Portal de Periódicos da CAPES

The 50th Anniversary of the Kefauver‐Harris Amendments: Efficacy Assessment and the Randomized Clinical Trial

2012; Wiley; Volume: 14; Issue: 11 Linguagem: Inglês

10.1111/jch.12012

1751-7176

J. Rick Turner,

Pharmaceutical industry and healthcare

It is possible for a practicing physician to initially think that randomized clinical trials (RCTs) are the province of clinical research scientists and physicians who act as principal investigators (being in charge of clinical trial execution at investigative sites at which trials are conducted), and not the province of those exclusively engaged in providing patient care. Given a few moments’ thought, however, an alternative viewpoint recommends itself. The importance of the RCT to all physicians is that clinical research informs clinical practice and evidence-based medicine, and practicing physicians can benefit from having sufficient knowledge about RCTs to understand their role in placing new drugs within their treatment armamentaria and generating the evidence contained in treatment practice guidelines. This commentary does not presume to make any reader an expert in the methodology underpinning the RCT; rather, it aims to increase awareness of the central importance of RCTs in the practice of biopharmaceutical medicine and to acknowledge the Kefauver-Harris Amendments as a powerful engine that drives the development of the statistical and ethical sciences supporting the contemporary RCT. Since hypertension is the most common chronic disease in the world,1 and there are now many antihypertensive agents approved for the treatment of hypertension,2 this journal seems a particularly suitable venue in which to offer the following commentary. The Kefauver-Harris Amendments to the US Federal Food, Drug, and Cosmetics Act of 1938 (FFDCA) were signed into law by President Kennedy a half-century ago on October 10, 1962. The signing into law of the FFDCA, which revised the Pure Food and Drug Act of 1906 and had been under congressional discussion for several years, was galvanized by the elixir sulfanilamide tragedy in which more than 100 people were fatally poisoned by an ingredient in the elixir following its ingestion. While all such acts are more complex than can be fully described in a short article, the fundamental requirement introduced by the FFDCA was the provision of compelling evidence of a drug’s safety at the time of application for marketing approval. Similarly succinctly, the central novel requirements of the Kefauver-Harris Amendments concerned the provision of compelling evidence of efficacy. Signing of the amendments into law was spurred on by the thalidomide tragedy in Europe. Thalidomide was first marketed in 1956 in Germany for the treatment of insomnia and vomiting in early pregnancy. In 1961, a sizeable increase in the incidence of congenital birth defects was noted. These defects were typically an absence or reduction of the long bones of the limbs with normal or rudimentary hands and feet. The association of these conditions with thalidomide was not recognized for several years after the drug was marketed, and several thousand babies in Europe were born with this congenital condition. These events led to the UK’s Medical Act of 1968. While public awareness of the events in Europe was a powerful motivator for congressional action for further drug law reform, the United States did not experience the same tragedy from the use of thalidomide. Having seen the reports from Europe, Dr Frances Kelsey, a newly appointed reviewer at the US Food and Drug Administration (FDA), undertook considerable research and, as a result, took a firm stance against the drug’s approval. In recognition of her diligence she was awarded the President’s Award for Distinguished Federal Civilian Service 2 months before the amendments were signed.3 The RCT (more accurately but less commonly called the randomized concurrently controlled clinical trial, because a concurrent control treatment arm is a fundamental aspect of the trial’s comparative nature) became the key methodology for providing the compelling evidence of efficacy required by the amendments. The amendments and the RCT can therefore be regarded as very closely related in the history of modern biopharmaceutical medicine. Derenzo and Moss4 captured the importance of ethical considerations in all aspects of clinical studies as follows: Each study component has an ethical aspect. The ethical aspects of a clinical trial cannot be separated from the scientific objectives. Segregation of ethical issues from the full range of study design components demonstrates a flaw in understanding the fundamental nature of research involving human subjects. Compartmentalization of ethical issues is inconsistent with a well-run trial. Ethical and scientific considerations are intertwined. From the scientific perspective, an inappropriate study design and/or a poorly conducted trial is generally incapable of answering a research question. Participants voluntarily take part in clinical research with the understanding that their participation may not benefit them directly (eg, they may be randomized into a placebo treatment arm), but it will provide information that will be useful to a much larger group of people should the drug be approved for marketing and made available to perhaps millions of patients. This is one of the “benefits” that is weighed against the “risks” of participants’ exposure to a drug under development. From an ethical perspective, therefore, three consequences of poorly designed and conducted clinical trials become