Effects of Mild Hypothermia on Blood–Brain Barrier Disruption during Isoflurane or Pentobarbital Anesthesia

This study was approved by the Institutional Animal Care and Use Committee of the University of Medicine and Dentistry of New Jersey–Robert Wood Johnson Medical School. Twenty-eight male Sprague-Dawley rats (Hilltop Lab Animals, Scottsdale, PA) weighing 300–340 g were divided into four groups of seven, i.e. , the normothermic isoflurane (NI), hypothermic isoflurane (HI), normothermic pentobarbital (NP), and hypothermic pentobarbital (HP) groups. Isoflurane groups were anesthetized with 1.4% isoflurane in an air and oxygen mixture (fraction of inspired oxygen, 0.25–0.3). For the pentobarbital groups, 50 mg/kg pentobarbital was administered intraperitoneally, and additional pentobarbital (15 mg/kg administered intraperitoneally) was administered 40–50 min after the initial dose. The lungs were mechanically ventilated through a tracheal tube. A femoral artery and a femoral vein were catheterized, and the femoral arterial catheter was connected to a Statham P23AA transducer (Gould Instruments, Cleveland, OH). Blood pressure was continuously monitored and recorded on a Beckman R-611 recorder (Fullerton, CA). The right common, internal, and external carotid arteries were exposed. The external carotid artery was catheterized with a polyethylene tube after its branches were ligated. The tip of the catheter was placed in the external carotid artery approximately 1 mm distal to the carotid bifurcation. The femoral venous catheter was used to administer drugs and radioactive tracers. For the normothermic groups, body temperature was maintained at 37°C with a heat lamp and a servo-controlled rectal thermistor throughout the experimental period. Pericranial temperature was also monitored with a probe in the temporalis muscle with a thermocouple thermistor (Omega Engineering Inc. Model 450ATT. Stamford, CT) and was maintained at 37 ± 0.2°C. For the hypothermic groups, the temperature was decreased to 32°C with an ice-water bag and ambient temperature. The pericranial temperature was maintained at 32 ± 0.2°C. Arterial blood samples (0.2 ml) were drawn anaerobically and were analyzed for partial pressure of oxygen and carbon dioxide, and p  H, using a blood gas analyzer (ABL330, Radiometer America, Westlake, OH).

One hour after stabilizing the body temperature, a 25% mannitol solution was infused to open the BBB. For the normothermic groups, the mannitol solution was warmed to 37°C, filtered, and infused through the catheter in the carotid artery for 30 s at a rate of 0.25 ml · kg⁻¹· s⁻¹. This dose has been reported to produce a reversible BBB disruption without significant neuronal damage. 12For the hypothermic groups, the mannitol solution was warmed to 32°C and filtered before infusion. To determine the Ki, 2 min after mannitol infusion, 20 μCi of ¹⁴C-α-AIB (Amersham, Arlington Heights, IL) was rapidly injected intravenously and flushed with 0.5 ml of normal saline. Blood samples were collected from the femoral arterial catheter at 15-s intervals for the first 2 min and then every minute for the next 8 min. Five minutes after injecting ¹⁴C-AIB, 20 μCi of ³H-dextran (70,000 Da; Amersham) was injected intravenously and flushed with 0.5 ml of normal saline. After collecting the 10-min arterial blood sample, the animals were decapitated, and their brains were quickly frozen in liquid nitrogen. The following brain regions were dissected: ipsilateral cortex (where mannitol was injected), contralateral cortex, pons, and basal ganglia. Brain samples were solubilized in tissue solubilizer (Soluene; Packard, Downers Grove, IL) before counting the radioactivity. Arterial blood samples were centrifuged, and the plasma was separated and processed for scintillation. Plasma and brain samples were counted on a liquid scintillation counter that was equipped for dual-label counting. Quench curves were prepared using carbon tetrachloride, and all samples were automatically corrected for quenching. The blood-to-tissue transfer coefficient for ¹⁴C-AIB was determined, assuming a unidirectional transfer of ¹⁴C-AIB during a 10-min period of the experiment, using the following equation described by Gross et al.  22:

where Am is the amount of ¹⁴C-AIB radioactivity in the tissue per gram, Vp is the volume of plasma retained in the tissue or volume of dextran distribution, Cp(t) is the arterial concentration of ¹⁴C-AIB over time t, and CT is the arterial plasma concentration of ¹⁴C-AIB at the time of decapitation. In this equation, Vp × CT is a correction term that accounts for the ¹⁴C-AIB retained in the vascular compartment of the tissue Am. Vp is determined from the results of the ³H-dextran data and the following equation:

where A′m is the amount of ³H-dextran radioactivity in the tissue per gram, and C′p is the concentration of ³H-dextran in the plasma at the time of decapitation. To calculate the Ki of the ipsilateral cortex, the Vp of the contralateral cortex was used in this study.

