| 1. Purpose:(1) To identify which breast lesion descriptors in the BI-RADS MRI lexicon are most strongly associated with the diagnosis of breast cancer when performing breast MR imaging and to assess whether the BI-RADS category can distinguish between benign and malignant lesions.(2) To assess the performance of DWI for the differential diagnosis of breast lesions by analysis of the signal intensity and the apparent diffusion coefficient (ADC) and to investigate whether adding diffusion weighted imaging (DWI) to BI-RADS category could improve the diagnostic value of breast MRI.(3) To correlate the apparent diffusion coefficient (ADC) value of invasive ductal carcinoma (IDC) with the histopathologic grade.2. Materials and methods:(1) Study population①From May2012and October2013,89patient were diagnosed as breast lesions, in which26cases were excluded due to chemotherapy or radiotherapy (n=11), or lack of surgical confirmation (n=15). The remaining63patients with a mean age±standard deviation of43.78±10.20years were included in the analysis. After examination, all lesions were histologically verified by lumpectomy, mastectomy, or biopsy at the Institute of Pathology in Provincial Hospital of Shandong University.②From May2012and October2013,89patient were diagnosed as breast lesions, in which53cases were excluded due to chemotherapy or radiotherapy (n=11), lack of surgical confirmation (n=15), benign breast lesions (n=22), or other malignant lesions (n=5). The remaining36patients with a mean age±standard deviation of44.50±10.73years were included in the analysis. After examination, all lesions were histologically verified by lumpectomy, mastectomy, or biopsy at the Institute of Pathology in Provincial Hospital of Shandong University.(2)MRIAll breast MR examinations were performed on a3.0-T system (MagnetomVerio, Siemens, Germany) with an eight-channel dedicated breast coil. The patients were in the prone position, and both breasts were imaged simultaneously. Imaging was performed between day7and day14of the menstrual cycle for premenopausal women. The MRI protocols included the T1-weighted, T2-weighted, DWI and a fat-saturated3D T1-weighted FLASH sequence. DWI was obtained using single-shot echo planar imaging with b value of0s/mm2and800s/mm2.(3) Image analysis①63MRI investigations were categorised according to BI-RADS category. The results were compared with histology. All MRI studies were interpreted by two radiologists. Sensitivity, specificity and diagnostic accuracy were assessed.②The same two radiologists visually assessed the signal intensity of all breast lesions and classified them into isointense, slightly hyperintense or obviously hyperintense by mutual agreement and SI value of each lesion were measured.ADC value were obtained for normal breast tissue (n=36), benign lesions (n=22) and malignant lesions (n=41). The results were compared with histology.According to the method previously reported by Pinker et al, all lesions were adjusted BI-RADS category according to the corresponding ADC threshold.③Apparent diffusion coefficient (ADC) maps were reconstructed. The ADC value of the breast cancer was calculated and correlated with pathologic grade.(4) HistopathologyThe histopathologic grade was assessed using the Nottingham modification of the Bloom-Richardson system considering tubule formation, nuclear pleomorphism and mitotic count. Scoring (score3-5was considered grade Ⅰ,6-7was grade Ⅱ and8-9was grade Ⅲ). (5) Statistical analysisStatistical analysis was performed by SPSS version17.0. The results of measured values were recorded in the format of mean size±SD. P<0.05was considered statistically significant, and P<0.001was considered highly statistically significant.The difference of the morphological and kinetic features between malignant and benign masses was compared by corrected Pearson’s Χ2test and Fisher’s exact probability test. The sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of the BI-RADS category for the presence of malignant masses were calculated.Independent sample (student-t test) was performed to study the difference of the SI value between benign and malignant lesions and of the ADC value between less aggressive and more aggressive IDC. Fisher’s exact probability test was performed to assess the visually evaluated signal intensity on DWI between benign and malignant lesions.The difference of the ADC value was compared among normal breast tissue, benign lesions and malignant lesions using the one-way ANOVA and LSD test. ROC analysis was used to analysis the diagnostic value of ADC value in differentiating malignant lesions from benign lesions. The optimal cutoff point of ADC value and the corresponding sensitivity and specificity were determined. When BI-RADS category combined ADC, the corresponding sensitivity and specificity were calculated.The difference of the ADC value was compared with the histopathologic grade using the one-way ANOVA and LSD test. ROC analysis was used to analysis the diagnostic value of ADC value in differentiating less aggressive IDC from more aggressive IDC. The optimal cutoff point of ADC value and the corresponding sensitivity, specificity and accuracy were determined.3. Results:(1) Significance was raised with irregular shape (87.80%), spiculated (43.90%) or irregular (46.30%) margin, heterogeneous enhancement (60.98%), fast initial increase (90.20%) and Type III curve (48.78) result. Highly significant for benignity was a round mass (40.90%) or oval mass (36.30%), circumscribed margin (77.27%), homogeneous enhancement (59.09%). Slow initial increase, internal septations and type I curve refers to benignity. Rim enhancement was in93.33%a marker for a malignant lesion. Out of a total number of23tumours with type III curve,91.3%were categorized as malignant. The corresponding sensitivity, specificity, positive predictive value, negative predictive value, and the accuracy of BI-RADS category were97.56%,77.27%,88.89%,94.44%and90.47%, respectively.(2) The mean ADC value of malignant breast lesions, benign breast lesions and normal breast tissue was (0.978±0.128)×10-3mm2/s,(1.404±0.164)×10-3mm2/s,(1.689±0.344)×10-3mm2/s respectively. There was a statistically significant (F=88.31, p<0.001, one-way ANOVA). With a cut-off value of1.218×10-3mm2/s for ADC in receiver operating characteristic analysis,97.56%sensitivity、81.82%specificity and92.06%accuracy were achieved for differentiating between benign and malignant lesions. When BI-RADS category combined ADC, the corresponding sensitivity, specificity and accuracy was97.56%,86.36%,93.65%.(3) The mean ADC value of grade I was (1.116±0.113)×10-3mm2/s, of grade Ⅱ was (0.955±0.047)×10-3mm2/s, and of grade III was (0.826±0.579)×10-3mm2/s. A statistically significant (F=37.662, p<0.001, one-way ANOVA) inverse correlation was disclosed between the ADC value and tumour grading. The mean ADC value of the less aggressive group of IDC was (1.116±0.113)×10-3mm2/s, whereas the mean ADC value of the more aggressive group of IDC (0.912±0.080)×10-3mm2/s。4. Conclusions:(1) Irregular shape, spiculated or Irregular margin, heterogeneous or rim enhancement, fast initial increase, Type III or II curves in the BI-RADS MRI lexicon are mostly strongly associated with breast cancer. The BI-RADS MRI category can distinguish between benign and malignant lesions, and the corresponding sensitivity, specificity and the accuracy were97.56%、77.27%and90.47%, respectively.(2) DWI of the breast can help differentiate benign and malignant breast lesions. Furthermore, the diagnostic value of breast MRI can increase when adding diffusion weighted imaging (DWI) to BI-RADS category, and the corresponding sensitivity, specificity and the accuracy were97.56%、86.36%and93.65%, respectively.(3) Apparent diffusion coefficient (ADC) value is correlated with histopathologic grade of invasive ductal carcinoma. The lower ADC values are associated with higher histological grade. The ADC value can be considered as a promising prognostic parameter that may identify highly aggressive cancer. |
PDF Full Text Request