Skip navigation

Tamoxifen trials, tribulations and truths

Tamoxifen has been used in the treatment of breast cancer for about 25 years, but it has only been used in prevention trials, involving large numbers of women, for about 6 years, (although the first pilot study was started in 1986). The effectiveness of tamoxifen in treatment of early breast cancer and its role in prevention of breast cancer in higher risk women have recently been featured in a number of publications.

Treating early breast cancer

The Early Breast Cancer Trialists Collaborative Group led by Richard Peto and Rory Collins from the Clinical Trials Service Unit in Oxford have released an updated overview of 55 worldwide randomised trials [1] involving 37,000 women treated for early breast cancer and followed for at least 5 years, many more than 10. The overview is thought to include about 87% of the worldwide relevant evidence.

In nearly 8,000 of the women known to have a low or zero level of oestrogen receptors in their primary tumour the effects of tamoxifen treatment appeared to be small and these women were excluded from the subsequent analysis. Of the remainder, 18,000 women were known to have oestrogen receptor +ve tumours but in nearly 12,000 women receptor status was not measured (about 8,000 of these would be expected to have +ve tumours).


In women treated with tamoxifen for 1, 2 or about 5 years, the reduction in recurrence rates during up to 10 years follow up was 21%, 29% and 47% respectively, the reduction in the incidence of contralateral breast tumours was 13%, 26% and 47% and the reduction in mortality was 12%, 17%, and 26%. The NNTs calculated from the paper for the main results for recurrence and mortality are given in Table 1. For every eight women given five years of tamoxifen treatment one would have a recurrence prevented - an NNT of 8 (95%CI 7 to 10).
Table 1: Tamoxifen in early breast cancer treatment
Outcome Years of tamoxifen NNT (95%CI)
Prevent recurrence 1 18 (13 to 30)
  2 16 (13 to 26)
  5 8 (7 to 10)
Prevent death 1 28 (18 to 66)
  2 30 (21 to 49)
  5 22 (15 to 36)
    NNH (95%CI)
Endometrial cancer 5 97 (68 to 168)


The incidence of endometrial cancers was higher with tamoxifen; in absolute terms these tumours were only about half the number of contralateral breast tumours prevented. Overall, in three large trials which used tamoxifen treatment for an average of five years, the number needed to harm to produce one extra case of endometrial cancer was 97 (95%CI 68 to 168).

Looking good

The report notes that whereas in earlier studies younger (pre-menopausal) women seemed to benefit less than older (post-menopausal) women, the benefits found in this overview apply about equally to women of all ages. Although the report does not make specific recommendations about who should or should not be treated, (because, it argues, such decisions involve factors such as side effects and cost not reviewed in this study) nevertheless the evidence presented is strongly in favour of advising all women with oestrogen receptor positive early breast cancer to have "some" years of adjuvant tamoxifen treatment. In a separate publication Rory Collins is quoted as saying that such treatment given worldwide could save up to 20,000 lives per year. This has to be good news.

Preventing breast cancer

In stark contrast to this rather clear and relatively easily understood study stands the controversy provoked by the early disclosure of results by the US Breast Cancer Prevention Trial. This trial was started in April 1992 and closed enrolment in September 1997 with 13,388 women. The data were regularly reviewed by an independent Endpoint Review, Safety Monitoring and Advisory Committee and at its meeting in March 1998 the committee decided that in its view the benefits of tamoxifen in reducing the incidence of breast cancer had been clearly demonstrated and it recommended that the trial should be unblinded, participants and their physicians should be told which pills they had been taking, and those on placebo, and other women at increased risk, should be offered tamoxifen. This recommendation was endorsed by the National Surgical Adjuvant Breast and Bowel Project and the National Cancer Institute and the decision to close the trial was announced on 6th April.

The results on which this decision was based are shown in in Table 2. There was a 45% reduction in breast cancer, based on a total of 239 cases, but the tamoxifen treatment group had more cases of endometrial cancer, pulmonary embolism and deep vein thrombosis. There was no significant difference in heart attack rates.

Table 2: Outcomes from US breast cancer prevention trial
  Number of events recorded in the
Outcome Tamoxifen arm Placebo arm
Invasive breast cancer 85 154
Endometrial cancer 33 14
Pulmonary embolism 17 6
Deep vein thrombosis 30 19
Fractures (hip, spine, wrist) 47 71
Deaths from breast cancer 3 5
In the trial 13,388 women were followed for a mean of 3.5 years
The way in which the news of this trial was trumpeted around the world in the media suggested that adopting tamoxifen for prevention of breast cancer was a "no brainer". A quiet look at the numbers of events actually recorded in the trial might give one pause: there were very few. Even more pause comes from the number of adverse events recorded. People can choose themselves how to weight these outcomes, but looking at "all bad things", it is difficult to see all that much difference.

Two more trials

UK experts in this field immediately voiced their disquiet at the US decision to terminate the trial 14 months before originally planned. They felt that the follow up was too short to get a secure view of the likely long term benefits and complications. They were also concerned about the effect of the US decision on their own ongoing tamoxifen (IBIS) trial, making it more difficult to recruit women willing to be randomised to tamoxifen or placebo groups.

These concerns seem to have been vindicated by the publication of the interim results of two chemoprevention studies with tamoxifen in the Lancet of July 11th. That by Powles and colleagues at the Marsden Hospital [2] is a trial which has been running longer than any other. It was originally set up as a pilot study and only extended when regulatory problems delayed the start of the UK multicentre trial. The second report is of a trial undertaken by Veronesi and colleagues [3], largely in Italy. The UK study has recruited 2471 women and followed them for an average of nearly 6 years; the Italian study has 5408 women and an average follow up of just under 4 years.

In neither of these studies was there a statistically significant reduction in the number of breast tumours found in the tamoxifen treated women. A number of explanations for the differences have been offered but none seem convincing. Smaller studies obviously have less power to detect small differences, but both the UK and the Italian studies would have been capable of detecting a difference as great as that found in the US study. The UK study had more younger women with a stronger family history; perhaps these womens' tumours are less susceptible to prevention. They were also treated and followed up for longer than the US study and it has even been suggested that longer treatment with tamoxifen might be counter productive. The Italian study was different from the US one in that it recruited women who had had a hysterectomy, younger on average than the Americans and there was a high rate of drop-out rate, 26%.


The end point that is ultimately the most important outcome is likely to be mortality. None of these studies can yet tell us anything very useful about that. Longer follow up of the completed US trial and the ongoing ones in the UK and Europe is essential. The decision of the investigators and data monitoring committee of the IBIS trial not to unmask the results and indeed to continue to recruit patients seems a wise one.

This situation illustrates well, if slightly sadly, how difficult it is to reach universally acceptable decisions on what might at first sight might appear rather clear results. As far as the tamoxifen prevention "truth" is concerned the jury is still definitely out.


  1. Early Breast Cancer Trialists' Collaborative Group. Tamoxifen for early breast cancer: an overview of the randomised trials. Lancet 1998 351: 1451-67.
  2. T Powles, R Eeles, S Ashley et al. Interim analyisis of the incidence of breast cancer in the Royal Marsden Hospital tamoxifen randomised chemoprevention trial. Lancet 1998 352: 98-101.
  3. U Veronesi, P Maisonneuve, A Costa et al. Prevention of breast cancer with tamoxifen: preliminary findings from the Italian randomised trial among hysterectomised women. Lancet 1998 352: 93-7.

previous or next story in this issue