An official website of the United States government
NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.
Vink R, Nechifor M, editors. Magnesium in the Central Nervous System [Internet]. Adelaide (AU): University of Adelaide Press; 2011.
The rationale for trialing magnesium as a neuroprotective agent following cerebral ischemia has been based both on its role in maintaining brain tissue homeostasis and on its known cellular actions that are likely to counteract damaging ischemic processes. A number of studies using animal models of cerebral ischemia, seizure, perinatal hypoxia-ischemia, subarachnoid hemorrhage and traumatic brain injury have reported positive outcomes with magnesium therapy. However, scrutiny of the animal cerebral ischemia data shows that about 46% of studies have not shown a neuroprotective effect. Furthermore, the IMAGES clinical trial found magnesium to be largely ineffective in treating strokes. In this review, we present the majority of published cerebral ischemia animal studies (focal and global) that have used magnesium as a neuroprotective therapy, and discuss the possible reasons for the inconsistent results. Our examination suggests that, in the majority of experiments, post-ischemic hypothermia has probably been a confounding factor in producing the positive outcomes. In addition, experimental design has not always been appropriate with respect to magnesium dosage, and to the time and route of magnesium administration. Moreover, data from our own laboratory indicates that magnesium is only neuroprotective when combined with post-ischemic hypothermia. Finally, additional information regarding the efficacy of magnesium as a stroke treatment will be available on completion of the FAST- Mag trial, but in the meantime the neuroprotective potential of magnesium should be explored when combined with post-ischemic hypothermia, and potentially with other agents, in cerebral ischemia models.
More than any other organ, the brain is dependent on the aerobic metabolism of glucose for energy, making it exceedingly sensitive to ischemic disease. Inadequate supply of oxygen- ated blood leads to a rapidly developing energy crisis that, within minutes, results in the death of cells comprising the neurovascular unit. Restoration of blood supply, or the presence of collateral circulation, may preserve some tissue only for it to die over the hours or days that follow an ischemic event. Such delayed death may occur from one or more of a range of pathological processes; glutamate excitotoxicity, calcium dysregulation and overload, apoptosis, autophagy, inflammation, oxygen free radical generation and edema have all been implicated. Consequently, interruption of these processes presents an opportunity for therapeutic intervention to protect the damaged cells within the neurovascular unit after cerebral ischemia. Many therapies have been found to offer neuroprotection in experimental models of cerebral ischemia, magnesium among them.
Since magnesium is important in many normal cellular processes such as protein synthesis, energy metabolism, maintenance of ionic gradients and vascular smooth muscle tone, abnormalities in serum and CNS magnesium status before, during and after cerebral ischemia can be expected to have an effect on neurological outcome. Indeed, clinical and experimental studies have shown that subjects with low serum or CSF magnesium have worsened neurological outcomes following brain ischemia and trauma (Lampl et al., 1998; McIntosh et al., 1988; Vink et al., 1988). There is also ample evidence demon- strating that there are marked changes in intracellular and extracellular free magnesium concentrations in brain tissue following ischemic and traumatic insults (Helpern et al., 1993; Vande Linde and Chopp, 1991; Vink et al., 1996; Lee et al., 2002). As such, maintaining normal serum magnesium concentration should be considered a part of standard supportive care in such patients. On top of that, however, the Mg2+ cation has properties that may specifically counteract a number of the damaging process associated with cerebral ischemia. In particular, magnesium can reduce synaptic glutamate release (Nowak et al., 1984), decrease calcium influx via voltage gated calcium channels (Iseri and French, 1984), stabilise mitochondria (Kowaltowski et al., 1998; Xu et al., 2002), increase cerebral blood flow (Chi et al., 1990) and reduce free radical production (Garcia et al., 1998).
