INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

HYPOGLYCEMIA is a common problem in critically ill neonates, and when severe it results in permanent neurological dysfunction (28). Hypoglycemic neuronal injury in mature animal models is associated with progressive energy failure characterized by ATP depletion and decreased cerebral oxygen consumption (CMR_O2) (32). In these models, the onset of energy failure coincides with the appearance of isoelectric electroencephalogram (EEG), the duration of which correlates directly with neuropathological damage. EEG recovery following glucose restitution is delayed compared with metabolic recovery in proportion to the duration of the coma. Temperature influences metabolic rate during and after hypoxic-ischemic brain injury, but its effect on energy metabolism during and after hypoglycemia is not well understood. In adult rats, body temperature falls during hypoglycemic coma (10), and this mild degree of hypothermia during hypoglycemic coma may be neuroprotective in select brain regions (4). Our prior studies showed CMR_O2 transiently increases during hypoglycemic coma in piglets maintained at a brain temperature of 38.5°C by radiant warming (23). This is normothermic compared with that in awake unanesthetized piglets in our laboratory, which have rectal temperatures of 38.5-39.5°C. Comparable temperatures have been reported for normothermic anesthetized piglets by other investigators (11, 30). The mechanism and pathological importance of increased CMR_O2 and its relation to brain temperature during hypoglycemia are not known. Hypoglycemic neuronal injury in mature animal models involves excitotoxic mechanisms (48). Excitotoxicity during ischemia is temperature dependent (20), but the influence of temperature on excitotoxic mechanisms during hypoglycemia is not known. Appreciation of the clinical significance of temperature in acute brain injury has grown dramatically in recent years. This is particularly relevant for newborns and infants because radiant warming is standard practice to maintain normothermia during the course of critical illness.

Our objectives in these studies were to evaluate forebrain CMR_O2 and excitatory amino acids (EAA) concentrations in cortical microdialysates during and after hypoglycemic coma in piglets maintained normothermic compared with those allowed to become mildly hypothermic during coma. We tested the hypotheses that during hypoglycemic coma in the immature brain 1) increased microdialysate EAA concentrations coincide with increased forebrain CMR_O2, and 2) maintaining brain normothermia by radiant heating increases CMR_O2 and microdialysate EAA concentrations compared with the unwarmed condition. We further evaluated hypoglycemia-induced perturbations by defining the temporal profile of the recovery of forebrain CMR_O2, microdialysate EAA concentrations, and EEG after glucose restitution. We tested the hypotheses that glucose restitution in piglets after an hour of hypoglycemic coma restores excessive EAA microdialysate concentrations and forebrain CMR_O2 to normal but that neuronal function as measured by EEG is persistently and severely impaired in a temperature-dependent manner.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Animal preparation. Piglets 1-2 wk old and weighing 3-4 kg were anesthetized with pentobarbital (65 mg/kg ip then 10 mg · kg¹ · h¹ iv) and mechanically ventilated via tracheostomy with a Harvard small animal ventilator to maintain normoxia (Pa_O2, 120-160 mmHg) and normocapnia (Pa_CO2, 35-40 mmHg). Paralysis was maintained with pancuronium 1 mg/kg im. During surgical preparation, rectal temperature was monitored and regulated to 38.5 ± 0.5°C with a heating blanket. Catheters were placed in the descending aorta via the femoral artery for withdrawal of the radiolabeled microsphere reference sample, axillary artery for monitoring mean arterial blood pressure (MAP) and arterial blood sampling, and inferior vena cava via femoral veins for fluid and drug infusion. A catheter was placed in the left atrium via a thoracotomy for radiolabeled microsphere injection. A catheter was placed in the sagittal sinus for withdrawing cerebral venous blood. A 4-Fr balloon tip occluding catheter (Fogarty, American Edwards, Puerto Rico) was placed in the descending aorta distal to the microsphere withdrawal catheter for temporary support of cerebral perfusion in the event of systemic hypotension. Intracranial pressure was monitored via a 2.5-mm Silastic ventricular catheter (Cordis, Miami, FL) placed in the left lateral ventricle via a burr hole. A temperature probe (Mon-A-Therm, LaBarge, St. Louis, MO) was placed in the frontal cortex at a depth of 1.5 cm from the skull surface, and brain temperature was regulated according to the experimental protocol with heating pads and overhead lamps. Microdialysis cannulas were implanted through a burr hole and small dura incisions using a stereotactic micromanipulator in the right parietal cortex, 1 cm lateral and 1 cm caudal to the bregma, at a depth of 10 mm and at an angle of 30° perpendicular to the cortical surface. Cannula position was ascertained by infusion of methylene blue and brain dissection at the conclusion of the experiment. Cannulas so placed were reproducibly found to occupy the full thickness of the cortical gray matter and did not appreciably penetrate the subcortical white matter. These studies were approved by the institutional Animal Care and Use committee.

