| Part Ⅰ Comparative Study of Three-Dimensional Arterial Spin Labeling and Dynamic Susceptibility Contrast Perfusion Imaging for Quantitative Perfusion Measurements of Brain TumorsPurpose:. We comparatively analyzed the difference between three-dimensional arterial spin labeling (3D-ASL) and the conventional dynamic susceptibility contrast (DSC) perfusion imaging in the setting of assessing brain tumor perfusion.Materials and Methods:We retrospectively evaluated3D-ASL perfusion MRI and DSC-MRI in patients (n=28, M/F:13/15; mean age:49.3years old) with histopathologically proved brain tumors. All patients underwent standard MRI,3D-ASL and DSC-MRI scanings on GE DISCOVERY MR750system. Maximal relative tumor perfusion was obtained by use of the region of interest (ROI) method. A quantitative analysis was performed with normalised tumour perfusion values (ASL nTBF, DSC nTBF, DSC nTBV) by dividing tumour rCBF and rCBV into the values obtained from its mirror region (M), contralateral normal white matter (WM) and contralateral normal gray matter (GM). Paired t tests was applied to compare ASL nTBF measurements with DSC nTBF and DSC nTBV. Linear regression and Pearson’s correlation were applied to evaluate the correlations among quantitative results.Results:①The absence of significant differences were noted between ASL nTBF and DSC nTBF when normalized to M and GM (P=0.106and0.093, respectively). When normalized to WM, ASL nTBF and DSC nTBF had significant difference with P value of0.②There were significantly differences between ASL nTBF and DSC nTBV (P=0.006and0normalized to M and WM, respectively), and there was no significant difference (P=0.616) when normalized to GM though.③When normalized to M, GM and WM, there was a highly significant correlation between ASL nTBF and DSC nTBF (R=0.807,0.737and0.729respectively).④When normalized to M, there was a highly significant correlation (R=0.780) and when normalized to GM and WM, there was moderate correlation (R=0.621and0.616,respectively) between ASL nTBF and DSC nTBV.Conclusion:3D-ASL was proved to possess a potential to be a non-invasive alternate to DSC-MRI in the evaluation of perfusion in brain tumours, especially in patients with renal failure and patients required followed up, since no contrast injection was required. Part Ⅱ The application of stretched exponential model of diffusion weighted imaging in grading gliomas and the correlation between its quantitative parameters and Ki-67labeling indexPurpose:To assess the the application of stretched exponential model of diffusion weighted imaging in grading gliomas and the correlation between its quantitative parameters and Ki-67labeling index (LI), and then to determine the noninvasive method for evaluating the proliferative potential of gliomas.Materials and Methods:Seven-two patients (M/F:43/29, mean age:41.1years old, age range:6-64years old) with histopathologically proved primary cerebral gliomas (WHO grades Ⅱ:N=34, WHO grades III:N=15, WHO grade IV=23) underwent conventional MRI, traditional DWI and multi-b value DWI before surgery and chemoradiotherapy. ADC maps were created using b-values of0and1000s/mm2and DDC and a maps were created by applying the stretched-exponential model using b-values of0,1000,2000, and3000s/mm2.ADC, DDC of solid regions of tumors and their mirror regions, a values of the whole tumor and its mirror region in different groups were measured. In pathology, Ki-67is one of the nuclear markers used to demonstrate the proliferative phase of the cell cycle.It indicates the proliferative activity and is important in evaluating the the tumor grade, the possibility of recurrence and malignancy. One-way ANOVA, t-test, Pearson’s correlation, and Receiver operating characteristic (ROC) curve were used for statistical analysis.Results:①There were statistically significant differences of DDC、aand ADC values between tumors and their mirror regions, and corresponding values were (1.239±0.647)×10-3,(0.810±0.191)×10-3, p<0.05;0.689±0.060,0.729±0.057, p<0.05;(1.135±0.436)×10-3,(0.809±0.137)×10-3, p<0.05.②There was statistically significant difference of DDC value between any two groups of WHO grade Ⅱ,Ⅲ and Ⅳ gliomas, and correponding DDC values were (1.693±0.630)×10-3,(1.081±0.324)×10-3,(0.670±0.123)×10-3in ascending order.③There was statistically significant difference of a value between WHO grade II and Ⅲ or IV gliomas (0.713±0.060,0.665±0.044, p<0.05;0.713±0.060,0.669±0.056, p<0.05, respectively), but there was no significant difference between WHO grade Ⅲ and IV grade gliomas. There was statistically significant difference of a value between low grade (WHO grade Ⅰ and Ⅱ) and high grade (WHO grade Ⅲ and IV) gliomas (0.713±0.060,0.667±0.051, p<0.05).④There was statistically significant difference of ADC value between WHO grade Ⅱ and Ⅲ or Ⅳ gliomas ((1.433±0.433)×10-3,(1.000±0.242)×10-3, p<0.05;(1.433±0.433)×10-3,(0.784±0.126)×10-3, p<0.05,respectively) but there was absent of significant difference between WHO grade Ⅲ and Ⅳ grade gliomas. There was statistically significant difference of ADC value between low-grade and high-grade gliomas ((1.433±0.433)×10-3,(0.869±0.207)×10-3, p<0.05).