TitleDifferentiating benign and malignant inflammatory breastlesions: Value of T2 weighted and diffusion weighted MRimages( Dissertation_全文 )

Author(s) Kanao, Shotaro

Citation 京都大学

Issue Date 2019-01-23

URL https://doi.org/10.14989/doctor.r13218

Right

Type Thesis or Dissertation

Textversion ETD

Kyoto University

Contents lists available at ScienceDirect

Magnetic Resonance Imaging

journal homepage: www.elsevier.com/locate/mri

Original contribution

Differentiating benign and malignant inflammatory breast lesions: Value ofT2 weighted and diffusion weighted MR images

Shotaro Kanaoa,b,⁎, Masako Kataokaa, Mami Iimaa, Debra Masako Ikedab, Masakazu Toic,Kaori Togashia

a Department of Diagnostic Imaging and Nuclear Medicine, Kyoto University Graduate School of Medicine, 54 Kawahara-cho Shogoin Sakyo-ku, Kyoto 606-8507, JapanbDepartment of Radiology, Stanford University School of Medicine, 875 Blake Wilbur Dr. Stanford, CA 94305, USAc Department of Breast Surgery, Kyoto University Graduate School of Medicine, 54 Kawahara-cho Shogoin Sakyo-ku, Kyoto 606-8507, Japan

A R T I C L E I N F O

Keywords:Breast cancerInflammatory breast cancerGranulomatous mastitisDiffusion weighted imagingT2 weighted imaging

A B S T R A C T

Objectives: Benign and malignant inflammatory breast lesions demonstrate similar findings on both T2 weightedimaging (T2WI) and dynamic contrast enhanced (DCE) images. We hypothesized that benign inflammatorylesions might be differentiated form malignancies using a combination of apparent diffusion coefficient (ADC)values derived from diffusion weighted images (DWI) and T2WI.Materials and methods: We retrospectively reviewed 162 patients undergoing breast MRI (T2WI, DWI and DCEimages) between 2008 and 2015 who had breast lesions with high T2WI signal intensity (High T2 SI) including14 benign inflammatory lesions, 69 benign non-inflammatory lesions, 16 malignant inflammatory lesions and 63malignant non-inflammatory lesions. On the High T2 SI and low T2WI signal intensity (Low T2 SI) areas in thesebreast lesions, we calculated ADC values from b values of 0 and 1000 s/mm2.Results: The mean ADC values ± standard deviation (10−3 mm2/s) of the High T2 SI areas in benign in-flammatory, benign non-inflammatory, malignant inflammatory and malignant non-inflammatory breast lesionswere 0.75 ± 0.18, 1.77 ± 0.33, 2.06 ± 0.32 and 1.88 ± 0.41, respectively. Those of the Low T2 SI areas inbenign inflammatory, benign non-inflammatory, malignant inflammatory and malignant non-inflammatory le-sions were 0.89 ± 0.15, 1.31 ± 0.28, 0.87 ± 0.20 and 0.94 ± 0.27 respectively. ADC values of High T2 SIareas of the benign inflammatory lesions were significantly lower than those of benign non-inflammatory,malignant inflammatory, and malignant non-inflammatory lesions (p < 0.001). ADC values of Low T2 SI areasin benign inflammatory lesions were not significantly different from those of malignant inflammatory (p=0.99)or malignant non-inflammatory lesions (p=0.72).Conclusion: For breast lesions with High T2 SI, segmenting the High T2 SI for ADC mapping distinguishes benignfrom malignant inflammatory conditions. Using ADC mapping of the Low T2 SI areas will not result in thisdistinction.

1. Introduction

There are benign and malignant inflammatory lesions of the breast.One benign inflammatory breast lesion is idiopathic granulomatousmastitis (IGM), also known as granulomatous lobular mastitis or gran-ulomatous mastitis. IGM is a chronic inflammatory disease of unknownetiology [1–3], presenting with galactorrhea, inflammation, breastmass, tumorous indurations, and skin ulcerations. IGM has clinical,radiological and dynamic contrast enhanced magnetic resonance image(DCE-MRI) findings similar to breast cancer [4]. Another benign

inflammatory breast entity is infectious mastitis, usually caused byStaphylococcus aureus [5]. Inflammatory breast carcinoma (IBC), is arare, aggressive cancer in which tumor cells embolize and are found indermal lymphatics [6–8]. The term “inflammatory” comes from theclinical findings of a swollen, red, or inflamed breast with skin in-duration caused by the tumor, which simulates mastitis.

DCE-MRI is the most sensitive imaging modality for detecting breastcancer, and is a valuable tool for breast care [9]. Breast lesion mor-phology and dynamic contrast enhancement patterns from high-re-solution DCE-MRI post-contrast images are used to diagnose breast

https://doi.org/10.1016/j.mri.2018.03.012Received 30 October 2017; Received in revised form 10 March 2018; Accepted 10 March 2018

⁎ Corresponding author at: Department of Diagnostic Imaging and Nuclear Medicine, Kyoto University Graduate School of Medicine, 54 Kawahara-cho Shogoin Sakyo-ku, Kyoto 606-8507, Japan.

