Arehlvu of
TOXICOLOGY
Arch. Toxieol. 43, 27-33 (1979)
9 Springer-Verlag 1979
Toxicological Barriers to Providing Better Drugs Louis Lasagna Departments of Pharmacology and Toxicology,and Medicine, University of Rochester School of Medicine and Dentistry, Rochester, New York, 14642, USA
Abstract. The unmet needs of the sick demand that toxicologic requirements do not stifle the rational search for new and better remedies. A number of conceptual problems hamper the rational use of toxicological testing. These include: a misplaced confidence in the value of animal testing, a failure to make sophisticated risk-benefit analyses, the proliferation of new tests of uncertain validity, and improperly executed retrospective case control studies. Regulatory barriers include the ever increasing bureaucratic demand for toxicological testing, the unseemly willingness of regulatory agencies to yield to hysterical or cynical consumer group pressures, the unreasonable demand for "superiority" of new products before the granting of registration, and the temptation to institute expensive but untested post-marketing surveillance schemes. Economic obstacles to new drug development have become formidable, and new demands for toxicologic studies in animals and humans are adding to these problems. Finally, some examples of unwise regulatory decisions involving saccharin, spray adhesives, Depo-Provera, and a new anti-metabolite are given. Key words: Toxicology - Mutagenicity - Carcinogenicity cals.
Pharmaceuti-
Introduction In the United States, and to some extent also in other countries, the toxic effects of chemicals seem to preoccupy the public, the news media, and the politicians. Much less attention is paid to the need for new and better medicines, although such need is painfully obvious in any doctor's office or hospital. Despite the great progress made by the Pharmaceutical Revolution that began a few short decades ago, there is no area of medicine that does not suffer by reason of inadequate drugs. Many cancer patients can expect little from our present therapies.
0340-5761/79/0043/0027/$ 01.40
28
L. Lasagna
Schizophrenics would welcome chronic treatment that does not carry with it the risk of irreversible neurologic damage. Severely afflicted arthritics find present remedies far less effective and more toxic than desired. Microorganisms can be expected to spawn strains that are resistant to widely used antibiotics. The muscular dystrophies do not respond significantly to available drugs. The list goes on and on, to the frustration of the sick, their relatives, and physicians, who want to help but often cannot. While few scientists would disagree with the proposition that patients need better drugs than those now available, there is considerable controversy over the amount and nature of toxicologic testing in animals and man that is optimal. I propose to examine some of the conceptual, economic, and regulatory aspects of such testing, with a particular view to identifying toxicologic requirements that cannot be logically defended.
Conceptual Problems Let me begin with some conceptual problems. A prime difficulty is the misplaced confidence that many place in animal testing. Such tests have long been extremely useful in detecting toxic effects that are dose related and not species specific. But they cannot be relied upon to detect toxicity that does not possess these attributes. This pathetic illusion gives rise - quite logically if you believe the premise - to the idea that simply doing enough animal testing will predict all human toxicity. A second conceptual difficulty arises from the growing failure of certain groups in our society to conduct risk-benefit analyses in determining whether or not to pursue a drug in man. The essence of toxicity testing is to delineate the extent and nature of the risk associated with a drug as one varies dose and duration of treatment, and thus to calculate the risk not only under conditions of massive ingestion, but at those levels adequate to produce the desired therapeutic benefit. For certain kinds of toxicity, such as carcinogenicity, the attitude of some seems to be that any risk of cancer, no matter how remote, should rule out the use of a drug in man. In the case of food additives, the Delaney Amendment has embodied this concept in law in the United States. A third conceptual difficulty relates to the willingness of the scientific community to accept, as a basis for rejection of drugs for human use, new tests of unknown specificity or validity simply to satisfy the needs that we all feel for ways of predicting teratogenicity or carcinogenicity in man. The Ames test and a variety of other procedures for mutagenicity and carcinogenicity assessment are now being employed everywhere in the world. The operating procedure generally has become that such tests - if positive - will ordinarily toll the death knell of a chemical. This despite the fact that the validity of such predictions -- perhaps especially those based on bacterial systems - is far from clear. Advocates of the bacterial tests for mutagenicity point out that since these procedures are not expensive to run and do not take much time, thousands of chemicals can be screened yearly. They assert that 90% or so of carcinogens are also mutagens, and that false positives occur perhaps only 15% of the time [6, 7].
