General strategies
Anti-excitotoxic interventions
Glutamate and aspartate are known as excitatory neurotransmitters that stimulate NMDA and AMPA receptors (Ca2+ and Na+ influx, respectively). As the activation of these receptors initiates catabolic intracellular processes, blockade of NMDA and AMPA receptors may protect cerebral tissue. N- Methyl- d-aspartate receptors are generally comprised of NR1/NR2 sub-units, which are hyperactivated during neuronal injury. Te neonatal NR3 subunit reduces the excitability and the Ca2+ permeability of NMDA-associated ion channels, which makes exogenously added NR3 an interesting neuroprotective therapeutic.
Ketamine, MK-801 (dizocilpine), aptiganel, dex-trorphan and Mg2+ represent non-competitive NMDA-receptor antagonists. In animal models of focal (but not global) cerebral ischaemia and head injury, ketamine as well as MK-801 reduced neuronal injury and improved outcome. Likewise, infusion of the competitive NMDA-receptor antagonist CGS 19755 (selfotel) reduced infarct size following focal ischaemia. Clinical trials using MK-801 were terminated due to toxic side effects and the induction of mitochondrial vacuolization. Te clinical development of the antitussive agent dextror-phan was also terminated because of side effects such as hallucination, agitation and sedation. In clinical phase III trials in patients with acute stroke using the non-competitive NMDA-receptor antagonist aptiganel (Cerestat, CNS-1102) or the competitive selfotel, side effects were minimized by choosing low drug concentrations (far below the protective dose range of preclinical studies). All trials showed no improvement in primary outcome and mortality was higher in the treatment groups. In contrast to the direct NMDA-receptor antagonists, the NMDA-receptor glycine site antagonist gavestinel had an acceptable safety profile; however, two phase III multicentre trials (GAIN Americas Trial and GAIN International Trial) in stroke patients were not able to show any neuroprotective effect. Another approach to minimize excitotoxic damage is blockade of the AMPA receptor. However, in a phase III trial, the AMPA antagonist YM872 (ARTIST+ trial) was abandoned after failing an interim futility analysis. Decreased excitotoxicity might also be achieved by activation of the inhibiting y-aminobutyric acid (GABA) receptor. Te GABA agonist clomethiazole, which has been promising in preclinical studies, unfortunately failed to improve neurological outcome when administered within 12 h after the onset of stroke. In summary, to date, no anti-excitotoxic drug has been effective in clinical multicentre trials.
Anti-inflammatory interventions
Poststroke inflammation is associated with progression of neuronal damage. Terefore, anti-inflammation represents a neuroprotective strategy, which includes inhibition of inflammatory mediators such as cytokines (e.g. IL-1, IL-6, nuclear factor-KB) and adhesion molecules (e.g. ICAM, VCAM, integrins). Despite many successful experimental studies, only a few clinical studies have been performed. Te endogenous, highly selective IL-1 receptor antagonist (IL-1ra) suppressed markers of inflammation in 17 patients after stroke and improved clinical outcome at 3 months compared with 17 placebo-treated patients. Te ICAM-1 antibody enlimomab reduced leucocyte adhesion and infarct size in many experimental stroke studies, but there was deteriorated neurological outcome in stroke patients after 90 days, causing significantly more infections and fever (Enlimomab Acute Stroke Trial). Hypoxia increases the in vitro expression of the neu-trophil integrin CD11. In rats, administration of anti-CD11 monoclonal antibodies reduced infarct volume and apoptosis, which was associated with a decreased accumulation of neutrophils. Unfortunately, in a clinical trial in stroke patients, inhibition of neutrophils using an anti-integrin therapy with CD11/CD18 antibodies showed no protection and the study was prematurely terminated. To date, no multicentre trial has been able to prove the validity of anti-inflammatory therapeutic strategies in patients with neuronal injury. Possibly, cerebral inflammation after brain damage also has some beneficial effects. Inflammation seems to be important for clearing of damaged tissue, stimulating the process of angiogenesis, tissue remodelling and regeneration. hus, it is possible that certain kinds of infl ammatory reactions are neuroprotective and neu-roregenerative. herefore, an accurate balance between inflammation and anti-inflammation is necessary to assure the removal of cell debris and to avoid secondary cell damage. New therapeutic targets could be designed to obtain a correct modulation of the immune system and to reduce cerebral damage after brain injury.
