflatoxin B₁.

Aflatoxin B₁ is acutely toxic in all species studied, with an LD₅₀ ranging from 0.5 mg/kg for the duckling to 60 mg/kg for the mouse (Wogan, 1973). Death typically results from hepatotoxicity. Aflatoxin B₁ is also highly mutagenic, hepatocarcinogenic, and possibly teratogenic. A problem in extrapolating animal data to humans is the extremely wide range of species susceptibility to aflatoxin B₁. For instance, whereas aflatoxin B₁ appears to be the most hepatocarcinogenic compound known for the rat, the adult mouse is essentially totally resistant to its hepatocarcinogenicity.

Aflatoxin B₁ is an extremely reactive compound biologically, altering a number of biochemical systems. The hepatocarcinogenicity of aflatoxin B₁ is associated with its biotransformation to a highly reactive electrophilic epoxide, which forms covalent adducts with DNA, RNA, and protein. Damage to DNA is thought to be the initial biochemical lesion resulting in the expression of the pathological tumor growth (IARC, 2002). Species differences in the response to aflatoxin may be due in part to differences in biotransformation and susceptibility to the initial biochemical lesion (Monroe and Eaton, 1987).

Trichothecenes

Trichothecenes are toxic sesquiterpenoid compounds that have an epoxide functionality that is apparently crucial for their toxicity (Desjardins et al., 1993). They include many different chemical entities that all contain the trichothecene nucleus (Ellison and Kotsonis, 1975) and are produced primarily by Fusarium, but also by a number of commonly occurring molds, including Myrothecium, Trichothecium, Stachybotrys, and Cephalosporium. The trichothecenes were first discovered during attempts to isolate antibiotics, and although some show antibiotic activity, their toxicity has precluded their use as therapeutic agents. Fusarium head blight, caused by trichothecene-producing Fusarium species, is a destructive disease of cereal grain crops that has a worldwide economic impact (Fouroud and Eudes, 2009). Consumption of contaminated grain has been associated with intestinal irritation in mammals and can lead to feed refusal in livestock and other toxic responses (Eriksen and Pettersson, 2004). There have been many reported cases of trichothecene toxicity in farm animals and a few in humans. One of the most famous cases of presumed human toxicity associated with the consumption of trichothecenes occurred in Russia during 1944 around Orenburg, Siberia. Disruption of agriculture caused by World War II resulted in millet, wheat, and barley being overwintered in the field. Consumption of these commodities resulted in vomiting, skin inflammation, diarrhea, and multiple hemorrhages, among other symptoms. About 10% of the population was affected and mortality rates were as high as 60% in some counties (Ueno, 1977; Beardall and Miller, 1994), and was subsequently identified as alimentary toxic aleukia (CAST, 2003). Trichothecenes are protein synthesis inhibitors known to bind to ribosomes. The acute LD₅₀s of the trichothecenes range from 0.5 to 70 mg/kg, and although there have been reports of possible chronic toxicity associated with certain members of this group, more research will be needed before the magnitude of their potential to produce adverse human health effects is understood (Sato and Ueno, 1977). The extent of toxicity associated with the trichothecenes in humans and farm animals is poorly understood, owing in part to the number of entities in this group and the difficulty of assaying for these compounds (JECFA, 2001).

Fumonisins

Fumonisins are mycotoxins produced by Fusarium verticillioides (formerly known as F moniliforme) and several other Fusarium species. Corn products contaminated with F verticilliodes are responsible for agriculturally important diseases in horses and swine (ICPS, 2000) and are actively being evaluated to determine how great a threat they pose to public health. Initial evidence of the involvement of F verticilliodes–produced toxins in human disease was reported by Marasas et al. (1988), who found that an increased incidence of esophageal cancer was associated with the consumption of contaminated corn (maize) by humans in a region in South Africa. Fumonisins have been associated with cancer, reproductive toxicity (neural tube defects), and acute disease outbreaks where low-quality corn is consumed on a regular basis (Cousins et al., 2005). Fumonisins target different organs in different species, but the underlying mechanism is a disruption of lipid metabolism by inhibition of ceramide synthetase, an enzyme integral to the formation of complex lipids for use in membranes (ICPS, 2000; IARC, 2002).

Corn borer insect pests cause damage to the developing grain, which enables spores of the toxin-producing fungi, Fusarium, to germinate. The fungus then proliferates, which leads to ear and kernel rot and the production of potentially hazardous levels of fumonisins. Corn varieties, which express the Bt insect control proteins, have been shown to contain significantly reduced levels of fumonisin because the Bt protein significantly reduces the corn borer–induced tissue damage in corn products (Munkvold et al., 1997, 1999; Masoero et al., 1999; Hammond et al., 2004; Papst et al., 2005; Ostry et al., 2010; Folcher et al., 2010).

Ochratoxin A

This mycotoxin is primarily produced by Aspergillus ochraceus, A carbonarius, and Penicillium verrucosum, and human exposure occurs as the result of contamination of small grains (barley, wheat, and corn), coffee beans, and grapes. The effects of ochratoxin A were discovered as the result of feeding the mycotoxin to pigs, who subsequently drank copious amounts of water, urinated near continuously, and exhibited pain in the area of the kidney. Ochratoxin A is nephrotoxic and carcinogenic in mice and rats. Ochratoxin A is absorbed from the GI tract and enters the enterohepatic circulation; it is also absorbed by the proximal and distal tubules of the kidney. It binds tightly to albumin in the blood and can therefore have a very long serum half-life. Epidemiological evidence indicates nearly half the European population is exposed to ochratoxin A and there is an association of endemic nephropathy and renal tumors in humans in parts of Eastern Europe (CAST, 2003).

Ethyl Carbamate

Ethyl carbamate or urethane, the ethyl ester of carbamic acid, was used for many years as an intravenous anesthetic until its mutagenic and carcinogenic properties became known. It has since been classified by the International Agency for Research on Cancer (IARC) as “possibly carcinogenic to humans” (Group 2B) and “reasonably anticipated to be a human carcinogen” by the NTP (2004a). The primary use of urethane is as a chemical intermediate in the preparation of amino resins, with lesser uses as a solubilizer in the manufacture of pesticides, fumigants, and cosmetics and as an intermediate for pharmaceuticals and biochemical research. It is allowed in some anticonvulsant drugs at a level of 1 ppm and is still used as a veterinary anesthetic. Urethane has been found in fermented foods and beverages including liquor, wine, beer, bread, soy sauce, and yogurt. Diethylpyrocarbonate, an inhibitor of fermentation, can form ethyl carbamate.

Ethyl carbamate is easily absorbed and undergoes CYP2E1-mediated metabolic activation to vinyl carbamate epoxide, which binds covalently to nucleic acids and proteins, producing adducts. Ethyl carbamate is a multisite carcinogen with a short latency period. Single doses or short-term oral dosing at 100 to 200 mg/kg BW/day has been shown to induce tumors in mice, rats, and hamsters. Intake estimates from food and alcoholic beverages range from a mean of 0.015 μg/kg BW/day to 0.080 μg/kg BW/day for high-end users. The benchmark dose lower confidence limit as set by JECFA is 0.3 mg/kg BW/day, which yields a margin of safety of 20,000 (JECFA, 2005).

