Proton MR spectroscopy in predicting the increase of perfusion MR imaging for WHO grade II gliomas

To investigate the correlation between the metabolite ratios obtained from proton magnetic resonance (MR) spectroscopy and those obtained from MR perfusion parameters (relative cerebral blood volume [rCBV]) in a cohort of low-grade glioma (LGG).

Patients underwent prospectively conventional MR, proton magnetic resonance spectroscopy ((1) HMRS), and perfusion-weighted images (PWI). Statistical analyses were performed to determine the correlative and independent predictive factors of rCBVmax and the metabolite ratio thresholds with optimum sensitivity and specificity.

Thirty-one patients were included in this study. Linear correlations were observed between the metabolic ratios (lactate [Lac]/creatine [Cr], choline [Cho]/N-acetyl-aspartate [NAA], free-lipids/Cr) and rCBVmax (P < 0.05). These metabolic ratios were determined to be independent predictive factors of rCBVmax (P = 0.027, 0.011 and 0.032, respectively). According to the receiver operating characteristic curves, the cutoff values of the metabolic ratios to discriminate between the two populations of rCBVmax (<1.7 versus = 1.7) were 1.72, 1.54, and 1.40, respectively, with a sensitivity = 75% and a specificity >95% for Lac/Cr.

This study demonstrated consistent correlations between the data from (1) HMRS and PWI. The Lac/Cr ratio predicts regional hemodynamic changes, which are themselves a useful biomarker of clinical prognosis in patients with LGG. As such, this ratio may provide a new parameter for making improved clinical decisions.