Main

Perinatal hypoxia-ischemia is a significant cause of neonatal encephalopathy and later neurodevelopmental impairment(1). In the days after birth asphyxia, infants who later develop neurologic impairments have significant abnormalities of cerebral energy metabolism which can be detected by both proton (1H) and phosphorus-31 (31P) MRS(2–8): the cerebral metabolite concentration ratio of PCr to Pi falls, and the peak area ratio of lactate to Cr rises, suggesting that the normal generation of ATP has been disrupted. The magnitude of these early changes in cerebral phosphate metabolism predicts the severity of later neurodevelopmental impairment(9).

In a preliminary study lactate was detected by 1H MRS in the brains of two severely affected infants some weeks after birth asphyxia(10), and others have reported the detection of lactate in the brain of one infant 3 mo after a focal cerebral lesion(11). The present study thus tested the hypothesis that lactate can be detected later than 1 mo after birth in the brains of infants who display severe neurodevelopmental impairment 1 y after transient perinatal hypoxia-ischemia.

METHODS

Permission for this study was granted by the Hammersmith Hospitals Research Ethics Committee (No. 93/4047). Parental consent was obtained in all cases.

Subjects. Eighteen infants of median (range) gestational age 40 (37-42) wk and birth weight 3245 (2060-4250) g were studied. All showed evidence of perinatal hypoxia-ischemia and neurologic signs consistent with hypoxic-ischemic encephalopathy as graded by Sarnat and Sarnat(12) (Table 1). Six further infants of median (range) gestational age 38 (37-38) wk and birth weight 2905(2380-3610) g, born after uneventful pregnancies and who had no evidence of fetal distress or hypoxic-ischemic encephalopathy, were studied as normal control subjects.

Table 1 Clinical details of birth asphyxiated infants
Full size table

1H MRS. 1H MRS was performed during natural sleep or, if required and in asphyxiated infants only, after sedation with oral or rectal chloral hydrate (50-75 mg·kg-1). Throughout each examination the infants were monitored using MRS-compatible pulse oximetry and electrocardiography, and supervised by a pediatrician experienced in MRS. Infants were clinically stable during examinations, with no evidence of seizures, and not requiring medical support.

An enveloping quadrature transmitter/receiver coil tuned to 64.047 MHz was used, and three-dimensional spectroscopic chemical shift imaging used in the SE data acquisition mode with TE 130 ms (and TE 270 ms in some cases), at the level of the basal ganglia, as previously described(8). Peaks were assigned to Cho, Cr, aspartate, NAA, glutamine/glutamate, lactate, and propan-1,2-diol on the basis of chemical shift, measured in parts/million.

Peak areas for Cr, NAA, and lactate were calculated objectively using automated software (NMR1), and metabolite ratios were expressed relative to Cr. Four spectra each localized to 4.5 cm3 voxels within the basal ganglia were analyzed from each infant, and the results averaged to give an overall value for each subject. The criteria for lactate detection were a peak centered at 1.3 ppm which was inverted at TE 130 ms, with J-coupling where spectral resolution permitted, and which also showed a phase change of 180° if an additional sequence was acquired using a TE 270 ms. Lactate was considered not detectable when such a resonance was not observed at 1.3 ppm. Data were considered unreliable if spectra from the basal ganglia showed a confounding fat signal and/or poor signal-to-noise, principally due to patient movement, and were excluded from further analysis.

Study protocol. Infants with a clinical diagnosis of perinatal hypoxia-ischemia were studied prospectively. Evidence of the hypoxic-ischemic insult was collected during the neonatal period and progress observed for 1 y, when infants were categorized as either neurodevelopmentally normal (group 1) or abnormal (group 2). 1H MRS studies were performed between 4 and 68 (median 11) wk after birth in these infants. Control infants(group 3) were studied by 1H MRS at 5-8 wk of age. The end point of the study was a comparison of the prevalence of cerebral lactate detection in these three groups.

Neonatal evidence of perinatal hypoxia-ischemia: Hypoxicischemic cerebral injury was characterized during the neonatal period in groups 1 and 2 by one or more of the following methods: 1) neurologic examination was carried out to detect the characteristic features of hypoxic-ischemic encephalopathy; 2) EEG was recorded at median 18 (range 8-336) h;3) MRI was performed using a 1.0 Tesla Picker Vista system at 6.5(0-28) d on 32 occasions using age-related inversion recovery (IR3800/30/950), T1-weighted (SE860/20) and T2-weighted (SE3000/120) SE sequences; 4) cerebral lactate/Cr was calculated by 1H MRS using a 1.5 Tesla Picker prototype system at 1 (1-16) d after birth; and 5) measurements of the PCr/Pi ratio were obtained by 31P MRS using a 1.5 Tesla Picker prototype system at 60 (33-111) h after birth. No infant had evidence of a congenital malformation, and investigations for inherited metabolic disorder and infection were negative.