E-mail addresses: kanaos@kuhp.kyoto-u.ac.jp (S. Kanao), makok@kuhp.kyot-u.ac.jp (M. Kataoka), mamiiima@kuhp.kyoto-u.ac.jp (M. Iima), dikeda@stanford.edu (D.M. Ikeda),toi@kuhp.kyoto-u.ac.jp (M. Toi), ktgashi@kuhp.kyoto-u.ac.jp (K. Togashi).

Magnetic Resonance Imaging 50 (2018) 38–44

0730-725X/ © 2018 Published by Elsevier Inc.

T

cancers and differentiate them from benign lesions [10,11]. Authorsalso report the value of using non-contrast T2 weighted imaging (T2WI)in this context [12–14].

DWI is another promising MR imaging tool for the diagnosis ofbreast lesions [15]. DWI is usually evaluated combined with T2WI orshort tau inversion recovery (STIR) sequences [16,17]. Benign in-flammatory lesions usually contain abscesses or inflammatory fluidcomponents showing high signal intensity (SI) on T2WI (High T2 SI)with a low apparent diffusion coefficient (ADC) value calculated fromDWI scans [18]. While breast cancers may have High T2 SI componentsdue to necrosis, extracellular matrix, edema or cystic components[19,20], the ADC value of the necrotic/edematous/cystic component isnot predicted to be as low as an abscess or cystic fluid. Therefore, theADC value of High T2 SI areas within breast lesions may help distin-guish inflammatory breast lesions as benign or malignant, reasoningthat the ADC value of the T2 bright region would be low in benigninflammatory lesions and high in malignant inflammatory lesions.

The purpose of our study is to evaluate the diagnostic performanceof T2 weighted and diffusion weighted MR images to distinguish benigninflammatory lesions from other breast lesions, especially from malig-nant inflammatory lesions.

2. Materials and methods

2.1. Patient population

This study was approved by our institutional review board. Writteninformed consent was waived because of its retrospective observationalnature. We retrospectively searched our 2028 consecutive breast re-ports from June 2008 to December 2015. Among them, there were 652lesions with pathologically confirmed diagnosis and without pre-surgical systemic therapy, including 17 benign inflammatory, 174 be-nign non-inflammatory, 16 malignant inflammatory and 445 malignantnon-inflammatory lesions. Among the 652 lesions, we identified 162(24.8%, 162/652) lesions that met the following criteria; 1) contains anevaluable (> 1.0 cm in diameter) component of T2WI high signal

intensity (High T2 SI), i.e. equal to water signal intensity on T2WI, and2) the High T2 SI area demonstrated signal intensity equal to or lowerthan surrounding breast tissue on non-contrast T1 weighted images(T1WI) to exclude fat containing lesions and subacute hematomas.These 162 lesions comprise the study. They were an average of 3.5 cmin size, range 1.0 cm to 14 cm, median size 2.5 cm.

2.2. MRI acquisition

MRI was performed on Trio Tim 3.0 Tesla Scanner (Siemens AG,Erlangen, Germany) with 16 channel dedicated breast coil or Avanto1.5 Tesla Scanner (Siemens AG, Erlangen, Germany) with 4 channeldedicated breast coil.

The 3.0-T MRI with 16 channel coil parameters were as follows:T2WI (axial orientation; 2D-turbo spin echo with fat suppression; re-petition time/echo time [TR/TE], 5500/77ms; FOV, 330× 330mm;matrix, 448×336; thickness, 3.0 mm), T1WI (axial orientation; volu-metric interpolated breath-hold examination [VIBE]; TR/TE, 4.83/2.45ms; FOV, 330× 330mm; matrix, 448× 399; thickness, 1.5 mm),diffusion-weighted images (DWI) (axial orientation; single-shot echoplanar imaging [EPI]; TR/TE, 7000/62ms; FOV, 330×160mm: ma-trix, 166× 80; thickness, 3.0 mm; NEX, 3), T1-weighted DCE imagesscanning at pre-contrast state and at 0–1, 1–2 and 5–6min after ga-dolinium injection (axial orientation; VIBE with fat suppression; TR/TE,3.70/1.36ms; flip angle [FA], 15°; FOV, 330× 330mm; matrix,384× 346; thickness, 1.0 mm).

The 1.5-T MRI with 4 channel coil parameters were as follows: T2WI(axial orientation; 2D-turbo spin echo with fat suppression; TR/TE,5500/83ms; FOV, 330×330mm; matrix, 448×336; thickness,3.0 mm), T1WI (axial orientation; VIBE; TR/TE, 7.33/4.76 ms; FOV,330× 330mm; matrix, 448×358; thickness, 1.5 mm), DWI (axialorientation; single-shot EPI; TR/TE, 9000/78ms; FOV, 330× 165mm:matrix, 150×72; thickness, 3.0 mm; NEX, 3), T1-weighted DCE imagesscanning at pre-contrast state and at 0–1, 1–2 and 5–6min after ga-dolinium injection (axial orientation; VIBE with fat suppression; TR/TE,4.00/1.43ms; FA, 15°; FOV, 330× 330mm; matrix, 448× 336;

Table 1Patient and lesion characteristics of 162 breast lesions with and without high T2-weighted signal intensity on breast MRI scans.

