Review of the Fialuridine (FIAU) Clinical Trials (1995)

A large number of new drugs have been added to the therapeutic armamentarium during the past three decades. They have cured, controlled, or ameliorated a variety of human diseases; their use has saved the lives of numerous individuals and positively affected the quality of life of many others. These drugs have been used successfully to suppress rejection following organ transplantation and to treat a variety of infectious diseases, hypertension, various forms of heart disease, peptic ulcer disease, rheumatoid arthritis, asthma, and prostatic hypertrophy, to mention just a few. Physicians prescribe these medications extensively, and the public gladly accepts the benefits that accrue from their use. However, before they can be approved for use considerable testing for safety and efficacy must be performed.

Clinical trials are of vital importance in the development of new drugs for use in humans. Preliminary experimentation in animals is important in identifying potentially effective interventions in animal models human diseases and in identifying the efficacious and nontoxic amounts of a pharmaceutical to be given. However, no amount of testing in animals can substitute for carefully planned and carefully conducted studies in humans. In clinical studies in humans a particular medication or intervention is administered to individuals with a specific disease under conditions in which beneficial or detrimental effects can be identified during a reasonable period of observation. There are many examples of medications that appear to be effective or ineffective in animal models but have been demonstrated to have the opposite effects in humans. For specific medications or interventions, certain side-effects in humans may apply to humans generally or only to selected individuals. There are always certain risks in the administration of new medications for the first time, since efficacy cannot be assumed and the relevance of side effects can usually be established only on the basis of the human trial. Individuals who agree to participate in clinical trials must be adequately informed about the rationale for the use of a particular drug or intervention and about potential benefits and adverse effects that might occur. Moreover, the individuals should understand that some side effects that occur may not be predictable and may be serious. Unfortunately, there is no alternative to careful evaluation of new medications and interventions in human trials if one is to gain the needed knowledge concerning their efficiencies and possible undesirable effects.

While accepting the benefits from all the new drugs introduced in the past 4 to 5 decades, it is important to recognize what is required in the way of subject participation in the clinical investigation of such new drugs. As an example, consider antimicrobial agents. In Phase I trials (pharmacology; dosage studies) recent practice has involves about 100-300 subjects (Gilbert, 1987). In Phase II trials (controlled clinical trials; relative safety/efficacy),

300-1,000 subjects have usually been involved. In Phase III trials (expanded trials, both controlled and uncontrolled; effectiveness, specific indications, and precise definition of side effects), recent practice involved 1,000-3,000 subjects. The estimated time to complete such trials is 4 to 7 years (Phase I; 1-2 years; Phase II, 1-2 years; Phase III, 2-3 years). Overall, the total number of new commercial applications submitted to the Food and Drug Administration (FDA) by sponsors each year is large. For example, 371 new applications were submitted in 1992 (FDA, 1993c). Of each 100 drugs for which applications are submitted to the FDA, about 70 successfully complete Phase I trials, about 33 complete Phase II trials, and 25-30 clear Phase III trials. About 20 of the original 100 are ultimately approved for marketing.