Fluoride

Fluorine, in the form of fluoride, is nearly ubiquitous in nature. Primary human exposure is via drinking water, although it is also present in some foods, notably some teas, vegetables, and marine fish. Exposure also occurs via processed food made with fluoridated water or produce washed with fluoridated water (the so-called ‘halo’ effect). The greatest nondietary source is fluoridated toothpaste (NRC, 2006). Fluoride taken in water has a high degree of bioavailability with an absorption of 90%, whereas fluoride taken in food is approximately 50% absorbed. Consumption of fluoride results in uptake by bone and teeth, where enamel crystallites form fluorhydroxyapatite in place of the naturally formed hydroxyapatite; the former being stronger and more acid-resistant than the latter, resisting and even reversing the initiation and progression of dental caries. Ionic fluoride rarely exists in blood, most is trapped by bone tissue, where new bone growth is stimulated and this mechanism has served as the basis of some treatments for osteoporosis (WHO, 1996).

Fluorosis occurs as the result of high fluoride intake and may be complicated by low calcium intake; it is cumulative and endemic to some areas of the world (eg, China and India). Fluorosis is dose-responsive, producing a range of effects from cosmetic (mottling of teeth) to adverse functionality (skeletal fluorosis). Enamel fluorosis occurs as the result of high fluoride consumption prior to tooth eruption (ie, in children up to the age of eight years, exposed to water with a fluoride content of ≥4 mg/L) and can range from a mild discoloration of the tooth surface to severe (brown) staining and pitting of the teeth to the point of enamel loss (NRC, 2006). In skeletal fluorosis, in the asymptomatic, preclinical stage, patients have an increased bone density. Stage 1 skeletal fluorosis is characterized by occasional stiffness, pain in joints, or some osteosclerosis of the pelvis and vertebra. Stage 2 skeletal fluorosis is characterized by sporadic pain, stiffness of joints, and osteosclerosis of the pelvis and spine, although mobility is not severely affected. In Stage 3 (rarely seen in the United States), there may be crippling, dose-related calcification of ligaments, osteosclerosis, exotoses, osteoporosis of long bones, muscle wasting, and neurological effects due to hypercalcification of vertebrae (at this point, bone ash fluoride may be two to three times that of the bones of normal subjects). Although it is agreed that skeletal fluorosis is the result of prolonged exposure to increased amounts of fluoride, because the incidence of crippling skeletal fibrosis continues to be rare even in geographic areas of high exposure, unidentified intervening metabolic or dietary factors may have rendered skeletons more or less susceptible. Other effects attributed to excess fluoride include lower IQs and decreased thyroid function, increased calcitonin activity, increased parathyroid hormone activity, secondary hyperparathyroidism, impaired glucose tolerance, and possible effects on timing of sexual activity (NRC, 2006). The reports of possible carcinogenic effects of fluoride, including those of the bone (osteosarcoma), are tentative and mixed (NRC, 2006; Douglass and Joshipura, 2006).

Guidelines for fluoridation of the public water supply recommend addition at levels of 0.7 to 1.2 mg/L, to achieve target Adequate Intake (AI) levels based on a 2 L water/day intake by adults, with adjustments in warmer regions where water intake is high in the summer months, or where fluoride occurs naturally at high levels (eg, some areas of Colorado, 11.2 mg/L; Oklahoma, 12.0 mg/L; New Mexico, 13.0 mg/L; and Idaho, 15.9 mg/L). Although the essentiality of fluoride has not been described, an AI has been established for various age groups as a balance between caries resistance and possible fluorosis of teeth. For example, the AI for infants is 0.01 mg/day, for adult females and males is 3 and 4 mg/d, respectively, and there is a range of graduated AIs for intervening age groups (IOM, 1997).

Toxins in Fish, Shellfish, and Turtles

There are a number of marine (seafood) toxins (to be distinguished from marine venoms), many of which are not confined to a single species (over 400 species have been associated with ciguatera toxicity). Some of these toxins occur with sporadic frequency and nonpredictability, indicating an environmental influence and can often be traced to the presence of an algae (including dinoflagellates) or commensurate bacteria. However, some marine toxins appear to be specific to a single genus or species and are therefore innate to that taxonomic group.

Marine Toxins

Ciguatera Poisoning

The “cigua” in ciguatera toxin is derived from the Spanish name for the sea snail Turbo pica in which the symptoms were first reported. Ciguatera and related toxins (scaritoxin and maitotoxin) are ichthyosarcotoxic neurotoxins (anticholinesterase) and are found in 11 orders, 57 families, and over 400 species of fish as well as in oysters and clams. The penultimate toxin (gambiertoxin) is produced by the dinoflagellate Gambierdiscus toxicus, commonly isolated from microalgae growing on or near coral reefs that have ingested the dinoflagellate. The pretoxin appears to pass through the food chain and is biotransformed upon transfer to or by the ingesting fish to the active form, which is consumed by mammals. Other toxins, including palytoxin and okadaic acid, unrelated to gambiertoxin, may be present in ciguarteric fish and may contribute to toxicity. The asymptomatic period is three to five hours after consumption but may last up to 24 hours. The onset is sudden and symptoms may include abdominal pain, nausea, vomiting, and watery diarrhea; muscular aches; tingling and numbness of the lips, tongue, and throat; a metallic taste; temporary blindness; and paralysis. Deaths have occurred. Recovery usually occurs within 24 hours, but tingling may continue for a week or more. The intraperitoneal (i.p.) LD₅₀ of maitotoxin in mice is 50 ng/kg (Bryan, 1984; Liston, 2000).

Palytoxin Poisoning

Palytoxin is produced by the zoanthid soft coral of the genus Palythoa, and fish, crabs, and polychaete worms, living in close association with or eating this mass, may become contaminated with palytoxin. The toxin is not part of the stinging nematocyst of the coral, but may be produced by female polyps and mature eggs of the organism, possibly requiring the presence of symbiotic algae (possibly the dinoflagellate Ostreopsis siamensis). Palytoxin, in various forms, is produced by any number of species, including P tuberculosa in the tropical waters in the Pacific and Japan, P mammilosa and P caribaeorum in the West Indies and Puerto Rico, and in the Bahamas, P vestitus and other Palythoa spp. On occasion, the coral becomes detached from its anchorage and becomes a soft floating mass with a seaweed or moss-like appearance, a very attractive feeding ground for fish. Indigenous peoples of Hawaii knew this as limu-make-o-Hana (the deadly seaweed of Hana) and some are said to have smeared the moss on spear points to enhance their utility as a weapon (Onuma et al., 1999; Tan and Lau, 2000; Tosteson, 2000).

The toxin has been reported in mackerel, parrotfish, and several species of crabs. Victims report a bitter, metallic taste from the meat (most often muscle, liver, ovary, and digestive tract), followed immediately by nausea, vomiting, and diarrhea. Within several hours, symptoms include myoglobinuria, a burning sensation around the mouth and extremities, muscle spasms, dyspnea, and dysphonia. Cause of death may be the result of myocardial injury, although it is known in vitro to be a powerful hemolysin.

