Changes in magnetic resonance mammography due to hormone replacement therapy

Observational studies have shown that women receiving hormone replacement therapy (HRT) have a reduced risk for osteoporosis [1, 2] by preventive slowing of the decrease of bone mineral density in postmenopausal women. Protective effects against heart disease have also been claimed [3, 4]. In contrast, Grimes and Lobo recently reported an increased risk for cardiovascular disease and breast cancer in the Women's Health Initiative trial of HRT [5]. The relation of HRT and the increased risk for breast cancer has been controversially discussed for several years [6, 7].

It is well known that mammographic density of the breast is increased during HRT and potentially affects the diagnosis of breast cancer on mammograms [8, 9]. Increased density may therefore obscure mammographic masses or may decrease the detection rate of microcalcifications, resulting in more false-negative findings in mammography and, consecutively, resulting in a decrease in the sensitivity of mammographic screening [10]. Even the specificity of mammography seems to be reduced with postmenopausal HRT [11]. There are differences between women using combination HRT compared with women treated with estrogens alone [12].

The ultrasound examination of the breast is less influenced by HRT. It has been reported that formation of cysts detectable by ultrasound does not significantly correlate with HRT [13]. In women with breast tumors receiving HRT, an increase in the number of intratumoral vessels and an alteration of the resistance and pulsatile index were associated with the menopausal status [14].

Several reports have described hormonal influences and alterations of the contrast enhancement kinetics in dynamic magnetic resonance imaging of the breast. Alterations in T1 and T2 relaxation times depending on the phase of the menstrual cycle were reported in 1985 [15]. In contradiction, Martin and El Yousef did not observe differences of the T2 relaxation time in different cycle phases [16]. Regarding the contrast enhancement patterns in magnetic resonance mammography (MRM), it has been reported that during the second half of the menstrual cycle a generally higher contrast medium (CM) uptake of fibrocystic formations [17], and even of normal breast parenchyma [18], takes place. Diffuse or focal enhancement may occur in postmenopausal women who are taking HRT [19]. Other authors report volume changes of the breast or alterations of the tissue component such as the ratio between fatty tissue and breast parenchyma, the water content, and the fibroglandular fraction during the menstrual cycle [20, 21].

The purpose of the present study was to evaluate the effects of HRT on contrast-enhancement kinetics in MRM in postmenopausal women.

Between December 1994 and April 1999, 1420 patients who had lesions of uncertain dignity in X-ray mammography or ultrasound of the breast underwent MRM at our institution. The patients gave their written consent prior to the examination. Two hundred and fifteen postmenopausal women among these patients (mean age, 52 ± 12.2 years) were treated with HRT. They were retrospectively divided into four groups (groups A–D) according to the time of onset or withdrawal of HRT and of MRM: 150 patients received 1 MRM during HRT (group A), 13 patients underwent 2 MRMs under HRT (group B), 30 patients received 1 MRM during HRT and 1 MRM within 1–3 months after HRT withdrawal (group C), and 22 women received 1 MRM within 1–3 months after withdrawal of hormonal medication (group D) (Table 1).

All MRM studies were performed on a 1.5 Tesla magnetic resonance scanner (ACS II; Philips, Best, The Netherlands). Patients lay in a prone position on a dedicated double breast coil. Subsequent to acquisition of scout images in three orthogonal orientations, a T1-weighted two-dimensional gradient echo sequence (repetition time/echo time/flip angle = 12/4.1/30; thickness, 4 mm; field of view, 350 × 350 mm²; matrix, 256 × 256; seven slices; acquisition time, 1 min and 25 s) in the coronal orientation and a T2-weighted turbo spin echo sequence (repetition time/echo time/flip angle = 14286/300/90; turbo factor, 29; slice thickness, 4 mm; gap, 0.4 mm; field of view, 350 × 350 mm²; matrix, 256 × 256; 24 slices; acquisition time, 3 min and 5 s) in the transverse orientation were acquired. Dynamic contrast-enhanced scans were performed using a multislice, T1-weighted, two-dimensional gradient echo sequence in the transverse orientation (repetition time/echo time/flip angle = 96/5/80; slice thickness, 4 mm; gap, 0.4 mm; field of view, 350 × 350 mm²; matrix, 256 × 256; 24 slices; acquisition time, 1 min) at 1-min intervals before and after intravenous administration of a bolus of 0.1 mmol/kg dosage of Gd-DTPA (Magnevist; Schering, Berlin, Germany) for 8 min.