In a review of clinical design characteristics in studies of new antimicrobial agents reported in various symposia over the 17-year period from 1969 through 1985, a total of 207 individual trials involving 21 different antimicrobial agents were described (Gilbert, 1987). If one applies the number of subjects commonly studied in each clinical investigation of a new antimicrobial from 1969 to 1985 to the 21 different antimicrobial agents approved by the FDA in that period, then a sizeable number of subjects, somewhere between 29,400 and 90,300, have been involved in the clinical trial process for approval of these new antimicrobial drugs. The process is thus a lengthy one (4-7 years) and one which involves many study subjects. The time involved is lengthy because of the necessary stepwise nature of study progression. The argument can be raised that the trial duration can be too lengthy, particularly when the underlying disease is a life-threatening one for which therapies are limited. This has been raised, for example, in the case of clinical trials of nucleoside analogs and other drugs investigated for therapy of human immunodeficiency virus (HIV) infection. One must balance safety considerations and the need for a particular treatment. Similar considerations must apply as well in defining therapeutic efficacy for a new drug and markers of adverse effects during clinical trials. If too stringent a requirement for efficacy is applied, ultimately useful drugs may be discarded prematurely. Similarly, if adverse effects are defined inappropriately, then introduction of useful drugs may be delayed or proscribed. If the most stringent criteria are applied to adverse events then it may be possible to avert any toxicity. That is certainly an important consideration; some (e.g., Jonas, 1970) would consider it an overriding obligation. However, the other side of the coin must be recognized in tallying up the pros and cons. While avoiding a single serious adverse reaction is an obviously desirable goal, the premature termination of a study for laboratory test abnormalities of uncertain significance may also have undesirable effects by causing the cessation of further development of that drug. The evaluation of this step should consider the potential saving of one life by stopping the further testing of the drug but should also consider the alternative, loss of life by prematurely holting development of a drug that may save many lives. This consideration should necessarily be weighed in the evaluation of adverse events in the course of clinical drug testing for all new drugs, whatever the type (antimicrobial, cardiovascular, psychotropic, etc.)

Thus, there are many precedents in clinical medicine for the urgent need to develop improved therapies for patients with serious illnesses. There are definite risks to such patients as they receive new medications and interventions, and these need to be acknowledged by FDA and involved physician-scientists and to be well understood by the patient. Nevertheless, there have been major clinical gains from improved therapy. These gains have led to reductions in morbidity and mortality on the basis of observations made in clinical trials; they simply could

not have been made without accepting the inherent risks in conducting the necessary clinical trials.

Before the advent of trials, treatments were judged to be safe and effective via personal observation and anecdote. George Washington is reported to have had his demise hastened by blood letting for the treatment of hypertension (Donaldson and Donaldson, 1980; Knox, 1933). The treatment was judged to be beneficial on the basis of logic and clinical impression.

Surgeons in the days of Ambrose Paré treated gunshot wounds with boiling oil. It was his treatment of choice on the field of battle to capture the castle of Vilaine in 1537. Fortunately for those to follow the battle was so heated as to have caused Paré to deplete his supply of boiling oil. He was forced to an alternative—an ointment made of egg yolk, oil of roses, and turpentine. To his great surprise he found that those so treated fared better than those treated with boiling oil. He wrote:

It would have been unthinkable in Paré's day to have subjected the treatment of boiling oil to test in a trial. To have done so almost certainly would have been at odds with the prevailing wisdom of the time and would have likely exposed him to wilting criticisms of his peers. Had he had the vision to want to evaluate it in the context of a trial before the Battle of Vilaine in 1537, he most likely would have been castigated for experimenting on his patients. The fact is, however, that his patients would have been better off had he had the vision and wherewithal to do the evaluation. They would have been exposed to less pain and suffering than by having waited for the events of the minute to cause him to change his treatment approach.

A trial is an experiment carried out on human beings for the purpose of determining whether or not a specified treatment regimen is safe or effective. A clinical trial is such an experiment performed on people having some defined health condition or disease and that is undertaken to determine whether or not the indicated health condition or disease can be cured, ameliorated, or prevented.

One of the things that sets clinical trials apart from other members of the class of trials is the obligation for care. Investigators undertaking them have obligations to ensure that the patients enrolled in the trial are adequately cared for, even if doing so conflicts with or is at odds with required data collection and treatment procedures set forth in the protocol for the trial. In short, their first responsibility is to the patients who enroll in the trial and in ensuring that the patients has adequate care, even if meeting that responsibility results in violations of

the treatment protocol and discontinuation of the patient's participation in the trial. Indeed, investigators are expected to deviate from the protocol if doing so is considered in the best interest of the patient. The dual responsibility of meeting the requirements of the Hippocratic Oath and a study protocol creates a tension that, of necessity, should settled in favor of individual patients when there is conflict. The goal is to design the study protocol so as to minimize such conflicts by the inclusion of safeguards to allow investigators to meet the requirements of the protocol while also meeting their responsibilities under the Hippocratic Oath.

