Background and Purpose: For women diagnosed with hormonal receptor positive breast cancer, adjuvant hormonal therapy is the standard of care treatment. For pre-menopausal women, tamoxifen is the most prescribed drug; for post-menopausal women, letrozole and aromatase inhibitors are used. Tamoxifen was shown to reduce breast cancer recurrence by 40% and the annual cancer death rate by 25–30%, irrespective of chemotherapy status and age [1]. Despite that, many patients still developed progressive diseases, raising the question of individual responsiveness [1]. Moreover, tamoxifen is known to be associated with many side effects; thus there is a strong interest to find biomarkers that can predict the responsiveness for each individual patient. Hormonal regimens may affect breast tissue with the change of breast volume or composition. It was noted that tamoxifen may induce a decrease in stromal, and an increase in adipose percent area [2]. In adjuvant setting, mammographic density (MD) change during short-term use of tamoxifen was a significant predictor of long-term recurrence. One study reported that women showing a relative reduction of 20 % in MD were associated with a 50 % risk reduction of BC-specific mortality [3]. Another study showed that the recurrence rate in women with no change in MD was more than doubled compared with those who had at least a 10 % decrease in MD [4]. Due to the limitation of two dimensional MD, a recent study investigated the use of 3D MRI to assess the density change following tamoxifen treatment, and showed similar findings [5]. But the authors did not describe detailed segmentation methods that were used for the density analysis on MRI. The aim of this study was to apply a well-established breast and fibroglandular tissue segmentation method to analyze the density changes in patients receiving adjuvant hormonal therapy. Results between pre-/peri- and post- menopausal women receiving different drugs were compared.

Materials and Methods: From October 2012 to November 2014, 81 women (age range 34-73, mean 51 y/o) with hormonal positive breast cancer were studied. The pre- and peri- menopausal women received daily Nolvadex-D (Tamoxifen, 20mg); and post-menopausal women received daily Femara or Lenara (Letrozole, 2.5mg) or Arimidex (Anastrozole, 1mg). All women had baseline (pre-treatment), and at least one follow-up (post-treatment) breast MRI studies. The duration from the start of hormonal treatment to the follow-up MRI ranged from 119 to 686 days (mean 288 days). MRI was acquired using a 3.0T Philips scanner. For MR imaging, non-fat-suppressed, non-contrast-enhanced T1-weighted images (T1WI) in axial section were acquired first, followed by the dynamic contrast-enhanced (DCE) MRI. The imaging sequence and parameters for the non-enhanced T1WI were: spin echo, TR/TE 620/10 msec, matrix 332 x 332, field of view 200x340 mm, slice thickness 3 mm, and gap 1 mm. The breast and fibroglandular tissue segmentation was performed using our template-based automatic segmentation method [6]. The difference in the percent density between the baseline and follow-up MRI were calculated. In this study, only the contralateral normal breast was analyzed.

Results: Fig.1 shows baseline and follow-up MRI from a 43 y/o pre-menopausal woman who demonstrates a clear reduction in the segmented density. Fig.2 shows the results from a 58 y/o post-menopausal woman with very little change in density. Fig.3a illustrates a clear decreasing trend of percent density with age. Fig.3b shows a moderate correlation between the absolute density reduction with the baseline density. Fig.3c plots the change of density with age and the received drugs. As noted in the figure, there was a high variation in women taking tamoxifen, from 5% increase to 20% decrease. The results in post-menopausal women taking all three drugs were similar, showing very little reduction, presumably due to the very low baseline density to begin with and not leaving much room for further decrease. The mean density reduction in pre-/peri-menopausal women was 3.1%±4.3%, which was significantly higher than the reduction in post-menopausal women (1.1%±1.5%, p<0.01). The change in density was not correlated with the duration of treatment (r<0.1).

Conclusions: The results show that pre-menopausal women had a higher density reduction, presumably due to their more abundant fibroglandular tissues that can be decreased, but a high variation was observed. Pharmacogenomics approach can be used to study how an individual's genetic inheritance affects the body's response to drugs. For example, Cytochrome P450 2D6 Polymorphisms were known to affect the metabolism of tamoxifen to the active metabolite Endoxifen that has a potent therapeutic effect for cancer control. The density reduction assessed by 3D MRI may be used as a surrogate marker to correlate with metabolic genotyping, and further used in combination to better predict patient’s prognosis.