Main

Cerebral hypoxia results in a change to anaerobic metabolism with conversion of pyruvate in the cytosol to lactate to produce ATP. An increase in lactate is an early indicator that the cell's activity has exceeded its oxygen supply, but high energy phosphate production may still be adequate for cell function, if glycolysis can be increased. After reoxygenation, lactate dehydrogenase metabolizes lactate back to pyruvate, which can enter the mitochondrial tricarboxylic acid (Krebs) cycle where oxidative phosphorylation gives a much higher yield of high energy phosphate per molecule of substrate. Thus in the posthypoxic phase, lactate levels may be raised as the conversion of lactate to pyruvate takes time, but one expects to find maintenance of a normal L/P ratio. Finding a persistently raised L/P ratio after reoxygenation suggests either a continuation of the hypoxic/ischemic state or a derangement of energy metabolism probably at the mitochondrial level (1).

Induction of seizures in newborn animals by chemical substances such as bicuculline or fluorethyl produces a rapid increase in glucose utilization, a rise in brain lactate, and a decrease in phosphocreatine (storage form of high energy phosphate), but brain ATP levels are generally well maintained even after 1 h of seizures (2–4). High cerebral lactate levels above 16-20 mmol/L are thought to be toxic (5,6). The marked rise in cerebral lactate with chemically induced epileptic activity is used as one argument for the urgency of treating posthypoxic seizures with anticonvulsant drugs (7). However, it is not known whether posthypoxic neonatal seizures produce an additional rise in lactate over and above that produced by the initial hypoxia. It is important to understand more about potentially damaging cellular changes in posthypoxic seizures because it is still not certain whether neonatal seizures, especially subclinical ones, are damaging in their own right (8,9).

Newborn infants and newborn pigs with posthypoxic encephalopathy have been studied by proton MRS, and this has shown prolonged elevations of cerebral lactate (10,11), but simultaneous MRS and EEG recording has not been possible and so the relationship between lactate changes and paroxysmal or other EEG changes could not be examined. Furthermore, pyruvate cannot be measured in brain by MRS because of the low concentration. White matter injury is an important part of brain damage in the unmyelinated brain and subsequent disability in the neonatal period (12,13). Differentiation of gray/white matter has not been possible with MRS, and previous studies of cerebral metabolism by microdialysis, biopsy, or freezing have rarely localized sampling at a microscopic level. Thus little is known about the time course of lactate/pyruvate changes in white matter versus gray matter during neonatal posthypoxic encephalopathy.

The objectives of this study were: 1) to use precisely localized microdialysis probes to determine whether posthypoxic lactate and pyruvate concentrations and the L/P ratio in gray and white matter show different time courses from those found in blood; 2) to correlate cerebral lactate, pyruvate, and the L/P ratio with simultaneous EEG findings; and 3) to determine whether the onset of ECA is associated with an increase in brain lactate or L/P ratio.

METHODS

The hypoxic model has been previously described (14), and this study has been approved by the Norwegian Animal Research Committee. Fourteen Landrace piglets (median age, 35 h; range, 13-49 h; median weight, 1.6 kg; range, 1.0-1.9 kg) were supplied on the morning of the study by local breeders.

Animals were ventilated and instrumented during halothane anesthesia for full cardiovascular and EEG monitoring as previously described (14). Two or three microdialysis probes were stereotactically inserted into the cerebral hemisphere frontally (8 mm lateral to the midline and 5 mm anterior to the bregma) or parasagittally (8 mm lateral to the midline and 5 mm posterior to the bregma) in white (10 mm depth) and gray (2 mm depth) matter.

The hypoxic insult. Hypoxia lasted for 45 min and was induced by reducing FIO2 to the maximum concentration at which the EEG amplitude was below 7 µV (low amplitude EEG). There were transient increases in FIO2 during the hypoxic period to correct bradycardia or hypotension. After 45 min the animals were ventilated with the oxygen fraction needed to obtain SaO2 > 95%. Animals with an arterial pH less than 7.0 were buffered (base excess × weight × 0.2 mmol) with NaHCO3. At the end of hypoxia the animals were kept under halothane anesthesia for 6 h and randomized to either normothermia (which is a rectal temperature of 39.0°C for piglets) or mild hypothermia (35.0°C). The purpose of this particular study was not to compare the effects of posthypoxic hypothermia with normothermia. The results from the two groups of animals were similar as shown in Table 1. Arterial blood gases were analyzed from the umbilical artery catheter and the target PaCO2 was 4.5-5.5 kPa. Arterial pressure was measured continuously from the umbilical artery catheter and a minimum MABP of 5.3 kPa (40 mm Hg) was maintained. The cardiovascular parameters throughout the experimental period are shown in Table 1. There was no significant difference in MABP, blood gases, or severity of the insult between the hypothermic and the normothermic group. Heart rate was lower in the hypothermic group from 3 h onward, and this is likely to reflect reduced metabolism. Two animals (one hypothermic and one normothermic) received dopamine to keep MABP above 5.3 kPa (one normothermic needed 10 µg kg-1 min-1 in reducing doses for 2 h, another hypothermic 10-15 µg kg-1 min-1 for the whole posthypoxic period). The need for dopamine did not relate to the severity of the insult.

