Early stopping of a clinical trial when there is evidence of no treatment benefit: Protocol B-14 of the National Surgical Adjuvant Breast and Bowel Project

Although several randomized clinical trials in the 1980s indicated a benefit from the use of tamoxifen in the treatment of early-stage breast cancer, questions have remained regarding the optimal duration of drug administration. In 1982, the National Surgical Adjuvant Breast and Bowel Project (NSABP) initiated a randomized trial to compare 5 years of tamoxifen to placebo among breast cancer patients with estrogen receptor-positive tumors and no evidence of axillary node involvement. By 1987, evidence of a substantial benefit for tamoxifen led the NSABP to extend this trial to determine whether longer duration tamoxifen therapy would be additionally beneficial. This study randomized patients who had completed 5 years of tamoxifen free of breast cancer recurrence or other events to either tamoxifen or placebo for an additional 5 years. By 1994, 1172 women had entered the study and accrual was closed. In late 1995, the trial was terminated on the basis of interim findings indicating that a benefit for continuing tamoxifen would not be realized. The closure has prompted controversy among cancer researchers, because there are currently at least three tamoxifen duration trials in progress, whereas results from two other studies evaluating 5-year duration therapy versus longer therapy were recently published. Here, we provide details of the statistical rationale contributing to our decision to recommend early closure of the study. We then consider other possible approaches to assessing the appropriateness of early termination in the face of evidence against a benefit, including Bayesian methods, which can be used to incorporate a range of prior beliefs regarding the efficacy of a treatment with accruing information from the trial. We also briefly discuss results of the other published studies. Controlled Clin Trials 1998;19:575-588 (C) Elsevier Science Inc. 1998.