The protocol specifies details as to the route of administration, dose and dosages, and the points at which treatments are to be suspended or altered, depending on the outcome. The administration may involve a single dose or multiple doses. The period of treatment may consist of a single exposure to the drug or repeated exposures. The exposures may involve use of the same dose administered according to a specified schedule or differing dose administered according to some scheme (e.g., as in a crossover trial in which each person enrolled receives a low or a high dose followed by a high or a low dose).

Only patients judged to be eligible (as determined via defined eligibility criteria) may be enrolled in the trial, and among those, only those who consent can be enrolled or those for whom permission to enroll is obtained from appropriate representatives. Persons are under no obligation to enroll or to continue once they are enrolled. Investigators are required to make these facts known to all who are approached for enrollment prior to enrollment and the start of treatment as part of the consent process.

The amount of data collected on each person enrolled depends on the nature of the drug being tested, the nature of the persons enrolled (i.e., whether or not they themselves have the disease or condition for which the drug is intended), and the period of time that persons are to be under the care and observation of study investigators. Typically, that period is relatively short in the case of Phase I and II drug trials and is usually measured in days or weeks as opposed to months or years. As a rule, it ends when a person is separated from the trial.

All trials involve data collection at various time points over the course of follow-up. The requirement for repeated observation obligates patients to a defined data collection schedule (except for trials done in hospitals or other settings involving resident patients or trials that can be performed in one session) that calls for a series of visits to the site of the trial for observation and treatment. The first visit or series of visits will be for the purpose of determining eligibility, collecting essential baseline data, obtaining consent, and the initiation of treatment. Visits thereafter will be for continued administration of treatment and/or collection of essential follow-up data. The schedule of follow-up visits will be timed from the point of initiation of treatment and will be on a defined time schedule (e.g., once every week), with provisions for interim (unscheduled) visits when they are required or deemed necessary in caring for those enrolled.

Judgments regarding the merits of treatment are made in different ways depending on the design employed and on the kinds of outcome measures used to judge safety or efficacy. All trials, even if they do not involve a control treatment, involve assessment of change for individual patients or the collection of patients by subtraction of a person's baseline value from the value observed at a specified point during the course of follow-up. For example, one

monitors for weight change by comparing a patient's weight on entry with that observed over the course of follow-up.

Evaluation of the treatment is based on results from all patients enrolled in the trial. That evaluation is relatively straightforward in randomized controlled trials involving comparison of one or more test treatments versus a control treatment. The control treatment in the case of drug trials may be a placebo, or it may be a standard medical treatment or another test treatment. Typically, the judgment is based on comparison of the different treatment groups with standard statistical tests on the outcome or outcomes of interest.

Judging the safety or efficacy of a treatment is much more problematic in the case of trials not involving a designed comparison group—usually the case in most early evaluations of a new drug. The problem in those settings is compounded by the fact that those trials are typically of short duration and are small in size. That problem is most acute in cases, such as with fialuridine (FIAU), in which drugs are being tested on people with a life-threatening disease that results in its own morbidity and increased risk of death. In those settings one is often left in a quandary as to what to make of isolated cases of morbidity indicated by observed clinical events or intimated by changes in specified laboratory tests. Are the changes due to the disease itself or to the drug? Even the cause of death becomes difficult to interpret in the presence of a relatively high background death rate from the disease or other characteristics of those enrolled. There is always room for doubt as to cause. Was it the natural outcome of the disease or was it induced by the treatment? The issue is rarely clear until in retrospect sufficient information has accumulated to discount underlying disease as the likely explanation or until one sees an unusual clustering of deaths and morbid events like those in trial H3X-MC-PPPC.

