This study shows a remarkable change in age-specific breast cancer incidence in Geneva. In developed countries, other than Japan, the typical age incidence curve of breast cancer described a progressive increase of risk with age, with a slope down around the menopause age, called the Clemmensen's hook. This typical curve by age is no longer observed in Geneva, where the risk of developing breast cancer does not increase with advancing age anymore. The highest breast cancer risk is now observed among women aged 60–64 years with strongly decreased risks among older women. The incidence peak of women aged 60–64 years was only observed in early stage disease and in oestrogen receptor positive tumours. In the subgroup of women for whom information on HRT use and mammography screening was available, the incidence peak was only present among women who ever used HRT.
The prevalence of screening and HRT use are high in our study population and we had the opportunity to examine the respective effects of each of these factors on the change in age of breast cancer occurrence. However, we realise that our study has several shortcomings. Information on HRT use was available only for approximately 50% of the women diagnosed in the period 1997–2000. Although these women were recruited in the context of a population-based case-control study, they might not be completely representative of all breast cancer patients. We compared them with the other 50–74 years old breast cancer patients recorded at the cancer registry and they appeared to be similar in terms of age, social class, stage distribution and tumour characteristics. Therefore, we believe that selection bias does not explain the fact that the new age distribution is only present among women who used HRT, irrespective of their mammography screening status. In addition, we have no information on the duration of HRT utilisation, on HRT cessation, or on time elapsed since the stop of HRT. It is therefore not possible to evaluate if the incidence peak occurred among current or ex-users of HRT. Also, we were not able to evaluate incidence patterns according to type of HRT.
To our knowledge, only three studies have reported important changes in age-specific breast cancer incidence [14–16]. In Marin Country, Bay area, San Francisco, Prehn et al. observed an increase in breast cancer incidence among women aged 45–64 years for the period 1991–1997, whereas the incidence among older and younger women remained stable [14]. They ruled out screening mammography as the most important reason, because the incidence increase involved also advanced stages and lobular cancers (which are difficult to detect mammographically), and because women aged 65–70 (an age-category also covered by screening) showed no incidence increase. The authors suggested HRT, frequently used by the relatively wealthy, well-educated female population of this area, as a possible explanation for the increasing breast cancer incidence among women aged 45–64. It was in this same region (Bay area) that the increase of endometrial adenocarcinoma, associated with postmenopausal oestrogen use, was particularly high [17]. More recently, Hemminki et al. also reported an important change in age-specific breast cancer incidence in Sweden [16]. They observed a strong incidence increase among women aged 50–69 years, but only a slight decrease for women older than 75 years. As this change in age-specific incidence coincided with the introduction of organised mammography screening programs, the authors attributed this incidence pattern change to mammography screening. Fuglede et al, reported on breast cancer incidence in a population of unscreened women in Denmark from 1973–2002, using data of the nationwide Danish Cancer Registry [15]. They showed that the age-specific incidence rates of breast cancer increased throughout the whole period. In addition, they observed marked changes in the age-specific incidence pattern: between 1973–1981, there was a plateau and change of slope around the age of 46–48, which shifted to 64–66 years in 1994–2002. This was not due to screening as they confined their study to a non-screened population [15].
In our population, we believe that both HRT and, to a lesser extent, mammography screening explain the change in incidence pattern. Mammography screening advances diagnosis by detecting breast cancer at an early pre-clinical stage. This can explain why the change in pattern concerned mainly the early stages. In this study however, we did not observe the typical incidence peak among never users of HRT with screen-detected tumours. In fact, only ever users of HRT showed an incidence peak at 60–64 years, regardless whether the tumour was screen-detected or not.
It has been suggested that HRT particularly increases the risk of oestrogen receptor positive tumours [18]. In our study, the changed incidence pattern involved only women with oestrogen receptor positive tumours, which supports our hypothesis of a potential effect of HRT use on the change in age-specific breast cancer risk.
Recent studies showed that current users of HRT are at increased risk of developing breast cancer [5, 19]. These studies generally involved patients between 50–65 years and the effect of ever use of HRT on breast cancer incidence among elderly women (aged ≥ 75 years) has, to our knowledge, never been examined.
Our results suggest that HRT use might not only increase breast cancer risk among middle-aged women. It may also bring forward the clinical appearance of breast cancer by several years by stimulating the growth of oestrogen sensible lesions and by inhibiting spontaneous slow down of tumours when sex hormone levels drop during menopause. This hypothesis is supported by a previous study, which showed that postmenopausal women using HRT were on average five years younger upon breast cancer diagnosis than never users [20].
Use of HRT could have eliminated the temporary protective effect of menopausal oestrogen depletion.