Table 1 Mean (SD) values for cardiovascular and blood chemistry results during the insult followed by 5 h of either normothermia (NT) n = 7 (39°C rectal temperature) or mild hypothermia (HT) n = 7 (35°C rectal temperature)
Full size table

End tidal halothane concentration was reduced from 1% during normothermia to 0.6% during hypothermia to obtain the same level of anesthesia (15). Before and during the insult, glucose 35 g/L in NaCl 4.5 g/L was given at 7.5 mL kg-1 h-1. At the end of the insult and for the next 6 h, glucose at 5 mL kg-1 h-1 was given at varying concentrations (35 to 50 g/L) to keep the blood glucose between 2 and 10 mmol/L. Anticonvulsant medication was not given despite the occurrence of ECA and clinical seizures. At the conclusion of the experiment (6 h post reoxygenation), halothane anesthesia was increased, and the brains were perfusion fixed via both common carotid arteries with 4% phosphate-buffered formaldehyde run for 20 min followed by formaldehyde immersion. The microdialysis probes were left in situ for 24 h to ensure that the probe canals could be visualized on later histologic examination.

EEG analysis. Two-channel EEG from each hemisphere was continuously recorded on tape (Oxford Medilog system 9000) before, during, and for 5 h after the insult and examined with respect to background activity and presence of pathology at each 15-min epoch, corresponding to the corresponding microdialysis sampling period. The technique and analysis were based on those of Connell et al. (16). ECA was defined as spike or sharp wave activity with an amplitude more than double the background activity (amplitude of background activity was taken from compressed EEG, ignoring peaks) at a regular frequency and lasting longer than 20 s. Clinical seizures were defined as rhythmic pathologic movements accompanied by ECA. Burst suppression was defined as periodic spells of activity followed by near isoelectricity. Spike or sharp wave activity was defined as irregular occurrence of spike or sharp waves, with duration less than 20 s but with a frequency of at least one spike or sharp wave per min. Nonparoxysmal activity contained no spikes or sharp waves and was characterized by the background activity. Low amplitude EEG was defined as a peak-to-peak amplitude of less than 7 µV.

Because EEG was recorded continuously, the onset and duration of ECA could be timed. The six animals who resumed continuous EEG activity after the insult had significantly lower median duration of low amplitude EEG during the insult (32.1 min) compared with those with a frankly pathologic EEG, two animals with posthypoxic burst suppression (41.7 min), four with ECA (39.2 min), and two with low amplitude EEG (38.3 min).

ECA started between 1.5 and 2.5 h after reoxygenation in four animals. No anticonvulsive drugs were given, and ECA was virtually continuous with only very short pauses until the end of the experiment.

Histology. Coronal blocks around the probes were embedded in paraffin, subserially sectioned at 5 µm, and stained with hematoxylin and eosin. The locations of the probe canal in white or gray matter were accurately described. Six-hour survival is too short for full histopathologic damage to have developed, but cortical damage was assessed using the same score as reported previously (14). The damage was evaluated around the probe canal as well as in other areas of the cerebral cortex. The examiner was blinded to the mode of treatment. All four groups had similar FIO2 and MABP during the insult. In each of the 14 animals, histopathologic examination showed that at least one microdialysis probe was localized in parasagittal gray matter, and one probe was localized in frontal white matter. The results presented here are from the probes localized in this manner.

Figure 1 shows that the longer the duration of low amplitude EEG the higher the neuropathology score r = 0.77 (p = 0.0014). Those who developed ECA, burst suppression, or had persistently low EEG amplitude after the insult had the greatest neurologic damage.

Figure 1
figure 1

The vertical scale is a subjective grading of cortical neuronal damage. Grade 0, no damage, grade 4.0, >75% of neurons injured. The horizontal scale is the duration in minutes of low (<7 µV) background activity of the EEG trace during the 45-min period of hypoxic ventilation. Each animal is coded according to its main EEG pattern after reoxygenation. An asterisk (*) is added to those who were hypothermic (35 vs 39°C) after the insult.