Progress: The subjects were observed for at least 1 y by:1) repeated MRI at 10 (5-52) wk on 30 occasions using age-related inversion recovery [(IR3800/30/950) at <3 mo and (IR3600/30/800) at 3-12 mo], T1-weighted (SE860/20), and T2-weighted (SE3000/120) SE sequences; 2) measurement of head growth velocity from birth to 1 y of age; 3) measurement of NAA/Cr at 11 (4-68) wk by 1H MRS.

Study end points: The two primary end points of the study were: 1) detection of lactate in the brain by MRS after 4 wk of age. 2) Neurodevelopmental outcome at 1 y of age, characterized by structured neurologic examination, the administration of the Griffiths developmental examination and a neurologic optimality score(13) adapted from a proforma used by Kuenzle(14). Infants with a developmental quotient of greater than 85%, an optimality score of>20 and no neurologic abnormality was regarded as having a normal outcome(group 1). All other infants, i.e. those with a developmental quotient of less than 85%, a low optimality score, or at least one major neurologic impairment, were deemed to have an abnormal outcome (group 2).

Data analysis. MRS and neurodevelopmental data were collected and analyzed independently. Data distributions were inspected for normality and homoscedacity, and parametric or nonparametric tests were applied as appropriate. Differences in the frequency of lactate detection in each group were examined by χ2 analysis. Differences in measurements of cerebral metabolite ratios and head growth velocity between the groups were compared by analysis of variance. Differences between groups were considered significant if p < 0.05.

RESULTS

Exclusions. In addition to the subjects reported, a further seven asphyxiated infants and two control subjects were examined but excluded because spectra of analyzable quality could not be obtained, generally because of motion artifact. At 1 y of age, two of these asphyxiated subjects were neurodevelopmentally abnormal and five were normal.

1H MRS spectra from the examination in the neonatal period of two study infants, and single examinations of two other study infants scanned on more than one occasion outside the neonatal period were not of analyzable quality and were consequently excluded. The median coefficient of intraindividual variation was 77%.

Neurodevelopmental outcome at 1 y of age. Ten of the 18 asphyxiated infants were categorized as group 1. The remaining eight were neurodevelopmentally abnormal and placed in group 2. All of the six control infants (group 3) were normal at 1 y.

Neonatal evidence of perinatal hypoxia-ischemia. First, MRI was undertaken in 7/10 subjects in group 1 and 8/8 subjects in group 2. All infants imaged at less than 1 wk of age showed some degree of brain swelling. The MRI images were either normal or showed transient changes of uncertain significance in all but one of the seven subjects studied in group 1, and in that case generalized white matter changes were observed. The MR images were abnormal in all eight subjects in group 2, and were consistent with perinatal hypoxic-ischemic injury.

Second, EEG was recorded in 7/10 subjects in group 1 and 8/8 subjects in group 2. In group 1, it was normal in four subjects, showed some discontinuity in two, and seizure activity in one. In group 2, it showed marked suppression of activity and/or seizures in seven subjects. In the eighth, who was not studied until 2 wk of age, the EEG showed sharp waves but no overt seizures.

Third, cerebral lactate was detected during the neonatal period by1 H MRS in 8/10 in group 1 and 6/7 in group 2. Median (range) lactate/Cr in the infants studied was 0.30 (not detected, 0.42) in group 1 and 0.80 (0.05-1.42) in group 2 (p < 0.05).

Fourth, measurements of the PCr/Pi ratio were obtained in 6/10 in group 1 and 4/8 in group 2. Median (range) PCr/Pi in the infants studied was 0.82 (0.75-1.37) in group 1 and 0.64 (0.22-0.75) in group 2 (p< 0.05) These data are given in Table 2.

Table 2 Results of investigations within the neonatal period
Full size table

Progress during the first year of life. Details of 26 follow-up 1H MRS examinations, follow-up MRI results, and neurodevelopmental outcome of the 18 asphyxiated infants are summarized in Table 3. The head growth velocity was significantly lower in group 2 compared with groups 1 and 3 (p < 0.001). NAA/Cr(median, range) at 4-14 wk was significantly lower in group 2 (1.29, 0.65-1.62) than in group 1 (1.66, 1.54-2.63) (p < 0.005).

