Potential conflict of interest: Nothing to report.
Supported by The Rigshospitalet Research Foundation, University of Copenhagen, Denmark. Savvaerksejer Jeppe Juhl and Wife Ovita Juhls Foundation.
Intravenous infusion of magnesium sulfate prevents seizures in patients with eclampsia and brain edema after traumatic brain injury. Neuroprotection is achieved by controlling cerebral blood flow (CBF), intracranial pressure, neuronal glutamate release, and aquaporin-4 (Aqp4) expression. These factors are also thought to be involved in the development of brain edema in acute liver failure. We wanted to study whether hypermagnesemia prevented development of intracranial hypertension and hyperperfusion in a rat model of portacaval anastomosis (PCA) and acute hyperammonemia. We also studied whether hypermagnesemia had an influence on brain content of glutamate, glutamine, and aquaporin-4 expression. The study consisted of three experiments: The first was a dose-finding study of four different dosing regimens of magnesium sulfate (MgSO4) in healthy rats. The second involved four groups of PCA rats receiving ammonia infusion/vehicle and MgSO4/saline. The effect of MgSO4 on mean arterial pressure (MAP), intracranial pressure (ICP), CBF, cerebral glutamate and glutamine, and aquaporin-4 expression was studied. Finally, the effect of MgSO4 on MAP, ICP, and CBF was studied, using two supplementary dosing regimens. In the second experiment, we found that hypermagnesemia and hyperammonemia were associated with a significantly higher CBF (P < 0.05, two-way analysis of variance [ANOVA]). Hypermagnesemia did not lead to a reduction in ICP and did not affect the brain content of glutamate, glutamine, or Aqp-4 expression. In the third experiment, we achieved higher P-Mg but this did not lead to a significant reduction in ICP or CBF. Conclusion:Our results demonstrate that hypermagnesemia does not prevent intracranial hypertension and aggravates cerebral hyperperfusion in rats with PCA and hyperammonemia. (HEPATOLOGY 2011;)
Acute liver failure (ALF) is a condition with a substantial mortality rate because of a high risk of multiple organ failure. Of special interest are the cerebral complications in ALF that in the most severe cases can progress to cerebral edema, intracranial hypertension, and ultimately cerebral incarceration. The genesis for these cerebral complications is among other factors related to persistent hyperammonemia,1, 2 systemic inflammation,2, 3 and electrolyte disturbances4 and is associated with cerebral hyperperfusion and loss of autoregulation of the cerebral blood flow (CBF).5 The complex and multifactorial nature has made the understanding of the pathogenesis of brain edema difficult.
Both the therapeutic and prophylactic strategies related to brain edema in ALF are few and limited in efficacy. During surges of high intracranial pressure (ICP), intervention with mannitol, hypertonic saline, hyperventilation, hypothermia, and indomethacin has demonstrated a temporary beneficial effect on intracranial hypertension.6, 7 Recently, ammonia-lowering therapy with the compound of L-ornithine and phenylacetate has shown promising data in animal studies,8 but no human data have yet been published. Ammonia-lowering therapy with the amino acid compound L-ornithine L-aspartate has shown disappointing results in a large randomized study of patients with ALF.9 Because stabilization of the patients during spontaneous recovery or bridging to transplantation currently is a complicated task with a high risk of fatal outcome, the introduction of new ways to secure brain viability in ALF is required.
