Randomized Phase II Studies and Their Ethics

Abstract

TO THE EDITOR: Randomized phase II studies are being conducted more frequently and have recently been reviewed from a scientific perspective. These reviews of these important studies need to be complemented by some of the associated ethical considerations. Randomized phase II studies have come into being with two different aims. Randomization may be between different experimental arms so as to efficiently identify the most promising experimental treatments for further phase III study. Alternatively, a control group may be included to address some of the problems associated with a comparison to historical controls caused by selection bias. In practice however, there is overlap in the conduct and interpretation of different approaches to phase II studies, as well as phase III trials. Data on survival or other indicators of patient benefit are collected and statistical comparisons made, despite the fact that most randomized phase II studies are not powered to demonstrate statistically significant differences. This is valid provided that the prospectively defined comparative scale (eg, hazard ratio or odds ratio) and population (eg, all randomized patients) are still used, even if the power on that scale was low. These comparative statistics, along with their confidence intervals, then provide unbiased estimates of treatment effects and are, in fact, more appropriate than within-arm summary statistics, such as median survival or response. Indeed, statisticians generally argue that trials must stress comparisons. Low power is irrelevant to validity, and indeed should significant differences be found, they will not and should not be ignored. It is noted that comparative statistical analyses are not in themselves unbiased and may be biased when they are the best of several secondary and/or different comparative analyses. Ethical codes and good clinical practice guides do not differentiate between phase II and III studies. Ethical guidelines, which have been applied routinely in phase III trials, need to be considered when conducting a randomized phase II study. These ethical questions are in the realm of good science and differ from the concerns expressed at misinterpretation or over interpretation of phase II studies, as cited in a recent journal editorial. The nature of a possible control arm needs to be clarified in randomized phase II studies. Article 29 of the Helsinki Declaration of 2004 states, “The benefits, risks, burdens and effectiveness of a new method should be tested against those of the best current prophylactic, diagnostic, and therapeutic methods.” This requires careful attention with designs that do not include the current standard of care as a treatment arm, particularly when an unequivocal commitment to evaluate encouraging findings with a phase III trial cannot be made. In addition, the well known ethical commitment to patient care at the conclusion of a clinical study holds for all studies. Patients should have access to the best proven therapeutic methods identified by the study (Helsinki Declaration, article 30).