Study Of 3.0T MR Perfusion Imaging And ~1H MR Spectroscopy In Differential Diagnosis Between Recurrent Glioma And Radiation-induced Brain Injury

Objective1.To explore the value of 3.0T MR perfusion weighted imaging(PWI) in the differential diagnosis between recurrent glioma and radiation-induced brain injury by.2.To explore the value of 3.0T ~1H MR spectroscopy(MRS) in the differential diagnosis between recurrent glioma and radiation-induced brain injury.Materials and Methods1.Subjects41 cases of brain glioma were performed with MR perfusion weighted imaging (MR-PWI) and ~1H MR spectroscopy(MRS).Among them,30 cases were males and 11 females,aging from 14 to 67 years old with average age 40.5 years old.All the cases were required to satisfy the following conditions:1) All the subjects were surgical resection and were confirmed by pathology;2) All the patients experienced postoperative radiotherapy from 20 days to 6 years ago,and the radiation dose was 10 -40Gy;3) After radiotherapy,all the patients showed enhanced lesions in MRI examination.2.Groups and standardsRecurrent glioma group:1) 10 cases were confirmed by operation as recurrent glioma,2) 54 cases were diagnosed by increased lesion's extent,edema and space occupy effect by MRI follow-up.Radiation-induced brain injury:1) 2 cases were confirmed by operation as Radiation-induced brain injury,2) 15 cases were diagnosed by decreased or unchangeable of the lesion's extent,edema and space occupy effect by MRI follow-up.The duration of follow-up was longer than 3-6 month.2.Equipment and scanning sequenceAll the patients were performed with GE Signa EXCITE HD 3.0T superconducting magnetic resonance scanner with GE-ADW 4.3 workstation for data analysis..All the patients underwent conventional axial MR spin-echo T1-weighted imaging(T1WI),fast spin-echo T2-weighted imaging(T2WI),fluid-attenuated inversion recovery(FLAIR),and contrast-enhanced axial,sagittal and coronal spin-echo T1WI.(1) MR-PWI:Dynamic susceptibility contrast-enhanced MR imaging(DSC) sequences were used with GRE-EPI,TR/TE1500/30ms,5mm thickness with 1.5mm gap,50 times with totally scan time lmin20sec.(2) MRS:Two-dimensional,multi-voxel spectral imaging acquisition mode was used,with pulse sequences of point-resolved surface coil spectroscopy,(PRESS), TR/TE:1000/144ms.3.Data processing and analysis(1) Data processing:All the primitive data were sent to ADW4.3 workstation for post-processing.Relative cerebral blood volume(rCBV) map,relative cerebral blood flow(rCBF) map and(mean transit time,MTT) map were automatically generated by using GE Functool brain perfusion software for DSC.The Chmical Shift Image(CSI), spectrum image(SI),metabolic image(MI),and anatomical image pluse metabolic image(AI+ MI) was automatically generated by using GE Functool software packge.(2) Selection of ROIs:Referring to conventional MR imaging,maximal rCBV, rCBF rMTT and were obtained by identifying regions of maximal perfusion from color maps.Three different circular regions of interest were placed on the central enhancement region(central enhancement region),1-2 centimeters around enhancement region(marginal region) and normal white matter region(normal region).The maximal rCBV,rCBF and rMTT values were then normalized by dividing theses values to the values from the contralateral healthy gray matter.And thus the maximal rCBV,rCBF and rMTT ratios of the lesions were obtained.For MRS,the distribution of each metabolite were obtained on(AI+MI) images,and the Cho/Cr,NAA/Cr and NAA/Cho ration were calculated automatically..4,Statistics analysisSPSS 13.0 software package was applied to statistical analysis.All data were recorded as(mean±standard deviation).The statistically significant difference was set at P<0.05.Use R×C chi-square test to compara the datas such as the enhanced shape, enhanced width and the existing time.MR perfusion and MRS parameters obtained by DSC and MRS were compared with general linear model repeated measures.If P<0.05,the best classification threshold was found by analyzing ROC curve,and then