In the past several decades there has been exponential growth in the number of clinical trials conducted to test innovations in the treatment of disease. A search of the 1999 Medline database alone found reports of 19,587 such trials (Meinert, 2000). In addition, in the past 10 years clinical trials of drugs and other interventions have become more than a process required to judge the safeties and efficacies of potential treatments. They
What Is a Clinical Trial?
A clinical trial is defined as a prospective study comparing the effect and value of intervention(s) against control in human beings (Friedman, Furberg, and DeMets, 1996).
A controlled experiment having a clinical event as an outcome measure and done in a clinic or clinical setting and involving persons having a specific disease or health condition (Meinert, 1996).
An experiment is a series of observations made under conditions controlled by the scientist. A clinical trial is an experiment testing medical treatments on human participants (Piantadosi, 1997).
The term clinical trials may be applied to any form of planned experiment which involves patients and is designed to elucidate the most appropriate treatment of future patients with a given medical condition (Pocock, 1984).
have also become part of the health care system, in that patients often view participation in a clinical trial as their best hope of achieving a cure (Zivin, 2000).
Although there are different interpretations of the term “clinical trial,” in general, it is defined as an experiment designed to assess the safety or efficacy of a test treatment (Meinert, 2000) ( Box 1-1). Clinical trials refer to planned experiments that involve human participants and that are designed to elucidate the most appropriate treatments for future patients with a given medical condition (Pocock, 1984). Perhaps the most essential feature of a clinical trial is that it aims to use results for a limited sample of patients to make inferences about how treatment should be administered to the general population of patients who will require therapy in the future (Pocock, 1984).
WHEN THE STANDARD APPROACH TO CLINICAL TRIALS IS NOT FEASIBLE
Adequately powered randomized clinical trials (RCTs) and double blind RCTs are generally regarded as the most authoritative research methods for establishment of the efficacies of therapeutic interventions. By allocating sufficient numbers of individuals to groups—for example, an experimental or a control group—investigators can estimate or determine with some degree of certainty the effect of a given intervention.
However, when the available population of research participants does not allow the conduct of an RCT with adequate statistical power, there might still be a need to design and perform clinical research (e.g., treatments are not available for a rare disorder or a unique patient population or studies require the participation of patients with terminal or severely debilitating or incapacitating disorders). Some distinctive research populations—such as astronauts or members of a small, isolated community—may consist of less than five individuals. For example, a study focused on assessing the effects of microgravity on bone mineral density loss during space missions would have to rely on data for a few individuals (see Box 1-2). This report defines this research situation as a small clinical trial and explores the various design and analytical strategies one might consider to approach a small clinical trial.
Obtaining sufficiently large control groups for research with small numbers of participants can be difficult for research involving individuals with severe, debilitating, or incapacitating conditions, and the use of untreated or placebo control groups can raise ethical dilemmas (Altman, 2000; Delaney, 2000; Emond, 2000). Historically, drug developers and federal regulators have been wary of small clinical trials for a number of reasons, but primarily
Case Study: Effects of Long-Term Microgravity Exposure on Weight-Bearing Bones of Cosmonauts
Microgravity-induced bone mineral density (BMD) loss was first suspected in the 1970s, but systematic investigations with astronauts and cosmonauts did not commence until the 1990s. Investigation is made difficult because there are few potential participants and the conditions of microgravity, in terms of length of exposure, vary with the lengths of space missions and the availability of individual astronauts or cosmonauts to participate in clinical investigations while they are on those missions.
Observational studies suggest that the loss in BMD may be on the order of 1 to 2 percent per month during extended exposure to microgravity. Recently, Vico et al. (2000) reported measurements of BMD at the distal radius and tibia in 15 cosmonauts on the Russian Mir space station who had sojourned in space for 1 month (n = 2), 2 months (n = 2), or 6 months (n = 11) since 1994. BMD was measured before launch and the week after landing. Each cosmonaut who spent at least 2 months in space was allowed a recovery period equal to the duration of the corresponding space mission.
The findings demonstrate striking interindividual variations in BMD responses and indicated that the BMD of neither cancellous bone nor cortical bone of the radius was significantly changed at any time. In the weight-bearing tibial site, however, cancellous BMD loss was already present after the first month and deteriorated with mission duration. In tibial corticies, BMD loss was noted after a 2-month mission. In the group who had been in space for 6 months, cortical BMD loss was less pronounced than cancellous BMD loss. In some individuals the tibial BMD deterioration was marked. Variations in tibial BMD loss were also large during recovery, and the loss persisted in some individuals. These studies have found a mean BMD loss of about 2 percent after 6 months of space travel, with a standard deviation of approximately 1.2 percent. If the clinically important effect of an intervention that limits the total BMD loss to 1 percent is to be detected, the total sample size needed is 50 (significance level = 5 percent with power of 90 percent [as determined by a one-sided t test]). With only three to six astronauts per flight and one or two long missions per year, it would take up to 10 years to evaluate a single intervention in a traditional clinical trials.
