Effects of radiotherapy and surgery for early breast cancer

Abstract

We have the following comments to make on the latest EBCTCG overview (Dec 17, p 2087). One of the key issues facing women who are diagnosed with early breast cancer is the choice of surgery. Without a clear analysis of the types of surgery to address interactions with radiotherapy, the main question has been skipped. The overview describes extensive subgroup analyses (and omits overall results). As well as splitting patients by breast surgery, axillary surgery, and nodal status, they are then further divided into those who presented less than 10%, 10–20%, or more than 20% absolute reduction in the 5-year local recurrence risk. These sub-subgroups are then used to analyse recurrences and breast cancer mortality. Using an outcome (recurrence) to predict itself is not only of dubious value but also raises concerns of how any individual patient’s data were assigned to each of the three categories. The overview argues that one breast cancer death is avoided for every four local recurrences avoided. This assumes a causal relation between local recurrences and breast cancer deaths. But neither the individual trials nor a meta-analysis can say whether recurrences are indeed causal, or whether recurrences are only a marker of risk. A cancer can recur because the disease is more aggressive. However, prevention of recurrence will not necessarily render the disease less aggressive. The overview argues that large absolute reductions of recurrence or death are seen only when the absolute risk of recurrence or death is high. The paper’s conclusion is self-contradictory: the absolute breast cancer mortality at 15 years is reduced from 60·1% to 54·7% (5·4% absolute reduction) after mastectomy in women with nodepositive disease and from 35·9% to 30·5% after breast-conserving surgery (5·4% absolute reduction; fi gure 6). Whether the absolute risk of death is 30% or 60%, the absolute reductions in breast cancer deaths shown by the paper are of the same order of magnitude. It is unclear why the overview has to speculate on “hypothetical other causes of death”, when the data clearly show that the reduction of all-cause mortality far exceeds that of any hypothetical causes. Mortality from any cause is reduced from 40·5% to 35·2% (breast-conserving surgery), and from 72·3% to 68·8% (mastectomy)—ie, an absolute reduction in all-cause mortality of 3·5–5·3%. The overview of 2000 presented results at 20-year follow-up. It found some survival advantage but argued that the “average hazard seen in these trials would, however, reduce this 20-year survival benefi t in young women and reverse it in older women”. In the 2005 overview, results are presented only as a 15-year follow-up. With more years of follow-up, surely something substantial could be said about the previous interpretation?