European Journal of Radiology 83 (2014) 2137–2143

Contents lists available at ScienceDirect

European Journal of Radiology journal homepage: www.elsevier.com/locate/ejrad

Impact of fibroglandular tissue and background parenchymal enhancement on diffusion weighted imaging of breast lesions Chiara Iacconi a,∗,1 , Sunitha B. Thakur b,2 , David D. Dershaw c,3 , Jennifer Brooks d,4 , Charles W. Fry e , Elizabeth A. Morris c,5 a

Breast Unit, USL1 Massa-Carrara, Piazza Monzoni 2, Carrara 54033, Italy Department of Medical Physics, Memorial Sloan-Kettering Cancer Center, NY 1275 York Avenue, New York, NY 10065, USA c Department of Radiology – Breast Imaging Center, Memorial Sloan-Kettering Cancer Center, NY 1275 York Avenue, New York, NY 10065, USA d Department of Epidemiology and Biostatistics, Memorial Sloan-Kettering Cancer Center, 307 East 63rd Street, New York, NY 10065, USA e Memorial Sloan-Kettering Cancer Center, NY 1275 York Avenue, New York, NY 10065, USA b

a r t i c l e

i n f o

Article history: Received 28 March 2014 Received in revised form 7 August 2014 Accepted 11 September 2014 Keywords: Breast MRI DWI Fibroglandular breast tissue (FGT) Background parenchymal enhancement (BPE)

a b s t r a c t Purpose: To evaluate the influence of the amount of fibroglandular breast tissue (FGT) and backgroundparenchymal enhancement (BPE) on lesion detection, quantitative analysis of normal breast tissue and of breast lesions on DWI. Materials and methods: IRB approved this retrospective study on focal findings at contrast-enhanced (CE) breast MR and DWI performed during July–December 2011. Patients with cysts, previous irradiation, silicone implants and current chemotherapy were excluded. DWI with fat suppression was acquired before dynamic acquisition (b factors: 0.1000 s/mm2 ) using 1.5 and 3 T scanners. Using correlation with dynamic and T2 images, ROIs were drawn free-hand within the borders of any visible lesion and in contralateral normal breast. Fisher’s exact test to evaluate visibility and Wilcoxon-rank-sum test for comparison of ADC values were used. The amount of FGT and BPE was visually assessed by concurrent MRI. Analysis was stratified by menopausal status. Results: 25/127 (20%) lesions were excluded for technical reasons. 65/102 (64%) lesions were visible on DWI (median diameter: 1.85 cm). Mass lesions (M) were more visible (43/60 = 72%) than non-mass enhancement (NME) (22/42 = 52%) and malignant lesions were more visible (55/72 = 76%) than benign (10/30 = 33%). BPE and FGT did not influence visibility of M (p = 0.35 and p = 0.57 respectively) as well as of NME (p = 0.54 and p = 0.10). BPE and FGT did not influence visibility of malignant (p = 0.96 and p = 1.0) and benign lesions (p = 1.0 and p = 0.10). Results were confirmed adjusting for menopausal status. The ADC value of normal breast tissue was not influenced by BPE, while it was lower in predominantly fatty breasts compared to dense ones (p = 0.002). Conclusions: FGT affects the quantitative evaluation of ADC in normal breast tissue whereas BPE does not. Furthermore, both BPE and FGT do not influence visibility of benign or malignant findings, including both mass lesions and non-mass enhancement, on DWI. © 2014 Elsevier Ireland Ltd. All rights reserved.

1. Introduction

∗ Corresponding author. Tel.: +39 3287694612; fax: +39 0585655463. E-mail addresses: [email protected] (C. Iacconi), [email protected] (S.B. Thakur), [email protected] (D.D. Dershaw), [email protected] (J. Brooks), charles [email protected] (C.W. Fry), [email protected] (E.A. Morris). 1 Department of Radiology, Breast Imaging Section, Memorial Sloan-Kettering Cancer Center, 1275 York Avenue, New York, NY 10065, USA. 2 Tel.: +1 646 888 4816. 3 Tel.: +1 646 888 4505. 4 Tel.: +1 646 735 8068; fax: +1 646 735 0032. 5 Tel.: +1 646 888 4510. http://dx.doi.org/10.1016/j.ejrad.2014.09.004 0720-048X/© 2014 Elsevier Ireland Ltd. All rights reserved.

Screening mammography is recognized as the gold standard for screening of breast cancer [1] with a significant reduction in mortality rates [2,3]. However, mammography is less sensitive in women with dense breasts (e.g., pre-menopausal women), limiting its effectiveness as a screening modality in this population. Among high-risk women, ultrasonography screening [4,5] slightly improves cancer detection but the highest sensitivity is achieved with contrast-enhanced breast MR [6–8]. Results with contrast-enhanced MR are independent of breast density but may be negatively impacted by background

