| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 University of Maryland School of Medicine, Baltimore, Maryland
Correspondence and requests for reprints should be addressed to Henry Silverman, M.D., M.A., University of Maryland School of Medicine, Department of Medicine, 10 S. Pine Street, Suite 800, Baltimore, MD 21201. E-mail: hsilverm{at}medicine.umaryland.edu
ABSTRACT
While the focus of ethical clinical trial conduct has been on protocol review in advance of the research, there has been a recent emphasis on monitoring subject welfare during the conduct of research. The ethical conduct of a clinical trial does not end with the formulation of study design or the obtainment of a signature on the informed consent form. This article defines and describes the necessary monitoring responsibilities to ensure the adequate protection of the rights and welfare of human subjects and the four parties who share such responsibilities: the institutional review board, the data monitoring committee (or the data safety and monitoring board), the sponsor, and the investigator. This article also reviews the numerous challenges associated with monitoring—such as overlapping responsibilities, communication gaps, and lack of standards—and attempts to provide recommendations to address some of these issues.
Key Words: ethics • clinical trial • guidelines • clinical trials data monitoring committees • research personnel
The ethical conduct of a clinical trial does not end with the formulation of study design and a signature on the informed consent form. Protecting the rights, interests, and safety of research subjects must continue throughout the study duration. Subject safety monitoring is the responsibility of several groups, including research ethics committees (RECs) or institutional review boards (IRBs), investigators and their research staffs, sponsors, and data monitoring committees (DMCs), also called data and safety monitoring boards (DSMBs), especially in the United States. Reports during the last few years of the deaths of research subjects and deficiencies in the monitoring of clinical trials have raised serious concerns regarding the systems and processes by which subject safety is currently monitored (1–5). The unexpected catastrophic adverse events in a "first-in-human trial" involving a drug called TGN1412 highlighted ethical issues in clinical study design and safety monitoring. Despite the findings from the Medicines and Healthcare products Regulatory Agency (MHRA) investigation, which found no association between the adverse incidents and drug administration, concerns remain regarding whether the drug was administered to the participants within a shorter time than what was approved (20 minutes rather than 2 hours) and whether the drug company should have known the drug would provoke the catastrophic reactions seen in humans (6).
In general, commentators have remarked on the duplication of efforts across several participating parties, overemphasis on the monitoring ability of some groups (e.g., RECs and investigators), and underemphasis by other groups (e.g., DSMBs and sponsors). Essentially, there is confusion about monitoring responsibilities, imperfect communications, and lack of integration among these groups. There are also inadequate standards for DSMBs (7).
Monitoring ongoing research ensures that the research is conducted as planned. Such monitoring entails several activities: an adequate process of informed consent; adherence to approved protocols; the proper collection of adverse events, ensuring the integrity of the collected research data; and the annual review of continuing research to assess the appropriateness of subject selection and the continued adequacy of a reasonable risk/benefit profile. Essentially, one needs to ensure that the integrity and safety of the research remains intact and that both anticipated and unanticipated harms can be rapidly detected and contained (i.e., being able to revise the protocol or stop the study if necessary).
Of the groups responsible for safety monitoring, we will focus on the four parties who have major monitoring responsibilities in a clinical trial: IRBs, the DSMBs, the study sponsors, and investigators and their research staffs.
INSTITUTIONAL REVIEW BOARDS (IRBs)
The IRB has numerous protection responsibilities that include initial and continuing review of the study protocol and related documents, review of the documentation of informed consent (though it is particularly difficult for the IRBs to adequately monitor the informed consent process, even with unannounced "spot checks"), and review of reports of unanticipated problems (UAPs) and of adverse events (AEs).
The reporting of UAPs and AEs is an important mechanism of ensuring subject safety. The AE report needs to present an accurate and expanding picture of harms occurring during the study. When first formed, IRBs were expected to be able to monitor the progress of a clinical trial at their site, but such expectations have become unreasonable for multicenter studies and complex single-center studies (8).