Although there are several isoforms and possibly minor toxins associated with palytoxin (depending on the producing species), the predominant action is as a ouabain-sensitive Na⁺K⁺ ATPase inhibitor. Unlike ouabain, palytoxin has no effect on H⁺-, Ca²⁺-, or H+/K+-transporting ATPases. The toxin is quite effective with intravenous LD₅₀ of 0.078, 0.45, 0.033, and 0.089 μg/kg for monkeys, mice, dogs, and rats, respectively. The standard assay is measured in mouse units (MU), the time taken to kill a mouse weighting 20 grams in four hours following intraperitoneal (i.p.) injection of 0.25 mL (Tan and Lau, 2000; Tosteson, 2000).

Abalone Poisoning (Pyropheophorbide)

Abalone poisoning is caused by abalone viscera poison (located in the liver and digestive gland) of the Japanese abalone, Haliotis discus and H sieboldi, and is unusual in that it causes photosensitization; Hashimoto et al. (2010) report the toxin was also found in the midgut glands of cultured scallops (Patinopecten yessoensis) gathered in early spring in Japan. The toxin, pyropheophorbide a, is stable to boiling, freezing, and salting. The development of symptoms is contingent on exposure to sunlight. The symptoms are of sudden onset and include a burning and stinging sensation over the entire body, a prickling sensation, itching, erythema, edema, and skin ulceration on parts of the body exposed to sunlight (Bryan, 1984; Shiomi, 1999). Paralytic shellfish toxin (PST) has been detected in abalone, probably through consumption of the mossworm, a plankton feeder that also clings to seaweed and some shellfish (Takatani et al., 1997).

Pheophorbide b ethyl ester has been isolated from a Chlorella vulgaris dietary supplement (Chee et al., 2008), and pheophorbide a and phyloerythrin were putatively identified as the causative substance in photosensitization and erythematopurpuric eruptions on the skin of patients with a history of consuming chlorella dietary supplements (Jitsukawa et al., 1984). Pyropheophorbide, pheophorbide, and similar phototoxic agents are breakdown products of chlorophyll.

Dinoflagellate Poisoning (Paralytic Shellfish Poisoning or PSP; Saxitoxin)

The etiological agent in dinoflagellate poisoning is saxitoxin or related compounds and is found in mussels, cockles, clams or soft shell clams, butter clams, scallops, and shellfish broth. Bivalve mussels are the most common vehicles. Saxitoxin, originally isolated from toxic Alaskan butter clams (Saxidomus giganteus) is actually a family of neurotoxins and includes neosaxitin and gonyautoxin one through four. All block neural transmission at the neuromuscular junction by binding to the surface of the sodium channels and interrupting the flow of Na⁺ ions; apical vesicles (AV) nodal conduction may be suppressed, there may be direct suppression of the respiratory center and progressive reduction of peripheral nerve excitability. Saxitoxin produces parathesia and neuromuscular weakness without hypotension and lacks the emetic and hypothermic action of tetrodotoxin. Moderate symptoms are produced by 120 to 180 μg/person and are reversible within hours or days, whereas 80 μg of purified toxin per 100 g of tissue (0.5–2 mg/person) may be lethal, due to asphyxiation, usually within 12 hours of ingestion. The toxin is an alkaloid and is relatively heat-stable. The toxin is produced by several genera of plankton (Gonyaulax [now known as Alexandrium] catenella, G acatenella, and G tamarensis, Pyrodinium spp, Ptychodiscus brevis, Gymnodinium catenatum, and others), and during red tide blooms may reach 20 to 40 million/mL. Toxic materials are stored in various parts of the body of shellfish. Digestive organs, liver, gills, and siphons contain the greatest concentrations of poison during the warmer months. Distribution is worldwide (Bryan, 1984; Clark et al., 1999; Liston, 2000). The tolerance for Paralytic Shellfish Poisoning (PSP) for clams, mussels, and oysters is 80 μg/100 g meat (Compliance Policy Guideline, 540.250).

Neurotoxic Shellfish Poisoning

Traditionally limited to the coast of Florida, Gymnodinium breve form red tide blooms containing polycyclic ether toxins called brevetoxins (based on the backbone structure of the molecule, they are generally divided into Type 1 or Type 2), with Type 2 the most often found. Brevetoxins bind to voltage-dependent sodium channels and strength of binding varies with the specific affinity of the toxin and thus the relative potency. Symptoms of Neurotoxic Shellfish Poisoning (NSP) include nausea, tingling, and numbness of the oral area, loss of motor control, and severe muscular ache, all of which resolve in a few days and no deaths have been reported, unlike PSP. An additional route of entry for mammals may result from inhalation of aerosolized toxin as the result of the relative ease of lysis of the unarmored G breve organism during the breaking of waves on the shore. Symptoms of this type of exposure are seen as irritation of the throat and upper respiratory tract. A “kill” of nearly 150 manatees was reported during an unprecedented large outbreak of the toxin, although the specific mode of transmission is uncertain. Human exposure is primarily via consumption of filter-feeding organisms, which may concentrate the toxin (Van Dolah, 2000). An FDA Action Level for NSP in food has been set at 0.8 ppm brevetoxin-2 equivalent (SNIC, 2007).

Amnesic Shellfish Poisoning (Domoic Acid)

Consumption of mussels harvested from the area off Prince Edward Island in 1987 resulted in gastroenteritis, and many older consumers or those with underlying chronic diseases experienced neurological symptoms including memory loss. Despite treatment, three patients (71–84 years old) died within 11 to 24 days. The poisoning was attributed to domoic acid produced by the diatom Nitzschia pungens f multiseries (now called Pseudonitzschia multiseries), which had been ingested by the mussels during the normal course of feeding. Occurrence of domoic acid has also been reported in California shellfish and produced by Nitzchia pseudodelicatissima and in anchovies (resulting in pelican deaths) produced by N pseudoseriata (now called Pseudonitzchia australis). Domoic acid has been reported in shellfish in other provinces of Canada, and in the United States, in Alaska, Washington, and Oregon and may be as frequent as PSP toxins. Domoic acid has also been reported in seaweed. Domoic acid was reported in Japan in 1958 and was isolated from the red algae, Chondria armata.

In the Canadian outbreak, mice injected with extracts (as in the PSP assay) died within 3.5 hours. The mice exhibited a scratching syndrome uniquely characteristic of domoic acid that was followed by increasingly uncoordinated movements and seizures until the mice died. Levels of domoic acid >40 μg/g wet weight of mussel meat caused the mouse symptoms (Canadian authorities require cessation of harvesting when levels approach 20 μg/g.). Mice and rats can generally tolerate 30 to 50 mg/kg. Domoic acid is dose-responsive in humans, with no effect at 0.2 to 0.3 mg/kg, mild GI symptoms at 0.9 to 2.0 mg/kg, and the most serious symptoms at 1.9 to 4.2 mg/kg with GI effects and neurological effects, including dizziness, disorientation, lethargy, seizures, and permanent loss of short-term memory. Although rodents appear to be more tolerant, the fatalities in humans were likely associated with underlying illness. Domoic acid is an analog of glutamine, a neurotransmitter, and of kainic acid; the toxicity of all three is similar, as they are excitatory and act on three types of receptors in the CNS with the hippocampus being the most sensitive. Domoic acid may be a more potent activator of kainic acid receptors than kainic acid itself. The stimulatory action may lead to extensive damage of the hippocampus, but less severe injury to the thalamic and forebrain regions (Todd, 1993; Clark et al, 1999; Van Dolah, 2000). An FDA Action Level for amnesic shellfish poisoning has been set at 20-ppm domoic acid, except in the viscera of Dungeness crab, where 30 ppm is permitted (SNIC, 2007).