Full size image

Microdialysis. Microdialysis probes (CMA 11, CMA Microdialysis, Stockholm, Sweden) with 1-mm membrane length were used (17). Probes with the smallest available membrane length were chosen to sample selectively from gray or white matter. The newborn piglet cortex is less than 2 mm thick. The probes were perfused at 2 µL/min with a solution containing 150 mM NaCl, 3 mM KCl, 1.3 mM CaCl2, 1.2 mM MgSO4 using a CMA 100 microinjection pump (CMA Microdialysis, Stockholm, Sweden). At this flow rate a probe with a 1-mm long membrane will yield a typical in vitro recovery of about 2.5% for lactate and pyruvate (data not shown). Fifteen-minute collection periods gave 30-µL sample volumes. The samples were stored at -70°C until analysis. There was a 3.5-h period of stabilization between the probe insertion and the hypoxic insult. Hillered et al. (18) using the same make of probes as in this study, have shown in rats with middle cerebral artery occlusion that baseline lactate and pyruvate values are stable 1 h after probe implantation, much earlier than are excitatory amino acids.

Chromatography. The microdialysate was injected via a CMA 200 autoinjector into a HPLC system equipped with a polymeric resin based column (Brownlee Applied Biosystems) according to the method of Hallström et al. (19). The mobile phase consisted of 2 mM sulfuric acid and the flow rate was 0.3 mL/min. A UV detector at 214-nm wavelength was used.

Blood lactate and pyruvate analyses.. One-milliliter blood samples were taken before hypoxia, after 45 min of hypoxia, 15 min after hypoxia, and then at 90, 150, 210, and 270 min. The samples were mixed with 0.5 mL of perchloric acid before centrifugation. Due to a technical error plasma samples from only 10 out of the 14 animals could be analyzed (five normothermic and five hypothermic). There was no significant difference in any cardiovascular values or severity of insult between the group of four (two normothermic and two hypothermic) not analyzed and the remaining 10. Lactate and pyruvate were measured by quantitative enzymatic detection at 340 nm (Sigma Chemical Co. Diagnostics).

Data analysis. For descriptive statistics, mean and SD or median and total range were used. Between group comparisons were performed using ANOVA with Bonferroni correction for repeated testing (20). Where appropriate, paired t tests were used (Statview 4.5, Abacus Concepts Inc., Berkeley, CA). The level of significance was set at 0.05 (two-sided).

RESULTS

Figure 2, A-C, shows serial values for plasma lactate, pyruvate, and the L/P ratio in 10 animals. The plasma lactate concentrations peaked at 20-24 mmol/L at the end of hypoxia and returned to values not significantly different from those before hypoxia after 3.5 h (Fig. 2A). Plasma pyruvate levels (Fig. 2B) kept rising after hypoxia and peaked by 2-3 times baseline values from 15 to 90 min after the insult. The L/P ratio in plasma (Fig. 2C) rose from a mean value of just over 20 to 85-165 during hypoxia. The plasma L/P ratio was not significantly above the prehypoxic value by 90 min after reoxygenation, but at the end of the insult the L/P ratio in plasma was significantly higher in animals who later developed a burst suppression pattern or persistently low amplitude EEG.

Figure 2
figure 2

(A-F) Serial values (mean ± SE) for arterial plasma lactate (A), pyruvate (B), and L/P ratio (C) before, at the end of 45 min of hypoxia, and repeatedly for 5 h after hypoxia in 10 pigs. The simultaneously sampled cerebral dialysate lactate (D), pyruvate (E), and L/P ratio (F) in all 14 animals are shown.

Full size image

The dialysate lactate concentration in gray matter was always higher (2-3 times) than the simultaneous level in white matter. In gray and white matter the dialysate concentrations of lactate showed a rise to 4 times the prehypoxic value by 30 min after reoxygenation (Fig. 2D). Cerebral dialysate lactate remained high at 1.5 h. By 4.5 h, dialysate lactate values in gray and white matter for all 14 animals were still twice the prehypoxic median, and these differences were statistically significant. Dialysate pyruvate values (Fig. 2E) were consistently higher in gray matter than in white matter and rose after hypoxia much less at 1.5 h than did lactate, returning to prehypoxic levels thereafter. The L/P ratio in gray matter was not significantly different from white matter (Fig. 2F). The L/P ratio rose approximately 3 times in gray and white matter at 30 min and remained significantly elevated in gray matter and white matter at 4.5 h post reoxygenation. Thus the brain showed a different time course with a much more prolonged increase in lactate and L/P ratio than did plasma.