Intravenous magnesium sulfate (MgSO4) was introduced as an anticonvulsant for pregnant women with eclampsia more than 80 years ago.10 More recently, animal models of ischemic and traumatic brain injury have shown neuroprotective features of systemically administered MgSO4 seen as a reduction in brain edema,11 tissue damage,12 and metabolic derangement.13 Magnesium sulfate may act as a neuroprotector by reducing the extracellular release of the excitatory neurotransmitter glutamate, down-regulating the expression of the water channel Aquaporin-4 (Aqp4), blocking induction of mitochondrial permeability transition, and attenuating generation of free radicals—all pathogenic mechanisms also thought to be involved in the cerebral complications of ALF.16-18
The use of hypermagnesemia as a neuroprotectant in ALF has not been studied. Therefore, we decided to evaluate the effect of hypermagnesemia, achieved by administration of MgSO4 on ICP and CBF with three different dosing regimens. We used a well-established rat model of hepatic encephalopathy and brain edema induced by acute hyperammonemia after construction of a portacaval anastomosis (PCA).19 Our study consists of the following experiments:
Experiment A: Dose-finding study in healthy rats
Experiment B: Study of the effect of hypermagnesemia on ICP and CBF achieved by two doses of MgSO4 on rats with PCA and hyperammonemia
Experiment C: Study of the effect of hypermagnesemia on ICP and CBF using two additional dosing regimens of MgSO4 on rats with PCA and hyperammonemia
In experiment B, we also wanted to study the potential mechanisms of action, and we therefore measured the cortical content of glutamate, glutamine, and the expression of Aqp4 in the experimental groups receiving ammonia infusion and either MgSO4 or vehicle injections.
We hypothesized that hypermagnesemia would reduce ICP, CBF, cortical content of glutamine, and the expression of Aqp4 as well as partially restore cortical glutamate levels.
ALF, acute liver failure; ANOVA, analysis of variance; Aqp4, aquaporin-4; CBF, cerebral blood flow; ICP, intracranial pressure; MAP, mean arterial pressure; PCA, portacaval anastomosis; P-Mg, total plasma magnesium concentration.
Materials and Methods
All procedures involving laboratory animals were conducted in accordance with the European Communities Council Directive of 24 November 1986 (86/609/EEC) and approved by the Danish Animal Experiments Inspectorate. The experiments were carried out in the animal facilities associated with the Hepatology Laboratory, Rigshospitalet, Copenhagen, Denmark.
Male Wistar rats (Charles River, Sulzfeld, Germany) were housed in plastic cages with free access to water and rodent chow and kept at constant room temperature and humidity with a 12/12-hour light/dark cycle.
Experiment A included 17 healthy anesthetized animals divided into the following groups:
1.A single dose of 1.6 mmol/kg MgSO4 intraperitoneally at t = 0 (n = 4)
2.Two intraperitoneal doses of MgSO4: 1.6 mmol/kg at t = 0 and 0.8 mmol/kg at t = 1 hour (n = 6)
3.Three intraperitoneal doses of MgSO4: 1.6 mmol/kg at t = 0, 0.8 mmol/kg at t = 1 hour, and 0.8 mmol/kg at t = 2 hours (n = 3)
4.Intravenous infusion of MgSO4: 0.8 mmol/kg intravenously over 10 minutes and at t = 30 minutes continuous infusion of 0.8 mmol/kg/hour IV for 210 minutes (n = 4)
In experiment B, we used an intraperitoneal double-dosing regimen of MgSO4 with 1.6 mmol/kg at t = 0 and 0.8 mmol/kg at t = 1 hour and included 24 rats divided into four groups:
1.PCA + ammonia infusion + vehicle (n = 7)
2.PCA + ammonia infusion + MgSO4 (n = 6)
3.PCA + saline infusion + vehicle (n = 5)
4.PCA + saline infusion + MgSO4 (n = 6)
Experiment C included 12 rats divided into two groups receiving MgSO4 by either an intraperitoneal triple dosing regimen (1.6 mmol/kg at t = 0, 0.8 mmol/kg at t = 1 hour, and 0.8 mmol/kg at t = 2 hour) or IV infusion (0.8 mmol/kg IV over 10 minutes and at t = 30 minutes continuous infusion of 0.6 mmol/kg/hour IV for 210 minutes):
Groups 1 and 2 in experiment C were compared with groups 1 and 2 in experiment B.
Surgical Procedure (Experiment B+C).
The PCA was done as an end-to-side anastomosis. In isoflurane anaesthesia, the rats underwent laparotomy. The portal vein and vena cava were isolated, and after the portal vein was ligated and cut, the distal part was sutured onto a hole in the side of the vena cava. The anastomosis was completed in less than 15 minutes, and the abdomen was sutured in two layers. Buprenorphine was given intramuscularly as postoperative analgesic. The animals then returned to their housing, and the actual experiment started 24 hours later.
Experimental Procedure (Experiment B+C).