Innate Marine Toxins

There are naturally occurring toxins in some species that do not involve marine algae or other environmental influences, but are innate to the particular marine species. The first example is Escolar (Lepidocybium flavobrunneum), and Oilfish or Cocco (Ruvettus pretiosus), a marine fish of the snake mackerel family, which are sometimes sold under the category of “butterfish,” and contain a strong purgative oil, that when consumed can cause diarrhea known as gempylid fish poisoning, gempylotoxism, or keriorrhea (FDA, 2010a). The toxin consists of wax esters (C32, C34, C36, and C38 fatty acid esters), the primary component of which is C34H66O2 (Ukishima et al., 1987); these constitute a substantive portion of the lipid present in these fish (14%–25% by weight). Escolar oil contains >90% wax esters (Nichols et al., 2001). Ingestion of fish containing wax esters in large amounts, coupled with their indigestibility and low melting point, results in diarrhea (Berman et al., 1981). No tolerances have been established, and the FDA recommends avoidance of these fish (Dolan et al., 2010; FDA, 2010a).

A second example of an innate toxin is tetramine found in the salivary glands of Buccinum, Busycon, or Neptunia spp, a type of whelk or sea snail that is distributed in temperate and tropic waters and has long been a food source for humans. These whelks are associated with a heat-stable neurotoxin, tetramine, which upon ingestion by humans causes, among other symptoms, eyeball pain, headache, dizziness, abdominal pain, ataxia, tingling in the fingers, nausea, and diarrhea (Reid et al., 1988; Kim et al., 2009). Power et al. (2002) report that the highest concentration of tetramine is in the salivary gland (up to 6530 μg/g), but varies according to season. Reid et al. (1988) reported levels of 37.5 μg tetramine/g of salivary gland tissue (Reid et al, 1988). Because the whelk is a predator of bivalves, it is assumed the toxin is used for food procurement (Power et al., 2002). Although the FDA recommends removal of the salivary gland to avoid possible intoxication (FDA, 2010b), tetramine is present in other tissues, albeit at lesser concentrations (Anthoni et al., 1989; Dolan et al., 2010).

The third and last example is the meat of the Greenland shark (Somniosus microcephalus) and the related member of the dogfish family, the pacific sleeper shark (Somniosus pacificus), is known to be poisonous to both man and dogs. The causative agent is trimethylamine oxide, which breaks down to trimethylamine in the gut, probably by enteric bacteria. The result is absorption of trimethylamine, which acts as a neurotoxin, producing ataxia in both humans and dogs. However, the flesh may be consumed if boiled several times with changes of water, or as the Inuit people prepare it, by burying it in the ground and allowing the meat to go through several freezing and thawing cycles (Anthoni et al., 1991; Benz et al., 2004; Idboro, 2008; Dolan et al., 2010).

Microbiological Agents—Preformed Bacterial Toxins

Although the United States likely has the safest and cleanest food supply in the world, most food-related illnesses in the United States result from microbial contamination. Food-borne disease outbreaks are tracked by the Centers for Disease Control and Prevention (CDC) in Atlanta, Georgia. The CDC reports that there are approximately 400 outbreaks of food-borne disease per year involving 10,000 to 20,000 people. However, the actual frequency may be as much as 10 to 200 times as high because (1) an outbreak is classified as such only when the source can be identified as affecting two or more people and (2) most home poisonings are mild or have a long incubation time and are therefore not connected to the ingested food, go unreported, and are often felt to be only a “24-hour bug.” Naturally, because of differences in virulence and opportunity, some species are more likely than others to cause outbreaks.

If all the microbiological food-borne health concerns could be divided into two categories—poisonings and infections—the former would include chemical poisonings and intoxications, which may have a plant, animal, or microbial origin. In the infections category, food acts as a vector for organisms that exhibit their pathogenicity once they have multiplied inside the body. Infections include the two subcategories: enterotoxigenic infections (with the release of toxins following colonization of the GI tract) and invasive infections in which the GI tract is penetrated and the body is invaded by organisms.

There are a number of food toxins of microbial origin; however, discussion in this chapter will be limited to preformed bacterial toxins—that is, those toxins elaborated by bacteria concomitant to their residence and growth in or on the food prior to ingestion. Importantly, the bacteria need not be present for the intoxication to take place because the bacteria may have been killed by heat while the toxin survives. Bacterial toxins may be divided on the basis of activity: emetic toxins (ie, Bacillus cereus), which produce their effect by binding to specific receptors in the duodenum, neurotoxins (whose action is self-explanatory), and enterotoxins, which are protein toxins having action on the enteric cells of the intestine. Enterotoxins can be subdivided into cytotoxic enterotoxins which disrupt the cell membrane or other vital functions of the cell and cytotonic enterotoxins, which enter the epithelial cell and cause diarrhea without direct membrane disruption or cell death (Granum, 2006). Bacterial toxins may also be divided on the basis of their origin: an endotoxin is generally a lipopolysaccharide membrane constituent released from a dead or dying Gram-negative bacteria, and these toxins are nonspecific and stimulate inflammatory responses from macrophages including, but not limited to prostaglandins, thromboxanes, interleukins, and other mediators of immunity; exotoxins, which are synthesized and released (usually by Gram-positive bacteria) and are not an integral part of the organism, but may enhance its virulence. Some bacteria, such as Shigella spp, Staphylococcus aureus, or Escherichia coli, can elaborate both endotoxin and exotoxin.

