Pediatric Annals

Carcinogens and Other Toxic Hazards: How Accurate Are Animal Studies?

Daniel B Menzel, PhD


Abbreviations in this article

DNA: Deoxyribonucleic acid

FM: Fast metabolizer

MFO: Mixed-function oxidase system

PM: Poor metabolizer

PMA: 4-Methoxyamphetamine (a hallucinogen)

RNA: Ribonucleic acid


Abbreviations in this article

DNA: Deoxyribonucleic acid

FM: Fast metabolizer

MFO: Mixed-function oxidase system

PM: Poor metabolizer

PMA: 4-Methoxyamphetamine (a hallucinogen)

RNA: Ribonucleic acid

Children are special targets for the action of toxicants. Exposure in childhood leads to a longer exposure period during a lifetime, and consequently a greater likelihood of reaction between toxicant and receptors.

How valuable are studies in animals in determining whether a given substance will be harmful to an infant or child - now or later? Extrapolating the results of animal studies to human beings is one of the most challenging problems of toxicology. Some have even questioned the usefulness of employing animals at all, suggesting that only human beings are suitable models.

Such a view, of course, ignores the long history of experimental medicine. The hazards of assessing new drugs and experimental compounds in man are too great; it is essential that responses in animals be quantità ted first, and related to the quantitative responses in man.

Restated in more explicit terms: What is the minimal toxic dose to man? Will the compound produce mutations in man? Will the compound produce cancer in man? Is it safe to expose persons in specific age groups or disease groups? Are the risks worth the benefits?

These are some of the questions the toxicologist attempts to answer. And in this article, we will explore some of the ways he arrives at conclusions about the relative danger, or safety, of carcinogens and toxicants to developing human beings.


The judgment that a drug or chemical is a hazard is not determined solely by animal toxicology. There are four sources of information that can be used to estimate the hazards to man of a given entity - animal studies, human clinical studies, epidemiologic studies, and analytic chemistry.

Animal studies have the greatest flexibility, since they are not limited by dose, route of administration, time, or end point. The basis for the use of animals is their similarity to man. While certain anatomic differences clearly exist between man and animal surrogates, the basic biochemical effects of toxicants differ little. Because of the irreversible nature of mutations and genetically mediated symptoms, animals are a vital, irreplaceable necessity in the testing of drugs and foreign compounds before clinical trials or commercial use.

Human clinical studies and observations during the controlled intentional exposure of a person to a given compound need no caveats or extrapolations for risk assessment, but they suffer from the ethical limitations that are inherent in any human experimentation. The line of ethical limits in dealing with developing human beings from conceptus to the age of majority is poorly delineated in the United States,* resulting in the unfortunate use of exposure data that are inadequately controlled.

Epidemiologic studies are vital counterparts of clinical studies, differing in the uncontrolled nature of the exposure. As a consequence, their results are inherently limited and confounded by numerous interactions.

Figure 1. Pathways of oxidative metabolism of carcinogens through the MFO.

Figure 1. Pathways of oxidative metabolism of carcinogens through the MFO.

Analytic chemistry, lastly, supplies us with unifying ideas that reinforce our certainty about the observation. As I will discuss below, the greatest single predictive advance in toxicology has been the reactive-intermediary theory. This theory would not have been possible without sensitive, sophisticated analytic chemistry. Nowadays one can detect as little as 10 10 to 10~12 gm. of the parent compound with ease. More important, one can trace in detail the pathways of metabolism, in both man and animals, to assure their similarity or to detect their dissimilarity.

Prediction is firm when the observations from all four disciplines coincide. Unfortunately, few compounds have been studied with detail and perspective to provide the coincidence of observations.


Carcinogenesis by chemical compounds is one of the most intensive areas of toxicological study. Early, the Millers1 observed that certain carcinogens had to be activated metabolically. The key to this scheme is the oxidative metabolism of the carcinogen through the mixed-function oxidase system (MFO) (Figure 1). The MFO depends for its catalytic activity on cytochrome P450.