After induction of anesthesia with isoflurane, 0.2 to 0.3 mL pentobarbital (50 mg/mL) was administered in a tail vein. Every 10 minutes, the reaction to claw pinching was checked and supplementary pentobarbital given if necessary. Arterial and venous catheters (PE-50) were inserted in femoral vessels for monitoring blood pressure, intravenous drug administration, and blood sampling. The arterial catheters were flushed with 500 IU heparin and one connected to a pressure transducer. The rats were then tracheotomized and mechanically ventilated (Hallowel EMV, E-vet, Haderslev, Denmark) with a respiratory frequency of 65 breaths/minute and a tidal volume of 5 to 10 mL adjusted to maintain physiological arterial partial pressure of oxygen and partial pressure of carbon dioxide. The animals were placed in a stereotactic frame and, with the head fixated, a midline scalp incision was made and two small boreholes were drilled in the skull. One borehole was used for the placement of a catheter (PE-10) in the cisterna magna for ICP monitoring by connection to a pressure transducer. In the second borehole, we placed a laser Doppler probe (Probe 407; Perimed, Stockholm, Sweden) on the surface of the brain. The probe was connected to a Periflux Laser Doppler System 5000 monitor, allowing us to measure arbitrary values of blood velocity for later calculation of relative changes in CBF. Continuous measurements of mean arterial pressure (MAP), ICP, and blood velocity were recorded on a computer using the software Perisoft (Perimed, Stockholm, Sweden). During the experiment, body temperature was monitored with an intraabdominal thermistor and maintained at 37°C with a heating blanket. Arterial blood samples were taken regularly and pO2 and pCO2 analyzed (ABL 505; Radiometer, Copenhagen, Denmark). After an initialization period, stable baseline values were recorded, and intravenous ammonium acetate infusion 55 μmol/kg × min or saline infusion 2 mL/hour (0.9 mg/mL) was initiated (t = 0). Hypermagnesemia was induced in the appropriate groups by administration of MgSO4 as stated previously. The control groups in experiment B received an equal volume of saline (vehicle) intraperitoneally
Arterial blood samples were taken at t = 2 hours and t = 4 hours for measurement of total plasma magnesium (P-Mg), which were determined on a Roche Modular P analyzer with the use of colorimetry according to the manufacturer's instructions. In addition, plasma levels of alanine aminotransferase, coagulation factors II+VII+X (PP), and ammonia were determined at t = 4 hours. The experiment was ended after 4 hours, and the animals were sacrificed while anesthetized. After decapitation, cerebral cortex was removed from the brain and immediately frozen in liquid nitrogen and stored at −80°C for later analysis of glutamate, glutamine, and Aqp4 content.
Analysis of Cortical Glutamate and Glutamine Content (Experiment B).
The cerebral cortical tissue was weighed and homogenized in a sixfold amount of ice-cold 1 mol/L HClO4. The homogenate was centrifuged and the supernatant neutralized by ice-cold 1.6 mol/L KOH containing 0.4 mol/L K2CO3. The concentrations of glutamate and glutamine were then measured in the supernatant by an enzymatic method using a YSI 2700 (YSI, Yellow Springs, OH), and the actual cortical concentration in the unit mmol/100 g could then be calculated.
Analysis of Aqp4 Protein Expression by Western Blotting (Experiment B).