Clostridium botulinum, C butyricum, and C baratti

Food botulism rarely causes illness because the confluence of conditions required for its occurrence—Clostridia in the presence of low acidity, high water activity, absence of preservatives, ambient temperature, and anaerobic environment—such a combination rarely occurs in foods, but botulinum poisoning remains important, the result of its potency (Sobel et al., 2004). All Clostridia are Gram-positive, spore-forming anaerobes. Botulism is a product of the toxins: A (the predominant form in the United States), B (the predominant form in Europe), E (the predominant form in Northern latitudes), and F that may be produced by one or more strains of C botulinum, C butyricum (Type E only), and C baratti; toxins C and D cause botulism in animals. Type G has not caused any human cases. C botulinum toxins are categorized as Group I to IV on the basis of toxin produced; additionally, Group I is proteolytic in culture (liquefying egg white, gelatin, and other solid proteins). The toxin is elaborated in foods, wounds, and infant gut and is neurotoxic, interfering with acetylcholine at peripheral nerve endings. Botulinum neurotoxins induce blockage of voluntary motor and autonomic cholinergic neuromuscular junctions, which prevent motor fiber stimulation. Clinical illness is characterized by cranial nerve palsies, followed by descending flaccid muscle paralysis, which can involve the muscles of respiration. Although ptosis and dysarthria may be mistaken for signs of encephalopathy, patients are fully alert, and the results of a sensory examination are normal. Recovery often takes weeks to months (Sobel et al., 2004). Although the spores are among the most heat-resistant, the toxins are heat-labile (the toxin may be rendered harmless at 80°C–100°C for five–10 minutes). Botulinum toxins are large zinc metalloproteins of ~150,000 Da, composed of two parts, a 50,000-Da piece, the catalytic subunit, and the 100,000-Da piece containing an N-terminal translocation domain and a C-terminal binding domain. The structural features are similar to tetanus toxin. For Types B, D, F, and G (and tetanus toxin), the target protein is vesicle-associated membrane protein (VAMP/synaptobrevin), a protein associated with the synaptic vesicle. Types A and E cleave a protein associated with the presynaptic membrane, ANAP25. Botulinum toxin C cleaves SNAP25 and syntaxin, another protein involved in exocytosis. Although intracellular mechanisms of botulinum and tetanus toxins are similar, symptoms are different because different populations of neurons are targeted. The symptoms may include respiratory distress and respiratory paralysis that may persist for six to eight months. The case fatality rate in the United States is 4% (Sobel et al., 2004) and the poison is fatal in three to 10 days; a lethal dose is approximately 1 ng.

Current methods for detecting botulinum toxin include a mouse bioassay and an enzyme-linked immunosorbent assay. The mouse bioassay is the accepted standard, where the mouse is injected with a lethal dose, the signs of which should develop in eight hours and, if not, the mouse is observed for four days. The mouse bioassay can also be used to differentiate between the toxin types by mixing neutralizing antibodies with the sample, prior to injection. Determining which mice survive following specific combinations of toxin and antisera, determines the specific toxin type. The absolute amount of toxin detected in the mouse bioassay is not well defined but is thought to be 10 to 20 pg/mL for type A (Barr et al., 2005).

Sources and reservoirs for Clostridia include soil, mud, water, and the intestinal tracts of animals. Foods associated with botulinum toxin include improperly canned low-acid foods (green beans, corn, beets, asparagus, chili peppers, mushrooms, spinach, figs, baked potato, cheese sauce, beef stew, olives, and tuna). The toxin also may occur in smoked fish, fermented food (seal flippers, salmon eggs), and improperly home-cured hams. An increasing source of poisonings is from the use of flavored oils or oil infusion, most typically in garlic-in-oil preparations. In 1993, FDA required acidification of such preparations to prevent the growth of Clostridia (FDA, 2005a). Whereas a proteolytic strain of C botulinum (Group I) may cause the food to appear and smell “spoiled” (by-products include isobutyric acid, isovaleric acid, and phenylpropionic acid), this is not the case with nonproteolytic strains, many of which can flourish and elaborate toxin at temperatures as low as 3°C (Loving, 1998; Belitz and Grosch, 1999; Crane, 1999; Lund and Peck, 2000).

The successful use of nitrates in meat to prevent spoilage by C botulinum resulted in the petitioning of FDA by the USDA to have sodium and potassium nitrate approved for use by “prior sanction” (21 CFR 181.33). The mechanism of nitrates is believed to be due to an inactivation by nitric oxide of iron-sulfur proteins such as ferrodoxin and pyruvate oxidoreductase within the germinated cells. The activity is dependent on the pH and is proportional to the level of free HNO₂; 100 mg nitrate/kg of meat is necessary for the antimicrobial effect, although this effect can be enhanced with ascorbates and chelating agents. Other antibacterials that prevent C botulinum include nisin (used in cheese spreads), parabens, phenolic antioxidants, polyphosphates, and carbon dioxide (Belitz and Grosch, 1999; Lund and Peck, 2000).

Clostridium perfringens

Unlike C botulinum, the primary reservoir for C perfringens is the intestinal tract of warm-blooded animals (including humans). Most incidences of C perfringens food poisoning are associated with the consumption of roasted meat that has been contaminated with intestinal contents at slaughter, followed by roasting and inadequate storage, allowing C perfringens growth and enterotoxin Clostridium perfringens enterotoxin (CPE) to be elaborated (although some CPE may actually be released during a “second-sporulation” process in the stomach of the victim). Virtually all food poisoning is produced by type A strain, although a particularly severe form (a necrotic enteritis called “pig-bel” among indigenous peoples of the New Guinea highlands or in Germany known as “Darmbrand”) is produced by type C strain, which has a mortality rate of 15% to 25% even with treatment. The toxin is normally trypsin-sensitive, but people with low intakes of protein or who consume trypsin-inactivating foods (eg, sweet potatoes) are more at risk than carnivorous people with normal trypsin levels (Granum and Brynestad, 1999; Granum, 2006).

CPE is enterotoxic and follows an ordered series of events, first causing cellular ion permeability, followed by macromolecular (DNA, RNA) synthesis inhibition, morphological alteration, cell lysis, and villi tip desquamation and severe fluid loss. This is manifested by abdominal cramping, and diarrhea occurs within eight to 16 hours, although symptoms are of short duration, one day or less. Foods associated with C perfringens poisoning include cooked meat or poultry, gravy, stew, and meat pies. C perfringens is also associated with the production of another 11 toxins, including those associated with gas gangrene (Hobbs et al., 1953; Hauschild, 1971; Walker, 1975; Hobbs, 1976; Crane, 1999; Labbe, 2000).

Bacillus cereus

Bacillus cereus is also a Gram-positive, spore-forming rod, but is an aerobe or facultative anaerobe. Bacillus cereus is a causative agent of emetic or diarrheagenic exo- and enterotoxins elaborated in food. The emetic thermostable toxin (surviving 259°F for 90 minutes) is called cerulide (a small cyclic peptide, 1.2 kDa, that acts on 5-HT₃ receptors stimulating the vagus afferent nerve) and is produced by serotypes 1, 3, and 8; it is also resistant to pH and proteolysis, but is not antigenic. The diarrheagenic thermolabile toxin (surviving 133°F for 20 minutes) is produced by serotypes 1, 2, 6, 8, 10, and 19 and may also be produced in situ in the lower intestine of the host. The diarrheal form may actually consist of three toxins, one of which is hemolytic (Granum, 2006). Reservoirs are soil and dust. Foods associated with this organism and its toxic properties include boiled and fried rice (principally the emetic form), while the diarrheal form has a wider occurrence and may be found in meats, stews, pudding, sauces, dairy products, vegetable dishes, soups, and meat loaf (Goepfert et al., 1972; Gilbert, 1979; Bryan, 1984; Crane, 1999; Granum and Lund, 1997). The foods associated with the two types somewhat reflect the geographic distribution of the types, as the emetic type predominates in Japan, whereas in North America and Europe, the diarrhea type is most often seen.

