Evaluation And Application Of Vessel Size Imaging Derived From MR Perfusion In Glioma

Glioma is the most common primary brain tumor. According to the classification basedon WHO central nervous system (CNS) in2007, gliomas were classified to low-gradegliomas (WHO grade Ⅰ and Ⅱ) and high-grade gliomas (WHO grade Ⅲ and Ⅳ).Different grades of gliomas have different biological behaviors, and this has become thefocus on clinical research for diagnosis. Tumor angiogenesis is strongly associated with themalignant degree of tumors and considered as an important factor to pathological gradingand prognosis. Perfusion weighted magnetic resonance imaging (PWI), a functionalimaging approach, can be used to assess microvascular distribution and perfusion. Analysisto glioma angiogenesis not only contributes to preoperative classification and clinicaltreatment programs, but also is beneficial for monitoring the efficacy of anti-gliomasdrugs. Cerebral blood volume (CBV) has been widely applied in clinic, and shows a certainadvantage in the preoperative grading of gliomas. Recent years, increasing animalexperiments revealed that vessel size index (VSI) could quantitatively describe vascularremodeling in tumors with a non-invasive approach, and surpassed the indirect indicatorsprovided by CBV or cerebral blood flow (CBF), while it is still in pre-clinical researchstage. In the present study，we measured the microvascular diameters of human gliomas viaSE-GE-EPI perfusion imaging techniques, comparing with CBV and histopathology.Furthermore, we analyzed the angiogenesis features in gliomas at different grades, and theresults of PWI-VSI detection. Moreover, we assessed the accuracy of PWI-VSI detectionfor glioma diagnosis, and identified the correlations between microvascular diameters andglioma grading.Objectives:To assess the effectiveness and accuracy of using MR-PWI to measure vessel sizeindex in gliomas, and analyze the angiogenesis features in gliomas at different stages, andfurther explore the correlations between VSI, or CBV and glioma grading, in order to improve the accuracy of MRI detection for evaluation of preoperative grading of gliomasand provide more information for monitoring the therapy of gliomas in clinic.Methods:1. Patients and samplesPatients with glioma:35patients with glioma confirmed by histopathology wereenrolled in this study, including15men and20women, and median age,48years (range,18-69years). All patients were first diagnosed and did not undergo any invasive ornon-invasive treatment. According to the classification in CNS,10patients with low-gradeglioma (including six patients with diffuse astrocytoma, one patient with obesityastrocytoma and three patients with oligodendroglioma) and25patients with high-gradeglioma (including five patients with anaplastic oligodendroglioma, one patient withanaplastic oligoastrocytoma, three patients with anaplastic astrocytoma and16patients withglioblastoma) were included.2. Equipment and parameterMRI Equipment: the routine MRI scan and SE perfusion scan were carried out byusing SIEMENS lagnetom Verio3.0T, and the SE-GE-EPI perfusions were performed inthe same patients via GE SIGNA HDX1.5T superconductive MR scanner, eight-channelcoil after24h. The maximal and average values of VSIMRIand maximal blood volumeperfusion (CBVmax) values were measured in gliomas, respectively. The VSIMRIandCBVmaxwere also measured in the contralateral normal-appearing white matter in the samepatients.3. HistopathologyThe paraffins (2-8per patient) were made from the tumor tissues of35patients withglioma, and then the slices were prepared (one paraffin provided one slice). The slices werestained with H&E and immunohistochemistry (with CD34), followed bythe microvasculardensity and morphology (including microvascular long and short diameter, and microvesselarea) in gliomas were quantitatively analyzed via Image-Pro Plus5.0.4. Data analysesPearson analysis was used to evaluate the correlations among VSIMRIvalues, CBVmaxvalues, microvascular long or short diameters, and microvessel area. Mann-Whitney U testwas used to analyze the maximal and average values of VSIMRIand rCBVmaxvalues in different grades of gliomas. A P value of less than0.05was considered to be statisticallysignificant. Receiver operating characteristic (ROC) curve was used to determine the cutoffof maximal andaverage values of VSIMRIand rCBVmaxvalues in distinguishing the patientswith low-grade glioma from those with high-grade glioma, and the sensitivity andspecificity were determined.Results:1. Immunohistochemical representation in different grades of gliomas: microvesselslabeled with CD34showed brown. The microvascular diameters of low-grade gliomas werehomogeneous, and the microvascular diameters were short (long diameters,18.93±5.11μm,and short diameters,9.62±2.68μm), larger than those of the normal brain tissues (longdiameters,8.79±5.10μm, and short diameters,5.51±2.63μm). In the low-grade gliomas, themicrovascular diameters in oligodendrogliomas (short diameters,8.06±0.51μm) weregenerally shorter than that in diffuse astrocytoma (short diameters,10.41±3.03μm),however, the MVD values in the former were larger than those in the latter. themicrovessels in high-grade gliomas were intensive and had long diameters (long diameters,43.59±13.26μm, and short diameters,33.36±11.43μm). Glioblastoma especially had morevascular malformations, branch-like blood vessels and glomeruloid vessels. The tumorareas which had the longest diameters of microvessels just had low number of microvessels.In the