Cerebral blood flow and oxygen consumption. We used the radiolabeled microsphere technique (22) to measure regional cerebral blood flow (CBF). Radiolabeled microspheres (0.3 ml, 1.5 × 10⁶ spheres, 16 ± 0.5 µm diameter, DuPont-NEN, Boston, MA) were injected via the left atrial catheter. Six different spheres (¹⁵³Gd, ^114mIn, ¹¹³Sn, ¹⁰³Ru, ⁹⁵Nb, and ⁴⁶Sc) were injected in random order, and reference samples were withdrawn at 1.94 ml/min. At the end of the experiment, the piglet was killed with a left atrial injection of supersaturated KCl, and the brain was placed in 10% physiologically buffered Formalin. After 2-7 days fixation, the brain was sectioned and analyzed in an autogamma scintillation spectrometer (Minaxi Auto-Gamma 5000 Series, Packard Instruments). The energy windows were set (in keV) at 68-170 for ¹⁵³Gd, 174-230 for ^114mIn, 360-440 for ¹¹³Sn, 450-560 for ¹⁰³Ru, 690-820 for ⁹⁵Nb, and 830-1,200 for ⁴⁶Sc. The overlap of activity from high-energy isotopes into low-energy windows was corrected by differential spectroscopy, and regional CBF was calculated by the reference sample technique (22). (Regional CBF is expressed in ml · min¹ · 100 g tissue¹ by correcting for tissue weight.) The brain was sectioned to determine flow in the following regions: brain stem, cerebellum, thalamus, caudate, hippocampus, periventricular white matter, superficial cortical gray matter, forebrain, and whole brain. Forebrain CBF was calculated as the total of all supratentorial structures excluding thalamus. Forebrain CMR_O2 was calculated from forebrain blood flow and arterial and venous oxygen contents: CMR_O2 = CBF × arteriovenous O₂ content difference. Forebrain CBF was used in calculating forebrain CMR_O2 based on the assumption that sagittal sinus venous O₂ content reflects brain metabolic activity in the supratentorial compartment.

Blood analysis. Pa_O2, Pa_CO2, and pH were measured with a Radiometer analyzer (ABL 3, Copenhagen, Denmark). Arterial and venous O₂ content and hemoglobin concentration were determined with a Radiometer Hemoximeter (model OSM3). Arterial glucose concentration was measured with a glucose autoanalyzer (model 2300, YSI, Yellow Springs, OH). All machines were calibrated periodically throughout the experiments.

EEG. A single bipolar EEG channel was recorded using silver ball electrodes seated in the frontal skull and a needle electrode in the tongue as a ground. Tracings were made on a Grass model 6 EEG machine (Quincy, MA), with high and low filters set at 70 and 1 Hz, respectively, and sensitivity set at 3 µV/mm. Simultaneous electrocardiogram was recorded in a separate channel to assist detection of artifact. EEG was evaluated at 5- to 10-min intervals throughout the experiment. The onset of coma was defined as the time at which EEG became isoelectric. EEG recovery was evaluated visually, and abnormalities were classified in a manner blinded as to the experimental group, as previously described (23). Scores were assigned on a scale of 0-4 where 0 is isoelectric, 1 is burst suppression or very severe voltage and/or frequency depression, 2 is moderate voltage and/or frequency depression, 3 is mild voltage and/or frequency depression, and 4 is baseline anesthetized recording.

