The science of cerebral ischemia and the quest for neuroprotection: navigating past failure to future success

Abstract

Ischemic stroke remains a leading cause of morbidity and death for which few therapeutic options are available. The development of neuroprotective agents, a once promising field of investigation, has failed to translate from bench to bedside successfully. This work reviews the ischemic cascade, agents targeting steps within the cascade, and potential reasons for lack of translation. Additional therapeutic targets are highlighted and areas requiring further investigation are discussed. It is clear that alternative targets need to be pursued, such as the role glia play in neurological injury and recovery, particularly the interactions between neurons, astrocytes, microglia, and the vasculature. Similarly, the biphasic nature of many signaling molecules such as matrix metalloproteinases and high-mobility group box 1 protein must be further investigated to elucidate periods of detrimental versus beneficial activity.

Decades of research on ischemic stroke have produced a heightened understanding of its pathophysiology, providing an extensive array of potential therapeutic targets. Although many of these proposed targets have been manipulated through various pharmacological compounds, many of which have progressed to clinical trials, there remains an unmet need for alternative or supplemental treatments besides tPA for stroke. In addition to the complex pathophysiology of stroke, the lack of successful translation from bench to bedside has resulted in controversy concerning the most appropriate preclinical models of ischemic stroke and assessment of potential therapeutics. This work highlights the pathophysiology of stroke, with an emphasis on the time-dependent nature of the ischemic cascade and the pharmacological agents tested preclinically and/or clinically targeting events within the cascade. Preclinical modeling and alternative therapeutic approaches and targets are discussed at length with the goal of improving the likelihood of successful translation from bench to bedside.

Background

Significance of Stroke

Stroke remains a leading cause of morbidity and death in the developed world, with tPA being the only approved pharmacological treatment. Due to many risks associated with tPA use, stringent guidelines must be followed; most specifically, treatment within 3 hours of symptom onset, resulting in less than 5% of patients receiving tPA treatment.⁵ Therefore, the development of alternative treatments is highly desirable.

Concept of Neuroprotection

The notion that neuroprotection may be possible following neurological injury, and more specifically ischemia, is based largely on the findings of Astrup et al.³ The onset of ischemia produces a so-called core region in which flow is severely reduced, by almost 90%. This sharp reduction in blood flow results in deficient ATP levels and subsequent ionic disruption and metabolic failure, progressing to cell death via necrosis within minutes. Surrounding this core region lies the so-called penumbra, in which flow is less severely reduced, typically in the range of 35% of baseline.³ This territory is characterized by the loss of action potential firing but maintenance of proper resting membrane potential. On restoration of blood flow to the penumbra, normal function returns. If flow is not restored acutely, tissue originally in the penumbral region dies and the core infarct region evolves to encompass what was previously penumbral tissue.⁵⁹ Although the general consensus remains that the infarct core is not salvageable, it is believed that penumbral tissue can be salvaged through flow restoration and/or restoration of cellular homeostasis through manipulation with pharmacological compounds. Potential neuroprotective agents target a variety of aspects of cellular physiology ranging from membrane integrity to mitochondrial function and the regulation of apoptosis. In addition to a targeted approach, other novel techniques and agents for the induction of hypothermia have been applied to patients afflicted with either ischemic stroke or other forms of neurological injury such as traumatic brain injury. Hypothermia is believed to slow the injury cascade, potentially slowing damage and providing an expanded window of opportunity for treatment. Because hypothermia is not believed to target a specific molecular mechanism, it is beyond the scope of this review.

Current Treatment Options

This concept resulted in the development of intravenous tPA, the most widely used therapeutic option for ischemic stroke treatment. Unfortunately, intravenous tPA can only be applied to a relatively small subset of those afflicted with ischemic stroke, less than 5%, as a consequence of numerous contraindications and the late presentation time of most patients after stroke onset.⁵ In an attempt to provide thrombolytic therapy to a broader patient population, other therapeutic techniques have been developed. These include intraarterial delivery of tPA, the Penumbra System, and the Mechanical Embolus Removal in Cerebral Ischemia (MERCI) clot retrieval device. Each of these alternative techniques, while expanding the therapeutic window of opportunity, requires specialists found primarily at advanced academic medical centers and are rarely available at smaller community hospitals, making them less than an ideal therapeutic option.

Targeting the Ischemic Injury Cascade

Pharmacological agents selected to interrupt the ischemic injury cascade that can be administered safely to a larger patient population are therefore desirable. Numerous compounds have been investigated as potential neuroprotective agents. These compounds have targeted events primarily occurring rapidly after ischemia onset. After cerebral blood flow is reduced, energy failure occurs due to a decoupling of oxidative phosphorylation, resulting in a series of events including glutamate release and ionic imbalance, which subsequently causes membrane depolarization.⁶⁰ Calcium rapidly enters the cell, causing the induction of various enzymes and mitochondrial damage. Further downstream, apoptotic mediators are released, free radicals produced, and inflammatory mediators released.¹⁴ The events of the injury cascade and the pharmacological agents designed to prevent them will be discussed in further detail in subsequent sections.

Current Status of Neuroprotectant Development

To date there have been more than 1000 preclinical studies identifying neuroprotective compounds, with more than 100 progressing to clinical trials. Of those progressing to clinical trials, none have proved efficacious in the human population.⁷⁸ Consequently, funding for the field has declined from both pharmaceutical corporations and government agencies.