calculating its sensitivity,specificity and accuracy.Results1.Morphological characteristics of conventional MRIThe shape,width and existing time of enhancement on contrast enhanced T1WI were different between tumor recurrence group and radiation-induced brain injury group,with statistically significant differences(P<0.05).Nodular-like enhancement was particularly prevalent in the recurrence group while map-like and ring-like enhancement existed most in the radiation-induced brain injury group(χ~2=14.508,P =0.001).The width of enhancement≧5mm was particularly prevalent in the recurrence group while width <5mm was particularly prevalent in the radiation-induced brain injury group(χ~2=9.259,P=0.002).The existing time of enhancement≧3 months was particularly prevalent in the recurrence group while <3 months was particularly prevalent in the radiation-induced brain injury group(χ~2= 29.556,P=0.000).The diagnosis of sensitivity(75.0%),specificity(82.4%),accuracy(78.5%), positive predictive value(85.7%) and negative predictive value(46.7%) in tumor recurrence was based on nodular-like enhancement.The diagnosis of sensitivity (75.0%),specificity(70.6%),accuracy(73.2%),positive predictive value(78.3%) and negative predictive value(66.7%) in tumor recurrence was based on the width of enhancement≧5mm.The diagnosis of sensitivity(95.8%),specificity(88.2%), accuracy(92.7%),positive predictive value(92.0%) and negative predictive value (93.8%) in tumor recurrence was based on existing time of enhancement≧3 months.2.MR perfusion imaging analysis(1) Comparison of central enhancement region:The mean maximal rCBV and rCBF ratios in the central enhancement regions of the recurrent glioma group and radiation-induced brain injury group were 5.279±1.588,4.876±1.436 and 2.244±0.698,2.051±0.992 respectively.Statistically significant differences(P<0.05) existed in the mean maximal rCBV and rCBF ratios between two groups,with higher perfusion parameters at the recurrent glioma group(P=0.000).(2) Comparison of marginal region:The mean maximal rCBV and rCBF ratios in the marginal regions of the recurrent glioma group and radiation-induced brain injury group were 4.091±1.087,3.676±1.138,and 1.611±0.738,1.314±0.568 respectively.Statistically significant differences(P<0.05) existed in the mean maximal rCBV and rCBF ratios between two groups,with higher perfusion parameters at the recurrent glioma group(P=0.000).(3) Analysis of ROC curve:The classification threshold for the values of the mean maximal rCBV and rCBF ratios between the recurrent glioma group and radiation-induced brain injury group in the central enhancement regions were 3.149, 2.906 respectively,with the sensitivity,specificity and accuracy of 95.8%,94.1%, 98.0%and 91.7%,88.2%,92.0%respectively.The classification threshold for the values of the mean maximal rCBV and rCBF ratios between the recurrent glioma group and radiation-induced brain injury group in the marginal regions were 2.585, 1.996 respectively,with the sensitivity,specificity and accuracy of 91.7%,88.2%, 89.1%and 91.7%,82.4%,83.2%respectively.3.MRS analysis(1) Cho/Cr ratio:The mean Cho/Cr ratio in the central enhancement region and marginal regions of the recurrent glioma group and radiation-induced brain injury group were 2.914±0.460,1.964±0.324 and 1.648±0.345,1.299±0.271 respectively.Statistically significant differences(P=0.000) existed in the mean Cho/Cr ratio between two groups,with higher Cho/Cr ratio at the recurrent glioma group.By ROC curve analysis,take the average Cho/Cr ratio of the central enhancement region as identification index of differential diagnosis between tumor recurrence and radiation-induced brain injury.When the threshold value is 2.125,the sensitivity,specificity and accuracy being 100%,94.1%and 92.8%respectively.It will be recurrent glioma if the ratio above this threshold,and radiation-induced brain injury if below the threshold.