Type of lesion Benign inflammatory lesions Benign non-inflammatorylesions

Malignant inflammatory lesions Malignant non-inflammatory lesions Remarks

Number of Patients 14 69 16 63Age, years ± SD 42 ± 13 44 ± 13 51 ± 11 59 ± 15Size cm ± SD 4.3 ± 2.3 2.0 ± 1.6 9.1 ± 2.2 3.4 ± 1.7Edema 71.4% (10/14) 2.9% (2/69) 100% (16/16) 52.4% (33/63)Skin Edema 71.4% (10/14) 0.0% (0/69) 100% (16/16) 23.8% (15/63) a

Type, nNME 4 18 4 24Mass 10 51 12 39

Source of diagnosis, nOperation 3 20 0 30Biopsy 10 49 16 33Cytology 1 0 0 0

Cause of T2 high signal, nMyxoid matrix 0 60 0 30Necrosis 0 0 12 23Fluid 0 9 0 10Abscess/inflammation 14 0 0 0Edema 0 0 4 0

Histological diagnosis (n) IGM (12)Infection (2)

FA (47)IDP (8)Benign PT (6)FCC (4)PASH (3)Adenolipoma (1)

IDC (16) IDC (27)MBC (26)IPC (5)MPC (2)ACC (1)Malignant PT (1)Sarcoma (1)

Abbreviations IGM: Idiopathic Granulomatous Mastitis. FA: Fibroadenoma. IDP: Intraductal Papilloma. PT: Phyllodes tumor. FCC: Fibrocystic change. PASH: Pseudoangiomatous StromalHyperplasia. IDC: Invasive Ductal. MBC: Mucinous Breast Carcinoma. IPC: Intraductal Papillary Carcinoma. MPC: Metaplastic Carcinoma. ACC: Adenoid cystic carcinoma.

a Benign and malignant inflammatory lesions showed significantly higher rate of skin edema than non-inflammatory lesions.

S. Kanao et al. Magnetic Resonance Imaging 50 (2018) 38–44

39

thickness, 1.5 mm).Gadolinium contrast material used for intravenous contrast en-

hancement included either Gadoteridol (ProHance, Eisai Inc., Tokyo,Japan) or Gadopentetate Dimeglumine (Magnevist, Bayer, Inc., Osaka,Japan) using a 0.1 moll/kg dose. Contrast was intravenously adminis-tered using a power injector at 2.0 ml/s and flushed with 20ml of salineat the same rate.

2.3. Image analysis

ADCs were calculated automatically per pixel at a scanning consoleusing b values of 0 and 1000 s/mm2, displayed as an ADC map. All theimages were transferred to the Picture Archiving and CommunicationSystem (Centricity PACS, GE Healthcare, Milwaukee, USA) and fusionsoftware (Aquarius Net, Tera Recon, Foster City, USA). A radiologist(SK) viewed T2WI scans of each breast lesion on PACS, placing aminimum 3.0× 3.0mm circle ROI at ADC map over areas of highsignal intensity on T2WI (High T2 SI) and areas of hypo-/iso-intensesignal intensity on T2WI (Low T2 SI) in each breast lesion. Fusionimages of DWI (b=0) and ADC map were helpful because the imagingcontrast of DWI (b=0) was almost the same as T2WI and there waspoint to point spatial correspondence between DWI (b=0) and ADCmap. We then calculated the minimum ADC values within the pre-viously selected ROIs of High T2 SI and Low T2 SI areas within eachbreast lesion.

2.4. Pathological analysis

Pathological specimens were reviewed by pathologists with morethan ten years of experience in breast pathology. Radiologic–pathologiccorrelations were made by a radiologist (SK) and pathologists to classifythe cause of High T2 SI components of the breast lesions into five ca-tegories; edema, cystic fluid, necrosis, abscess (including organizedabscess) and mucinous or myxoid extracellular matrix.

2.5. Statistical analysis

All statistical analyses were performed with EZR (Saitama MedicalCenter, Jichi Medical University, Saitama, Japan), which is a graphicaluser interface for R (The R Foundation for Statistical Computing,Vienna, Austria) [21]. The Kruskal Wallis test and post hoc Steel-Dwasstest was used to assess the distribution of ADCs and skin edema of eachpathological condition. Two-sided p values < 0.05 were consideredstatistically significant.

3. Results

Case summaries are described in Table 1. The diagnoses of the 162cases were benign inflammatory lesions (n=14, 8.6%), benign non-inflammatory lesions (n=69, 42.5%), malignant inflammatory lesions(n=16, 9.9%), malignant non-inflammatory lesions (n=63, 38.9%).We could successfully choose all ROIs of ADC map corresponding T2

Table 2Patient, lesion and imaging characteristics of 17 benign and 16 malignant inflammatory breast lesions, including 3 benign inflammatory lesions without T2 high signal intensity.

