| BackgroundBreast cancer is the most commonly diagnosed types of cancer among women and alone is expected to account for29%of all new cancer cases among women in2012. Compared with the conventional mammography and ultrasonography, DCE MRI of the breast has clearly been demonstrated to increase the sensitivity of MRI for cancer detection, which has been supposed to be beneficial for patients. However, a meta-analysis of MR imaging in the diagnosis of breast lesions in2008shown that DCE MRI has a high sensitivity of90%, but a lower specificity of72%in patients referred for biopsy of a breast lesion. According to another meta-analysis about the accuracy of MRI in suspicious breast lesions published in2011, the Pooled sensitivity and specificity were90%and75%, respectively. Therefore, a further research is still required to improve the diagnostic specificity and accuracy for patients with breast lesions.Diffusion-weighted imaging (DWI) is one of the functional MRI techniquewhich provides unique information about the Brownian motion of water molecules and permits evaluation of tissue architecture and pathological changes on a cellular level. Water movements in tumors with high cellularity are usually restricted, which results in higher signal intensity on DWI and smaller ADC value. Therefore, the quantitative measurement of ADC values can be used to differentiate malignant from benign breast lesions and improve the diagnostic accuracy and the positive predictive value for detection of breast cancer. A meta-analysis shown the sensitivity and specificity in one homogenous subgroup of studies was0.84and0.84, respectively, when using a maximum b value of1000s/mm2. The application of DWI in breast lesions has shown overall better specificity than DCE MRI and improved accuracy for lesion diagnosis on DCE MRI. The American College of Radiology imaging network has used a standard protocol for breast DWI as part of a multicenter clinical trial. However, DWI techniques are still not standardized and uniform method of interpretation is not established, such as the choices of b values, ADC cutoff value for differentiation malignant from benign lesions. Therefore, challenges exist when performing DWI in clinical trials.As we know, the conventional ADC of water molecules in biological tissues is derived from diffusion images with a postulation that the water molecular diffusion is a random motion, which is known as Gaussian diffusion featured by a simple mono-exponential decay fitting with2b values, such as0and1000s/mm2. In fact, the measured diffusion signals in living tissues were not simply influenced by the motion of water molecule, but also influenced by the perfusion of blood microcirculation in capillary network at low b-value, and the intravoxel incoherent motion (IVIM) model was recommended to explain molecular diffusion driven by thermal motion as well as perfusion-related pseudo-diffusion. IVIM, first described by Le Bihan et al, is an interesting imaging technique for the separation of perfusion and diffusion, using multi-b-value DWI with a bi-exponential curve fitting. Such analysis from IVIM can result in perfusion-related incoherent microcirculation (D*) and microvascular volume fraction (f), separately, along with pure diffusion coefficient (D). D*describes the incoherent movements of blood in the capillary network and f describes the fraction of incoherent signal that comes from the capillary network. D represents the pure water molecular diffusion outside the microcirculation. IVIM provides an interest for the evaluation of perfusion without the use of exogenous contrast. Recently, with the development of MR hardware, I VIM has been used in several parts of body organs, such as prostate, kidney, live, salivary gland, placenta and so on. The applications of IVIM in breast lesions have been reported, and D decreased in malignant tumors when compared with normal tissues and benign lesions. However, f and D*has remained controversial in differentiating breast malignant tumors from benign lesions. Encouragingly, our preceding work and another study shown that a combination of D*and f can provide higher AUC for differentiation between malignant and benign lesions than ADC. IVIM