For semiquantative analysis of the dynamic study, time–signal intensity (SI) curves were calculated and subtraction of the precontrast images from the postcontrast images was performed using the built-in software of the magnetic resonance scanner. Regions of interest were placed in each area or spot that showed CM enhancement. Five curve types concerning the amount of CM enhancement were predefined according to the report of Kaiser and colleagues [22]. The curves were characterized by minimal enhancement (up to 25% increase of the SI after administration of Gd-DTPA; curve type I), weak continuous enhancement (increase of SI between 25% and 60%; curve type II), strong continuous enhancement (increase of SI between 60% and 80%; curve type III), and a steep initial slope (more than 80% increase of SI) followed by a plateau phenomenon (curve type IV) or by a washout effect (curve type V) (Fig. 1). Curve type IV is suspicious and curve type V is highly suspicious for a malignant tumor in MRM. Curve types I–III are more typical for benign lesions [22]. In cases where several areas or spots with different curve types were present in the same breast, the curve type most suspicious for malignancy was chosen for evaluation.

The Kruskal–Wallis test for single-ordered row × column tables was used to evaluate statistical differences between the incidence of the five curve types of all patients examined during HRT (group A, group B and group C) and those patients who underwent MRM without hormonal medication (group C and group D). This test was also used for testing whether group D, which was examined without HRT, differed notably in its SI curve distribution compared with all other patients receiving HRT (group A, group B and group C) and whether group D would thus be representative for that sample. For statistical evaluation of differences between the curve types achieved in the two examinations during and after withdrawal of HRT in group C, an exact Wilcoxon signed rank test for dependent samples was performed. The same test was used to evaluate differences between the two measurements during HRT in group B.

Of all patients receiving HRT at the time of examination (group A; group B, first examination; group C, first examination), 60 of 193 patients (31.1%) showed curve type I, 88 patients (45.6%) were referred to as curve type II, and 45 patients (23.3%) showed curve type III (Fig. 2). CM enhancement characteristics were homogeneous or spotted granular, were continuous over the time of dynamic imaging and were symmetrical in both breasts. One of the patients with curve type II had an additional segmental spotted granular CM enhancement according to curve type IV in the right breast. The histological evaluation after surgical excision revealed a ductal carcinoma in situ in the right breast. One of the patients with curve type III revealed a breast lesion on the right side, which presented CM enhancement according to curve type V with an increase of 112% of SI after CM administration suspicious of breast cancer. Histological evaluation after surgery revealed an invasive lobular cancer.

Patients who received MRM under HRT (n = 193) showed highly significant differences in the contribution of the curve types compared with the patients who were free of hormone medication at the time of MRM (group D and group C after withdrawal of HRT) (P < 0.0001). Forty-two of these 52 patients (80.8%) showed curve type I, 8 patients (15.4%) showed curve type II, and 2 patients (3.8%) were referred to as curve type III (Fig. 2).

Twenty-eight of 30 patients of group C changed to lower curve types in the second MRM session after withdrawal of HRT. Twelve patients dropped by two curve types after withdrawal of HRT (Figs 3 and 4). Sixteen patients dropped by one curve type after stopping HRT and two patients remained at curve type II. The Wilcoxon signed rank test revealed that the curve types under HRT are highly significant different compared with the curve types after withdrawal of hormone medication (P < 0.0001).

The current study demonstrates bilateral, symmetrical, synchronous, and progressive CM enhancement in the breast in dynamic MRM according to the findings of the majority of the patients receiving HRT. Classifying these patients in time signal intensity curve types revealed enhancement patterns according to curve types I, II, and III that are not suspicious of malignancy.

The statistically significant difference in curve types with and without HRT underlines the presence of hormonal effects in MRM. Neither a plateau phenomenon nor a washout effect, which are known to be typical for invasive breast cancer [23], occurred during HRT. HRT, in our study therefore did not result in mimicking breast cancer typical findings in dynamic MRM. Carcinomas in situ (CIS), on the contrary, might present with similar findings to hormonal effects in MRM and therefore may complicate differential diagnosis. CIS very often have unilateral, focal formed, local limited, and multifocal signal enhancements. In cases of bilateral CIS, which is reported to be up to 40% in lobular CIS [24], differentiation may be uncertain. The detection of CIS may thus be obscured, especially with low-grade CIS, because they may present with patchy continuous enhancement just like parenchymal CM enhancement under HRT in the present study [25]. Stopping HRT prior to a second MRM is thus advisable.