For these reasons, restoration of magnesium homeostasis and exploration of the potential positive effects in the ischemic brain have prompted the investigation of magnesium as a neuroprotective agent following such disorders as traumatic brain injury (Heath and Vink, 1996; Vink et al., 1988), seizure (Cotton et al., 1993), subarachnoid hemorrhage (Van den Bergh et al., 2002) and cerebral ischemia (Muir, 1998). Here we overview the experimental animal studies that have to date tested magnesium as a neuroprotectant in the major animal models of cerebral ischemia, specifically focal (ischemic stroke) and global (cardiac arrest) brain ischemia models.
This review will present experimental findings from these studies, offer potential explanations for conflicting or inconsistent results and consider the future application of magnesium in cerebral ischemia.
The results from 16 studies using magnesium in animal models of focal cerebral ischemia are summarized in Table 1. As can be seen in this table, there is considerable variability in study design (stroke model, animal species used, magnesium dose, route and timing of admin- istration, outcome measures), which precludes direct comparison of outcomes. There were 5 studies using permanent focal ischemia models, 9 using transient models and single studies using an embolic and an endothelin-1 (lacunar stroke) model. As a broad summary, 10 of the 16 studies reported a neuroprotective effect for magnesium based on infarct volume reduction. Magnesium treatment in the endothelin-1 model did not reduce infarct volume, but did improve some behavioral outcomes. Thus, there is slightly better than equivocal support for the claim that magnesium has a neuroprotective effect in these models.
Summary of animal studies using magnesium following focal cerebral ischemia
In the 5 studies using permanent middle cerebral artery occlusion (MCAO), 3 reported positive outcomes with magnesium treatment. Two of the studies, 1 using rats (Izumi et al., 1991) and 1 using mice (Roffe et al., 1996), were otherwise similar in their experimental procedures, yet reported contrasting results. Izumi et al., (1991) observed both a neuroprotective effect and a hyperglycemia associated with administration of magnesium chloride; if the hyperglycemia was managed with insulin, there was a further reduction of infarct volume. In contrast, Roffe et al., (1996) found no neuroprotection, and in fact reported that magnesium chloride treatment increased edema in the infarcted hemispheres. While increased edema was not evident in animals treated with magnesium and insulin, the combined treatment did not reduce infarct volume.
Two other studies (Lee et al., 1999; Chung et al., 2004), which were similar with respect to magnesium dose and timing of administration, but different in animal species used and route of administration, reported positive findings, while a study from our laboratory was negative (Campbell et al., 2008a). Among these, our study was the only trial to use a dosing regimen (IV loading dose and 24 hour infusion) similar to the IMAGES human stroke trial. Furthermore, we paid careful attention to body temperatures post- surgery to ensure that animals did not become hypothermic during the recovery period or thereafter. The possibility of post-ischemic hypothermia having confounded these experimental results, especially in the earlier animal studies, will be discussed in a later section.
It is of interest that Lee et al., (1999) commented that, in unpublished experiments, if the intra- arterial route of magnesium administration was changed to an intravenous route, magnesium was no longer protective.
In the 10 studies using transient focal ischemia, 6 reported positive outcomes with magnesium treatment. As in the permanent focal ischemia studies, some of the trials, while similar in key respects, reported contrasting results. For example, Marinov et al., (1996) reported positive findings when either of 2 doses of magnesium sulphate (0.75 or 0.25 mmol/kg) were administered intra-arterially before 2 or 1.5 hours of MCAO in rats. In contrast, Zhu et al., (2004a) reported negative findings in trials consisting of different doses of magnesium sulphate (0.18, 0.36, 0.72 or 0.74 mmol/kg) administered either intra-arterially or intravenously before 45 minutes or 2 hours of MCAO in rats. In a subsequent experiment from the Zhu laboratory (Campbell et al., 2008b), rats administered 0.36 mmol/kg magnesium intravenously before MCAO, and allowed to self-regulate their body temperatures post-ischemia, experienced up to 4 hours of mild hypothermia and had significantly reduced striatal infarct volumes.
Similarly, Lee et al., (2005) observed a modest, magnesium-induced hypothermic effect in that magnesium treatment attenuated the post- ischemic hyperthermia normally observed in their focal model. When administered immediately after reperfusion, magnesium reduced infarct volumes and improved behavioral outcomes in rats. They ruled out magnesium’s hypothermic effect as being responsible for the neuroprotection by inducing the same level of hypothermia in a separate group of animals, showing that these animals did not have either reduced infarct volumes or improved behavioral outcomes.
