INTRODUCTION

Hypoxic–ischemic encephalopathy (HIE) independent of its origin is a major cause of neurologic morbidity and mortality.¹ The outcome of term infants with HIE varies and prediction of neurological outcome in these infants is difficult.² Standard clinical information used to determine prognosis include the 5-minute Apgar score,³ neurologic examination,⁴ evoked potentials,⁵ electroencephalography,⁶ cerebral blood flow,⁷ and neuroimaging.⁸ Unfortunately, none of these methods reliably predict neurological outcome.^2,9

Proton magnetic resonance spectroscopy (¹H-MRS) has been shown to predict neurological outcome after various central nervous system (CNS) insults.¹⁰ Recent studies in newborns have correlated poor neurological outcomes with low N-acetyl-aspartate/creatine (NAA/Cre) and N-acetyl-aspartate/choline (NAA/Cho) ratios in infants and children post-CNS injury.^{11,12,13,14,15} Presence of lactate after CNS injury has been associated with poor neurologic outcome.^{10,11,12,13,14,15} However, the majority of spectroscopy studies have only reported outcomes at relatively short time intervals after the neonatal insults, usually 1 to 12 months. More recent studies have suggested that long-term neurologic and cognitive outcome is more accurately predicted at 24 rather than 6 to 12 months postinjury.^16,17

We asked whether spectroscopy would improve the prediction of neurological outcome of term neonates 24 months after CNS injury beyond the abilities of standard clinical assessment.

METHODS

Patients

This study is part of a larger study evaluating the clinical use of spectroscopy in pediatric patients with various CNS injuries. We studied term infants who had sustained HIE in the first month of life and who were considered to be at high risk for long-term neurologic impairment. HIE was defined by clinical history and a neurologic examination consistent with an acute global ischemic insult of the CNS, including: (1) initial arterial pH<7.15, (2) Apgar score <5 at 5 minutes, and (3) neurologic findings of generalized hypotonia, lethargy, poor sucking and feeding, and respiratory failure not because of medications or other CNS etiologies. These infants were prospectively followed from the time of injury. This study was approved by the Institutional Review Board of Loma Linda University Children's Hospital and informed consent was obtained from either the parents or the guardians for each infant.

Variables

Clinical data including gestational age, time to spectroscopy, length of time for which the neonate required mechanical ventilation, and duration of the hospitalization were recorded. The standard clinical criteria for predicting outcome were the neurologic examination, electroencephalography, 5-minute Apgar score, initial arterial pH, and initial blood glucose.

Electroencephalogram results were stratified as normal, mild (slowing or mild suppression), moderate (focal seizures), or severe (burst suppression or near electrocerebral silence).³ Neurologic examinations by a pediatric neurologist (blinded to the MRS results) were obtained at the time of injury and at 24 months. The neonatal neurologic examination was stratified using the classification system proposed by Sarnat and Sarnat.⁴ The 24-month-old neurologic examination was scored by the Pediatric Cerebral Performance Category Scale (PCPCS), a six-point outcome scoring system modified from the Glasgow Outcome Scale.¹⁸ We stratified the infants using the six-point PCPCS scores into two groups: good/moderate (normal, mild, or moderate disability) and poor outcome (severe disability, persistent vegetative state, or death).

Spectroscopy variables measured from ¹H-MRS were N-acetyl aspartate (NAA), creatine (Cre), choline (Cho), and lactate. Peak area ratios were computed for NAA/Cho, NAA/Cre, and Cho/Cre.¹² The NAA/Cre, NAA/Cho, and Cho/Cre metabolite ratios of each infant were compared with a mean value obtained from control infants. The study infants were then stratified by the number of standard deviations from the control patients' mean. Values greater than two standard deviations from the mean were considered to be abnormal.

MR Imaging and MR Spectroscopy Techniques

Magnetic resonance imaging and spectroscopy were obtained using a conventional 1.5T whole-body imaging system (Magnetom SP4000, Numaris 2.3, Siemens Medical Systems, Erlangen, Germany) as previously reported.¹⁹ MR images were used to define optimal placement of the volume of interest (VOI) for spectroscopy. Localized water-suppressed proton spectra were obtained using a stimulated echo acquisition mode (STEAM) sequence. Proton spectra were acquired in an 8 cm³ VOI placed in the occipital region of the brain consisting primarily of gray matter. Volumes of interest were not placed in areas with structural injury (e.g. infarction, thrombosis, and hemorrhage). Our previous work has shown that placement of volumes of interest in areas of structural injury are not predictive of global outcome.^11,12,13 Additional eight acquisition spectra without water suppression were obtained and used for the correction of eddy current-induced phase shifts at each voxel location.²⁰ Localized shimming and optimization of the Gaussian pulse amplitude for maximum water suppression was adjusted prior to acquiring the spectra. Total study time averaged 60 to 70 minutes.