Microdialysis and amino acid analysis. Cerebral microdialysis was performed using techniques modified by Van Wylen et al. (47). Cannulas were constructed of two 60-mm long segments of silica tube (ID 75 µm, OD 150 µm) sealed in a 15-mm segment of dialysis tubing (ID 300 µm, molecular mass cutoff 5,000 Da) to give a dialysis distance of 10 mm. Warmed artificial cerebrospinal fluid prepared fresh for each experiment was bubbled with 95% N₂-5% CO₂, to a PO₂ of 20-30 mmHg, then infused using gas-impermeable 1-ml glass syringes and a Harvard microinfusion pump set at 1.0 µl/min. Twenty-minute samples were collected, placed immediately on ice, then frozen and stored at 70°C until biochemical analysis. Dialysate amino acids were initially quantitated by HPLC (models 710 Autosampler and 510 Solvent Delivery System) using the recently developed AccQ-Tag assay (15). AccQ-Fluor reagent (6-aminoquinolyl-N-hydroxy-succinimidyl carbamate) was added to dialysate samples. The derivatized amino acids were separated by reversed-phase chromatography using a gradient of 0-60% acetonitrile in 140 mM sodium acetate (pH 6.43), detected fluorimetrically (excitation and emission wavelengths 250 and 350 nm, respectively), and quantified by comparison to known standards treated identically. In later experiments, dialysate amino acids were measured by a HPLC electrochemical detection system (16). An ESA (Coulchem) HPLC pump, an ESA electrochemical detector, and an automatic integrator were used with a reverse phase column. Dialysates were derivatized using o-pthalaldehyde via a Gilson autosampler. After 2 min, 20 µl of sample were injected onto the column, and the derivatized amino acids were monitored with the electrochemical detector. Total area under the peak was integrated and compared with the homoserine internal standard. Microdialysate concentrations of EAA measured in baseline conditions were the same when measured by either method. Glutamate, aspartate, and glutamine were measured in all experiments. The nonneurotransmitter amino acids serine and histidine were measured in samples from four warmed animals for comparison with glutamate and aspartate.

Protocol for hypoglycemic coma. After surgical preparation and equilibration of microdialysis cannulas for 90 min, baseline measures for physiological parameters and EEG were obtained. Collection of microdialysis samples was begun at this time and continued for 20 min/sample during the protocol. Insulin was then given at 200 IU/kg iv, and EEG was monitored at 5- to 10-min intervals until the appearance of isoelectric EEG, designated time 0. CBF and CMR_O2 were measured at 0, 30, and 60 min of coma. After the 60-min coma measurements, blood glucose was restored with a 0.5-ml/kg bolus of 50% dextrose iv, followed by a continuous infusion of 25% dextrose to maintain arterial glucose levels at 60-90 mg/dl. Recovery measurements of regional CBF and forebrain CMR_O2 were obtained at 30 and 120 min after glucose restitution. Control animals did not receive insulin and underwent the same anesthesia, instrumentation, and measures of physiological variables, regional CBF, CMR_O2, cerebral microdialysis, and EEG at comparable time points as hypoglycemic animals.

The influence of temperature was evaluated in two groups of animals. In "warmed" animals, brain temperature was monitored and experimentally maintained at normothermic levels of 38.5 ± 0.5°C using overhead lamps and a heating pad throughout the protocol. We considered 38.5°C to represent normal brain temperature because normal awake piglets have a rectal temperature of 38.5-39.5°C in our laboratory, and brain temperature measured under anesthesia as described above at the start of the experiment is within 0.5°C of the rectal temperature (data not shown). Similar cerebral and rectal temperatures in normothermic piglets have been described by other investigators (30). In "unwarmed" animals, brain temperature was monitored but radiant heat was not used to control brain temperature at any particular value. Rectal temperature was monitored in these animals and regulated using a heating pad to maintain rectal temperature at 38.5 ± 1°C. Control animals were warmed to maintain normothermic brain temperatures comparable to the "warmed" hypoglycemic animals. We studied 16 hypoglycemic animals (9 warmed, 7 unwarmed) and 5 control animals.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Brain temperature was constant over time in all three groups of animals. Therefore, brain temperatures were pooled across time for comparison between animal groups. Brain temperature in warmed hypoglycemic animals was higher than in unwarmed animals (38.4 ± 0.05 vs. 37.6 ± 0.12°C) and slightly higher than control values (38.2 ± 0.04°C; means ± SE, P < 0.01, Student's t-test corrected for multiple comparisons). Rectal temperature was constant over time in all groups and was the same between groups: warmed animals 38.7 ± 0.07°C vs. unwarmed animals 38.4 ± 0.09°C. Isoelectric EEG occurred at comparable time intervals after insulin administration in warmed and unwarmed animals (99 ± 12 min vs. 97 ± 17 min, respectively). There were no differences between groups or over time in arterial pH (7.36 ± 0.01), arterial PCO₂ (40 ± 1 mmHg), or arterial PO₂ (141 ± 9 mmHg). There were no differences between groups or within groups over time with respect to MAP, intracranial pressure (ICP), and cerebral perfusion pressure (CPP) (Table 1). Arterial glucose concentration was stable over time in control animals. There were no differences between warmed and unwarmed hypoglycemic animals in arterial glucose concentrations during any phase of the protocol (Table 1). CPP decreased during coma compared with baseline measures in both warmed and unwarmed groups but remained >60 mmHg at all time points during coma. The change in CPP was due to a slight rise in ICP, while MAP remained stable over time. Hemoglobin was stable over time in controls (11.3 ± 0.3 g/dl) and decreased slightly over time in both hypoglycemic groups from baseline values of 11.2 ± 0.4 in warmed animals, and 11.9 ± 0.5 in unwarmed animals at baseline, to 8.9 ± 0.5 and 9.2 ± 0.5, respectively, at 120 min of recovery.

