Cookies on this website

We use cookies to ensure that we give you the best experience on our website. If you click 'Accept all cookies' we'll assume that you are happy to receive all cookies and you won't see this message again. If you click 'Reject all non-essential cookies' only necessary cookies providing core functionality such as security, network management, and accessibility will be enabled. Click 'Find out more' for information on how to change your cookie settings.

Accept all cookies Reject all non-essential cookies Find out more

9.1 Introduction

Breast cancer is the commonest type of cancer among women in most developed countries. There are about a million new cases diagnosed each year worldwide, and around 35,000 new cases annually in the United Kingdom alone (Quinn et al. 2001). For such a common disease, widely practicable treatments that produce only moderate effects on long-term survival, such as increasing the number of women surviving for more than ten years after diagnosis from 50% to 55%, could result in the avoidance of many thousands of deaths each year. Therefore, it is important to be able to distinguish such treatments from those that have no effect, or even a deleterious effect, on overall survival.

In developed countries, most women who are diagnosed with breast cancer are diagnosed when the disease is at an early stage and is detected only in the breast itself—‘node-negative’ women—or in the breast and the lymph nodes near the affected breast—‘node-positive’ women. The primary treatment for most such women is surgery. However, the extent of the surgery considered necessary has varied substantially at different times and in different countries. There is also a wide variety of ‘adjuvant’ treatments that can be given in addition to surgery, and many hundreds of trials comparing the different treatments and combinations of treatments have been undertaken. The design of most of these trials is that women who satisfy a pre-specified set of entry criteria, for example in terms of age or extent of disease at diagnosis, are allocated at random to one of two possible treatment ‘arms’ that differ only with respect to the treatment being evaluated. For example the extent of surgery might be identical in both trial arms, but the women allocated to one of the trial arms might also be offered radiotherapy.

Where there are several trials that address similar, although not necessarily identical, questions, it is possible to obtain estimates of the differences between treatments by combining the data from them. This approach is much more precise than the estimates given by any individual trial. Inevitably, trials with extreme results tend to receive more attention than those with more moderate results. This produces a natural tendency for unduly selective emphasis on those trials or subcategories of patients where, by chance alone, the results are misleadingly positive or misleadingly negative. Most such biases can be avoided by appropriate combination of the results of all trials that address similar questions. This combination cannot be done satisfactorily from published data alone (Stewart and Parmar 1993), and the inclusion of unpublished as well as published data is necessary to avoid bias. Furthermore, the information available from the published trials is not sufficient to allow a uniform analysis of all the available data with appropriate stratification for factors that will affect survival such as age, time since diagnosis, or nodal status. Thus, analysis based on individual patient data is necessary.

With these issues in mind, collaboration was sought in 1983–1984 between the co-ordinators of all randomized trials of the treatment of early breast cancer that satisfied certain criteria, and in 1985 the Early Breast Cancer Trialists’ Collaborative Group (EBCTCG) was initiated. It has continued since then in five-yearly cycles. At the time of writing, the analyses resulting from the fourth cycle of the EBCTCG have been finalized, and initial preparations are being made for the fifth cycle, which will include data up to 2005. The fourth-cycle analyses now available via the University of Oxford Clinical Trial Service Unit website, www.ctsu.ox.ac.uk, and in the published literature. See the note added in proof on page 196. The present chapter summarizes the main results of the earlier cycles of the EBCTCG, both in Table 9.1 and in the text, and it also comments on recent trends in breast cancer mortality. The results presented here are mainly those for overall survival. However, the original publications also consider other endpoints, including mortality from breast cancer and mortality from other specific causes of death, as well as breast cancer recurrence, that is, the return of the original cancer after a period of remission. In almost all cases, where a treatment has a beneficial effect on overall survival, this occurs following an earlier and larger effect on breast cancer recurrence.

Table 9.1

The main results for overall survival in the randomized controlled trials considered by the Early Breast Cancer Trialists’ Collaborative Group. Standard errors (se) in percent are given in parentheses.

Cycle Data included Main results for overall survival and references
First Trials started before 1985. Follow-up to 1985.

Tamoxifen (28 trials involving 16,500 randomized women): Highly significant (p <0.0001) reduction in mortality in trials of ‘tamoxifen versus no tamoxifen’ over about five years of follow-up (EBCTCG 1988).