Toxicological Barriers to Providing Better Drugs
29
In contrast, a distinguished committee appointed by the Council of the Environmental Mutagen Society said the following three years ago [2]: 1. " . . . only the intensive accumulation of comparative information can indicate the reliability of extrapolation procedures . . . . " 2. " . . . no information exists proving or disproving the existence of a threshold response to a mutagen in any system . . . . " 3. " . . . the identification of a compound as a mutagen is only the first step in estimating the hazard it may pose to man . . . . " 4. " . . . quantitative extrapolation from microbial systems, or from reversion tests in which tissue cultured mutant mammalian cells are used, does not yet appear to be possible . . . . " 5. " . . . the test systems now available do not provide a very broad base for screening, either in their abilities to detect all forms of heritable mutations or in their relevance to mutation in man. We therefore urge that a number of new systems be developed . . . . " Later in that same year, the environmental carcinogenesis subcommittee of the National Cancer Advisory Board pointed out that "the short-term (or in vitro) tests cannot be used to define a carcinogen." The subcommittee further warned of certain positive results even in experimental animal tests which in themselves would not be definitive evidence of carcinogenicity. These included assays in which the strain of animals had a greater than 10% spontaneous frequency of a particular tumor; or in which known carcinogens or other foreign material or modifying factors were used in addition to the test compound; or in which test animals were subjected to unphysiologic conditions [3]. Riley, also in 1975, reported the fantastic differences that could be achieved in C3H/He strain mice carrying an oncogenic virus by manipulating the type of housing and handling [8]. These observations suggest the need for great caution in the design and conduct of earcinogenicity tests in rodents. But to go back to the Ames test. It is acknowledged that false negatives would not be surprising, and hence "passing" the Ames test is considered to provide only slight assurance about a drug's safety, so that more classical tests using mammals are usually required to "establish" a drug's safety. Hence these new quick and inexpensive tests now often serve to kill a drug, but not to grant it developmental life. Will such a philosophy yield harm to society? No one knows. To test the potential losses would require money and time, and involve legal and ethical problems. What we do know is that if a compound does not get to market, we will never be able to assess its true potential for human benefit (or harm). We have plenty of examples of drugs that produced pleasant (as well as unpleasant) surprises after registration, when new and important uses were found by accident for them - lidocaine and phenytoin for cardiac arrhythmias, diazepam for status epilepticus, estrogens and progestins for contraception, etc. Clearly the earlier in a drug's development the decision is made to kill a drug, the more information will be lost about the drug's potential. Carcinogenicity testing also suffers from a failure to consider the logical statistical consequences of lifetime studies in animals known to develop cancers spontaneously. Control Swiss-Webster mice, for example, have been reported in three sepa-
30
L. Lasagna
rate 14-month studies to show tumors in 15 to 26% of animals. The majority of these tumors were malignant [11]. The incidence of tumors went up to 28 to 37% in three separate 24-month studies. Salsburg has analyzed the implications of such data [10]. Let me quote him: During the past four or five years, both new and old drugs have been subjected to a toxicology "screen" involving a procedure which has never been fully evaluated to determine its relevance to man, which has a high probability of producing false positive results, and which adds about a quarter of a million dollars to the cost of a drug's development and up to three years additional time on to that development. Not only is the procedure expensive and of unknown value, it also violates fundamental principles of good biological experimentation and it can easily produce results which cannot be analyzed without serious statistical bias and a resulting confusion between direct carcinogenesis and extraneous biological events. This procedure has cast suspicion onto such marketed drugs as a widely used sedative (phenobarbital), the first line of defense in the treatment of tuberculosis (isoniazid), a standard treatment for trichomoniasis (metronidazole), a potent diuretic (spironolactone), and a whole class of psychotherapeutic agents (the phenothiazines). A poll taken at the Biopharmaceutical subsection of the American Statistical Association meetings in 1976 indicated that, up to that time, at least 10 new drugs had been withdrawn from investigation in the U.S. due to findings from such studies against less than 30 such studies completed. Using a Monte Carlo computer simulation technique, Salsburg has calculated that 20-month feeding experiments in mice will yield false positive conclusions about carcinogenicity at least 50% of the time, as will two-year studies in male rats. His cogent analysis [9, 10] is, in my opinion, required reading, especially since even our best statisticians seem unable to grasp the notion that running one or more control groups will not solve all of these problems. Briggs has recently criticized the mandatory use of beagle clogs for chronic toxicity testing of contraceptive steroids [1]. He believes that the beagle, unlike rodents or monkeys, tends to falsely incriminate progestogens that are in fact safe for the human. Let me give one additional conceptual problem - the tendency to rely on human case control studies to assess the capability of a drug to produce rare serious side effects. The case control method can be highly useful in such instances, but it has certain inherent dangers, especially the possibility of bias in drug history, in case selection, and in the construction of a control group. It now appears, for example, that the widely publicized studies incriminating reserpine as a cause of breast cancer were a fluke of some sort, and that Rauwolfia alkaloids probably do not cause breast cancer. A similar controversy now rages over the suggestion that estrogens may cause endometrial cancer. There are regulatory barriers as well. Regardless of what science might recommend as defensible toxicology requirements, there is a growing tendency to require more and more tests, regardless of their cost or value. For the regulatory bureaucrat, this poses no problem, since he is rarely criticized for failing to approve a drug, but is often berated for having registered a drug whose toxic potential turns out to be greater than anticipated.