In contrast to the inflammatory reaction of the brain, the systemic immune system is depressed after neuronal damage and thereby increases the susceptibility to infection. his central nervous system injury-induced immunodepression was first detected in mice, which showed that experimental stroke propagates bacterial aspiration from harmless intranasal colonization to harmful pneumonia due to systemic immunosuppression. Although the exact mechanism is unclear, it became evident that activation of the hypothalamic-pituitary axis and of the sympathetic nervous system triggers downregulation of the systemic immune response af er brain damage. To prevent pneumonia as a consequence of stroke-induced immu-nodepression, three clinical trials tested the protective effect of antibiotic therapy with the fluocinolone levo-floxacin (ESPIAS trial), the fluocinolone moxifloxacin (PANTHERIS trial) and the tetracycline minocycline. While the ESPIAS trial revealed that prophylactic administration of levofloxacin does not prevent infection in patients with acute stroke, the PANTHERIS trial suggested that preventative administration of moxifloxacin is superior in reducing infections after severe non-lacunar ischaemic stroke compared with placebo. he conflicting results of both studies may be explained by patient selection and differences in antibiotics and dosages. Larger trials are needed to answer the question of whether survival of patients after neuronal trauma can be positively affected by preventative anti-infective therapy. he minocycline treatment in acute stroke improved outcome (Ranking Scale and Barthel Index), which was attributed to the prevention of infection but also blockade of other steps of secondary brain damage, such as reduction of micro-glial activation, an anti-NMDA effect and inhibition of apoptotic cell death. After a successful dose-finding study, the neuroprotective effect of minocycline after stroke is currently under investigation in two phase II studies, which should be completed in 2011.
Antioxidants
High concentrations of free radicals cause cellular degeneration and disruption of the brain-blood barrier. Antioxidants have been neuroprotective in several in vitro models of oxidative cell damage and in vivo models of cerebral ischaemia. However, none has been successfully translated into clinical practice. One of the most recent failures was disodium 4-[(tert-butylimino)methyl]benzene-1,3-disulfonate N-oxide (NXY-059), a nitrone-based free radical trapping agent, which initially appeared to be a promising compound, yet was later dismissed for not reaching predefined clinical goals. Consistently, in patients with intracerebral haemorrhage, NXY-059 also showed no neuroprotective potential. Similar failures of initially claimed promising compounds occurred with ebselen and edaravone, in spite of the fact that the latter compound has been approved for clinical use in Japan.
Superoxide dismutase is a physiological free radical scavenger that occurs in almost all organisms. Studies in cell cultures and transgenic mice overexpressing SOD have shown that SOD produces a maximum reduction in cellular oxidative stress. Unfortunately, SOD has an extremely low penetration through the blood-brain barrier and cellular membranes. Consequently, SOD has been conjugated with polyethylene glycol (PEG) to enhance the bioavailability. In patients with severe head injury (phase II trials), PEG-SOD reduced mortality when given in high concentrations. However, a consecutive phase III trial in 463 head-injured patients failed to confirm the neuroprotective effects of PEG-SOD.
Tirilazad mesylat (U-74006F), a non-glucocorti-coid 21-aminosteroid, is a potent inhibitor of oxygen free radical-induced lipid peroxidation. Most laboratory investigations have shown that tirilazad reduces infarct size and improves neurological outcome in models of transient or permanent focal or global cerebral ischaemia, cardiopulmonary arrest, spinal cord injury and traumatic brain injury (TBI), even when infused after the insult. In contrast, phase III clinical trials in patients with acute stroke, subarachnoid haemorrhage (SAH) and head injury failed to confirm the experimental neuroprotective evidence.