A major concern with interpreting AEs is that the random variation in background rates of adverse medical outcomes (due to underlying illnesses) can preclude an accurate assessment of causality of individual AEs. In other words, is the AE due to the toxic effects of the drug or to the natural history of the disease itself? IRBs are often flooded with individual reports of all AEs from theirs and the other study sites in a multicenter trial; yet such reports are largely uninterpretable, because the IRB does not know enough about their context. Determining causality of the AE with the study intervention would require measuring the excess events in the intervention group compared with the AE rate in the control group, which would require the receipt of aggregated data on AEs and the number of subjects currently enrolled in the clinical trial as a whole (providing a denominator for calculating the incidence of AEs). Because IRBs do not receive such data, they cannot know the potential causality of an AE occurring during the administration of an experimental drug.
IRBs also receive little guidance on how to handle such reports. Regulatory inconsistencies lead to confusion regarding what AEs must be reported and how they should be reported (7). As a result, IRBs are indicating that they do not want to be responsible for collecting and reviewing AE reports from distant sites. IRBs should collect relevant information and be responsible for considering the continued acceptability of trials at the study site(s) they oversee. IRBs are also calling for more restricted definitions of AE (i.e., "probably" or "definitely related" to the study intervention), which would be determined by the investigator and/or sponsor. They also are asking for clearer Federal guidance on how to handle AE reports.
Commentators have recommended that IRBs should have three major roles in a multicenter trial, in addition to their review of the research protocol and informed consent form (4). First, IRBs should ensure that a monitoring plan exists for each individual study site, and that there is a data monitoring program for multi-site studies. Second, the IRB should certify investigators' understanding of compliance with regulations governing the subjects' safety during the trial, including AE reporting and the collection of relevant study documents. Finally, the IRB should review the DSMB reports and query investigators as needed to determine whether additional safeguards are necessary. These recommendations seem reasonable, because the structure and function of IRBs would enable them to perform these roles better than the other entities involved in the monitoring of clinical trials. In addition, the suggested roles would add an efficient framework to the role that IRBs can play in the safety monitoring of human subjects during the course of a clinical trial.
DATA SAFETY MONITORING BOARDS (DSMBs)
The establishment of DSMBs was based on the recognition in the 1960s that independent means of interim monitoring of accumulating data were essential to determine ongoing subject safety in a trial. Essentially, individuals closely involved in trial design and conduct might not be fully objective in reviewing interim data for emerging concerns with harms to trial subjects. To provide the necessary monitoring, DSMBs usually consist of individuals with pertinent expertise in the disease under study, as well as statisticians, ethicists, and sometimes community representatives.
The increasing use of DSMBs has occurred due to: (1) the increasing number of industry-sponsored trials with mortality or major morbidity endpoints; (2) heightened awareness in the scientific community of problems in analysis that might lead to bias or inaccurate results; and (3) the previously mentioned concerns that IRBs are unable to properly monitor subject safety in multicenter trials.
The focus of DSMBs is on the total safety experience in a trial. The members of the DSMB therefore review aggregate data at predefined intervals and consider differences in the rates of clinical endpoints to determine whether clear benefits or harm might be occurring. They also review individual reports of AEs and consider the frequency, severity, and types of AEs and Serious Adverse Events (SAEs). A decision to stop a trial is made either when, using preplanned statistical analyses, significant differences in either benefits or harms are observed among the study arms or when there have been an excessive number of AEs in one of the study groups (see below).
Several commentators have raised concerns about DSMBs (7). For example, there are no commonly accepted standards for the composition and functions of DSMBs. Also, if the DSMB is formed late in the process, the committee may not be responsible for the structure and content of the data monitoring plan. There is also a concern about the degree of independence of DSMBs and the communication pathway from the DSMB to the Steering Committee to the IRBs. The US Food and Drug Administration (FDA) issued in March 2006 guidance for clinical trial sponsors for the establishment and operation of clinical trial data monitoring committees (FDA Guidance at http://www.fda.gov/cber/gdlns/clindatmon.pdf).
