Overview of the main outcomes in breast-cancer prevention trials

BACKGROUND: Early findings on the use of tamoxifen or raloxifene as prophylaxis against breast cancer have been mixed; we update available data and overview the combined results.

METHODS: All five randomised prevention trials comparing tamoxifen or raloxifene with placebo were included. Relevant data on contralateral breast tumours and side-effects were included from an overview of adjuvant trials of tamoxifen versus control.

FINDINGS: The tamoxifen prevention trials showed a 38% (95% CI 28-46; p<0.0001) reduction in breast-cancer incidence. There was no effect for breast cancers negative for oestrogen receptor (ER; hazard ratio 1.22 [0.89-1.67]; p=0.21), but ER-positive cancers were decreased by 48% (36-58; p<0.0001) in the tamoxifen prevention trials. Age had no apparent effect. Rates of endometrial cancer were increased in all tamoxifen prevention trials (consensus relative risk 2.4 [1.5-4.0]; p=0.0005) and the adjuvant trials (relative risk 3.4 [1.8-6.4]; p=0.0002); no increase has been seen so far with raloxifene. Venous thromboembolic events were increased in all tamoxifen studies (relative risk 1.9 [1.4-2.6] in the prevention trials; p<0.0001) and with raloxifene. Overall, there was no effect on non-breast-cancer mortality; the only cause showing a mortality increase was pulmonary embolism (six vs two).

INTERPRETATION: The evidence now clearly shows that tamoxifen can reduce the risk of ER-positive breast cancer. New approaches are needed to prevent ER-negative breast cancer and to reduce the side-effects of tamoxifen. Newer agents such as raloxifene and the aromatase inhibitors need to be evaluated. Although tamoxifen cannot yet be recommended as a preventive agent (except possibly in women at very high risk with a low risk of side-effects), continued follow-up of the current trials is essential for identification of a subgroup of high-risk, healthy women for whom the risk-benefit ratio is sufficiently positive.

Contributors

Cuzick, J. and T. Powles, U. Veronesi, J. Forbes, R. Edwards, S. Ashley, P. Boyle

Publication

Journal: Lancet
Volume: 361
Issue: 9354
Pages: 296 - 300
Year: 2003
DOI:

Further Study Information

Current Stage: Not Applicable
Date:
Funding source(s): J Cuzick and J Forbes have served as occasional consultants and advisory board members to AstraZeneca. T Powles serves on advisory boards for Eli Lilley and Pfizer. J Forbes has also served occasionally as consultant to Pharmacia and is the principal investigator for trials for which his institutions receive funding from AstraZeneca. J Cuzick is the principal investigator for a proposed new prevention study for which his institution has received an unrestricted supporting educational grant from AstraZeneca.


Health Area

Disease Category: Cancer

Disease Name: Breast cancer

Target Population

Age Range: Unknown

Sex: Female

Nature of Intervention: Drug

Stakeholders Involved

- None

Study Type

- Systematic review of outcomes measured in trials

Method(s)

- Literature review

Data from the International Breast Cancer Intervention Study I (IBIS-I), National Surgical Adjuvant Breast and Bowel Project P-1 Study (NSABP-P1), and MORE trial are based on published material, whereas those from the Royal Marsden Hospital and Italian studies update published reports. The adjuvant-trial data are taken from the 2000 overview and update a previous report, but they have not been previously published. Follow-up is until January, 2002, for IBIS-I and the Royal Marsden trial, February, 2001, for the Italian trial, and about January, 2000, for the adjuvant trials. No further follow-up is available for NSABP-P1 after March 31, 1998, or for the MORE trial after November, 1999. Results from a secondary prevention study on the use of the retinoid fenretinide on contralateral tumours are not included in this overview.

This report focuses on breast-cancer incidence, endometrial-cancer incidence, vascular events, all-cause mortality, and mortality from the above causes. All data were obtained directly from the original investigators or from published reports. The primary analysis was based on a fixed-effect model stratified by trial, by Poisson regression with woman-years of follow-up to cancer or death as the denominator. For this analysis, each event contributed equally, so the larger trials had more effect. Additional analysis with a random-effects model allowing for heterogeneity between trials was also undertaken and gave more equal weight to each trial.

Linked Studies

    No related studies


Related Links

    No related links