In the assessment of supratentorial glioma grade: the combined role of multivoxel proton MR spectroscopy and diffusion tensor imaging

Aim: To detect a difference in the parameters derived from proton magnetic resonance spectroscopy ((1)H-MRS) and diffusion tensor imaging (DTI) between low-grade and high-grade gliomas, and to evaluate whether the combination of these two techniques can improve the diagnostic accuracy of conventional magnetic resonance imaging (MRI) in supratentorial glioma grading.

Materials and methods: Thirty patients with histologically proved supratentorial brain gliomas (12 low grade, 18 high grade) were prospectively evaluated with contrast material-enhanced MRI, DTI, and multivoxel (1)H-MRS (135 ms echo time). The tumour grades determined using the three methods were then compared with those obtained at histopathology. Receiver operating characteristic (ROC) analyses were performed to determine the optimum thresholds for glioma grading. Independent sample t-test, Spearman's rank correlation, and the Fisher's exact test were also carried out for statistical analysis. p<0.05 was considered statistically significant.

Results: Statistically significant differences were found between the low-grade and high-grade gliomas for the choline (Cho)/creatine (Cr), N-acetylaspartate (NAA)/Cr, NAA/Cho ratio in the tumours (p<0.01), apparent diffusion coefficient (ADC) value (p<0.01), and fractional anisotropy (FA) value (p<0.05) in the tumours. The NAA/Cr and NAA/Cho ratios and the calculated ADC value significantly correlated with the histological grading of the gliomas (p<0.01). Using a threshold value of 0.66 for tumour NAA/Cr, 0.265 for NAA/Cho, 1118.1×10(-6) mm(2)/s for the calculated ADC value, corresponding to the maximum Youden index from the ROC curve of the above-selected parameters, the resultant sensitivities, specificities, positive predictive values (PPVs), negative predictive values (NPVs), and Kappa values were all higher and the fraction of misclassified tumour was lower when compared with conventional MRI. However, only NAA/Cho and ADC calculation contributed to the significant difference (p<0.01) in the assessment of glioma grade compared to conventional MRI alone, and the grading results of statistical tests comparing those two parameters were highly consistent (kappa value=0.798).

Conclusion: Thresholds for NAA/Cho and calculated ADC values, corresponding to maximum Youden index from ROC curve analyses, helped to improve the accuracy of supratentorial glioma grading when compared with conventional MRI alone. In addition, a combination of NAA/Cho and ADC calculation were more useful together than each alone in a clinical setting to evaluate glioma grade preoperatively and provide a means for guiding treatment.