When the piglets subjected to mild hypothermia were compared with the piglets held normothermic, there was no significant difference for cerebral dialysate lactate, pyruvate, or L/P ratio at any time. The animals with continuous EEG activity, burst suppression, or ECA showed a clear downward trend (Fig. 3, A, C,and D) in both gray and white matter lactate levels from 1.5 to 4.5 h, whereas the two animals with persistently low amplitude activity showed persistently elevated or rising lactate concentrations at 4.5 h (Fig. 3B). Figure 3C shows the serial gray matter lactate concentrations in four animals with documented ECA. Because EEG was recorded continuously, the onset of ECA could be timed precisely. After hypoxia, there was an initial rise in gray and white matter lactate before the onset of ECA, followed by decline after the onset of ECA.

Figure 3
figure 3

Individual values shown for each animal with corresponding symbols for gray and white matter. Serial values for lactate in gray (black symbols) and white (white symbols) matter in: (A) six pigs who maintained continuous background activity on EEG after hypoxia, (B) two pigs who had persistently low amplitude EEG without seizures after hypoxia, (C) four pigs with ECA after hypoxia, and (D) two pigs who developed burst suppression after hypoxia.

Full size image

The L/P ratio before hypoxia was similar in gray and white matter. The time course of simultaneous L/P ratios in gray and white matter was similar within individual animals. Figure 4, A-D, shows the L/P ratios according to the four different EEG patterns. In the three groups showing either continuous activity, burst suppression, or ECA, the peak L/P value occurred at 30 min after the insult, normalizing by 4.5 h and with no difference between gray and white matter. ECA was not associated with a rise in L/P ratio, and the L/P ratio was not significantly different from prehypoxic levels even after 2 h of seizures (Fig. 4C). Figure 4B shows the L/P ratio in the two animals with persistently low amplitude EEG. These showed not only the highest L/P ratios, but a different time course from all the other animals with ratios increasing, instead of decreasing, after 1.5 h and being over 10 times baseline at 4.5 h.

Figure 4
figure 4

Serial values for L/P ratio in gray (black symbols) and white (white symbols) matter before, at the end, and after hypoxia. Individual values are shown for each animal with corresponding symbols for gray and white matter. (A) Six pigs who maintained continuous background activity on EEG after hypoxia, (B) two pigs who had persistently low amplitude EEG without seizures after hypoxia, (C) four pigs with ECA after hypoxia starting from 1 to 1.5 h after the insult, and (D) two pigs who developed a burst suppression EEG pattern after hypoxia.

Full size image

DISCUSSION

This study uses a global hypoxia-ischemia model in which the whole animal becomes hypoxic with some secondary circulatory impairment and develops clinical, EEG, and neuropathologic evidence of posthypoxic encephalopathy (14). Thus our model is much more like the clinical situation of birth asphyxia than pure ischemia models using vessel ligation or epilepsy models using electric shock or chemicals. The studies currently described were not extended for longer than 6 h after hypoxia and thus, a full assessment of neuropathologic damage could not be made. However, in cortex and white matter, which is the first region to be damaged in this model, we found a similar relationship between pathology score and duration of low amplitude EEG, thus validating the model (14). Neuropathologic examination after 6-h survival cannot reveal whether hypothermia is protective or not because we do not know to which extent further (delayed) cell death will occur in the two groups. Microscopic examination enabled us to know retrospectively which probes were definitely sampling from within gray or white matter.

Microdialysis has been extensively used in metabolic investigations but has two main drawbacks. One is that acute intracerebral implantation leads immediately to a number of disturbances in the vicinity, such as damage to the blood-brain barrier and disrupted glucose metabolism (21) causing initial high lactate values in the microdialysate. We have previously shown that extracellular lactate by intracerebral microdialysis becomes stable within 1 h of probe insertion in the rat (18). There is little reason to suspect that the consequences of probe insertion are radically different in the pig. In our experiments a total stabilization and baseline period of 210 min (3.5 h) was allowed before the hypoxic insult. We have not yet developed a chronic microdialysis technique in the piglet. Such a technique also has its limitations as repeated microdialysis perfusions have been shown to affect brain tissue with hypercellularity and granulocyte infiltration, with the suggestion that blood-brain barrier permeability is changed (22). If there is an error after only 3.5 h of stabilization this would probably give a slight underestimation of the real changes in L/P ratio. The second concern is that the percentage recovery of biologic molecules can vary between sites and subjects. The absolute concentrations in micromoles/L of lactate and pyruvate in the dialysis fluid are much lower than the true concentrations in the extracellular fluid in the brain, but the trends over time within a subject should be reliable, and the L/P ratio should be reliable when comparing animals. Thus we have emphasized L/P ratios and trends over time in our results.

