Main

Perinatal asphyxia is reputed to be the most important cause of acute neurologic injury in the newborn and occurs in 6 per 1000 term live births(1,2). It is a major cause of neonatal death, brain injury, and neuromotor and intellectual impairment, and 10-20% of survivors suffer from mental retardation, cerebral palsy, and seizure disorder later in childhood(3,4).

Studies of animal models have suggested that whereas some neuronal loss may occur during the hypoxic-ischemic insult, a cascade of processes is also initiated that may result in the attainment of critical thresholds, hours or days later, which will trigger damage and further expression of cell injury and death(5).

The evolution of cerebral energy metabolism during and after HI has been studied in vivo by means of MRS(6). During HI in animal models, there was impairment of oxidative phosphorylation with an increase in the concentration of Pi and a reduction in the concentration of PCr in the brain, an increase in the cerebral concentration of lactate, and an acidic pHi(7–9). Pi/PCr is a surrogate measure of ADP concentration ([ADP]) which accumulates as ATP concentration ([ATP]) falls, and which stimulates glycolysis and oxidative phosphorylation(10). Pi/PCr thus reflects intracellular phosphorylation potential(11,12). After resuscitation, these variables briefly normalized, but despite adequate oxygen and substrate supply to the brain, a period of "delayed energy failure" began 8-24 h later, characterized by a second reduction in phosphorylation potential and increase in lactate, but with a rise in pHi(8,9,13).

In newborn infants who suffered severe birth asphyxia, there was a similar delayed decline in cerebral high energy phosphates, an increase in brain lactate, and an alkaline pHi during the days subsequent to birth(14,15). The severity of this delayed energy failure(16) correlated with subsequent severity of adverse outcome and reduced head growth at 1 and 4 y of age(17,18); similarly elevated cerebral lactate within 18 h of birth asphyxia predicted adverse outcome at 1 y of age(15).

Recent studies have suggested that the metabolic consequences of HI can persist for much longer than this in infants who develop neurodevelopmental impairment. Lactate was detected in the brain after 4 wk of age in seven of eight infants with abnormal neurodevelopmental outcome at 1 y after perinatal asphyxia; none of the infants with normal outcome at 1 y showed evidence of persisting lactate(19). One infant demonstrated an alkalotic pHi in the brain for 10 months after severe perinatal asphyxia(20).

The present study therefore addressed the hypothesis that after NE thought to be secondary to perinatal HI, cerebral energy metabolism remains abnormal for a prolonged period, with increased lactate concentrations that are accompanied by increased pHi and decreased cell phosphorylation potential. A secondary aim was to determine whether there was any difference in these variables between the normal and abnormal neurodevelopmental outcome groups.

SUBJECTS AND METHODS

Permission for this prospective study was granted by the Hammersmith Hospital Research Ethics Committee (no. 93/4047), and parental consent was obtained for every examination.

Subjects. Forty-three infants, born at median (range) gestational age 39.5 wk (36-42 wk) and birth weight 3340 g (2290-4450 g), were selected by the following criteria: clinical suspicion of birth asphyxia based upon a history of fetal distress (late decelerations on cardiotocography, meconium-stained liquor, acidemia (i.e.. pH < 7.1 and/or base deficit > 12 mmol/L) in umbilical cord blood or arterial blood within 30 min of birth), and abnormal postnatal neurologic signs consistent with NE(21). Six infants were small for gestational age. In all cases, appearances on magnetic resonance imaging at 2-14 d of age were consistent with hypoxia-ischemia damage(22), and findings included one or more of the following: brain swelling, loss of the posterior limb of the internal capsule, abnormal signal in the posterior lentiform nucleus and thalamus, cortical highlighting, and/or loss of gray/white matter differentiation.

1H and 31P MRS. To measure Pi, PCr, and pHi phosphorus-31 (31P) MRS was used. The peak area ratios of lactate/Cr and NAA/Cr were measured with proton (1H) MRS. Cr was chosen as the metabolite of reference because of its relative stability after HI, and so lactate/Cr is a reflection of lactate concentration(23). Brain volumes always encompassing the basal ganglia, an area particularly susceptible to damage after HI, were assessed(24).

1H and 31P MRS was performed after sedation with oral or rectal chloral hydrate (50-75 mg/kg-1). Throughout each examination, the infants were monitored by MRS-compatible pulse oximetry and ECG and were supervised by a pediatrician experienced in neonatal resuscitation and MRS.

