Dietary Supplements: A Framework for Evaluating Safety (2005)

It is scientifically acceptable and appropriate to use information about safety concerns of related substances to inform a decision about the risk associated with a dietary supplement ingredient, especially in the absence of information about the activity of the ingredient in question in humans, animals, or in vitro experiments.¹ Information about substances related to the dietary supplement ingredient of interest may be helpful when predicting risk in one of the following ways:

• Chemical relatedness—similarity to known toxic chemicals or presence of constituents similar in structure to known toxicophores. Chemical structures associated with potential adverse effects;

• Taxonomic relatedness—similarity to known toxic plant species, genus, or family; and

• Functional relatedness—the dietary supplement ingredient of interest is related to another substance because they share a common biological target or mechanism of action that is clearly tied to a toxic effect. This includes endogenous substances and mimetics of endogenous substances when the effect of increasing the amount of an endogenous substance is linked to an adverse health effect.

The value and utility of these types of information, taken together, to predict risk depends on the type of dietary supplement ingredient that is being considered. Cause for concern with a botanical dietary supplement may be derived from information about risk associated with known chemical constituents, as well as information about risk associated with related toxic plants. Similarly, information about the potential risk of dietary supplements that are pure single chemical compounds may be derived by reviewing a list of known risk-associated chemical compounds and chemical moieties (toxicophores) that raise concern of safety. However, for information about what might occur following ingestion of substances that are normally present in the human body (endogenous substances), it is helpful to understand what the substances do in the body at normal concentrations and to understand their mechanisms of action well enough to shed light on what might occur if the normal concentrations are exceeded. Certainly, for particular dietary supplement ingredients, such information could be more useful than reviewing a list of unrelated toxic chemical structures or substances that are not endogenous. Finally, especially when dietary supplements have undefined chemical composition² but information about biological activity is available, it may be helpful and it is appropriate to consider whether the exhibited biological activity is the basis for safety concerns of other substances that are considered potentially harmful. Provided below are guiding principles and further descriptions of the different types of “relatedness” information, including discussion of when and why it is appropriate to use this type of information and specific questions that may help in extrapolating the most useful information.

It is well known that plants produce secondary metabolites with biological activities in mammals, and that plant toxicities are due to chemical constituents in plants. Indeed, the rationale for the use of botanical dietary supplements is that they are likely to affect human function. The challenge in assessing risk in the use of dietary supplements is to establish whether the plant compounds present a hazard to humans and, if so, whether the conditions of use suggest risk.

Risk is always considered a function of two factors: hazard and exposure. In the case of botanical ingredients, hazard relates to the presence of biologically active metabolites produced naturally by biosynthetic processes within the plant. In contrast, exposure may be a consequence of the amount of any particular substance produced by the plant, its concentration or dilution during manufacture, and user intake level and bioavailability (see Chapter 3). Thus consumption of a botanical containing a high level of potentially dangerous bioactive substances, consumed at high dosages or for prolonged periods, will significantly increase risk.

It is possible to make educated estimates of the potential hazard of any given botanical through consideration of the types of biologically active compounds that may be present in the plant (constituents of concern) and the nature of the plant (taxonomic relationships). The goal is to consider two likely scenarios that could provide some guidance regarding the possible toxicity of a botanical dietary supplement ingredient (1) where a known constituent of the plant is, or is structurally similar to, a known toxic compound; and (2) where a plant genus or species is, or is closely related to, a plant known to be toxic. When there is evidence that a botanical is taxonomically related to known poisonous plants and that particular constituents are established as having deleterious effects, the convergence of these factors compels detailed consideration of the potential risks associated with the use of the ingredient.

Information about the potential biological activity of a plant-derived dietary supplement ingredient is obtained by reviewing information about the plant’s individual chemical components to determine if any of the constituents raise concerns. Given that related plants have related chemical composition, with more closely related plants generally having more similar chemical constituents, it is therefore also appropriate to consider the activity of other plants in the same plant family or genus to predict composition and potential toxicity.

This approach of considering the taxonomic relatives of the dietary supplement ingredient has its limitations, however. That is, not all genera of a given family will contain similar toxic components. Furthermore, the concentration of potentially dangerous compounds in the final product will be affected by the plant part being utilized and the manner of preparation, processing, and formulation, as well as by growth conditions that can produce variation in chemical constituents (e.g., climate, season, soil conditions).

Secondary metabolites of plants are generally low-molecular-weight compounds (~ < 1000 Daltons), originally thought to be biosynthesized by the plant primarily for purposes other than basic nutritional and metabolic requirements for normal growth and reproduction (Harborne, 1993). When produced, these metabolites confer “fitness” on the plant, enabling it to respond to and counteract external influences, such as competition for resources, environmental stresses, herbivory, and microbiological or viral attack. The biosynthetic mechanisms by which certain of these compounds (phytoalexins) are produced may essentially shut down unless there is an external stimulus that triggers their production for defensive purposes (Fong, 2002; Harborne, 1993). Other compounds may always be present because evolutionary pressures have established their necessity.