Toxicological Barriers to Providing Better Drugs
31
Recently, a consumerist group prodded our Secretary of Health, Education and Welfare into invoking, for the first time, the so-called "imminent risk" provision to withdraw phenformin from the market. Whatever one's feelings may be about this drug, it is hard to see what new scientific insights at that particular moment in history created the need for action. The relationship of phenformin to some cases of lactic acidosis had been known for years, and the true incidence is no better known today than it has ever been. What this incident shows, however, is the power of public agitation, even by a group with no large constituency, to move politicians. A troublesome new development in the United States is the tendency to disapprove a new drug application because a drug is "no more effective than older drugs but more toxic". At In'st glance this seems sweetly reasonable. But is it? How many anti-epileptics would be off the market if we used these guidelines? How many anticancer drugs? Is it not abundantly clear that the optimal practice of medicine often requires individualization of therapy, and that one man's best drug may be different from another's? Do we not need "backstop" drugs whose use can be invoked when first- or second-line drugs prove either ineffective or toxic? One other aspect of toxicologic scrutiny is becoming more important. I refer to the growing tendency to demand, perhaps routinely, postmarketing surveillance of new drugs [12]. I have long believed that we now neglect the proper study of drugs after registration, and that such postmarketing scrutiny is crucial because many troubles simply cannot be detected prior to marketing. The "naturalistic" study of drugs in ordinary clinical use can potentially tell us a great deal about the overuse, underuse, misuse and abuse of drugs that cannot be easily studied (if at all) in the premarketing phases. It is in the marketplace that we should be quantifying - as best we can - the benefits as well as the costs of drugs. Therefore in principle I am in favor of moves in this direction. What bothers me, however, is the haste to make such studies mandatory when we do not know how best to do them, and the temptation to do them only to check on toxicity rather than on both toxicity and benefit. These studies can be enormously expensive if large numbers of patients are to be followed carefully, and it would be a pity to delude ourselves into believing that rare or long-delayed side effects are going to be readily detected by such schemes.
Economic Aspects The economic obstacles to new drug development have become formidable. Our own studies suggest that $55 million and 7 - 8 years are required in the U.S. to move a drug to market from the inception of human studies. As a result, effective patent life is probably more like 9 - 1 0 years than 17. Every new toxicity test that is added on as a requirement will add to both the length and the expense of drug development. At the same time, there are pressures all over the world to constrain the prices of drugs and to limit exclusivity in the market place. Is it possible that these various trends may summate to diminish adventuresomeness and creativity in the drug industry? It has been suggested that follow-up studies should be conducted - perhaps for life - in humans who have received even small doses, for short periods of time, of a
32
L. Lasagna
compound for which the specter of carcinogenicity has been raised by animal studies. Such follow-up surveys involve formidable technical and economic problems, to say nothing of the expense.