A total of 77 examinations were obtained from the patient group: 25 infants were less than 2 wk old; 16 were more than or equal to 2 wk and less than or equal to 4 wk; 25 were more than 4 wk and less than or equal to 30 wk; and 11 were more than 30 wk old. Eighteen of the infants were studied more than once. Proton data from subject no. 30 at 48 wk of age have been reported previously(19).

Cerebral MR data were obtained with a Picker prototype 1.5 Tesla MRS system (Picker International, Cleveland, OH), using a double-tuned pediatric birdcage coil. Data were collected between June 1996 (when the double-tuned pediatric birdcage coil became available) and December 1997 (when the system was upgraded to a Picker Eclipse configuration). A set of localized 1H MR spectra were obtained from a 16 × 16 or 32 × 16 grid covering a transverse plane at the level of the basal ganglia by using three-dimensional chemical shift imaging in the spin-echo data acquisition mode with TE of 130 ms and 270 ms(25). The nominal voxel size was 4.5 cm3. To process data, cosine filtering was used in each spatial domain and an exponential line-broadening filter function of 1.2 Hz in the spectral domain before three-dimensional Fourier transformation. A knowledge-based algorithm was used for the removal of the water residuum and for baseline flattening(26). A set of localized 31P MR spectra were then collected by two-dimensional chemical shift imaging. Data were collected from planes including the basal ganglia, each nominally 20 mm thick, with TR 1 s, pulse angle 45°, and 32 data collections at each gradient permutation. This data set was filtered with a cosine filter in the spatial domain and an exponential filter of 5.3 Hz in the spectral dimension.

Peak areas of lactate, Cr, and NAA were measured in the 1H MR spectra using inverse polynomial lineshape functions in NMR1 (New Methods Research Inc, Syracuse, NY). The criteria for lactate detection were an inverted peak at TE 130 ms, centered at 1.3 ppm; doublet structure (J-coupling 7 Hz) if adequate spectral resolution; and a phase change of 180° at TE 270 ms, allowing assignment of lactate to be confirmed when there was ambiguity as to the extent of peak overlap with lipid signals. Lactate/Cr and NAA/Cr were generally measured from four spectra localized to the basal ganglia, and the results were averaged to give a single value for each subject. Lactate was considered detected when lactate/Cr was greater than 0.05.

Pi/PCr was determined by fitting the peaks to Lorentzian lineshape functions using NMR1. pHi was measured from the difference in the chemical shifts of Pi and PCr(27).

Statistical considerations. Infants were grouped according to normal and abnormal outcome at 1 y of age. Data collected during three time periods were compared: <2 wk, ≥ 2 to ≤ 4 wk, >4 to ≤ 30 wk. The data were not normally distributed, and thus lactate/Cr, NAA/Cr, Pi/PCr, and pHi were transformed to normality by using log (x-k), where k was -0.02, -0.50, 0.50, and 6.88, respectively(28). Normality was confirmed for each transformed variable by the skewness/kurtosis test for normality(29). Linear and multiple regression analyses were used to define the relationship between the transformed variables for the group overall. Regression analyses were then performed to compare the normal and abnormal outcome groups at the different time points. Robust regression methods were used to take account of multiple measurements in some infants(30).

Neurodevelopmental outcome. Neurodevelopmental outcome was assessed at 1 y of age, using Griffith's development scales(31) and a detailed neurologic examination. Neurologic signs were quantified with an optimality score(32,33). This score includes items for tone, posture, passive and elicited motility and tone, interaction, reflexes, and vision and hearing responses. The maximum possible optimality score was 21.

Neurodevelopmental outcome was categorized as normal or abnormal. Infants were classified as normal if they had no abnormal neurologic signs and achieved a Griffith's DQ of >85. All other infants were regarded as having an abnormal neurodevelopmental outcome. MRS data were compared between normal and abnormal groups, but for illustrative purposes, the severity of neurodevelopmental impairment was classified as mild, moderate, severe, or no discernible development. Infants with minor neurologic signs, such as hypotonia or asymmetry of tone but normal development (DQ > 85), were classified as mildly abnormal. Infants with DQ 75-84 were classified as mild, those with a DQ of 50-74 as moderate, those with a DQ of 20-50 as severe, and those with DQ <20 as having no detectable development. All infants who died did so because of neurologic problems, and in this study, they were considered together with those with severe neurologic impairment.