The chemical form of the reactive intermediary is not restricted to arene oxides' as originally described. Free radicals, for example, arise from carbon tetrachloride or halothane metabolism. The diversity of enzymatically catalyzed reactions is often confusing, since the total enzymatic activity of the tissue is most likely the result of a mixture of different cytochrome-P4J0 isoenzymes. At present, one can characterize no single form of cytochrome P450 within a tissue by a single enzymic reaction.

Oxidation on carbon is the favored reaction. Most of the absorbed compound finds its way to some metabolite, but a significant amount (20 to 40 percent) is often excreted unchanged. The excretion of unchanged compound is as important a pathway of detoxification as is metabolism. A very small fraction of the reactive intermediary reacts with macromolecules of the cell. Some of the reactive intermediary may react with proteins and lipids, while an exceedingly small fraction may react with cellular deoxyribonucleic acid (DNA) or ribonucleic acid (RNA). The total amount bound is very small, being of the order of one part of foreign compound to 109 to 1012 parts of cellular constituent.

Mutations and carcinogenesis are associated with the reaction of the foreign compound with DNA or RNA. It is now possible to isolate the nucleotide bases that have reacted covalently with the reactive intermediary of the foreign compound. Radiolabeled carcinogen can be found covalently bound at the 8 position to guanine. Breaks in DNA are also detected and correlate with the amount of covalently reacted foreign compound to DNA.

These observations from animal cells in culture and intact animals support the use of microbialmutagenesis-assay systems, such as the Ames assay, as a preliminary screen for carcinogens. The microbial systems measure mutagenic reactions with DNA indirectly by measuring the number of mutations produced by the carcinogen or mutagen. Since microbial cells, such as the Salmonella used in the Ames assay, do not posses's cytochrome P45o for activation of the putative carcinogen or mutagen, rat-liver homogenates are usually added. These are derived from rats pretreated with inducers of the aryl hydrocarbon hydroxylase system (cytochromeP448 isoenzyme).

Unfortunately, the microbes also possess metabolic enzymes of a different type from those of mammalian cells, so it is not always clear exactly what the chemical identity of the reactive intermediary is. This disadvantage has serious implications. It is my view that a positive or negative test in a microbial system is not definitive and requires additional experiments to either prove or refute the mutagenic capacity of any compound. It provides little insight into the true carcinogenic capacity of the compound in man.

The number of false positives (presumptive carcinogen /mutagen) found by these systems is being reported only poorly, since a large number of compounds are now being screened that are in the developmental stages. Commercial producers are unwilling to reveal such information, and it is not possible to judge how this simplistic approach of correlating mutagenesis with carcinogenesis may be hampering development of important compounds or protecting the public from great hazard.

The important lesson to be learned here is that oxidative metabolism is a general means by which carcinogens are activated to their reactive and therefore ultimate form. This reaction appears to be carried out by the cytochrome-P450 system, but it is not exclusive in this action. Other enzymes are present that may also play an important role and are yet to be investigated as thoroughly as cytochrome P450. Still, it is safe to say that almost 90 percent of the oxidative metabolism of all foreign compounds proceeds through the cytochrome-P4So system.


The oxidative metabolism of drugs is just as important in measuring toxicity as it is in making judgment on environmental carcinogens. Toxic reactions are better correlated with covalent reactions between the drug and proteins than with DNA or RNA. In the case of drugs metabolized by dehalogenation, reaction with lipids is better correlated with toxicity. Since it is not now possible to identify the catalytic nature of the proteins with which the drug has reacted, it is currendy difficult to describe the chemical nature of the toxic reaction resulting from covalent reaction with cellular proteins.

One can imagine a variety of possible mechanisms. Inhibition of a specific enzyme, acetylcholinesterase, has proved the cause, for instance, of organophosphate-insecticide poisoning. By analogy, covalent reaction at or near the catalytic site of some one or more metabolically important enzymes could be toxic. The cellular membrane is a highly likely site for toxic reactions produced by covalent reactions with both the lipid and protein macromolecüles. The cytoskeleton is an intriguing target for toxicity, especially from compounds producing delayed neuromuscular toxicity.