Regulatory Aspects Let me close with a few examples that illustrate some of the problems in specific terms. First, saccharin. This is the only artificial sweetener available on the U.S. market, yet its removal has been recommended by both consumerists and the FDA Commissioner. Indeed, if Congress had not intervened, saccharin would almost certainly have been banned by now. Yet an expert advisory committee to the FDA Commissioner concluded that the case against saccharin was unproven, and that responsible scientists could render a verdict neither of "innocent" nor "guilty". The committee proposed exactly what you might expect in such a situation - another study. So, on January 25, 1978, the FDA and the National Cancer Institute announced plans for an 18-month, $1,375 million case control study involving 9000 persons. The study, I am told, has already been criticized, and I cannot see how this can do anything but add one more study to the several already in the literature. Why should this study be expected to settle the issue? Should we not, instead, be asking ourselves about the collective costs of obesity and the benefits of saccharin - whatever they may be - as an aid to "girth control"? In 1973 the U.S. Consumer Product Safety Commission reported an association between some spray adhesives and birth defects. Six months later, the ban was withdrawn, but only after considerable anguish in the minds of many pregnant women and their spouses about the need for abortions to prevent the birth of a possibly malformed child. Several women did, in fact, undergo abortion without justification [5]. A similar epidemic of needless anxiety followed the spurious report, later retracted, of an association between tricyclic antidepressants and congenital anomalies. Our government has decreed that Depo-Provera (injectable medoxyprogesterone) can be shipped overseas but not used in the U.S. Why? Let me quote Commissioner Kennedy: " . . . the benefits of Depo-Provera used in many situations exceeds the risk, but not for the U.S. We have a variety of alternatives available." Thus one of the few effective ways to prevent conception in promiscuous feeble-minded women is denied us, and it will be only a matter of time before were are indicted again for the need to consider abortions in such retarded citizens. In other words, science and society are wrong to use this progestogen to prevent pregnancy, but they are also wrong to allow retardates to become unintentionally pregnant[ Freireich has recently recounted [4] his experiences with an antimetabolite derivative which was found to kill animal tumors while it had no effect on normal cells. FDA said that it could not be given to humans until preclinical toxicity testing was done. Because it was a racemic compound, the alpha form (the one of interest) was purified at great expense. The world's total supply was then given to animals and produced no toxicity. FDA rejected the plans to proceed because the toxic dose was
Toxicological Barriers to Providing Better Drugs
33
not known. One can readily imagine similar bureaucratic demands being made of those who wish to pursue enkephalin or endorphin research in man. In conclusion, I hope that I have made clear some of the toxicological obstacles that I see to the development of better drugs. The obstacles can be classified as economic, regulatory, and conceptual, although these categories are obviously interrelated and interdependent. It is not true that more and more toxicity testing is necessarily better for the sick of the world. We must be both realistic and honest in our appraisal of the costs and benefits of such testing, and select those approaches that seem to represent the best long-term interests of the public.
References 1. Briggs, M.: The beagle dog and contraceptive steroids. Life Sci. 21, 275-284 (1977) 2. Committee 17 Report: Environmental mutagenic hazards. Science 187, 503-514 (1975) 3. Environmental Carcinogenesis Subcommittee: Draft of criteria for establishingthe carcinogenicity of chemicals. Drug Research Reports 18, No. 48, Supplement 5-2-4 (1975) 4. Freireich, E. J.: The bottleneck in anticancer drug development. Med. World News, March 20, 1978, p. 103 5. Hook, E. B., Healy, K. M.: Consequences of a nationwide ban on spray adhesives alleged to be human teratogens and mutagens. Science 191, 566-567 (1976) 6. McCann, J., Choi, E., Yamasaki, E., Ames, B. N.: Detection of carcinogens as mutagens in the Salmonella/microsometest: assay of 300 chemicals. Proc. Natl. Acad. Sci. USA 72, 5153-5139 (1975) 7. McCann, J., Ames, B. N.: Detection of carcinogens as mutagens in the Salmonella/mierosometest: assay of 300 chemicals; discussion. Proc. Natl. Acad. Sci. USA 73, 950-954 (1976) 8. Riley,V.: Mouse mammary tumors: alteration of incidence as apparent function of stress. Science 189, 465-467 (1975) 9. Salsburg, D.: The use of statistics when examining life-time studies in rodents to detect carcinogenicity. J. Toxicol. Environ. Health 3, 611--628 (1977) 10. Salsburg, D.: Are carcinogenicity tests useful? In: Controversies in Therapeutics (Lasagna, L., ed.). W. B. Saunders, Philadelphia(in press, 1979) 11. Sher, S. P.: Tumors in control mice: literature tabulation. Toxicol. Appl. Pharmacol. 30, 337--359 (1974) 12. Special Article. Towards a more rational regulation of the developmentof new medicines. Eur. J. Clin. Pharmacol. 11, 233--238 (1977) Received October 30, 1978
TOXICOLOGY
Arch. Toxieol. 43, 27-33 (1979)
9 Springer-Verlag 1979
Toxicological Barriers to Providing Better Drugs Louis Lasagna Departments of Pharmacology and Toxicology,and Medicine, University of Rochester School of Medicine and Dentistry, Rochester, New York, 14642, USA
Abstract. The unmet needs of the sick demand that toxicologic requirements do not stifle the rational search for new and better remedies. A number of conceptual problems hamper the rational use of toxicological testing. These include: a misplaced confidence in the value of animal testing, a failure to make sophisticated risk-benefit analyses, the proliferation of new tests of uncertain validity, and improperly executed retrospective case control studies. Regulatory barriers include the ever increasing bureaucratic demand for toxicological testing, the unseemly willingness of regulatory agencies to yield to hysterical or cynical consumer group pressures, the unreasonable demand for "superiority" of new products before the granting of registration, and the temptation to institute expensive but untested post-marketing surveillance schemes. Economic obstacles to new drug development have become formidable, and new demands for toxicologic studies in animals and humans are adding to these problems. Finally, some examples of unwise regulatory decisions involving saccharin, spray adhesives, Depo-Provera, and a new anti-metabolite are given. Key words: Toxicology - Mutagenicity - Carcinogenicity cals.