ID Age Size (cm) Skin edema ADC Method Final diagnosis Remarks

T2-high T2-low

Benign inflammatory lesions9 43 7.0 Yes 0.71 0.89 VAB IGM32 56 5.5 Yes 1.08 0.94 CNB IGM37 34 2.8 No 0.59 0.89 CNB IGM43 68 1.1 No 0.85 1.25 Surgery IGM53 27 6.7 Yes 0.73 0.88 CNB infection54 34 5.0 Yes 0.78 0.88 CNB IGM59 27 6.2 Yes 0.56 0.77 Cytology infection95 35 1.4 Yes 0.52 0.68 CNB IGM102 43 1.8 No 0.72 0.69 Surgery IGM117 41 7.0 Yes 0.51 1.10 VAB IGM121 40 6.2 Yes 0.77 0.82 CNB IGM138 28 2.0 Yes 0.71 0.89 Surgery IGM155 43 5.0 Yes 1.05 1.00 VAB IGM156 64 1.9 No 0.91 1.06 CNB IGM163 62 0.8 No – 1.18 VAB NBIL a

164 41 1.4 No – 0.84 CNB GLM a

165 37 1.1 No – 0.92 VAB GLM a

Malignant inflammatory lesions2 33 8.1 Yes 2.31 0.72 VAB IBC6 47 7.7 Yes 2.23 0.85 VAB IBC19 49 7.0 Yes 2.40 1.01 CNB IBC22 72 7.0 Yes 1.85 0.75 CNB IBC27 57 9.8 Yes 1.43 0.82 VAB IBC34 40 13 Yes 2.20 1.07 CNB IBC41 39 11 Yes 2.31 1.44 CNB IBC69 57 12 Yes 2.20 0.80 VAB IBC78 34 14 Yes 2.14 0.87 VAB IBC89 51 8.0 Yes 1.92 1.10 CNB IBC94 48 9.5 Yes 1.42 1.27 VAB IBC97 64 7.4 Yes 1.75 0.95 CNB IBC106 56 8.1 Yes 2.00 0.77 CNB IBC126 58 7.6 Yes 1.97 0.86 VAB IBC130 63 7.0 Yes 2.37 0.85 VAB IBC131 41 8.9 Yes 2.41 0.88 CNB IBC

Abbreviations ADC: Apparent Diffusion Coefficient. VAB: Vacuum Assisted Core Needle Biopsy. CNB: Core Needle Biopsy. IGM: Idiopathic Granulomatous Mastitis. NBIL: Non-specificbenign inflammatory lesion. IBC: Inflammatory breast cancer.

a These lesions were not included in 162 cases of evaluating ADC of High or Low T2 SI.

S. Kanao et al. Magnetic Resonance Imaging 50 (2018) 38–44

40

High or Low SI using fusion software. The details of the benign andmalignant inflammatory cases are described in Table 2. Fig. 1 shows atypical case of a benign inflammatory breast lesion. Fig. 2 shows a ty-pical malignant inflammatory breast lesion.

Average ADC values of High T2 SI areas in benign inflammatorylesions, benign non-inflammatory lesions, malignant inflammatory le-sions and malignant non-inflammatory lesions were 0.75 ± 0.18,1.77 ± 0.33, 2.06 ± 0.32 and 1.88 ± 0.41(10−3 mm2/s), respec-tively (Fig. 3A). Average ADC values of High T2 SI regions in benigninflammatory lesions were significantly lower than those of benign non-inflammatory, malignant inflammatory and malignant non-in-flammatory breast lesions (p < 0.001). The benign inflammatory le-sions containing High T2 SI areas can be confidently differentiated frombenign non-inflammatory lesions and malignancies using the cutoffvalue of ADC=1.2×10−3 mm2/s calculated from the High T2 SI area.

Of the 162 High T2 SI breast lesions, 75 lesions had no Low T2 SIareas for comparison, therefore were excluded from the evaluation ofADC in areas of Low T2 SI. This left 87 (53.7%) High T2 SI study lesionswhich also contained evaluable areas of Low T2 SI for ADC evaluation.The diagnoses of the 87 cases which had Low T2 SI areas were benigninflammatory lesions (n=14/87, 16.1%), benign non-inflammatorylesions (n=14/87, 16.1%), malignant inflammatory lesions (n=16/87, 18.4%) and malignant non-inflammatory lesions (n=43/87,49.4%). The average ADC values of the Low T2 SI areas in benign in-flammatory, benign non-inflammatory, malignant inflammatory andmalignant non-inflammatory lesions were 0.89 ± 0.15, 1.31 ± 0.28,0.87 ± 0.20 and 0.94 ± 0.27 (10−3 mm2/s), respectively (Fig. 3B).The ADC values of Low T2 SI areas of benign non-inflammatory lesionswere significantly higher than those of benign inflammatory(p=0.003), malignant inflammatory (p=0.006) and non-

Fig. 1. 40-year-old female with idiopathic granulomatous mastitis of the right breast.A. Axial early phase post contrast enhanced MRI image shows an enhancing irregular mass containing an unenhanced regions and skin thickening.B Axial T2 weighted noncontrast MRI shows high signal intensity corresponding to the unenhanced regions shown in Fig. 1A, thought to be due to abscesses. Note skin thickening andhigh signal intensity from skin edema.C Axial DWI (b= 1000) breast MRI image shows high signal intensity in the same regions that were bright on the T2 Weighted sequence in Fig. 1B.D ADCs of the regions of T2-high signal intensity and region of T2-low signal intensity were 0.78 and 0.89, respectively.