Peak area metabolite ratios (NAA/Cre, NAA/Cho, Cho/Cre) were calculated (blinded and independently) on two separate occasions and the measurements averaged as previously described.^19,21 The presence of lactate was determined by identifying a characteristic peak doublet with 7 Hz splitting at 1.3 ppm relative to NAA on the proton spectra.^20,22

Normal values for the occipital metabolite ratios were obtained from six full-term control infants. Normal mean metabolite valuesSD were: NAA/Cre 0.970.26, NAA/Cho 0.830.21, and Cho/Cre 1.190.27. Our values are similar to those previously published.²² Lactate was not observed on any of the control patients.

Statistical analysis

Clinical, neurodiagnostic, and ¹H-MRS data for neonates were compared between the good/moderate and poor outcome groups by independent two-tailed t-test, ², and discriminant analyses. The NAA/Cho, NAA/Cre, and Cho/Cre metabolite ratios for each infant were compared with the mean values of control neonates. Ratios were considered abnormal if they were beyond two standard deviations from the mean. Results of discriminant analysis were reported as the percentage of correctly predicted cases when the different combinations of input variables were grouped together. Sensitivity, specificity, positive, and negative predictive values for each method of predicting outcome were calculated. We analyzed the data using SPSS for Windows (Release 10, SPSS, Inc. Chicago, IL). Differences were considered significant at p0.05.

RESULTS

In all, 33 full-term appropriate for gestational age neonates were enrolled with HIE suffered within the first 48 hours of life from July 1993 to December 1998. All of these infants had perinatal asphyxia, none had traumatic or hemorrhagic injury. There were 15 male and 18 female infants. There were eight infants in the good outcome group, seven in the mild, four in the moderate, eight in the severe, and six infants died. There were no significant differences between the good/moderate and poor outcome groups in gestational age, days to MRS, and days in the hospital. The poor outcome group received longer mechanical ventilation (Table 1).

Table 1 - Patient Population.

Full table

As expected, the standard clinical methods for predicting outcome showed that neonates in the poor outcome group had significantly higher glucoses and more abnormal Sarnat and EEG scores. There were no significant differences in the initial arterial pH or the Apgar score at 5 minutes (Table 2).

Table 2 - Clinical Variables and Spectroscopy Data By Outcome Group.

Full table

Neonates with poor outcomes had a significantly lower NAA/Cho ratio and were more likely to have lactate present. There were no significant differences in NAA/Cre or Cho/Cre ratios (Table 2).

Discriminant analysis showed that individually, the Sarnat score and the NAA/Cre ratios had a 100% specificity, but relatively low sensitivities. The addition of the spectroscopy data (NAA/Cho, NAA/Cre, Cho/Cre, and lactate) to the Sarnat and EEG scores significantly improved sensitivity, but had no impact on specificity (Table 3).

Table 3 - Prediction of Outcome by Individual and Grouped Variables.

Full table

DISCUSSION

The goal of our study was to determine if adding spectroscopy information to known clinical predictors of outcome would improve sensitivity and specificity when predicting outcome in neonates with HIE. Our results show that neonates with a poor outcome have a lower NAA/Cho ratio and had lactate present. Combining commonly used clinical methods for predicting outcome with spectroscopy significantly improved sensitivity, but not specificity for predicting neurological outcome in term neonates with HIE at 24 months of age.

N-acetyl aspartate is considered to be a neuronal marker even though it is found in the precursors of oligodendrocytes.^23,24 Our data at 24 months of follow-up confirm the correlation between reduced cerebral NAA and poor outcome suggesting severe neuronal loss and poor neurodevelpmental outcome.^{11,14,20,22,25,26,27} As in other studies, we found decreased NAA/Cho ratios, but no difference in NAA/Cre ratios.²⁷ Our results differ from studies with less than 1 year of follow-up, which showed correlation between decreased NAA/Cre and poor outcome.^11,26 The length of follow- up time may account for the difference in results. Three of the patients in our current study initially had a poor outcome at 6 months of age. On serial examination, these patients' motor and language skills improved so that they were reclassified to the good/moderate outcome group. This improvement is most likely because of the neuronal plasticity seen in young children.^28,29

Lactate production occurs during anaerobic metabolism during the initial insult to the brain and during secondary energy failure.^30,31 As in other studies, our study showed that the presence of lactate did not always indicate a poor outcome.^23,32,33,34 Since overlying lipid and protein resonance makes lactate difficult to observe in short echo-time sequences, our results may be different from others who use long echo-time sequences. Seizures may increase lactate and decrease NAA for up to 3 hours and therefore may have caused spurious results.^32,35 One of our patients had a clinical seizure in the MRI scanner.

