HRT and the incidence of ovarian and breast cancer

HRT and the incidence of ovarian and breast cancer

Two major epidemiological studies published today look at the possible increased risk of ovarian and breast cancers associated with hormone replacement therapy (HRT). They have, inevitably, generated significant media interest.

A study published early online by the Lancet found that current use of HRT for prolonged periods is associated with a 20% increase in risk of ovarian cancer. The authors used data from the Million Women Study, a large prospective cohort study of UK women aged 50 and over that is investigating a range of factors that affect women's health. Data collected at recruitment includes whether the woman is using HRT, confirmed by subsequent questionnaire three years later if possible. Each participant is followed-up for death, emigration, or cancer registration, so that relevant events are reported to the study investigators.

This analysis aimed to determine whether use of HRT affected risk of ovarian cancer; it involved women from the cohort who were post-menopausal, had no history of cancer or bilateral oophorectomy, and for whom data on use of HRT was available. They were followed for averages of 5.3 years for incident ovarian cancer and 6.9 years for death. Primary outcome of the analysis was relative risk for ovarian cancer, adjusted for a range of relevant factors.

A total of 1.3 million women were recruited into the Million Women Study cohort. Of these, 948, 576 were included in the analysis. When they last reported data to the study, 50% reported current or past use of HRT - 30% were current users and 20% past users: the two groups were fairly similar with only past use of oral contraceptives and hysterectomy being major differences between them (both higher in HRT users). Mean duration of use was 7.7 years.

There were 2,273 ovarian cancers in total and 1,591 deaths from ovarian cancer. Women who were current users of HRT had a statistically significant increase in risk both of ovarian cancer (relative risk 1.20, 95% CI 1.09 to 1.32; p=0.0002) and death from ovarian cancer (RR 1.23, 95% CI 1.09 to 1.32; p=0.0002). Risk increased with duration of use, however past users were not at significantly increased risk. Crude incidence rate in the study population as a whole was 2.2 per 1,000 women; in never-users it was 2.2 per 1,000, and in users 2.6 per 1000 (rates for death from ovarian cancer 1.3, 1.3, and 1.6 per 1,000 respectively). From these figures, assuming that the differences are due to HRT the authors calculate that over a five year period there would be about one extra case of ovarian cancer for every 2,500 users (and one extra death per 3,300 users).

The authors conclude that women who use HRT are at greater risk of ovarian cancer and of death due to it. The risk increases with duration of use: mean duration of use in the study population was 7.7 years and this was associated with a 20% increase in risk. It is related to current use, not past use, and falls back to baseline level soon after use is stopped. An accompanying Comment discusses the paper and its implications; the author discusses some potential mechanisms for the effect, and notes the difference between pre-menopausal use of the same hormones as oral contraceptives, which is protective, and the data on HRT. He notes that although the absolute increase in risk is small, the large number of women who used HRT until recently meant that the toll in terms of cancers and mortality was significant.

The second paper reports that the rate of breast cancer in the US fell appreciably in 2003 compared to the rate in 2002; while a number of explanations are possible, the decrease seems to be related to reporting of results from the Women's Health Initiative in 2002 that was followed by a decrease in HRT use among post-menopausal women in the US.
The authors use data from a program (SEER) run by the US National Cancer Institute: this collects data from nine cancer registries reporting on 9% of the total US population. Data from the registries were adjusted for reporting delays and standardised to the US female population in 2000. Annual incidence rates were plotted and compared to determine trends.

The results show a sharp fall in incidence of 6.7% in 2003 compared to 2002; the incidence stabilised in 2004 with little further decrease. Analysis showed that the decrease actually started in mid-2002 and had begun to level off by mid-2003, and comparison between 2001 and 2004 showed an overall decrease of 8.6% (95% CI 6.8 to 10.4%). The decrease was age-specific and occurred only in women aged over 50; there was no significant change in incidence in younger women. It was also primarily in oestrogen-receptor positive tumours. Examination of trends since 1975 showed a steady increase in women aged over 50 from the late 1970's onwards, with rates rising by about 0.5% per year during most of the 1990's. Rates in women under 50 were substantially the same over the whole of this period.

The authors discuss a number of possible reasons for the changed incidence, but conclude that reductions in HRT use are the most likely. There is evidence that stopping HRT may cause clinically occult oestrogen-receptor positive breast cancers to stop growing or even regress soon after withdrawal, and evidence of studies involving hormonal manipulation therapies (e.g. tamoxifen) supports this. HRT prescriptions in the US declined sharply during 2002 and 2003 and stabilised at a lower level in 2004, from around 60 million to 21 million prescriptions per year.

The MHRA has issued a statement on the two papers. The new evidence is being reviewed by experts, but is unlikely to result in major changes in current advice: this is that HRT is effective for short-term relief of menopausal symptoms but should be used in the lowest effective dose and the minimum duration.

Ovarian cancer study: Lancet, published early online 19 April 2007; DOI:10.1016/S0140-6736(07)60534-0 (link to abstract) and Lancet, published early online 19 April 2007; DOI:10.1016/S0140-6736(07)60535-2 (Comment; link to full text, may be available to subscribers only);
breast cancer study: New Engl J Med 2007; 356: 1670-4 (link to abstract, full text available free at time of posting)
The MHRA statement is available here
BBC News report;
there is full information on the Million Women Study on its website