No matter how promising the theory or impassioned the pleas for early use in humans, all potential drugs must go through a series of tests in animals before they can be tested in human beings. The object of these studies is to evaluate the potential toxic properties of the test drug in such a way that possible adverse reactions are identified before introduction of the drug into humans. Consequently, one is interested in (1) as complete an inventory as possible regarding potential toxic events, (2) the exposure conditions that are likely to lead to the appearance of toxicity, (3) the estimated toxic dosage range in laboratory animals, and (4) an estimate of the likelihood that adverse events might occur in human subjects. The testing procedures are designed in such a way that toxicity will occur at some dosage level in laboratory animals; relatively large doses are used intentionally in an attempt to elicit the complete inventory of adverse effects as accurately as possible. There are no standard established protocols for preclinical toxicity testing. However, it is essential to have knowledge regarding the possible adverse effects that might occur when the drug is administered acutely (administered once only) or repetitively (multiple treatments over days, weeks, or months). Furthermore, some assessment of potential adverse effects of the test drug on genetic material (DNA) is required, as well as its adverse effects on reproduction (fertility, fetal viability, fetal development, etc.)

It is necessary to establish the various dose-effect relationships that exist when the drug is administered acutely or repetitively (chronically). Toxicologists wish to establish at which dose levels adverse events are observed. One also wishes to determine the dose levels where drugs are unlikely to exert adverse effects (the so-called no-observable-adverse-effect-level). Finally, one wishes to establish dosing regimens for humans that will be devoid of adverse effects.

Since there is no standard series of toxicity studies formally required before going into clinical trials, the final package is determined by the sponsor of the investigational new drug (IND) request and the FDA. The specific requirements for preclinical toxicity testing depend on the type of drug in question and the particular illness for which it is intended. For an antiviral agent like FIAU, it is likely that the FDA would require toxicity information on the test substance when it is administered acutely to at least two species. The route of administration employed in the tests ideally would be the route envisioned for use in the clinical treatment situation. Furthermore, drug effects after multiple-dose treatment of two species would be required. The duration of treatment in these studies in animals would be at least equal to the treatment period envisioned for the clinical trial. They could vary in length from a period of several weeks to a month or up to 6 months. In addition, it is likely that genotoxicity data as well as reproductive studies in laboratory animals would be required.

Compounds considered to lack promise for use in human beings do not come to testing in trials. They are dropped from considerations before reaching that stage. There must be a plausible basis for believing that compound will be useful in the treatment of human beings and sufficient preclinical data to indicate that the compound may be effective and that it is safe relative to the risks associated with the condition or disease of interest.

Testing of new drugs in human beings typically proceeds in three, and sometimes four phases, defined as follows:

Phase I: Usually the first stage in testing performed as part of an IND application to FDA, the trial is done primarily to generate preliminary information on the chemical action and safety of the new drug and is usually not controlled; that is, it is done without the benefit of a concurrently observed comparison group that does not receive the drug. Subjects are most often normal volunteers, but sometimes patients with the target disease.

Phase II: Usually the second stage of testing, the trial is carried out on persons having the disease or condition of interest, and thus provides preliminary information on the efficacy of the drug along with additional information on safety. The design may include a control treatment and random assignment of patients to treatment.

Phase I/II: A trial combining the features of both a Phase I and a Phase II trial, designed to provide preliminary information on both safety and efficacy.

Phase III: Usually the third and final stage of testing, this sort of trial is concerned with assessment of dosage effects, efficacy, and safety. The design usually includes a control treatment and random assignment to treatment. Once this phase is completed (or nearly completed) the drug manufacturer or sponsor may request permission to market the drug for the indication or condition covered in the testing. This is done by submitting a new drug application (NDA) to FDA.

Phase IV: A fourth stage of testing, carried out after approval of NDA, can take a variety of forms, from passive surveillance to carefully planned experimental studies. Goals may be simply examining long-term safety and efficacy, or the utility of the drug for an indication or population not specifically addressed during the IND period. Efficacy and safety are typically measured against a designated control treatment.