A factorial analysis of variance was used to assess the effects of temperature and anesthetics on Ki, plasma volume, and vital signs among the various examined brain regions and groups. Multiple comparisons using a Student-Newman-Keuls test were performed to analyze differences among the groups and the brain regions. The effects of blood pressure on Ki were also evaluated using a correlation analysis. All data are expressed as mean ± SD, and significance was defined as P < 0.05.

Arterial blood pressures were similar among the four experimental groups. Decreasing body temperature did not significantly affect blood pressure during isoflurane or pentobarbital anesthesia. There was no correlation between the change of blood pressure and Ki. Heart rate was significantly lower in the NI than the NP group (P < 0.005). In the pentobarbital-anesthetized rats, decreasing body temperature decreased heart rate (P < 0.001). There was no difference in the heart rate between the HI- and HP-anesthetized rats (table 1).

The Ki of the ipsilateral cortex of the NI- or NP-anesthetized rats was significantly higher than that of the contralateral cortex (NI:P < 0.005; NP:P < 0.0001), where the BBB was not disrupted. The Ki of the ipsilateral cortex of the NI group was significantly lower than that of the NP group (P < 0.05;fig. 1).

Mild hypothermia decreased the Ki of the ipsilateral cortex in both anesthetic groups (NI vs.  HI, P < 0.05; NP vs.  HP, P < 0.05). The Ki was still higher in the ipsilateral than contralateral cortex in rats anesthetized with HI (P < 0.05) and HP (P < 0.005). The Ki of the ipsilateral cortex of the HI group was significantly lower than that of the NP or HP groups (P < 0.001 and P < 0.005, respectively;fig. 1).

In the contralateral cortex, there was no significant difference in the Ki between any of the experimental groups (fig. 1). The Ki of the pons and basal ganglia was similar to that of the contralateral cortex in all experimental groups. The Ki of the pons of the HI group was lower than that of the HP group (P < 0.05). The Ki of the basal ganglia of the HI was lower than that of the HP (P < 0.05) or NI groups (P < 0.05;table 2).

In the normothermic condition, the volume of the dextran distribution of the ipsilateral cortex was greater than those of the rest of the brain regions (contralateral cortex, basal ganglia, and pons) in both isoflurane- and pentobarbital-anesthetized rats (P < 0.05 and P < 0.005, respectively). Between these two groups, there was no statistically significant difference in the volume of dextran distribution in any of the brain regions that were studied (table 3and fig. 2).

In the hypothermic condition, the volume of dextran distribution of the ipsilateral cortex was not significantly different from those of the rest of the brain regions in either of the isoflurane- or pentobarbital-anesthetized rates (table 3and fig. 2). The volume of dextran distribution of the ipsilateral cortex was smaller in the HI-than HP-anesthetized rats (P < 0.05;fig. 2). The value of dextran distribution was not significantly different between these two groups in any of the rest of the brain regions.

Our data demonstrated that the degree of the BBB disruption by hyperosmolar mannitol was less with isoflurane than pentobarbital anesthesia in the normothermic groups. Mild hypothermia decreased the BBB disruption during anesthesia with both anesthetic agents, but the disruption was still less during isoflurane anesthesia.

Hyperosmolar agents are thought to affect the BBB by causing shrinkage of endothelial cells, opening tight junctions, or increasing pinocytotic transfer. 16,17,23,24However, electron microscopy failed to demonstrate any consistent changes to elucidate the mechanisms. 12The degree of osmotic disruption of the BBB could be affected by osmolarity of a substance and duration of its intraarterial infusion, regions of the brain, anesthetic agents, blood pressure, carbon dioxide partial pressure, or use of steroids. 18,19,25–27An excessively high osmolarity or long duration of infusion could cause an irreversible BBB disruption and neuronal damage, resulting in massive cerebral edema. 12In this study, arterial blood pressure and blood gases, including carbon dioxide partial pressure, were similar between the experimental groups. All of the animals in this study were given an infusion of 25% mannitol at the same rate and duration to open the BBB. This dose has been reported to produce a reversible opening of the BBB without serious immediate or delayed neurotoxicity. 12,25 

Because transfer of ¹⁴C-AIB across the BBB is usually independent of cerebral blood flow, it was not determined in this study. 27–29The Ki reflects the product of permeability and the perfused capillary surface area. For ¹⁴C-AIB, the Ki is almost equal to the value of the product. 13,14,28Because of technical difficulties in measuring the perfused capillary surface area, in most studies on BBB, only Ki is measured to reflect BBB permeability, as was the case in this study. ¹⁴C-AIB was used because its molecular size is physiologically relevant to those of other amino acids, ions, or hormones. Once it crosses the BBB, back flow is minimal during the experimental time period. 21,22The plasma volume obtained with dextran in the ipsilateral cortex represents the sum of retained plasma volume and the volume of dextran that leaked into the brain tissue. To calculate the Ki of the ipsilateral cortex, the Vp (volume of dextran distribution) of the contralateral cortex was used to obtain more accurate data.