The prospect of neuroprotection remains at the forefront of the stroke field, despite the aforementioned failures, with the vast majority of researchers still believing it has merit as a potential therapeutic option. Due to the economic burden of stroke, with an estimated $28.3 billion in direct costs and $25.6 billion in indirect costs in the US in the year 2010,³⁷ there is simply too great a need for additional treatments and for the development of neuroprotective agents to give up the search.

The failed translation from bench to bedside of therapeutic agents has resulted in much speculation about the requirements for achieving success. This work seeks to summarize the ischemic cascade (Fig. 1), agents that have targeted a given step yet failed in clinical trials, and potential reasons for past failures. Additional discussion highlights potentially promising therapeutic targets and areas or topics in need of further investigation.

Fig 1.

A diagrammatic representation of the ischemic injury cascade from the onset of ischemia to cell death.

Pathophysiology of Stroke and Corresponding Therapeutics

Acute Period

Glutamate-Mediated Excitotoxicity

Blood flow cessation as a result of ischemia onset induces a shortage of oxygen and glucose delivery. As the main sources of energy production via oxidative phosphorylation, diminished oxygen and glucose supply results in a decoupling of oxidative phosphorylation and, subsequently, reduced ATP production and disruption of ionic pump function. These events occur rapidly after onset of ischemia; in fact, ATP levels are decreased substantially within 2 minutes.⁹⁶ Diminished levels of ATP impede the functioning of Na-K ATPase, preventing the maintenance of normal ionic gradients. Therefore, elevations in cytosolic Na⁺ and a decrease in cytosolic K⁺ are seen, leading to a depolarization of the neuronal membrane.⁹⁶ Voltage-gated calcium channels are activated by depolarization, allowing for rapid influx of Ca⁺⁺, which results in release of excitotoxic amino acids such as glutamate. Under normal homeostatic conditions, glutamate is rapidly cleared from the synapse via presynaptic and astrocyte uptake, but this response is altered in ischemia. In ischemia, the excess glutamate within the synaptic cleft binds ionotropic NMDA and AMPA receptors on the postsynaptic membrane, prompting depolarization of the postsynaptic neuron and propagating the wave of depolarization.⁶⁰

Blocking Excitotoxic Events

Attempts to mitigate excitotoxicity have encompassed a wide array of targets, ranging from prevention of glutamate release to blocking glutamate receptors and facilitating the opening of ion channels that counteract the excitotoxicity (Fig. 2). One of the most widely investigated targets is antagonism of the NMDA receptor on the postsynaptic membrane. The NMDA receptor is activated by the binding of both glutamate and glycine to their respective sites, and is further modulated by an allosteric site.⁵² Once activated, the NMDA receptor acts as a nonspecific cation channel, allowing the passage of Ca⁺⁺, Na⁺, and K⁺ into the cell.⁹³ By preventing activation of the NMDA receptor, cation flux through the ion channel is reduced, preventing depolarization. Numerous agents have been investigated for this purpose and include noncompetitive antagonists such as aptiganel hydrochloride (Cerestat; CNS-1102),^2,52,69 dextrorphan hydrochloride, ^1,25,32,52 and magnesium;^72,87,113 competitive antagonists such as Selfotel (CGS-19755);^19,52,64 glycine-site antagonists like gavestinel (GV-150,526);^8,52,55,76 and indirect blockers exemplified by lubeluzole (Prosynap).^20,22,52

Fig 2.

A representation of targets pursued for preventing neuronal excitotoxicity, often attributed to excessive glutamate release. These targets include the following: 1) NMDA receptor antagonists; 2) AMPA receptor antagonists; 3) kappa opiate receptor antagonists; 4) GABA_A receptor agonists; 5) 5-HT_1A receptor agonist; and 6) potassium channel openers.

The AMPA receptor, an ionotropic receptor for glutamate, is another well-documented target in the search for neuroprotection via reducing excitotoxicity. The AMPA receptor has multiple glutamate-binding sites that when bound activate the receptor and open the cation channel, allowing passage of cations nonspecifically.¹¹² Logic in targeting the AMPA receptor is similar to that of the NMDA receptor, in that antagonizing the AMPA receptor should prevent cation influx and subsequent depolarization. One widely investigated therapeutic agent antagonizing the AMPA receptor is zonampanel (YM872).^52,99

In addition to targeting glutamate-binding receptors and preventing substrate-receptor binding, another alternative is to counteract the increase in cation influx. One way of doing this is to increase the influx of anions such as Cl⁻. Clomethiazole (Zendra), a GABA^A receptor agonist, was evaluated for this purpose.^61,66,91 Similarly, activation of the neuronal 5-HT^1A receptor has been shown to elicit hyperpolarization due to increasing inward K⁺ current. A compound activating the 5-HT^1A receptor that has been evaluated in not only preclinical experimental stroke studies but also in clinical trials is repinotan (Bayx3702).^67,89,104 Maxipost (BMS-204352), a potassium channel opener, has also been evaluated for the treatment of ischemic stroke. Once again, the logic behind application of this compound was that an increase of potassium influx would hyperpolarize the membrane, preventing depolarization.^16,43,52

A final target for preventing excitotoxicity is the opiate receptor, the kappa receptor in particular. This receptor results in the initiation of a G-protein signaling cascade when stimulated.⁷¹ Although the details of this cascade are not entirely elucidated, it is clear that the pathway is involved in the release of excitatory neurotransmitters, release of free fatty acids, and the presence of antioxidants. The agent nalmefene (Cervene) was a kappa receptor antagonist investigated for the treatment of stroke.^17,28

The target, preclinical development, clinical evaluation, and results of these agents are highlighted in Table 1. Notably, each of these agents failed when reaching clinical trials. The precise reason for each of these failures is unlikely to be identified, but numerous weaknesses in preclinical design and the translation from bench to bedside are evident. For example, most preclinical studies use young-adult animals despite the role of age in the clinical stroke population. Furthermore, preclinical work often involves drug administration prior to or shortly after onset of ischemia, yet clinical trials often have a much longer exclusion window to increase the number of eligible patients. Other potential shortcomings of studies are discussed in greater detail below.