The challenge is to conduct clinical trials with small populations to evaluate the efficacies of various interventions for the prevention of BMD.
because of their lack of statistical power and generalizability ( Table-1.1). Thus, in general, a small clinical trial is conducted because of external constraints, not necessarily by choice. Nonetheless, the general requirements for small clinical trials are no different than those for adequately powered “large clinical” trials; that is, they must be sufficiently designed and appro priately analyzed to provide a reasonable measure of the effect of an intervention. They should be designed to have an outcome measure for determination of success, a baseline measure that can used to determine changes, and a means to monitor the changes (Meinert, 2000). Because of the design
and analysis constraints of small clinical trials and because of uncertainties inherent to small clinical trials, it is likely that they will require at least as much—and probably more—thought than traditional, large clinical trials.
In some cases, however, properly designed small clinical trials can contribute to substantial evidence of efficacy; however, those conclusions may require the use of assumptions and inferences given the paucity of data (Siegel, 2000). Small clinical trials may successfully be used to study diseases or conditions with a well-described natural history with little variation; when sensitive pharmacodynamic effects are directly related to pathophysiology; when good nonhuman models are available; and when the intervention has a large effect on efficacy, produces a predictable relationship between measurable drug levels and effects, and has been applied to a related condition (Siegel, 2000) ( Table 1-2). Traditionally, small studies are more likely to be conducted to test surgical procedures than to test drugs (Delaney, 2000; Emond, 2000) ( Box 1-3). They are least likely to be useful for the study of complex disease syndromes with highly variable outcomes (e.g., some chronic diseases such as arteriosclerotic cardiovascular disease), for drugs with less than dramatic effects in vitro, for illnesses in which correlates of success are unclear, in situations in which the risk of short-term death is high, and for surgical procedures for which there are many complex and confounding factors (Delaney, 2000; Faustman, 2000; Mishoe, 2000).
A SEARCH FOR ALTERNATIVES
New approaches to protocol design are needed for trials with small sample sizes that can assess the potential therapeutic efficacies of drugs,
biologics, devices, and other medical interventions. For example, a possible alternative is to assess the therapeutic results in a single treated population by sequentially measuring whether the intervention results in outcomes that
Case Study: Clinical Trial of Organ Transplantation in HIV-Positive Individuals
Human immunodeficiency virus (HIV)-positive individuals now live long enough to be considered candidates for cardiac or liver transplantation (Gow and Mutimer, 2001). A very small number of HIV-positive organs become available for transplantation into HIV-positive individuals. Whether the immunosuppressive agents required for organ transplantation will accelerate or slow the progression of HIV infection or AIDS is an unresolved research question. A study is underway to evaluate the outcomes of transplants of hearts and livers from HIV-positive donors into HIV-positive recipients. There are a small number of candidates for such a study, and the only alternative is to not do a transplant. In that event, death is the ultimate outcome since long-term alternative supports—for example, dialysis for renal failure—are not available for patients with liver failure.
The research question to be answered in a clinical trial is whether transplantation will benefit patients with HIV infection or AIDS with a diminished life expectancy from cardiac or liver failure due to their impaired organs if they do not receive a transplant? The ideal study should determine the risk of transplantation (and associated interventions) on the progression of HIV infection or AIDS and its effect on life expectancy. The projected life expectancy is approximately 6 months without transplantation, and only 5 percent of the participants are expected to be alive after 12 months. The 1-year survival rate after standard heart and liver transplantation in non-HIV infected individuals approaches 90 percent. A survival rate of 25 percent at 1 year in the HIV-infected or AIDS patients who receive a transplant would be a clinically important advance. With a significance level of 0.05 and a power of 0.9, the sample size needed to detect an increase in the 1-year survival rate from 5 to 25 percent is 150. At the current rate of organ availability, such a study would take many years.
fall above or below a preestablished probability range for an efficacious outcome. Such a clinical trial could be considered to have demonstrated efficacy when the cumulative observed results fall within or above the prescribed confidence range, or the trial could be stopped when the cumulative observed effect falls below the preestablished level of confidence ( Box 1-4). A major question, however, for this and other approaches is whether the science base of alternative methods alone or in combination is sufficiently developed for these nonrandomized clinical trials to be effective in demonstrating efficacy in studies with small sample size.