Known chemicals and classes of chemicals that are botanical constituents and warrant concern for safety are listed in Box 6-1, a list generated largely from consideration of plant genera of concern identified in the next section of this chapter. Some of the mechanisms of these compounds, as well as information about plants containing them, are described in the discussion of plant families in Appendix C. (Other compounds or classes of compounds act through mechanisms that are only theoretical or are not understood.) Appendix C describes how some of these compounds are ingested in conventional foods where the amounts ingested are limited or are in different forms due to processing (e.g., cooking). This list is not intended to be all inclusive, but rather to highlight some of the compounds that may result in adverse effects from ingesting plants. Some of these compounds cause more serious deleterious effects than others and some compounds are more potent than others. It is suggested that this list be taken as a general guideline helpful to the Food and Drug Administration (FDA) in determining which botanical substances may warrant higher priority attention. Evidence that one or more of these chemical constituents is

present in a botanical dietary supplement should be considered as an indicator of increased concern for potential toxicity of the specific botanical product, except when consumed as constituents of conventional foods, unless additional information mitigates concern. Further investigation may result in mitigated concern if it is found that circulating concentrations of constituents resulting in adverse effects are substantially lower than circulating concentrations reached with dietary supplement ingestion or if quality animal toxicity studies suggest that the effects are unlikely to occur from the amounts or preparations ingested as dietary supplements. As indicated, some of the substances listed are classes of compounds rather than individual chemical constituents. In this case, some members of a given class may be of less or no concern (see also Appendix C), as will be uncovered by a search of the available literature. For example, a literature search may reveal conclusive evidence that specific structural features required for toxicity are not present for some members of a given class.

Of all classes of botanical toxic compounds, those classified as alkaloids predominate in causing concern because a large proportion have been associated with biological activities and/or toxic effects in mammals (Harborne, 1993; Seawright et al., 1985). Thus particular attention is warranted for dietary supplement ingredients containing alkaloids. Although most chemists recognize and agree on whether a particular compound is an alkaloid, there has been considerable discussion as to how to define such compounds simply because they do not conform to a single structural type. The most workable definition is probably that of Pelletier (1983), which states that “an alkaloid is a cyclic organic compound containing nitrogen in a negative oxidation state which is of limited distribution among living organisms.”³ This definition excludes simple amines, amino acids, peptides, proteins, nucleic acids, and nucleotides, which are ubiquitous, as well as nitro compounds such as aristolochic acid, in which the nitrogen is not in a negative oxidation state. It is particularly noteworthy that the definition does not carry a requirement for pharmacological activity. This is appropriate because many newly isolated alkaloids may not have been tested, and even those of long standing will not have been evaluated for each and every type of activity. Nevertheless, alkaloid-containing plants should always be suspected of being capable of pharmacological activity and should be considered as risk factors.

In addition to known chemicals highlighted in Box 6-1, other structurally related chemical constituents should also be of concern, unless there is convincing information suggesting that particular structural features are required for toxicity and these are not present. It is important to note that if a botanical is known to contain a chemical constituent that is structurally related to a chemical that is regulated (e.g., as a drug), this is a reason for concern and should be investigated. It is not possible to specifically define all the ways that different chemicals may be related, but this concept can be illustrated with an example. Substances with similar chemical structures, such as ephedrine and amphetamine, are structurally related, and substances that stimulate or inhibit activity at the same cellular receptors or other biological targets are “functionally” related (see discussion later in this chapter). Similarity of dietary supplement ingredients to biologically active metabolic intermediates, such as cytokines or hormones, may also be important if the actions of metabolic intermediates provide clues about the activity of a dietary supplement ingredient; this concept is discussed in the “endogenous substances” section.

Frequently, information about the chemical constituents or the distribution of chemical constituents throughout a plant used to make a dietary supplement will not be complete. In this case, it will be helpful and appropriate to consider whether a botanical is related to plants that are of concern.⁴ The system of naming, ranking, and classifying plants and other

organisms based on morphology is now being guided by analysis of metabolites and molecular genetics. There is no doubt that, as progression is made to an increasing degree of specialization through the hierarchy sequence of family, genus, species, subspecies/variety/cultivar, and plant organ, there will be a corresponding increase in congruence, not only in physical appearance, but also in the nature of secondary metabolites produced and sequestered by the plant. Therefore, as any group of plant species becomes more closely related, the compounds biosynthesized will become more similar in both structural types and specific constituents. Thus, in summary, evidence that a botanical bears a close taxonomic relationship to known toxic plants should be used to evaluate potential human risk in the absence of scientific information that such data are not relevant. (See Box 6-2 for a summary of questions to be asked and Box 6-3 for notes on botanical nomenclature.)