The Schmid-Elsaesser laboratory has been exploring the efficacy of magnesium following cerebral ischemia, alone and in combination with other treatments (e.g., tirilazad, hypothermia), dating back to 1999 (Schmid-Elsaesser et al., 1999; Zausinger et al., 2003ab; Westermaier et al., 2003; 2005). To summarize their several studies, they have shown that when magnesium is administered before, during or after MCAO it results in infarct volume reductions ranging from 25 - 42%, with reductions of >31% being statistically significant. After surgery, they routinely monitored animals’ body temperatures for 1 hour after reperfusion and kept them in a warm environment for the first eight hours (though without further monitoring) to minimize the possibility of hypothermia (Robert Schmid- Elsaesser, personal communication).
Yang et al., (2000) assessed the neuroprotective efficacy of intravenously administered magnesium at different time points following ischemia (2, 6 or 8 hours) using an embolic stroke model in rats. They found that magnesium treatment administered 2 or 6 hours, but not 8 hours, after ischemia significantly reduced infarct volumes. In addition, magnesium treatment appeared to improve animal survival and neurological outcome. While animals’ body temperatures were monitored during surgical recovery, it was not reported what measures were taken to control their temperatures or to ensure they did not become hypothermic until the 72-hour trial end-point.
To assess the effects of magnesium on white matter injury associated with lacunar stroke, Lecrux et al., (2008) injected the vasoconstrictive peptide endothelin-1 into the internal capsule in rats (the study was prompted by post-hoc analysis of the IMAGES stroke trial data, which revealed that magnesium provided a small, but significant benefit in lacunar stroke patients (IMAGES, 2004)). In this study, relatively high subcutaneous doses of magnesium sulphate were given, which increased magnesium serum levels nearly four fold and reduced blood pressure. The treatment did not influence infarct volumes, but motor functional impairments were reduced in magnesium treated animals when assessed at 3 and 10 days.
The results from 9 studies using magnesium in models of global cerebral ischemia are summarized in Table 2. As in the focal ischemia studies, there is considerable variability in study design and protocols. At the histological level, 4 of the studies found a neuroprotective effect, 2 studies did not, while 2 studies reported a positive outcome only when magnesium treatment was combined with post-ischemic hypothermia. One study, using an aortic occlusion model in dogs, reported improved neurological outcomes with magnesium treatment. Again, this is moderate support for the contention that magnesium is neuroprotective in these models.
Summary of animal studies using magnesium following global cerebral ischemia
In the first study, by Blair et al., (1989), intravenous magnesium chloride appeared in fact to exacerbate CA1 neuronal injury. In this study, the magnesium dose was relatively high (5 mmol/kg), and was considered responsible for elevating serum glucose levels from 150 mg/dl to
220 mg/dl, and thus contributing to increased CA1 neuronal loss. When hyperglycemia was controlled by the simultaneous administration of magnesium and insulin, CA1 injury was no different to saline treated controls. As a measure to ensure the delivery of magnesium to the target tissue, Tsuda et al., (1991) administered magnesium chloride (1µl; 50 mM solution) directly to the CA1 sector of the hippocampus at 10 minutes before ischemia or at 0, 2, 12, 24 or 48 hours after ischemia. A neuroprotective effect was observed in CA1 neurons at all time points except 48 hours post-ischemia. A lower dose of magnesium chloride (1 µl; 10 mM) administered at the 24 hour time point also showed a neuroprotective trend, but was not significant at the P < 0.01 level. Animal body temperatures were not monitored during recovery, so a potential confounding factor is that the animals might have become hypothermic during this period. The potential for hypothermia will also have been compounded since the animals were re-anaesthetized in those groups receiving magnesium post-ischemia.