After approval drugs remain under surveillance to determine if they cause serious effects not detected during testing. Broadly referred to as postmarketing surveillance, this involves the collection of reports of adverse events via systematic reporting schemes, surveys, and observational studies.

Phase I and II trials by nature tend to be small and are typically performed by one or a few investigators located at the same institution or at a small number of institutions. Typically, the number of people studied in any individual trial will number in the tens as opposed to hundreds. The primary question in a Phase I trial is whether or not the drug can be safely administered to people and in a manner and at a level likely to be beneficial. As a result, trials in this class are generally noncomparative in the usual sense of that term. That is, as a rule they are not randomized trials because all persons enrolled receive the study drug. If there is any randomization it relates to the dosage administered or to the order in which different dosages of the same drug are given (e.g., as in the case of crossover designs). Furthermore, the period of treatment tends to be short, measured in days or weeks as opposed to months or years. As a rule, data collection and follow-up cease on or soon after the last administration of treatment.

The Phase II trial generally includes only slightly more subjects than a Phase I trials, and the period of treatment is only slightly longer than its Phase I counterpart. The main difference is in focus. As the emphasis moves from safety to efficacy, the emphasis also shifts from designs involving dose escalation via some systematic or Bayesian process to designs involving a fixed sample size design (sample size is usually determined by pragmatic considerations as opposed to formal calculations based on type I and II errors) and with each person enrolled receiving the same dosage or with different patients receiving different dosages. The designs are usually parallel treatment designs if more than one treatment regimen or dosage is studied.

The sample size and durations of treatment and follow-up tend to increase as one moves from Phase II to Phases III and IV trials. The latter two phases are likely to have sample sizes numbering in the hundreds. As a rule, they are controlled, have fixed sample size designs, parallel treatment structure, and protocols involving a fixed dosage schedule.

All trials, regardless of phase, are subject to review and approval by a local institutional review board (IRB) before implementation and are subject to periodic review by the IRB of record so long as they remain in force. Investigators undertaking trials have obligations and responsibilities to obtain such reviews and approval before implementation and to seek the review and approval of the responsible IRBs before implementing amendments to the protocol. They have a responsibility, as well, to inform the IRB of any untoward events in the conduct of the trial and to report to such boards any conditions or events believed to be related to the treatment.

All trials of the sort discussed above are done under an IND application held either by a named individual at a performance site or by the sponsor of the drug. The sponsor the is typically a drug company, but in the broader sense of usage of the term it is simply the person or agency holding an IND application regardless of location. Personnel at the performance site or sites are named on the Form 1572 accompanying IND applications. People named are expected to be familiar with FDA regulations in regard to the conduct of trials of an IND and are obliged to comply with those regulations. The regulations, among other demands, make a number of stipulations regarding informed consent by patients. FDA regulations on informed consent are similar, but not identical, to those of the U.S. Department of Health and Human Services and other agencies of the federal government. In addition, FDA regulations are very specific about reporting adverse events.

Under an IND application investigators are obliged to file safety reports to FDA for an adverse experience that is both serious and unexpected. The regulations (Code of Federal Regulations [CFR]; 21, parts 300 to 499, Revised 1 April 1992) specify that

In regard to telephone reports the regulations specify that:

An unexpected adverse experience is

A serious adverse experience, as described in the CFR for drugs is

Ethical justification of research involving human subjects requires responsiveness to the ethical norms or rules that are embodied in ethical codes and regulations. The substantive norms presented in ethical codes and regulations consist of various expressions of six general behavior-prescribing statements (Levine, 1986): There should be (1) good research design, (2) competent investigators, (3) a favorable balance of harms and benefits, (4) informed consent, (5) equitable selection of subjects, and, (6) in some ethical Codes but not in U.S. federal regulations, compensation for research-induced injury. In addition to these six substantive norms there are various procedural rules designed to ensure responsiveness to the substantive rules. Most important for the present considerations are that there should be (1) review and approval by an IRB and (2) written documentation of informed consent.