The limitations of the current study are as follows. The study design was retrospective. The size of the four patient groups with and without hormonal medication was therefore different. However, the particular findings of contrast enhancement were proven statistically significant. Additionally, the type of hormone medication is rather heterogeneous (e.g. combination therapy versus mono therapy or oral application versus transdermal application). Reichenbach and colleagues found an elevated parenchyma/fat tissue ratio in women who were administered combined hormonal therapy [21]. Growing volumes of breast parenchyma might be an explanation for the increase of CM enhancement because breast parenchyma has a higher microvessel density than fatty tissue, contributing to extravascular CM passage into the interstitial space. CM enhancement in malignant tumors may also be related to an increased fractional volume of the extravascular extracellular space [26].

Studies of breast cell proliferative activity showed that most proliferation, as well as the highest breast density on mammograms, occurs in the luteal phase of the menstrual cycle, which is partially characterized by high serum levels of progesterone and estrogens [27]. Breast density is also increased in HRT [28] and elevated proliferative activity of breast parenchyma probably results in increased contrast enhancement during dynamic contrast-enhanced MRI.

Several reports deal with the menstrual cycle and its effects on relaxation times, breast tissue composition, and CM enhancement in MRM [15‐19]. Graham and colleagues [23] and Fowler and colleagues [29] found increased water content and rising fibroglandular tissue volume of the breast during the end of the menstrual cycle and the menses. Rieber and colleagues reported an increase of CM enhancement during the second half of the menstrual cycle [17]. In this secretory cycle phase, the levels of estrogene and progesterone are physiologically elevated. This situation could thus be interpreted as an endogenous hormonal stimulation, and it seems to have similar effects on dynamic MRM compared with exogenous hormone administration. On the contrary, the higher CM enhancement during HRT may obscure overlaying pathologies. However, two cases with histologically proven cancers demonstrated that the detection of an invasive cancer or a ductal carcinoma in situ was not obscured inevitably in these cases.

Kuhl and colleagues found that 26 of 60 enhancing foci demonstrated enhancement velocity beyond the malignancy threshold of more than 80% increase of the signal intensity after the first minute after contrast media administration [30]. But these cases did not show any washout effect, which in our opinion is the strongest indicator for malignancy, and a plateau phenomenon was present in only two out of 60 lesions. Taking this into account, the study results are comparable and similar. However, the few differences between our findings and other reports may be due to different imaging protocols because MRM still is not standardized.

Hormonal effects are well known in X-ray mammography as well and do not similarly affect all women. Breast density on mammograms during HRT increases in only 20–35% of postmenopausal women [10, 11]. In contrast to MRM, increased breast density in patients under HRT may even cause a decrease of sensitivity of breast cancer detection in screening mammography [12, 13].

The statistically nonsignificant differences between the first and the second examination in group B reflect the equivalence of the two MRM examinations during HRT. The differences also underline the reproducibility of enhancement patterns during HRT.

The statistically highly significant differences in group C indicate that alterations of CM enhancement patterns caused by HRT are reversible after stopping hormonal medication, comparable with the cyclic reversible changes in MRM during the secretory phase of the menstrual cycle [23].

Two patients showed higher curve types (curve type III) during HRT than during treatment with selected estrogene receptor modulators. These findings again underline that hormone effects are reversible under selected estrogene receptor modulator therapy as well, but the patient groups are too small for statistically proven comparison of hormone withdrawal and following treatment using selected estrogene receptor modulators. Bogin and Degani, in contrast, recently reported higher levels of vascular endothelial growth factor after treatment with tamoxifen compared with HRT resulting in higher tumor vascular permeability, leading to increased CM enhancement of MCF7 breast cancers. Tumor vessels seem to behave in a different manner to vessels of normal breast tissue, which may be due to the pathological high leakage of tumor vessels [31].

The majority of postmenopausal patients receiving HRT resulted in characteristic changes with bilateral, symmetrical, and continuous CM enhancement in dynamic MRM. Cancer-typical findings like a plateau phenomenon or a washout effect could not be observed, and in two examples even breast cancer diagnosis was not affected. Repeated examinations under HRT resulted in reproducible hormone effects in MRM, which were reversible after termination of HRT. In cases of uncertain suspicious findings, which may be rarely present during HRT, MRM could thus be repeated sufficiently after hormone withdrawal.