Forebrain CMR_O2 doubled in the warmed animals at the onset of coma but did not change in unwarmed animals (Fig. 1). Forebrain CBF increased three- to fourfold above baseline, remained elevated throughout the 60-min coma, and returned toward baseline values with glucose restitution (Fig. 2). CBF and CMR_O2 were stable over time in control animals. The increase in CBF was similar in the warmed groups compared with the unwarmed groups in all regions evaluated, although baseline CBF was slightly higher in warmed animals (Fig. 2). CBF values from both temperature groups were pooled at each time point for comparison among brain regions. Hindbrain and subcortical gray matter nuclei showed greater increases in CBF than cortical gray matter, as follows (CBF shown as %baseline): thalamus 672 ± 46% > brain stem 586 ± 64% > caudate 512 ± 57% > hippocampus 357 ± 56% = cerebellum 352 ± 32% = cortical gray matter 305 ± 24% > white matter 260 ± 22% (P < 0.05 by ANOVA on ranks with post hoc comparisons). The increase in CBF during coma was sustained longer in hindbrain and subcortical structures than in cortical gray matter.

View larger version (19K): [in this window] [in a new window]   Fig. 1.   Forebrain cerebral O2 consumption (CMRO2) increases at onset of coma (time 0) in warmed (n = 9 piglets) but not in unwarmed animals (n = 7 piglets). Control animals were stable over time (n = 5). Data are means ± SE. EEG, electroencephalogram. * Differs from baseline by ANOVA with repeated measures, P < 0.05. + Differs from warmed group by t-test, P < 0.05.

View larger version (19K): [in this window] [in a new window]   Fig. 2.   Forebrain cerebral blood flow (CBF) increases to same level at onset of coma (time 0) in warmed (n = 9) and in unwarmed animals (n = 7). Control animals were stable over time (n = 5). Data are means ± SE. * Differs from baseline by ANOVA with repeated measures, P < 0.05. + Differs from warmed animals by t-test, P < 0.05.

We observed increases in EAA microdialysate concentrations during hypoglycemic coma (Fig. 3). EAA microdialysate concentrations were stable throughout the protocol in normoglycemic controls. During the preisoelectric phase of hypoglycemia, EAA microdialysate concentrations in warmed and unwarmed animals were stable at levels similar to normoglycemic controls. With the onset of isoelectric EEG, there was a 15- to 20-fold increase in glutamate and aspartate and a 3- to 5-fold increase in glycine. Levels were generally maximal at 30 min of coma and remained elevated or began to fall slightly during the 60 min of coma. All EAA values returned quickly to normal after glucose restitution, reaching baseline values between 30-60 min of recovery. Microdialysate concentrations of glutamate during coma were greater in warmed than in unwarmed animals, whereas aspartate and glycine were the same in the warmed compared with unwarmed animals. Analysis of the nonneurotransmitter amino acids serine and histidine shows only minor (1- to 2-fold) or no increases during the hypoglycemic coma (Fig. 4). Glutamine decreased to 40% of baseline values during coma in both warmed and unwarmed animals.

View larger version (22K): [in this window] [in a new window]   Fig. 3.   Microdialysate concentrations of neurotransmitter amino acids increase at onset of coma (time 0). Increases in aspartate (A) are similar for warmed (n = 9) and unwarmed (n = 7) animals. Glutamate efflux (B) is greater in warmed than unwarmed animals. Glycine efflux (C) is similar in warmed and unwarmed animals. Amino acid efflux was stable over time in control animals (n = 5). Data are means ± SE for each 20-min epoch, plotted at its midpoint on the x-axis. * Differs from baseline by ANOVA for repeated measures, P < 0.05; + differs from warmed at same time epoch by t-test, P < 0.05.

View larger version (19K): [in this window] [in a new window]   Fig. 4.   Microdialysate concentrations of nonneurotransmitter amino acids are shown for warmed animals (n = 4) for serine (A) and histidine (B). Glutamine (C) decreases similarly during coma in warmed (n = 7) and unwarmed animals (n = 7). Amino acid efflux was stable over time in control animals (n = 5). Data are means ± SE for each 20-min epoch, plotted at its midpoint on x-axis. * Differs from baseline by ANOVA for repeated measures, P < 0.05.

