Proton MR spectroscopy after acute central nervous system injury: outcome prediction in neonates, infants, and children.

Abstract

To evaluate the usefulness of proton magnetic resonance (MR) spectroscopy in predicting 6-12-month neurologic outcome in children after central nervous system injuries. Localized single-voxel, 20-msec-echo-time MR spectra (including N-acetylaspartate [NAA], choline [Ch], creatine and phosphocreatine [Cr]) were obtained in the occipital gray matter in 82 patients and 24 control patients. Patient age groups were defined as neonates (< or = 1 month [n = 23]), infants (1-18 months [n = 31]), and children (> or = 18 months [n = 28]). Metabolite ratios and the presence of lactate were determined. Linear discriminant analysis-with admission clinical data, proton MR spectroscopy findings, and MR imaging score (three-point scale based on severity of structural neuroimaging changes)-was performed to help predict outcome in each patient. Findings were then compared with the actual 6-12-month outcome assigned by a pediatric neurologist. Outcome on the basis of proton MR spectroscopy findings combined with clinical data and MR imaging score was predicted correctly in 91% of neonates and in 100% of infants and children. Outcome on the basis of clinical data and MR imaging score alone was 83% in neonates, 84% in infants, and 93% in children. The presence of lactate was significantly higher in patients with poor outcome than in patients with good-moderate outcomes in all three age groups (neonates, 38% vs 5%; infants, 87% vs 5%; children, 64% vs 10% [chi 2 test, P < .02]). In children with poor outcomes, NAA/Cr ratios were significantly lower in infants (P = .006) and children (P < .001), and NAA/Ch ratios were significantly lower in infants (P = .001) and neonates (P = .05). Findings at proton MR spectroscopy helped predict long-term neurologic outcomes in children after central nervous system injury.