Pre-Empting Breast Cancer: Where Do we Go from Here?

‘…future attempts to improve breast cancer risk models by incorporating parameters such as breast density and estrogen levels should separate premenopausal from postmenopausal women and segregate risk by hormone-receptor status,’

Despite the recent decrease in breast cancer incidence reported in the USA [1], breast cancer remains the most common malignancy diagnosed in women in this county with 178,480 cases of invasive breast cancer estimated for 2007 [2]. There are now two agents with US FDA label indication for breast cancer risk reduction in women who are at increased risk for breast cancer: the selective estrogen-receptor modulator tamoxifen for both pre- and postmenopausal women [3], and raloxifene for use only in postmenopausal women [4]. Use of these agents in appropriate candidates reduce breast cancer incidence by approximately 40–50% [4]. To identify such candidates, several breast cancer risk-assessment tools are available, including the Tyler-Cuzick model [5] and the Breast Cancer Risk Assessment Tool of the National Cancer Institute (Gail model), which has recently been modified to more accurately predict risk in African–American women [6,101].

Against this background, it might be expected that breast cancer risk assessment would be a routine component of physician primary care visits, especially for postmenopausal women, as risk of hormone-receptor positive breast cancer increases with age, and both tamoxifen and raloxifene specifically target hormone-receptor positive disease. However, this is not the case [7–9]. In a recent year, only 16% of US primary care providers reported that ‘it is easy to determine’ who is eligible for breast cancer risk-reduction strategies [9], and only 11% of primary care providers in California, USA, reported having used the Gail model in the past year [8].

It is not clear why breast cancer risk assessment is not more commonly practiced and why use of approved agents to reduce breast cancer risk has been limited [10]. There has been concern regarding the side effects of tamoxifen use, especially the risk of developing endometrial cancer, and the association of tamoxifen as being a ‘cancer drug’. Full discussion of the risks and benefits of tamoxifen and raloxifene for breast cancer risk reduction is beyond the scope of this commentary. However, raloxifene is not a cancer drug, it is approved for osteoporosis prevention and therapy and was shown to reduce invasive breast cancer incidence by a similar degree as tamoxifen with less endometrial cancer risk in a head-to-head, randomized, Phase III clinical trial comparison [4]. Thus, raloxifene provides a more attractive agent for breast cancer risk reduction for postmenopausal women, especially those with no prior hysterectomy. In addition, the perceived complexity of the breast cancer risk-assessment process itself has also been seen as a barrier to clinical implementation of breast cancer risk assessment [11]. The Gail model requires answering a series of questions and subsequent use of a computer-based program to calculate the 5-year and lifetime breast cancer risk estimates. For entry into prevention trials, a 5-year breast cancer risk of 1.67% or greater has been required. While a variety of user-friendly devices to facilitate this calculation are available, the need for a calculation has hindered widespread use.

‘…only 16% of US primary care providers reported that ‘it is easy to determine’ who is eligible for breast cancer risk-reduction strategies.’

To potentially address this issue, we developed a simpler breast cancer risk model, which did not require formal calculation, to specifically target the risk of hormone-receptor positive breast cancers developing in postmenopausal women [12].

Our study population was the observational study and clinical trial cohort of the Women's Health Initiative (WHI), which included 147,916 eligible postmenopausal women with 3236 invasive breast cancers. Several logistic regression models were developed and ‘trained’ on the observational study cohort and tested on the clinical trial cohort. The new model that emerged incorporates only three risk factors – age, previous benign breast biopsy (regardless of findings) and breast cancer history in first-degree relatives. When evaluated using receiver operator curves and area under the curve statistics, the simpler WHI model was as effective as the Gail model in identifying postmenopausal women at increased risk of developing hormone-receptor positive breast cancer and does not require computer use. Breast cancer risk can be easily estimated from tabular displays.

Women who are 55 years of age or older with either a first-degree relative with breast cancer or a previous breast biopsy have a 5-year risk above 1.8%, which exceeds the Gail threshold of 1.67%. Postmenopausal African–American women are recognized to be at a lower risk of developing receptor-positive breast cancer [13]. When we applied this WHI model to our African–American population, women who were 60 years of age or older with either a first-degree relative with breast cancer or a previous breast biopsy were found to have a 5-year risk above 1.8% of developing a receptor-positive invasive breast cancer.

Using the WHI model, postmenopausal women could easily determine for themselves if they would be potential candidates for breast cancer risk-reduction interventions if this simpler model received wide, general dissemination. The use of this simpler model and the availability of an agent, raloxifene, with favorable influence on both bone health and breast cancer risk in postmenopausal women provides an opportunity to move breast cancer risk reduction more broadly into general clinical practice.

Breast cancer is now recognized as representing several distinct entities and the influence of risk factors have been shown to differ depending on menopausal status [12,14]. For breast cancer investigators, future attempts to improve breast cancer risk models by incorporating parameters such as breast density and estrogen levels should separate premenopausal from postmenopausal women and segregate risk by hormone-receptor status. For physicians of any speciality providing primary care to postmenopausal women, the availability of this new simple WHI model to identify women who could benefit from available breast cancer risk-reduction interventions should be considered for use in clinical practice.