model, specifically, can obtain the perfusion information without intravenous contrast media, which is especially impractical for patients with severe allergies who cannot receive intravenous gadolinium-based contrast media or renal dysfunction. A recent study demonstrates that there is a significant correlation between D*or f and histological microvessel density (MVD). However, comparison of IVIM biomarkers with DCE-MRI initial enhancement indicate s only a moderate correlations (r=0.42) and it is unclear how closely the perfusion-related coefficients f and D*derived from IVIM models relate to the pharmacokinetic parameters [Ktrans (the volume transferconstant), Kep (rate constant), Ve (the extracellular volume fraction of the imaged tissue) and Vp (the blood volume fraction)] rising from quantitative DCE MRI.Breast cancer is a heterogeneous lesion with highly variable biological behavior. The cytological and morphological patterns of tumor correlate with the degrees of malignancy. The correct classification of breast cancer subtypes can improve the treatment and defines the patient’s outcome. The markers to describe the aggression of breast cancers include the histological typeand tumor size, lymph node status and molecular markers, includingER, PR, HER2and Ki-67. DCE MRI and DWI have been used to evaluate the correlation between dynanmic MRI parameters or ADC values and prognostic factors, but the results are still in controversy.ObjectiveTo demonstrate the characteristics of signal decay curves of normal and pathologic breast tissues and obtain perfusion as well as diffusion information in normal breast tissues and breast lesions from IVIM imaging with bi-exponential analysis of multiple b-value diffusion-weighted imaging (DWI) and compare these parameters to apparent diffusion coefficient (ADC) obtained with mono-exponential analysis in their ability to discriminate benign lesions and malignant tumors.To prospectively demonstrate the IVIM diffusion and perfusion characteristics of locally advanced breast invasive ductal carcinomas and compare the correlations of IVIM-derived parameters with the histologic grades of breast IDCs and assess the relationships between IVIM-derived perfusion-related parameters f and D*and the pharmacokinetic parameters Ktrans,Kep, Ve and Vp from quantitative DCE MRI.To evaluate the correlations of IVIM parameters D, f and D*values of breast IDCs withpathological prognostic factors.Materials and methodsIVIM in Evaluation of Breast Lesions DWI examinations were performed in84patients with a total of111breast lesions. Benign lesions included30patients with41lesions and20patients with30simple cysts. The malignant tumors included34patients with40lesions. In addition, healthy breast tissues on the39contralateral sides were chosen as a contrast. MRI was performed by using a1.5-T MR scanner, including axial T1WI, fat-suppressed T2WI, dynamic contrast-enhanced MRI and delayed coronal MRI. Axial IVIM DW images were obtained by using EPI sequence with spectral pre-saturation inversion recovery. Diffusion sensitization in the X, Y and Z directions applied with weighting factors of b0,10,20,30,50,70,100,150,200,400,600and1000sec/mm2. PRIDE DWI Tool was used for IVIM and mono-exponential analysis. The bi-exponential model from an IVIM sequence was expressed by the following equation, as described by Le Bihan: Sb/So=(1-f)exp(-bD)+fexp[-b(D*+D)]. ROI was manually placed on each lesion at the level of maximum transverse diameter of lesions and were chosen to be as large as possible, consistent with minimal contaminations from surrounding unintended tissues. Large cystic or necrotic areas by visual inspection were excluded from ROI in order to focus upon viable tumor tissue in deriving IVIM parameters and ADC values maps. D, f, D*and ADC values were measured by2independent observers.SPSS Statistics V17.0and medcalc software were used for the statistics analysis. The intra-class correlation coefficient (ICC) was calculated and Bland Altman plots were drawn to derive the data variability for the2different observer. Test of Normal Distribution for each group and each IVIM parameter and ADC value was performed. The IVIM parameters were compared in different groups by K Independent