Institutional review board (IRB) review and approval of each protocol for the conduct of research involving human subjects are required by federal regulation for all research sponsored by any agency of the federal government and for all research regulated by FDA. Most institutions that conduct research involving human subjects have voluntarily extended the reach of the requirement for IRB review to all such research in their multiple project assurance MBA—documents filed with the Office for Protection from Research Risks of the National Institutes of Health that provide the details of the institution's plans to comply with federal regulations for the protection of human subjects. The purpose of IRB review is to ensure compliance with all of the other procedural and substantive ethical norms.

Federal regulations set forth a detailed account of the minimum standards for the composition, administration, reporting requirements, and substantive responsibilities of IRBs; most university hospital and research hospital-based IRBs exceed these minimum standards. For example, although federal regulations require at least five members, one of whom must be "primarily concerned" with nonscientific areas, the typical university hospital-based IRB has 20 or more members representing medicine, science, law, ethics, and other professions and other members who are not professionals (Levine, 1986).

Federal regulations also provide detailed requirements for the composition and use of consent forms. The requirement for consent forms is a procedural requirement designed to provide evidence of compliance with the substantive requirement for informed consent. Federal regulations specify circumstances in which the requirement for written documentation of consent may be waived; none are relevant to the FIAU research reviewed by the Institute of Medicine (IOM) Committee.

The leading international codes and guidelines in the field of research involving human subjects, The Nuremberg Code, The Declaration of Helsinki, and the International Ethical Guidelines for Biomedical Research Involving Human Subjects (promulgated in 1993 by the Council of International Organizations of Medical Sciences in collaboration with the World Health Organization each clearly presents the requirements for good scientific design and competent investigators as ethical requirements (Levine, 1986). This point notwithstanding, it is customary to dissociate the discussion of these matters from discussion of the other substantive ethical norms. This report reflects this customary division; considerations of scientific merit and investigators' competencies are discussed in other sections of this report.

Ethical justification of research requires a determination that the balance of risks and anticipated benefits is—in the words of federal regulations—"favorable" or "reasonable." this determination, which must first be made by the investigators and then ratified by the IRB, must be accomplished at the outset, before the research is begun. All too often, media addressed to the general public look at the results after the research has been completed and challenge the ethical justification on grounds that the "risks" were not justified by the benefits. At this point, what was identified at the outset as "risks''—a statement of probability that harms of a certain type could occur—either have or have not materialized as "injuries."

The process of ethical justification before the research is begun should always include an assumption that there is some statistical probability of injury and should generally include an attempt to calculate that probability. In general, the probability and magnitude of injury should be less than the probability and magnitude of anticipated benefit. It should always be understood, however, that adverse events with a low probability of occurrence will eventually occur if the conditions that create the risk of their occurrence are repeated sufficiently often. Thus, the fact that injury has occurred does not mean that the research project in which it occurred was unethical. Research is justified ethically by the relation of risks to anticipated benefits at the outset, not by the ratio of injuries to actual benefits at the conclusion of the research.

In research designed to evaluate a new drug, the initial estimation of risks must also take into account the possibility of occurrence of novel injuries—harms that cannot be

predicted on the basis of prior clinical or preclinical testing. Fortunately, injuries that are both serious and unanticipated are very unusual (Levine, 1986; Williams, 1990).

Responsiveness to this norm also requires that there be careful monitoring of all events that have a bearing on the assessment of both risks and benefits. It is the joint responsibility of the sponsor and the investigator to see to it that all reasonable measures are put in place to ensure the prompt identification of adverse reactions. With regard to research in the field of drug development, there are two major purposes of such activity: (1) to ensure the timely identification of adverse reactions so that appropriate interventions can be implemented in time to minimize injury to individual subjects; such interventions are exemplified by discontinuation of a subject's participation in the study and by the administration of antidotes to toxic agents; and (2) to ensure accurate estimates of the nature, probability and magnitude of side effects and adverse events (a) to inform subsequent judgments as to whether a drug should be approved for commercial distribution and, if so, (b) to determine what information should be put on the package label. An accurate assessment of benefits also contributes to the effective functioning of the drug approval and labelling process.