RESULTS

Exclusions. In addition to the studies reported, data from 19 examinations of 12 of the study infants were not of analyzable quality or the study protocol could not be completed, generally because of motion artefact. These data were consequently excluded. This included four infants with normal outcome (one examination in the neonatal period, three examinations of three infants between 4 and 30 wk, and two examinations of two infants after 30 wk) and eight infants with abnormal outcome (four examinations of four infants in the neonatal period, six examinations of five infants between 4 and 30 wk, and three infants after 30 wk). No data from infants with normal outcome aged >30 wk were obtained, so a comparison between data from infants with normal and abnormal outcome could be made only up to 30 wk of age.

Neurodevelopmental outcome. The perinatal clinical details, Griffith's General Quotient, and Optimality Score from 17 asphyxiated infants with normal outcome at 1 y are shown in Table 1.

Table 1 Clinical details of birth-asphyxiated infants with normal outcome at 1 y
Full size table

The perinatal clinical details, Griffith's General Quotient, and Optimality Score of 26 asphyxiated infants with abnormal outcome (mild, moderate, severe, and no detectable development) at 1 y, or those who died, are shown in Table 2.

Table 2 Clinical details of birth-asphyxiated infants with abnormal outcome at 1 y or who died (ordered by outcome)
Full size table

One infant with a normal outcome and five infants with an abnormal outcome were small for gestational age.

Spectroscopy findings. Fig. 1 (A and B) provides an illustrative example of proton and phosphorus spectra from one infant after severe NE, showing the presence of lactate coinciding with elevated Pi/PCr and alkaline pHi at 2 d of age. Fig. 1 (C and D) shows the persistence of lactate and an alkaline pHi at 9 wk of age in the same infant.

Figure 1
figure 1

Representative 1H (A, C) and 31P MR (B, D) spectra from a term infant with severe birth asphyxia. The 1H MRS data were collected with TE 130 ms. At 2 d of age (A, B) lactate was obviously present (mean lactate/Cr from four spectra in the basal ganglia was 0.8), Pi/PCr was 2.44, and pHi was 7.23. At 9 wk of age (C, D), lactate was still present (mean lactate/Cr in the basal ganglia was 0.26), Pi/PCr was decreased to 0.86, and pHi was less alkaline (7.08). At 0.3 wk of age, a signal was detected from propan-1,2-diol (pd), a carrier for phenobarbitone.

Full size image

The spectral data from the two outcome groups are summarized in Fig. 2 (a and b) and in Fig. 3. Lactate, Pi/PCr, and pHi were most abnormal at 0-2 wk and decreased with increasing age. Univariate linear regression revealed highly significant relationships, lactate/Cr to pHi (p < 0.001) and lactate/Cr to Pi/PCr (p < 0.001), for the group as a whole. These relationships were also significant considering separately the normal outcome group [lactate/Cr to pHi (p = 0.004) and lactate/Cr to Pi/PCr (p < 0.001)] and the abnormal outcome groups [lactate/Cr to pHi (p < 0.001) and lactate/Cr to Pi/PCr (p < 0.001)]. Adding either Pi/PCr or age to the regression equation predicting lactate/Cr from pHi did not significantly improve the fit (p = 0.17, 0.46, respectively).

Figure 2
figure 2

Lactate/Cr plotted against pHi for normal (a) and abnormal (b) outcome. Each baby is identified by the patient number. Linear regression analysis showed a significant relationship between lactate/Cr and pHi. The relationship was the same, irrespective of outcome.

Full size image
Figure 3
figure 3

Mean and 95% confidence intervals of data collected at the different time periods for the spectral parameters, lactate/Cr, pHi, Pi/PCr, and NAA/Cr in the normal and abnormal outcome groups. The p values obtained by comparing the means of the normal and abnormal outcome groups at different time periods are also illustrated. NS = not statistically significant.

Full size image

For the lactate/Cr to pHi relationship, the regression lines were similar in the normal and abnormal outcome groups with similar intercepts (R2 = 0.52). For the lactate/Cr to Pi/PCr relationship, the regression lines of the two groups were parallel (R2 = 43%), but the intercepts were significantly different (p = 0.003), differing by 0.81. In other words, despite the very strong relationship between these variables, for a given Pi/PCr, the lactate/Cr was higher in the abnormal group than in the normal group.