Plants in the same genera will not necessarily produce compounds with exactly the same structure, but they are likely to produce the same structural classes of compounds. For example, different species of the genus Senecio (Asteraceae), in spite of being widely distributed in many parts of the world and growing under vastly different climatic conditions, are invariably found to contain pyrrolizidine alkaloids on phytochemical examination (Hartmann and Witte, 1995). Since chemical structure and biological activity are intimately related, novel pyrrolizidine alkaloids should be assumed to possess at least some degree of the hepatotoxic activity established for the most common members of this class (Hartmann and Witte, 1995) if information to prove otherwise is not available. (In this case, data suggest that hepatotoxicity of pyrrolizidine alkaloids depends on unsaturated 1,2 bonds in one of the rings [Hartmann and Witte, 1995].) For the purposes of this framework, taxonomic classification helps in identifying plants that are likely to have similar chemical components. Therefore, much information can be gained by reviewing what is known about plants that are taxonomically related to the dietary supplement ingredient under consideration.

The chemical composition of a given plant species can vary depending on the conditions under which it was grown. However, it is rare for a chemical compound to be observed in one specimen of a species, but not in another specimen of the same species, except due to artifactual differences in analysis techniques. It is more likely that differences in the levels of particular compounds will be observed (Fong, 2002; Harborne, 1993). This is because the array of phytochemicals that a given species may contain is under genetic control; thus each plant has the potential to create the same range of phytochemicals. While the environment and growth conditions may impact phytochemicals found in a given plant, plants of a species known to contain harmful phytochemicals under some conditions should

be assumed to have them even if grown under different conditions. Analyses that suggest specimens of a given plant species do not contain a hazardous phytochemical usually associated with the plant should be carefully considered to ensure that the analysis techniques are appropriately sensitive.

The presence of toxic compounds has been traditionally associated with a number of plant genera and families (e.g., Liliaceae are known to contain cardiac glycosides, Euphorbiaceae are known to contain phorbol esters and toxic diterpenes). The ability to anticipate the presence of specific classes of compounds based on plant family and genus knowledge may be helpful in predicting potential toxicity. Table 6-1⁵ highlights some of the plant genera to which FDA may want to give attention. Also important are the nuances of information about each plant family, which are discussed in Appendix C. It is important to note that this table is not intended to provide a complete reference or to be inclusive, but it serves to provide FDA with a starting point of plant genera that warrant concern. The primary difficulty in using information about related plants to infer information about the toxicity of a particular plant arises when the family encompasses both valuable food plants and species capable of producing toxic compounds (see discussion of traditional use as a food plant below and specific examples in Appendix C).

There are a number of considerations that may mitigate or exacerbate concerns raised by the taxonomic relationship of a dietary supplement ingredient to a hazardous botanical or knowledge that a botanical contains chemical constituents of concern. These are described here, followed by discussion of how these and other factors should impact the use of historical consumption information as a mitigator of concern.

Chemical compounds are differentially distributed in various parts of plants. When secondary metabolites are biosynthesized for the purpose of

providing a protective function within the plant, they tend to be concentrated in young, tender leaves, shoots, and roots, or in reproductive structures (e.g., flowers and seeds). For example, livestock poisoning episodes have shown that there is often a bimodal distribution of toxic hazard with very young plants and plants at the reproductive stage being toxic, whereas at other growth stages no problems occur. (For a more complete discussion, see Appendix C and resources listed.) Frequently, compounds are continuously biosynthesized in a particular part of the plant, such as mature leaves where photosynthesis is at a maximum, but then they are transported and accumulated in other organs where the protective function conferred by such substances is required (Harborne, 1993).

Although it is possible for plants to contain completely different chemical entities in different parts, it is generally more likely that they will contain the same compounds or compounds that have undergone relatively minor structural transformations. The situation with respect to structural types of constituents is often under a state of continuous flux in response to environmental conditions and ecological factors. It is therefore appropriate to assume, in the absence of other information to the contrary, that a plant part marketed as a dietary supplement ingredient contains toxins that are found in other parts of the plant. That is, if a toxic chemical is present in one part of the plant, it will generally be present in the other parts of the plants, even if at lower concentrations. Indeed, concentrations of toxins

may vary and thus be less problematic in some plant parts, but the assumption should be that all parts of a plant containing toxins pose a risk unless there is credible evidence suggesting that dangerous levels of toxins are not present in the part marketed as a dietary supplement. In this case, selection of plant material at a specific growth stage to avoid incorporation of potentially toxic plant parts is desirable.