Frozen cortical brain tissue was homogenized in a Potter Elverhjem (B. Braun, Melsungen, Germany) at high speed for 4 minutes on ice in dissection buffer containing 0.32 M sucrose, 50 mM 4-(2-hydroxyethyl)-1-piperazine ethanesulfonic acid buffer (pH 7.4) (Invitrogen Tåstrup, Denmark), two mM ethylenediaminetetra-acetic acid (Invitrogen), and one dissolved tablet of protease-inhibitor cocktail completeMINI (Roche Diagnostics, Copenhagen, Denmark). The homogenate was centrifuged in an Eppendorf 5415c centrifuge (Eppendorf, Hamburg, Germany) at 4000 g for 15 minutes at 4°C to remove whole cells, nuclei, and mitochondria. Subsequently, the supernatant was centrifuged at 200,000 g in a Beckmann 50.3 Ti Centrifuge (Beckman Coulter, Fullerton, CA) for 30 minutes to produce a membrane-enriched pellet.20 The resultant pellet was resuspended in 50 mM 4-(2-hydroxyethyl)-1-piperazine ethanesulfonic acid buffer, 2% sodium dodecyl sulfate and one dissolved tablet of protease-inhibitor cocktail, for 2.5 hours at 20°C. The protein concentration was measured using BIORAD DC Kit (Bio-Rad Laboratories, Copenhagen, Denmark) and a photometer Beckman Coulter DU730 (Ramcon, Birkerød, Denmark). The membrane-enriched protein samples were loaded onto 4% to 12% Invitrogen mini-cell-system (Invitrogen) (150 V, 50 minutes) with 3 μg protein per lane. SeeBlue PLUS 2 (Invitrogen) standard marker was also loaded. No heating before sample loading was performed, and all procedures were performed under denaturing conditions. Protein was transferred to a polyvinylidene fluoride membrane (Invitrogen) by electroelution (30 V, 60 minutes, 20°C). After blocking with 5% low-fat milk in Tris-buffered saline/Tween-20 (10 mM Tris-HCl, 150 mM NaCl, 0.1% Tween 20, pH 7.4) for 1 hour, the polyvinylidene fluoride membrane was incubated overnight at 4°C with the primary antibody diluted in 5% low-fat milk and Tris-buffered saline/Tween-20 buffer (Polyclonal antibody, SC9888, 1:5000, Santa Cruz Biotechnology, Heidelberg, Germany). The membrane were subsequently washed in Tris-buffered saline/Tween-20 buffer for 30 minutes and then incubated at room temperature for 2 hours with horseradish peroxidase–conjugated secondary antibody (SC2020, 1:5000; Santa Cruz Biotechnology) diluted in the same buffer. After incubation, the membrane was washed with Tris-buffered saline/Tween-20 for 30 minutes. Finally, detection of bound antibody was performed using the enhanced chemiluminescence system (PerkinElmer, Waltham, MA) and camera detecting system LAS 9000 with software ImageGauge 2006 Software (FujiFilm, Stockholm, Sweden).
Analysis of mRNA Expression by Dot Blotting (Experiment B).
Total RNA was isolated with RNeasy mini lipid kit (Qiagen Sciences, Gaithersburg, MD). The total amount of RNA in the samples was measured using a photometer Beckman Coulter DU730 (Ramcon). Messenger RNA quantification of Aqp4 was performed by dot-blot analysis with specific complementary DNA probes against rat Aqp4 (Primers: Aqp4 forward: 5′ ccccccagcgtggtgggaggattggg 3′, Aqp4 reverse: 5′ gccagcacagcgcctatgattggtccaaccc 3′). The 32PdCTP-labeling of the Aqp4 complementary DNA probe was performed with the in vitro transcription, using a Maxiscript in vitro transcription kit (Amersham Biosciences, Hillerød, Denmark) followed by purification on NICK Spin Columns (Stratagene, La Jolla, CA). Five micrograms total RNA was immobilized onto polyvinylidene fluoride nylon membranes using a Schleicher & Schuell Minifold SRC 072/0 (Schleicher & Schuell, Dassel, Germany). Prehybridization was performed at 68°C for 30 minutes followed by hybridization with the probe at 68°C for 1 hour in Quick Hybridisation Solution (Boehringer Mannheim, Ingelheim, Germany). Subsequently, the membrane was washed at 25°C in 2× saline sodium citrate (NaCl + Na3Citrate) and 0.1% sodium dodecyl sulfate for 15 minutes and second at 68°C in 0.1× saline sodium citrate and 0.1% sodium dodecyl sulfate for 30 minutes. Signals were detected using Imaging Plate BASIII, and the hybridization signal was analyzed by FUJI FLA 9000 STARION (FujiFilm). The signals were densitometrically evaluated using ImageGauge 2006 software (FujiFilm).