Evidence is accumulating that other species of Bacillus may elaborate food toxins, including Bt, B subtilis, B licheniformis, and B pumilis (Crane, 1999; Granum and Baird-Parker, 2000; Granum, 2006). A notable exception is B cereus var toyoi, a naturally occurring, nontoxigenic, and nonpathogenic strain. B toyoi has been tested in a variety of systems, including conventional toxicity studies and tests for enterotoxicity and genotoxicity, and determined to be safe for its intended use in animal feed as a probiotic to promote digestive health (Williams et al., 2009). B toyoi has been approved for addition to swine, bovine, poultry, and rabbit feed in the European Communities.

Staphylococcus aureus

Staphylococcal intoxication includes staphyloenterotoxicosis and staphylococcus food poisoning. S aureus produces a wide variety of exoproteins, including toxic shock syndrome toxin-1 (TSST-1), the exfoliative toxins ETA and ETB, leukocidin, and the staphylococcal enterotoxins (SE) (SEA, SEB, SECn,17 SED, SEE, SEG, SHE, and SEI). TSST-1 and the SE are also known as pyrogenic toxin superantigens on the basis of their biological characteristics. There is a relatively wide degree of molecular diversity among SE toxins and this is thought to be the result of adaptation to allow for a broad range of potential hosts (Monday and Bohach, 1999). Some, but not all, SE require Zn⁺⁺ for superantigen activity. Although enterotoxemia only develops from ingestion of large amounts of SE, emesis is produced as the result of stimulation of the putative SE receptors in the abdominal viscera, followed by a cascade of inflammatory mediator release. All the SE toxins share a number of properties: an ability to cause emesis and gastroenteritis in primates, superantigenicity, intermediate resistance to heat and pepsin digestion, and tertiary structural similarity, including an intramolecular disulfide bond. Induction of emesis separates the SE toxins from TSST-1, but the induction of emesis is not directly correlated to superantigen activity (Granum, 2006). The exact link between superantigenicity and lethality by the SE toxins and TSST-1 is not known, but may be dependent upon cytotoxicity for certain cells, possibly in the kidneys, liver, or vascular endothelium (Monday and Bohach, 1999). Sources of Staphylococcus include nose and throat discharges, hands and skin, infected cuts, wounds, burns, boils, pimples, acne, and feces. The anterior nares of humans are the primary reservoirs. Other reservoirs include mastitic udders of cows and ewes (responsible for contamination of unpasteurized milk) and arthritic and bruised tissues of poultry. Foods usually are contaminated after cooking by persons cutting, slicing, chopping, or otherwise handling them, and then keeping them at room temperature for several hours or storing them in large containers. Foods associated with staphylococcal poisoning include cooked ham; meat products, including poultry and dressing; sauces and gravy; cream-filled pastry; potatoes; ham; poultry; fish salads; milk; cheese; bread pudding; and generally high-protein leftover foods (Cohen, 1972; Bryan, 1976, 1984; Minor and Marth, 1976; Crane, 1999; Dinges et al., 2000).

E coli

Although E coli does not produce a preformed toxin, it deserves mention because of the overwhelming publicity the emergent strain O157:H7 has received (H and O refer to flagellar antigens and virulence markers). There are four categories of E coli associated with diarrheal disease: enteropathogenic, enterotoxigenic, enteroinvasive, and Vero cytotoxin–producing E coli (VTEC). The classification VTEC also includes “shiga-like toxin”–producing E coli and “shiga toxin”–producing E coli (STEC). Enterohemorrhagic E coli (EHEC) refers to those strains producing bloody diarrhea and are a subset of VTEC. The reference to shiga toxin is the result of the clinical similarity of the bloody diarrhea caused by EHEC to that caused by Shigellae. Each of the diseases presented by the four categories is also associated with one or more toxins (Willshaw et al., 2000). The symptoms of STEC infections vary, but commonly include severe stomach cramps, diarrhea (often bloody), and vomiting, and about 5% to 10% of those diagnosed with STEC infections develop hemolytic uremic syndrome, a potentially life-threatening complication because the kidneys may stop working (CDC, 2010).

Cattle are a significant reservoir of E coli; therefore, it is logical that most outbreaks in the United States have been associated with hamburgers and other beef products, although raw vegetables (often fertilized with manure) and unpasteurized apple cider and juice have been reported as sources of outbreaks. Outbreaks in Europe are more often associated with contamination of recreational waters (swimming pools, lakes, etc). Other sources of contamination include person-to-person contact (especially in families and among institutionalized persons) and contact with farm animals especially following educational farm visits (Karch et al., 1999).

The subject of organic food has increasingly captured the public interest. As a result, numerous studies have been done comparing the nutritional and health benefits of organic and conventional foods. When currently available literature was reviewed, there was no compelling evidence that there are nutrition-related health effects from the consumption of organically produced foods (Dangour et al., 2010, Gueguen and Pascal, 2010).

Within this issue, there is a debate concerning the use of organic fertilizers (eg, cow manure) in organic and conventional farming, which may contain E coli O157:H7 (Stephenson, 1997). Data reported to the US CDC in 1996, and tabulated in a CDC document entitled “Clusters/Outbreaks of E coli O157:H7 reported to CDC in 1996,” show that approximately 10% of all E coli O157:H7 infections reported that year were from organically grown lettuce, although organic foods apparently account for less than 1% of the total food supply.

At the basis of the potential problem is the use of inadequately treated manure for fertilizer. Human cases of E coli O157:H7 infection have been reported from consumption of contaminated lettuce, potatoes, radish sprouts, alfalfa sprouts, cantaloupe, and unpasteurized apple cider and juice (Karch et al., 1999). Adequate treatment of manure requires composting the manure for a minimum of three months during which the heap must reach a temperature of 60°C and although this may be adequate to kill vegetative pathogens, it will not destroy spore-formers such as Clostridium perfringens or C botulinum. Survival of viruses and protozoa during composting is not known (Anonymous, 1999).

Bovine Spongiform Encephalopathy

Bovine spongiform encephalopathy (BSE) was first identified in Great Britain in 1986. BSE is a neurological disease classified as a transmissible spongiform encephalopathy (TSE) and is similar to TSEs in other species including scrapie (sheep and goats), transmissible mink encephalopathy (ranch-bred mink), chronic wasting disease (CWD) (mule deer and elk), exotic ungulate encephalopathy (captive exotic bovoids such as bison, orynx, kudu), and feline spongiform encephalopathy (domestic cats and zoo Felidae). TSEs among humans include kuru, Creutzfeldt–Jakob Disease (CJD) and “new variant” CJD (nvCJD), and Gerstmann–Sträussler–Scheinker syndrome (to be distinguished from CJD by an earlier onset and that it tends to run in families). There is compelling epidemiological and laboratory evidence of a causal association between the BSE outbreak in cattle in Great Britain and the new human prion disease nvCJD (CDC, 2011).