S. Kanao et al. Magnetic Resonance Imaging 50 (2018) 38–44

41

inflammatory lesions (p=0.015). However, ADC values of Low T2 SIareas in benign inflammatory lesions were not significantly differentfrom those of malignant inflammatory (p=0.99) or malignant non-inflammatory lesions (p=0.72).

Skin edema was seen in 71.4% (10/14) of benign inflammatorylesions, 0% (0/69) of benign non-inflammatory lesions, 100% (16/16)malignant inflammatory lesions and 23.8% (15/63) in malignant non-inflammatory lesions. Both benign and malignant inflammatory lesionsshowed significantly higher rate of skin edema than non-inflammatorylesions (p < 0.001).

On direct review of the pathology in correlation with the MRIfindings, the causes of High T2 SI within benign inflammatory lesionswere due to inflammatory change or septic/aseptic abscesses (n=14/14, 100%). High T2 signal intensity in 69 benign non-inflammatory

lesions were classified as due to myxoid extracellular matrix, as mightbe seen in a fibroadenoma (n=47/69, 68.1%), phyllodes tumor(n=6/69, 8.7%), pseudo-angiomatous stromal hyperplasia (n=3/69,4.3%), other benign pathology (n=5/69, 7.2%) and intracystic fluid aswas seen in papillomas (n=8/69, 11.6%). The High T2 SI areas ofmalignant inflammatory lesions were thought to be due to necrosis(n=12/16, 75.0%) and edema (n=4/16, 25.0%). Pathology showedthat malignant non-inflammatory lesions were T2 bright in areas ofmucinous components or myxoid extracellular matrix (n=30/63,47.6%), necrosis (n=23/63, 36.5%) and or fluid in intracystic carci-nomas (n=10/63, 15.9%) (Table 1). We note that among the original652 cases there were three histologically proven benign inflammatorylesions which had High T2 SI too small to evaluate by ADC (< 1.0 cm ofHigh T2 SI in two IGM lesions and one non-specific inflammatory

Fig. 2. 50-year-old female with inflammatory breast cancer of the left breast.A. Axial early phase post contrast enhanced MRI image shows an irregular mass with rim enhancement surrounding a central region of no enhancement, and skin thickening withenhancement of the medial skin.B. Axial T2 weighted noncontrast breast MRI shows high signal intensity area inside the mass in the region that was unenhanced in Fig. 1A. The high T2-weighted signal was due tocentral necrosis. Note skin thickening and high T2 signal within the skin.C. Axial DWI (b= 1000) breast MRI image shows peripheral rim like hyper-intensity and heterogeneous central hypo-intensity within the mass, and high intensity signal of the thickenedskin.D. Axial ADC map showed a peripheral low value and central high value within the cancer.

S. Kanao et al. Magnetic Resonance Imaging 50 (2018) 38–44

42

lesion) (Table 2). These three lesions were excluded from the study.

4. Discussion

Our study shows that a low ADC value (below 1.2× 10−3 mm2/s)within areas of High T2 SI differentiates benign inflammatory lesionsfrom benign non-inflammatory, malignant inflammatory and malignantnon-inflammatory breast lesions. This result suggests that the combi-nation of T2WI and DWI could be a useful noninvasive tool in diag-nosing inflammatory breast lesions as benign or malignant. Our resultswere similar to those of Wang et al. [22] showed that breast cancers hada higher ADC value than benign inflammatory lesions within the central

part of the lesion. Their patient population was comprised of rim-en-hancing masses with central cystic changes whereas our populationconsisted of benign and malignant inflammatory and non-inflammatorybreast lesions, which is much broader population.

There was no significant difference in the ADC values of Low T2 SIareas in benign inflammatory lesions compared to malignant in-flammatory (p=0.99) and malignant non-inflammatory lesions(p=0.72). Therefore, to differentiate benign inflammatory lesionsfrom malignancy, it is important to segment the high T2 signal intensityregion of the tumor for ADC mapping. Yilmaz et al. reviewed the ADCvalues of 37 IGM, 42 mastitis cases and 42 invasive breast cancersshowing no significant difference, however their ADC values werecalculated over the enhancing part of the lesions, and not over regionsof high T2 weighted signal intensity [23]. Aslan et al showed IGM le-sions showed significantly lower mean that ADC values compared tonormal contralateral breast parenchyma, but did not compare IGM tobreast cancer [24].

DWI plays an important role in the differentiation of breast cancerfrom benign lesions, the characterization of malignancy, the detectionof lesions in unexpected sites, and evaluating tumor extension[16,17,25–27]. Partridge et al showed improvement of diagnostic ac-curacy by combining DWI and DCE-MRI [28], showing that breastcancer tends to show washout on DCE MRI and a low ADC value onDW-MRI. However, DCE-MRI provides limited information in non-en-hancing areas within the breast such as abscess, cystic fluid, or necrosis.Our study shows that combined analysis of high signal T2WI findingsand ADC values of High T2 SI areas can provide additional and com-plimentary information for these unenhanced areas.

The evaluation and analysis of ADC values are variable for variousstudies using the average of ADC ROIs, lowest ADC value, multiple ADCROIs, or difference between highest and lowest ADC values [20,29,30].In our study, we proposed the method of measuring the ADC valueswithin ROIs placed on areas of high T2WI signal intensity. Based on ourdata, ADC values were different between High T2 SI and Low T2 SIareas in assessing for malignancy. Our data support the importance ofassessing ADC values within regions of high T2WI signal intensity, withlow ADC values within bright T2 signal regions suggesting benign in-flammatory lesions.