EEG was available through 2 h of recovery in eight of nine warmed animals and six of seven unwarmed animals. Representative recordings of two animals depict the sequence of changes from baseline to isoelectric coma, followed by variable partial recovery during 2 h of glucose restitution (Fig. 5). Recovery of EEG activity was delayed in warmed animals compared with unwarmed animals (Table 2). EEG was stable over time in control animals.

View larger version (17K): [in this window] [in a new window]   Fig. 5.   Representative EEG recordings are shown from piglets at baseline normoglycemic normothermic conditions (A), during hypoglycemic coma (B, D), and during recovery (C, E). Tracings from an unwarmed animal show isoelectric EEG with electrcardiogram artifact at onset of coma (B) and partial recovery at 2 h after glucose restitution (C). This record was assigned grade 2 due to moderate amplitude reduction and slowing. Tracings from a warmed animal show minimal recovery at 2 h (E) compared with isoelectric record at onset of coma (D).

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Our experiments are the first in an immature animal model of hypoglycemia to measure the temporal correlation of EAA microdialysate concentrations with forebrain CMR_O2 through an interval of sustained isoelectric EEG and recovery after glucose restitution. This paradigm was chosen because earlier studies by others in mature animals have shown that hypoglycemic brain injury is directly related to the duration of isoelectric EEG (8) and is selectively localized to neurons in superficial cortical laminae (7). In the present study, we have shown that hypoglycemic coma in the immature brain causes marked increases in cortical EAA microdialysate concentrations. Our data further show that microdialysate concentrations of glutamate are selectively sensitive to small differences in brain temperature. These experiments have shown that increased forebrain CMR_O2 coincides with marked increases in EAA microdialysate concentrations and depends on maintaining normothermia. EEG recovery was delayed in animals experiencing increased forebrain CMR_O2 and greater glutamate microdialysate concentrations, suggesting more severe neuronal injury in these animals. These findings support the possibility that hypoglycemic neuronal injury in the immature brain is mediated in part by excitotoxic mechanisms, as is well established in the mature brain (48). Furthermore, we have shown that higher brain temperature may exacerbate hypoglycemic brain injury as a result of the combination of stimulated oxidative metabolism and enhanced excitotoxicity.

The temporal profile of EAA microdialysate concentrations we observed in piglets closely resembles those seen in adult rat hippocampus during hypoglycemia (41). Baseline microdialysate concentrations of neurotransmitter amino acids in piglet cortex were in the low micromolar range for glutamate, submicromolar range for aspartate, and the midmicromolar range for glycine. These values are similar to those reported in adult rat hippocampus (41) and rat pup striatum (44). Increased EAA microdialysate concentrations began to appear at the onset of isoelectric EEG, progressed over the next 30 min, and were sustained through 60 min after the onset of coma at levels 15- to 20-fold above baseline. EAA microdialysate concentrations recovered quickly to baseline values after glucose restitution. The greater magnitude of EAA elevation in our model than that seen in the rat pup (44) may be related to differences between the two models in species (rat vs. piglet), sampling site (striatum vs. cortex), brain maturity, or brain temperature. The EAA elevations we observed are similar in time course but greater in magnitude compared with those seen in other injury models such as ischemia or trauma, which cause 2- to 10-fold increases in glutamate in cerebral microdialysates (20).

The mechanisms underlying the rise in EAA microdialysate concentrations during hypoglycemic coma in our model are not known, and our study was not designed to evaluate mechanisms. However, the temporal profile of microdialysate levels and differences among amino acids help distinguish among several possible mechanisms. Possible explanations for increased microdialysate concentrations of amino acids include: 1) increased transport into the brain due to nonspecific mechanisms such as altered blood-brain barrier function or volume shifts; 2) increased presynaptic release; and 3) failure or reversal of reuptake.