The lactate, pyruvate, and L/P ratios in plasma showed the expected conversion to anaerobic metabolism during hypoxia with a 12-fold increase in lactate and 5-fold increase in L/P ratio. After reoxygenation, plasma lactate decreased more rapidly than pyruvate, and the plasma L/P ratio normalized within 90 min. This is consistent with reoxygenation facilitating the reconversion of lactate to pyruvate, which can then be metabolized within the mitochondria via the Krebs cycle. The brain showed an early rise in lactate and a later smaller rise in pyruvate, but in contrast to blood, the brain showed a persistently elevated lactate concentration and L/P ratio in gray and white matter for 4.5 h after reoxygenation. It appears that the body as a whole shows good evidence of recovery of energy metabolism after hypoxia, but the brain has a more persistent disturbance of energy metabolism despite reoxygenation. The two animals who showed the highest sustained L/P ratios were those with persistently low amplitude EEG, i.e. almost no cerebral activity. Persistently raised L/P ratios could be due to prolonged cerebral hypoxia-ischemia. However, there is no evidence to support this. Not only were arterial blood pressure and PO2 levels adequate after reoxygenation (Table 1), but the preparation for microdialysis had involved opening the skull, thus preventing any severe rise in intracranial pressure. Furthermore, postmortem dissection showed no occlusion in the carotid arteries.

This finding is best explained by mitochondrial injury. In an adult cat hypoxia model, the onset of neurologic signs 3-5 h after reoxygenation correlated with evidence of mitochondrial dysfunction including markedly reduced cytochrome oxidase (23,24). The glial cells in white matter are capable of storing glycogen, which can be released as glucose and then metabolized to pyruvate and lactate within the glial cells. White matter contains far fewer neurons than does gray matter. Neurons cannot store glycogen and are dependent on a constant supply of substrate, i.e. glucose, lactate, or ketones. Glial cells within gray matter are capable of releasing glucose from stored glycogen, which can then be taken up by the specific glucose transporter molecules on the surfaces of neurons. Neurons involved in ECA would be expected to consume more energy while discharging. Experimental seizure models using electric shock or chemicals such as fluorethyl or bicuculline to provoke seizures in previously normal animals (including newborns) have consistently shown a large (100-200%) increase in cerebral lactate as well as a decrease in cerebral glucose levels (2–4). In our study, the animals did not increase cerebral lactate or the L/P ratio with the onset of ECA.

One possible reason for the lack of increase in lactate is that posthypoxic seizure activity in the newborn brain involves a relatively smaller proportion of the brain's neurons than either adult type epilepsy or electrically or chemically induced fits. Posthypoxic neonatal seizures are often subclinical or subtle and are not generalized tonic-clonic seizures involving all four limbs, frothing at the mouth, and incontinence of bladder and bowel (25), and this gives some support to the above suggestion. A more likely explanation relates to the ability of the newborn brain to use lactate as a fuel under certain conditions with good oxygenation and perfusion (26). Indeed, recent work has shown that lactate is an obligatory energy substrate for recovery of synaptic function in the posthypoxic rat hippocampus (27). The increased energy demand of ECA might thus increase consumption of lactate with the net cerebral lactate concentration not increasing but showing a downward trend, as we observed. If lactate were being used as a substrate for oxidative phosphorylation, lactate would have to be continuously converted to pyruvate and one would expect a decreasing L/P ratio, which was, in fact, found in our study.

Thus our main conclusions are the following. 1) Neonatal posthypoxic cerebral energy metabolism is disturbed in both white matter and gray matter and shows impaired metabolic recovery compared with the body as a whole. 2) Elevated cerebral extracellular L/P ratios 2.5-4.5 h after reoxygenation indicate a persistent disturbance in energy metabolism, probably at the mitochondrial level. The most disturbed energy metabolism was found in the two animals who had no posthypoxic seizures but only continuous low amplitude EEG. 3) Posthypoxic neonatal seizures are not accompanied by a rise in gray or white matter lactate or L/P ratio.

The marked rise in cerebral lactate demonstrated with the chemical induction of seizures has been thought to be toxic (2–6) and has supported the urgency of controlling posthypoxic seizures (7). Our finding that neonatal posthypoxic seizures are not accompanied by any rise in extracellular cerebral lactate or L/P ratio thus reinforces the doubt as to whether neonatal posthypoxic seizures are damaging in their own right or are merely a sign of disturbed neuronal function.