Determining the Need for a DSMB
Every clinical trial requires a safety monitoring plan, but not all trials require an external formal committee to perform the monitoring. The factors to consider when determining the need for a DSMB relate primarily to safety, practicality, and scientific validity (see FDA Guidance, http://www.fda.gov/cber/gdlns/clindatmon.pdf), as outlined in Table 1. These criteria explain the rationale behind the National Institutes of Health's (NIH) requirement that a DSMB be established for all phase III trials and a Safety Monitoring Plan for phase I and phase II trials. The FDA requires a DSMB for long-term trials with mortality or major morbidity outcome measures, when SAEs are expected, with novel and/or potentially high-risk treatments, when very little prior information about the study treatment is available, when studying an at-risk population consisting of vulnerable subjects (e.g., elderly or pediatric patients), or with a multicenter or long-term study. Conversely, an external DSMB is not required in early phase trials (with the exception of gene therapy trials), trials with symptom-only endpoints, and short-term trials.
|
Stopping rules exist to ensure that current and future subjects are not exposed to unreasonable risk. Indeed, one of the major responsibilities of a DSMB is to evaluate the relative treatment effects based on protocol-specified endpoints to determine if the trial is meeting its objectives. Statisticians have long recognized that reviewing accumulating data frequently would increase the chances of a false-positive error (Type I error), simply because of the increased number of opportunities to look for a statistical significance (9). This is especially true when frequent data reviews are coupled with a plan to stop a trial the first time the data comparison shows a statistically significant difference. Procedures should specify a statistical approach in advance of the trial's initiation that permits multiple interim reviews while maintaining the Type I error rate at the desired level.
DSMBs may also desire to devise a statistical plan to suggest early termination of a trial on the basis of futility—that is, when the probability, given the interim results, that the trial will ultimately be unable to demonstrate the effectiveness of the investigational product. In this case, a DSMB may recommend early termination on the grounds that the trial is unlikely to meet its objectives and there is therefore no basis for continuing enrollment and/or follow-up. DSMBs might also consider stopping a trial if accrual rates are too low or noncompliance is too great to provide adequate power for identifying the specified benefit. Stopping trials on the basis of futility runs the risk of drawing falsely negative conclusions.
DSMBs might also consider stopping a trial early if the investigational intervention is causing excessive risks compared with the control group. Because many types of adverse reactions cannot be anticipated based on prior trial data, DSMBs should generally be provided with interim summaries of AEs observed in each treatment arm, not limited to those identified in advance. DSMBs might recommend measures to reduce the risk of AEs—for example, changing the inclusion criteria, altering the product dosage, instituting screening procedures that could identify those at increase risk of a particular AE, and informing current and future study subjects of newly identified risks via changes in the informed consent form. Again, such interim assessments have potential implications for falsely concluding an adverse effect. Nonetheless, it is usually appropriate to require less rigorous proof of harm for early termination than would be appropriate for a finding of benefit. However, if a beneficial effect emerges, more rigorous proof of harm to determine an appropriate benefit-to-risk calculation might be appropriate.
There are different philosophies for stopping a trial early. Some suggest (especially those in the European Union) that studies should rarely be stopped early, mainly to ensure that any positive outcome endpoints will not be questioned, thereby ensuring that the study has a clinical impact. Others, mostly in the United States, have argued that once the pre-specified statistical endpoints have been reached and the study questions answered, the study should be stopped (10). Another issue regarding trial stoppage is exemplified by differences between major morbidity events and mortality outcome during the Heart Protection Study, in which the primary endpoint was mortality, and stroke and nonfatal myocardial infarction were secondary endpoints. By year 2 of this 5-year study, there was a significant difference in the secondary endpoints but not the primary mortality endpoint (11). Migrino and Topol argued that the trial should have been stopped based on group differences in secondary endpoints (12). However, the trial continued for 5 years and ultimately showed significant difference between all endpoints (13).
An important issue is the appropriate length of follow-up for the collection of safety data after the discontinuation of study drug (14). Another issue occurs when the decision to stop early due to favorable efficacy can diminish potential safety data. This occurred in the case of natalizumab for multiple sclerosis, in which a phase IV study showed AEs (progressive multifocal leukoencephalopathy) that did not appear in earlier trials (15). On the other hand, if the study is stopped early for safety reasons, the cases with SAEs should be of high quality (again underscoring the importance of quality data for IRBs and DSMBs), especially if the study's intervention has the potential of achieving a large amount of benefits. Such a scenario raises the issue of whether drugs that have safety concerns should still warrant approval due to the large degree of benefits that can be given to a select group of patients.