Table 3 Results of 1H MRS and MRI investigations at 4-68 wk of age and clinical outcome of birth asphyxiated infants
Full size table

Spectroscopic evidence of cerebral lactate later than one month after birth. Representative spectra obtained during the neonatal period and between 4 and 14 wk from infants with normal and abnormal neurodevelopmental outcome are illustrated in Figures 1 and 2. Lactate/Cr ratios obtained at 4-68 wk are given in Table 3.

Figure 1
figure 1

Representative 1H MRS spectra from a 4.5 cm3 voxel within the basal ganglia of a birth asphyxiated infant (no. 2) with normal outcome at (A) 8 h and (B) 6 wk of age. Resonances were assigned to Cho at 3.2 ppm, Cr at 3.0 ppm, and NAA at 2.0 ppm. A signal from lactate (Lac) at 1.3 ppm was detected in the spectrum obtained on d 1, but not in the spectrum obtained at 6 wk.

Full size image
Figure 2
figure 2

Representative 1H MRS spectra from a 4.5 cm3 voxel within the basal ganglia of a birth asphyxiated infant with abnormal outcome (infant no. 13) at (A) 3 d and(B) 8 wk of age. Resonances identified to Cho at 3.2 ppm, Cr at 3.00 ppm, NAA at 2.0 ppm, and lactate (Lac) at 1.3 ppm. A signal from Lac was identified in both spectra.

Full size image

No signal from lactate was detected in any of the asphyxiated infants with normal outcome (group 1) studied after 4 wk of age. Nine of these infants were studied at a median age of 10 wk (range 6-14) and one was studied at 43 wk.

In the eight asphyxiated infants with abnormal outcome a total of 16 examinations were performed after the 1st mo of life, five infants being studied on more than one occasion. Lactate was detected in 7/8 infants in 10 of these examinations. Median (range) of lactate/Cr was 0.24 (not detected, 0.67) at 12 wk (5 wk to 16 mo). A small signal from lactate was detected in one control infant (lactate/Cr 0.04) at 8 wk. No lactate was detected in the remaining five. The difference between the three groups in the presence of lactate after 4 wk of age was significant at p < 0.001(χ2).

DISCUSSION

Subjects. This study looked for evidence of lactate in the brain beyond the neonatal period in infants suffering perinatal hypoxia-ischemia. Control studies and most examinations of asphyxiated infants were made 4-14 wk after birth, although some infants were studied at up to 68 wk of age. Data could not be acquired from all infants examined, but the neurodevelopmental outcome of these excluded infants was similarly distributed to the studied infants, making ascertainment bias unlikely.

Prospective enrollment of patients allowed the spectrum of insult severity included in the study to range from mild to very severe. The severity of insult was examined using a variety of techniques and, although not every one was used for every infant, the results were typical of perinatal hypoxic-ischemic injury. Infants with abnormal neurodevelopmental outcome displayed characteristic MRI(14,15) and EEG(16,17) and they showed lower PCr/Pi, higher lactate/Cr, and later a reduced NAA/Cr ratio and head growth velocity(4,6,8,9,18). No infants displayed clinical features consistent with inherited mitochondrial disease, and no evidence of these conditions was found.

Segregation into normal and abnormal outcome groups at 1 y was by validated and widely used schemes of neurodevelopmental assessment. The Griffiths development quotient scores in isolation are not always considered appropriate to assess infants with severe motor handicap, and the optimality score therefore was used in addition. Assignment to group 2 was made if one of the three measures of outcome was abnormal; but in fact all infants in group 2 were abnormal in all three, which gives confidence in the accuracy of the grouping. Although the sequelae of birth asphyxia are sometimes first diagnosed later than 1 y after birth, other researchers have shown that, in a cohort of full-term infants studied prospectively, examination at 1 y correlates well with the child's status at 5 y of age(19). It is therefore unlikely that the prevalence of severe neuromotor deficit has been significantly underestimated.

Most infants with abnormal neurodevelopmental outcome and persistent lactate also had lesions seen in the basal ganglia by MRI. However, a small number with persisting lactate did not demonstrate MRI abnormalities; the reason for this is unclear but may relate to the timing of investigations or the sensitivity of MRI.

1H MRS. 1H MRS data were acquired from the basal ganglia because these structures are vulnerable to hypoxic-ischemic injury in the perinatal period(15), and increased lactate/Cr ratios have been demonstrated in this region soon after birth asphyxia(8). Spectra from the deep ganglia are also less susceptible to interference from extracranial tissue, allowing optimal signal quality and consistency. This defined volume of interest may help explain why lactate was detected frequently in this study.