It has been recognized for some time that RCTs—although highly desirable—are neither practical nor feasible as a means of answering all clinical research questions. A variety of other methods, such as non-RCTs, observational methods, naturalistic studies, and case-control studies, have been used in clinical investigations. In addition, there has been increasing discussion over the past decade about the value of measuring surrogate markers rather than traditional clinical endpoints in clinical trials.
Case Study: Clinical Trial for Treatment of Sickle Cell Disease
Sickle cell disease is a red blood cell (RBC) disorder that affects 1 in 200 African Americans. Half of all individuals living with sickle cell disease die before age 40. The most common complications include stroke, renal failure, and chronic severe pain. Patients who have a stroke are predisposed to having another one.
Mixed donor and host stem cell chimerism (i.e., the sickle cell host recipient has a cellular constitution made up of both the donor and the recipient subject's cells) is curative far sickle cell disease (Krishnamutri, Blazar, and Wagner, 2001). Only 20 percent donor RBC production (with 80 percent recipient RBC production) is required to cure the abnormality. Conditioning of the recipient is required for the bone marrow transplant to be successfully established. The degree of human leukocyte antigen (HLA) mismatch, as well as the sensitization state (i.e., chronic transfusion immunizes the recipient), influences how much conditioning is required to establish 20 percent donor chimerism.
In patients who have an HLA-identical donor and who have not been heavily transfused, 200 centigrays of total body irradiation (TBI) is sufficient to establish donor engraftment. This dose of irradiation has been shown to be safe and well tolerated. In heavily transfused recipients who are HLA-mismatched, more conditioning will probably be required. The optimal dose of TBI for this cohort has not been established. The focus of this hypothetical study is to establish the optimum dose of TBI dose to achieve 20 percent donor chimerism in patients enrolled in the protocol ( Chapter 3).
In 1990, the Institute of Medicine (IOM) published Medical Intervention at the Crossroads: Modern Methods of Clinical Investigation, which discussed the benefits and drawbacks of non-RCTs. However, the issue of when and how to conduct a small clinical trial continues to challenge many areas of biomedical science.
IOM COMMITTEE PROCESS AND STATEMENT OF TASK
In response to the growing need for reliable and valid methods for clinical research with small populations, the National Aeronautics and Space Administration requested that the IOM undertake a study of strategies for small-number-participant clinical research trials. However, the term “small number” may convey different meanings to different constituencies; consequently, the committee prefers to use the phrase “small clinical studies” or “small clinical trials.”
The committee, consisting of individuals with expertise in biostatistics, clinical pharmacology, clinical research, ethics, and research design, was charged with the following specific tasks:
Assess the current methodologies for conducting clinical trials with small populations. The analysis of methods used to conduct small clinical trials will include assessment of the published literature and commissioned papers on various strategies such as (1) meta-analysis to combine disparate information from several studies, including Bayesian techniques, as in the confidence profile method, and (2) other alternatives to RCTs, such as assessments of therapeutic results for a single treated population by sequentially measuring whether the outcomes from the intervention fall above or below a preestablished probability outcome range and meet predesigned specifications as opposed to incremental improvements.
Convene a 1-day conference during which participants from federal research and regulatory agencies, industry, academia, and other areas of clinical research and practice will discuss the progress being made in the strategies and the state of the science in the design, conduct, and evaluation of clinical trials of drugs, biologics, devices, and other medical interventions in populations with small numbers of individuals. Methods including RCTs, meta-analysis, decision analysis, and sequential clinical trial approaches will be considered in terms of their potentials and problems. The discussions will include, where possible, ethical and statistical evaluations and comparisons.
The committee, through consideration of background materials and presentations at the conference, will review the methodology for clinical trials with small populations and make recommendations for future research to continue development of this area of medical research.
ORGANIZATION OF THE REPORT
This report is organized around the two major issues in the performance of any clinical trial: design and analysis. Chapter 2 addresses the fundamental tenets of clinical trial design and how they are challenged and possibly addressed by studies with small numbers of participants. Chapter 3 focuses on several statistical approaches that can be used to analyze small clinical trials. Each chapter provides recommendations and suggests research needs.
It should be noted that the various strategies and methodologies presented in this report are by no means an exhaustive list of those methods that are applicable to small clinical trials. Indeed, other strategies not presented here may be useful on a case-by-case basis. Moreover, this is a developing area of research, and new approaches to small clinical trials will arise in the future and may well be in progress. Finally, the amount of attention paid to a particular area described in the report is not necessarily proportional to its importance or utility but, rather, may have been motivated by a particular example that the committee used to illustrate a small-sample-size clinical problem and potential solution. The committee's goal is to provide a balanced overview and analysis of various methods in use and to suggest new ones where appropriate.