Nonspecific mechanisms such as blood-brain barrier breakdown or volume shifts might be expected to affect neurotransmitter amino acids (glutamate, aspartate, and glycine) as well as nonneurotransmitter amino acids similarly in a given condition. The fact that temperature selectively altered glutamate and not aspartate makes these mechanisms unlikely. The 10-fold greater increase in microdialysate concentrations of neurotransmitter (glutamate, aspartate) compared with nonneurotransmitter (serine, histidine) amino acids further supports the specificity of this phenomenon. Increased presynaptic release of neurotransmitters may occur at the onset of coma in association with widespread membrane depolarization. However, this is an energy-consuming calcium-dependent process that is unlikely to be sustained as coma advances and energy substrate depletion progresses. Kauppinen et al. (26) showed that glutamate release by glucose-deprived synaptosomal preparations is largely calcium independent and coincides with ATP depletion and collapse of membrane ion gradients. This observation was extended by Geng et al. (18) who treated hippocampal neurons with pyridoxal phosphate to reverse ATP depletion and thereby prevent extracellular glutamate elevation during glucose deprivation. The appearance of isoelectric EEG during glucose deprivation coincides with the onset of ATP depletion in adult rats (32) and in piglets (24) and is associated with widespread sustained membrane depolarization and progressive decline in Na⁺-K⁺-ATPase activity (27). Reuptake of neurotransmitter amino acids from the extracellular space occurs via neuronal and glial transporters, their function of which depends on intact membrane ion gradients. Under conditions of marked ATP depletion, membrane ion gradients collapse, transporters fail, and EAAs passively diffuse down their concentration gradients into the extracellular space (33). The temporal profile of EAA microdialysate levels we observed is most compatible with transport failure because it begins with the onset of isoelectric EEG, is progressive and sustained through the entire 1-h coma period, and quickly normalizes after glucose restitution. Rapid restoration of EAA levels to baseline with glucose restitution in our model is consistent with the rapid time course of recovery of adenylate energy charge, membrane potential, and extracellular K⁺ observed by Katsura et al. (25) after 30 min of hypoglycemic coma in rats.

The selective effect of temperature on hypoglycemia-induced changes in glutamate microdialysate concentra-tion compared with other amino acids may reflect differences in their subcellular compartmentation and metabolic fate during glucose deprivation. The extracellular accumulation of EAAs in the circumstance of transporter failure would likely be determined in part by their intracellular concentrations and subcellular compartmentation. Agardh et al. (2) described the effects of hypoglycemic coma on the brain content of amino acids. Tissue concentrations of glutamate decrease, whereas aspartate increases as a consequence of glutamate deamination to -keto-glutarate, providing reducing equivalents through the TCA (Krebs) cycle as a substrate for oxidative phosphorylation. Santos et al. (42) further showed that hypoglycemia-induced glutamate release by synaptosomes correlates well with ATP depletion and membrane depolarization, whereas aspartate release does not. Glutamatergic nerve terminals are highly enriched in glutamate localized to synaptic vesicles through the action of high-affinity vesicular transporters that are specific for glutamate. It is possible in our experiments that ATP depletion proceeded more rapidly at a higher temperature, which would be more likely to correlate with glutamate than aspartate release if the region sampled was more enriched in synaptosomal vesicle glutamate than aspartate. Although we did not measure this directly, the consistently higher baseline microdialysate concentrations of glutamate than those of aspartate would support the notion that the region of cortex we sampled was more enriched in glutamate than aspartate terminals. The effect of temperature on tissue concentrations of aspartate and glutamate during hypoglycemic coma has not been studied. It is possible that temperature effects on microdialysate concentrations of glutamate compared with aspartate reflect differences in the effects of temperature on their intracellular metabolism during hypoglycemia.

We observed a decrease in microdialysate glutamine during the coma similar to that occurring in adult rats (41), which parallels total tissue depletion of glutamine seen by Norberg and Siesjo (38). This has been attributed to an increase in conversion of glutamine to glutamate, which is then converted to -keto-glutarate as an alternative substrate for the Krebs cycle. Our data do not explain the absence of a temperature effect on glutamine levels. Glutamine metabolism includes both astoglial and neuronal components, which provide for conversion of glutamate to glutamine in astroglia, followed by transport of glutamine from astroglia to neurons where it is converted back to glutamate (2). Both steps are energy-consuming processes. Enhanced ATP depletion at higher temperatures may alter glutamine metabolism at multiple sites in the pathway, resulting in no net effect on microdialysate levels.

The effect of hypoglycemia on glycine metabolism has not been studied. Increased microdialysate glycine concentrations paralleled those of glutamate and aspartate in our experiments. Nakashima and Todd (37) showed that glycine release during ischemia is linked to membrane depolarization in a manner comparable to glutamate and aspartate release and concluded that transporter-mediated processes are involved. Glutamate and asparate have similar affinities for a family of Na⁺ cotransporters, whereas glycine is taken up by a distinct Cl cotransporter. The similar temporal profiles we observed for glycine compared with glutamate and aspartate levels support the possibility that they involve similar mechanisms of transporter failure. The reason for a lack of a temperature effect on glycine levels is not evident from our data. There may be differences in the effects of temperature on the development of transporter failure. Nakashima and Todd (37) showed that microdialysate glycine levels during ischemia were significantly less sensitive to temperature changes than were glutamate and aspartate. Alternatively, it is possible that a predominance of glutamate among synaptic vesicle pools of neurotransmitter amino acids, or its central role in intermediate metabolism, account for its unique sensitivity to small temperature changes during a hypoglycemic coma.

