Distinction Between Recurrent Glioma And Radiation Injury Using MR Spectroscopy In Combination With Diffusion-weighted Imaging

Objective:On routine follow-up MR images, the differentiation of recurrent tumor from radiation injury in subject previously resected and irradiated glioma is problematic, although both the lesions can be associated with more specific characteristics, such as the former with corpus callosum involvement or multiple enhancing lesions and the latter with "soap bubble" or "Swiss cheese" pattern of enhancement. The aim of this study is to explore the diagnostic effectiveness of two-dimensional (2D) proton MR Spectroscopy (¹-MRS) with diffusion-weighted imaging (DWI) on the evaluation of the recurrent contrast-enhancing areas at the site of the treated gliomas.Methods:Fifty-five patients with previously resected and irradiated glioma were performed MRI examination by 3.0T MR scanner (Signa EXCITE II; GE Medical Systems, Milwaukee, WI). The MRI examination included the conventional MRI, DWI and 2D ¹H-MRS.1 .The conventional MRI examinationThe conventional MR images consisted of axial T1-weighted (500/8 ms [TR/TE]) spin-echo (SE), T2-weighted (4500/102 ms) fast SE, and fluid-attenuated inversion-recovery (FLAIR) (9000/120/2250 ms [TR/TE/TI]) images obtained by using 6-mm section thickness, 240-mm field of view (FOV) and 320×224 matrix, and then DWI examination was performed. At last, contrast-enhanced T1-weighted SE images were obtained in axial, coronal and sagittal planes after intravenous administration of gadopentetate dimeglumine (Magnevist; Schering, Germany) at a dose of 0.1-mmol per kilogram of body weight.2.DWI examinationDWI was obtained by using an axial echo-planar SE sequence (5000/65 ms [TR/TE]; one average; 6-mm section thickness; diffusion gradient encoding in 3 orthogonal directions; b = 1000 s/mm²; 240-mm FOV; 160×192 matrix) in 1.00min.Postprocessing of ADC maps was performed by using standard software on a workstation (Functool 3.1 software; Sun, GE Healthcare). In brief, regions of interest (ROIs) were drawn manually onto the obtained apparent diffusion coefficient (ADC) maps in the region corresponding to the enhancing area on contrast-enhanced T1-weighted images. The ADC value was calculated automatically by the Functool 3.1 software. For normalizing ADC levels, an ADC ratio was calculated as the quotient of the ADC values of the enhancing region and the matching structure of same-size ROIs in the contralateral hemisphere.3.Two-dimensional ^H-MRS examinationTwo-dimensional ¹H-MRS was always performed as an additional sequence within 24-48 hours after contrast-enhanced MR imaging to minimize the influence of gadolinium on MR Spectroscopy and to insure voxel placement over the area of contrast enhancement.The following parameters were used for 2D MR Spectroscopy: a point-resolved Spectroscopy sequence (PRESS), which included water and outer volume saturation pulses; 1500/144 ms [TR/TE]; 16-cm FOV; 16×16 matrix; 10-mm slice thickness; acquisition, 1 average; scanning time, 4 min 20 sec. The volume of interest (VOI) was placed on axial T1-weighted images corresponding to the contrast-enhancing area on contrast-enhanced axial T1-weighted images. Automatic prescanning was performed before each spectroscopic scan to ensure adequate water suppression. The full-width half-maximum was kept under 15 Hz and water saturation between 95% and 99%.The details of the postprocessing used for MR Spectroscopy was as following: Within the obtained VOI, separate 1cm×1cm×1cm voxels were individually placed in the area of enhancement corresponding to contrast-enhanced axial T1-weighted image. Metabolite peaks used were as follows: N-acetylaspartate (NAA) at 2.02-ppm, choline-containing compounds (Cho) at 3.22-ppm, (phospho-) creatine (Cr) at 3.01-ppm, lipid-containing compounds (Lip) in the range of 0.9-1.3 ppm, and lactate (Lac) at 1.35-ppm (an invertedβ-methyl doublet). Metabolite values were calculated automatically from the area under each metabolite peak by the Functool 3.1 software. Metabolite ratios (NAA/Cr, Cho/Cr, Lip/Cr, Lac/Cr, and Cho/NAA) were manually calculated.4.Statistical analysisStatistical analysis utilized the software SPSS for windows release 11.5 (SPSS Inc., Chicago, IL, USA). Metabolite ratios and ADC parameters (ADC value and ADC ratio) between the recurrent tumor group and radiation injury group were compared by using unpaired two-tailed Student t test. A forward stepwise discriminant analysis was undertaken to assess the power of metabolite ratios and ADC parameters to distinguish tumor