clear. First, if the trial does not permit the best possible information to be obtained, the participants’ expectations have been violated. Second, a poor trial may lead to a drug failing to be approved for marketing when in reality it is safe and effective. Patients who would have benefited from the drug will therefore be denied the opportunity to receive it. Third, a poor trial may lead to a drug being approved when in reality it is not acceptably safe, where the term “acceptably safe” refers to a favorable benefit-risk balance despite risks being present.5,6 The amendments played a seminal role in the creation of clinical research ethics. The practice of informed consent and the creation of institutional review boards, for example, are directly traceable to them, and ongoing discussions led to the Belmont Report, which addressed the “basic ethical principles” of respect for persons, beneficence, and justice.7 Biological considerations are at the heart of biopharmaceutical drug development in several ways. First, clinical trials investigate topics of clinical relevance, and clinical relevance is intimately related to biological relevance. The ultimate goal in this field of clinical research is to develop a new compound that is biologically active, acceptably safe, well tolerated, and useful in the treatment of biological states that are or may become of clinical concern. A second central importance of biological considerations is individual variation: not everyone responds to the same drug in the same way, a statement that remains true when factors such as sex, age, weight, and ethnicity are taken into account. This observation provides a direct link between biological science and statistical science in the setting of clinical research. One reason for individual variation in drug response is individual variation in drug metabolism. The tricyclic antidepressant nortriptyline provides an illuminating example of such genetic influence. The majority of patients taking this drug require around 75 to 100 mg/d to reach the intended steady-state blood plasma concentration. Poor metabolizers (whose abnormal genetic variant enzymes break down the drug less efficiently than usual) require only 10 to 20 mg/d. At the other end of the continuum, ultra-rapid metabolizers (whose abnormal variant enzymes break down the drug more efficiently than usual) require around 300 to 500 mg/d to achieve the same blood concentration.8 A second reason why individuals respond differently to the same drug is genetic differences in the structure of the target receptor, the biological structure (usually a macromolecule) with which the drug is intended to interact to produce the desired beneficial response. Additionally, there can be genetic differences in the structure of off-target receptors, biological structures with which the drug is not supposed to interact. Such interaction leads to adverse drug responses, the severity of which can vary among individuals. The process of randomization involves randomly assigning participants in RCTs to one of various treatment arms in a parallel-groups clinical trial (where each participant participates in just one of the treatment arms: in a cross-over design each participant completes all treatment arms). The goal of randomization is to control for the many potentially confounding influences that cannot be controlled for (eg, height and weight) or cannot be determined by simple visual observation (eg, the nature of an individual’s metabolic pathways and on-target and off-target drug receptors). That is, the goal is to minimize to the greatest extent possible any selection bias in participant assignment to the treatment groups. In statistical language, participants in the trial have an independent chance of receiving the different treatment arms. The simplest design is one in which half of the participants are randomized into the drug treatment arm and the other half are randomized into the control treatment arm. The randomization procedure occurs after a potential participant’s eligibility for a clinical trial has been determined (ie, the study protocol inclusion and exclusion criteria have been satisfied by that individual) and before the individual provides any trial data. As for other test statistics, F has to reach a certain size for the result of the analysis to attain statistical significance. This size varies from trial to trial and is dictated by the total number of participants included. However, it can be stated for all cases that, to attain statistical significance, an F value must always be >1 (unity). If the difference between the mean response in the drug treatment group and the mean response in the control treatment group, that is, the drug’s treatment effect, attains statistical significance, the process of randomization has facilitated the provision of compelling evidence that administration of either the test drug or the control drug to each participant, ie, a systematic influence on responses, has influenced drug (biological) responses to a degree that is over and above any difference that could be ascribed to chance factors alone. That is, compelling evidence of the drug’s efficacy has been provided. Fisher’s work on scientific experimental methodology was initiated in the field of plant breeding and agriculture, another domain in which natural variation is evident and randomization is therefore necessary. References to more of his writings11,12 and to introductory conceptual and computational statistics books discussing experimental research methodology13,14 are provided. Professor Paul Meier was very influential in the