Lipids and macromolecules can be identified with shorter echo-time sequences such as the simulated echo acquisition mode (STEAM) technique at resonances of 0.9 to 1.3 ppm. In areas of ischemia, these metabolites will be elevated and may obscure the lactate doublet at 1.33 ppm. Long echo-time acquisitions such as the point-resolved spectroscopy (PRESS) sequence suppress the lipid component and therefore allow greater sensitivity to lactate. Previous studies have shown that in neonates, the PRESS technique had a higher percentage of correct outcome predictions compared to the STEAM technique at 6 months after injury.^19,36 The STEAM does not allow for accurate quantification of lactate because of the overlying lipid resonances. When this study was initiated, the PRESS technique was not yet available.

When we followed patients for only 6 months in our previous study, we had 100% specificity with no falsely predicted poor outcomes for patients who had abnormal NAA/Cho, NAA/Cre and Cho/Cre ratios.¹¹ Our current study did not duplicate these results. While our ratios had very high specificities, 95, 100, and 89% respectively, they did not completely predict the outcome. This maybe because of the fact that four patients who had good/moderate outcomes at 6 months of age had minimal-to-no improvement in their developmental skills by 24 months and therefore were reclassified in the poor outcome group.

Significant improvement in sensitivity when combining the Sarnat and EEG scores with the spectroscopy variables suggests that a scoring system may be constructed for predicting the outcome. Future studies using chemical shift imaging may be able to localize specific areas of neurologic injury so that at-risk children may receive early intervention.

CONCLUSION

We conclude that spectroscopy significantly improves the sensitivity of accepted clinical predictors of neurologic outcome. The NAA/Cho ratio and lactate were significantly different between the good/moderate and poor outcome groups. The presence of lactate had the best sensitivity for predicting neurological outcome. The NAA/Cho ratio had the best specificity.

