Recurrent glioblastoma multiforme versus radiation injury: a multiparametric 3-T MR approach.

Objective The discrimination between recurrent glioma and radiation injury is often a challenge on conventional magnetic resonance imaging (MRI). We verified whether adding and combining proton MR spectroscopic imaging (1H-MRSI), diffusion-weighted imaging (DWI) and perfusion- weighted imaging (PWI) information at 3 Tesla facilitate such discrimination. Materials and methods Twenty-nine patients with histologically verified high-grade gliomas, who had undergone surgical resection and radiotherapy, and had developed new contrast-enhancing lesions close to the treated tumour, underwent MRI, 1H-MRSI, DWI and PWI at regular time intervals. The metabolite ratios choline (Cho)/normal( n)Chon, N-acetylaspartate (NAA)/NAAn, creatine (Cr)/ Crn, lactate/lipids (LL)/LLn, Cho/Crn, NAA/Crn, Cho/NAA,NAA/Cr and Cho/Cr were derived from 1H-MRSI; the apparent diffusion coefficient (ADC) from DWI; and the relative cerebral blood volume (rCBV) from PWI. Results In serial MRI, recurrent gliomas showed a progressive enlargement, and radiation injuries showed regression or no modification. Discriminant analysis showed that discrimination accuracy was 79.3 % when considering only the metabolite ratios (predictor, Cho/Crn), 86.2 % when considering ratios and ADC (predictors, Cho/Crn and ADC), 89.7 % when considering ratios and rCBV (predictors, Cho/ Crn, Cho/Cr and rCBV), and 96.6 %when considering ratios, ADC and rCBV (predictors, Cho/Chon, ADC and rCBV). Conclusions The multiparametric 3-T MR assessment based on 1H-MRSI, DWI and PWI in addition to MRI is a useful tool to discriminate tumour recurrence/progression from radiation effect