⑤A strongly positive correlation was demonstrated between DDC and ADC values in the solid part of tumors (R=0.872; P<0.05).⑥AUC (area under curve) of DDC value in differentiating WHO grade Ⅱ and Ⅲ WHO grade Ⅲ and Ⅳ were0.836and0.968, respectively. Optimal thresholds of DDC value for differentiating WHO grade Ⅱ and Ⅲ WHO grade Ⅲ and Ⅳ were1.065×10-3mm2/s and0.811×10-3mm2/s, respectively, and the corresponding diagnostic sensibility and specificity were94.1%,73.7%and93.3%,87.0%.⑦AUC of DDC、α and ADC values for differentiating high-grade from low-grade gliomas were0.918、0.743、0.918, respectively. Optimal thresholds of DDC、α and ADC values for differentiating high-grade from low-grade gliomas were1.065×10-3mm2/s、0.698and1.022×10-3mm2/s, and the corresponding diagnostic sensibility were94.1%、64.7%and85.3%, specificity were89.59、81.6%'194.7%.⑧A significantly positive correlation was found between Ki-67LI and increasing grade (R=0.517, P<0.05) ⑨A statistically negative correlation was noted between DDC and Ki-67LI values (R=-0.305, P<0.05).⑩A negative correlation was seen between aand Ki-67LI values, but the correlation was no statistically significant(R=0.636, P<0.05).Conclusion:The stretched exponential model of diffusion weighted imaging was more accurate in grading gliomas than the traditional mono-exponential model preoperatively and its quantiative parameters DDC and a values were negatively correlated with PCNA and ki67LI which reflects tumor proliferation, so it can be a a new imaging marker in noninvasively differentiating the grade of gliomas and evaluating the effectiveness of therapy response. Part Ⅲ The preliminary application of diffusion kurtosis imaging in grading gliomas and its quantitative parameters correlation with Ki67Purpose:To assess the preliminary application of diffusion kurtosis imaging in grading gliomas and the correlation between its quantitative parameters and Ki67.Materials and Methods:Fifty-two patients (M/F:32/20, age range:7～62years old; mean age:41years old) with histopathologically proved primary cerebral gliomas. The conventional diffusion parameters (fractional anisotropy(FA), mean diffusivity (MD), axial diffusivity (Da), radial diffusivity(Dr)) and kurtosis parameters (mean kurtosis(MK), axial kurtosis(Ka), radial kurtosis(Kr)) were compared in the solid parts of22high-grade gliomas and30low-grade gliomas with Mann-Whitney test. P<0.05is considered the significance level. The conventional diffusion parameters and kurtosis parameters were also normalized to the corresponding values in contralateral normal-appearing white matter (NAWM),contralateral posterior limb of the internal capsule(PLIC) and contralateral thalamic normal-appearing gray matter (NAGM) after age correlation and were compared among high-grade and low-grade gliomas. We also analysed the correlation between Ki67which indicates the proliferative activity of tumor and FA, MD, Da, Dr, MK, Ka and Kr.Results:①Before normalization, MD, Da and Dr were significantly lower in high-grade gliomas than in low-grade gliomas (P=0.028, P=0.015, and P=0.043, respectively), MK, Ka and Kr were significantly higher in high-grade gliomas than in low-grade gliomas (P=0.001, P=0.002and P=0.002, respectively). FA did not significantly differ between glioma grades.②When normalized to the values in the contralateral NAWM, MK, Ka and Kr were were significantly different between high-grade and low-grade gliomas (MK, P=0.001; Ka, P=0.001and Kr, P=0.003). When normalized to the values in the contralateral PLIC, Da, MK, Ka and Kr were significantly different between high-grade and low-grade gliomas (Da, P=0.032; MK, P=0.002; Ka, P=0.004and Kr, P=0.002). When normalized to the values in the contralateral NAGM, all values, except for axial kurtosis, were significantly different between high-grade and low-grade gliomas (MD, P=0.028; Da, P=0.017; Dr, P=0.047; MK, P=0.003; Ka, P=0.001and Kr, P=0.030.③The highest sensitivity and specificity for discriminating between high-grade and low-grade gliomas were found for MK unnormalized (86.4%and66.7%, respectively), Ka normalized to the value in the contralateral NAWM (63.6%and83.3%, respectively), MK normalized to the value in the contralateral PLIC (81.8%and66.7%, respectively) and Ka normalized to the value in the contralateral NAGM (68.2%and80.0%, respectively).④Optimal thresholds for MK unnormalized, Ka normalized to the contralateral NAWM, MK normalized to the contralateral PLIC and Ka normalzied to the contralateral NAGM that distinguished high-grade from low-grade gliomas were0.482,0.921,0.432and0.829, respectively.⑤There was a negative correlation between Ki67LI and FA, MD, Da and Dr (P>0.05). There was a strongly postive correlation between Ki67LI and MK, Ka and Kr (R=0.648, P=0.001; R=0.606, P=0.002; R=0.641, P=0.001, respectively).Conclusion:There were significant differences in kurtosis parameters between high-grade and low-grade gliomas and the higher sensitivity and specificity were seen at kurtosis parameters. So compared with conventional diffusion parameters, better separation was achieved with kurtosis parameters. Its strongly postive correlation with Ki67makes kurtosis parameters new noninvasive biomarkers in the grading of gliomas. |
PDF Full Text Request