Pharmaceuti-
Introduction In the United States, and to some extent also in other countries, the toxic effects of chemicals seem to preoccupy the public, the news media, and the politicians. Much less attention is paid to the need for new and better medicines, although such need is painfully obvious in any doctor's office or hospital. Despite the great progress made by the Pharmaceutical Revolution that began a few short decades ago, there is no area of medicine that does not suffer by reason of inadequate drugs. Many cancer patients can expect little from our present therapies.
0340-5761/79/0043/0027/$ 01.40
28
L. Lasagna
Schizophrenics would welcome chronic treatment that does not carry with it the risk of irreversible neurologic damage. Severely afflicted arthritics find present remedies far less effective and more toxic than desired. Microorganisms can be expected to spawn strains that are resistant to widely used antibiotics. The muscular dystrophies do not respond significantly to available drugs. The list goes on and on, to the frustration of the sick, their relatives, and physicians, who want to help but often cannot. While few scientists would disagree with the proposition that patients need better drugs than those now available, there is considerable controversy over the amount and nature of toxicologic testing in animals and man that is optimal. I propose to examine some of the conceptual, economic, and regulatory aspects of such testing, with a particular view to identifying toxicologic requirements that cannot be logically defended.
Conceptual Problems Let me begin with some conceptual problems. A prime difficulty is the misplaced confidence that many place in animal testing. Such tests have long been extremely useful in detecting toxic effects that are dose related and not species specific. But they cannot be relied upon to detect toxicity that does not possess these attributes. This pathetic illusion gives rise - quite logically if you believe the premise - to the idea that simply doing enough animal testing will predict all human toxicity. A second conceptual difficulty arises from the growing failure of certain groups in our society to conduct risk-benefit analyses in determining whether or not to pursue a drug in man. The essence of toxicity testing is to delineate the extent and nature of the risk associated with a drug as one varies dose and duration of treatment, and thus to calculate the risk not only under conditions of massive ingestion, but at those levels adequate to produce the desired therapeutic benefit. For certain kinds of toxicity, such as carcinogenicity, the attitude of some seems to be that any risk of cancer, no matter how remote, should rule out the use of a drug in man. In the case of food additives, the Delaney Amendment has embodied this concept in law in the United States. A third conceptual difficulty relates to the willingness of the scientific community to accept, as a basis for rejection of drugs for human use, new tests of unknown specificity or validity simply to satisfy the needs that we all feel for ways of predicting teratogenicity or carcinogenicity in man. The Ames test and a variety of other procedures for mutagenicity and carcinogenicity assessment are now being employed everywhere in the world. The operating procedure generally has become that such tests - if positive - will ordinarily toll the death knell of a chemical. This despite the fact that the validity of such predictions -- perhaps especially those based on bacterial systems - is far from clear. Advocates of the bacterial tests for mutagenicity point out that since these procedures are not expensive to run and do not take much time, thousands of chemicals can be screened yearly. They assert that 90% or so of carcinogens are also mutagens, and that false positives occur perhaps only 15% of the time [6, 7].