Pre-post-conditioning
An unconventional way to protect the brain seems to enhance its tolerance to ischaemic or anoxic insults by pre-conditioning. By a brief sublethal episode of ischaemia, cells become resistant to subsequent lethal events, which was first proven for the myocardium and later for the brain. Shortly after pre-conditioning (early pre-conditioning after minutes) or after a delay (delayed pre-conditioning after hours or days), the brain develops tolerance towards the same or even different subsequent injury. Te pre-conditioning cascade includes stimuli that, via sensors and transducers, activate downstream transcription factors, ultimately modulating gene expression. Novel proteins may then act as effectors, enhancing the resistance of neurons to ischaemia, and/ or pre-existing proteins may be modulated by post-translational modification acting as effectors. Tolerance induced by pre-conditioning changes the expression of genes involved in the suppression of metabolic pathways, immune responses, ion-channel activity and blood coagulation. Tese protection concepts include the antagonization of mechanisms of damage, increase in substrate delivery, decrease in energy use and improvement of recovery. Besides the classical pre-conditioning using short sublethal ischae-mia drugs such as anaesthetic agents, molecules such as glutamate, inflammatory cytokines and caspases, or ATP-sensitive potassium (KATP)-channel openers, can also pre-condition brain tissue and increase its tolerance towards injury.
If the concept of pre-conditioning is clinically valid, patients with transient ischaemic attacks should have an attenuated severity of subsequent stroke and an improved outcome compared with patients without a preceding event. Most prospective and retrospective clinical studies have been able to show this improvement. However, two more recent studies by Johnston in 2004 and Della Morte and colleagues in 2008 were not able to find a correlation between previous transient ischaemic attack and low stroke severity, thereby challenging the neuroprotective effects of preconditioning. Te clinical benefit of pharmacological pre-conditioning might include patients undergoing cardiovascular or brain surgery or experiencing SAH (delayed ischaemic deficits due to vasospasm) who are at risk of developing brain injury in the near future. Of particular clinical and pharmacological interest are the mitochondrial KATP -channel openers, erythropoietin and volatile anaesthetics. Te effect of pre-conditioning using erythropoietin after SAH has been under investigation. Although this study was terminated in February 2009 because of concerns about increased mortality in stroke patients, the results have not yet been published (http://clinicaltri-als.gov/ct2/show/NCT00626574). Furthermore, in China a clinical trial on the pre-conditioning effects of sevoflurane in patients with intracranial aneurysm surgery was initiated (http://clinicaltrials.gov/ct2/ show/NCT01204268). Slow intermittent reperfusion performed immediately after opening of the occluded vessel can reduce cerebral infarction after experimental stroke, which is described as post-conditioning. Tis concept of post-conditioning can be induced by a broad range of stimuli or triggers, and may even be performed as late as 6 h after focal ischaemia and 2 days after transient global ischaemia. Several clinical studies are currently being performed to investigate the protective effect of remote conditioning, which describes the fact that a 5 min ischaemia of a remote limb might protect the brain .
Specific drugs with multiple approaches
Calcium-channel blockers
The proposed mechanisms of neuronal protection by calcium-channel blockers include cerebral vaso-dilation, prevention of vasospasm, reduced Ca2+ influx and modulation of free fatty acid metabolism. Unfortunately, the results in animal models are rather contradictory. When calcium-channel blockers were infused within 15 min after the onset of cerebral ischae-mia, neuronal injury and neurological outcome were positively influenced, while after that time window calcium-channel blockers have been ineffective.