In addition to concentration of toxic compounds in particular plant parts, levels of toxins in plants may also be influenced by growth stage, time of collection, environmental stress, herbivory, and a multitude of other factors (Fong, 2002). Blending of plant material from a number of locations will tend to dilute toxin levels that are higher in some plants if other plants are lower in toxin levels. However, in the absence of comprehensive studies, it is not possible to delineate precisely the overall influence of such conditions on constituent levels, although their role must be recognized in evaluating the safety of dietary supplements. When sporadic adverse incidents occur in association with a botanical supplement ingredient with no previous indication of risk, it may well be possible that environmental changes have resulted in an increase in levels of toxic constituents. If a plant’s content of a hazardous phytochemical varies significantly with environ-

mental and growth conditions, then it is appropriate to consider the plant’s use in dietary supplements as a risk unless quality control or other actions are implemented to ensure that levels of compounds associated with risk are not reached in raw materials or finished products.

When considering the risk associated with a dietary supplement, it is important to consider whether the method of preparation is likely to concentrate toxic constituents or otherwise increase the consumption or bioavailability of toxic compounds. Toxic plant constituents that are normally present below a given threshold of toxicity can be concentrated by a variety of processing methods. Some methods of preparation may make specific toxins more readily available or even concentrate them (see also discussion of relevance of historical use in Chapter 4). An additional risk is that there may be a tendency to consume more of the plant material in an encapsulated form than if it were consumed in its “native” form.

Some methods of extract preparation can raise levels of toxic constituents to levels of greater concern. Whereas preparation of teas (i.e., aqueous infusions) is a method designed to concentrate specific constituents, many low-molecular-weight phytochemicals⁶ are not particularly water-soluble,⁷ and exceptionally high levels are not likely to be attained. Furthermore, hydrolytic changes may occur that can detoxify or reduce the levels of toxic compounds. It is also difficult to consume large volumes of teas. In contrast to aqueous extracts, extraction of plant material with alcohol or aqueous alcohol in which low-molecular-weight compounds are generally very soluble will likely concentrate toxic components many fold. Thus, such extraction of botanicals that may contain hazardous constituents should be a cause for concern unless there is credible evidence to the contrary. Such procedures underlie the process by which natural products chemists isolate specific bioactive substances from plant material for further purification and identification. Given the potential for extraction to impact the constituents consumed, it is not appropriate to assume that safety of an aqueous extract of a botanical indicates that alcoholic extracts (or other organic solvent extracts) of the same botanical are not of concern.

Certain plant families, while being major sources of food plants, also contain some of the most toxic plant species (see discussion in Appendix C). Species of Fabaceae and Solanaceae that are toxic have been established (probably a reflection of extensive investigations of the secondary metabolite compositions of these families because they are primary food sources), but even some species that contain toxins are used as food plants. For example, plants in the following families are used as foods: Liliaceae (e.g., onion, garlic, asparagus), Apiaceae (e.g., carrots, celery, parsnips), Brassicaceae (e.g., green leafy vegetables), Fabaceae (e.g., peas and beans), and Solanaceae (e.g., potatoes, tomatoes, eggplant). In most cases, the toxin tends to be concentrated in specific plant parts (e.g., potato sprouts) that are not consumed as conventional foods (Drager et al., 1995). Levels in plant parts conventionally eaten as foods are generally known and, in some cases, regulated (e.g., steroidal alkaloids in potatoes); when present, the levels are sufficiently low that they can be metabolized and excreted without adverse effects.

Some general principles apply when considering the relevance of historical or traditional use of a plant. It is important to consider the points outlined in the previous section, particularly the importance of plant part and whether the dietary supplement ingredient preparation will allow excessive amounts to be consumed or will concentrate toxins (in capsules vs. teas, or alcoholic vs. aqueous extracts, for example). A dissimilar amount of ingested toxin could result from changes, such as a switch from cooked to raw consumption, inclusion of plant parts not traditionally consumed, excessive use of one particular vegetable foodstuff, preparation of infusions or extracts, or ingestion of plant preparations historically only applied externally. For example, plant foods are typically consumed after cooking by either dry heat or boiling in water, a process that often destroys toxins because they are thermally labile, hydrolyzed, or extracted.

That particular concern should arise when nontraditional or excessive levels of plant parts are consumed can be illustrated by several examples. Increased consumption of potato skins without the flesh, because nutrients are generally considered more concentrated in this part, may lead to ingestion of high levels of glycoalkaloids and glycosidase-inhibitory calystegines known to be concentrated in the skin, which can seriously affect digestive processes (Asano et al., 1997). Similarly, excessive use of vegetables of the family Brassicaceae can result in toxic effects due to isothiocyanates and other hydrolytic products from glucosinolates, which can cause goiter and a general inhibition of iodine uptake by the thyroid. Internal consumption of plants historically used externally, as with comfrey (Symphytum spp.), a pyrrolizidine alkaloid-containing plant, is especially relevant. Medieval

herbalists prescribed the use of comfrey as a poultice, but more recent use as a salad vegetable or tea has resulted in evidence of liver damage (Coulombe, 2003).