While one cannot point to a single reaction as being toxic, hepatocellular necrosis is well correlated with the extent of covalent reactions. A general rule is that reduced-glutathione levels are also depleted during these reactions, suggesting that reduced glutathione reacts with the reactive intermediary to prevent subsequent toxic reactions with cellular macromolecüles. Genetic, nutritional, or environmental factors lowering the intracellular reduced-glutathione content of the cells then tends to increase the toxicity of drugs acting in this manner. A threshold or saturation dose also is observed, suggesting that toxicity occurs when reduced-glutathione levels have been exhausted in the cell.

Figure 2. Incidence of normal (fast metabolizers) and defective (poor metabolizers) oxidative drug metabolism in man.

Figure 2. Incidence of normal (fast metabolizers) and defective (poor metabolizers) oxidative drug metabolism in man.

Acetaminophen toxicity is an example of the threshold-dose concept, whereby reducedglutathione levels fall precipitously once the dose reaches a specific level. While the threshold-dose phenomenon appears in drug toxicity, no such threshold is apparent in carcinogenesis.


Genetic polymorphism has been found in the acetylation of some sulfonamide and hydrazine drugs; in the hydrolosis of esters, such as the drug succinylcholine and the insecticide paraoxon; and in the glucuronidation of a number of substrates representative of drug and foreign-compound metabolites.2,3 Mahgoub and his co-workers4,5 have demonstrated the genetic polymorphism of oxidative metabolism. Using the antihypertensive drug debrisoquin (3,4-dihydro-2[lH]-isoquinoline-carboxamidine), they were able to detect a single autosomal gene controlling the rate of oxidative metabolism of debrisoquin by alicyclic hydroxylation. Two pheno types were found: fast metabolizers (FMs) and poor metabolizers (PMs). In a survey of English Caucasians, the incidence of PMs was about 6 percent of the population. FMs converted 50 to 70 percent of the oral dose of debrisoquin to its alicyclic hydroxylated metabolite, while PMs converted only 1 to 3 percent of the drug. Family studies demonstrated that the incidence of the PM phenotype was 6 percent in West African blacks and about 1 percent in Egyptians (Figure 2).

Subsequent studies revealed that the genetic polymorphism was not restricted to the alicyclic hydroxylation of debrisoquin but also included the aromatic hydroxylation of guanoxan and oxidation of the a-carbon of phenacetin. The defect in oxidative metabolism of these patients is profound and extends to a wide variety of drugs whose single feature of metabolism is oxidation at a carbon center. This pathway is, however, the dominant pathway of drug and foreign-compound metabolism in man.

The gene responsible for oxidative metabolism is not linked to that already established for acetylation. Slow acetylation has an incidence of about 60 percent in whites. The possibility of having both traits is quite high - about 1 in 20 in whites (0.6x0.09).


How does metabolism by different human phenotypes compare to that of test animals? Let us consider 4-methoxyamphetamine (PMA), a potent hallucinogen that has been identified as a cause of death when it is misused. PMA is oxidatively metabolized by O-demethylation, N-oxidation, deamination, and /3-oxidation. A significant fraction of the drug may be excreted unchanged in the urine. Kitchen et al.6 compared the metabolism of this drug in the rat, guinea pig, human PMs, and human FMs, characterizing quantitatively the extent and chemical nature of the metabolites found. In this way, it was possible to make a clear comparison between the two phenotypes in man and their potential surrogates in rodents. In Figure 3 I have shown schematically the results obtained. Both the rat and guinea pig metabolized PMA extensively through the oxidative O-demethylation. The rat also metabolizes PMA by N-oxidation, a pathway not found in guinea pigs.

In the FM phenotype of man, O-demethylation represented 70 to 84 percent of the metabolism of PMA. N-oxidation and /3-oxidation accounted for about 5 to 7 percent each, but up to 20 percent of unmetabolized drug was also excreted. The rat then represents a fair surrogate to 90 percent of the human population as far as PMA metabolism is concerned.

In contrast to the FM phenotype in man and its animal surrogates, the metabolism of PMA by the PM human phenotype is very different.

