Can you give us an outline of what findings you presented to the European Breast Cancer Conference this month?
We carried out an individual patient data meta-analysis of 14 clinical trials in which women with early stage breast cancer, who had been given mastectomies along with the surgical removal of lymph nodes under the arm (axillary dissection), were randomised to receive either radiotherapy to the chest wall and surrounding regions or to no radiotherapy.
We found radiotherapy was of little benefit in the 700 women with low risk of breast cancer recurrence. There was no effect on 10-year recurrence or 20-year breast cancer mortality. In the 1800 women at high risk, radiotherapy did reduce recurrence and breast cancer mortality. However, these findings were not unexpected; it is among the 1300 women of intermediate risk that our findings are new. For these women we saw that radiotherapy reduced both recurrence and breast cancer mortality.
How long have you and your team been working on this project?
The Clinical Trial Service Unit, Oxford University, is the base of the secretariat of the Early Breast Cancer Trialists’ Collaborative Group (EBCTCG) which is a worldwide collaboration of about 600 trialists. It was initiated by Prof Sir Richard Peto in the mid-1980s.
The main function of EBCTCG is, every 5 years or so, systematic overviews of all the available randomised trials of local and systemic adjuvant breast cancer therapy. We have been working on this latest study for several years. It takes a long time to get all the fine details of the individual patient data correct and consistent as well as latest follow-up information.
How did you go about segmenting the 3786 women who took part in the study in order to measure the effects of radiotherapy on patients with different characteristics?
The main determinant of recurrence risk amongst the women in these trials is how far the cancer has spread into the lymph nodes. So this was the chief characteristic in segmenting the women into low, intermediate, and high risk. Broadly speaking this corresponds to how clinicians categorise risk and so this makes the results from the study of most use to them.
Your study looked at data from the beginning of trials in 1964 up until data from 2009 on deaths and recurrences. How important was it that you were looking a data that spanned many decades?
It takes many years to get a full picture of the benefits, and hazards, of cancer treatments. Breast cancer has a long natural history. So, having trials starting decades ago allows us to look over extended time periods. Of course the drawback is that state of art therapy in the 1970s is not state of the art therapy in the 2010s.
Chemotherapy and hormonal therapy have evolved over the period that you studied, how did this affect the data and what challenge did it present in measuring the effect of radiotherapy?
Not only has systemic therapy evolved over the intervening period but so to have breast cancer surgery, lymph node staging and screening. This means that for a woman seen in the clinic today her risk of recurrence without radiotherapy is likely to be somewhat lower than for women in the trials we analysed.
Therefore, we emphasised the proportional estimates in our results. If a clinician estimates absolute recurrence risk, based on characteristics of the tumour, our study will enable them to calculate the expected absolute reduction in risk due to radiotherapy.
What did you find the greatest statistical challenges with the study to be?
Getting all the available data and standardising it makes the statistics easier than it would otherwise be. This, pretty much, lets us use standard methods. The greatest challenge then becomes trying to show the most important results clearly and communicating the relative usefulness of our estimates of absolute risk reductions and proportional risk reductions.