TABLE 1.

Examples of proposed neuroprotectants attempting to mitigate excitotoxicity, and the progression from preclinical experimental stroke models to clinical trials^*

Reference No.	Drug	Target	Model	Time of Admin	Phase Reached	Max Incl Time	Result/Notes
2, 52, 69	aptiganel HCl	NMDA receptor (noncomp)	young, focal, transient	<1 hr	Phase III	6 hrs	failed
1, 25, 32, 52	dextrorphan HCl	NMDA receptor (noncomp)	young, focal, transient	<1 hr	Phase II	48 hrs	no improvement & adverse events
72, 87, 113	magnesium	NMDA receptor (noncomp)	young, focal, permanent	<8 hrs	Phase III	<2 hrs	ongoing
19, 52, 64	Selfotel	NMDA receptor (competitive)	young, global, transient	<30 min	Phase III	6 hrs	failed
8, 52, 55, 76	gavestinel	NMDA receptor (glycine site)	young, focal, permanent	<6 hrs	Phase III	6 hrs	failed
20, 22, 52	lubeluzole	NMDA receptor (indirect)	young, focal, permanent	<6 hrs	Phase III	8 hrs	failed
52, 99	zonampanel (YM872)	AMPA receptor	young, focal, transient	2 hrs	Phase III	<2 hrs (ARTIST+) or <6 hrs (MRI)	both trials failed
17, 28	nalmefene	κ opiate receptor	young, global, transient	15 min preischemia	Phase III	<6 hrs	failed
61, 66, 91	clomethiazole	GABAA receptor	young, global, transient	<4 hrs	Phase III	<12 hrs	failed
67, 89, 104	repinotan	5-HT1A receptor	young, focal, permanent	immediately	Phase IIb	<4.5 hrs	failed
16, 43, 52	BMS-204352	potassium channels	young, focal, permanent	2 hrs	Phase III	<6 hrs	failed

^*Admin = administration;

ARTIST+ = AMPA Receptor Antagonist Treatment in Ischemic Stroke Trial; Max Incl = maximum inclusion; noncomp = noncompetitive.

Effects of Intracellular Calcium

The excessive calcium influx results in activation of a multitude of enzymes including phospholipases and proteases, mediated through calcium serving as a messenger molecule for numerous downstream signaling pathways.^40,50 Activation of phospholipases and proteases results in membrane and protein degradation, decreasing cellular integrity and survival. Endonucleases are also activated, cleaving DNA and leading to further cellular changes and potentially cell death. Calcium influx also potentiates glutamate-mediated excitotoxicity as depolarization occurs and additional glutamate is released. The release of excessive amounts of glutamate results in propagating waves of depolarization, called periinfarct depolarizations, originating in the core and spreading to the peripheral penumbra. ⁷³ Periinfarct depolarizations, at least in the initial 3 hours of ischemia, have been shown to correlate with final infarct volume in preclinical studies.⁴¹ Other studies have shown wave propagation for at least 6–8 hours after ischemic onset, although the correlation with infarct volume at these later time points has not been established.⁵¹ Agents interrupting the waves of depolarization through targeted blockade of glutamate receptors (AMPA and NMDA) prevent progression of the penumbra to infarct in experimental models of ischemic stroke. In contrast, the creation of additional waves of depolarization through administration of potassium chloride causes expansion of the infarct core. While manipulation of periinfarct depolarization has occurred in preclinical models, the propagating waves of depolarization have been shown to occur in humans afflicted with stroke and, equally important, each wave has been shown to correlate with expansion of the infarct. It is unclear whether the spreading depolarization may have some beneficial effects, though, because it may induce the release of protective factors.

In addition to roles in protease, phospholipase, and endonuclease activation, and serving as an intermediary in downstream signaling, calcium influx is involved in mitochondrial damage and subsequent ROS generation.²⁴ Following a rise in intracellular cytosolic calcium concentration occurring following ischemia, calcium enters the mitochondria.⁹⁷ Calcium balance in the mitochondria is achieved via intake through the uniporter and output through 2Na⁺-Ca⁺⁺ exchange. This balance is disrupted in ischemia, with calcium entering via the uniporter far more quickly than the 2Na⁺-Ca⁺⁺ exchanger can function.⁵⁰ Accumulation of excess calcium within the mitochondria most often results in opening of the permeability transition pore found on the inner mitochondrial membrane. The permeability transition pore opening is associated with loss of membrane potential and subsequent osmotic swelling, leading to outer-membrane rupture and release of cytochrome c.⁹⁴ Although the mechanism has not been elucidated entirely, calcium influx and associated mitochondrial dysregulation may result in initiation of apoptosis and generation of ROS. This process plays a role not only in the initial ischemic injury but also in subsequent reperfusion and reperfusion-associated injury (Fig. 3).

Fig 3.

Schematic showing that ischemia results in increased intracellular calcium levels through not only the NMDA and AMPA receptors but also through a slow Ca⁺⁺ channel. Increased cytosolic calcium results in activation of endonucleases and phospholipases as well as proteases. The MPTP allows calcium to enter the mitochondria, which causes production of ROS and further damage. Calcium in the mitochondria also causes swelling and eventual rupture, releasing cytochrome c, long known to play a role in apoptosis.