Clinical trials have tested the neuroprotective effects of the L-type calcium-channel blocker nimodipine, especially the prevention of vasospasm, in patients with acute ischaemic stroke and aneurysmatic or traumatic SAH. According to a meta-analysis of nine placebo- controlled trials with a total of 3700 patients with acute stroke, oral administration of nimodipine appears to be associated with a favourable outcome as long as the treatment commences within the first 12 h following the onset of the symptoms. However, calcium-channel blockers may induce arterial hypotension below the individual ischaemic threshold of the patients, and any relevant decrease in arterial blood pressure will reverse the neuroprotective effects of the intended treatment. Consistently, recent analyses by the Cochrane Foundation identified no beneficial effect of nimodipine in patients with ischaemic stroke or traumatic haemorrhage. However, in patients with aneurysmal SAH, oral nimodipine reduces the risk of poor outcome and secondary ischaemia, while there is no evidence for intravenous nimodipine or other calcium-channel blockers.
Anaesthetic agents
The proposed mechanisms of anaesthetic protection include reduction of cerebral metabolism and ICP, and suppression of seizures and sympathetic discharge. Additionally, anaesthetics may be neuroprotective by inhibiting synaptic glutamate release, activating inhibitory GABA and glycine receptors, and minimizing intracellular Ca2+ concentration and free radical accumulation. While the protective effect of anaesthetic agents after neuronal injury in animals is quite evident, a multicentre trial in patients undergoing carotid end-arterectomy with general or local anaesthesia revealed no difference in the occurrence of stroke at 30 days after surgery. Despite some limitations of this impressive trial such as statistical underpower, lack of standardization of the local anaesthetic used and varying levels of sedation in the local anaesthetic group, the data confirm the view that one single pharmacological approach may have little effect on the multiple pathological events in patients with parallel variability of coexisting disease.
Inhalational anaesthetics
Isoflurane, sevoflurane and desflurane produce cerebral metabolic suppression at end-tidal concentrations >0.5 minimum alveolar concentration (MAC), suggesting that volatile anaesthetics may correct for the imbalance between reduced oxygen supply and demand during focal cerebral ischaemia. Animal studies with focal or incomplete hemispheric ischaemia have shown that isoflurane, sevoflurane and desflurane may decrease infarct size and improve neurological outcome when given prior to the ischaemic challenge. hese experimental data are consistent with studies in sevoflurane-anaesthetized patients undergoing carotid endarterectomy showing increased tolerance to lower levels of cerebral blood flow with preserved neuronal function during carotid cross-clamping when compared with halothane or enflurane. In contrast, volatile anaesthetics have no neuroprotective properties in the setting of global cerebral ischaemia and when given af er the insult. It is questionable whether the anti-necrotic effects of volatile anaesthetics seen in different ischaemia models are permanent. As indicated above, neurons may die from apoptosis if the tissue is exposed to a lesser degree of hypoxia or ischaemia. Studies in rats subjected to focal cerebral or forebrain ischaemia have shown that necrotic cell death was substantially reduced at 2 or 5 days from ischaemia in isoflurane-anaesthetized rats compared with awake controls or fentanyl-NO-anaesthetized animals. However, cortical and subcortical damage was not different between isoflurane and awake state or fentanyl-NO anaesthesia at 14 days, 3 weeks or 3 months after ischaemia. his suggests that volatile anaesthetics economize ischae-mic energy consumption with consecutive reduction in immediate necrotic cell death. However, the metabolic shift to less energy deprivation is insufficient to entirely restore neuronal integrity, and initiation of apoptosis (an energy-requiring process) will reverse the initial neuroprotection.
The noble gas xenon, which is currently undergoing trials as an anaesthetic agent in patients, reduces the extent of cerebral infarction in rodents subjected to focal ischaemia possibly by NMDA-receptor blockade. However, clinical trials investigating the neuroprotec-tive potency of xenon are still lacking.