The O-demethylation pathway is much reduced. Excretion of unmetabolized PMA is the dominant pathway for some 58 percent of the drug. N-oxidation now represents an important pathway of some 20 percent. But the important difference, lies in the occurrence of an entirely different pathway of deamination not found in the common FM phenotype in man or in his animal surrogates. The poor overall metabolism of PMA by the PM phenotype results in higher blood levels of this drug for prolonged periods. This may have been the reason for mortality from the abuse of this drug. Unfortunately, the phenotype of patients dying from PMA intoxication is not known, so this hypothesis is still speculative.


In neonates the total metabolism of drugs is much lower than it is in adults, whether by oxidative, hydrolytic, or conjugation pathways. The various oxidative pathways develop at different rates. The excretion of unchanged drugs or toxicants then assumes a more important role in the neonate than in the adult, and the newborn who is genetically defective in both oxidative and conjugative metabolism is doubly compromised.

The maternal metabolism of drugs applied during parturition, such as analgesic agents (pethidine, meperidine), may be adequate, while that of the neonate may not be. Prior identification of the phenotype of mother and neonate would be helpful, but is difficult to implement at present.

Given the incidence of defective pathways of metabolism, the long-term prescription of drugs to infants and children is especially worrisome, as is exposure to environmental toxicants. Infants and children who poorly metabolize these substances will have persistently higher concentrations of foreign compounds.


Animal surrogates of man metabolize foreign compounds and drugs through the same pathways as does man. If the reactive-intermediary theory is correct, the same toxic and carcinogenic reactions should also occur in animals as in man. Clearly, there are genetic differences in oxidative metabolism in man not known to be present in animals. Only through precise analytic chemistry are these differences detectable. As more compounds are studied, generalizations can be made about aberrant pathways of metabolism in man and better surrogates can be found in animals by searching for strains of animals with these traits.

If drug and foreign-compound metabolism and toxicity are qualitatively similar in animals and man, the job of extrapolation is much easier. The problem of risk assessment becomes one of scale or size. Mathematical models are particularly useful here. However, simple mortality studies (counting dead animais administered a given dose) do not provide the necessary rate data needed to construct models of the metabolism, distribution, and reaction of a toxicant. Metabolic studies are essential here. From dose-response and metabolic data, one can construct a mathematical model predicting the tissue dose of toxicants and relating that dose to the toxic effect. The effect can then be scaled to man and the toxic effect estimated more precisely.

Figure 3. Comparative metabolism of PMA by rodents and two phenotypes of human oxidative drug metabolism. (FM man, fast metabolizers; PM man, poor metabolizers.)

Figure 3. Comparative metabolism of PMA by rodents and two phenotypes of human oxidative drug metabolism. (FM man, fast metabolizers; PM man, poor metabolizers.)

One such application has been made for ozone by Dr. F. Miller. Mathematical models are easily altered to fit different rates of metabolism caused by genetic defects, disease, age, or state of development and are thus adaptable for estimating toxicity in even small populations.

What is the likely incidence of defective oxidation metabolism in man and is it a significant problem? It is clear from the work of Smith and colleagues that the metabolism by some people is both less and different. 2·3·5 One can calculate that in a town of some 200,000 population about 10,000 adults would have such a defective oxidative-metabolism trait. In a practice having some 3,000 patients, this amounts to 150 defective individuals. The clinical implications have been mostly related to the elevated levels of drugs in affected persons. "Hypersensitivity" to the antihypertensive properties of debrisoquin, for example, precipitated these studies. Methemoglobin formation has also been detected in PM phenotypes. Smith has suggested that defective metabolism may well be the reason for most of the side effects of drugs.

With regard to the broader question of determination of toxic hazards to man from chemicals in the environment, these studies suggest that a small but not insignificant fraction of the U.S. population is at greater risk from toxic chemicals than the general population.

Carcinogenesis from organic compounds in animals provides sure evidence that it will occur in man. So far no threshold effect has been shown for carcinogens. Thus, dose is not as important as incidence of exposure. The question of cancer risk from carcinogens is still open, but the safest course is to avoid all known carcinogens.