development of RCTs. In an instructive 1975 paper15 he discussed issues that were controversial at the time, noting that randomization continued to be one of the most controversial: Randomization is seen by the classically trained statistician as an almost automatic requirement for a scientific experiment, and it is frequently seen as an act of ethical irresponsibility by the classically trained physician. Various issues surround this statement, but one of the most straightforward is concern over why anyone (eg, researchers conducting a clinical trial in which the control drug is a placebo, which is pharmacologically inactive) would want to give some participants a drug that may well help them while giving others placebo. In clinical practice, a patient would never be administered a treatment that was known to be inferior to another available treatment. How, then, are RCTs in which the control drug is a placebo to be ethically justified? The answer to this question is that, at the time the trial starts, it is not known whether the drug is superior in efficacy to a placebo: the whole reason for conducting the trial is to determine whether the drug is superior. Thus, the participants who receive the placebo are not knowingly being given an inferior treatment. This state of affairs is called clinical equipoise.16 Clinical research is quite different from clinical practice. While the latter focuses on providing the best possible care to an individual patient by implementing a treatment regimen specifically tailored to that patient, clinical trials are conducted for the greater good. As noted previously, participants voluntarily take part in clinical research with the understanding that their participation may not benefit them directly, but it will provide information that will be useful to a much larger group of people should the drug be approved for marketing and then available to perhaps millions of patients, including individuals who participated in clinical trials of the drug and were randomized to a placebo arm. In the same paper discussed in the previous section, Meier15 also commented as follows: If the considerations are those of informativeness and efficiency, I see no substitute for randomized studies in many areas of clinical research. We may elect to forego new knowledge of various kinds, rather than face up to the ethical problems inherent in clinical experiments, but in many cases we will be unable to get that knowledge in other ways. It is fair to say that the generalizability of results obtained in tightly controlled RCTs to patient care in real-world settings has been the topic of much discussion over the years, and both the challenges and the benefits of postmarketing trials and pharmacovigilance (both the traditional passive type and more recent active paradigms) have been highlighted by various authors. In his 2007 paper entitled “In Defense of Pharmacoepidemiology: Embracing the Yin and Yang of Drug Development,” Avorn17 commented as follows: We forget how difficult it was to establish the rules of the road for conducting randomized trials. In terms of design theory and public policy, drug-epidemiology research is now where randomized trials were in the 1950s. We have much to learn about methods, transparency, and protecting the public’s interest. But that work can be done, and we often have no other way of gathering vital insights. While there are limitations to the RCT in the full spectrum of learning everything we want to know about a drug and its influence on the millions of patients for whom it may potentially be prescribed, and complementary approaches from the discipline of pharmacoepidemiology should be acknowledged here, the RCT brings a uniquely powerful tool to the table. By authoritatively championing its employment, Dr Meier has left us with a rich legacy.18 In his 1999 textbook, Matthews19 commented as follows: Over the last two to three decades, randomized concurrently controlled clinical trials have become established as the method which investigators must use to assess new treatments if their claims are to find widespread acceptance. The methodology underpinning these trials is firmly based in statistical theory, and the success of randomised clinical trials perhaps constitutes the greatest achievement of statistics in the second half of the twentieth century. This section notes some milestones in the development of RCTs. Credit for conducting the first pharmaceutical RCT is often given to Sir Austin Bradford Hill and his 14 colleagues (the Streptomycin in Tuberculosis Trials Committee, chaired by Dr Geoffrey Marshall) for his work in the late 1940s on the United Kingdom Medical Research Council’s (MRC’s) trial of the effects of the aminoglycoside antibiotic streptomycin on pulmonary tuberculosis.20 The control group treatment consisted of bed rest (the standard of care at the time) and the streptomycin group treatment consisted of bed rest plus intramuscular administration of 2 g/d of streptomycin, given in 4 injections at 6-hour intervals. Compelling evidence of efficacy was provided, and streptomycin subsequently became the first antibiotic treatment for this disease. Control groups had certainly been used in medical research prior to this trial but often the method of allocating participants to one of two treatment groups was alternate allocation, simply placing the next individual entering the trial in the alternate treatment group to the one entered by the previous individual.21 The method of allocating (randomizing) participants to either treatment group on this occasion was as follows:20 Determination of