Decision making in adjuvant trials in breast cancer: the NCIC CTG MA.17 trial as an example
Decision-making regarding early closure and reporting of clinical trial results became a topic of intense debate following the reporting of the MA.17 adjuvant endocrine therapy trial. This trial was terminated at the first planned interim analysis when a highly significant improvement in disease free survival (DFS) was found. It has been suggested that the criterion for early stopping be made stricter when DFS is the primary study endpoint by ensuring that the targeted effect size is excluded. Our purpose is to examine this approach and to determine whether applying such a criterion would have affected the decision to terminate MA17. The sample size assumptions and the interim analysis of MA17 were reviewed and an appropriate method employed to convert hazard ratios (HR) to absolute differences. Expressed in relative terms, the effect size of MA17 was an HR of 0.78, and the upper boundary (0.75) of the confidence limits around the observed HR (0.57) excluded this value. In absolute terms, the lower confidence limit (3.1%) of best estimate of the 4 year difference in DFS (5.4%) excludes the difference (2.5%) used in the calculation of the targeted HR. We conclude that although the decision to release the results of the interim analysis of MA17 and allow patients on placebo to take letrozole was justified, methods for analyzing and interpreting interim results can be improved.