This ethical requirement is usually construed to mean that vulnerable persons should not be enrolled as subjects without suitable justification. "Vulnerable subjects" is a category often equated with those classes identified in federal regulations as the special populations, special in that they have "limited capacities to consent." Populations so identified in regulations are children, fetuses, and prisoners; in addition, regulations were once proposed for the protection of persons with limited capacities to consent by reason of mental infirmity. In general, members of the special populations may not be recruited as research subjects without special justifications. Such justifications include, but are not limited to, the following: the research goals could not be realized by using less vulnerable subjects; the research is designed to develop knowledge or therapeutic products that will be of benefit to the vulnerable class from which the subjects are recruited; the assent of the subject must be obtained (when feasible and appropriate) along with the permission of a parent or other legally authorized person; and interventions or procedures that do not hold out the prospect of direct benefit to the individual and that present more than minimal risk require even more stringent justification.

According to The Nuremberg Code, the first international code of research ethics, there are four criteria for valid consent: it must be voluntary, legally competent, informed, and understood. Most commentators on research ethics agree that persons having diminished capacity to understand or to refuse participation should be regarded as at least potentially vulnerable even if they do not have the attributes that define the special populations (see earlier discussion). To the extent that such persons are actually vulnerable, careful consideration should be given to applying some or all of the special justifications specified earlier for the recruitment of vulnerable persons as research subjects, even though such justifications are not explicitly required by regulations (Levine, 1986).

Particularly germane to the IOM committee's mandate is a recognition that persons with serious and potentially disabling or deadly diseases are vulnerable in several important respects.

They may lack sufficient voluntariness—in the words of The Nuremberg Code, they may not be "so situated as to be able to exercise free power of choice." They may be willing to assume unreasonable risks out of desperation to find a way to prevent or postpone death or disability. They may fear that if they do not cooperate with their doctors or with the health care establishment they might be subjected to some sort of retaliation. Even though they are assured, as required by federal regulations (45 CFR 46.116 (a)(8), that they are free to refuse to participate in research or to withdraw without the loss of any benefits to which they are otherwise entitled, they may fear retaliation. Scientists and IRBs must be alert to these possibilities as they attempt to insure that patients make free and informed choices.

Another condition that undermines the capability of a person to make an unconstrained choice is poverty. Many people with serious chronic diseases are actually or potentially impoverished. Without the free medical care that is commonly made available to research subjects, many would suffer the plight of the medically indigent. Accordingly, one must be especially attentive to avoid what are commonly called "undue inducements." A frequently used guideline calls for limiting cash payments to vulnerable research subjects to the minimum wage for the actual time spent plus reimbursement for out-of-pocket expenses such as parking fees and stipends for baby-sitter (Levine, 1986).

Several prestigious groups advisory to the federal government have recommended that there should be "no-fault compensation" for research-induced injury (Levine, 1986). These groups include but are not limited to the U.S. Department of Health, Education, and Welfare Task Force on the Compensation of Injured Research Subjects (1977), the National Commission for the Protection of Human Subjects of Biomedical and Behavioral Research (1976), and the President's Commission for the Study of Ethical Problems in Medicine and Biomedical and Behavioral Research (1982). These recommendations notwithstanding, the federal government has limited its involvement in compensation to a requirement that informed consent must include "an explanation as to whether any compensation and an explanation as to whether any medical treatments are available if injury occurs and, if so, what they consist of, or where further information may be obtained" (45 CFR 46.116(a)(6).

