Clinical Trials Should NOT Be Stopped Early for Benefit Based on Interim Analyses Group 6: David Hottman,Fatima Khan, Kelley Knapek, Rebecca Kruc,Andrew Timmons, Lisa Vaughan, Ashwini Venkatasubramaniam, Xiaoyue Zhao
Ethical Requirements • Participant Consent and Respect for Participants • Scientific Validity • Social/Scientific Value and Favorable Risk-Benefit Ratio • Independent Review
Participant Consent and Respect for Participants Requirements for informed consent: Completeness of information: • Adequate information should be provided to the participant • Mutually beneficial, continued involvement in the research should help meet their goals and ideals. Timely communication of changes: • Investigators should provide updates and inform participants of changes in the trial in a timely manner. .
Disadvantages of Stopping Trials Early Misleading Interim results: • Once the choice of a superior treatment has been made, equipoise ceases to exist • Clinicians and patients would choose the treatment deemed to be “superior”. Unblinded trials: • Once patients are informed of interim results, the patients would be unblinded. • They might leave the trial or cross over to the “superior” treatment
Disadvantages (Continued) Long delays in treatment dissemination: • Treatment deemed “superior” might not be administered immediately. Higher chance of receiving superior treatment with trial continuation: • Continuation of the trial at least gives patients a 50% chance, as opposed to if the trial is stopped early. Reduction in trial’s scientific validity: • Compromised with patients being informed of early results.
Scientific Validity • Aim of clinical trials is to closely approximate the true effect of the treatment by minimizing both random and systematic error • Stopping trials early may compromise this and introduce error • Example:review of 143 trials that were stopped for apparent benefit • It was found that trials accruing fewer ends points before being stopped estimated larger relative risks (strong inverse association between number of events and estimated treatment effect) • This was consistent at the median and 75th percentile of events and at the median and top quartile of RR estimated • Suggests that stopping trials early may introduce systematic overestimation of treatment effects http://jama.jamanetwork.com.ezp3.lib.umn.edu/article.aspx?articleid=201802
Social or Scientific Value and Favorable Risk Benefit Ratio Impact on a larger population than the study must be considered Overestimation of primary endpoints • Often stopped for implausibly large results; stopped at a random high • Subsequent misinterpretation of results by clinicians could lead to unnecessary and improper treatment for the wider population Lack of data on other important endpoints • Safety, secondary outcomes, etc. • Lack of data for future research
Social or Scientific Value and Favorable Risk Benefit Ratio (Continued) Example: Twelfth Medical Research Council acute myeloid leukemia trial • Data monitoring and ethics committee found a large benefit of the 5-drug regimen compared to the 4-drug regimen for acute myeloid at interim analyses • Trial was continued and later showed non-significant results • If the trial had been ended early, the 5 treatment regimen may have been adopted, thus exposing patients to higher toxicity
Example: Trials in Oncology • Specific scenario • One of the goals is to get accelerated and conditional approvals, leading to quicker access of new drugs to patients. • It is often accepted that the rules for early stopping in oncology trials evaluating a chronic treatment intervention should be based on two major criteria: • Development of prohibitive toxicity • Improvement in either overall survival (OS) or quality of life (QOL). • Improvement in Progression-free survival (PFS) or disease-free survival (DFS) should not generally be used as an early stopping criterion.
What Really Happens? A study to assess the use of interim analyses in randomised controlled trials (RCTs) testing new anticancer drugs focused on trials stopped early for benefit. They looked at all clinical trials regarding anticancer drugs that were published from January 1997 to October 2007
What Really Happens? Of the 93 papers selected as having been stopped after an interim analysis: • 28 (30%) were stopped early for benefit (OS evaluated in 40%) • 28 (30%) for futility • and 4 (4%) for harm • Ethical issues arise not only when stopping clinical trials early for benefit, but more generally, in stopping clinical trials early for the wrong reasons. Ann Oncol (2008) 19 (7): 1347-1353.
Independent Review Stakeholders may want a trial to be stopped early for their own benefit: • Large treatment effects get published more often-may stop early for investigator or sponsor benefit • May stop early to decrease total cost, even though the full picture of the results may not be clear at that time • Trials stopped early for benefit often receive greater publicity-would be beneficial for researchers (greater prestige), trial sponsors (may sell more drugs), or journals (greater prestige, more copies sold) • Even participants and patients may want a trial stopped early for benefit so that the treatment can be made available to everyone
Independent Review (Continued) • Trials stopped early for these reasons may not have the full picture about treatment benefit • The decision to stop trials early should be done by an independent review board • This minimizes bias that would occur if the decision to stop early is influenced by the stakeholders
Example 1 : EARLY • Investigators conducted a trial that involved patients undergoing vascular surgery • Trial stopped early when 2 of the 53 patients randomly assigned to receive the beta-blocker and 18 of the 59 control patients had major cardiovascular events (relative risk reduction, 90% [CI, 59% to 98%]) • These results then contributed to recommendations by the American Heart Association to administer beta-blockers to patients with cardiac risk factors • These results contradict those of 2 much larger subsequently published trials, neither of which suggested that beta-blockers reduce cardiac risk in patients undergoing noncardiac surgery
Example 2: The Candesartan in Heart Failure Assessment of Reduction in Mortality and Morbidity • Stringent threshold for stopping early (p<0.0001) • First three interims were highly significant but did not meet threshold (p=0.0007, p=0.0002) • Final outcome was borderline (p=0.055) http://circheartfailure.ahajournals.org/content/5/2/294.full
Recommendations More stringent early stopping rules: • Demand that a large number of events (patient-important events in trials using composite endpoints) accrue before investigators or data monitoring committees examine interim data (200-400 to be conservative) • Invoke a stopping rule with a low P-value (ex: 0.001) • Continue enrollment and follow-up for a further period to ensure that the trend continues
Recommendations Journals should require that reports of truncated trials describe the rationale for early stopping, including the statistics and rules applied
Recommendations Investigators and data monitoring committees should consider the context of the illness and the treatment being studied when deciding to stop a trial early • Approximately 25% of new drugs eventually prove to have serious adverse effects that were unexpected when they entered the market.
Conclusion Stopping a randomized trial early for apparent benefit is often unethical and can be justified only under restricted circumstances!
References • http://www.bioethicsforum.info/files/2008-1-Mueller-CT.pdf • http://jama.jamanetwork.com.ezp3.lib.umn.edu/article.aspx?articleid=201802 • http://www.ncbi.nlm.nih.gov/pubmed/12559643 • http://circheartfailure.ahajournals.org/content/5/2/294.full • Ann Oncol (2008) 19 (7): 1347-1353. • Mueller, Paul S, Victor M Montori, Dirk Bassler, Barbara Koenig, and Gordon H Guyatt. "Ethical Issues in Stopping Randomized Trials Early Because of Apparent Benefit." Annals of Internal Medicine. 146.12 (2007): 878-881.