The three metabolites measured in our study were Ch, Cr and NAA. The concentration of Ch, a component of cell membrane, reflects myelination and cell membrane turnover.28
The normal decrease in Ch during early brain development likely reflects the incorporation of Ch into macromolecules associated with myelin.15,29
The metabolite NAA, which is primarily stored and synthesized in neurons,30
increases as the brain develops. Relative decreases in NAA likely reflect decreased neuronal viability, neuronal function or neuronal loss.15
The single Cr resonance represents both Cr and phosphocreatine. Cr is converted to phosphocreatine, a high energy compound critical for maintaining cellular energy-dependent systems.15
In adults, Cr is often used as a reference metabolite based on the assumption that Cr is relatively unaffected by age or various pathologies.31
In our study, NAA/Ch in the thalamus and basal ganglia increased significantly between 35 and 43 postmenstrual weeks (). This finding is consistent with the literature and indicates an increase in NAA and/or a decrease in Ch as a function of age. Roelants-van Rijn et al
reported increased NAA/Ch in both periventricular white matter and the basal ganglia in preterm infants at 32 and 41 postmenstrual weeks. Similarly, Kreis et al
reported a significant increase in the amount of NAA measured in the cortical gray matter, thalamus and white matter between 32 and 43 postmenstrual weeks. This same study did not find significant changes in Ch during the same developmental period. In an antenatal MRS study, Heerschap et al
found increased NAA between 30 and 41 weeks of gestation. Our study found the positive correlation between postmenstrual age and NAA/Ch when all infants were included, and when the analyses were limited to infants with a normal MDI and a normal PDI. A significant increase in NAA/Ch was not found when analyses were limited to infants with an abnormal MDI or an abnormal PDI. Although this result may be influenced by the limited number of children with developmental delay, it may well reflect abnormal changes in metabolite concentrations near term age in these infants. Neurodevelopmental outcome was not reported in the studies by Kreis et al
or Heerschap et al
but Roelants-van Rijn et al
reported an abnormal outcome in 2 of the 40 infants in the study.
Numerous studies have found decreased NAA/Cr and/or NAA/Ch to be predictive of abnormal neuromotor outcome in term newborns with hypoxic–ischemic encephalopathy scanned at term age.19-24
These decreased ratios were measured in the basal ganglia,19,21,22,24
and intervascular boundary zone.21,22
However, in one study of preterm infants with a mean gestational age of 27.9±3.1, no difference in NAA/Cr or Ch/Cr was found at a postnatal age of 9.8±4.1 weeks between infants with and without white matter damage on MRI scan.32
Another study compared children younger than 2 years with developmental delay (1.5, 0.5 to 2 years; mean, range) with age-matched control subjects (1, 0.5 to 2 years), and compared children older than 2 years with developmental delay (5, 3 to 10 years) with age-matched control subjects (5.7, 3 to 10 years).29
Significant differences in NAA/Cr and Ch/Cr between children with developmental delay and age-matched control subjects were only found in the older group. In children older than 2 years with developmental delay, the NAA/ Cr ratio was decreased in frontal and parieto-occipital subcortical white matter, and the Ch/Cr ratio was increased in the parieto-occipital subcortical white matter. A future longitudinal study could allow a more accurate assessment as to whether the developmental changes in metabolites over time are abnormal in the poor outcome group. Our study is consistent with the limited data available and suggests that although NAA/Cr and/or NAA/Ch are useful in predicting outcome of asphyxiated neonates scanned at term, the ratios are not predictive of future development in VLBW preterm infants scanned near the same PMA.
Lactate accumulates during acute cerebral hypoxic–ischemic insults due to anaerobic metabolism. Numerous studies of term neonates with encephalopathy have found an association between increased lactate and poor developmental outcome.19-24
In the majority of these infants, MRS was performed within 1 week after delivery. As would be expected in these studies, lactate levels were found to be increased because injury was still acute or subacute at the time of MRS. Given the SNR of the data acquisition and the postnatal age of the infants (9.5±3.0 weeks) in our study, lactate peaks were not distinguished from the noise. In one study, however, comparing metabolite ratios of preterm infants scanned at a postnatal age of 9.8±4.1 weeks, Robertson et al
reported a significant difference in lactate/Cr in the periventricular area between infants with and without white matter damage observed on MRI. In our study, 12 infants had MRI abnormalities, but no infants had classic periventricular leukomalacia. Lactate peaks were not significantly above the noise in any of the infants in our study, perhaps because compared with the study by Robertson et al
., the white matter damage was not as severe, the SNR was lower, fewer MRS spectra were acquired at the level of the thalamus and basal ganglia or selected regions of interest were different.
Magnetic resonance imaging did not correlate with metabolite ratios measured by MRS in our study. MRI has been shown to be more sensitive than ultrasonography in predicting white matter abnormalities.8-10
In our previous study using the same MRI classification criteria, the sensitivity, specificity, positive predictive value and negative predictive value of MRI in predicting CP at 20 months were reported as 71, 91, 50 and 96%, respectively.8
The values reported in our study at predicting normal and abnormal outcomes based on the MDI and PDI at 18–24 months adjusted age were 42, 71, 42 and 71%, and 63, 73, 42 and 86%, respectively. These differences likely reflect the differing outcome measure, CP versus Bayley developmental indices. No infants in our study were diagnosed with CP, and only 4 and 2 infants had abnormal MDI and PDI scores at least two standard deviations below the mean (score <70), respectively. Earlier studies by Cook et al.33
have shown that after removing children with CP from a population of preterm infants, MRI even at 8 years of age was not predictive of school performance.
There were several notable limitations to this retrospective study. The sample size was relatively small and few infants exhibited abnormal developmental outcome. Only one MRI/MRS scan was performed on each subject; data at multiple developmental time points would strengthen the conclusions. Because MRS scans were performed after a series of MRI scans, it was not always possible to obtain spectra from all subjects who were not compliant for the full duration of the exam. Ratios, rather than absolute concentrations,34,35
were computed from the spectra. Owing to certain factors, such as magnetic field inhomogeneity over a large volume and patient motion, it was necessary to average signals over a large region of interest to obtain sufficiently high SNR for quantitative analysis. Volumetric MRS imaging acquisition techniques36-38
for increased spatial coverage, as well as shorter echo times39
and phased-array head coils40
for an increased SNR, would improve the metabolite data. MRI data was not included in determining the accuracy of predicting long-term outcome; this study was intended to focus solely on the potential of MRS to make clinical diagnoses, whereas in the clinical setting, MRS is typically used as supplemental data to aid in clarifying MRI findings.
A normal Bayley exam at 2 years of age may not necessarily reflect normal cognitive and motor function at an older age.1
Another study of term neonates with encephalopathy found that 15% of the children considered normal at 2 years had minor neurological dysfunction and/or perceptual motor difficulties at 5 to 6 years of age.41
As a result, follow-up at a later date is necessary.