recurrence and radiation injury. Then, diagnostic accuracy was compared between metabolite ratios and metabolite ratios plus ADC parameters for discriminating the two entities by using the x² test. The level of significance was set at p<0.05.Results:1.The identity of the recurrent contrast-enhancing lesionHistopathologic examination of 39 tissue specimens was performed by the board-certified neuropathologists. Thirteen were classified as radiation injury and 26 as tumor recurrence. Apart from histopathologic examination to prove the nature of contrast-enhancing lesion, 16 patients underwent additional follow-up. Six were classified as tumor recurrence and 10 as radiation injury.2.Findings of MR SpectroscopyThe Cho/Cr, Cho/NAA and Lac/Cr ratios of the contrast-enhancing lesions in the recurrent tumor group were significantly higher than those in the radiation injury group (p < 0.01, p < 0.01, and p = 0.01, respectively), while the NAA/Cr and Lip/Cr ratios of the contrast-enhancing lesions in the recurrent tumor group were significantly lower than those in the radiation injury group (p < 0.01 and p = 0.01, respectively). The mean values of the metabolite ratios in respective lesions are summarized in Table 1.Abbreviations: NAA = N-acetyl-aspartate; Cho = Choline-containing compounds; Cr = (phospho-) creatine; Lip = Iipid-containing compounds; Lac = lactate; SD = standard deviation3.Findings of DW imagingThe recurrent tumor group showed significantly lower ADC value (1.20±0.08×10^-3 mm²/s, mean±SD) compared with the radiation injury group (1.39±0.09×10^-3 mm²/s). Significance level at p< 0.01 was obtained.We subsequently obtained data normalized by creating the ratio of ADC of the enhancing lesion to ADC of matching structure in the contralateral hemisphere (i.e., ADC ratio). ADC ratio was significantly lower in the recurrent tumor group (1.42±0.10) compared with that in the radiation injury group (1.69±0.08; p < 0.01).4.Findings of MR Spectroscopy in conjunction with DW imagingTo analyze the power of correct classification, two discriminant analyses were carried out separately for 1) metabolite ratios and 2) metabolite ratios in conjunction with ADC parameters. In the first analysis, which only focused on the available metabolite ratios, NAA/Cr, Cho/Cr, Lip/Cr, Lac/Cr, and Cho/NAA were used as independent variables. In the second analysis, all variables (metabolite ratios and ADC parameters) were used as independent variables. For both analyses, the findings of follow-up or histopathology served as the group variable.In the first analysis, two significant variables were revealed. Cho/NAA emerged as the first variable to differentiate tumor recurrence from radiation injury. In order to see the classification ability of other variables, Cho/NAA was removed as a candidate. Cho/Cr emerged as the second variable. When Cho/Cr was forced out, other variables couldn't contribute significantly. When both variables (Cho/NAA and Cho/Cr) were subjected together for classification, 85.5% of total subjects were classified into correct groups (radiation injury, 91.3%; tumor recurrence, 81.3%; Table 2).Abbreviations: NAA = N-acetyl-aspartate; Cho = Choline-containing compounds; Cr = (phospho-) creatine; ADC ratios = quotient of apparent diffusion coefficient of contrast- enhancing lesion and matching structure in the contralateral hemisphereIn the second analysis, three significant variables were identified. The first variable correlated highly with Cho/NAA, the second with ADC ratio and the third with Cho/Cr. The ADC value couldn't contribute significantly to differentiate tumor recurrence from radiation injury. When all the three variables (Cho/Cr, Cho/NAA, and ADC ratio) were subjected together for classification, 96.4% of total subjects were classified into correct groups (radiation injury, 100%; tumor recurrence, 93.8%; Table 2). Compared with the first analysis, the second analysis could further increase the accuracy of discriminant analysis for the assessment of the recurrent enhancing lesion (X²= 3.96, p <0.05). On the basis of discriminant analysis, two different equations were derived (Table 2).Conclusion:our results suggest MR Spectroscopy allows the non-invasive differentiation of recurrent glioma from radiation injury in patients presenting with suspicious findings on routine follow-up MR images. Adding the ADC ratio to MR Spectroscopy, however, substantially improves the ability to differentiate the two entities. The accuracy of differential diagnosis of MR Spectroscopy in conjunction with ADC ratio is higher than that of MR Spectroscopy.