Multiparametric MR assessment of pediatric brain tumors

MR assessment of pediatric brain tumors has expanded to include physiologic information related to cellular metabolites, hemodynamic and diffusion parameters. The purpose of this study was to investigate the relationship between MR and proton MR spectroscopic imaging in children with primary brain tumors. Twenty-one patients (mean age 9 years) with histologically verified brain tumors underwent conventional MR imaging, hemodynamic MR imaging (HMRI) and proton MR spectroscopic imaging (MRSI). Fourteen patients also had diffusion-weighted MR imaging (DWMRI). Metabolic indices including choline-containing compounds (Cho), total creatine (tCr) and lipids/lactate (L) were derived by proton MRSI, relative cerebral blood volume (rCBV) by HMRI, and apparent tissue water diffusion coefficients (ADC) by DWMRI. Variables were examined by linear regression and correlation as well as by ANOVA. Cho (suggestive of tumor cellularity and proliferative activity) correlated positively with rCBV, while the relationship between Cho and ADC (suggestive of cellular density) was inverse (P<0.001). The relationship between rCBV and ADC was also inverse (P=0.004). Cho and lipids (suggestive of necrosis and/or apoptosis) were not significantly correlated (P=0.51). A positive relationship was found between lipids and ADC (P=0.002). The relationships between Cho, rCBV, ADC and lipids signify that tumor physiology is influenced by the tumor's physical and chemical environment. Normalized Cho and lipids distinguished high-grade from low-grade tumors (P<0.05). Multiparametric MR imaging using MRSI, HMRI and DWMRI enhances assessment of brain tumors in children and improves our understanding of tumor physiology while promising to distinguish higher- from lower-malignancy tumors, a distinction that is particularly clinically important among inoperable tumors. (orig.)