It is well known that abscesses in other organs show high signalintensity on T2WI and a low ADC value [18]. Previous papers that didnot include MRI showed that the percentage of IGM breast lesionscontaining abscesses were 30–50% [1–3]. But the studies that includeMRI [4,23] and our study showed that the percentage of IGM con-taining abscesses were approximately 80%. These results are probablybecause MRI has a high sensitivity of detecting abscess componentswithin IGMs.

Inflammatory breast cancer (IBC) is a very aggressive disease inwhich tumor emboli are identified within dermal lymphatic vessels,with clinical findings characterized by skin induration with an er-ysipeloid edge [8]. The presence of subcutaneous and pre-pectoraledema on T2WI is an important finding of IBC [7,31]. In our case series,benign and malignant inflammatory lesions showed a significantlyhigher rate of skin edema. Since the clinical presentation of IBC is si-milar to benign inflammatory lesions, the ADC of the high signal in-tensity areas on T2WI helps differentiate the two conditions.

Importance of lesions with high signal intensity on T2-weightedsequences has been evaluated by other groups [12–14,26]. Yuen et al.performed detailed analysis of breast carcinoma containing stronghyper-intensity on T2WI [13] showing that mucinous carcinomas andother invasive breast cancers may have high T2 bright signal similar tocellular fibroadenomas. There were differences in the inclusion criteriabetween our study and that of Yuen et al. We excluded breast carci-nomas with small areas of high T2 signal intensity because of the needto include a large enough region to evaluate ADC values of the T2 highintensity areas accurately. This means that apocrine carcinoma or in-vasive micropapillary carcinoma, which sometimes shows a High T2 SI

Fig. 3. Box-plots of ADC values of areas of high- or low -T2 signal intensity in benign andmalignant inflammatory and non-inflammatory breast lesions.A. Box-plot of ADC values of T2-high signal areas within inflammatory (14 benign, 16malignant) and non-inflammatory (69 benign, 63 malignant) breast lesions (totalN=162). The plot showed significantly lower ADC values of T2-high signal areas in thebenign inflammatory breast lesions compared to those in benign non-inflammatory le-sions (p < 0.001), malignant inflammatory lesions (p < 0.001), and malignant non-in-flammatory lesions (p < 0.001).B. Box-plot of ADC values of areas of T2-low signal intensity in 87 breast lesions withinflammatory (14 benign, 16 malignant) and non-inflammatory (14 benign, 43 malig-nant). 87/162 (53.7%) lesions in Fig. 3A had areas of low T2 signal intensity. Benign non-inflammatory lesions showed significantly higher ADC than those of benign inflammatorylesions (p=0.003), malignant inflammatory lesions (p=0.006) and malignant non-in-flammatory lesions (p=0.016). But benign inflammatory lesions showed no significantdifference in ADC values compared to malignant inflammatory lesions and malignantnon-inflammatory lesions (p=0.99 and p=0.72).

S. Kanao et al. Magnetic Resonance Imaging 50 (2018) 38–44

43

with a mesh-like low intensity septum could possibly show a low ADCbut were excluded in our study because of our inclusion size criteria ofat least 3 mm×3mm of high T2 weighted signal intensity for the ROI.

There are some limitations of this study. One limitation is thevariety of tumor conditions. In many cases, the area of high T2 signalintensity does not consist of cancer or inflammatory cells but is due tosecondary changes such as necrosis, fluid collection, edema or abscess.We have no cases where malignancy and infection (abscess) co-exist.So, in a possible rare case, high T2 signal intensity inside a cancer mayshow a low ADC value such as carcinoma with infection [32] or epi-dermal inclusion cyst, entities of which are not included in our study. Inaddition, carcinoma with subacute hematoma may show a low ADCvalue in regions of high T2 signal intensity; however, subacute hema-toma could be identified by correlation of clinical history and evalua-tion on non-contrast T1WI scans. Subacute hematomas were specifi-cally excluded from our study during patient selection using non-contrast T1 WI sequences.

Another limitation is that not all inflammatory lesions have anevaluable high signal intensity area on T2WI. Thus, we cannot applyour methods for lesions which do not have an evaluable region of highT2-weighted signal intensity. In our case series, 3 of 17 benign in-flammatory lesions had no evaluable regions of high T2-weighted signalintensity (Table 2). However, these three lesions were relatively small(0.8 to 1.4 cm) and had no skin edema. Thus, these small lesions wouldbe difficult to differentiate from benign and malignant non-in-flammatory lesions, but not from malignant inflammatory lesion whichis usually large and displays extensive skin edema.

5. Conclusion

In breast lesions with high signal intensity on T2 weighted se-quences, segmenting the T2 bright areas in the lesions for ADC mappingdistinguishes benign from malignant conditions. Using ADC mapping ofthe areas of low T2 weighted signal intensity will not result in thisdistinction.

Acknowledgements

We thank Drs. Tastuki Kataoka, Yoshiki Mikami and Takaki Sakuraifrom the Department of Pathology, Kyoto University Hospital for theirhelp in this project.