In the NI group, our data of the Ki of the contralateral cortex where BBB was not disrupted was similar to those of our previous studies. 18,19,30Saija et al.  15demonstrated that the Ki of ¹⁴C-AIB was significantly lower during pentobarbital or ketamine anesthesia than during the awake state. They found that there was no significant differences in the Ki between these two anesthetic conditions. Our data also demonstrated that when the BBB is not disrupted, the transport of ¹⁴C-AIB across the BBB was similar between isoflurane and pentobarbital anesthesia.

The possibility of varying degrees of disruption of the BBB by hyperosmolar solution during different anesthetic agents has been suggested. Evans-blue staining after hyperosmolar BBB disruption was more prominent during pentobarbital and isoflurane anesthesia than during fentanyl and methoxyflurane anesthesia. 16The mean methotrexate transport was less during isoflurane than during pentobarbital anesthesia. 17Our study also demonstrated that the volume of dextran distribution was similar between isoflurane and pentobarbital groups. However, the Ki of a much smaller molecule, ¹⁴C-AIB, was significantly lower in the isoflurane than in the pentobarbital groups. These data suggest that, depending on the molecular size of the indicator, even at the same degree of the BBB disruption, the transport of the indicator would vary. Small molecules would pass through, but large molecules cannot. The mechanism of the variation in the degree of hyperosmolar BBB disruption among anesthetic agents has not been studied systematically. The hemodynamic changes during anesthesia was suggested as one of the mechanisms. 16,18However, in this study, the blood pressure was similar between the pentobarbital and the isoflurane groups despite the difference in the degree of BBB disruption. At least blood pressure does not appear to be the main mechanism of the difference. Anesthetic agents may have different effects on carrier systems. The higher affinity of isoflurane for lipids, which allows its higher dissolution into biologic membranes, may affect the size and shape of endothelial vesicles and intercellular tight junctions. 31,32Perhaps a higher lipid solubility of isoflurane could make the BBB less vulnerable to disruption by hyperosmolar mannitol. 33Rapoport 34demonstrated that the osmolar threshold for the opening of the BBB increased with increasing lipid solubility. During isoflurane or pentobarbital anesthesia, mild hypothermia significantly decreased the transfer coefficient of ¹⁴C-AIB. These data are similar to those of other studies in which mild hypothermia protected the BBB during ischemia and brain injury. 3–6Although the mechanism of BBB disruption could be different between ischemia and hyperosmolar disruption, these data suggest that mild hypothermia also protected the BBB from hyperosmolar disruption.

How does hypothermia attenuate disruption of the BBB? The mechanism is unknown and may involve multiple factors. Mild hypothermia may decrease the activity of carrier systems and stabilize cellular membrane. Hypothermia-induced reduction in permeability of radio-labeled tracer solutions across the BBB has been reported. 35It was suggested that the effects of hypothermia on the BBB were primarily caused by the condition of cerebrovascular membrane.

In vitro  studies have demonstrated that endothelial transport could be temperature-dependent. 36,37During cooling, the membrane lipids may undergo transition from their dynamic fluid state to a more highly ordered gel state, thereby possibly progressively reducing the mobility of membrane proteins and enzyme complexes. 38The physicochemical effects of temperature may affect free diffusion. However, the effect is minimal in the range of the body temperature that was used in our study. 39In this in vivo  study, the body temperature may have influenced other factors affecting diffusion independent of BBB disruption. Biochemical alteration occurring during mild hypothermia, such as reduction in release of neurotransmitters and vasoactive substances, may all affect BBB permeability. 40,41 

At the same body temperature (32°C), the Ki of the ipsilateral cortex of the pentobarbital group was approximately three times that of isoflurane. The volume of dextran distribution that reflects both retained plasma volume and extravasated dextran in the HP group was 2.2 times that of the HI group. These results indicated that isoflurane was more effective in attenuating the BBB disruption by hyperosmolar mannitol during mild hypothermia.

In conclusion, our data demonstrated that the degree of the BBB disruption by hyperosmolar mannitol was significantly smaller in mild hypothermia than in normothermia during both isoflurane and pentobarbital anesthesia. The degree of disruption was less during isoflurane than pentobarbital anesthesia in both normo-thermia and hypothermia. The mechanism is not clear. However, our data demonstrated that blood pressure was not the main mechanism of attenuation of the BBB disruption during isoflurane anesthesia or during mild hypothermia.

The authors thank Patricia A. Sheffield, M.A. (Administrative Director, Department of Anesthesia, University of Medicine and Dentistry of New Jersey–Robert Wood Johnson Medical School, New Brunswick, NJ), for expert editorial assistance.