In our comparison of the means of the normal and abnormal outcome groups at different time periods (Fig. 3), lactate/Cr was significantly higher in the abnormal group at <2 wk (p = 0.005) and >4 to ≤ 30 wk (p = 0.01). Pi/PCr was significantly higher in the abnormal outcome group at <2 wk (p < 0.001), but there was no significant difference beyond this time. pHi was significantly more alkaline in the abnormal outcome group at <2 wk (p < 0.001) and 2-4 wk (p = 0.02), and >4 to ≤ 30 wk (p = 0.03). There was no significant difference between groups for NAA/Cr in the first 2 wk of age; beyond this time NAA/Cr was significantly higher in the normal outcome group (p = 0.02, p = 0.03).

DISCUSSION

The major finding of the study was that the increased lactate in the basal ganglia of infants months after NE was associated with an alkaline pHi and an increased Pi/PCr. These findings decreased with increasing time after the insult but could be detected for up to 1 y after the hypoxic-ischemic insult. The relationship between these variables was significant in both the abnormal and normal outcome groups.

The 1H MR data were obtained from a nominal 4.5 cm3 volume encompassing the basal ganglia, whereas the 31P MR data were obtained from a much larger volume, a transverse plane nominally 2 cm thick at the level of the basal ganglia. The volumes could not be matched because of the lower sensitivity of 31P MRS compared with 1H MRS. Cr was selected as the 1H metabolite of reference, because its concentration is thought to remain unaltered immediately after HI injury(23), although it is not clear how the Cr concentration changes in the months subsequent to birth asphyxia. It is important that the infants are adequately still during the MR examination, because motion would cause a decreased signal-to-noise ratio in the spectra and maybe uncertainty as to the accuracy of the metabolite ratios. Data from 19 additional examinations of 12 of the study infants were excluded generally because of motion.

Cerebral lactate concentrations in neonates have been shown to increase with the degree of growth restriction and to be inversely related to gestational age(34). However, infants whose birthweight is appropriate for gestational age and those who are small for gestational age have similar phosphorus metabolites and pHi(35). Cerebral lactate has been shown to be elevated during various pediatric neurologic diseases such as stroke(36), hypoxia-ischemia(25), metabolic and mitochondrial disorders(37), CNS injury(38), and in near-drowning(39). Neonates, infants, and children with acute CNS insults and elevated lactate peaks detected acutely by 1H MRS were more likely to have a poor long-term neurologic outcome than were those in whom lactate was not detected(40).

In adults, several studies have demonstrated persisting lactate in the chronic stage of brain infarction - in one case, as late as 2 y after a middle cerebral artery infarct(41–48). Patients with large and persistent elevations of cerebral lactate in the acute to subacute phase had significant neurologic impairment, whereas those with lesser changes had a more benign course(45,49). The progression from intracellular acidosis during the acute phase to alkalosis during the subacute/chronic phase of adult ischemic brain injury in humans and in animal models has been illustrated by several groups(50–54). Positron emission tomography has also confirmed this crossover from acute acidosis to alkalosis subsequent to adult stroke; a significant pHi increase in subacute to chronic infarcts correlated with a reduced oxygen extraction fraction and blood perfusion exceeding metabolic demand (i.e., luxury perfusion), causing alkalosis by removing excess CO2(55–58). In the months after perinatal HI, this rebound alkalosis has also been demonstrated in one infant followed serially until the age of 10 months: however, it was not possible to measure cerebral lactate levels at the time of this study(20).

Possible mechanisms leading to lactic alkalosis. There are several mechanisms that could explain the chronic lactic alkalosis seen in our infants in the months after NE. One possibility is that lactate is a stagnant pool produced during the acute phase of injury and that it diminishes gradually with time. This, however, is unlikely, according to the physiologic estimates of lactate washout after ischemia(59), blood brain barrier permeability to lactate(60), and the well- characterized biphasic response of lactate after HI(9). After adult stroke, such lactate has been shown to be actively produced in the stroke region by the formation of lactate from 13C-labeled glucose(61). The lactate seen during the period of secondary energy failure therefore must be due to a renewed production.