Blocking Actions of Excess Calcium

Nimodipine, a calcium channel blocker frequently used in other forms of cardiovascular disease, was assessed as a potential therapeutic agent in the Phase III trial VENUS.³⁹ Preclinical studies as well as previous clinical trials reported mixed findings regarding the efficacy of nimodipine.^31,84,108 A subgroup analysis of a previous clinical trial in which nimodipine was used suggested improvement in outcome when drug administration occurred within 18 hours of symptom onset; hence the VENUS trial was initiated. Unfortunately, the VENUS trial failed to show improved outcome, in both primary and secondary measures, at its termination.³⁹

Oxidative Stress–Induced Damage

Oxidative stress has long been recognized as a potential target for therapeutic development following ischemia. Oxidative stress, induced by the production of ROS, can be attributed largely to mitochondrial dysfunction.⁹⁷ In addition to permeability transition pore opening and release of cytochrome c, inhibition of normal respiration occurs, along with the release of nucleotides.⁹⁷ Furthermore, the glutathione necessary for neutralization of ROS is reduced. Free-radical production leads to peroxidation of plasma membranes and intracellular organelle membranes.⁴¹ Release of biologically active free fatty acids, such as arachidonic acid, is a direct consequence of free-radical production. Production of ROS also leads to damage of intracellular organelles such as the endoplasmic reticulum and mitochondria, and induces DNA fragmentation. Disturbances to homeostasis induced by ROS contribute to the progression of downstream injury pathways and apoptosis, often similar to those induced by excess calcium influx.⁴⁴

Free-Radical Scavenging

Tirilazad mesylate, a compound inhibiting lipid peroxidation, was evaluated for treatment of stroke in a total of 6 trials (4 of which were published).¹⁰⁵ Tirilazad failed to reduce morbidity or death in acute ischemic stroke. Reasons for failure remain unknown but preclinical evaluation of the drug produced mixed findings, often depending on time of drug administration. ^38,70 When administered prior to ischemia tirilazad appeared efficacious in preclinical studies, but when administered poststroke no positive effect was found.^38,70

The most recent well-documented failure of a proposed neuroprotective agent is that of disufenton sodium (Cerovive; NXY-059), a free-radical spin-trapping agent. Believed by many to have undergone the most rigorous preclinical assessment ever of a potential therapeutic agent, disufenton sodium progressed to clinical trials (SAINT I and II).⁸⁸ In the SAINT I trial, disufenton sodium appeared to be neuroprotective because disability, assessed with the modified Rankin scale, was reduced in the treatment group.⁹⁰ In the confirmatory trial, SAINT II, disufenton sodium failed to reduce disability when administered within 6 hours of ischemic onset.²³ This lack of efficacy was also true for subgroup analysis and hemorrhage associated with tPA administration.⁹² A critical review of not only preclinical studies but also the design of the clinical trials reveals numerous discrepancies between studies, ranging from inconsistent time of administration to inadequate animal models and insufficient functional assessment.⁸⁸

Subacute Period

Apoptosis

Evolution of the infarct does not occur based on necrosis or apoptosis individually, but is rather a complex interplay between the aforementioned forms of cell death.^12,60,63,115 In the infarct core, where ATP is severely depleted, necrosis may be the predominating form of cellular death, whereas in the penumbra apoptosis may be the predominant form of cell death.¹² The onset of ischemia induces apoptosis through two primary methods, one of which is through altered gene expression of apoptotic mediators such as Bax, Bad, Bid, Bcl-2, Bcl-XL, and so on, and the second of which is the influx of calcium and subsequent mitochondrial damage, ultimately resulting in opening of the permeability transition pore. Mitochondrial damage and changes in mitochondrial membrane integrity can cause release of toxic mediators such as cytochrome c and apoptosis-inducing factor. This process, called the intrinsic arm of the apoptosis program, leads to the activation of proteolytic caspases and ultimately cell death. In contrast, the extrinsic arm of the apoptosis program uses a series of interactions between cell-death factors, such as Fas ligand, and celldeath receptors, such as the Fas receptor. After binding of ligand and receptor, an adapter protein (FADD) is recruited, resulting in the activation of the caspase cascade.^11,12

Inhibiting Apoptosis

As a result of the previously discussed cell stressors produced following ischemia, many of which are involved in mediating apoptosis, the pro- versus antiapoptotic protein balance is shifted toward the induction of apoptosis. One of the proapoptotic proteins, Bax, has been demonstrated to play an essential role in permeabilization of the outer mitochondrial membrane. ¹¹¹ By preventing the Bax-mediated disruption, the release of proapoptotic proteins can be avoided. When a Bax-inhibiting peptide was used in a neonatal model of ischemia, outcome was improved.¹¹¹

Inhibition of caspases is another alternative target in the attempt to reduce apoptosis and thus improve outcome from cerebral ischemia.⁷⁷ Caspases are proteases, first synthesized in the procaspase form, that are rapidly activated in response to earlier events in the apoptotic cascade such as apoptosome formation or death receptor signaling.⁸³ Caspase activation ultimately results in DNA fragmentation via cleavage and inactivation of inhibitor of caspaseactivated DNase, producing the hallmark “DNA ladder” of apoptotic cells.⁴⁷ In experimental neonatal focal ischemia, application of a pan-caspase inhibitor failed to reduce infarct volume, despite cell death shifting toward necrosis.⁴⁷