(2) NAA/Cr ratio:The mean NAA/Cr ratio in the central enhancement region and marginal regions of the recurrent glioma group and radiation-induced brain injury group were 0.856±0.141,0.937±0.158 and 1.064±0.156,1.132±0.895 respectively.Statistically significant differences(P=0.000) existed in the mean NAA/Cr ratio between two groups,with lower NAA/Cr ratio at the recurrent glioma group.By ROC curve analysis,take the average NAA/Cr ratio of the central enhancement region as identification index of differential diagnosis between tumor recurrence and radiation-induced brain injury.When the threshold value is 0.931,the sensitivity,specificity and accuracy being 70.8%,76.5%and 73.0%respectively.It will be recurrent glioma if the ratio below this threshold,and radiation-induced brain injury if above the threshold.(3) NAA/Cho ratio:The mean NAA/Cho ratio in the central enhancement region and marginal regions of the recurrent glioma group and radiation-induced brain injury group were 0.327±0.056,0.650±0.303 and 0.781±0.056,1.067±0.131 respectively.Statistically significant differences(P=0.000) existed in the mean NAA/Cho ratio between two groups,with lower NAA/Cho ratio at the recurrent glioma group.By ROC curve analysis,take the average NAA/Cho ratio of the central enhancement region as identification index of differential diagnosis between tumor recurrence and radiation-induced brain injury.When the threshold value is 0.951,the sensitivity,specificity and accuracy being 91.7%,82.4%and 79.2%respectively.It will be recurrent glioma if the ratio below this threshold,and radiation-induced brain injury if above the threshold.Conclusions1.Conventional MRI plays a certain role in the differential diagnosis between recurrent glioma and radiation-induced brain injury.In the recurrence group,most of the cases showed nodular-like enhancement of tumor,enhanced width≧5mm, existing time of enhancing≧3 months.The diagnostic value of existing time≧3 months was the highest,with its sensitivity(95.8%),specificity(88.2%),accuracy (92.7%),positive predictive value(92.0%) and negative predictive value(93.8%).2.The Dynamic susceptibility contrast-enhanced MR perfusion imaging can,in a non-invasive way,reflects the hemodynamic information of proliferation of brain tissue vascular.Both the mean ratio of rCBFmax and of rCBVmax can be used in the differential diagnosis between recurrent glioma and radiation-induced brain injury. Both the mean ratio of rCBFmax and of rCBVmax in the central enhancement region and marginal ragion are higher in recurrent glioma than that in the radiation-induced brain injury group,with statistically significant differences.3.When it takes threshold as 3.149 and 2.906 respectively in the central enhancement region,the sensitivity,specificity and accuracy of mean rCBVmax and rCBFmax in the differential diagnosis between recurrent glioma and radiation-induced brain injury are 95.8%,94.1%,98.0%and 91.7%,88.2%,92.0%respectively.When it takes threshold as 2.585 and 1.996 respectively in the marginal region,the sensitivity,specificity and accuracy of mean rCBVmax and rCBFmax in the differential diagnosis between recurrent glioma and radiation-induced brain injury are 91.7%,88.2%,89.1%and91.7%,82.4%,83.2%respectively.4.~1H-MRS is a non-invasive method to study tissue metabolites changes in vivo. 1H-MRS metabolite ratios were significantly different in the recurrence group and radiation-induced brain injury group.In the recurrence group,Cho increases significantly and Cho/Cr ratio is higher than the radiation-induced brain injury group, however,NAA decreases,both NAA/Cho and NAA/Cr ratios lower than the group of radiation-induced brain injury.5.Cho/Cr,NAA/Cr and NAA/Cho ratio of ~1H-MRS metabolites play a significant role in the differential diagnosis between recurrent glioma and radiation-induced brain injury.The higher the Cho/Cr ratio is,the more possible the tumor recurrence.The lower the NAA/Cr and NAA/Cho ratios are,the more possible the tumor recurrence.Cho/Cr ratio is of most valueable in the differential diagnosis. When the threshold is 2.125,the sensitivity,specificity and accuracy of tumor recurrence are the highest,being 100%,94.1%and 92.8%respectively.