Barbiturates and propofol
Barbiturates as well as propofol reduce infarct size and improve neurological outcome following focal or incomplete global cerebral ischaemia as long as physiological variables are controlled during the experiments. Similar to volatile anaesthetics, this neuroprotective effect seems to be sustained in models of mild neuronal damage, while the neuroprotective effect after severe neuronal injury is not preserved after 1 week. While experimental data support the preventative neuropro-tective effects of hypnotic agents, the clinical evidence is less convincing. In patients undergoing cardiac surgery with normothermic cardiopulmonary bypass, the infusion of thiopental (total dose during extracorporeal circulation: 39.5±8.4 mg kg-1 IV) was able to minimize post-operative neuropsychological deficits. In contrast, barbiturates infused to comatose patients within the first hour following cardiopulmonary resuscitation were ineffective in reducing mortality as well as neurological deficits in survivors compared with standard ICU treatment. Tese data are consistent with the view that the infusion of hypnotics prior to focal but not global ischae-mic insults may increase the ischaemic tolerance of neurons. Barbiturates may be also beneficial in patients with severe head injury and refractory intracranial hypertension. Tis conclusion is related to a series of clinical studies where infusion of barbiturates was effective in reducing ICP and probably the mortality rate following brain trauma as long as systemic haemodynamic stability was maintained. More recently, propofol was suggested as an alternative to barbiturates for sedation of patients with head injury due to a favourable context-sensitive half-time. Similar to barbiturates, propofol turned out to be effective in treating elevated ICP following head injury, and compared with an opioid-based sedative regimen, propofol was more effective in controlling ICP, although with a similar neurological outcome.
Glucocorticoids
Glucocorticoids ameliorate oedema associated with brain tumours and improve outcome in patients with bacterial meningitis. Besides these proven clinical indications, further neuroprotective effects of glucocorticoids have been investigated. After neuronal injury, the proposed neuroprotective mechanisms of glucocorticoids include an increased order of lipid bilayers, a decrease in intracellular Ca2+ concentration, free radical scavenging and prevention of free fatty acid accumulation by inhibition of lipid peroxi-dation. However, in patients with acute stroke, after traumatic head injury or following cardiac arrest, glucocorticoids (e.g. dexamethasone or methylpred-nisolone) did not reduce neuronal damage and even resulted in a deteriorated outcome. In contrast, following spinal cord injury, infusion of methylprednisolone (30 mg kg-1 bolus, 5.4 mg kg-1 over 24 h within 3 h of injury; or 30 mg kg-1 bolus, 5.4 mg kg-1 over 48 h within 3-8 h from injury) may slightly reduce motor deficit and improve function of sensory tracts.
Nitric oxide
Nitric oxide is a messenger molecule with impact on a variety of motor deficit intra-, extra- and intercellular processes. Nitric oxide occurs during the conversion of L-arginine to citrulline by the enzyme NO synthase (NOS). Tree isoforms of NOS have been identified.
Constitutive NOS isoforms I (neuronal NOS, nNOS) and III (endothelial NOS, eNOS) are present in neurons, astrocytes, perivascular nerve fibres and endothelial cells. Te inducible isoform NOS II (inducible NOS (iNOS)) is present in leucocytes and macrophages. Nitric oxide is a diffusible, highly reactive molecule with a half-life in the order of a few seconds. During hyp-oxic ischaemic conditions, NO exerts positive as well as negative effects on neuronal functions and structure. Endothelial NOS dilates cerebral vessels and is crucial for improving cerebral blood flow. Te inducible NOS in leucocytes and macrophages may contribute to the formation of peroxynitrite and hydroxyl anions after neuronal damage, releasing NO at concentrations that are cytotoxic by inhibition of mitochondrial enzymes and DNA trauma. Experiments in nNOS knockout mice revealed a relevant reduction in infarct size following focal cerebral ischaemia compared with wild-type mice. In contrast, infarct size was increased in eNOS knockout animals. Lubeluzole, the S-isomer of a novel 3,4-difluorobenzothiazole, downregulates the glutamate-activated NOS pathway, thereby mediating neuroprotection in models of focal and global cerebral ischaemia models. However, a Cochrane Database meta-analysis reviewing five trials with 3510 stroke patients revealed no effect on functional outcome or mortality in the lubeluzole-treated patients.