The second possibility is that persisting lactic alkalosis is attributable to ongoing ischemia. This too is unlikely. Levels of cerebral blood flow after perinatal HI have been shown to be high(62–64) and to be associated with low oxygen extraction(62,65). Positron emission tomography scans demonstrate a period of localized increased glucose uptake or hypermetabolism in the early post-asphyxial period (2-5 d) in areas that go on to develop an infarct(66). After this acute period, there is a local decrease in cerebral blood flow in the parasagittal regions(67) and a halving of the cerebral metabolic rate for glucose(68). In subacute to chronic adult infarcts, an alkaline pHi correlated with a reduced oxygen extraction fraction and blood perfusion exceeding metabolic demand, i.e., luxury perfusion(55,56,69).

A more likely explanation for the persisting lactic alkalosis is an altered redox state caused by inability to regenerate NAD+. Lactate and its redox partner pyruvate have a central position as the terminal metabolites of glycolysis within the cytosol, and pyruvate as the initial substrate for the mitochondrial tricarboxylic acid cycle(70). Lactate is both formed and used by lactate dehydrogenase in a near equilibrium reaction: [lactate]+[NAD+]= [pyruvate]+[NADH]+[H+] or Equation 1 where [NADH] = dihydronicotinamide adenine dinucleotide concentration, [NAD+] = nicotinamide adenine dinucleotide concentration, [H+] = hydrogen ion concentration, and K = equilibrium constant for lactate dehydrogenase.

A decrease in NAD+ or an increase in NADH causes a shift in the equilibrium either toward lactate formation, or a fall in [H+], or both. Consequently, a persisting reduction of the redox potential will lead to increased lactate and an alkaline pHi. Mitochondrial impairment could exacerbate this decreased NAD+/NADH by leading to increased [ADP] which will drive glycolysis and produce NADH, and by an inability to regenerate NAD+ from NADH along the electron transport chain. Further evidence of mitochondrial damage is that children with inborn errors of mitochondrial metabolism have increased levels of cerebral lactate(71,72). The pHi and phosphorylation potential of mitochondrial disorders have not been explored fully; however, a study (with a mouse model) determined the brain pHi to be alkaline, by 31P MRS(73). This situation may be analogous to that occurring in the months that follow perinatal asphyxia(74). Interventional strategies that use dichloroacetate are being developed at present for mitochondrial disorders-dichloroacetate having been shown experimentally to lower lactate levels and improve morbidity by accelerating the oxidation of glucose, lactate, and pyruvate to acetyl-CoA(75).

Activation of poly (ADP-ribose) (PARP) by cellular injury will also consume NAD+ as a substrate to transfer ADP-ribose groups to nuclear proteins in an attempt to effect repair(76–78). Excessive activation of PARP and depletion of its NAD+ might contribute to cell death through consequent depletion of ATP, inasmuch as four molecules of ATP are needed to generate one molecule of NAD+. In vitro studies have shown that inhibition of PARP activity spares the cell from energy loss, thus providing neuroprotection(79). In vivo administration of a PARP inhibitor led to a significant reduction in infarct volume in focal ischemia(80), and genetic deletion of PARP or inhibition of PARP attenuates tissue injury and conserves NAD+(81).

A fourth possibility is that tissue alkalosis enhances glycolysis and lactate production(82). The major regulator of glycolysis is phosphofructokinase which has an alkaline pHi optimum(83). Glycolysis of glucose to pyruvate generates NADH which shifts the cytosolic redox state toward lactate formation. Lactate production, therefore, could be stimulated by alkalosis. Brain lactate increases during hypocapnia in animals(84) and humans(85). However, in severe hypocapnia, oxygen delivery is reduced and may be insufficient to maintain oxidative metabolism, and thus the lactate increase may result from cerebral ischemia(86). The alkalosis resulting from hypocapnia may further limit oxygen delivery to the brain by shifting the oxygen dissociation curve to the left.

A fifth possibility that could explain the lactic alkalosis is the presence of phagocytes. Phagocytic cells have a high rate of anaerobic glycolysis which leads to elevated lactate(87). Brain macrophages (microglia, astrocytes, and leukocytes) have been shown to appear 1-3 d after an infarction and slowly disappear over many months(88). Histopathology of 10- and 15-d-old strokes has confirmed the presence of macrophages with few other cells throughout the infarction core center, blending into border zones of reactive astrocytes(89).

A sixth possible cause of intracerebral alkalosis is proliferation of glial cells to replace necrotic or apoptotic neurons after perinatal HI. Alkalosis may be associated with the stimulation of glial cell growth, as growth factor reportedly increases the pHi in fibroblasts(90). Increased pHi has also been reported in glial cell tumors(91).