Samples Test, respectively and Mann-Whitney U Test was used for a further comparisons between specific group pairs (Malignant vs. Benign, Malignant vs. Cyst, Malignant vs. Normal, Benign vs. Cyst, Benign vs. Normal, Cyst vs. Normal) groups(P<0.008was thought to be statistically significant because of6times of comparisons between groups). ROC curve analyses were performed to assess the utility of the measures for the detection of benign and malignant lesions and to identify thresholds to be used in a test to detect malignant tumors. Parallel test [SensitivityA+(1-Sensitivity A)×Sensitivity B)] was used to calculate the sensitivity of combining D and f values. Comparison of ROC curves for IVIM parameters and ADC value was done by Medcalc. D values were compared with ADC values for different groups by Paired-Samples Test.IVIM imaging for locally advanced IDCs and an evaluation of the correlations ofperfusion parameters derived from IVIM and quantitative DCE MRIDWI and quantitative DCE MRI were prospectively performed in patients with suspicious breast cancers. The final patients included47patients with49IDCs and18patients with20benign lesions as a contrast. The IDC lesions consist of10lesions with Grade1,15lesions with Grade2and24lesions with Grade3. The final diagnosis was confirmed on the basis of histopathology.The conventional MRI subsequences included axial T1WI, fat-suppressed T2WI and delayed coronal MRI. The IVIM imaging and post-processing were the same as above. Data for a T1map were acquired with a3D FFE sequence with a five-degree flip angle. The dynamic scans used the same parameters with a flip angle of15°. Each40-slice set was collected at35time points. The pharmacokinetics from quantitative DCE MRI were calculated by the following equation Ct(T)Ktrans.(?)t0Cp(t). e-(Ktrans/Ve).(T-t)dt+Vp. Cp(t)+....(Ktrans:the volume transferconstant, Kep:rate constant, Ve:the extracellular volume fraction, and Vp:the blood volume fraction). The quantitative DCE MRI postprocessing software was used to generate Ktrans, Kep, Ve and Vp maps. ROIs on D, f, D*maps were kept as close as possible to those on Ktrans, Kep, Ve, Vp. SPSS Statistics V17.0and MedCalc were used for the statistics analysis. The Bland-Altman plot was used to derive the data variability for2different observers. Mann-Whitney U test was used to compare DCE-MRI parameters (Ktrans, Kep,Ve, Vp) and IVIM parameters(D, f and D*) between IDCs and benign tumors and K Independent Samples Test was used to compare different grade IDCs (WHO grade1,2and3). The variables from Mann-Whitney U test were selected for ROC curve analysis to assess the diagnostic performance for differentiating IDCs from benign lesions and comparisons of ROCs were performed for IVIM parameters. Spearman correlation analyses were used among the DCE-MRI and IVIM parameters. A p-value<0.05was set as statistical significance.Correlations of IVIM parameters with prognostic factors for breast cancersDWI and conventional DCE MRI were prospectively performed in patients with suspicious breast cancers. The final patients included69patients with76IDCs and6patients with6ductal carcinomas in situ(DCIS). The IDC lesions consist of9lesions with Grade1,31lesions with Grade2and36lesions with Grade3. All the breast cancers were performed by pathologic histology and immunohistochemisty. The conventional MRI subsequences included axial T1WI, fat-suppressed T2WI, DCE MRI and delayed coronal T1WI. The IVIM imaging and postprocessing were the same as above. The method described by Elston was applied for evaluation of histological grades of IDCs, including grade1, grade2and grade3. Lymph node specimens were obtained by resection followed by histologically assessing on routinely stained sections. Imunohistochemical analysis was performed for ER, PR, HER2and Ki-67.①The ER and PR status was considered to be positive if expression was>1%.②The HER2expression was recorded as negative,1+,2+or3+. Cancers with0or+expression were classified as HER2negative and3+were HER2positive. If Cancers shown2+expression, fluorescence in situ hybridization(FISH) was performed for a further evaluation and HER2was positive when FISH demonstrated positive and vice versa.③The positive expression of Ki-67needed staining≥20%.