Research Article: Ischemic stroke: experimental models and reality

Date Published: January 7, 2017

Publisher: Springer Berlin Heidelberg

Author(s): Clemens J. Sommer.


The vast majority of cerebral stroke cases are caused by transient or permanent occlusion of a cerebral blood vessel (“ischemic stroke”) eventually leading to brain infarction. The final infarct size and the neurological outcome depend on a multitude of factors such as the duration and severity of ischemia, the existence of collateral systems and an adequate systemic blood pressure, etiology and localization of the infarct, but also on age, sex, comorbidities with the respective multimedication and genetic background. Thus, ischemic stroke is a highly complex and heterogeneous disorder. It is immediately obvious that experimental models of stroke can cover only individual specific aspects of this multifaceted disease. A basic understanding of the principal molecular pathways induced by ischemia-like conditions comes already from in vitro studies. One of the most frequently used in vivo models in stroke research is the endovascular suture or filament model in rodents with occlusion of the middle cerebral artery (MCA), which causes reproducible infarcts in the MCA territory. It does not require craniectomy and allows reperfusion by withdrawal of the occluding filament. Although promptly restored blood flow is far from the pathophysiology of spontaneous human stroke, it more closely mimics the therapeutic situation of mechanical thrombectomy which is expected to be increasingly applied to stroke patients. Direct transient or permanent occlusion of cerebral arteries represents an alternative approach but requires craniectomy. Application of endothelin-1, a potent vasoconstrictor, allows induction of transient focal ischemia in nearly any brain region and is frequently used to model lacunar stroke. Circumscribed and highly reproducible cortical lesions are characteristic of photothrombotic stroke where infarcts are induced by photoactivation of a systemically given dye through the intact skull. The major shortcoming of this model is near complete lack of a penumbra. The two models mimicking human stroke most closely are various embolic stroke models and spontaneous stroke models. Closeness to reality has its price and goes along with higher variability of infarct size and location as well as unpredictable stroke onset in spontaneous models versus unpredictable reperfusion in embolic clot models.

Partial Text

On average every 40 s someone in the USA suffers a stroke, which drastically demonstrates the omnipresence and frequency of this devastating disease [92]. Acute cerebral stroke is caused in the great majority of cases by occlusion of a supplying arterial vessel, whereas vessel rupture with associated hemorrhage accounts for a minority of about 15% [92]. The extent of the resulting brain injury after focal cerebral ischemia depends on many factors such as the severity and duration of ischemia or collateral blood flow to mention the most important. This is in contrast to global cerebral ischemia, which mostly occurs in the setting of cardiac arrest, where—under normothermic conditions—relatively well-defined survival times of various neuronal subpopulations, glial and endothelial cells exist [53]. For the sake of clarity, this review will focus on experimental models of ischemic stroke due to focal cerebral ischemia only. Because of its heterogeneous etiology with a broad spectrum of manifestations, a large repertoire of models is required to address specific facets of this complex disorder. The deep gap between ischemic stroke in experimental models and the reality of stroke in human patients becomes painfully illustrated by the fact that with the exception of recanalization of the occluded vessel none of the hundreds of experimental neuroprotective strategies could be translated into the clinic up to now [61]. Nevertheless, many aspects of the pathophysiology of focal brain ischemia have been primarily identified in experimental models and could be confirmed later also in the setting of human stroke [90]. The development of the concept of the penumbra, the identification of spreading cortical depolarizations, the detection of post-stroke neurogenesis and the phenomenon of preconditioning are prominent examples of fruitful experimental stroke research [25].