This research did not receive any specific grant from fundingagencies in the public, commercial, or not-for-profit sectors.

References

[1] Tse GMK, Poon CSP, Ramachandram K, Ma TKF, Pang L-M, Law BKB, et al.Granulomatous mastitis: a clinicopathological review of 26 cases. Pathology2004;36:254–7. http://dx.doi.org/10.1080/00313020410001692602.

[2] Akcan A, Akyildiz H, Deneme MA, Akgun H, Aritas Y. Granulomatous lobularmastitis: a complex diagnostic and therapeutic problem. World J Surg2006;30:1403–9. http://dx.doi.org/10.1007/s00268-005-0476-0.

[3] Hovanessian Larsen LJ, Peyvandi B, Klipfel N, Grant E, Iyengar G. Granulomatouslobular mastitis: imaging, diagnosis, and treatment. AJR Am J Roentgenol2009;193:574–81. http://dx.doi.org/10.2214/AJR.08.1528.

[4] Kocaoglu M, Somuncu I, Ors F, Bulakbasi N, Tayfun C, Ilkbahar S. Imaging findingsin idiopathic granulomatous mastitis. A review with emphasis on magnetic re-sonance imaging. J Comput Assist Tomogr 2004;28:635–41. http://dx.doi.org/10.1097/01.rct.0000131927.82761.40.

[5] Moazzez A, Kelso RL, Towfigh S, Sohn H, Berne TV, Mason RJ. Breast abscessbacteriologic features in the era of community-acquired methicillin-resistantStaphylococcus aureus epidemics. Arch Surg 2007;142:881–4. http://dx.doi.org/10.1001/archsurg.142.9.881.

[6] Hance KW, Anderson WF, Devesa SS, Young HA, Levine PH. Trends in inflammatorybreast carcinoma incidence and survival: the surveillance, epidemiology, and endresults program at the National Cancer Institute. J Natl Cancer Inst 2005;97:966–75.http://dx.doi.org/10.1093/jnci/dji172.

[7] Le-Petross HT, Cristofanilli M, Carkaci S, Krishnamurthy S, Jackson EF, Harrell RK,et al. MRI features of inflammatory breast cancer. AJR Am J Roentgenol2011;197:W769–76. http://dx.doi.org/10.2214/AJR.10.6157.

[8] van Uden DJP, van Laarhoven HWM, Westenberg AH, de Wilt JHW, Blanken-Peeters CFJM. Inflammatory breast cancer: an overview. Crit Rev Oncol Hematol2015;93:116–26. http://dx.doi.org/10.1016/j.critrevonc.2014.09.003.

[9] Mann RM, Kuhl CK, Kinkel K, Boetes C. Breast MRI: guidelines from the EuropeanSociety of Breast Imaging. Eur Radiol 2008;18:1307–18. http://dx.doi.org/10.1007/s00330-008-0863-7.

[10] Kuhl CK, Mielcareck P, Klaschik S, Leutner C, Wardelmann E, Gieseke J, et al.Dynamic breast MR imaging: are signal intensity time course data useful for dif-ferential diagnosis of enhancing lesions? Radiology 1999;211:101–10.

[11] Blume J, Debruhl N, Harms S, Heywang-ko SH, Hylton N, Kuhl CK, et al. Diagnosticarchitectural and dynamic features at breast MR imaging: multicenter study.Radiology 2006;238:42–53. http://dx.doi.org/10.1148/radiol.2381042117.

[12] Kuhl CK, Klaschik S, Mielcarek P, Gieseke J, Wardelmann E, Schild HH. Do T2-weighted pulse sequences help with the differential diagnosis of enhancing lesionsin dynamic breast MRI? J Magn Reson Imaging 1999;9:187–96.

[13] Yuen S, Uematsu T, Kasami M, Tanaka K, Kimura K, Sanuki J, et al. Breast carci-nomas with strong high-signal intensity on T2-weighted MR images: pathologicalcharacteristics and differential diagnosis. J Magn Reson Imaging 2007;25:502–10.http://dx.doi.org/10.1002/jmri.20845.

[14] Ballesio L, Savelli S, Angeletti M, Porfiri LM, D'Ambrosio I, Maggi C, et al. BreastMRI: are T2 IR sequences useful in the evaluation of breast lesions? Eur J Radiol2009;71:96–101. http://dx.doi.org/10.1016/j.ejrad.2008.03.025.

[15] Partridge SC, DeMartini WB, Kurland BF, Eby PR, White SW, Lehman CD.Quantitative diffusion-weighted imaging as an adjunct to conventional breast MRIfor improved positive predictive value. AJR Am J Roentgenol 2009;193:1716–22.http://dx.doi.org/10.2214/AJR.08.2139.

[16] Moschetta M, Telegrafo M, Rella L, Capolongo A, Stabile Ianora AA, Angelelli G. MRevaluation of breast lesions obtained by diffusion-weighted imaging with back-ground body signal suppression (DWIBS) and correlations with histological find-ings. Magn Reson Imaging 2014;32:605–9. http://dx.doi.org/10.1016/j.mri.2014.03.009.