2138

C. Iacconi et al. / European Journal of Radiology 83 (2014) 2137–2143

parenchymal enhancement (BPE) which can impact the specificity of the technique, as well as personal allergy to contrast in a small number of women [9]. In the last 10 years, the role of breast MR without contrast medium both for lesion detection and characterization has been evaluated [10]. Diffusion weighted imaging (DWI) has the potential to characterize breast lesions [11–13] showing reduced diffusion of water molecules within malignant lesions. When motion probing gradients are applied, the random and thermal movement of water molecules is restricted in tissue with high cellular density (e.g., cancers, abscesses) and appears to be less restricted where there is a low cellular density or altered membrane cells (e.g., cysts, necrotic areas) [14]. Prior studies [15,16] have evaluated whether breast density can influence DWI with the hope that this type of MR sequences, without contrast medium, could be used to screen women with dense breasts in whom mammography is less effective. Our study was undertaken to evaluate if both the amount of fibroglandular breast tissue (FGT) and the background enhancement (BPE) influence the detection of lesions and their quantitative analysis in DWI. 2. Methods The Institutional Review Board approved with a waiver of informed consent this retrospective study. 2.1. Patients cohort All the women who had a contrast-enhanced (CE) breast MR with DWI between July and December 2011 with a reported focal finding (BIRADS 2–6) on CE MRI were included in the study (n = 96). Women ranged in age from 27 to 82 years (median: 50 years). Those with lesions greater than 0.5 cm were included. Smaller lesions could be critical for partial volume averaging with adjacent tissue. Exclusion criteria were focal findings of cysts, previous irradiation of the chest or of the breast, silicone breast implants, or patients currently receiving chemotherapy. In case of patients with positive margins, only the breast without surgery was evaluated. All lesions were assessed by tissue sampling or had at least one year of followup with breast MRI to confirm lesion status and absence of lesion in the contralateral breast (‘normal’ breast tissue). Table 1 shows the reason for breast MR with the majority (73%) of women undergoing MRI as part of the pre-surgical evaluation. 2.2. MRI protocol Sequence: MRI was performed using GE – 1.5 T scanner (Discovery MR450; GE Healthcare, Waukesha, WI) with a dedicated eight-channel phased array breast coil (InVivo Corporation, Orlando, FL) and GE – 3 T scanners (Discovery MR750; GE Healthcare, Waukesha, WI) using a dedicated 16-channel phased array receiver coil (Sentinelle Vanguard; Sentinelle Medical, Toronto, ON, Canada). Conventional T1- and T2-weighted images were acquired with and without fat suppression (slice thickness, 3 mm). Then, axial DW MR images were performed using 2D DW single-shot dual spin echo echo-planar imaging (EPI) sequences (TR: 6000 ms; TE: 56.4–120.7 ms; flip angle: 90◦ ; number of excitations: 3; acquisition matrix: 98 × 98 or 128 × 128; reconstructed matrix: 256 × 256; field of view, 28–38 cm; slice thickness: 4 or 5 mm; slice gap: 0–1 mm; fat suppression: enhanced; parallel imaging: ASSET; acquisition time: approximately 2 min for 2 b-values). Dual shim volumes were applied on each breast side to optimize the B0 homogeneity [17] and fat suppression was achieved applying frequency selective spectral spatial radio frequency pulses [18]. B = 0.1000 were considered from multi b acquisition.

DCE-MR images (slice thickness, 3 mm) were acquired using sagittal 3D T1-weighted gradient echo VIBRANT sequences before and at three points at 60-s intervals after an injection of 0.1 mmol/kg of gadopentetate dimeglumine (Magnevist; Bayer HealthCare Pharmaceuticals, Wayne, NJ). Subsequently, axial 3D T1-weighted gradient echo VIBRANT delayed CE images (slice thickness, 0.8–1 mm, FOV = 28–38) were performed [18]. This FOV is close to DWI FOV and the slice image with the location of lesion on DWI was selected to match or close to the axial contrast-enhanced MR image. DWI gradients were applied in three orthogonal directions. The DWI data were collected with nine b-values (acquisition time of 5–6 min). The present study used DW images corresponding to bvalues = 0.1000 to generate the isotropic ADC parametric map. 2.3. Image analysis GE advantage workstation (ADC 4.6) was used for DW images visualization and for generating ADC map selecting b0 and b1000 series from multi b-data. Dynamic contrast-enhanced images and T2-weighted images were available to correlate with DWI and were used to confirm the identification of the tumor and normal breast tissue and to assess the amount of FGT and BPE. Region of interests (ROI) were drawn free-hand within the borders of any visible lesion and in contralateral normal breast both in the retroareolar region and upper outer quadrant for each patient. Normal breast tissue was systematically evaluated in the breast without MR detected findings. It was defined basing on absence of enhancement in T1 dynamic sequence and absence of findings in T2 weighted images. T1 non-fat saturated MR images were also used to make sure to exclude fat tissue for ROI placements in normal breast tissue. Lesions were identified in DW images and confirmed in ADC map by a fellowship trained reader with 7 years of experience in breast imaging. Correlation with delayed axial contrast-enhanced images to match or close to DWI slice location, sagittal dynamic images, and T2 images was available and used for all patients basing on anatomical reference and morphology of the lesion. ROIs within the lesion were drawn in DW images and automatically copied in ADC map. The ROIs were drawn encompassing the lesion, carefully staying within the borders, to avoid partial volume averaging with adjacent tissue, necrosis or hematomas. In the retroareolar region an ROI one third of the total breast depth posterior to the nipple, encompassing only normal breast tissue was selected in DW images and ADC map. Similarly we identified the upper outer quadrant in DW images and ADC map and put the ROI within the visible normal breast tissue. The absence of lesions was confirmed comparing DWI images to dynamic sequence and T2 images. The amount of FGT and BPE were assessed by concurrent MRI and were visually classified by two fellowship trained readers with a minimum of 3 years of experience in breast imaging. When the two readings differed a consensus reading was determined. FGT was assessed according to BIRADS lexicon and then for statistical reason grouped into two classes: fatty breasts (