4–5 Breast tumours following combined hormone replacement therapy express favourable prognostic factors

Breast Diseases: A Year Book Quarterly. 2008;18(4):347-348, R.L. Prentice

The aim of the present study was to evaluate the association between different types of hormone replacement therapy (HRT) and risk of specific breast cancer subgroups. A population-based prospective cohort study including 12,583 peri- or postmenopausal women were followed using record-linkage with national cancer registries. During an average follow-up of 4.5 years, 332 cases of invasive breast cancer were diagnosed. Tumour samples were available from 283 cases. These tumours were re-evaluated according to histological type, grade, and mitotic index. Evaluation of tumours included estrogen and progesterone receptor status (ERα, ERβ and PgR), as well as expression of Ki67, HER2, cyclin D1 and p27. The incidence of breast cancer in current users of combined HRT (CHRT) was significantly higher than in non-users. The difference corresponded to an adjusted relative risk (95% confidence interval) of 3.01 (2.35–3.84) as obtained using a Cox's proportional hazards analysis. CHRT was associated with lobular tumours (3.48:1.99–6.10), grade 1 tumours (4.46:2.79–7.13) and tumours with a low mitotic index (4.35:2.99–6.34). CHRT was not related to any specific subgroup in terms of ERα-, ERβ- or PgR-expression. CHRT was associated with low proliferating tumours, defined by the Ki67 index (3.58:2.60–4.93), HER2 amplified tumours (4.40:1.93–10.06), low expression of the oncogene cyclin D1 (3.14:2.32–4.23) and high expression of the tumour suppressor gene p27 (3.47:2.40–5.01). Use of estrogen-alone HRT (ERT) was not associated with any statistically significant risk of breast cancer. We conclude that the use of CHRT is associated with an increased incidence of breast tumours with comparatively favourable prognostic factors.

This article by Borgquist and colleagues adds to a considerable body of evidence that CHRT substantially increases breast cancer risk. This finding is in contrast with that for postmenopausal ERT, which evidently has a much smaller effect, if any.

An increased risk of breast cancer was the principal trigger for the early stopping of the Women's Health Initiative (WHI) CHRT trial after a mean follow-up of 5.6 years.1,2 The elevation in breast cancer risk was accompanied by increased incidences of stroke, venous thromboembolism, and early coronary heart disease, as well as reduced incidences of hip and other fractures. The collective data led the authors to conclude that the health risks exceeded the benefits for the daily regimen of CHRT that was studied (0.625 mg conjugated equine estrogens plus 2.5 mg medroxyprogesterone acetate).1

Because an increased risk of breast cancer is among the major safety concerns with HRT, it would be of interest if the tumors attributable to CHRT were of comparatively favorable prognosis, as Borgquist and colleagues report. However, this type of analysis is difficult for various reasons: (1) the hazard ratio associated with CHRT increases with time from initiation of use, (2) CHRT may hinder breast cancer diagnosis, and (3) mammographic screening practices tend to differ between CHRT users and nonusers, which plays a major role in determining tumor size and disease stage at diagnosis.

Borgquist and colleagues ascertained CHRT or ERT use between 1991 and 1996 and followed patients through the end of 2001, a follow-up period of 5–10 years. Interpreting the magnitude of relative risks in such an analysis requires knowledge of the duration of CHRT at the time of enrollment, in view of strong hazard ratio dependencies on duration of use.2

In the absence of CHRT influences on mammographic screening effectiveness, the higher frequency of screening among CHRT users could cause a substantial upward bias in relative risk estimates, and the authors do not appear to have standardized for this important factor (eg, by censoring follow-up times when a woman has not had a mammogram in more than 2 years). The authors note that an upward bias of this type could be offset by a downward bias because CHRT-induced mammographic densities may hinder breast cancer diagnosis. Whether this latter factor would yield an upward or downward bias, however, depends on the duration of CHRT use at enrollment. For example, if most current users had been using CHRT for some years at the time of enrollment, the breast cancer incidence in the user group might be in a “catch-up” phase during cohort follow-up, resulting in further relative risk overestimation. In addition, biases related to screening effectiveness and diagnosis could vary with the size and growth characteristics of the tumor.

Other features of this study that complicate the interpretation of findings involve the apparent inclusion of prevalent cases at study enrollment, the inclusion of perimenopausal women, the use of a common nonuser group regardless of uterine status for both ERT and CHRT, the lack of allowance for nonadherence to baseline HRT status during cohort follow-up, and the inability to stratify results by the specific HRT regimen used. However, these limitations and biases seem to be related to the magnitude of relative risks rather than to comparative relative risks among tumors of differing characteristics. For example, the somewhat higher relative risk for lobular tumors than for ductal tumors, even though not seen in the WHI trial, has been observed in a number of other observational studies and may provide a useful lead for understanding the effects of CHRT in the breast. Also, the higher relative risk associated with tumors with a low metastatic index could reduce concern about breast cancer induced by CHRT, although tumors diagnosed in the CHRT group seemed to be slightly more aggressive in terms of metastasis than did those arising in the placebo group in the WHI trial.2