In summary, plant foods and spices, when consumed in a conventional manner, can generally be assumed to be safe if the same plant part is consumed, but caution should be applied to restrict higher than conventional ingestion of those known to have some degree of toxicity. Similarly, the assumption of safety may or may not apply to plants prepared differently. Of course, not all foods are safe for all persons. If adverse events are reported, then the level of concern should be higher. Most plant species containing overtly toxic compounds are not generally consumed, and consequently there will be no normal food intake data available. For plants that are not commonly consumed by humans, the best data as to adverse effects will come from observations and/or studies with herbivorous animals (see Chapter 5).

When considering the safety of a botanical dietary supplement ingredient, information about its chemical constituents may provide important clues as to the potential toxicities of the substance. Similarly, consideration of related plant species, especially those in the genera listed in Table 6-1, can provide information about chemical constituents that may be present in the ingredient, as well as toxicities associated with these chemicals. Some information about the potencies is also provided in Appendix C. The concept that it is important to consider chemical constituents and related plants of possible concern is summarized in the guiding principle at the beginning of this section: it is appropriate to consider risk by considering “… whether the plant source of the botanical dietary supplement is itself a toxic plant or is taxonomically related to a known toxic plant.” Several corollary guiding principles are important to remember as individual supplements are considered:

• Plant foods and spices consumed in a conventional manner can generally be assumed to be safe if the same plant part is consumed, but caution applies to higher than conventional ingestion of those known to have some degree of toxicity.

• Alkaloid-containing plants should always be suspected of possibly being pharmacologically active and should be considered as a risk factor.

• Unless there is evidence to the contrary, the assumption should be that all parts of a plant containing toxins pose a risk unless there is credible evidence suggesting that dangerous levels of toxins are not present in the part marketed as a dietary supplement.

• If production of a hazardous phytochemical in a botanical appears to be particularly sensitive to environmental and growth conditions, then it may be appropriate to consider its use in dietary supplements as a risk unless quality control or other actions are implemented to assure that toxic levels of compounds are not reached in raw materials or finished products.

• Preparation affects toxicity. Materials that are traditionally consumed in cooked form may not have the same safety profile as in uncooked form, compounds that are concentrated or otherwise altered by the method of preparation will present a hazard that is of greater concern than for unprocessed material, and knowledge about the safety of one plant preparation should not be applied prima facie to different preparations of the same plant.

• Particular concern should arise when nontraditional or excessive levels of plant parts are consumed. There may be a tendency to consume more of the plant material in an encapsulated form than if it were used in its “native” form.

• Extraction of plant material with alcohol or aqueous alcohol, in which low-molecular-weight compounds are generally very soluble, will likely concentrate toxic components several fold. Thus, for botanicals containing toxic compounds, a shift from aqueous to alcoholic extracts should be a cause for concern unless there is credible evidence to mitigate this concern.

The physical–chemical properties and biological effects of a substance are derived from its chemical structure. If the chemical structure of a dietary supplement is known, but additional insight into the biological activity is needed, then it is scientifically appropriate to consider the information about the biological activity of structurally related substances. It is assumed that the biological effects of chemicals, including toxic effects, are implicit

in their molecular structures (referred to as toxicophores when they are associated with toxic effects). This concept is most clearly illustrated with the example of ephedra, which is considered by some scientists to have similar physiological actions, although less potent, to the chemically related substance amphetamine, as well as the recently banned pharmaceutical agent phenylpropanolamine (FDA, 2004; Furuya and Watanabe, 1993; Lake and Quirk, 1984).

Along these lines, FDA and other agencies have developed chemical structure classes of concern to describe well-known toxicophores. The structure of regulated chemicals (e.g., pesticides or food additives) is compared with structural classes of concern to predict which may cause adverse effects. For ingredients regulated by premarket approval, structures of higher concern classes lead to a requirement for particular studies to provide additional information about the likelihood of adverse effects occurring in humans. The structural class approach is discussed in Box 6-4 and in the FDA Redbook II (OFAS, 2001). When the structure of a dietary supplement ingredient or its constituents belongs in one of the higher classes of concern identified via the Structure Category Assignment, it should be considered as a potential risk if mitigating information suggesting other-

wise does not exist. The rationale behind this recommendation is that the dietary supplements with structures of concern are no less likely to produce adverse effects than other ingested substances. Ideally, the use of information about the structural classes of concern will provide guidelines for manufacturers or other scientists to study the ingredient’s safety in more depth.