Erythropoietin
In the brain, the glycoprotein erythropoietin is produced in the hippocampus, internal capsule, cortex, endothe-lial cells and astrocytes, and its receptors are expressed by neurons, microglia, astrocytes and cerebral endothelial cells. Hypoxia and ischaemia have been recognized as important driving forces of erythropoietin expression in the brain, suggesting that erythropoietin is part of a self-regulating physiological protection mechanism to prevent neuronal injury. Systemic application of the growth factor erythropoietin stimulates neurogenesis and neuronal differentiation, and activates brain neurotrophic, anti-apoptotic, anti-oxidant and anti-inflammatory signalling. Furthermore, erythropoietin seems to possess pre- and post-conditioning effects. Tese multiple protective approaches were confirmed in animal models of focal and global cerebral infarction and of TBI. In a small study in 13 patients with cerebral ischaemia, erythropoietin proved to be well tolerated and was associated with an improvement in clinical outcome after 1 month. Based on these promising results, the German Multicenter EPO Stroke Trial was started in 2003, aiming at the inclusion of over 500 patients. After the trial concluded in 2008, the authors realized that, during the study, the standard treatment of stroke changed and that over 60% of the patients additionally received recombinant tissue plasminogen activator (rtPA). Subgrouping the patients into rtPA and non-rtPA populations complicated the evaluation of the data. However, a preliminary analysis of the data showed that the results of the initial erythropoi-etin stroke trial can be reproduced. Another promising result derives from a single-centre study investigating 80 patients with aneurysmal SAH. In these patients, 90,000 IU of erythropoietin seemed to reduce delayed cerebral ischaemia following subarachnoidal haemorrhage by decreasing the severity of vasospasm and shortening of impaired autoregulation.
Magnesium
The potential neuroprotective mechanisms of magnesium include reduction of presynaptic glutamate release, blockade of NMDA receptors, improvement of mitochondrial calcium buffering, blockade of calcium entry via voltage-gated channels and relaxation of smooth muscles, which might be beneficial in patients with vasospasm after SAH.
After successful experimental studies in focal ischaemia models, a clinical study was performed in stroke patients (IMAGES), but magnesium infusion within 12 h after stroke failed to reduce the chances of death or disability. As a consequence, a shorter time window, 2 h after the onset of stroke symptoms, for magnesium infusion is currently being investigated (FAST-MAG trial, started in 2005). After a pilot study showing that intravenous magnesium might reduce the incidence of delayed cerebral ischaemia after SAH, a phase III trial failed to show any beneficial effect of magnesium infusion in aneurysmal SAH (IMASH trial). Continuous infusion of magnesium to patients within 8 h of TBI was not shown to be neuroprotective and might even have a negative effect in the treatment of significant head trauma. A promising innovative concept in the treatment of spinal cord injury might be the intrathecal application of magnesium sulfate, which improved neurological function in a model of spinal cord ischaemia .
Conclusion
Over the last few decades, the pathological mechanisms of neuronal damage have been characterized and many drugs have been tested successfully in reproducible and physiologically controlled animal models of cerebral ischaemia or brain trauma. However, translation from the bench to the bedside has failed and none of these drugs has proved its neuroprotective potency in clinical multicentre trials. Possibly, many agents were brought to clinical trial without a sufficient solid or relevant evidence-based preclinical foundation. In contrast to experimental studies, where the drug was administered shortly after or even before the ischaemic challenge, in most of the clinical trials, the drug was only applied as early as 6 h after the neuronal injury. Furthermore, in some cases, it is not possible to achieve the preclinical efficacious drug doses or plasma levels in humans due to unfavourable side effects or different pharmacokinetics. Another problem is the insufficiency of clinically available outcome measures. Tese and other pitfalls may be responsible for the lack of evidence for the effectiveness of potential neuroprotective drugs in clinical trials. To overcome these problems by improving the quality of the preclinical studies, guidelines for experimental studies have been established by the Stroke Terapy Academic and Industry Roundtable (STAIR). Tese criteria might help in future to choose the right drug for clinical investigations.