Who Should See the Interim Data?
Sponsors, as funders of clinical trials, typically want as much information about trial results as early as possible. Unfortunately, the involvement of sponsors in collecting, analyzing, and interpreting clinical trial data during the conduct of a trial raises issues with conflict of interest (7). The sponsor's desire of access to interim data for planning purposes is understandable, but may render trial results uninterpretable. Thus, the sponsor should see minimal interim information. The standard operating procedures are that only identified individuals within the sponsor organization who need to know about the interim data would have access to interim data.
Finally, the amount of interim data that should be available to investigators and potential subjects is an unanswered question. While some argue that potential subjects need to know the interim data to make an informed decision, others argue that such data are not conclusive and that subjects may not understand the importance or clinical significance of interim data analysis. Disclosure of inconclusive interim data could seriously jeopardize the clinical trial by disturbing the assumption of clinical equipoise, which could then make physician-investigators hesitant to enroll further patients in the trial (16). To be sure, significant adverse events that have occurred in the study should be shared with current and potential study participants in order for them to make an informed decision regarding continued enrollment or initiating participation, respectively. Failure to make available such data would be ethically inappropriate.
Interim Data Analysis: The Value of a Control Group
In critical care trials, the control group serves an essential function to help distinguish study endpoints (e.g., mortality) or AE rates between the study treatment and control treatment (e.g., usual care). Without the control group, it would not be possible to detect these differences, if/when they exist. Control groups of usual care exist to determine whether one has disturbed the clinical equipoise between the intervention and usual care. Accordingly, investigators can use the control group to determine when to stop a trial if one of the interventions is causing harm (as discussed in this issue by Dr. Robert Wise). Because critical care trials involve endpoints involving mortality or major morbidities, and because diseases can be rapidly progressive and lethal, it would be appropriate for a DSMB to require less rigorous proof of harm to stop a trial. A particular concern arises if the control group is not representative of usual care (e.g., in trials that compare contrasting strategies, such as the ARDSnet Tidal Volume trial or the Hemoglobin Study) (17, 18).
Such studies of contrasting strategies are performed to enhance their explanatory power in assessing physiologic mechanisms. However, if none of the study arms represent usual care, there is less assurance that any of the study arms will have mortality rates that do not exceed usual care practices (19). Also, a "control" group selected to ensure a large separation in the study outcome between the two study groups runs the risk that the experimental group will be compared with a "control" group with higher risks of harms, thus falsely attributing benefit to the experimental group (20).
SPONSORS
Sponsor responsibilities are numerous and extend far beyond financing. Sponsors must demonstrate in pre-clinical studies that an intervention will be safe enough to justify clinical research, provide medical expertise, ensure that the trial design minimizes harms, participate in investigator selection, review and initially interpret AEs, indemnify subjects against injury claims arising from the trial, provide applications, notifications, and submissions to regulatory authorities, confirm IRB review, monitor to ensure Good Clinical Practice (GCP, described below) during the trial, supply and handle the investigational agent, maintain records and keep them available for inspection, perform safety evaluation and reporting, monitor and audit the data that have been collected during the study, secure compliance and report noncompliance, prepare trial reports, and monitor study sites.
The major role for sponsors in ensuring subject safety serves two purposes. First, because the sponsor has the greatest access to information about the study agent, they are in the best position to protect study subjects by making information about risks readily available to investigators and IRBs both before the trial (in the form of the investigator brochure) and during the trial, and then to the FDA after the trial. The major concerns of the Vioxx trial involved claims of incomplete provision of trial information to the FDA (21). Similarly, one of the issues in the Jesse Gelsinger Pennsylvania gene-transfer study was the failure to report complete pre-clinical information regarding liver toxicity risks in the informed consent form (1).
Second, sponsors are able to report serious and unexpected AEs to IRBs with detailed interpretation of the likeliness of association with the intervention. Of note, sponsors cite an excessive focus on data collection and data auditing and over-reliance on auditing completeness of case report forms to ensure GCP, such that these processes slow the dissemination of data to the investigator brochure. To be sure, the true purpose of data audits is to ensure the integrity of the safety and efficacy data and completeness of data submitted to the FDA, rather than to identify AEs.