Based on the understanding that biological effects are implicit to molecular structures, computational programs have been developed to predict the biological activity of less-characterized chemicals by comparing their chemical structures with other well-characterized compounds. Computer programs designed to assist in predictive toxicology are useful in predicting the potential propensity for a chemical to cause particular effects. For example, The Open Practical Knowledge Acquisition Toolkit program (AIAI, 2003) uses chemical structures and a variety of models to estimate carcinogenicity and teratogenicity, among other toxicological endpoints, and is used by the Cosmetic Industry Review in setting priorities for review. An endorsement or comparative evaluation of individual programs is beyond the scope of this report, but these types of programs in general are believed to have value in providing insight into the potential for a dietary supplement ingredient to demonstrate toxicological outcomes, especially if there is a paucity of experimental or other data relevant to the ingredient’s safety.

Computational prediction is most useful for predicting biological activities of pure compounds because it is possible to circumvent the multitude of variables that influence toxicological endpoints, such as the presence of unknown chemicals, the animal species, strains, experimental conditions, and other factors. Even then, however, care must be exercised in selecting data that may be considered valid. Additional complexities, even vagaries, such as metabolic transformation and the multitude of modulating factors that determine a biological outcome under prescribed conditions, may obviate any meaningful predictive conclusion.

Moreover, other confounding factors are the often unique susceptibilities of individuals in animal and human populations that are dependent on complex genetic, environmental, and lifestyle factors. Even more difficult is the reliability of predictions of toxic response where there are effects of simultaneous exposures to a variety of compounds. Even the common example in humans of combining alcohol and tobacco smoke can exercise a striking influence on the toxic manifestations of another (test) material (Izzotti et al., 1998).

In summary, the understanding that toxic effects result from molecular structures that act on biological targets provides a good rationale for comparing chemical structures of a dietary supplement ingredient with other chemical structures to predict possible toxicities. Nonetheless, the

practicalities of how to make such comparisons in a systematic way and the limitations inherent in systematic prediction software may limit the usefulness of this approach, as may the important fact that small changes in chemical structure can result in major changes in physiological activity.

A number of dietary supplement ingredients may be structurally or functionally related to endogenous substances. Such substances include hormones, metabolites and their precursors, and ingredients created as mimetics of these substances (see Table 6-2 for examples). Any safety issues of ingredients related to endogenous substances are based on the extent to which the ingredient’s similarity to an endogenous substance alters homeostasis. Concern about one of these ingredients is warranted when certain characteristics or qualities are present, as discussed below.

Physiologically relevant amount of ingredient ingested: It is important to consider whether the ingredient is delivered to potential sites of action at a concentration that is physiologically relevant. For example, a supplement ingredient ingested at an amount that is clearly a small fraction of the amount typically provided in the diet is less likely to pose a risk because the likelihood of physiological impact is low.

Concentration at the site of action that can cause harm: The next consideration is whether the substance reaches the site of action at a concentration that can cause harm, which is largely determined by the substance’s bioavailability, rate of metabolism, and excretion. For example, dietary supplement ingredients for which evidence suggests negligible uptake from the gastrointestinal tract would be unlikely to pose a risk beyond local effects, such as gastrointestinal upset. In contrast, ingredients that may result in concentrations of endogenous substances above the normal

range at the site of action are worthy of further consideration, because of the potential for their activities to be adverse.

Having determined whether an ingredient is likely to be absorbed from the gut at a concentration that could alter cellular, biochemical, or biological activities, the next issue to consider is whether the compound is readily metabolized or degraded to an inactive metabolite. For example, effects on endpoint cells and tissues are unlikely if there are efficient mechanisms for metabolizing these compounds into inactive compounds.

Sensitivity of the target system to variation in the endogenous substance: An important question to ask about endogenous substances is whether a homeostatic regulatory system would attenuate biological effects that could otherwise occur. If the target system is one that is not tightly regulated by feedback or other mechanisms to maintain homeostasis, then there is greater likelihood of potential risk.

Hormones are an illustrative example for how some ingredients related to endogenous substances can be evaluated. Exogenous hormones can be potent substances often used clinically as pharmaceuticals to treat specific deficiency states (e.g., insulin to treat diabetes, thyroid stimulating hormone to treat hypothyroidism, human growth hormone to treat dwarfism) in order to achieve physiological homeostasis. Use of dietary supplements containing hormones, hormone precursors, or hormone mimetics known to be highly potent raises the possibility of significant and substantial harm unless there is demonstrated hormonal insufficiency.