The finding of low NAA/Cr associated with neurodevelopmental impairment is in accord with previous data, as is the lack of significant difference between the groups in the neonatal data. We did not detect an increase in the normal infants with maturation, but this was because very few data were acquired late enough in this group for the increase to be apparent(5–7). Interpretation of the glutamine/glutamate peak was not part of this study because the excitation profile of the water pulse was optimized for NAA detection and the TE was selected for inversion of the lactate peak.

Some technical issues regarding quantification need to be considered. First, measurements of lactate/Cr have a relatively high coefficient of variation, 85% in our system. However, the use of multiple voxels in all calculations means that the SEM average is an acceptable 0.15, giving confidence in the accuracy of the quantification. Second, Cr was selected as the metabolite of reference because its concentration is thought to remain unaltered immediately after hypoxic-ischemic injury(20). It is unclear how the Cr concentration changes in the subsequent months after birth asphyxia. However, these issues would not have influenced the overall findings of the study, as the primary outcome was detection of a resonance attributable to lactate, not quantitation of lactate compared with a reference metabolite.

Cerebral lactate concentration after hypoxia-ischemia. In the hours after birth asphyxia, increased lactate/Cr is accompanied by other evidence of impaired cerebral energy metabolism(8). However, small amounts of lactate have been detected in some normal control subjects(8) and in preterm infants(21), which may represent either metabolic immaturity or mild hypoxia-ischemia during delivery.

The present study found an association between the persistence of lactate more than 4 wk after hypoxia-ischemia and neurodevelopmental impairment at 1 y of age. However, lactate was not detected on all occasions in all affected infants. Until a pathologic mechanism to explain the persistence of lactate can be demonstrated, these results cannot be fully understood. It is unlikely that lactate produced during hypoxia-ischemia would persist beyond the 1st mo of life, because dispersal into the circulation, enhanced by the increase in cerebral blood flow after asphyxia(22), should lead to a consistent decline in the cerebral concentration over time. Equally, diffusion of lactate from blood into the brain is also unlikely to be the sole explanation, because there is no evidence that blood lactate is persistently increased after cerebral hypoxia-ischemia, and normal neural cells metabolize lactate as an alternative fuel for oxidative phosphorylation(23).

Persistently increased lactate/Cr may have been due to continuing abnormalities in cerebral energy metabolism. Studies using positron emission tomography have demonstrated abnormal cerebral glucose metabolism in the weeks after asphyxia, with a persistent increase in the metabolic rate for glucose in the basal ganglia and other parts of the brain which are susceptible to hypoxic-ischemic injury(24), although not surprisingly this can be replaced by hypometabolism, presumably when tissue dies(25). In an MRS study of a single adult patient some weeks after a stroke, cerebral lactate was formed in the penumbra of the cerebral lesion from the metabolism of circulating13 C-labeled glucose(26).

There are several possible mechanisms that could lead to abnormal cerebral energy metabolism in the months after birth asphyxia. First, oxidative metabolism may have been impaired due to mitochondrial damage, with a decreased ATP flux and increased NADH/NAD+ leading to a change in the [lactate]/[pyruvate] ratio and an increase in intracellular pH. This hypothesis is under investigation in our laboratory(27). Mitochondrial respiration is persistently depressed in adult rat neural cells that survive an ischemic insult(28). Decreased NAA concentrations may provide circumstantial evidence of impaired respiration in these infants, as reduced mitochondrial function has been shown to lead to a reduction in NAA synthesis(29). Mitochondria might also be dysfunctional because of impaired microcirculation or due to undiagnosed inherited abnormalities of mitochondrial function. However, there was no evidence of circulatory abnormalities or of inherited mitochondrial disease in the infants studied. Second, in studies of experimental animals transient ischemia can lead to a reduction in the activity of the pyruvate dehydrogenase complex, which may in turn lead to increased lactate production(30). Third, it is possible that increased activity by microglia of infiltrating phagocytes might result in increased lactate production. Large numbers of phagocytic cells are found in the brain in the months after hypoxia-ischemia and it has been reported that these cells have a high concentration of lactate when activated(31).

These possible mechanisms suggest hypotheses for further studies to investigate the mechanism and significance of persistence of lactate in the brain, and to determine whether any therapeutic benefit can be gained by intervention in the pathologic processes involved. They may also call into question the definition of cerebral palsy as a nonprogressive condition.