④ Triple negative breast cancer(TNBC):All the ER,PR and HER2statuses were negative. SPSS Statistics V17.0and MedCalc were used for the statistics analysis.For evaluation of differentiation between DCIS and IDC with different grades, the Kruskal-Wallis test was used for analysis of IVIM parameters and IDCs. D, f and D*values for histological grade of IDC were also analyzed by using the Kruskal-Wallis test and Mann-Whitney U Test was used for a further comparisons between specific group. D, f and D*values were also compared with lymph node status, ER status, PR status, extent of HER2expression, Ki-67index, respectively, by Mann-Whitney U Test.ResultsIVIM in Evaluation of Breast LesionsThere was excellent interobserver agreement between the2observers in measure of mono-exponential fit parameter ADC value with ICC of0.995(95%CI:0.993-0.996). Similarly, there were excellent interobserver agreements in IVIM parameters D, with ICC of0.997(95%CI:0.996-0.998) and f with ICC of0.962(95%CI:0.948-0.972), and a relatively good ICC for D*with ICC of0.883(95%CI:0.842-0.914).The Bland-Altman Plots also suggested excellent agreement between the2observersThe relative signals of the normal and pathologic breast tissue decaying with a large-range b values had their distinctive signal decay curve profiles, which implied that they had unique IVIM parameters. DWI signal intensity decay for a cyst showed a linear relation, while normal breast tissue, benign and malignant lesions demonstrated a non-linear relation. The signal intensity decayed more quickly in malignant tumors than benign lesions, normal breast tissues and cysts within b<200s/mm2.D, f and D*values from IVIM from bi-exponential fitting and ADC value from mono-exponential fitting (b=0,1000s/mm2) of the normal breast tissues and breast lesions were different. The Kruskal-Wallis test demonstrated significant differences in IVIM parameters and ADC values among malignant tumors, benign lesions, simple cysts and normal breast tissues (X2=122.565, P=0.000for D, x2=87.585, P=0.000for f, x2=16.000, P=0.000for D*, x2=120.122, P=0.000for ADC).D values of malignant tumors were significantly smaller than those of benign lesions, simple cysts and normal tissues (Z=-7.006, P=0.000; Z=-7.122, P=0.000and Z=-7.649, P=0.000, respectively). D values were also statistically different between benign lesions and cysts or normal tissues (Z=-6.875, P=0.000and Z=-7.466, P=0.000, respectively), while there was no statistical difference of D values between cysts and normal tissues (Z=-0.097, P=0.923). Using a D value cutoff of1.06×10-3mm2/s, malignant tumors could be diagnosed with90%sensitivity and92.768%specificity.The f values of malignant tumors were significantly larger than that of benign lesions, simple cysts and normal tissues (Z=-3.457, P=0.001, Z=-7.097, P=0.000, and Z=-6.756, P=0.000, respectively), and it was also statistically different between benign lesions and cysts or normal tissues and between cysts and normal tissues (Z=-6.041, P=0.000, Z=-3.071, P=0.002and Z=-5.720, P=0.000, respectively). Using a cutoff of6.93%, malignant tumors could be diagnosed with87.5%sensitivity and53.66%specificity.D*values of malignant tumors were significantly smaller than that of normal tissues (Z=-4.746, P=0.000). However, there were no differences between any of the other groups. Similar trends were noted with ADC value calculated from mono-exponential fitting. ADC values of malignant tumors were significantly smaller than those of benign lesions, simple cysts and normal tissues (Z=-6.893, P=0.000, Z=-7.123, P=0.000, and Z=-7.650, P=0.000, respectively). ADC values were also statistically different between benign lesions and cysts or normal tissues (Z=-6.450, P=0.000and Z=-7.437, P=0.000). Using an ADC value cutoff≤1.181.05×10-3mm2/s, malignant tumors could be diagnosed with92.5%sensitivity and90.24%specificity.For the comparison between D and ADC values, D values were smaller than ADC values in malignant tumors (t=-8.481, P=0.000), benign lesions (t=-8.481, P=0.000) and normal breast tissues (t=-2.148, P=0.038), but they were almost the same as ADC values in simple cysts (t=1.410, P=0.1174). The comparisons of ROC curves of D, f, D*and ADC values for