Whereas attempts to improve outcome after ischemia through the modulation of apoptosis have failed to progress to clinical trials and successful therapeutic agents, investigation of the apoptotic cascades in neonate as well as young-adult animals has revealed fundamental differences associated with age. For example, caspase-3 has been described as being highly involved in apoptosis in the immature brain but plays a minor role in the adult brain.⁴⁷ Similarly, the mitochondrial permeability often responsible for inducing apoptosis may be mediated via different age-related mechanisms. In neonatal hypoxicischemic injury permeability is largely Bax-dependent, but in the adult brain the cyclophilin D transition pore plays a prominent role.¹¹¹

Besides appreciating age-related differences in apoptotic mechanisms, another challenge that must be overcome in the pursuit of pharmacological apoptotic mediators is the variety of pathways ending in the same result—apoptotic cell death. Many studies focusing on one specific apoptotic pathway may delay cell death but not entirely prevent it. Rather, a pan-apoptotic inhibitor may be required.⁴⁶

Inflammation

Onset of ischemia results in a rapid activation of inflammatory cells residing in the brain, which are called microglia.¹¹⁰ This is followed by influx of circulating inflammatory cells such as granulocytes, leukocytes, T cells, monocytes, and so on.⁴² Similarly, over the first few hours, an array of proinflammatory mediators such as cytokines and chemokines are released from the damaged tissue, inducing adhesion molecule expression and subsequent transendothelial migration of circulating inflammatory cells.⁴² Infiltration of leukocytes from the bloodstream into brain tissue results in release of additional cytokines and chemokines, ultimately leading to excessive oxidative stress and activation of MMPs.¹¹⁰ The MMP activation subsequently causes enhanced blood-brain barrier disruption, leading to further inflammatory cell recruitment and associated inflammation.⁴⁵ The inflammatory cascade is complicated by evidence that the initiating microglia can also produce neuroprotective and neurotrophic factors.⁴⁴

Mitigating Inflammation

An extensive variety of targets exists in the effort to manipulate the inflammatory response to achieve neuroprotection (Fig. 4). Pharmacological agents modulating inflammation fall into a few general classes. These include astrocyte modulators, blockade of inflammatory cell receptor complexes, inhibitors of microglia, and general immunosuppressive and/or antiinflammatory agents.

Fig 4.

Inflammation has been identified as a promising avenue of therapeutic development. Multiple aspects can be inhibited, including release of inflammatory mediators from microglia, preventing adhesion, and blocking migration of inflammatory cells outside of the vasculature and into the parenchyma. ICAM-1 = intercellular adhesion molecule–1.

The precise role of the astrocyte in ischemic stroke research has received much less emphasis than the neuron. This neuron-centered approach fails to consider the intimate relationship between neurons and supporting glia.⁸⁶ Astrocytes have been credited with contributing to formation of the blood-brain barrier, modulating neurotransmitter levels in the synaptic cleft, regulating ion homeostasis, and also they may alter inflammation via microglial influence. Furthermore, the neuron-astrocyte interaction is responsible for controlling the microcirculation of the brain.^15,116 The influence of astrocytes can be seen both in homeostasis and in both the acute and chronic injury response via astrogliosis.¹⁰³ Modulating astrocyte activity, via either fluorocitrate or arundic acid (ONO-2506), has produced mixed results.^35,80,95,103 Arundic acid administered at up to 48 hours after stroke onset appears promising based on early results, yet fluorocitrate treatment at 5 days after ischemia resulted in decreased vascular remodeling and worsened functional status.^35,103 Arundic acid has advanced to clinical trials and its development is highlighted in Table 2.^52,80

TABLE 2.

Examples of pharmacological agents tested preclinically (and many clinically), seeking to modulate inflammation^*

Reference No.	Drug	Target	Model	Time of Admin	Phase Reached	Max Incl Time	Result/Notes
35, 52, 80, 103	arundic acid (ONO-2506)	astrocyte activation	young, focal, permanent	24 hrs	Phase I completed, Phase II/III planned	24 hrs	ongoing, improvement trend seen
6, 9, 27, 52, 114	Hu23F2G	CD11/CD18 integrin	young, focal, transient	20 min	Phase III	12 hrs	failed
6, 9, 27, 52, 114	enlimomab	anti–ICAM-1	young, focal, transient†	15 min	Phase III	6 hrs	failed
62, 65, 101	tacrolimus (FK-506)	immunosuppression	young, focal, transient	<3 hrs	NA	NA	no clinical trial completed to date
29, 34, 54, 100	minocycline	antiinflammation, antiapoptosis	young, focal, transient	4 hrs	Phase III planned	4.5 hrs	early phase success (given up to 24 hrs after stroke)

^*NA = not applicable.

^†When given with tPA but also when given alone.