Curiously, little attention has been paid to the genetics of metabolism in laboratory animals aside from the mouse, which is seldom used in testing. The approach has been to standardize on one strain of rat, Fisher 244, but this strain has been poorly characterized. By such standardization procedures, research is being taken out of hazard evaluation, with the result that new developments in drug metabolism are ignored or slowly implemented. The National Toxicology Testing Program is an expensive and vast undertaking but will not provide the data needed for risk assessment unless the present trend to "cookbook testing'* is revised. We must look beyond counting tumors to the causes of the tumor.

A major problem of extrapolation of hazards to man from animals is the role of disease in human responses. Almost all animal data come from young, healthy animals. Diseased animals are excluded. A major new aspect of the U.S. Environmental Protection Agency Health Effects of Air Pollution Program will deal with the effects of pulmonary disease on the toxicity of air pollutants. This is a much-needed approach.

While the discovery of carcinogenesis has been recognized as an important objective, the general action of carcinogens as mutagens has been overlooked. Carcinogenesis undoubtedly is a multistep process and thus a relatively rare event. Other mutations, especially somatic mutations, are more likely to survive and may lead to chronic disease. The Benditt hypothesis of atherosclerosis is one such possibility.

Thus, as was noted at the beginning of this article, children are special targets for the action of toxicants. They metabolize foreign compounds more slowly than adults, with the result that there is a greater accumulation of these compounds in their bodies even though they have received no more exposure than have adults.

Exposure in utero is another area of concern. Here, studies in rabbits have proved most useful. Human mothers having defective metabolism will have higher blood levels of foreign compounds than normal women and thus present a higher grathent for diffusion of compounds across the placenta.

Pregnant women are now working longer in their occupations than was the case only a few years ago, and they are working in more diverse environments, raising the possibilities of greater occupational exposure to chemicals. What can be done to protect the fetus? One approach might be to restrict pregnant women - or even women of the reproductive ages - from any job where exposure is likely. Or these women could be phenotyped, and those restricted who had defects in metabolism. Such approaches are already being applied by some chemical corporations. They represent a potentially discriminatory, albeit paternalistic, solution. A better solution would be to rid the workplace of toxic hazards.

The close parallels between drug and carcinogen metabolism suggest that much can be learned about the origins of human cancer from studies of drug metabolism in animals and man. Present studies of the cytochrome-P45o system in cancer patients have been disappointing, snowing little correlation, but this may be due to the advanced stages of the disse itself. The low incidence of genetically defective drug metabolism in man makes detection of these individuals difficult and espidemiologic studies vague. Alert clinical observations to detect persons having a higher frequency of adverse reactions to drugs or "hypersensitivity" to drugs could form an important information base for epidemiologic studies.

As has so often been the case in medicine, a search for diversity, rather than an obsession with protecting the majority, may be the key in providing a safe, useful chemical environment for man.


1. Miller, J. A., and Miller, E C. Ultimate chemical carcinogens as reactive mutagenic electrophiles. In Hiatt, H. H., Watson, j. D., and Winsten, J. A. (eds.): Origins of Human Cancer, Book B. Cold Spring Harbor, N.Y.: Cold Spring Harbor Laboratory, 1977, pp. 605-627.

2. Price Evans, D. A. Human pharmacogenetics. In Parke, D. V., and Smith, R. L. (eds.): Drug Metabolism from Microbe to Man. London: Taylor and Francis, 1977, pp. 369-391.

3. Atlas, S. A., and Nebart, D. W. Pharmacogenetics and human disease. In Parke, D. V., and Smith, R. L. (eds.): Drug Metabolism from Microbe to Man. London: Taylor and Francis, 1977, pp. 393-430.

4. Mahgoub, A., et al. Polymorphic hydroxylation of debrisoquine in man. Lancet 2 (1977), 584-586.

5. Mahgoub, A., Idle, J. R., and Smith, R. L. A population and familial study of the defective alicyclic hydroxylation of debrisoquine among Egyptians. Xenobiotica 9 (1979), 51.

6. Kitchen, L, et al. Interindividual and interspecies variation in the metabolism of the hallucinogen 4-methoxyamphetamine Xenobiotica 9 (1979), 379-404.


Sign up to receive

Journal E-contents