References

1. Volpe JJ. Hypoxic–ischemic encephalopathy: clinical aspects. In: Volpe JJ, editor. Neurology of the Newborn. 3 ed. Philadelphia: WB Saunders, Co; 1995. p. 314–369.
2. Robertson CM, Finer NN, Grace MG. School performance of survivors of neonatal encephalopathy associated with birth asphyxia at term. J Pediatr 1989;114:753–760.
3. Nelson KB, Ellenberg JH. Apgar scores as predictors of chronic neurologic disability. Pediatrics 1981;68:36–44. | PubMed | ChemPort |
4. Sarnat HB, Sarnat MS. Neonatal encephalopathy following fetal distress. A clinical and electroencephalographic study. Arch Neurol 1976;33:696–6705. | PubMed | ISI | ChemPort |
5. Majnemer A, Rosenblatt B. Evoked potentials as predictors of outcome in neonatal intensive care unit survivors: review of the literature. Pediatr Neurol 1996;14:189–195.
6. Holmes G, Rowe J, Hafford J, Schmidt R, Testa M, Zimmerman A. Prognostic value of the electroencephalogram in neonatal asphyxia. Electroencephalogr Clin Neurophysiol 1982;53:60–72. | PubMed |
7. Rosenkrantz TS, Zalneraitis EL. Prediction of survival in severely asphyxiated infants. Pediatr Neurol 1991;7:446–451.
8. Barkovich AJ, Westmark K, Partridge C, Sola A, Ferriero DM. Perinatal asphyxia: MR findings in the first 10 days. Am J Neuroradiol 1995;16:427–438.
9. Nelson KB, Emery III ES. Birth asphyxia and the neonatal brain: what do we know and when do we know it? Clin Perinatol 1993;20:327–344.
10. Shevell MI, Ashwal S, Novotny E. Proton magnetic resonance spectroscopy: clinical applications in children with nervous system diseases. Semin Pediatr Neurol 1999;6:68–77.
11. Shu SK, Ashwal S, Holshouser BA, Nystrom G, Hinshaw Jr DB. Prognostic value of ¹H-MRS in perinatal CNS insults. Pediatr Neurol 1997;17:309–318.
12. Holshouser BA, Ashwal S, Luh GY, et al. Proton MR spectroscopy after acute central nervous system injury: outcome prediction in neonates, infants, and children. Radiology 1997;202:487–496.
13. Ashwal S, Holshouser BA, Tomasi LG, et al. 1H- magnetic resonance spectroscopy-determined cerebral lactate and poor neurological outcomes in children with central nervous system disease. Ann Neurol 1997;41:470–481.
14. Groenendaal F, Veenhoven RH, van der Grond J, Jansen GH, Witkamp TD, de Vries LS. Cerebral lactate and N-acetyl-aspartate/choline ratios in asphyxiated full-term neonates demonstrated in vivo using proton magnetic resonance spectroscopy. Pediatr Res 1994;35:148–151.
15. Barkovich AJ, Baranski K, Vigneron D, et al. Proton MR spectroscopy for the evaluation of brain injury in asphyxiated, term neonates. Am J Neuroradiol 1999;20:1399–1405.
16. Nelson KB, Ellenberg JH. Children who "outgrew' cerebral palsy. Pediatrics 1982;69:529–536.
17. Piper MC, Mazer B, Silver KM, Ramsay M. Resolution of neurological symptoms in high-risk infants during the first two years of life. Dev Med Child Neurol 1988;30:26–35.
18. Fiser DH. Assessing the outcome of pediatric intensive care. J Pediatr 1992;121:68–74. | Article | PubMed | ChemPort |
19. Holshouser BA, Ashwal S, Shu S, Hinshaw Jr DB. Proton MR spectroscopy in children with acute brain injury: comparison of short and long echo time acquisitions. J Magn Reson Imaging 2000;11:9–19.
20. Klose U. In vivo proton spectroscopy in presence of eddy currents. Magn Reson Med 1990;14:26–30. | PubMed | ChemPort |
21. Ashwal S, Holshouser BA, Hinshaw Jr DB, Schell RM, Bailey L. Proton magnetic resonance spectroscopy in the evaluation of children with congenital heart disease and acute central nervous system injury. J Thorac Cardiovasc Surg 1996;112:403–414.
22. Kreis R, Ernst T, Ross BD. Development of the human brain: in vivo quantification of metabolite and water content with proton magnetic resonance spectroscopy. Magn Reson Med 1993;30:424–437.
23. Urenjak J, Williams SR, Gadian DG, Noble M. Proton nuclear magnetic resonance spectroscopy unambiguously identifies different neural cell types. J Neurosci 1993;13:981–989. | PubMed | ISI | ChemPort |
24. Bhakoo KK, Pearce D. In vitro expression of N-acetyl aspartate by oligodendrocytes: implications for proton magnetic resonance spectroscopy signal in vivo. J Neurochem 2000;74:254–262. | Article | PubMed | ChemPort |
25. Peden CJ, Rutherford MA, Sargentoni J, Cox IJ, Bryant DJ, Dubowitz LM. Proton spectroscopy of the neonatal brain following hypoxic–ischaemic injury. Dev Med Child Neurol 1993;35:502–510.
26. Robertson NJ, Cox IJ, Cowan FM, Counsell SJ, Azzopardi D, Edwards AD. Cerebral intracellular lactic alkalosis persisting months after neonatal encephalopathy measured by magnetic resonance spectroscopy. Pediatr Res 1999;46:287–296. | ChemPort |
27. Roelants-Van Rijn AM, van der Grond J, de Vries LS, Groenendaal F. Value of ¹H-MRS using different echo times in neonates with cerebral hypoxia–ischemia. Pediatr Res 2001;49:356–362.
28. Foz FB, Lucchini FL, Palimieri S, et al. Language plasticity revealed by electroencephalogram mapping. Pediatr Neurol 2002;26:106–115.
29. Hertz-Pannier L, Chiron C, Jambaque I, et al. Late plasticity for language in a child's non-dominant hemisphere: A pre- and post-surgery fMRI study. Brain 2002;125:361–372.
30. Penrice J, Lorek A, Cady EB, et al. Proton magnetic resonance spectroscopy of the brain during acute hypoxia–ischemia and delayed cerebral energy failure in the newborn piglet. Pediatr Res 1997;41:795–7802. | PubMed | ChemPort |
31. Lorek A, Takei Y, Cady EB, et al. Delayed ("secondary") cerebral energy failure after acute hypoxia–ischemia in the newborn piglet: continuous 48-hour studies by phosphorus magnetic resonance spectroscopy. Pediatr Res 1994;36:699–6706. | PubMed | ISI | ChemPort |
32. During MJ, Fried I, Leone P, Katz A, Spencer DD. Direct measurement of extracellular lactate in the human hippocampus during spontaneous seizures. J Neurochem 1994;62:2356–2361. | PubMed | ChemPort |
33. Lanfermann H, Kugel H, Heindel W, Herholz K, Heiss WD, Lackner K. Metabolic changes in acute and subacute cerebral infarctions: Findings at proton MR spectroscopic imaging. Radiology 1995;196:203–210.
34. Petroff OA, Graham GD, Blamire AM, et al. Spectroscopic imaging of stroke in humans: histopathology correlates of spectral changes. Neurology 1992;42:1349–1354. | PubMed | ISI | ChemPort |
35. Hetherington H, Kuzniecky R, Pan J, et al. Proton nuclear magnetic resonance spectroscopic imaging of human temporal lobe epilepsy at 4.1 T. Ann Neurol 1995;38:396–3404.
36. Saunders DE, Howe FA, van den Boogaart A, Griffiths JR, Brown MM. Discrimination of metabolite from lipid and macromolecule resonances in cerebral infarction in humans using short echo proton spectroscopy. J Magn Reson Imaging 1997;7:1116–1121.