Following ischemia, peripheral inflammatory cells such as neutrophils are recruited to the region of infarction. Multiple attempts have been made to target adhesion molecules on neutrophils, as well as other inflammatory cells, to prevent infiltration following ischemic stroke. Examples include Hu23F2G (LeukArrest), an antibody recognizing the CD11/CD18 integrin, and enlimomab (R6.5), which targets intercellular adhesion molecule–1.^6,9,27,114 More information on the preclinical development and results of clinical trials in which these compounds were used can be found in Table 2.⁵²

Modulating general inflammation after stroke is another potential (and promising) therapeutic target. The effects of nonspecific immunosuppressive agents like tacrolimus (FK-506) as well as microglia-specific inhibitors such as minocycline are highlighted in Table 2.^{29,34, 54,62,65,100,101}

Chronic Period

Inflammation

Although it is initiated in the subacute period after onset of ischemia, inflammation has been shown to persist for an extended period, out to more than 30 days.⁴⁸ Inflammation is generally thought of as a detrimental process exerting further tissue damage, but recent findings, particularly those associated with MMPs, indicate a restorative role of inflammation in these extended time periods.¹⁴

Because most of the preclinical stroke models focus on the acute period, many questions remain unanswered regarding pathological events occurring in the chronic period, particularly in terms of inflammation. Future work needs to elucidate the role of inflammation in repair and, more specifically, the ways in which altering acute inflammation impacts chronic inflammation that may in some cases be required for recovery and repair. Furthermore, determining the time course of expression of inflammatory mediators and how this is altered with age and/or comorbidity across both the acute and chronic periods may lead to additional insight into the role inflammation plays in not only injury but also recovery.

Repair

Recent work has focused on the development of stem cell therapies for poststroke treatment. Early studies have appeared promising and have shown effects ranging from inhibition of cell death to vasculature regrowth, modulation of inflammation, and induction of plasticity and neurogenesis. The use of stem cells, of which there are many subtypes, remains a promising future avenue of work with many questions remaining to be elucidated. Future studies will probably need to ascertain the precise cell types most likely to demonstrate success (exogenous, neural, embryonic, induced pluripotent, and so on) and the optimal delivery of these cells. Furthermore, other questions remain, such as the way in which these cells integrate within the remaining intact neurological structures and the prevalence of long-term side effects such as tumor formation, among others. Additionally, how these implanted cells exert effects on the vasculature, surrounding neurons and glia, and the inflammatory process all remain to be elucidated.¹³

Challenge: Improving Preclinical Studies

The relative lack of success in translating therapeutic agents from bench to bedside resulted in the formation of the STAIR, an advisory board composed of numerous leaders in the field of stroke. The STAIR group identified weaknesses in preclinical models such as the nearly widespread use of young-adult, male animals; animals without comorbidity; and the need for functional assessment in addition to the more commonly reported histological measures.^30,98

Accounting for Age

Age is the greatest risk factor for stroke, yet it is rarely considered in preclinical models.^56,85 The aged brain is fundamentally different and responds to pathological insults in a detrimental manner, resulting in increased infarct volume and worsened functional outcome, when compared with young-adult animals.^18,24,33,85 This is particularly evident when considering inflammatory processes and astrocyte activation.^{4,74,75,82,107}

The aged brain is believed to exist in a state of chronic, low-grade inflammation that diminishes its ability to respond appropriately to pathological stimuli such as ischemia. Throughout normal aging, microglia, the primary immune cells of the brain, are primed following peripheral inflammation. This priming, characterized by morphological changes and by cell-surface protein expression, leads to increased proinflammatory molecule release following additional pathological insult.⁷⁴ Although the release of proinflammatory molecules results in a wide array of downstream effects beyond the scope of this review, the effect of aging on proinflammatory molecule expression has been well documented.¹⁰⁷ Furthermore, work in an aged animal model has shown differences, when compared with young-adult animals, in regulation of the JAK2/STAT3 pathway.²⁴ Notably, young-adult and aged animals differ in inflammatory response across a variety of time points poststroke. Badan et al.⁴ showed a gradual activation of microglia and astrocytes in the young-adult animal, peaking 14–28 days after ischemia. Alternatively, aged animals exhibit rapid activation of these glial cells, peaking within the 1st week postischemia. This is consistent with the findings of Dinapoli et al.,²⁴ in which morphological differences in astrocytes and microglia were seen when comparing young-adult and aged animals after stroke.

The mitigation of oxidative stress by using free-radical scavengers has proven to be successful in preclinical studies but failed to translate in clinical trials.^88,105 This may be related to methodology differences such as time of administration after stroke, type of stroke, delivery method, and others, or to differences in downstream effects associated with age. An example of age-related effects was seen with apocynin, an inhibitor of the NOX2 isoform of NADPH oxidase, which has contrasting effects when used in young-adult or aged animals. Apocynin was shown to be neuroprotective in young-adult animals but worsened stroke outcome in aged animals.⁴⁹

Although the precise downstream consequences of these differences remain to be elucidated, it is clear that age and the aging process fundamentally alter the response to pathological stimuli. To truly address the lack of translation from bench to bedside, and to increase the clinical relevance of preclinical models, using aged animals may prove fruitful.

Role of Gender

Another risk factor for stroke that is rarely considered is the role of gender. Female patients are more likely to be afflicted with ischemic stroke, yet the vast majority of preclinical studies are conducted in young-adult, male animals.⁸⁵ This fails to account for the role gender-related hormones may play in stroke outcome.⁵⁶ Although numerous studies have illustrated a neuroprotective effect of gender-associated hormones such as estrogen on stroke outcome, these studies have been conducted in young-adult animals.^68,106 There remains no indication that these same hormones, when administered to the aged animal, will be neuroprotective. In fact, clinical experience from hormone replacement therapy would indicate an increased stroke risk with estrogen replacement in the aged individual.¹⁰⁹