Compounds that appear to be structurally dissimilar may nonetheless affect the same biological targets or have the same mechanism of action and thus result in the same downstream adverse health effect. Thus, if data strongly suggest that similar biological activity or mechanisms of action

exist between an ingredient and a substance known to be dangerous, there is scientific merit in considering whether similar adverse health effects might also occur. This is especially true if the ingredient or its relevant metabolites are bioavailable at the target site.

Although the actual data for such consideration will fall into the categories of data described in the previous chapters (in vitro, animal, or human data), the consideration of functional relatedness is described here rather than in the other chapters because (1) the concept applies to all types of data and (2) because the concept is similar to the concept of considering substances related in other ways (either structurally or, for botanicals, taxonomically). That is, a safety evaluation should consider the relationship between the dietary supplement ingredient in question and compounds known to be toxic. This type of information may be most useful in assessing the safety of a dietary supplement ingredient for which chemical constituents are not known or, for botanicals, when not much is known about the plant genus.

Functionally related substances may have similar actions in vitro, such as genetic effects or effects on cellular processes (e.g., enzymatic effects, effects on intracellular cell signaling). One example of such functional relatedness illustrated in Chapter 11 is saw palmetto and the drug finasteride. Finasteride is considered unsafe for consumption during pregnancy because of effects on male genitalia development (Bowman et al., 2003; Clark et al., 1990, 1993; Kurzrock et al., 2000). This effect is due to inhibition of the 5-α-reductase enzyme, which is important in testosterone production (Anderson and Clark, 1990; Prahalada et al., 1997). Saw palmetto also inhibits 5-α-reductase, as shown in in vitro experiments (Bayne et al., 1999), and thus would be considered as functionally related to finasteride. Thus, in the absence of mitigating data suggesting that saw palmetto does not also lead to male genitalia development problems, it would be scientifically appropriate to consider saw palmetto as a risk for consumption by pregnant women.

When considering which substances a given dietary supplement may be functionally related to, the purported mechanism of action of the ingredient should be considered. For example, shark cartilage has been referred to as an angiogenesis inhibitor. If angiogenesis inhibition is considered as dangerous, or angiogenesis inhibiting drugs are only used with caution by pregnant women because of this mechanism, then it would be appropriate to consider whether shark cartilage is indeed a risk for the same reason.

When evaluating whether functional relatedness to other chemicals provides helpful information about the safety of a dietary supplement ingredient, it is very important to consider that overt expression of toxicity is dependent on exposure (i.e., amount ingested or dose). When animal or human data do not exhibit the toxic effects predicted based on functional relatedness, then it is necessary to consider whether the possible effects

would not have been detectable, even if they did occur (such as would be expected for genetic effects or latent effects) (see also see “lack of effects” discussion in Chapter 10). Similarly, the amount of dietary supplement ingredient necessary to produce the effect in humans should be compared to the amount actually consumed.

In summary, if there is information about the mechanism of action of a substance suggesting that it exerts action similar to other classes of substances that are either considered dangerous or restricted in their use, then it may be appropriate to extrapolate such information to the dietary supplement ingredient, especially if little other information about the action of the supplement ingredient is available.

This chapter emphasizes the importance and scientific appropriateness of including information about related substances when considering the safety of a particular dietary supplement ingredient. For botanicals, the taxonomic and chemical relationships questions can be considered in parallel. Although evidence from one relationship alone may be sufficient to cause an awareness of significant risk, a higher level of concern arising from consideration of both relationships will amount to compelling evidence that the risk has to be seriously examined.

When a dietary supplement contains a known toxic substance, the hazard of ingestion must be assumed to be high unless mitigated by information about dose. Similarly, if the complete toxicity information about a given compound is not known, but it falls into a structural class of known toxins, then that compound is also likely to be a risk unless there is mitigating information about bioavailability or exposure levels.

Table 6-3, the relative spectrum of concern for relatedness information, provides general guidelines about the relative amount of concern for example scenarios using the different types of relatedness information. The information in the right-hand column suggests a significant risk to public health even in the absence of direct human data documenting adverse effects caused by the dietary supplement ingredient. In many situations like those described in the left-hand column, a conclusion of imminent risk may require corroboration with other types of information.

AIAI (Artificial Intelligence Applications Institute). 2003. TOPKAT—The Open Practical Knowledge Acquisition Toolkit. Online. University of Edinburgh. Available at http://www.aiai.ed.ac.uk/~jkk/topkat.html. Accessed July 9, 2003.

Anderson CA, Clark RL. 1990. External genitalia of the rat: Normal development and the histogenesis of 5 alpha-reductase inhibitor-induced abnormalities. Teratology 42:483–496.

Asano N, Kato A, Matsui K, Watson AA, Nash RJ, Molyneux RJ, Hackett L, Topping J, Winchester B. 1997. The effects of calystegines isolated from edible fruits and vegetables on mammalian liver glycosidases. Glycobiology 7:1085–1088.

