Role of MRSI Major Metabolite Ratios in Differentiating Between Intracerebral Ring-Enhancing Neoplastic and Non-Neoplastic Lesions, High-Grade Gliomas and Metastases, and High-Grade and Low-Grade Gliomas.

Introduction The purpose of this study was to determine whether multi-voxel magnetic resonance spectroscopic imaging (MRSI) can differentiate between intracranial neoplastic and non-neoplastic and between neoplastic ring-enhancing lesions (RELs) based on differences in major metabolite ratios in their enhancing and peri-enhancing regions. Methods In a prospective observational study involving patients with an intracerebral RELs, MRSI using the two-dimensional multi-voxel point-resolved spectroscopy (PRESS) chemical-shift imaging (CSI) sequence at an echo time (TE) of 135 milliseconds (ms) was performed on a total of 38 patients. Of 38 lesions, 23 (60.5%) were neoplastic and 15 (39.5%) were non-neoplastic. Of the 23 neoplastic lesions, 12 were high-grade gliomas (HGGs), seven were metastases, and four were low-grade gliomas (LGGs). Major metabolite ratios, i.e., choline-to-N-acetylaspartate (Cho/NAA), choline-to-creatine (Cho/Cr), and N-acetylaspartate-to-creatine (NAA/Cr), were calculated in the enhancing and peri-enhancing regions of the RELs. A Mann-Whitney U test was run to determine differences in metabolite ratios at different voxel locations between neoplastic versus non-neoplastic lesions, HGGs versus metastatic lesions, and HGGs versus LGGs. A receiver operating characteristic (ROC) curve analysis was performed to derive cut-off values for Cho/NAA and NAA/Cr ratios in the enhancing and peri-enhancing portions of the lesions. Results The sensitivity, specificity, positive predictive value, and negative predictive value for categorizing an REL in either neoplastic or non-neoplastic lesions using MRSI with magnetic resonance imaging (MRI) were 91.3%, 73.3%, 84%, and 84.6%, respectively. There was a statistically significant difference between Cho/NAA (p = 0.006) and NAA/Cr (p = 0.021) ratios in the enhancing region of 23 neoplastic and 15 non-neoplastic lesions. In the voxel placed in the peri-enhancing portions, the differences between Cho/Cr ratios were just significant (p = 0.047). A cut-off score of Cho/NAA >1.67 in the enhancing regions gave a sensitivity of 82.6% and specificity of 60%. The cut-off score for NAA/Cr of <0.80 in the enhancing regions showed a sensitivity and specificity of 60.9% and 86.7%, respectively. Of the 23 neoplastic lesions, 12 HGGs and seven metastases were differentiated using the Cho/NAA ratio in the peri-enhancing region with a cut-off value of 1.21, sensitivity of 100%, and specificity of 85%. A cut-off value of Cho/Cr ≥1.45 in the peri-enhancing regions showed a sensitivity of 83% and a specificity of 71.4%. For discriminating between 12 HGGs and four LGGs both from the 23 neoplastic REL group, using the cut-off score for Cho/NAA in the enhancing portions ≥4.16 showed a sensitivity of 0.75 and specificity of 100%. In the peri-enhancing regions, a cut-off score of ≥2.07 provided a sensitivity and specificity of 83% and 100%, respectively. Conclusion Conventional MRI sometimes poses a diagnostic challenge in distinguishing between neoplastic and non-neoplastic lesions and other neoplastic RELs. Interpreting MRSI findings by comparing the major metabolite ratios in the enhancing and peri-enhancing regions of these lesions may enable distinction between the two.