Occlusion Methodology and Combination Treatment

Methodology for inducing ischemia has been discussed at length in the literature, with each technique having distinct advantages and disadvantages.^10,26,57,79 Although suture occlusion remains the prevalent technique for inducing ischemia and is advantageous in its ease and reproducibility, the 100% reperfusion that occurs immediately on suture withdrawal lacks clinical relevance. Additionally, reperfusion in clinical stroke occurs with administration of tPA, and this is a gradual reperfusion as the thrombus is broken down. This reperfusion profile seen clinically is replicated in preclinical models by using a thrombus and subsequent tPA administration. Because of the failure to use the only approved pharmacological treatment for stroke in the vast majority of preclinical models, it has been difficult to study combination therapies. Although combination therapies result in potentially more intricate experimental paradigms, it seems unlikely that any potential neuroprotectant can be entirely successful without some degree of reperfusion. Similarly, the ischemic cascade, as described above, is a complex process and may require multiple steps to be targeted simultaneously. Therefore, it seems plausible that future studies need to investigate combinatorial therapy, particularly with administration of tPA to achieve reperfusion.⁵⁸

Treatment Administration

To increase the translation from bench to bedside, potential therapeutic agents need to be administered at extended time points in preclinical studies, time points more consistent with patient presentation. In the decade since regulatory approval of tPA, less than 5% of patients have received intravenous tPA. The lack of use is related primarily to time of presentation, in that only 46% of patients present within 3 hours, and there is an extensive list of contraindications.⁵³ This reality necessitates that therapeutic agents capable of extending the thrombolytic window that can be administered safely, perhaps by first responders, be developed.

Selecting Appropriate Outcome Measures

Preclinical models of stroke emphasize volumetric assessments of stroke outcome, whereas clinical assessment is based almost entirely on functional outcome. This discrepancy may play a role in the failed translation of therapeutic agents from bench to bedside. Furthermore, functional assessment may prove to be a more sensitive indicator of recovery in preclinical models, raising the possibility that compounds previously thought to be failures based on volumetric assessment may be advantageous functionally. Commonly used functional evaluations in preclinical work, such as the modified Neurological Severity Score, were developed in young-adult animals. If aged animals are to be used to increase clinical relevance, it will be important to obtain baseline assessments and to determine the most appropriate functional and behavioral outcome measures for aged animals, because aged animals may require a unique battery of tests for detailed functional outcome assessment.

Future Directions

A Shifting Approach—From Neurons to Neurovascular Unit

For years stroke research has emphasized neural survival, and hence the term “neuroprotection” was coined. Perhaps neuroprotection is simply not enough for clinical success, and rather the focus needs to shift toward full “cerebroprotection,” in which glial cells and the vasculature are also considered (Fig. 5).^21,58 Not only has the neurocentric approach failed to deliver clinically successful therapeutics, but it also disregards the fact that glia outnumber neurons and play an integral role in maintaining homeostasis in the healthy brain and in restoring neurological function in the injured brain. Furthermore, it has been shown that glial cells play an integral role in protecting neurons from ischemia, and can also lead to detrimental downstream consequences.¹⁰² In fact, the question of whether neurons can even survive in the absence of supporting glia has been asked and remains to be addressed, further necessitating the need to consider and investigate more thoroughly the role of glia in injury.

Fig 5.

The concept of neuroprotection is perhaps misguided; the neurocentric approach has failed to produce effective therapeutic agents. Cerebroprotection is needed, and for that the entire neurovascular unit must be considered. Also, supporting glia play a crucial role in health and disease, necessitating inclusion in therapeutic development.

Astrocytes, long recognized for having a role in neurological homeostasis via actions on neurotransmitter and ionic balance, have also been implicated in control of the brain’s microcirculation through signaling with neurons.¹¹⁶ Responding to synaptic release of glutamate from neurons, increases in intracellular calcium concentration within the astrocyte are observed, ultimately resulting in release of ATP, the primary signaling molecule through which astrocytes communicate with other cells of the CNS.⁸⁶ The release of ATP by astrocytes has been demonstrated following injury, and results in microglial activation. Considering the proximity of astrocytic processes to neural synapses and cerebral arterioles, forming a possible relay from neuron to vasculature, it is easy to envision how astrocyte compromise as a consequence of ischemia may be deleterious neurologically. Similarly, neuronal dysfunction following ischemia results in excitotoxicity via glutamate release, which may then signal microglial activation through the astrocytic response.

The role astrocytes play in homeostasis and response to injury is probably increased in the aged animal due to an increased number of astrocytes, as well as pericytes, with age. Pilegaard and Ladefoged⁸¹ suggested that there is an increase of 20% in astrocyte and pericyte population in the aged brain. Besides an increase in number, astrocytes in the aged brain exhibit hypertrophy and altered expression of surface markers, intracellular antigens, and growth factors. Notably, these differences exist in the absence of neurological injury, providing further indication that the aged brain is a fundamentally different entity in comparison with the young-adult brain, furthering the need to consider age in preclinical stroke models.¹⁸

Understanding the precise role of microglia in both acute and chronic periods after ischemia, and how microglia influence not only neuronal function but that of the neurovascular unit, is another area of investigation that may lead to successful therapeutic development. Microglia comprise approximately 12% of cells in the brain and potentially exert both detrimental and beneficial effects in the ischemic brain, necessitating further investigation and elucidation of these mechanisms.⁷ Whereas they generally exist in an inactivated state in the young-adult brain, normal aging has been shown to produce increasing numbers of activated microglia as well as phagocytic microglial subtypes.⁷⁴ Similarly, microglial reactivity is increased in response to numerous types of injury in the aged brain, ranging from neurotoxin (MPTP) exposure to traumatic brain injury and ischemia. Microglial activation ultimately results in the production of many pro-inflammatory factors such as prostaglandin E₂ (PGE₂), interleukin-1B (IL-1B), tumor necrosis factor–α (TNFα), nitric oxide (NO), peroxynitrite (NOO^–), superoxide (O₂*^–), and hydrogen peroxide (H₂O₂), furthering the damage initiated by the initial pathological stimuli.⁷