While the implications of post-ischemic hypothermia will be discussed in more detail later, it should be emphasized that 3 studies (Milani et al., 1999; Zhu et al., 2004b; 2005) in which measures were taken to avoid post- ischemic hypothermia, including 2 from our laboratory, found no significant neuroprotection (CA1 neuronal survival) despite starting magnesium treatment before or early after cerebral ischemia and using multiple and continuous dosing regimens. Furthermore, in the 2 studies from our laboratory (Zhu et al., 2004b; 2005), when either spontaneous or controlled mild hypothermia occurred during treatment, a neuroprotective effect was observed. In the remaining studies (Okawa et al., 1992; Sirin et al., 1998; Miles et al., 2001; Zhou et al., 2003) no measures were reported to have been taken to avoid post-ischemic hypothermia, and all of these reported positive histological or behavioral outcomes.
On the evidence from these studies, to the question of whether magnesium is neuro- protective following cerebral ischemia, the answer must be: sometimes. A second question arises, then, as to what factor or factors determine when magnesium will be neuroprotective, and when it will not. Taking into account the differences in study design (species; ischemia model; post-surgical temperature control; dosage, route and time of magnesium administration) we propose several explanations that may reasonably account for the inconsistencies in the reported data.
As previously reported (Meloni et al., 2006) there is reason to believe that the discrepant results found in these studies most likely can be attributed to the confounding effects of post- ischemic hypothermia. In the magnesium studies described in Tables 1 and 2, the animals were invariably maintained normothermic during ischemia, but in most cases there is no mention of body temperature monitoring or maintenance during recovery from surgery and in the post- ischemic period. This is crucial because it is now accepted and documented that post-surgical hypothermia commonly occurs in these models of cerebral ischemia, and it has been shown that post-ischemic hypothermia has confounded the results of previous studies (Buchan and Pulsinelli, 1990; Corbett et al., 1990; Welsh et al., 1990; Dietrich et al., 1995; Behringer et al., 2002). This is especially relevant in magnesium studies since there is also evidence that magnesium itself has hypothermia inducing properties (Zweifler et al., 2004; Wadhwa et al., 2005). Therefore, in those studies that did not actively maintain animals normothermic after ischemia, the occurrence of spontaneous hypothermia cannot be ruled out, and in fact is highly likely.
Further than that, direct evidence of the synergistic effects of mild hypothermia with magnesium has been generated from several studies in our laboratory. In an early global ischemia study (Miles et al., 2001) we did not monitor or control for animal body temperatures after ischemia; in this study we observed a neuroprotective effect. In subsequent studies (Zhu et al., 2004b; 2005), however, using the same model and magnesium treatment, but using controlled normothermic conditions after ischemia, we did not see a neuroprotective effect. Moreover, we observed that if body temperature was only monitored without being controlled, the animals did become mildly hypothermic during surgical recovery, and magnesium treatment in these animals did significantly reduce CA1 neuronal death. Note that control animals also became hypothermic to a similar degree, yet they did not show neuroprotection.
In subsequent experiments, when we compared magnesium efficacy in normothermic animals and in animals rendered mildly hypothermic (35°C) for 6 hours immediately after global ischemia, we observed that treatment with magnesium and mild hypothermia together increased CA1 neuronal survival. Importantly, no neuro- protection was observed either in normothermic animals receiving magnesium or in animals rendered hypothermic for 6 hours post-ischemia. We have also obtained similar findings with focal ischemia experiments. For example, we have demonstrated that following focal ischemia, magnesium treatment in normothermic animals does not result in reduced infarct volumes. In contrast, if magnesium treatment is combined with a period of mild hypothermia (spontaneous or controlled) reduced brain infarcts are observed (Campbell et al., 2008ab). In our controlled hypothermia experiments using a permanent focal model (Campbell et al., 2008b) we did not observe a protective effect with mild hypothermia alone.
