Adjuvant radiotherapy for breast cancer: effects of longer follow-up

Abstract

Background and purpose: Recent and large trials of adjuvant radiotherapy for breast cancer have shown an overall survival benefit in favour of radiotherapy. However, with longer follow-up the late lethal toxicity of radiotherapy might reduce the overall survival benefits. In this paper we investigate more deeply this hypothesis.

Patients and methods: Overviews of the Early Breast Cancer Trialists' Collaborative Group provide uniform data on more than 50 unconfounded trials on adjuvant radiotherapy for early breast cancer. These data were published at regular intervals: 1987, 1990, 1995, and 2000. The odds ratios (death of any cause) were borrowed to compare the benefits of adjuvant radiotherapy between the early publications and the more mature data of the same trials. Statistical significance is calculated following logrank statistics. The comparison of odds ratios (radiotherapy versus surgery only) was done for the whole group of trials, for the older (patients accrual started in 1970 or earlier) and the more recent trials (patient accrual started after 1970), and for the large (>or=600 patients) and the small trials (<600 patients).

Results: Comparison of early with more mature data reveals that the odds ratios for overall survival remain stable as data become more mature. The analyses of trials' age and trials' size, as predictors of overall survival benefit, indicate that these factors become statistically more significant with increasing maturity of the trials. In the large recent trials an overall survival benefit due to radiotherapy (odds reduction) of 10, 10, 12 and 13%, respectively P<0.3, 0.2, 0.005 and 0.00005 is found in the successive publications. The difference in survival benefit of radiotherapy between the group of large recent trials and group of old or small trials becomes more significant at the successive updates: 10 via 9% and 12 to 13% (odds reductions), with respectively P=0.2, 0.2, 0.004 and 0.00005.

Conclusions: These results support the hypothesis that the survival benefit in the recent trials is an inherent characteristic of the recent and large trials, not influenced by follow-up duration. The effect of radiotherapy as performed in the large recent trials is clinically and statistically significantly different from the effect of radiotherapy in the old or small trials. As a consequence, predictions based on pooled data including old radiotherapy trials should not be extrapolated to modern radiotherapy.