Besides release of inflammatory mediators, microglia probably play a role in oxidative stress-induced damage. As microglia are activated, cytosolic subunits of the NADPH oxidase enzyme translocate to the cell membrane and assemble the active form of the enzyme, resulting in superoxide production. In addition to the oxidative damage caused by extracellular ROS, such as superoxide, these molecules can often act intracellularly, serving to amplify the inflammatory response.⁷

The inflammatory response, mediated primarily through effects of astrocytes and microglia, represents a more logical target clinically because inflammation persists for an extended period and consists of beneficial and detrimental effects. Both beneficial and detrimental inflammatory processes can be targeted therapeutically through activation of beneficial aspects and inhibition of detrimental components. Similarly, evidence indicates that events occurring later in the ischemic cascade such as apoptosis and neuroinflammation account for most of the penumbral tissue death, which is the region of interest in therapeutic development.⁶⁰

Lack of therapeutic translation from bench to bedside may be partially explained by the neurocentric approach taken toward investigating ischemia and subsequent cell death. It is clear that there is continual communication between neurons and astrocytes, neurons and microglia, and neurons and vasculature. This communication, probably bidirectional in nature, is responsible for normal homeostasis as well as response to injury. The previous focus on neuronal response alone fails to consider the complex relationships among the various elements of the neurological system. Shortcomings of the field led to formation of the Stroke Progress Review Group by the National Institutes of Neurological Disorders and Stroke. One of the primary points of emphasis of this group is the need for cerebroprotection, in which the entire neurovascular unit and neurological structure are considered, rather than solely neuroprotection. Therefore, future work needs to address the temporality and mechanistic details of signaling events within the neurovascular unit.

Understanding Biphasic Signaling

A significant challenge in therapeutic development is the fact that various messenger molecules have divergent actions temporally. For example, MMPs have been tied to hemorrhagic transformation in the acute period following ischemic injury, yet also are essential for long-term recovery.¹⁴ Similarly, high-mobility group box 1 protein promotes necrosis and inflammatory cell infiltration immediately after ischemia, yet is influential in the recovery phase.³⁶ Not only is it important to identify these biphasic pathways but also to elucidate how pharmacological manipulation alters outcome. These signaling molecules and associated pathways are presented here as examples of the potentially biphasic nature of inflammation and its ability to serve both detrimental and beneficial roles, depending on the time period postinjury. This reality not only complicates target selection for possible therapeutic agents but also raises this question: If inhibition is possible in the acute phase, preventing further damage, will the beneficial activity still occur later on? Future studies are indicated to address these questions and to further identify similar molecules and associated pathways.

Conclusions

The dearth of success in the translation of proposed therapeutic agents for ischemic stroke from bench to bedside necessitates a reassessment of experimental methodology. Proposed therapeutic agents have too often targeted events occurring rapidly after the onset of ischemia, such as glutamate excitotoxicity, calcium release, and oxidative stress. Investigating pathological events occurring at more delayed periods, such as inflammation and apoptosis, may provide more realistic targets and consequently translate more effectively from bench to bedside. Notably, early results of minocycline clinical studies appear promising. Minocycline is a potentially promising therapeutic agent exerting its effects through modulation of inflammation, a pathological event that while initiated rapidly after ischemia, persists for hours to days poststroke. Therefore, minocycline may exhibit an extended therapeutic window in comparison to previous studies targeting pathological events limited to the acute period. Additionally, the protection afforded by minocycline is not neurocentric in nature and rather acts via effects on surrounding glia. Modulating the response of glia to injury, much like targeting the neurovascular unit, is more likely to translate to effective therapeutic development due to the large role glia play in both health and disease. Ischemic stroke is a vascular disease impacting all cellular types within the nervous system, not just neurons. Similarly, the only approved therapeutic agent, tPA, exerts its effects through the vascular system rather than by direct effects on neurons. This alone indicates the potential to improve outcome without a neurocentric focus.

Future studies must more closely replicate the clinical environment to increase the likelihood of translational success. Although the ideal preclinical model, target, and dosing strategy will remain unknown until successful translation to humans, a closer approximation to the clinical reality can be achieved in preclinical models. This includes incorporating aged animals into preclinical studies as well as animals with comorbidities typically seen in the clinical stroke population, such as diabetes and hypertension. Furthermore, models can be improved by incorporating tPA for combination therapy. Agents extending the window of opportunity for tPA administration would represent a significant improvement in stroke treatment.

Despite past failures, stroke research must press forward based on the immense need. By thoroughly reassessing reasons for the previous lack of success and pursuing new avenues and approaches to treatment, new advances are not just possible but likely.

Abbreviations used in this paper

AMPA

α-amino-3-hydroxy-5-methyl-4-isoxazole propionic acid

ATP

adenosine 5’-triphosphate

ATPase

adenosine triphosphatase

GABA_A

γ-aminobutyric acid-A

MMP

matrix metalloproteinase

MPTP

1-methyl-4-phenyl-2,5,6-tetrahydropyridine

NADPH

nicotinamide adenine dinucleotide phosphate (reduced form)

NMDA

N-methyl-D-aspartate

ROS

reactive oxygen species

SAINT

Stroke Acute Ischemic NXY Treatment

STAIR

Stroke Therapy Academic-Industry Roundtable

tPA

tissue plasminogen activator

VENUS

Very Early Nimodipine Use in Stroke

5-HT

5-hydroxytryptamine