Taken together, these experiments illustrate two important points. The first is that ischemic control animals, though they might experience some level of hypothermia, are unlikely to show any evidence of neuroprotection. The second is that the combination of magnesium and mild hypothermia does not appear to act in an additive way, but rather synergistically to unmask a neuroprotective effect. These findings highlight the necessity to maintain post-ischemic animal body temperatures in drug evaluation studies and, since in the majority of magnesium/cerebral ischemia studies this was not reported to have been done, there is reason to question the validity of the results.
It is a matter of speculation as to the degree to which magnesium dosage and route of administration is another contributing factor for the discrepant findings, in large part due to the fact that the optimal dose of magnesium remains unknown.
As a rule of thumb, early experimental and clinical studies have aimed to achieve serum concentrations of about double the baseline (i.e., to increase serum magnesium from ≈ 0.8 to ≈ 1.5 mmol/l), which seems reasonable as magnesium at this level appears to be safe and well tolerated (Muir and Lees, 1995; 1998). However, data from our laboratory indicate that magnesium doses that double baseline serum levels may be too high. For example, magnesium doses that resulted in serum magnesium levels >1.4 mmol/L (e.g., using an IV loading dose of 0.36 mmol/kg, followed by an infusion of 0.24 or 0.48 mmol/kg/h) provided no or minimal CA1 protection. On the other hand, a lower dose (0.36 mmol/kg IV, then 0.12 mmol/kg/h), which resulted in serum magnesium from 1 - 1.2 mmol/l, provided marked CA1 survival following global ischemia (note that in these experiments the animals’ body temperatures were not monitored during the post-ischemia period; Miles et al., 2001). In addition, we have shown that the higher loading dose of 0.72 mmol/kg, when given before focal ischemia (in animals that were maintained normothermic in the 6 hour post- ischemic period), produced a trend towards increased infarct volume.
The time of administration will also have a marked influence on the potential neuro- protective effect of any intervention. Most of the studies here administered magnesium before or immediately after induction of cerebral ischemia; hence, increased magnesium levels were present in serum, and possibly the brain, at the time of ischemia. Post-ischemic treatments with magnesium, following both focal and global cerebral ischemia, have, however, produced positive outcomes, with treatment as late as 6 hours after focal ischemia (Yang et al., 2000) and 24 hours after global ischemia showing a benefit (Tsuda et al., 1991). Even allowing that hypothermia has likely confounded some outcomes, it is encouraging to note that delayed treatment with magnesium can be effective.
While different routes of administration (intravenous, intra-arterial, intraperitoneal, subcutaneous, intracranial) have been used to deliver magnesium, it is difficult to comment on how this may have influenced outcomes in the absence of measures such as the serum and CSF concentrations achieved. Again of particular interest is the positive study using intracranial delivery (Tsuda et al., 1991), which reported efficacy even at 24 hours post-ischemia. By bypassing the blood brain barrier, intracranial delivery would guarantee increased brain magnesium and thus maximize its effects.
Only 1 cerebral ischemia study has examined the extent to which exogenously administered magnesium penetrates into either brain or CSF. This study, by Okawa et al., (1992), used an IV loading dose (0.664 mmol/kg) followed by an infusion of magnesium sulphate (0.332 mmol/kg/h for 3 h, then 0.083 mmol/kg/h for 45
h) in dogs subjected to global cerebral ischemia. They showed a significant increase in CSF magnesium lasting from 20 minutes to 7 days after treatment. Interestingly, dogs not subjected to global ischemia, but treated with the same magnesium regimen, did not show significantly elevated CSF magnesium. In a study using rats not subjected to cerebral ischemia, Sjöström and Wester (1990) showed elevated brain and CSF magnesium shortly after IV administration of 2.0 mmol/kg magnesium chloride. The Okawa et al., (1992) study suggests that exogenously administered magnesium may have increased access to the brain following cerebral ischemia, making it likely that in cerebral ischemia experiments magnesium has been present in the target tissue, and thus able to exert both vascular and neural effects.