Clinically, these diseases present neurological deterioration and wasting, with the incubation period and interval from clinical onset to inexorable death determined by the dose of infective agent, its virulence, and the genetic makeup of the victim. The incubation of BSE in cattle is generally four to five years (range of 20 months–18 years) and an interval of 1 to 12 months from presentation of clinical signs to death. Characteristic histological lesions in the brain and spinal cord are vacuolation and “spongiform” changes. BSE fibrils (long strands of host glycoprotein called prion protein [PrP]) in spinal cord preparations may be seen with electron microscopy following detergent extraction and proteinase K digestion. BSE/scrapie tissues with highest infectivity are brain and spinal cord, followed by retina, spleen, tonsil lymph nodes, distal ileum, and proximal colon. The infective agent can be transferred using preparations of neural tissue from infected animals across species barriers. The most effective method of transfer is direct injection into the brain or spinal cord, but transfer has been reported with intraperitoneal injection and oral dosing. Vertical transfer (mother to offspring) has been reported among domestic cattle, and lateral transfer through biting or injury (especially among mink) has also been reported. Indirect transmission of CWD has been reported recently (Miller et al., 2004); CWD of mule deer (Odocoileus hemionus) can be transmitted from environments contaminated by excreta 2.2 years earlier or decomposed carcasses ~1.8 years earlier.

It is generally agreed that the infective agent is likely a variant of scrapie (endemic to sheep) and was transferred to cattle from rendered sheep via inadequately processed meat and bone meal protein supplement. There is strong evidence and general agreement that the outbreak was amplified and spread throughout the UK cattle industry by feeding rendered (contaminated) bovine meat- and-bone meal to young calves. Disputes have arisen about other details of BSE and its relationship to other TSEs and effects in humans because of an expectation of conformation by BSE to historical principles of disease transmission. Recently, the human susceptibility to sheep and goat passaged-BSE prions was evaluated using transgenic mice expressing PrP and evidence was provided suggesting that humans might be equally or more susceptible to sheep or goat BSE agent compared to that of cattle (Padilla et al. 2011).

Responsive to concerns about transmission, new enhanced BSE-related feed bans went into effect in Canada in 2007 and in the United States in 2009. The enhanced bans prohibit most proteins, which include potentially BSE infectious tissues from all animal feeds, pet foods, and fertilizers not just from cattle feed as required by bans instituted in 1997 (CDC, 2011).

The currently most accepted theory is that the infective agent is a modified form of a normal cell surface component known as prion protein PrP^c (α-helix form), which when introduced into an organism causes a conversion of PrP^c into a likeness of itself (ie, the isoform), but then designated as the pathogenic form, PrP* or PrP^sc for scrapie or PrP^res for protease resistant (β-pleated sheet form) (Flechsig and Weissmann, 2004; Frosch et al., 2005). The agent does not possess nucleic acid. The pathogenic form of the protein, PrP*, is both less soluble and more resistant to enzyme degradation than the normal form. The protein is resistant to heat, antimicrobials, ultraviolet-, or ionizing radiation, and is not consistently inactivated with alcohol, formaldehyde, glutaraldehyde, or sodium hydroxide. Phenol and sodium hypochlorite disinfection have had variable success.

Investigators have concluded that the agent in nvCJD and BSE is the same strain and may be the same agent in feline spongiform encephalopathy and exotic ungulate encephalopathy. Although this information might indicate a simple mode of transmission, workers with the highest potential incidence of exposure to BSE or TSE (sheep farmers, butchers, veterinarians, cooks, and abattoir workers) do not have an unusually high incidence of nvCJD (Prusiner, 1991; Collee, 2000). Likewise, hemophilic patients have not reflected an increased incidence of nvCJD, although CJD transmission has been documented as the result of injections of human growth hormone or gonadotrophin (derived from human pituitary gland), implantation of dura mater and corneas, and even infected electroencephalographic (EEG) electrodes and neurosurgical instruments (Prusiner, 1994; Lee et al., 1998; Collee, 2000).

Substances Produced by Cooking or Processing

Heterocyclic Amines

Tolerances cannot be set for contaminants that are produced as a result of an action taken by the consumer because the home is out of the jurisdiction of FDA. An example of this type of contaminant is heterocyclic amines (HCAs), which are generated during cooking. HCAs were discovered serendipitously by Japanese investigators who, while examining the mutagenicity of smoke generated by charred foods, found that the extracts of the charred surfaces of the meat and fish were quantitatively more mutagenic than could be accounted for by the presence of polycyclic aromatic hydrocarbons (Sugimura et al., 1989). Collectively, there are more than 20 HCAs. They are formed as a result of high-temperature cooking of proteins (especially those containing high levels of creatinine) and carbohydrates. Normally, as a result of such heating, desirable flavor components are formed, for example, pyrazines, pyridines, and thiazoles. Intermediates in the formation of these substances are dihydropyrazines and dihydropyridines, which in the presence of oxygen form the flavor components; however, in the presence of creatinine, HCAs are formed (Table 31-27) (Chen and Chiu, 1998; Schut and Snyderwine, 1999).

These substances are rapidly absorbed by the GI tract, are distributed to all organs, and decline to undetectable levels within 72 hours. HCAs behave as electrophilic carcinogens (Table 31-28). They are metabolized first by N-hydroxylation followed by further activation by O-acetylation or O-sulfonation to react with DNA. DNA adducts are formed with guanosine in various organs, including the liver, heart, kidney, colon, small intestine, forestomach, pancreas, and lung. Unreacted substances are subject to phase II detoxication reactions and are excreted via the urine and feces. In vitro, HCAs require metabolic activation, with some requiring O-acetyltransferase and others not. Although much of the mutagenicity testing has been carried out in TA98 and TA100, these substances are mutagenic in mammalian cells both in vitro and in vivo, Drosophila, and other strains of Salmonella (Munro et al., 1993; Skog et al., 1998; Sugimura and Wakabayashi, 1999).

Acrylamide

Prior to 2002, when Swedish investigators detected acrylamide in food (Table 31-29), it was of interest only to specialists in worker safety, as this chemical is an important intermediate in the manufacture of polyacrylamides. Although there are many industrial and manufacturing uses of polyacrylamides, the bulk of production are used as chemical flocculants for water treatment, oil recovery, and in construction of dam foundations, tunnels, and sewers—consumer exposure is largely incidental (NTP, 2004b). End users are exposed to polyacrylamide, which is not toxic as long as the monomer is not present. Acrylamide (the monomer) was known to be a neurotoxin, creating morphological changes in peripheral nerves at doses as low as 1 mg/kg BW/day. Much more is now known about acrylamide and its primary metabolite, glycidamide, which is produced from acrylamide by the enzyme CYP2E1; both acrylamide and glycidamide will form adducts with hemoglobin. Acrylamide is absorbed rapidly and extensively (23%–48% of the administered dose to rodents) from the GI tract. Both acrylamide and glycidamide are largely eliminated as mercapturic acid conjugates. Repeated dosing of acrylamide (in drinking water at 21 mg/kg BW/day for 40 days) produces morphological changes in the brain areas critical for learning, memory, and other cognitive functions (ie, cerebral cortex, thalamus, and hippocampus) (JECFA, 2005). Acrylamide was classified as “probably carcinogenic to humans” (IARC Group 2A) by the IARC (IARC, 1994) and “reasonably anticipated to be a human carcinogen” by the NTP (2004b).