whether a patient would be treated by streptomycin and bed-rest (S case) or by bed-rest alone (C case) was made by reference to a statistical series based on random sampling numbers drawn up for each sex at each centre by Professor Bradford Hill; the details of the series were unknown to any of the investigators or to the co-ordinator and were contained in a set of sealed envelopes, each bearing on the outside only the name of the hospital and a number. After acceptance of a patient by the panel, and before admission to the streptomycin centre, the appropriate numbered envelope was opened at the central office; the card inside told if the patient was to be an S or a C case, and this information was then given to the medical officer of the centre. Additional informative commentary on this trial is provided by Yoshioka21 and by other authors in a 1998 issue of the British Medical Journal published 50 years following the original publication of the trial’s results. Tuberculosis remains a global heath issue of considerable proportions and a very active area of clinical research. Fortunately, as Lienhardt and colleagues22 recently observed, “a portfolio of promising new compounds for the treatment of tuberculosis is on the horizon.” In the same issue of the Journal of Infectious Diseases, authors from the MRC’s Clinical Trials Unit23 noted that innovative trial designs should be considered to speed drug and combination treatment regimen development for the treatment of tuberculosis. The sophisticated designs they discussed bear witness to the continuing evolution of the RCT. Cook and DeMets24 observed that the era of modern clinical trials in the United States can be regarded as beginning with the Coronary Drug Project (CDP), which ran from 1966 to 1975.25 The trial was sponsored by the National Heart Institute, which later became the National Heart, Lung, and Blood Institute (NHLBI, which is part of the National Institutes of Health [NIH]). The trial’s focus was secondary prevention, and 8341 men between the ages of 30 and 64 years who had recently survived myocardial infarction were randomized to 1 of 6 treatment arms that included 5 active treatments and a placebo, with all participants also receiving the standard of care of treatment at that time. The active treatments were representative of the drugs used at the time: 2.5 mg/d of conjugated estrogens, 5.0 mg/d of conjugated estrogens, 1.8 g/d of clofibrate, 6.0 mg/d of dextrothyroxine, and 3.0 g/d of niacin. The participants were followed for the occurrence of another cardiovascular event such as death or a second heart attack. Fifty-three investigative sites were included, along with a coordinating center and many committees, including a Steering Committee and a Data and Safety Monitoring Committee. The two estrogen treatment arms and the dextrothyroxine treatment arm were terminated before the end of the trial because of adverse effects experienced by the participants in them. No evidence of efficacy was found for clofibrate, while niacin showed modest benefit in decreasing nonfatal recurrent myocardial infarction but did not decrease total mortality. A 15-year follow-up study26 found that mortality from all causes for participants who had received niacin was 11% lower than those who were in the placebo treatment arm (52.0% vs 58.2%; P=.0004). The authors commented that “This late benefit of niacin, occurring after discontinuation of the drug, may be a result of a translation into a mortality benefit over subsequent years of the early favorable effect of niacin in decreasing nonfatal reinfarction or a result of the cholesterol-lowering effect of niacin, or both.” In the years following the start of the CDP, the NHLBI started several other large trials using similar trial methodology to investigate other major cardiovascular risk factors,27,28 and then to study treatments for lung and blood diseases. NIH statisticians also shared the concepts and principles of the RCT with other institutes at NIH, including the National Cancer Institute. Large industry-sponsored RCTs did not become common until the late 1980s and 1990s.24 As an example, consider the Scandinavian Simvastatin Survival Study (4S).29 The trial was designed to evaluate “the effect of cholesterol lowering with simvastatin on mortality and morbidity in patients with coronary heart disease (CHD).” A total of 4444 patients with angina pectoris or a previous heart attack who also had high serum cholesterol levels and were on a lipid-lowering diet were randomized to the simvastatin treatment group or the placebo treatment group. Compared with individuals receiving standard of care, simvastatin produced highly significant reductions in the risk of death and morbidity in individuals with CHD followed for a median of 5.4 years. Importantly, “The improvement in survival produced by simvastatin was achieved without any suggestion of an increase in non-CHD mortality…. No previously unknown adverse effects were apparent in this trial. Thus, the substantial and sustained reduction of total and LDL cholesterol in the simvastatin group was not associated with any serious hazard.”29 In addition to providing compelling evidence of efficacy, RCTs can also demonstrate a lack of compelling evidence of efficacy even though previous investigations in a drug’s development program had suggested such evidence would be found. Torcetrapib was a novel cholesteryl ester transfer protein (CETP) inhibitor that was demonstrated to inhibit the development of atherosclerosis in nonclinical studies (a rabbit model) and in early-phase clinical studies to increase