other words, the tumor areas with maximal MVD is inconsistent with that withmaximal microvessel. Up to the increase of the glioma grading, more heterogeneity of themicovessels were found.2. VSIMRIimaging: the colors of VSIMRIimaging in low-grade diffuse astrocytoma andoligodendrocytes were uniform and blue-green, which denotes the low VSIMRIvalues.Little difference was found in the different regions of interest in the same patient. TheVSIMRIvalues in oligodendroglioma were less than that in diffuse astrocytoma(VSIMRIaverage values,20.80±3.83μm and56.70±15.69μm, respectively). The colors of VSIMRIimaging in the high-grading gliomas were uneven and red, which shows the high VSIMRIvalues (glioma with grade Ⅲ,118.03±22.53μm, and glioblastoma,144.05±23.09μm).Especially, a necrotic area was clearly observed in the central lesion in glioblastoma, inwhich the VSIMRIvalue was zero, while the largest VSIMRIvalue frequently appeared in theedge of the necrotic lesion, and these are consistent with histopathology. 3. Correlations between VSIMRIvalues and microvascular diameters (histopathologytechnology): Pearson analysis was used to assess the correlations between VSIMRIvaluesand microvascular diameters. The results showed that VSIMRIaverage and maximal values,VSIMRIaverage values and microvascular long diameters, VSIMRIaverage values andmicrovascular short diameters, VSIMRImaximal values and microvascular long diameters,VSIMRImaximal values and microvascular short diameters, and microvascular long andshort diameters had significantly positive correlations, respectively(P <0.01, r>0.7).Moreover, within the MR-VSI and histopathology approaches, the most correlation wasobtained between the VSIMRImaximal values and microvascular short diameters (r=0.8190).4. Comparison of VSIMRIvalues and the results of histopathology in different grades ofgliomas: The microvascular diameters were significantly longer in the high-grade gliomasthan in the low-grade gliomas. Pairwise comparisons revealed that the average values ofVSIMRIbetween the gliomas with WHO grade Ⅱ and those with grade Ⅲ, Ⅱand Ⅳ, orⅢ and Ⅳ had statistically significance (P <0.05). The significance in maximal VSIMRIvalues was also found between low-grade gliomas and high-grade gliomas (P <0.05).However, the correlation between maximal VSIMRIvalues between gliomas with grade Ⅲand those with grade Ⅳ had no significance. The MVD values of gliomas with grade Ⅳwere significantly higher than those of gliomas with grade Ⅱ (P <0.01). however, therewere no significant differences between gliomas with grade Ⅱ and Ⅲ, or Ⅲ and Ⅳ.5. ROC analyses: The sensitivity and specificity of the VSIMRIaverage and maximalvalues in distinguishing the different grades of gliomas were analyzed by using ROC curves.The sensitivity and specificity of VSIMRIaverage and maximal values in distinguishing thegliomas with grade Ⅱ and grade Ⅲ or Ⅳ were100%, respectively (P <0.01), and theAUC value was1. Moreover, the sensitivity and specificity of VSIMRIaverage values indistinguishing the gliomas with grade Ⅲ and Ⅳ were68.75%and88.89%, respectively(P <0.05), and the AUC value was0.7986, suggesting that the VSIMRIvalues coulddistinguish the low-grade gliomas from the high-grade gliomas, and these results were alsoconsistent with the results of histopathology. Moreover, VSIMRIaverage values coulddistinguish the gliomas with grade Ⅲ from those with grade Ⅳ.6. Comparison VSIMRIwith CBV:13patients were underwent two kinds of perfusion scans, VSI and CBV. Positive correlation between the rCBV values and the VSIMRIvalues,andthe average VSIMRIvalues and the microvascular diameters were observed, whilebetween the rCBV values and microvascular diameters had no significant correlation.According to the results of above ROC analyses, we used cutoff values of VSIMRIand rCBV(average VSIMRI=81.02μm, VSIMRI=109μm, rCBV=3.4) to assess the13patients withglioma, and the results showed that the accuracy rates in distinguishing the low-gradegliomnas from the high-grade gliomas were100%for VSIMRI, and92%for rCBV, whichsuggesting that VSIMRIis more accuracy than rCBV for the determination of gliomagrading.Conclusions:1. Strong correlations between VSIMRIvalues and microvascular diameters were observed,which were consistent with the results of animal experiments. Among the correlations, themost correlation was found between the short diameters (average diameters of all microvesselsmeasured under light microscopy (200×)) and the average values of VSIMRI, suggesting thatusing MRI-VSI technology could reveal the average diameters of microvessels in the measuredregions of gliomas, thereby the changes of microvascular structure in gliomas were reflected.2. The VSIMRIvalues in diferent grades of gliomas were significantly different. Thesensitivity and specificity of the average values of VSIMRIin distinguishing the low-gradegliomas from the high-grade gliomas were both100%(P <0.01). Moreover, the sensitivityand specificity of average values of VSIMRIin distinguishing the gliomas with grade Ⅲfrom those with Ⅳ were64.29%and85.71%, respectively (P <0.05), which suggestedthat VSIMRIvalues can distinguish low-grade gliomas from high-grade gliomas, and thesemight contribute to preoperative grade of gliomas.3. Microvascular diameters could more reactively reflect the malignant degree thanmicrovascular density in gliomas. Thus, we speculated that the MRI-VSI technologysurpassed the conventional indicators which describe angiogenesis, such as rCBV, andprovided more information for the diagnosis of glioma grade.