Breast Lesions Presented As Non-Mass Enhancemen On MRI:Comparison Of Imaging Features And Pathology

Part I:Application of MRI Breast Imaging Report and Data System in the Diagnosis of Breast Lesions Presented as Non-mass Enhancemen[Purpose]:To investigate the value of the 2nd edtion MRI breast imaging report and data system (BI-RADS) in the diagnosis of breast lesions presented as non-mass enhancement (NME), correlate with histopathology and differ from benign and malignant lesions.[Materials and methods]:Between October 2010 and November 2014, the suspected and risk patients with breast carcinoma received MR examination. Until March 2015,433 women (mean age 48.2 years; range 20～80 years),446 NME lesions proved by pathology or more than 1 year follow-up were analyzed on the basis of breast imaging reporting and data system(BI-RADS). Two radiologists independently evaluated the MR imgings. The interobserver agreement was measured using kappa statistc. Distribution, internal enhancement and time-signal intensity curve(TIC) type were campared between malignant and benign lesions by using x2 test. Mltivariate analysis was used to calculate odds ratios (ORs) for each of the predictor variables for malignancy. A receiver operating characteristic (ROC) curve analysis was performed to evaluate the diagnostic performance of MRI BI-RADS in breast NME lesions.[Results]:In 433 patients,446 NME lesions were analyzed.178 (39.91%) lesions were benign,268 (60.09%) lesions were malignant. The most common pathology of benign lesions was adenosis (62/178,34.83%) and of malignant lesions was ductal carcinoma in situ (130/268,48.51%). Substantial agreement was obtain for internal enhancement and BI-RADS category(κ=0.6972 and 0.6941, repectively). Perfect agreement was obtain for distribution and TIC type(κ=0.8247 and 0.8614, repectively). Significant differences (P< 0.05) were observed in focal distribution, linear distribution, segmental distribution, cluster ring enhancement and TIC type between benign and malignant NME lesions. The result of binary logistic regression analysis displayed that the most significant features predictive of malignancy were segmental distribution, clumped enhancement, cluster ring enhancement, TIC type Ⅱand Ⅲ(OR=2.088,2.193, 2.321,4.522 and 10.086, respectively). The sensitivity, specificity, negative predictive value and positive predictive value accuracy of MRI BI-RADS were 92.54%,58.43%, 84.00% and 77.26%, respectively. The areas under ROC curve was 0.755.[Conclusion]:MRI BI-RADS is useful in the differential diagnosis between the malignant and benign NME lesions. It provides an objective classification, reducing subjectivity and is convenient for communication between physicians. In evaluation of NME breast lesions, MRI BI-RADS has high specificity and relatively lower sensitivity. Segmental distribution, clumped enhancement, cluster ring enhancement, TIC type Ⅱ and Ⅲ are indenpended factors to predict malignant in breast NME lesions.Part II:Application of Diffusion-weighted Imaging in Evaluation of Breast Lesions Presented as Non-mass Enhancement[Purpose]:To study the differentiation of breast non-mass enhanced(NME) malignant lesions from benign lesions and normal breast tissue using apparent diffusion coeffcient(ADC) value of conventional single b-value monoexponential model diffusion weighted imaging(DWI) and pure diffusion coefficient(D), perfusion related diffusion coefficient (D*) and perfusion fraction(F)of the intravoxel incoherent motion(IVIM) DWI. To evaluate the diagnostic value of these parameters and to determin if additional information provided by DWI improves the diagnostic value of breast dynamic contrast-enhanced(DCE) MRI.[Materials and methods]:260 consecutive patients with suspected breast lesions were examined on a 3.0 T MR scanner including fat-suppressed T2WI, single b-value monoexponential model DWI, IVIM DWI and dynaic contrast-enhanced MRI.58 cases presented as NME lesions and 55 cases(55 lesions) including 19 benign and 36 malignant were confirmed by pathological result or follow-up. IVIM DWI was performed by using a single-shot echo-planar sequence with 12 b-values (0,10,20,30,50,70,100,150,200,400, 800,1000 s/mm2). DWI data were analyzed by monoexponential and biexponential models. ADC parametric maps was calculated by a monoexponential fit.The derived parameters of D, D*and F were calculated by biexponential IVIM analysis. Dynamic contrast-enhanced MR imagines were assessed according to the 2nd edition MRI BI-RADS categories. Parameters derived by conventional monoexponential model and IVIM DWI of malignant,benign NME lesions and normal breast tissue were compared by the independent-samples t test. Receiver operating characteristic(ROC) curve was used to analysis and compare the ability of the parameters in differential diagnosis benign and malignant lesions.[Results]:The Bland-Altmall plots suggested good interobserver agreement on the measurements between the 2 observers. D Value was smaller than ADC value in malignant lesions,benign lesions and contralateral normal breast tissues(P<0.0001). ADC and D values of malignant lesions were sigllificantly smaller than those of benign lesions and normal breast tissues(P<0.0001). ADC and D values of benign lesions were sigllificantly smaller than those of normal breast tissues (P<0.0001).There were no significant differences of D* value among malignant, benign lesions and normal breast tissues (P=0.153,0.8130,0.3503, respectively). F values of breast malignant and benign lesions were sigllificantly larger than that of normal breast tissues (P<0.0001,0.0464, respectively), but difference of F value between malignant and benign lesions s P=howed no statistical significance (P=0.716). The optimal threshold of ADC value, D value, D*value and Fvalue for the differential diagnosis of benign and malignant lesions were 1.35×10-3mm2/s,1.07×10-3mm2/s,24.57×10-3mm2/s,26.20%, respectively. The sensitivity of ADC value, D value, D*value and f value were 97.22%,97.22%,47.22%, 69.44% and 52.63% and the specificity were 57.89%,78.95%,57.89%, respectively. The combinations of ADC with BI-RADS and D with BI-RADS provided higher area under the curve (AUC) for discrimination between malignant and benign lesions (P=0.920 and 0.893, respectively). The combination of ACD with BI-RADS showed a sensitivity of 94.44%and a specificity of 84.21%. The combination of D with BI-RADS showed a sensitivity of 94.44% and a specificity of 89.47%.