Imaging Study Of Magnetic Resonance Spectroscopy And Texture Analysis In Glioma With IDH1Mutation

Part I:Expression of IDH1R132H mutation and erythropoietin receptor (EpoR) in glioma with diverse type and gradeAim:This study is to investigate the differential expression of erythropoietin receptor (EpoR) in glioma with IDH1R132H mutation expression or without.Methods:We collected299glioma samples obtained from patients operated between2004Jan and2005Aug in neurosurgery department of Huashan hospital affiliated to Fudan University. Tissue samples included of40pilocytic astrocytomas,55astrocytomas,24oligodendrogliomas,1oligoastrocytoma,54anaplastic astrocytomas,27anaplastic oligodendrogliomas,60glioblastomas,20ependymomas,17anaplastic ependymomas and1subependymoma in our group. After constructed in to tissue microarrays, the expression of IDH1R132H mutation and EpoR were assessed by immunohistochemistry(IHC). We used Kendall’s tau for assessing the correlation between IDH1R132H mutation and EpoR expression by SPSS.Results:IHC shows IDH1R132H mutation expressed in cytoplasm, zero expression in pilocytic astrocytomas, ependymomas and anaplastic ependymomas, with expression rate of43.4%in astrocytomas,38%in anaplastic astrocytomas,50%in oligodendrogliomas,34.6%in anaplastic oligodendrogliomas,6.78%in glioblastomas. EpoR expressed in membrane and cytoplasm, with expression rate of50%in pilocytic astrocytomas,56.6%in astrocytomas,64%in anaplastic astrocytomas,58.3%in oligodendrogliomas,61.5%in anaplastic oligodendrogliomas,59.3%in glioblastomas. Expression rate of EpoR is82.35%in glioma with IDH1R132H mutation expression and49.33%in glioma without IDH1R132H mutation expression. Kendall’s tau revealed IDH1R132H mutation is positive related to EpoR expression(r=0.385, p<0.001). In astrocytomas, oligodendro-gliomas, anaplastic astrocytomas, anaplastic oligodendrogliomas and glioblastomas without IDH1R132H mutation expression, EpoR expression is related to WHO grading(r=0.149,p<0.05).Conclusion:IDH1R132H mutation expression is wildly in astrocytomas and oligodendrogliomas, rarely in glioblastomas and not detected in pilocytic astrocytomas and ependymomas. EpoR expressed in all glioma types. High expression of EpoR can be found in glioma with IDH1R132H mutation expression. EpoR expression is related to WHO grading in glioma without IDH1R132H mutation expression. Part Ⅱ:Magnetic resonance spectroscopyin glioma with IDH1R132H mutation expression or withoutAim:This study was to investigate the magnetic resonance spectroscopic change in glioma with IDH1R132H mutation expression or without.Methods:Patients operated between2012July and2013Feb in neurosurgery department of Huashan hospital with supratentorial new-diagnosed glioma were included. NAA, Cho and Cr levels from MRS data was quantified in voxels with maximum ratio of Cho to NAA. Expression of IDH1R132H and MIB-1was assessed by IHC. We used Student’s test for evaluating the spectroscopic change and differential expression of MIB-1in glioma with IDH1R132H mutation expression or without by SPSS.Results:NAA level (r=-0.244, p<0.05), ratio of Cho to NAA(r=0.239, p<0.05) and expression of MIB-1(r=0.732, p<0.05) were related to WHO grades of glioma, but there was no statistical difference of MRS metabolic product levels and expression of MIB-1in glioma with IDH1R132H mutation expression and without.Conclusions:Mutation of IDH1R132H in glioma has little effect on NAA level, ratio of Cho to NAA and expression of MIB-1. Conventional MRS can not predict the mutation status of IDH1R132H in glioma. Part Ⅲ:Application of magnetic resonance imaging texture characteristic analysis based on support vector machine algorism to detect IDH1R132H mutation status in gliomaAim:This study was to research the possibility of using MRI texture characteristic to predict mutation status of IDH1in glioma.Methods:Patients with glioma operated between2011Jan and2013Feb in neurosurgery department of Huashan hospital were included. All cases have three sequences (T1-FLAIR, T2-FLAIR and T1with contrast).2-4regions of interest (ROIs) were selected on every MR case, consisting of32ROI with IDH1R132H mutation and18ROI without IDH1R132H mutation. Texture features (gray level co-occurrence matrix, gray level gradient co-occurrence matrix and run length matrix) were extracted in ROIs and classified with a support vector machine in a leave-one-out cross validation strategy. The performance of the method was evaluated by accuracy, sensitivity, specificity, positive predicting value, negative predicting value, ROC curve and area under curve(AUC).Results:Accuracy of classification of IDH1R132H mutation was0.67, sensitivity was0.61, specificity was0.76, positive predicting value was0.83, negative predicting value was0.52and AUC of ROC was0.69.Conclusions:The proposed method has the potential to predict mutation status of IDH1R132H in glioma.