We observed that maintaining brain normothermia via radiant heating potentiates the increase in glutamate microdialysate concentrations compared with the unwarmed condition during the hypoglycemic coma. This is consistent with the well-characterized temperature dependence of glutamate microdialysate concentrations in ischemia (14, 37) and trauma (19). Whereas the mechanism for this effect in hypoglycemia is unknown, its possible pathological significance is highlighted by the neuroprotective effect of mild hypothermia in a number of injury models (13, 19, 30, 36), including hypoglycemia (4). Additional temperature-sensitive processes occur during ischemic and hypoglycemic insults that may interact with excitotoxic mechanisms. These include ion homeostasis (29), membrane lipid peroxidation (9), free radical generation (21), and calcium flux (5). Thus temperature-related exacerbation of calcium accumulation, lipid peroxidation, or free radical generation may amplify damage to a host of cellular structures. The minimal degree of EEG recovery in our experiments in the face of prompt normalization of EAA microdialysate levels is consistent with a severe degree of neuronal injury. Furthermore, our observation that EEG recovery was prolonged in warmed animals compared with unwarmed animals supports the possibility that a temperature-related increase in glutamate microdialysate levels is pathologically significant.

Our data show that brain temperature can significantly influence forebrain CMR_O2 during hypoglycemic coma. The lack of a temperature effect on baseline normoglycemic values for CMR_O2 may be explained by the relative insensitivity of the technique and the small increment in temperature. Busija et al. (11, 12) have described the temperature dependence of CMR_O2 in uninjured anesthetized piglets to be ~0.18 ml · 100 g¹ · min¹ per °C. Because the groups in our study differed by a fraction of 1°C, a very small change in CMR_O2 at baseline would be beyond the sensitivity of this method to detect without a very large sample size. The mechanism for an effect of temperature on forebrain CMR_O2 during hypoglycemic coma is not directly evident from the results of our experiments. Our data are compatible with a number of possible explanations: 1) a temperature-sensitive increase in glutamate with secondary glutamate-driven stimulation of oxygen consumption; 2) temperature-sensitive uncoupling of oxidative phosphorylation; and 3) limited sensitivity of the technique for measuring CMR_O2. Small increases in temperature during hypoglycemia could amplify cellular responses to the injury that drive oxygen consumption. Exogenous glutamate application increases local energy substrate utilization in the normal piglet cortex (6). This may be mediated in part by astrocytes, which increase aerobic glycolysis during conditions of glutamate excess (40). Because astroglia contain the brain's only supply of glycogen stores, glial energy production may be selectively stimulated and persist for some time after neuronal synaptic failure as indicated by the onset of an isoelectric EEG. Higher temperature may stimulate more pronounced intracellular calcium overload, which in turn may "drive" mitochondrial respiration (35). The relative deficiency of reducing equivalents due to glucose deprivation (46) may combine with direct excitotoxic mitochondrial damage to uncouple oxidative phosphorylation in a manner that is temperature sensitive (49). Nonglucose substrates such as lactate, ketone bodies, or amino acids, in particular glutamate, serve to maintain oxidative phosphorlyation temporarily during glucose deprivation (31, 43). Agardh et al. (1) have shown that oxygen consumption is maintained during the early stages of hypoglycemic coma in part as a result of oxidation of unknown substrates that do not subserve ATP synthesis. Pathological oxidation or peroxidation of structural elements such as membrane lipids or transporter proteins may be enhanced under conditions of higher temperatures and higher extracellular glutamate levels. The more delayed recovery of EEG in warmed animals supports the notion that the temperature-related increases in CMR_O2, the increased microdialysate glutamate levels, or both may have injurious effects on neuronal function. Finally, the technique we used to measure forebrain CMR_O2 has limited sensitivity and represents an average of all forebrain regions. Metabolic activity is highly topographically organized in piglet cortex (34). Increases in CMR_O2 that are regionally selective or small in amplitude may not be detectable by our technique. As temperature increases, similar metabolic responses may be sufficiently widespread or of sufficient magnitude in selected regions to become detectable by our technique.