Toxicological Barriers to Providing Better Drugs
29
In contrast, a distinguished committee appointed by the Council of the Environmental Mutagen Society said the following three years ago [2]: 1. " . . . only the intensive accumulation of comparative information can indicate the reliability of extrapolation procedures . . . . " 2. " . . . no information exists proving or disproving the existence of a threshold response to a mutagen in any system . . . . " 3. " . . . the identification of a compound as a mutagen is only the first step in estimating the hazard it may pose to man . . . . " 4. " . . . quantitative extrapolation from microbial systems, or from reversion tests in which tissue cultured mutant mammalian cells are used, does not yet appear to be possible . . . . " 5. " . . . the test systems now available do not provide a very broad base for screening, either in their abilities to detect all forms of heritable mutations or in their relevance to mutation in man. We therefore urge that a number of new systems be developed . . . . " Later in that same year, the environmental carcinogenesis subcommittee of the National Cancer Advisory Board pointed out that "the short-term (or in vitro) tests cannot be used to define a carcinogen." The subcommittee further warned of certain positive results even in experimental animal tests which in themselves would not be definitive evidence of carcinogenicity. These included assays in which the strain of animals had a greater than 10% spontaneous frequency of a particular tumor; or in which known carcinogens or other foreign material or modifying factors were used in addition to the test compound; or in which test animals were subjected to unphysiologic conditions [3]. Riley, also in 1975, reported the fantastic differences that could be achieved in C3H/He strain mice carrying an oncogenic virus by manipulating the type of housing and handling [8]. These observations suggest the need for great caution in the design and conduct of earcinogenicity tests in rodents. But to go back to the Ames test. It is acknowledged that false negatives would not be surprising, and hence "passing" the Ames test is considered to provide only slight assurance about a drug's safety, so that more classical tests using mammals are usually required to "establish" a drug's safety. Hence these new quick and inexpensive tests now often serve to kill a drug, but not to grant it developmental life. Will such a philosophy yield harm to society? No one knows. To test the potential losses would require money and time, and involve legal and ethical problems. What we do know is that if a compound does not get to market, we will never be able to assess its true potential for human benefit (or harm). We have plenty of examples of drugs that produced pleasant (as well as unpleasant) surprises after registration, when new and important uses were found by accident for them - lidocaine and phenytoin for cardiac arrhythmias, diazepam for status epilepticus, estrogens and progestins for contraception, etc. Clearly the earlier in a drug's development the decision is made to kill a drug, the more information will be lost about the drug's potential. Carcinogenicity testing also suffers from a failure to consider the logical statistical consequences of lifetime studies in animals known to develop cancers spontaneously. Control Swiss-Webster mice, for example, have been reported in three sepa-
30
L. Lasagna
rate 14-month studies to show tumors in 15 to 26% of animals. The majority of these tumors were malignant [11]. The incidence of tumors went up to 28 to 37% in three separate 24-month studies. Salsburg has analyzed the implications of such data [10]. Let me quote him: During the past four or five years, both new and old drugs have been subjected to a toxicology "screen" involving a procedure which has never been fully evaluated to determine its relevance to man, which has a high probability of producing false positive results, and which adds about a quarter of a million dollars to the cost of a drug's development and up to three years additional time on to that development. Not only is the procedure expensive and of unknown value, it also violates fundamental principles of good biological experimentation and it can easily produce results which cannot be analyzed without serious statistical bias and a resulting confusion between direct carcinogenesis and extraneous biological events. This procedure has cast suspicion onto such marketed drugs as a widely used sedative (phenobarbital), the first line of defense in the treatment of tuberculosis (isoniazid), a standard treatment for trichomoniasis (metronidazole), a potent diuretic (spironolactone), and a whole class of psychotherapeutic agents (the phenothiazines). A poll taken at the Biopharmaceutical subsection of the American Statistical Association meetings in 1976 indicated that, up to that time, at least 10 new drugs had been withdrawn from investigation in the U.S. due to findings from such studies against less than 30 such studies completed. Using a Monte Carlo computer simulation technique, Salsburg has calculated that 20-month feeding experiments in mice will yield false positive conclusions about carcinogenicity at least 50% of the time, as will two-year studies in male rats. His cogent analysis [9, 10] is, in my opinion, required reading, especially since even our best statisticians seem unable to grasp the notion that running one or more control groups will not solve all of these problems. Briggs has recently criticized the mandatory use of beagle clogs for chronic toxicity testing of contraceptive steroids [1]. He believes that the beagle, unlike rodents or monkeys, tends to falsely incriminate progestogens that are in fact safe for the human. Let me give one additional conceptual problem - the tendency to rely on human case control studies to assess the capability of a drug to produce rare serious side effects. The case control method can be highly useful in such instances, but it has certain inherent dangers, especially the possibility of bias in drug history, in case selection, and in the construction of a control group. It now appears, for example, that the widely publicized studies incriminating reserpine as a cause of breast cancer were a fluke of some sort, and that Rauwolfia alkaloids probably do not cause breast cancer. A similar controversy now rages over the suggestion that estrogens may cause endometrial cancer. There are regulatory barriers as well. Regardless of what science might recommend as defensible toxicology requirements, there is a growing tendency to require more and more tests, regardless of their cost or value. For the regulatory bureaucrat, this poses no problem, since he is rarely criticized for failing to approve a drug, but is often berated for having registered a drug whose toxic potential turns out to be greater than anticipated.