Magnesium sulphate and magnesium chloride have both been used in models of cerebral ischemia, with both positive and negative results, yet no studies have directly compared their efficacy. The only study that has assessed both magnesium salts was an early global ischemia study from our laboratory (Miles et al., 2001). Our data appeared to favour greater efficacy for magnesium sulphate, but unfortunately the trials with the two magnesium salts were not performed concurrently, and once again there is the likelihood of post-ischemic hypothermia confounding the results. Interestingly, in our experiments animals treated with magnesium chloride did not become hyperglycaemic (unpublished observation), which is in contrast to two earlier studies (Blair et al., 1989; Izumi et al., 1991). A possible contributing factor to the hyperglycaemia reported by Blair et al., (1989) and Izumi et al., (1991) is that they found magnesium treatment was associated with hypothermia, which can reduce insulin secretion (Polderman, 2009) predisposing the animals to hyper-glycaemia.
In summary, the discrepancies in the outcomes of these studies are explicable in light of several unanswered questions that still surround magnesium treatment for cerebral ischemia. On the available evidence, it seems fair to say that magnesium does indeed have neuroprotective properties when a suitable dose is given to achieve moderately increased (above normal) levels in the target tissue in conjunction with a mild degree of hypothermia.
As it stands, in light of the failure of the IMAGES stroke trial to find a significant treatment effect with magnesium (IMAGES, 2004), the future use of magnesium for the treatment of stroke, and possibly other forms of cerebral ischemia, will depend heavily on the outcome of the FAST-MAG Phase 3 trial (Saver et al., 2004, Stroke Trials Directory, 2010). This trial is currently underway, and is assessing whether field administration of magnesium within 2 hours of stroke onset improves clinical outcomes. In essence, the FAST- MAG trial will address whether the long delay to magnesium treatment in the IMAGES trial (enrolment up to 12 hours after the onset of stroke) was the principal reason for its ineffectiveness. However, based on our own assessment of magnesium under normothermic conditions, we predict that if FAST-MAG patients are maintained normothermic this trial will also show little or no benefit. As discussed, the evidence suggests that in order for magnesium to produce a positive outcome after stroke/cerebral ischemia it needs to be combined with mild hypothermia (Meloni et al., 2009). Fortunately, hypothermia induction may only require a reduction in body temperature of 1 - 2°C involving basic cooling measures, which may be achievable in conscious patients. It is anticipated that future cerebral ischemia experimental studies will play a vital role in assessing magnesium in combination with hypothermia, and potentially with other agents as well, with successful outcomes guiding the design of clinical trials.
There is evidence from several models and using a number of treatment regimens that magnesium has neuroprotective properties that can reduce brain damage and neurological deficits after cerebral ischemia. Given that hypomagnesaemia is known to occur after stroke, magnesium supplementation has a place in the normal supportive care of such patients. Beyond that, the question remains as to whether, and by how much, supplementation should aim to increase, rather than just restore, serum magnesium levels. The FAST-MAG trial may provide part of the answer, though probably without being definitive.
The rationale for trialing magnesium as a neuroprotective agent following stroke has been based on its role in maintaining normal brain functions, on its known cellular actions that are anticipated to counteract damaging ischemic processes and subsequently on positive experimental data from animal models of cerebral ischemia. However, close scrutiny of the animal data shows that about 46% of studies have not shown a neuroprotective effect, and that the majority, if not all, of the positive studies were potentially confounded by post-ischemic hypothermia. In addition, experimental design has not always been well-founded with respect to magnesium dosage, and to the time and route of magnesium administration. Moreover, recent animal studies under controlled post-ischemic conditions indicate that magnesium is only neuroprotective when combined with hypothermia. Finally, additional information regarding the efficacy of magnesium as a stroke treatment will be available on completion of the FAST-MAG trial, but in the meantime the neuroprotective potential of magnesium remains a valid subject for investigation in cerebral ischemia models, with emphasis on its combination with post-ischemic hypothermia.
The authors would like to thank William Gow for compiling the reference list.
This book is copyright. Apart from any fair dealing for the purposes of private study, research, criticism or review as permitted under the Copyright Act, no part may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise without the prior written permission. Address all inquiries to the Director at the above address.
Your browsing activity is empty.
Activity recording is turned off.
See more...