All results are presented as means ± standard error of the mean. In experiment B, two-way analysis of variance (ANOVA) was used to evaluate the individual effects for ammonia infusion and hypermagnesemia on the biochemical parameters as well as MAP, ICP, and relative CBF at the end of the experiment. A paired Student t test was used to evaluate changes in ICP and CBF between baseline and the end of the experiment. Differences in Aqp4 expression, glutamine, and glutamate between groups 1 and 2 in experiment B were evaluated by the Student t test. Comparison between groups in experiments A and C was done using one-way ANOVA and Tukey's test for post hoc analysis. P values below 0.05 were considered significant.
A single intraperitoneal dose of MgSO4 (1.6 mmol/kg) gave an inadequate increase of P-Mg at t = 2 and 4 hours (Table 1). Adding a second dose of MgSO4 intraperitoneally (0.8 mmol/kg) after 1 hour (group 2), we found a P-Mg at 2 hours above 2 mM but less than 2 mM after 4 hours. Adding a third dose of MgSO4 at t = 2 hours (0.8 mmol/kg), the P-Mg at t = 4 hours further increased but was still less than 2 mM. In the intravenous infusion group (group 4), we found a P-Mg at t = 4 hours above 3 mM and significantly higher than the P-Mg in the other groups. The magnesium concentration in the cerebrospinal fluid at t = 2 and t = 4 hours did not increase significantly with increased doses of MgSO4 but was 13% to 33% above baseline at t = 4 hours.
Table 1. Concentration of Magnesium in Plasma and Cerebrospinal Fluid Measured in Experiment A
T = 2 Hours
T = 4 Hours
T = 2 Hours
T = 4 Hours
Data presented as means ± standard error. Group 1 dosing: 1.6 mmol/kg MgSO4 at t = 0. Group 2 dosing: 1.6 + 0.8 mmol/kg of MgSO4 at t = 0 and t = 1 hour. Group 3 dosing: 1.6 + 0.8 + 0.8 mmol/kg MgSO4 at t = 0, t = 1, and t = 2 hours. Group 4 dosing: 0.8 mmol/kg bolus + continuous infusion of 0.8 mmol/kg/hour MgSO4.
Significantly different from group 1, P < 0.01.
Significantly different from group 1, P < 0.05 (one-way ANOVA, Tukey's test).
Significantly different from group 1, P < 0.001.
Significantly different from groups 1, 2, and 3, P < 0.001.
No significant differences were found at baseline between groups with regard to mean animal weight, arterial pH, or pCO2 (Table 2). Both groups of rats with ammonia infusion (groups 1 and 2) had significantly higher plasma levels of ammonia and alanine aminotransferase and also a lower PP compared with rats receiving saline infusion. Ammonia infusion also led to a significantly lower MAP after 1 hour of ammonia infusion (group 1: 83.0 ± 3.8 mm Hg, group 2: 82.1 ± 4.2 mm Hg, group 3: 98.6 ± 2.2 mm Hg, group 4: 92.2 ± 2.9 mm Hg (F[1,21] = 12.6, P < 0.01, two-way ANOVA), but at the end of the experiment MAP did not differ significantly (group 1: 95.1 ± 10.5 mm Hg, group 2: 89.7 ± 6.2 mm Hg, group 3: 93.2 ± 2.7 mm Hg, group 4: 90.7 ± 2.6 mm Hg, NS). An insignificant effect was seen for hypermagnesemia on blood pressure, where groups receiving MgSO4 had a lower MAP than groups receiving vehicle.
Table 2. Baseline Data, Magnesium Levels, and Liver-Related Blood Biochemistry of Experiment B
Mean P-Mg were not significantly different between the groups receiving MgSO4 at t = 2 hours (P = 0.53) or t = 4 hours (P = 0.12). Conversely, we found a significant drop in P-Mg in both groups from t = 2 to t = 4 hours (PCA + ammonia infusion + MgSO4: 0.85 mM [P < 0.05] and PCA + ammonia infusion + MgSO4: 0.79 mM [P < 0.01]) (Table 2).
Intracranial Pressure and Cerebral Blood Flow.