Chemotherapy (40 trials involving over 13,000 randomized women): Highly significant (p = 0.003) reduction in mortality in trials of ‘any chemotherapy versus no chemotherapy’, and also (p = 0.001) in trials of ‘polychemotherapy versus single-agent chemotherapy’, both over about five years of follow-up. Chemotherapy for 12–24 months may offer little survival advantage over 6 months (EBCTCG 1988).

Second

Trials started before 1985. Follow-up to 1990.

Tamoxifen (40 trials involving 30,000 randomized women): Highly significant 17% (se 2; p <0.00001) reduction in overall mortality rate in trials of ‘tamoxifen versus no tamoxifen’ over about 10 years of follow-up. Indirect comparisons suggest longer term treatment (∼2–5 years) better than shorter. Polychemotherapy plus tamoxifen clearly better than polychemotherapy alone at ages 50–69 (EBCTCG 1992).

Chemotherapy (31 trials involving 11,000 randomized women): Highly significant 16% (se 3; p <0.00001) reduction in overall mortality rate in trials of ‘polychemotherapy versus no chemotherapy’ over about 10 years of follow-up. No advantage in long-term treatment (∼12 months) over shorter term (∼6 months). Tamoxifen and polychemotherapy may be better than tamoxifen alone at ages 50–69 (EBCTCG 1992).

Ovarian ablation (10 trials involving 3000 randomized women): Highly significant 25% (se 7; p = 0.0004) reduction in overall mortality rate for women treated at age < 50 over at least 10 years of follow-up. No significant effect for those aged 50+ when treated. (EBCTCG 1992).

Immunotherapy (24 trials involving 6300 randomized women): No significantly favourable effects of immunotherapy found (EBCTCG 1992).

Local therapies (64 trials involving 28,500 randomized women): Radiotherapy reduced rate of local recurrence by factor of three and breast conserving surgery involved some risk of recurrence in remaining tissue, but no definite differences in overall survival at 10 years (EBCTCG 1995).

Third

Trials started before 1990. Follow-up to 1995.

Tamoxifen: (55 trials involving 37,000 women): No significantly favourable effect in women with ER-negative tumours, but for 30,000 women with ER-positive or untested tumours highly significant effects with 12% (se 3), 17% (se 3) and 26% (se 4) reductions in the overall mortality rate in trials of 1, 2, and 5 years respectively of tamoxifen versus no tamoxifen, over about 10 years of follow-up. Proportionate benefit applies regardless of nodal status, age, menopausal status, daily tamoxifen dose, and whether or not chemotherapy was given (EBCTCG 1998b). Chemotherapy (69 trials involving 30,000 women): Highly significant reductions in overall mortality rate of 25% (se 5; p <0.00001) in women aged < 50 and 11% (se 3; p = 0.0001) in women aged 50–69 in trials of “polychemotherapy versus no chemotherapy”. Proportionate benefit similar regardless of nodal status, menopausal status (given age), ER-status and whether or not tamoxifen had been given. No advantage of more than about 6 months of treatment (EBCTCG 1998a).

Ovarian ablation (12 trials involving 3500 women): No benefit for women aged over 50, but highly significant improvement in 15-year survival among those aged 50 or under when treated (15-year survival 52% versus 46%; p = 0.001) in trials of “ovarian ablation versus no ablation”. Further evidence needed on effect of ablation in the presence of other adjuvant treatments (EBCTCG 1996).

Radiotherapy (40 trials involving 20,000 women): Substantial reduction in breast cancer mortality largely offset by an increased risk of mortality from cardiovascular disease, so that overall 20-year survival was 37% with radiotherapy and 36% without (p = 0.06) in trials of “radiotherapy versus no radiotherapy”. The ratio of benefit (from reduced mortality from breast cancer) to harm (from increased mortality from cardiovascular disease) was strongly affected by nodal status, age and decade of follow-up (EBCTCG 2000).

Fourth

Trials started before 1995. Follow-up to 2000.

Data on ∼80,000 women randomised in trials of tamoxifen, ∼50,000 women in trials of chemotherapy, 10 000 women in trials of ovarian ablation and ∼38,000 women in trials of local therapies have recently been published (EBCTCG 2005); see page 196.