What is the “reality” of ischemic stroke experimental models should depict? In humans, there are three different major causes for ischemic stroke. About 50% of cases are due to large vessel atherosclerosis and rupture of an atherosclerotic plaque, while about 20% are caused by cardioembolism. About 25% manifest as lacunar infarcts due to small vessel disease and probably occlusion of deep perforating arteries [9]. Some additional rare causes such as vasculitis or extracranial artery dissection account for the remaining 5% [129]. These percentages represent mean values over all age groups but change depending on the age of stroke victims. Cardioembolic stroke becomes the most frequent subtype with increasing age, while small vessel disease is rarely responsible in young people [121]. Furthermore, there are substantial differences in the distribution of stroke subtypes among different ethnic groups [113]. Finally, the percentages refer to ischemic stroke with known etiology, but one must keep in mind that a substantial number of cases are of undetermined cause, also referred to as cryptogenic stroke [114]. While atherosclerosis and cardioembolism frequently cause infarcts in both gray and white matter, lacunar infarcts are typically seen only in the subcortical white matter or the deep gray matter. Since these major subtypes of ischemic stroke already substantially differ concerning their RNA expression profiles in blood [62], it seems plausible that the molecular pathways and mechanisms of brain injury following ischemia may also be considerably different. In fact, this issue is currently unresolved. The neurological deficit and clinical presentation after ischemic stroke again exhibit a high variability with respect to the cause, duration, localization and severity of ischemia as well as age and comorbidity. One important fact in human thromboembolic stroke ignored in most available animal models is the occurrence of a substantial degree of spontaneous reperfusion, which is documented in up to 17% within the first hours after stroke onset [65]. However, reperfusion is mostly protracted but not prompt, thus leading to substantially different pathophysiologic pathways. While in the former situation the infarct core is irreversibly damaged and the penumbra is consumed within about 3 h, prompt reperfusion in animal models may partly rescue even core tissue with secondary delayed injury developing up to 3 weeks after stroke onset depending on the ischemic interval [55]. Peri-infarct blood flow may also be maintained by collaterals through the circle of Willis and/or leptomeningeal anastomoses. Clinically, stroke with acute onset of symptoms must be distinguished from transient ischemic attacks (TIAs) where neurological symptoms disappear within 24 h. In contrast to stroke, TIAs are thought to leave no damage of brain tissue, an opinion that may work conceptually but does not fully reflect reality [119]. In addition to acute neurological deficits that can be directly attributed to the affected brain areas, additional cognitive and psychiatric long-term consequences may emerge that are not readily assignable to the injured brain region. Cognitive decline is a major problem that is often present, in particular after lacunar stroke(s), and is probably superimposed to the underlying small vessel disease [85]. Post-stroke depression is another serious problem that develops in about one-third of patients with chronic stroke [105]. The biological basis remains uncertain, which is also due to a paucity of appropriate animal models [75]. Another example for frequent complications of chronic stroke widely ignored in animal models is obstructive sleep apnea [123]. Apart from neurological and neuropsychiatric consequences, stroke induces an immediate immune depression that is reflected by the fact that about 60% of stroke patients develop fever within 3 days after stroke onset [45]. A complex neuro-immunological connection consisting of the sympathetic nervous system, the hypothalamic-pituitary-adrenal axis and the vagus nerve has been identified as the biological basis [89].

It is clear that the complex situation of ischemic stroke cannot be modeled in an in vitro system with single cells or pieces of brain tissue with the absence of intact blood vessels and blood flow as well as the lack of infiltration of leukocytes. Nevertheless, in vitro models allow the investigation of specific basic biochemical and molecular mechanisms under conditions of energy deficiency similar to ischemia. The fundamental critical control points and molecular pathways of necrotic cell death, programmed cell death and autophagy are also amenable to direct study in vitro [52]. Another advantage of in vitro models is the possibility of high-throughput analyses, which becomes relevant with respect to testing novel potentially neuroprotective pharmaceuticals. In this context, the possibility to use human or humanized cells becomes increasingly important.