After 4 hours of ammonia infusion, we observed a significant increase in ICP in groups 1 and 2 (from 1.6 ± 0.5 to 7.8 ± 1.1 mm Hg (P < 0.001, paired t test) and 1.7 ± 0.3 to 9.4 ± 2.2 mm Hg (P < 0.05), respectively). Likewise, the relative CBF increased significantly from baseline (100%) to 174% ± 24% and 241% ± 34% in groups 1 and 2 (P < 0.05 and P < 0.01), respectively. Two-way ANOVA analysis revealed that ICP increased significantly in groups receiving ammonia infusion (F[1,21] = 18.5, P < 0.01) (Fig. 1A) but was not affected significantly by hypermagnesemia. Conversely, both hyperammonemia and hypermagnesemia aggravated the changes in the relative CBF significantly (F[1,21] = 14.3, P < 0.01 and F[1,21] = 5.3, P < 0.05, respectively) (Fig. 1B). No significant interactions were seen between ammonia and hypermagnesemia on ICP or CBF.
Brain Cortex Glutamate and Glutamine in Groups 1 and 2.
The glutamate concentration (mmol/100 g) was 1.20 ± 0.12 in group 1 and 1.23 ± 0.12 in group 2, (NS, unpaired t test). The glutamine concentration (mmol/100 g) was 3.25 ± 0.19 in group 1 and 3.34 ± 0.07 in group 2 (NS) (Fig. 1C)
Expression of Aqp4 mRNA and Protein in Brain Cortex in Groups 1 and 2.
No significant differences were seen in the expression of Aqp4 mRNA between groups 1 and 2, and likewise no significant difference was seen in the protein level (Fig. 1D).
No significant differences were found at baseline between groups in regards to mean animal weight, arterial pH, partial pressure of carbon dioxide, alanine aminotransferase, ammonia, or PP (Table 3). In the triple dosing group, P-Mg was 2.59 ± 0.17 mM at t = 2 hours and 2.26 ± 0.30 mM at t = 4 hours. Based on the results from experiment A, we reduced the dosing in the intravenous infusion group from 0.8 to 0.6 mg/kg/hour as we targeted a P-Mg of above 2.0 mM, but not higher than 3.0 mM, at t = 4 hours. With this dose, P-Mg was found to be 2.27 ± 0.14 mM at t = 2 hours and 2.64 ± 0.26 mM at t = 4 hours. Compared with the magnesium levels achieved in group 2 of experiment B, we only found significantly higher levels at t = 4 hours in the intravenous infusion group (P < 0.05, Tukey's test on one-way ANOVA [F(2,15) = 5.42]).
Table 3. Baseline Data, Magnesium Levels, and Liver-Related Blood Biochemistry of Experiment C
At t = 4 hours, MAP was 78.7 ± 3.8 mm Hg in the triple dosing group and 94.6 ± 5.4 mm Hg in the intravenous infusion group. Compared with group 1 from experiment B, we found no significant differences (F[2,16] = 1.45, P = 0.26, one-way ANOVA). At t = 4 hours, ICP was 5.44 ± 0.8 mm Hg in the triple dosing group and 6.60 ± 1.6 mm Hg in the intravenous infusion group. Compared with group 1 from experiment B, we found no significant differences (F[2,16] = 0.99, P = 0.39, one-way ANOVA) (Fig. 2A). At t = 4 hours, the relative CBF was 228% ± 49% in the triple dosing group and 169% ± 22% in the intravenous infusion group. Compared with group 1 from experiment B, we found no significant differences (F[2,16] = 0.95, P = 0.41, one-way ANOVA) (Fig. 2B).
Based on local clinical recommendations for hypermagnesemia and relevant literature,21 we aimed at achieving a P-Mg > 2 mM after 2 hours. We also aimed at a CSF-Mg level of greater than 20% above baseline after 4 hours. The results of the dose finding study in experiment A demonstrated that groups 2, 3, and 4 achieved a P-Mg greater than 2 mM at t = 2 hours and that group 4 also had a P-Mg greater than 2 mM at t = 4 hours. Interestingly, we did not find that increased doses of magnesium led to higher CSF-Mg levels at t = 4 hours, even though P-Mg was higher. We concluded that the most appropriate dosing regimen was 1.6 mmol/kg MgSO4 intraperitoneally at baseline and 0.8 mmol/kg MgSO4 intraperitoneally after 1 hour, which fulfilled our criteria of acceptable plasma levels and relevant central nervous system bioavailability. Consequently, this dosing regimen was used in experiment B, in which we, apart from MAP, ICP, and CBF, also studied the potential mechanisms of action of hypermagnesemia.