Bayne CW, Donnelly F, Ross M, Habib FK. 1999. Serenoa repens (Permixon): A 5 alphareductase types I and II inhibitor–A new evidence in coculture model of BPH. Prostate 40:232–241.

Bowman CJ, Barlow NJ, Turner KJ, Wallace DG, Foster PMD. 2003. Effects of in utero exposure to finasteride on androgen-dependent reproductive development in the male rat. Toxicol Sci 74:393–406.

Clark RL, Antonello JM, Grossman SJ, Wise LD, Anderson C, Bagdon WJ, Prahalada S, MacDonald JS, Robertson RT. 1990. External genitalia abnormalities in male rats exposed in utero to finasteride, a 5 alpha-reductase inhibitor. Teratology 42:91–100.

Clark RL, Anderson CA, Prahalada S, Robertson RT, Lochry EA, Leonard YM, Stevens JL, Hoberman AM. 1993. Critical developmental periods for effects on male rat genitalia induced by finasteride, a 5 alpha-reductase inhibitor. Toxicol Appl Pharmacol 119:34–40.

Cordell GA, Quinn-Beattie ML, Farnsworth NR. 2001. The potential of alkaloids in drug discovery. Phytother Res 15:183–205.

Coulombe RA. 2003. Pyrrolizidine alkaloids in foods. In: Taylor SL, ed. Advances in Food and Nutrition Research. Vol 45. Amsterdam: Academic Press. Pp. 61–99.

Drager B, van Almsick A, Mrachtz G. 1995. Distribution of calystegines in several Solanaceae. Planta Med 61:577–579.

FDA (Food and Drug Administration). 2004. Final rule declaring dietary supplements containing ephedrine alkaloids adulterated because they present an unreasonable risk; Final rule. Fed Regis 69:6787–6854.

Fong HHS. 2002. Integration of herbal medicine into modern medical practices: Issues and prospects. Intergr Cancer Ther 1:287–293.

Furuya I, Watanabe S. 1993. Discriminative stimulus properties of ephedra herb (Ephedra sinica) in rats. Yakubutsu Seishin Kodo 13:33–38.

Harborne JB. 1993. Introduction to Ecological Biochemistry. 4th ed. London: Academic Press.

Hartmann T, Witte L. 1995. Chemistry, biology and chemoecology of pyrrolizidine alkaloids. In: Pelletier SW, ed. Alkaloids: Chemical and Biological Perspectives. Vol 9. Oxford, UK: Levier Science.

IPNI (International Plant Names Index). 2004. About Index Kewensis. Online. Available at http://www.ipni.org/ik_blurb.html. Accessed March 22, 2004.

Izzotti A, Balansky RM, Blagoeva PM, Mircheva ZI, Tulimiero L, Cartiglia C, De Flora S. 1998. DNA alterations in rat organs after chronic exposure to cigarette smoke and/or ethanol ingestion. FASEB J 12:753–758.

Kurzrock EA, Jegatheesan P, Cunha GR, Baskin LS. 2000. Urethral development in the fetal rabbit and induction of hypospadias: A model for human development. J Urol 164:1786–1792.

Lake CR, Quirk RS. 1984. CNS stimulants and the look-alike drugs. Psychiatr Clin North Am 7:689–701.

Mabberley DJ. 1997. The Plant Book. 2nd ed. Cambridge, UK: Cambridge University Press.

OFAS (Office of Food Additives Safety). 2001. Toxicological Principles for the Safety Assessment of Direct Food Additives and Color Additives Used in Food. Redbook II-Draft. Washington, DC: OFAS, Food and Drug Administration.

O’Neil MJ, Smith A, Heckelman PE, eds. 2001. The Merck Index: An Encylopedia of Chemicals, Drugs, and Biologics. 13th ed. Whitehouse Station, NJ: Merck and Co.

Pelletier SW. 1983. The nature and definition of an alkaloid. In: Pelliter SW, ed. Alkaloids: Chemical and Biological Perspectives. New York: Wiley. Pp. 1–31.

Prahalada S, Tarantal AF, Harris GS, Ellsworth KP, Clarke AP, Skiles GL, MacKenzie KI, Kruk LF, Ablin DS, Cukierski MA, Peter CP, VanZwieten MJ, Hendrickx AG. 1997. Effects of finasteride, a type 2 5-alpha reductase inhibitor, on fetal development in the rhesus monkey (Macaca mulatta). Teratology 55:119–131.

Seawright AE, Hegarty MP, Keeler RF, James LF, eds. 1985. Plant Toxicology. Brisbane, Australia: Queensland Poisonous Plants Committee.