Re: Tamoxifen for the Prevention of Breast Cancer: Current Status of the National Surgical Adjuvant Breast and Bowel Project P-1 Study

Extract

We read with great interest the recent report of Fisher et al. ( 1 ) on the longer-term follow up of the P-1 trial. This article underlines the importance of determining the effect of tamoxifen prophylaxis in the 5-year period following treatment. Unfortunately, interpretation of the data for P-1 is very difficult because the reduced difference in the breast cancer rates after treatment cessation appears to be due to a lower incidence of breast cancer in the control group in years 6 and 7 of follow-up [Fig. 3 of ( 1 ) ]. An important question is whether this reduced incidence in control subjects is a result of tamoxifen use in the control subjects after unblinding or whether some other mechanism is involved, possibly related to increased dropout in the control subjects.

The Fisher et al. article raises the issue of the conditions under which prevention trials should be stopped prematurely. The authors make several references to the International Breast Cancer Intervention Study (IBIS) studies, particularly in relation to the early stopping of prevention trials. A more complete representation of our views on this issue has recently been published ( 2 ) . We would like to emphasize that the reporting of statistically significant early results, which is an obligation to the patients in the trial, should not always entail enforced unblinding and early termination of the trial. In the IBIS-I trial, which was very similar to P-1, patients were informed of our early published results ( 3 ) and those of P-1 ( 4 ) , but neither the Independent Data Monitoring Committee nor the Steering Committee felt that the results were sufficient to change clinical practice. Accordingly, the women were told these results, and blinded continuation was recommended by the Steering Committee after written re-consent. Because of the level of communication we maintained with the participants, all but a handful of women agreed to continue in their original trial arm, and the trial remains blinded at this stage, with excellent compliance. We plan an updated analysis of IBIS I next year, with an 8-year median follow-up, and at that time we will review again whether blinded continuation is recommended.