Experiment B showed that induction of hypermagnesemia did not prevent development of intracranial hypertension or cerebral hyperperfusion. We used the well-characterized rat model with PCA and acute hyperammonemia22 and did indeed find cerebral hyperfusion and high ICP. In fact, with the dosing regimen of experiment B, we saw a higher CBF and a tendency toward a higher ICP in rats treated with MgSO4 compared with the corresponding group not receiving MgSO4.
Another interesting finding was that hyperammonemia led to a significant drop in MAP after 1 hour of ammonia infusion and that hypermagnesemia in both ammonia and saline infusion animals led to a tendency torward lower MAP; however, it was most pronounced in rats with hyperammonemia. This could indicate that the model itself (PCA + ammonia infusion) induces an initial vasodilatation that is worsened by hypermagnesemia and adds up to the substantial increase in relative CBF that was almost 50% higher in the PCA+NH3+MgSO4 group compared with PCA+NH3+vehicle.
To reduce the risk of false-negative results, we also performed experiment C with alternative dosing regimens of MgSO4. The PCA rats appeared to have a lower clearance of MgSO4 than the healthy animals in experiment A, most likely because of the hepatic shunt. We achieved a P-Mg above 2 mM at 2 hours and at the end of the experiment with both a triple-dosing regimen and intravenous infusion. This did not, however, lead to significant positive effects on ICP or CBF. We did observe a tendency toward a slightly lower ICP in the triple dosing group than in the vehicle group. This group, however, also had a markedly lower MAP at the end of the experiment, a circumstance that most likely affected the ICP as well as the CBF that also appeared higher in this group.
Taken as a whole, our findings are in conflict with our hypothesis of hypermagnesemia being a neuroprotectant. An explanation could be that the cerebral and systemic effects of hypermagnesemia superseded the theoretical neuroprotective effects or that the expected positive effect of hypermagnesemia might have been masked by postoperative stress after the PCA surgery. In addition, our study draws the attention to the fact that a systemic route of administration in combination with limited central nervous system bioavailability is making the use of hypermagnesemia as a neuroprotectant problematic,21 unless it is used in situations with cerebral vasoconstriction or reduced CBF.
Regarding the speculated specific mechanisms of actions of hypermagnesemia, we did not see any significant effects. We found it relevant to investigate the cortical levels of glutamate and glutamine and whether hypermagnesemia could influence the shift of the cerebral pool of glutamate toward glutamine, for two reasons: Hyperammonemia is known to heavily influence the glutamatergic neurotransmission18 and leads to acceleration of cerebral detoxification of ammonia by astrocyte glutamine synthesis from amidation of neuronal glutamate and ammonia.23 Also, others have reported that hypermagnesemia attenuates the excitatory release of neuronal glutamate,13 which would lead to lower glutamine levels and higher glutamate levels. Recent studies of the water channel Aqp4 indicate that Aqp4 has a role in the pathogenesis of brain edema in ALF models, although the up-regulation seems to be posttranslational.17, 24 We found that hypermagnesemia did not affect the messenger RNA or protein expression of Aqp4. This observation is in concordance with a study that found that hypermagnesemia did not affect cortical Aqp4 protein expression in a model of brain edema involving hypertensive pregnant rats,25 but rather is in contrast to a study that found that hypermagnesemia gave a restoration of cerebral Aqp4 immunoreactivity after traumatic brain injury.14
In conclusion, our results demonstrate that hypermagnesemia does not prevent intracranial hypertension and aggravates cerebral hyperperfusion in hyperammonemic rats. In our study, the effect of hypermagnesemia suggests a systemic and cerebral vasodilation that superseded the speculated beneficial effects on blood–brain barrier permeability and excitatory neurotransmission. We therefore recommend that the use of magnesium sulfate in patients with ALF be limited to cases with evidence of clinically significant hypomagnesemia or critical low cerebral perfusion.
The authors thank the laboratory animal technicians Bjørg Krogh and Mie Poulsen for their skillful and excellent work.