Contrast/Noise Ratio on Conventional MRI and Choline/Creatine Ratio on Proton MRI Spectroscopy Accurately Discriminate Low-Grade From High-Grade Cerebral Gliomas

Rationale and Objectives

Histopathology is the gold standard to establish the grade of brain tumors but biopsy and/or surgery are not always possible. The aim of this study is to determine whether histological grade of tumors may be predicted by means of conventional gadolinium-enhanced MRI and proton magnetic resonance spectroscopy (MRS).

Materials and Methods

In this study, we included 35 consecutive patients with single brain tumors and final histopathological verification: 12 had low-grade glioma, 16 had high-grade glioma, and 7 had single metastasis. Initially, we carried out T1 and T2 MRI paying attention to the following features: border definition, mass effect, heterogeneity of signal, perilesional edeme, hemorrhage, necrosis, and corpus callosum invasion. Gadolinium-enhancement was evaluated with the contrast-to-noise ratio (CNR). Next, single-voxel proton MRS was carried out to measure the absolute values of metabolites (N-acetyl-aspartate, creatine, choline, and myo-inositol) and their ratios in the area of maximum contrast enhancement.

Results

We found that gadolinium-enhancement measured with the CNR (CNR > 35.86) predicted malignancy at 82.6% sensitivity and 91.7% specificity (area under the curve, 0.88; 95% confidence interval [CI], 0.73−0.97). With regard to MRS a choline/creatine ratio higher than 1.56 predicted malignancy at 88.9% sensitivity and 91.7% specificity (area under the curve, 0.94; 95% CI, 0.78−0.99). When we combined the CNR value, the choline/creatine ratio, and the presence of lactates in a model of discriminant analysis the predictive power improved significantly with an area under the curve of 0.99% (95% CI, 0.87−1). However, the used techniques were unable to distinguish metastases from high-grade gliomas accurately.

Conclusions

The intensity of contrast enhancement measured with the CNR, the choline/creatine ratio, and the presence of lactate were the most powerful variables to predict malignancy in brain tumors. The CNR is a simple, objective, and useful tool in the initial assessment of gliomas and metastases.

Key Words:  Brain tumors , conventional magnetic resonance imaging , magnetic resonance spectroscopy , contrast-to-noise ratio