INVESTIGATORS
Like study sponsors, investigators and their research staffs have a wide range of responsibilities that include adhering to the protocol study design, conducting an ethically appropriate informed consent process, ensuring data integrity, maintaining the essential document files, and supplying interpretation of AEs within the context of known data about the intervention (7).
As individuals on the "front line" of the research endeavor, investigators and their research staffs are entrusted with assuring the proper application of the inclusion/exclusion criteria, the proper administration of study treatments, and the accurate monitoring and reporting of AEs. All of these tasks have important implications for subject safety.
Good Clinical Practice
The hallmark of sponsor and investigator responsibilities is the concept of good clinical practice, which is detailed in the International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use (ICH) Guideline #6 (22). GCP standards were developed to provide guidance to investigators that would result in common approaches to clinical trials performed in multiple countries. GCP forms a standard for the design, conduct, performance, monitoring, auditing, recording, analysis, and reporting of clinical trials, which provides assurance that the data and reported results are credible and accurate, and that the rights, integrity, and confidentiality of trial subjects are protected. GCP has several objectives concerning the protection of trial subjects, quality of data, and transparency of trial conduct (Table 2) (22). The principles of GCP are shown in Table 3. Of note, it has recommendations for all parties in the conduct of a clinical trial (22).
|
|
|
Given the numerous parties, multiple and overlapping responsibilities, possibilities for communication breakdown, and the ethical responsibility associated with research in human subjects, the following are recommended for system-wide improvement. Monitoring plans for clinical studies should be well designed, with a centralized review of large, multi-site studies concomitant with closer local scrutiny of single-institution studies. Sponsors need to pay greater attention to and take responsibility for monitoring AEs and disseminating that information. Performance standards need to be developed in GCP, by way of formal training programs for investigators, staffs, and IRBs. Likewise, DSMB composition and function needs to be standardized. Finally, there is a need for guidelines for monitoring phase I and phase II trials.
DISCUSSION HIGHLIGHTS
Role of Non-Study Investigators in Reporting SAEs
DR. ZAB MOSENIFAR (Cedars-Sinai Medical Center, Los Angeles, CA): Let's use, as an example, a trial X for emphysema. An investigator who was not linked to the study encounters a patient in the recovery room who has developed tension pneumothorax. Any reasonable person might say that this tension pneumothorax was linked to the study intervention. The PI felt that it wasn't related and didn't report it. What is the ethical responsibility of this investigator who has nothing to do with the study?
DR. SILVERMAN: I sometimes break down ethics into three categories: the duties that you are permitted to do, the duties that you are not allowed to do, and the duties that you are obligated to do. You are asking whether the non-study investigator has a special obligation to make a report. That gets a little tricky because it really depends on the interpretation of causality. I personally would talk to the other investigator and find out why he or she thinks that this is not a reportable SAE and then go from there.
DR. DUNCAN J. MACRAE (Royal Brompton Hospital, London, UK): I think anything that's potentially an adverse event should be reported. I don't think it should be up to the site investigator to cite whether it's relevant or not, because it's only when you get the whole population's reports together that you can actually start to make deductions. And it's particularly important when it comes to those events where it's not so obvious or easy to sort out the cause, like pneumonia.
DR. SILVERMAN: In that case, if the reason for the event is being misinterpreted, I think one has a special obligation to report that event. The hard cases come when you think there might or might not be causality. If there is any doubt, it should be reported.
FOOTNOTES
This article is funded by Boehringer Ingelheim and by Pfizer, Inc.
Conflict of Interest Statement: H.S. received $2,000 from Boehringer-Ingelheim in 2006 for a presentation at a workshop. Z.M. has received $4,500 in 2004, 2005, and 2006 from Boehringer-Ingelheim and $4,000 from Pfizer in 2006 for speaking at conferences organized by these companies. He is an investigator in the Uplift trial. The amount of this grant is $11,000 over a period of 4 years. However, Z.M. received no salary or consulting fees. D.J.M. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript.
(Received in original form January 3, 2007; accepted in final form February 26, 2007)
REFERENCES
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