Finally, cerebral alkalosis that follows NE may be due to altered buffering mechanisms in the remaining cells. All vertebrate cells, including neurons and glia, contain a Na+/H+ antiporter system and ischemia and associated acute lactic acidosis could cause an adaptive increase in brain buffering by this system. In some cells, Na+ influx and growth factor have been seen to stimulate the Na+/H+ antiporter to increase pHi(92): thus it is possible that this mechanism could contribute to the alkalosis.

This persisting lactic alkalosis therefore may be consistent with an altered redox state, the presence of phagocytic cells, gliosis, and/or altered buffering mechanisms. However, the relative influence of each of these factors is as yet unknown.

It is not known whether lactate is beneficial or detrimental to brain tissue after HI. Some studies suggest a possible role for lactate in supporting synaptic function after hypoxia(93). Several years ago it was demonstrated that brain tissue can respire when lactate is the substrate(94,95), and in vivo studies have shown that brain tissue produces lactate aerobically, especially under stimulating conditions(96–101). Schurr et al. has recently shown that lactate is preferred over glucose for recovery of synaptic function after HI(102). Based on studies with astrocytic and neuronal cultures, it has been hypothesized that glutamate uptake by astrocytes stimulates the production of glycolytic lactate and its aerobic utilization by neurons(103).

Relationship of MRS findings to neurodevelopmental outcome. As expected from previous work(15,104), in our study there was a significant relationship between Lac/Cr in the first 2 wk after birth, and the outcome. We have also confirmed a difference in Lac/Cr after 4 wk of age in the normal and abnormal groups(19).

pHi was significantly different between outcome groups at <2 wk (p < 0.001) and between ≥ 2 to ≤ 4 wk (p = 0.02) and >4 to ≤ 30 wk (p = 0.03). pHi appeared to be the strongest predictor of outcome at all times up to 30 wk postnatal age.

Confirming previous studies(16), the Pi/PCr ratio (in some studies expressed as PCr/Pi) was significantly different in the normal and abnormal groups at <2 wk (p < 0.001), and after this time was not significantly different. As discussed before, Pi/PCr is a surrogate measure of [ADP], which accumulates as [ATP] falls, and which stimulates glycolysis and oxidative phosphorylation. The significant difference between normal and abnormal groups only in the first 2 wk implies that a reduced phosphorylation potential may not be essential in the development of persistent lactic alkalosis.

NAA/Cr was not significantly different between outcome groups until <2 wk, after which time there was a significant difference between normal and abnormal outcome groups at ≥ 2 to ≤ 4 wk (p = 0.03) and >4 to ≤ 30 wk (p = 0.01). Other investigators have suggested that a reduced NAA/Cho was indicative of a poor prognosis after perinatal asphyxia, both neonatally (mean age 7 d)(104), at 3 mo(104), and at 14 d(105). However, a further study in infants at an earlier stage (median age 1.3 d, range 0.1-4.5 d) with age-matched controls showed no convincing reduction in NAA/Cho(106).

NAA is an amino acid found exclusively in the nervous system and is synthesized in brain mitochondria from acetyl-CoA and aspartate by the enzyme L-aspartate N-acetyl transferase. NAA has been used as a neuronal marker, because apart from oligodendrocyte type 2 cells, NAA is present primarily in neurons(107). However, the functions and roles of this amino acid remain unknown. NAA can be detected in the cerebral cortex and white matter of fetuses as early as 16 wk gestation(108) and shows an age-dependent increase in synthetic rate that is reflected in total brain levels of NAA(109,110). As demonstrated in animal and human adults, NAA/Cho ratios decline hours to days after HI(111), and this is reflected in our findings.

Conclusion. This study shows that, in the months after NE, there is a significant relationship between increased cerebral lactate levels and an alkaline pHi. In the infants with abnormal neurodevelopmental outcome at 1 y, lactate/Cr and pHi are generally significantly higher at >2 wk, compared with those infants with a normal neurodevelopmental outcome. Phosphorylation potential, however, is not significantly different between groups at >2 wk, which implies that a reduced phosphorylation potential may not be essential in the development of persistent lactic alkalosis. Further understanding of the mechanisms leading to these prolonged changes may lead us to other modalities of intervention, with the aim of improving outcome.