discriminating malignant tumors and from benign lesions demonstrated D and ADC values had the similar diagnostic efficacy (AUC=0.952and0.945, respectively, Z=0.9730, P=0.3306), while f and D*values had a lower diagnostic value (AUC=0.723and0.630). Combining f value, the sensitivity of D value had a sensitivitywas up to98.75%.IVIM imaging for locally advanced IDCs and an evaluation of the correlations of perfusion parameters derived from IVIM and quantitative DCE MRIBland-Altman plots demonstrate excellent agreements between2observers for the IVIM and DCE-MRI parameter measurements.D, f and D*values from IVIM bi-exponential fitting and Ktrans, Kep, Ve and Vp of benign lesions and IDCs were different. The Kruskal-Wallis test shown significant differences of IVIM parameters between benign breast lesions and IDCs (Z=-5.868, p=0.000for D, Z=-4.563, P=0.000for f, Z=-4.338, P=0.000for D*). D demonstrated the highest sensitivity (88%), specificity (90%) and accuracy (88%) in differentiating IDCs from benign lesions. D, f and D*values in different grade IDCs were also different.Further comparisons demonstrated significant differences of D, f d D*between Grade1and Grade3(Z=-2.837,P=0.005for D, Z=-3.213, P=0.001for f, Z=-2.910, P=0.004for D*), as well as between Grade2and3(Z=-3.077,P=0.002for D, Z=-2.483, P=0.013for f and Z=-2.483, P=0.013for D*), but not significant between Grade1and Grade2(Z=-1.556, P=0.120for D, Z=-1.109, P=0.267for f, Z=-0.943, P=0.367for D*).The ROC curves of D, f, D*for discriminating IDCs from benign lesions demonstrated a high diagnostic efficacy (AUC=0.946,0.852and0.835,respectively) and there was no significant difference for them in differentiating benign lesions from IDCs(Z=1.544, P=0.1226for D and f, Z=1.673, P=0.0942for D and D*, Z=0.344, P=0.7305for f and D*). Using a D, f and D*cutoff value of0.99×10-3mm2/s,7.67%,130.9mm2/s, malignant tumors could be diagnosed with a sensitivity of88%,78%,84%and specificity of90%,85%,80%, respectively.For comparison, the descriptive statistics of DCE-MRI parameters were also demonstrated. Ktrans, Kep and Vp were statistically different between IDCs and benign lesions (p<0.0001for Ktrans, Kep and Vp, respectively) and there was no difference for Ve between benign lesions and IDCs (p=0.8932). Comparisons of ROC curves of D, f, D*and Ktrans Kep Vp for IDCs and benign lesions were performed. There was significantly different between D and Kep (p=0.0257) or Vp (p=0.0099), and there was no statistical difference among other groups. Statistically, D was correlating negatively with Ktrans, Kep and Vp with Spearson’s correlation coefficientsrof-0.627(p=0.000),-0.509(p=0.000) and-0.673(p=0.000), respectively. Similar trends were found with D*and D*was also correlating negatively with Ktrans, Kep, and Vp with Spearson’s correlation coefficients r of-0.602(p=0.000),-0.443(p=0.000) and-0.700(p=0.000), respectively. On the contrary, f was negatively correlated with D (r=-0.638, p=0.000). Therefore, f was correlating positively with Ktrans, Kep and Vp with Spearson’s correlation coefficients r of0.790(p=0.000),0.536(p=0.000) and0.880(p=0.000), respectively. There were no statistical difference between Ve and D, f or D*.Correlations of IVIMparameters with prognostic factors for breast cancersThe interobserver agreements between2observers are shown with Bland-Altman plots which demonstrate excellent agreements between2observers for the IVIM parameter measurements.The mean size for all breast IDCs was30.31±1.83mm. The median D,f and D*values of breast cancers of less than2cm (n=29) were0.84×10-3mm2/s,9.14%,108.54×10-3mm2/s and of2-5cm(n=43) were0.88×10-3mm2/s,10.53%,101.09×10-3mm2/s and more than5cm (n=10) were0.85×10-3mm2/s,9.47%,93.94×10-3mm2/s, respectively. There were no statistical differences between D, f and D*values in different lesion sizes (p=0.896for D, p=0.465for f and p=0.169for D*). The D, f and D*values were also not associated with axillary lymph nodes status (p=0.338for D, p=0.718for f and p=0.328for D*). The median D values for DCIS, IDC Grade1, IDC Grade2and IDC Grade3were1.13.0.98.0.88、0.80×10-3mm2/s and D value was correlated with pathological grading of breast cancers (p=0.001). However, there was not statistically different