Acrylamide is formed in foods that are high in carbohydrate, but low in protein, which are subjected to processing temperatures of at least 120°C. These high-temperature processing conditions are largely the same as required for the Maillard reaction, which imparts a toasted, or baked (ie, “crust”), flavor to breads, toast, and other baked goods, breaded meats and vegetables for sautéing or frying, and production of French fries and potato chips. Most acrylamide is formed in the final stages of baking, grilling, or frying as the moisture content of the food falls and the surface temperature rises (JECFA, 2005). The presence of ammonium bicarbonate as a leavening agent increases the formation of acrylamide.

A critical element is the presence of asparagine, and amino acids competing with asparagine in the Maillard reaction, which reduces the levels of acrylamide in the final product. Strategies for mitigation focus on reduction of asparagine through use of asparaginase, breeding and selection of low-asparagine plants, and prolonged yeast fermentation; alternatively, processing temperature could be lowered (JECFA, 2005).

Acrylamide intake estimates range from 0.3 to 2.0 μg/kg BW/day for the average population, with the 90th to 97th percentile at 0.6 to 3.5 μg/kg BW/day and the 99th percentile at up to 5.1 μg/kg BW/day. Primary sources include french fries (1%–30%), potato chips (6%–46%), coffee (13%–39%), pastry and cookies (10%–20%), and bread and rolls/toasts (10%–30%). The NOEL for morphological change in nerves and for reproductive effects is 200 and 2000 μg/kg BW/day, respectively (JECFA, 2005).

Furan

Furan was once known only as an industrial chemical intermediate in the synthesis of polymers used to prepare temperature-resistant structural laminates and to prepare copolymers used in machine dishwashing products. Furan has recently been found to occur in a number of foods that undergo heat treatment, such as canned and jarred foods, including baby food. It is considered by IARC (1995) to be possibly carcinogenic to humans. According to the NTP, furan is hepatotoxic and shows clear evidence of carcinogenicity in both sexes and both species of mice and rats (NTP, 1993). Furan is produced in a variety of experimental systems, including heating of sugars (eg, glucose, lactose, fructose, xylose, rhamnose), heating sugars in the presence of amino acids or protein (eg, alanine, cysteine, casein), and thermal degradation of vitamins (ascorbic acid, dehydroascorbic acid, thiamine) (FDA, 2004).

Furan has been found in a small number of heat-treated foods, including coffee, canned meat, baked bread, cooked chicken, sodium caseinate, filberts (hazelnuts), soy protein isolate, hydrolyzed soy protein, rapeseed protein, fish protein concentrate, and caramel (FDA, 2004). Very little information has been developed on furan levels in food.

Miscellaneous Contaminants in Food

Despite normal precautions taken to protect ourselves from known toxins, some toxins have a propensity to appear unexpectedly in unfamiliar places. Cases in point here include honey poisoning, as documented by Xenophon, who when describing the “Retreat of the Ten Thousand” from Asia back to Greece, the soldiers entered an area rich in honeycombs and seized upon the food. Xenophon noted that the soldiers then “went off their heads” appearing to first be drunk, but then in a state a delirium—a description fitting what we know today about grayanotoxins from mountain laurel and other species.

Mountain laurel (Kalima spp), rhododendron, and azaleas (Rhododendron spp) all possess grayanotoxins of which there may be as many as 60, but the most potent are grayanotoxins I and III (Gunduz et al., 2008). The toxins are present in the shoots, leaves, twigs, and flowers. Honey made from flowers of these plants is toxic to humans, and after an asymptomatic period of four to six hours, salivation, malaise, vomiting, diarrhea, tingling of the skin, muscular weakness, headache, visual difficulties, coma, and convulsions occur. Symptoms are proportional to dose, and atropine administration is indicted as a primary treatment modality (Gunduz et al., 2008). Life-threatening bradycardia and arterial hypotension may occur. Needless to say, beekeepers maintain apiaries well away from these species of plants.

A similar poisoning can occur with oleander (Nerium oleander and N indicum), where honey made from the flowers, meat roasted on oleander sticks, or milk from a cow that eats the foliage can produce prostrating symptoms. The oleander toxin consists of a series of cardiac glycosides: thevetin, convallarin, steroidal, helleborein, ouabain, and digitoxin. Sympathetic nerves are paralyzed; the cardiotoxin stimulates the heart muscles similar to the action of digitalis, and gastric distress ensues (Anderson and Sogn, 1984; VonMalottki and Wiechmann, 1996). Although certainly as possible as the story captured by Xenophon, the stories about scouts sickened from roasting hot dogs on oleander sticks have not been documented and are likely apocryphal.

Closer to home and prevalent in the Midwest in the 18th and 19th centuries, was a mysterious scourge called “milk sickness” also known as “puking fever,” “sick stomach,” “the slows,” and “the trembles”; thousands of people have been reported as dying, including Abraham Lincoln’s mother, Nancy Hanks Lincoln. In humans, milk sickness is characterized by loss of appetite, listlessness, weakness, vague pains, muscle stiffness, vomiting, abdominal discomfort, constipation, foul breath, and finally, coma. For many years, the origin of milk sickness was unknown because there was nothing comparable in Europe (origin of most of the pioneers) and the outbreaks were sporadic. It was not recognized until the late 19th and early 20th century, that white snakeroot (Ageratina altissima née Eupatorium rugosum) and rayless goldenrod (Bigelowia spp, Haplopappus heterophyllus, and Isocoma pluriflora), when eaten by cattle, were the source. The sporadic nature of outbreaks became clear when it was realized that cattle would consume these plants in over-grazed pasture or in years of drought; additionally, the toxin levels in plants can vary considerably, making identification of the source of poisonings difficult. Tremetol or tremetone is the toxic agent and consists of a mixture of sterols and derivatives of methyl ketone benzofuran. The three major benzofuran ketones are tremetone, dehydrotremetone, and 3-oxyangeloyl-tremetone (Panter and James, 1990; Lee et al., 2009; NPS, 2010; Dolan et al., 2010).

Foods are regarded as such because they are edible. Therefore, they cannot be unpalatable or toxic, and foods must have nutritional, hedonic, or satietal value, otherwise there would be no point in consuming them. How then, could a food be toxic and still be considered a food—there are two principal means: (1) an ordinarily nontoxic food has become toxic (through some act of man or nature), if even for a small subpopulation (eg, allergy, intolerance); or (2) overconsumption of a food not ordinarily considered toxic at historic levels of use. This shift between toxic and nontoxic or toxic only for a select group would have had the potential for making the role of regulators charged with protecting the health of the public nearly impossible; however, the FD&C Act provides for some thoughtful and pragmatic solutions for achieving a balance of acceptable risk and unavoidable circumstances through the creative use of specifications, process and manufacturing controls, action levels, tolerances, warning labels, and outright prohibitions.

In the end, it is important to emphasize that the vast majority of food-borne illnesses are attributable to microbiological contamination of food and arise from the pathogenicity and/or toxicity of the contaminating organism. Thus, the overwhelming concern for food safety, in the United States and elsewhere, remains directed toward preserving the microbiological integrity of food. On a planet where for many people, food safety and food security is a daily challenge, FDA and other regulatory agencies have played a significant role in assuring a remarkably safe and available food supply in the United States.