Toxicological Barriers to Providing Better Drugs
31
Recently, a consumerist group prodded our Secretary of Health, Education and Welfare into invoking, for the first time, the so-called "imminent risk" provision to withdraw phenformin from the market. Whatever one's feelings may be about this drug, it is hard to see what new scientific insights at that particular moment in history created the need for action. The relationship of phenformin to some cases of lactic acidosis had been known for years, and the true incidence is no better known today than it has ever been. What this incident shows, however, is the power of public agitation, even by a group with no large constituency, to move politicians. A troublesome new development in the United States is the tendency to disapprove a new drug application because a drug is "no more effective than older drugs but more toxic". At In'st glance this seems sweetly reasonable. But is it? How many anti-epileptics would be off the market if we used these guidelines? How many anticancer drugs? Is it not abundantly clear that the optimal practice of medicine often requires individualization of therapy, and that one man's best drug may be different from another's? Do we not need "backstop" drugs whose use can be invoked when first- or second-line drugs prove either ineffective or toxic? One other aspect of toxicologic scrutiny is becoming more important. I refer to the growing tendency to demand, perhaps routinely, postmarketing surveillance of new drugs [12]. I have long believed that we now neglect the proper study of drugs after registration, and that such postmarketing scrutiny is crucial because many troubles simply cannot be detected prior to marketing. The "naturalistic" study of drugs in ordinary clinical use can potentially tell us a great deal about the overuse, underuse, misuse and abuse of drugs that cannot be easily studied (if at all) in the premarketing phases. It is in the marketplace that we should be quantifying - as best we can - the benefits as well as the costs of drugs. Therefore in principle I am in favor of moves in this direction. What bothers me, however, is the haste to make such studies mandatory when we do not know how best to do them, and the temptation to do them only to check on toxicity rather than on both toxicity and benefit. These studies can be enormously expensive if large numbers of patients are to be followed carefully, and it would be a pity to delude ourselves into believing that rare or long-delayed side effects are going to be readily detected by such schemes.
Economic Aspects The economic obstacles to new drug development have become formidable. Our own studies suggest that $55 million and 7 - 8 years are required in the U.S. to move a drug to market from the inception of human studies. As a result, effective patent life is probably more like 9 - 1 0 years than 17. Every new toxicity test that is added on as a requirement will add to both the length and the expense of drug development. At the same time, there are pressures all over the world to constrain the prices of drugs and to limit exclusivity in the market place. Is it possible that these various trends may summate to diminish adventuresomeness and creativity in the drug industry? It has been suggested that follow-up studies should be conducted - perhaps for life - in humans who have received even small doses, for short periods of time, of a
32
L. Lasagna
compound for which the specter of carcinogenicity has been raised by animal studies. Such follow-up surveys involve formidable technical and economic problems, to say nothing of the expense.