[Conclusion]:D Value is smaller than ADC Value in breast NME benign lesions, malignant lesions and contralateral normal breast tissues. ADC and D values of malignant lesions are smaller than those of benign lesions. The sensitivity,specificity and AUC of D value are higher than those of ADC value. D* and F value are not statistically different between malignant lesions and benign lesions.The combination of DWI and DCE-MRI has the potential to increase the specificity and dignostic performance of breast MRI, the combination of D with BI-RADS is the best.PartⅢ:Ductal Carcinoma in Situ of Breast:Dynamic Contrast-Enhanced MRI and Mammography Features and Correlation with Pathologic Findings[Purpose]:To characterize the MRI and mammography (MG) features of breast ductal carcinoma in situ(DCIS), analyze its relations with nuclear grade. To assess diagnostic sensitivity of MRI for DCIS compared with that of MG and further investigate the independent predictive factors of MRI and MG sensitivity.[Materials and methods]:Between October 2010 and January 2015, imagings of MG and MRI from 154 cases of DCIS were retrospectively reviewed. All patients were female with mean age 50.71 years and range from 28 to72 years.154 cases underwent MR examination and 111 cases underwent MG.156 lesions proved by pathology were analyzed on the basis of BI-RADS and the semi-quantitative parameters including maximum slope of increase (MSI), signal enhancement ratio at the second phase (SER2) and the peak (SERmax), positive enhancement integral (PEI), maximum slope of decrease (MSD), time to peak (TTP) were measured.156 lesions had histopathological grade diagnosis. ER, PR and HER-2 of 154 lesions were obtained by immunohistochemistry (IHC) SP method. Clinical and pathological characteristics, MRI features and its relations with nuclear grade, MG features of DCIS were analyzed by using Chi-square test and the Student t test. Multiple logistic regression analysi was used to calculate the independent predictive factors of MRI and MG sensitivity. P<0.05 considered statistical significant difference.[Results]:Of 156 DCIS lesions,152 lesions were enhanced.27 (17.76%)lesions showed mass enhancement and 125 (82.24%)lesions showed non-mass enhancement(NME). Among the NME lesions,16 (12.80%) lesions were focal distribution,24 (19.20%) were linear distribution,52 (41.60%) were segmental distribution,21 (16.80%) were regional distribution,10(8.00%) were multiple regional distribution,2(1.60%) were diffuse distribution; 21(8.00%)showed homogeneous enhancement,33(26.40%) showed heterogeneous enhancement,32(25.60%) showed cluster ring enhancement,39(31.20%) showed clumped enhancement. In 27 mass-like lesions,4(14.81%%) lesions showed oval,5(18.52%) showed round,18(66.67%) showed irregular; 6(22.22%) showed circumscribed margin,9(33.33%) showed irregular margin,12(44.44%) showed spiculated margin; 7(25.92%) showed homogenous enhancement,18(66.67%) showed heterogeneous enhancement,2(7.41%) showed rim enhancement. Among 152 lesions, type Ⅰ, type Ⅱ and type Ⅲ were demonstrated in 30, 74,48 lesions, respectively.52 lesions belonged to non-high-grade group,100 belonged to high-grade group. There were significant differences in TIC type (P=0.0375), SERmax (P=0.019) and SER2 (P=0.034), but no statistic differences for size, distribution pattern, internal enhancement, PEI, MSI, MSD and TTP between non-high-grade and high-grade group (P>0.05).111 cases with 111 lesions had MG examination.Only one abnormality was seen on MG in 74 patients. Combined two abnormalities on MG were seen in 21 patients. MG were normal in 16 patients. Calcification with or without other abnormalities were noted in 72 lesions, all of them were suspicious calcificaton. Fine pleomorphic(35 lesions) was the most common morphology, Clustered calcifications (26 lesions)was the most common distilbution. As far as the shape of mass(n=20) was concerned* the oval shaped lesion(11 lesions)was the most common, and the margin of the mass appeared as not circumscribed in 16 lesions, circumscribed in 4 lesions. Other non-calcification findings included local asymmetry(22 lesions), architecture distortion(2 lesions). There were statistic differences in ER expression, PR expression and nuclear grade (P=0.048,0.038,0.000,respectively) between calcification and non-calcification lesions, but no statistic differences for age and HER-2 expression. The sensitivity of MG was 96.32%(105/109) and MR1 was 74.31% (81/109). Logistic regression analysis showed nuclear grade was an independent factor influencing the sensitivity of MG (P=0.037, OR=3.1162). The sensitivity of MRI was not affected by clinical and pathological characteristics of DCIS.[Conclusion]:Non-mass-like enhancement, especially segmental distribution and clumped or cluster ring enhancement, are the most characteristic feature of DCIS. Different nuclear grade of DCIS is related to TIC type, SERmax and SER2. The primary presentation of DCIS on MG is as calcifications. DCIS with high-grade,ER-negative and PR-negative shows more frequently as calcifications. Sensitivity of MRI is higher than MG for the diagnosis of DCIS. Nuclear grade of DCIS influence the sensitivity of MG.