The late emergence of survival benefits in adjuvant trials of endocrine therapy

There is an overwhelming amount of data available evaluating the survival benefit from adjuvant tamoxifen. Vast numbers of women have been randomly assigned in trials between tamoxifen and no tamoxifen.

Initially, there was a definite difference in recurrence, with no definite difference in breast cancer mortality. Then there was a definite difference in recurrence and in breast cancer mortality.

In many of the breast cancer trials, the major differences in mortality become evident a decade after the treatment is initially given. There is little difference in the first few years and then a major difference emerges five, 10, or 15 years after treatment.

It could well be that this trend will hold true with the ATAC trial. There is a promising difference in early recurrence, and it might well be that over time, as with tamoxifen, that will translate into differences in breast cancer mortality and overall mortality.

Tamoxifen had almost no effect on mortality when the average follow-up in the trials was just two years. There is a definite, but small, effect on five-year mortality. However, the effect on 15-year breast cancer mortality is more than twice as great as the effect on five-year mortality.

The difference in mortality after the first five years is bigger than the difference in mortality during the first five years. So, what appears to be a small difference in mortality after a short-term follow-up may translate into quite a substantial difference in mortality at a 15-year follow-up.

ER status and breast cancer mortality

For women with hormone-sensitive breast cancer, the death rate from breast cancer remains high throughout the first and second decade. However, that does not hold true for ER-negative disease. In ER-negative disease, there is a very high death rate in the first five years, which declines 10 to 15 years after treatment.

Comparatively, the 15-year risk of breast cancer mortality is actually much the same for both ER-positive and ER-negative disease, although the five-year risk of death from breast cancer is much higher for ER-negative disease.

Because of the prolonged period of high mortality associated with ER-positive disease, it is important to take a 20-year perspective on treatment. The focus for therapy of breast cancer should extend beyond the first decade and into the second decade. This holds true for hormonal therapy, radiotherapy and chemotherapy.

In women under 50 years of age, chemotherapy has a very large effect on early recurrences. It has a moderate effect on five-year mortality, but it has twice the effect on 15-year mortality. In ER-positive disease, chemotherapy has the greatest effect on breast cancer mortality after the initial five years of treatment.

Declines in breast cancer mortality

During the 1990s there was a big drop in the national mortality rates from breast cancer, first in the United Kingdom and then in other countries, including the United States (4.1). This decrease in breast cancer mortality has remained steady or improved, while the incidence rate of breast cancer is slightly increasing as women have fewer children, have their first child at an older age and have more body fat after menopause.

In the 1980s there was an improvement in breast cancer treatment with widespread use of hormonal and chemotherapeutic adjuvant regimens. These improvements in treatment during the 1980s produced a reduction in breast cancer mortality during the 1990s. Further improvements in breast cancer treatment during the 1990s are going to keep breast cancer mortality rates falling during the present decade.

The mortality rate has already decreased by a third, and continues to fall. The change in practice in one decade produces the trends in mortality in the next decade. That is really the pattern and that’s what one should expect from trial results.

We have evidence that better local control produces a small but real difference in later mortality from breast cancer. Therefore, earlier diagnosis results in a reduction in breast cancer mortality. In addition to national screening programs, there’s been an increase of breast cancer awareness over the last few decades, which may have led to earlier diagnosis and downstaging.

In Britain, the national screening program began in 1991. So, the main benefits from screening are going to be seen during the present decade, as well as during the next decade.

The improvements in breast cancer control during the 1990s will translate into decreases in national death rates during the first and possibly the second decade of this century. Therefore, the decrease in breast cancer mortality during the 1990s is not chiefly due to screening; it is due to the changes in management that occurred during 1980s.

Meta-analysis of local therapy trials

Over the past few decades, there have been nearly 100 randomized trials of different methods of achieving local control in breast cancer: radiotherapy versus no radiotherapy or more surgery rather than less surgery or more surgery versus less surgery but adding radiotherapy. Some of these therapies do not make very much difference in local control, so it is not surprising that there is no material effect on long-term survival (Peto 2004).

When you look at treatments that involve a major difference — for example, a 30 percent local recurrence risk versus a 10 percent local recurrence risk — then you are going to get a mortality difference of around five percent — 50 percent mortality versus 45 percent mortality — at 15 years. So, there’s a real effect.

The conclusion is that local control does matter. It is not an enormous effect, but it is a real effect. A difference of about 20 percent in the five-year risk of local recurrence translates to approximately a five percent difference in the probability of death from breast cancer over the next two decades.

Select Publications

Professor Sir Richard Peto is Professor of Medical Statistics and Epidemiology and co-founder and co-director of the Clinical Trial Service Unit at the University of Oxford in the United Kingdom.