high-density lipoprotein cholesterol between 60% and 100% while at the same time lowering low-density lipoprotein cholesterol by up to 20%.30 Based on “conventional wisdom” regarding these two cardiovascular safety biomarkers (higher high-density lipoprotein cholesterol is good, and lower low-density lipoprotein cholesterol is good), this evidence suggested a cardioprotective effect of torcetrapib. The Investigation of Lipid Level Management to Understand its Impact in Atherosclerotic Events (ILLUMINATE) trial therefore tested the proposition that torcetrapib would reduce the risk of clinical cardiovascular events. However, torcetrapib was associated with an increased risk of major cardiovascular events and also increased mortality (from both cardiovascular and noncardiovascular causes). The drug’s sponsor terminated ILLUMINATE prematurely at the recommendation of the trial’s independent Steering Committee based on advice from the trial’s Independent Data and Safety Monitoring Board.31 In recent years, meta-analyses have attained increasing prominence in the evidence-based medicine literature. Meta-analysis facilitates a quantitative evaluation of the evidence provided by two or more individual trials that have addressed the same research question. It commonly involves the statistical combination of summary statistics from various trials (study-level data, ie, the study’s treatment effect and the variance associated with it are extracted from the published report of each trial), but it also refers to analyses performed on the combination of participant-level data (a more powerful strategy when possible). The conceptual basis of meta-analysis is straightforward: more data provide a better opportunity to get an optimum-quality answer to a research question. However, Turner and Durham32 commented as follows: If all of the components involved in conducting a meta-analysis are performed appropriately, and the extent to which the results are helpful is not overstated (ie, any limitations are appropriately acknowledged and shared whenever and wherever communicating the results), the results can be informative and instructive. Unfortunately, however, it is easier than one might suspect to conduct a meta-analysis inappropriately and then to overstate the results in a variety of circumstances. Kay33 expressed his concerns more colorfully, noting that to ensure that a meta-analysis is scientifically valid, the analysis must be planned and conducted in an appropriate way, and that “It is not sufficient to retrospectively go to a bunch of studies that you like the look of and stick them together!” Therefore, while an analysis is certainly conducted, the term meta-methodology usefully captures everything else required to provide a reliable answer.34 Different authors have differing views on the relative strength of evidence provided by individual clinical trials and meta-analyses, and there is not space enough here to debate this issue. Suffice it currently to make two observations. First, for evidence-based medicine to be practiced, sound evidence is required. Well-conducted individual trials and well-conducted meta-methodology and meta-analysis can both provide sound evidence. Second, the way in which the results of any research investigation employing either experimental methodology or meta-methodology are communicated to the scientific community, practicing physicians, and, increasingly, the general public is of great importance. Turner34 observed that “following the publication of their article in a journal, some meta-analysts disseminate their findings in the mass media with a bravado that markedly departs from calm, scientific, and clinical discourse, and seemingly with the expectation that the nation’s physicians will change their practice of medicine immediately.” Fortunately, many others are more judicious. Once drugs have been approved by regulators, and particularly when there are multiple classes of drugs available for prescription, evidence-based practice guidelines issues by professional societies and reputable organizations can be of considerable assistance to physicians in clinical practice.2 However, it ultimately falls upon individual physicians to make treatment decisions that are in their patients’ best interests. Katz35 captured this sentiment as follows: If our patient is older than, younger than, sicker than, healthier than, ethnically different from, taller, shorter, simply different from the subjects of a study, do the results pertain?... All of the art and all of the science of medicine depend on how artfully and scientifically we as practitioners reach our decisions. The art of clinical decision-making is judgment, an even more difficult concept to grapple with than evidence.” The more critically practicing physicians can evaluate the published results of randomized clinical trials and meta-analyses, the better they will be placed to decide whether the evidence presented constitutes a solid foundation for an individual patient’s treatment. The Kefauver-Harris Amendments are a landmark driving force in the development of the RCT, including both scientific and ethical components, which, as noted previously, cannot be meaningfully separated. It is therefore appropriate to acknowledge the amendments at their 50th anniversary. The statistical underpinnings of the RCT have withstood the test of time, and physicians who practice evidence-based medicine are well served by an understanding and appreciation of their history and importance. Funding: The author reports no specific funding in relation to this research. No editorial support was used.