Regulatory Aspects Let me close with a few examples that illustrate some of the problems in specific terms. First, saccharin. This is the only artificial sweetener available on the U.S. market, yet its removal has been recommended by both consumerists and the FDA Commissioner. Indeed, if Congress had not intervened, saccharin would almost certainly have been banned by now. Yet an expert advisory committee to the FDA Commissioner concluded that the case against saccharin was unproven, and that responsible scientists could render a verdict neither of "innocent" nor "guilty". The committee proposed exactly what you might expect in such a situation - another study. So, on January 25, 1978, the FDA and the National Cancer Institute announced plans for an 18-month, $1,375 million case control study involving 9000 persons. The study, I am told, has already been criticized, and I cannot see how this can do anything but add one more study to the several already in the literature. Why should this study be expected to settle the issue? Should we not, instead, be asking ourselves about the collective costs of obesity and the benefits of saccharin - whatever they may be - as an aid to "girth control"? In 1973 the U.S. Consumer Product Safety Commission reported an association between some spray adhesives and birth defects. Six months later, the ban was withdrawn, but only after considerable anguish in the minds of many pregnant women and their spouses about the need for abortions to prevent the birth of a possibly malformed child. Several women did, in fact, undergo abortion without justification [5]. A similar epidemic of needless anxiety followed the spurious report, later retracted, of an association between tricyclic antidepressants and congenital anomalies. Our government has decreed that Depo-Provera (injectable medoxyprogesterone) can be shipped overseas but not used in the U.S. Why? Let me quote Commissioner Kennedy: " . . . the benefits of Depo-Provera used in many situations exceeds the risk, but not for the U.S. We have a variety of alternatives available." Thus one of the few effective ways to prevent conception in promiscuous feeble-minded women is denied us, and it will be only a matter of time before were are indicted again for the need to consider abortions in such retarded citizens. In other words, science and society are wrong to use this progestogen to prevent pregnancy, but they are also wrong to allow retardates to become unintentionally pregnant[ Freireich has recently recounted [4] his experiences with an antimetabolite derivative which was found to kill animal tumors while it had no effect on normal cells. FDA said that it could not be given to humans until preclinical toxicity testing was done. Because it was a racemic compound, the alpha form (the one of interest) was purified at great expense. The world's total supply was then given to animals and produced no toxicity. FDA rejected the plans to proceed because the toxic dose was
Toxicological Barriers to Providing Better Drugs
33
not known. One can readily imagine similar bureaucratic demands being made of those who wish to pursue enkephalin or endorphin research in man. In conclusion, I hope that I have made clear some of the toxicological obstacles that I see to the development of better drugs. The obstacles can be classified as economic, regulatory, and conceptual, although these categories are obviously interrelated and interdependent. It is not true that more and more toxicity testing is necessarily better for the sick of the world. We must be both realistic and honest in our appraisal of the costs and benefits of such testing, and select those approaches that seem to represent the best long-term interests of the public.
References 1. Briggs, M.: The beagle dog and contraceptive steroids. Life Sci. 21, 275-284 (1977) 2. Committee 17 Report: Environmental mutagenic hazards. Science 187, 503-514 (1975) 3. Environmental Carcinogenesis Subcommittee: Draft of criteria for establishingthe carcinogenicity of chemicals. Drug Research Reports 18, No. 48, Supplement 5-2-4 (1975) 4. Freireich, E. J.: The bottleneck in anticancer drug development. Med. World News, March 20, 1978, p. 103 5. Hook, E. B., Healy, K. M.: Consequences of a nationwide ban on spray adhesives alleged to be human teratogens and mutagens. Science 191, 566-567 (1976) 6. McCann, J., Choi, E., Yamasaki, E., Ames, B. N.: Detection of carcinogens as mutagens in the Salmonella/microsometest: assay of 300 chemicals. Proc. Natl. Acad. Sci. USA 72, 5153-5139 (1975) 7. McCann, J., Ames, B. N.: Detection of carcinogens as mutagens in the Salmonella/mierosometest: assay of 300 chemicals; discussion. Proc. Natl. Acad. Sci. USA 73, 950-954 (1976) 8. Riley,V.: Mouse mammary tumors: alteration of incidence as apparent function of stress. Science 189, 465-467 (1975) 9. Salsburg, D.: The use of statistics when examining life-time studies in rodents to detect carcinogenicity. J. Toxicol. Environ. Health 3, 611--628 (1977) 10. Salsburg, D.: Are carcinogenicity tests useful? In: Controversies in Therapeutics (Lasagna, L., ed.). W. B. Saunders, Philadelphia(in press, 1979) 11. Sher, S. P.: Tumors in control mice: literature tabulation. Toxicol. Appl. Pharmacol. 30, 337--359 (1974) 12. Special Article. Towards a more rational regulation of the developmentof new medicines. Eur. J. Clin. Pharmacol. 11, 233--238 (1977) Received October 30, 1978