[17] Telegrafo M, Rella L, Stabile Ianora AA, Angelelli G, Moschetta M. Unenhancedbreast MRI (STIR, T2-weighted TSE, DWIBS): an accurate and alternative strategyfor detecting and differentiating breast lesions. Magn Reson Imaging2015;33:951–5. http://dx.doi.org/10.1016/j.mri.2015.06.002.

[18] Unal O, Koparan HI, Avcu S, Kalender AM, Kisli E. The diagnostic value of diffusion-weighted magnetic resonance imaging in soft tissue abscesses. Eur J Radiol2011;77:490–4. http://dx.doi.org/10.1016/j.ejrad.2009.08.025.

[19] Uematsu T, Kasami M, Yuen S. Triple-negative breast cancer: correlation betweenMR imaging and pathologic findings. Radiology 2009;250:638–47. http://dx.doi.org/10.1148/radiol.2503081054.

[20] Onishi N, Kanao S, Kataoka M, Iima M, Sakaguchi R, Kawai M, et al. Apparentdiffusion coefficient as a potential surrogate marker for Ki-67 index in mucinousbreast carcinoma. J Magn Reson Imaging 2015;41:610–5. http://dx.doi.org/10.1002/jmri.24615.

[21] Kanda Y. Investigation of the freely available easy-to-use software “EZR” formedical statistics. Bone Marrow Transplant 2013;48:452–8. http://dx.doi.org/10.1038/bmt.2012.244.

[22] Wang L, Wang D, Fei X, Ruan M, Chai W, Xu L, et al. A rim-enhanced mass withcentral cystic changes on MR imaging: how to distinguish breast cancer from in-flammatory breast diseases? PLoS One 2014;9:e90355http://dx.doi.org/10.1371/journal.pone.0090355.

[23] Yilmaz R, Demir AA, Kaplan A, Sahin D, Ozkurt E, Dursun M, et al. Magnetic re-sonance imaging features of idiopathic granulomatous mastitis: is there any con-tribution of diffusion-weighted imaging in the differential diagnosis? Radiol Med2016;121:857–66. http://dx.doi.org/10.1007/s11547-016-0666-x.

[24] Aslan H, Pourbagher A, Colakoglu T. Idiopathic granulomatous mastitis: magneticresonance imaging findings with diffusion MRI. Acta Radiol 2016;57:796–801.http://dx.doi.org/10.1177/0284185115609804.

[25] Yabuuchi H, Matsuo Y, Sunami S, Kamitani T, Kawanami S, Setoguchi T, et al.Detection of non-palpable breast cancer in asymptomatic women by using un-enhanced diffusion-weighted and T2-weighted MR imaging: comparison withmammography and dynamic contrast-enhanced MR imaging. Eur Radiol2011;21:11–7. http://dx.doi.org/10.1007/s00330-010-1890-8.

[26] Trimboli RM, Verardi N, Cartia F, Carbonaro LA, Sardanelli F. Breast cancer de-tection using double reading of unenhanced MRI including T1-weighted, T2-weighted STIR, and diffusion-weighted imaging: a proof of concept study. AJR Am JRoentgenol 2014;203:674–81. http://dx.doi.org/10.2214/AJR.13.11816.

[27] Arponen O, Masarwah A, Sutela A, Taina M, Könönen M, Sironen R, et al.Incidentally detected enhancing lesions found in breast MRI: analysis of apparentdiffusion coefficient and T2 signal intensity significantly improves specificity. EurRadiol 2016;26:4361–70. http://dx.doi.org/10.1007/s00330-016-4326-2.

[28] Partridge SC, Rahbar H, Murthy R, Chai X, Kurland BF, DeMartini WB, et al.Improved diagnostic accuracy of breast MRI through combined apparent diffusioncoefficients and dynamic contrast-enhanced kinetics. Magn Reson Med2011;65:1759–67. http://dx.doi.org/10.1002/mrm.22762.

[29] Hirano M, Satake H, Ishigaki S, Ikeda M, Kawai H, Naganawa S. Diffusion-weightedimaging of breast masses: comparison of diagnostic performance using variousapparent diffusion coefficient parameters. AJR Am J Roentgenol 2012;198:717–22.http://dx.doi.org/10.2214/AJR.11.7093.

[30] Kato F, Kudo K, Yamashita H, Wang J, Hosoda M, Hatanaka KC, et al. Differences inmorphological features and minimum apparent diffusion coefficient values amongbreast cancer subtypes using 3-tesla MRI. Eur J Radiol 2016;85:96–102. http://dx.doi.org/10.1016/j.ejrad.2015.10.018.

[31] Uematsu T. MRI findings of inflammatory breast cancer, locally advanced breastcancer, and acute mastitis: T2-weighted images can increase the specificity of in-flammatory breast cancer. Breast Cancer 2012;19:289–94. http://dx.doi.org/10.1007/s12282-012-0346-1.

[32] Cilekar M, Erkasap S, Oner U, Akici M, Ciftci E, Dizen H, et al. An atypical cause ofrapidly progressing breast lump with abscess formation: pure squamous cell car-cinoma of the breast. J Cancer Res Ther 2015;11:1023. http://dx.doi.org/10.4103/0973-1482.147735.

S. Kanao et al. Magnetic Resonance Imaging 50 (2018) 38–44

44