Hypoglycemic coma in this model causes marked increases in CBF in all regions, which is not affected by the small temperature difference observed in this study. There are several possible explanations for the lack of a temperature effect. First, the difference in temperature may have only affected the superficial cortex. We measured temperature in the parietal cortex and manipulated it by means of radiant heat from overhead lamps. Because the two groups did not differ with respect to rectal temperature, it is likely that temperature in subcortical and brain stem structures did not differ between the two groups. This could account for the finding of similar CBF in subcortical and brain stem structures when comparing warmed to unwarmed animals. The absence of a temperature effect on CBF in superficial cortical regions may be accounted for in other ways. It is possible that CBF regulatory mechanisms in hypoglycemic coma produce maximal vasodilation in the unwarmed condition. Support for this explanation comes from earlier studies showing that hypoglycemia abolishes vasodilatory responses to hypercapnea (45) and hypoxia (39) in piglets. A third explanation is that the systemic catecholamine response to hypoglycemia is likely to be similar in both groups because they had similar arterial glucose levels. Therefore, the contribution of circulating catecholamines to the cerebral vasodilation would likely be similar in both groups. We did not evaluate the role of catecholamines in the present experiments.

We observed a temperature-related increase in CMR_O2 without a corresponding increase in the CBF response to hypoglycemia. This suggests that the mechanisms that normally couple CBF with CMR_O2 are disrupted in this circumstance. Normally, CBF is coupled to CMR_O2 through a constellation of vasoactive signals. Some vasoregulatory signals reflect energy substrate utilization (adenosine), whereas others are the product of neuronal activity (nitric oxide, potassium fluxes). Our experiments were not designed to determine the mechanism of the increased CMR_O2 or its relationship to the CBF responses we observed. It is possible that maximal vasodilation occurs in the unwarmed condition, as discussed above, and that the circulation is insensitive to incremental changes in vasodilatory signals associated with a higher CMR_O2. Alternatively, the temperature-related increase in CMR_O2 may represent a pathological process due to uncoupling of oxidative phosphorylation or abnormal oxidation of substrates that do not subserve energy metabolism. The temperature-related enhancement of glutamate microdialysate levels that we observed may directly contribute to mitochondrial injury and stimulate abnormal oxidative processes, as previously discussed. Such nonphysiological oxygen consumption may be dissociated from vasodilatory signals that normally regulate oxygen delivery. The delayed EEG recovery in the warmed animals suggests that the higher CMR_O2 experienced by these animals represents an injurious rather than an adaptive process.

This study has several limitations. First, the invasive nature of the experiments and the requisite anesthetic may have altered the responses being measured. However, the range of MAP, heart rate, and arterial blood gas values maintained throughout the protocol was the same as is usually observed in our laboratory in awake unanesthetized normal piglets of the same age and size (data not shown here). Similar values for physiological, arterial blood gas, and temperature parameters in normoglycemic anesthetized piglets have been reported by other investigators (17, 30). Furthermore, data from the control animals shows that values for CBF, CMR_O2, EAA microdialysate concentrations, and EEG recordings remained stable throughout the period of observation despite the invasive nature of the measurements. Second, our interpretation of the significance of the results is limited by the short duration of recovery and the absence of pathological and long-term neurological functional end points. The invasive nature of the measurements precluded the performance of long-term survival studies on these animals. However, previous studies in adult animals showed that pronounced impairment of neurophysiological function following 60 min of hypoglycemic coma in adult rats (3) correlates with severe and widely distributed neuronal injury (2). The slow and incomplete EEG recovery we observed in piglets closely resembles that seen after a similar duration of coma in adult rats and would suggest the neurological injury is comparable in the two models. If this is the case, then the earlier EEG recovery we observed in unwarmed animals would be consistent with less severe neuronal injury (3). Long-term survival studies will be required to verify the significance of this finding.

In summary, hypoglycemic coma in the immature brain leads to a marked time-dependent increase in EAA microdialysate concentrations, which shows temperature dependence selectively for glutamate. This is associated with increased CMR_O2 if the brain temperature is maintained normothermic by radiant warming. CMR_O2 did not increase in unwarmed animals, the brain temperatures of which were allowed to fall to minimally hypothermic levels. There was more rapid recovery of EEG at lower brain temperatures. This is consistent with less severe neuronal injury and may be related to effects of temperature on oxidative metabolism, EAA concentrations, or both during hypoglycemic coma. These data suggest that small fluctuations in the brain temperature may significantly affect hypoglycemic brain injury in infants. This may be particularly relevant in critically ill newborns for whom it is standard practice to maintain body temperature using overhead radiant warmers.