between f and D*valueand pathological grading (p=0.721and0.669, respectively).D and D*values in ER-positive cancers were slightly lower than those of ER-negative cancers (0.86x10"3mm2/s for D value and102.46×10-3mm2/s for D*value) and f value (10.05%) of ER-positive cancers was slightly high. But D, f and D*values were not statistically different between ER-positive cancers and ER-negative cancers (p=0.246,0.844and0.829, respectively). The f value in HER2-positive cancers(8.9%) was lower than that in HER2-negative cancers(10.37%, p=0.046), but D and D*values were almost the same between HER2-negative and HER2-positive cancers(p=0.136and0.152, respectively).D value of Ki-67-positive cancers (0.82×10-3mm2/s) was significantly high than that of Ki-67-negative cancers (0.90×10-3mm2/s)(p=0.006), but f and D*values were not statistical different between Ki-67-negative Ki-67-positive cancers(p=0.059,0.154, respectively). D value was significantly negatively correlated with the percentage of Ki67expression (r=-0.270). D and D*values of triple-negative breast cancers (0.80×10-3mm2/s for D and95.07×10"3mm2/s for D*) were significantly lower than those of not-triple-negative breast cancers (0.89×10-3mm2/s for D and106.08×10-mm2for D*)(p=0.045for D,0.028for D*) and f value of triple-negative breast cancers (10.76%) was significantly higher than that of not-triple-negative breast cancers (9.6%)(p=0.039).ConclusionsDWI response curves in malignant tumors, benign lesions and normal fibroglandular tissues were found to be bi-exponential in comparison with the mono-exponential fit in simple cysts.Comparing to DWI with mono-exponential fit, IVIM provided separate quantitative measurement of D and f for cellularity and vascularity. D and f values are different among breast malignant tumors, benign lesions and normal fibroglandular tissues, but D*value was not statistically different between breast cancers and benign lesions.D value was smaller than ADC value in normal fibroglandular tissues, benign lesions and malignant tumors, but D value was almost the same as ADC value in simple cysts. D and ADC values had the same and the highest diagnostic values, and the f value and D*value. Combining with f value, D value can increase diagnostic sensitivity and may have a vital role in screening breast MRI in high-risk women. Thus, diffusion imaging of solid breast lesions analyzed with bi-exponential model with multiple low and high b values can provide microenvironment information, which could play an important role for diagnosis, prognosis, and prediction of treatment response in patients with breast lesions without the use of exogenous contrast agent.All of the three IVIM parameters D, f and D*values can be used to differentiate breast IDCs and benign lesions and they are statistically different in different grade IDCs and may be used to evaluate the degrees of breast IDCs.The perfusion-related parameters f and D*are statistically positively correlated with Vp, ktrans and Kep, derived from Gd-DTPA contrasted DCE-MRI. D value was negatively correlated with Ktrans, Kep, Vp. Thus, breast IVIM analysis can provide both diffusion and perfusion information. D, f and D*values from IVIM were correlated the parameters ofKtrans, Kep, Vp derived from Gd-DTPA-contrasted-enhanced MRI and f value can be used to evaluate the blood fraction in breast lesions.D value derived from IVIM was negatively correlated with histological grades of breast cancers and D value is lower in low grade IDCs than that in higher grade IDCs. D value is also negatively correlated with Ki67expression.The f value is higher in HER2-negative breast cancer than that in HER2-positive breast cancer.D and D*values of triple-negative breast cancers were significantly lower than those of not-triple-negative breast cancers and f value of triple-negative breast cancers was significantly higher than that of not-triple-negative breast cancers.D、f and D*values derived from IVIM are not statistically correlated with tumor size, axillary lymph nodes status, ER or PR expression. D and D*values were also not correlated with HER2positive or negative breast cancers. IVIM parameters can provide some prognostic information about breast cancers. |
PDF Full Text Request