The committee has already addressed the procedural standard that requires, in most research protocols, the signing of a consent form. Now attention is turned to the process of informed consent. The process of informed consent is designed to be responsive to the individual's right to self-determination. Through informed consent individuals are empowered to protect their own interests as they are provided with all of the information that is considered "material" to their judgments as to whether to enroll in a research study. The consent form, by contrast, is a document designed primarily to protect the interests of the institution and the investigator. It is, in effect, a signed receipt for the information (Levine, 1986).

The consent form consists of the minimum amount of information that the IRB finds necessary to divulge during the informed consent process. This minimum standard is actually quite comprehensive in that it includes all of the categories identified in the relevant federal regulations. Investigators are permitted to present more information but not less than what is on the document. Commonly, they present additional information in response to questions asked by the subjects.

When subjects are asked to recall what they were told during the process of informed consent, they quite often are unable to repeat many of the details. This is the case even when there is incontrovertible evidence that they have received and acknowledged a comprehensive account of all of the information deemed necessary by the IRB. Some commentators believe that this incomplete recall is due to a loss of memory, particularly of bits of information that subjects consider not worth knowing. This belief is supported by the results of research. For example, in studies in which subjects are given incentives to recall the consent information, their retention of information increases substantially in proportion to the value that they assign to the incentives (Levine, 1986).

Some commentators believe that the subjects' poor performances on tests of recall may reflect a trusting attitude on the part of the subjects. At some point in the encounter between an investigator and a prospective subject, the latter decides whether or not he or she trusts the investigator. If the subject does trust the investigator the subject becomes inattentive to the details of risks and benefits and chooses to participate on the basis of trust. If the subject does not trust the investigator, the prospective subject similarly becomes inattentive, having already decided to refuse on the basis of a lack of trust. Although the committee is not aware of any definitive study on the role of trust in the decision to consent or to refuse to participate in research, members of the committee with experience in recruiting research subjects believe that this is an important factor.

Clinical trials are of vital importance in the development of new drugs for use in humans. Preliminary animal experimentation is important in identifying potentially effective interventions and eliminating potentially dangerous ones, but no amount of animal testing can substitute for carefully planned and carefully conducted human studies in which a particular medication or intervention is administered to individuals with a specific disease under conditions where beneficial or detrimental effects can be identified during a reasonable period of observation.

Testing of new drugs in human beings typically proceeds in three, and sometimes four phases, beginning with attempts to generate preliminary information on the chemical action and safety of the new drug, generally using normal volunteers, through preliminary studies of efficacy carried out on persons having the disease or condition of interest, to large scale formal studies including a control treatment and random assignment of dozens or hundreds of patients to treatment.

The process is thus a lengthy one (4-7 years) and one which involves many study subjects. The time involved is lengthy because of the necessary stepwise nature of study

progression. The argument has been raised that the trial duration can be too lengthy, particularly when the underlying disease is a life-threatening one for which therapies are limited, e.g., AIDS. One has to balance safety considerations and the need for a particular treatment. If too stringent a requirement for efficacy is applied, ultimately useful drugs may be discarded prematurely. As is now, despite intensive preclinical screening, only about 70 of every 100 drugs for which applications are submitted to the FDA successfully complete phase I trials. About 33 complete phase II trials, and 25-30 clear phase III. About 20 of the original 100 are ultimately approved for marketing.

One of the features of clinical trials that distinguishes them from other experimentation with human subjects is the obligation for care. Investigators undertaking them have obligations to ensure that patients enrolled are adequately cared for, even if doing so is at odds with required data collection and treatment procedures set forth in the protocol for the trial. We have therefore include in our overview not only the FDA regulations governing reporting of adverse events, but also an ethical framework which we found useful in evaluating the FIAC\FIAU trials. This framework can be summarized by six substantive imperatives and two procedural norms. There should be 1) good research design, 2) competent investigators, 3) a favorable balance of harms and benefits, 4) informed consent, 5) equitable selection of subjects, and 6) compensation for research-induced injury. In addition to these six substantive norms, there should be 1) review and approval by an institutional review board and 2) written documentation of informed consent.