While in the beginning of experimental stroke research mainly higher species were used, this situation has completely changed within the last 3 to 4 decades. Today, mostly mice and rats are chosen for in vivo stroke models, which is easy to understand considering the lower costs of acquisition and keeping, simpler monitoring methods and tissue processing as well as ethical issues. The possibility to easily create transgenic animals is another major advantage of mouse models. However, one always has to keep in mind that the results of rodent models of stroke may only be half the truth. Due to the constant failure to translate experimental neuroprotective therapies into patients, the “stroke community” stands in the forefront of critically scrutinizing to what extent experimental models reflect reality. Starting with the STAIR recommendations 1999 [122], a series of landmark reviews considering this issue and providing recommendations to model stroke more realistically in the experimental setting has been published [26, 33, 36, 84]. In particular, there is the strong advice to reproduce successful stroke therapy in rodents in a higher species before starting clinical trials. The following paragraphs will briefly discuss the most important models with a focus on differences and similarities to the situations in humans (Table 1). For an extensive and excellent overview of currently available models with a focus on methodological aspects, see [23].Table 1Ischemic stroke: experimental models and realityExperimental modelClose to realityFar from reality/distorting realityIn vitro modelsPrincipal mechanisms and molecular pathways of cell death under ischemia-like conditionsAbsence of intact blood vessels and blood flowLack of infiltration of leukocytesEndovascular suture modelLocalization of the infarct (mostly MCAO), penumbra, blood-brain barrier injury, inflammatory processes and cell death pathways(Permanent and transient ischemia)No craniectomyLarge infarcts, mimics rather malignant infarction [16]Involvement of the hypothalamus with consecutive hyperthermia (rat) [79]Prompt reperfusion by withdrawal of the filament [55]Exception: mimics closely endovascular mechanical thrombectomyThromboembolism/thrombolysis not modeledAnesthesiaCraniectomy models with direct vessel occlusionPenumbra, blood-brain barrier injury, inflammatory processes and cell death pathways(Permanent and transient ischemia)Prompt reperfusion by reversal of the mechanical occlusion [55]Exception: mimics closely endovascular mechanical thrombectomyThrombembolism/thrombolysis not modeled CraniectomyAnesthesiaPhotothrombotic strokeSmall cortical infarcts and small subcortical infarcts(Permanent ischemia only)Recovery and plasticity mechanisms in chronic strokeModifications with stroke induction in freely moving rats and mice allow real-time analysis of a number of parameters in acute stroke without distortion through anesthesia [81, 136]Simultaneous development of cytotoxic and vascular edema with rapid breakdown of the blood-brain barrierNo penumbra (whether the “ring” model accurately models penumbra under discussion) [16]AnesthesiaEndothelin-1 modelInfarcts of variable sizes in nearly any brain regionSubcortical strokeRecovery and plasticity mechanisms in chronic stroke(Transient ischemia only)Minimal edema [112]Endothelin-1 and endothelin-1 receptors present also on neurons and astrocytes [94, 95]—may interfere with post-stroke recovery mechanisms [16]Thromboembolic clot modelsThromboembolic infarctsTransient ischemia with unpredictable time point of lysis of the embolusPathophysiology of embolic stroke including primarily cytotoxic edema superimposed later on by vasogenic edema with breakdown of the blood-brain barrier, presence of a penumbra, development of spreading depressions as well as an inflammatory responsePossibility to test thrombolytic therapies(Animal model)Microsphere models of embolic strokeThromboembolic infarcts(Permanent ischemia)(Mini-)penumbras, pathophysiology of ischemic cell death, inflammationPermanent ischemia without possibility of reperfusionMultiple vessels occludedCapillaries and arterioles are blocked resulting in redistribution of the blood flow and immediate disruption of the blood-brain barrier and vasogenic edema [132]Macrosphere models of embolic strokeThromboembolic infarcts(Permanent ischemia)Pathophysiology including penumbra, ischemic cell death, inflammationOcclusion can be postponed allowing to induce ischemia while the rat lies in an MRI or PET scannerPermanent ischemia without possibility of reperfusionSpontaneous stroke models: spSHR ratSubcortical infarcts(Small) vessel pathology(Animal model)

In particular, the persistent failure to translate successful neuroprotective strategies in rodents or even non-human primates into human patients suffering from ischemic stroke led to critical reflections about whether the animal models used are indeed suitable to cover the reality of human stroke [33]. Apart from possible experimental quality problems, the use of young, healthy, male rodents in the overwhelming number of experiments is a major distorting factor. Therefore, in the last years numerous attempts have been made to use animal models with the most frequent comorbidities such as diabetes mellitus, hypertension, atherosclerosis, hyperlipidemia, obesity or infection for stroke research. Further, aged animals and animals of both sexes have been used to capture reality more accurately. Age is the single most unmodifiable risk factor for stroke. From experimental studies with aged animals, it has become clear that neurological impairment increases, whereas the regenerative capacity is lowered compared to younger animals (for review, see [15]). There is growing evidence that the common link of age and the various modifiable risk factors mentioned above may be an elevated inflammatory profile resulting in a stroke-prone state [93, 108]. Sex is another important factor significantly affecting stroke incidence and outcome [51].

Non-human primates may have a lissencephalic brain, like the common marmoset and the squirrel monkey, or they may possess gyrencephalic brains, like the baboon, the Rhesus macaque and the Cynomolgus macaque, to mention the most relevant species in the context of stroke models. Of all primates, the macaque monkeys most closely resemble the human brain concerning anatomy with a specific cortical and subcortical organization, but most important for stroke research also with regard to vascular supply and collateralization [20]. Nevertheless, currently it is unclear which non-human primates most closely mimic the human conditions of stroke.

Although mechanisms of spontaneous recovery after stroke have been studied in both rodents and humans for a long time [27, 115], the frustrating translational failure of neuroprotective strategies in acute stroke gave new impetus toward studies on neuroregeneration. The challenges in models of post-stroke recovery differ from those in acute stroke models. Well-standardized lesions in defined brain areas are necessary to record the plasticity processes. Although principally the complete repertoire of stroke models can be used, the more suitable ones go along with low mortality and the possibility to produce small and standardized lesions. The peri-infarct cortex, closely connected ipsilateral areas but also contralateral and remote regions are of particular importance. Therefore, in particular the photothrombosis model and the endothelin-1 model have been used to analyze recovery processes. Although the principal events leading to post-stroke recovery are widely concordant in humans and animals, the kinetics differ considerably. While recovery processes in rodents are almost completed 4 weeks post-stroke, in humans they primarily occur within about 3 months but may even continue for years (for review, see [17, 138]). The exact reason for this difference is not known up so far [74].

On the one hand, ischemic stroke is principally a simple and well-defined pathological condition with interruption of blood flow and consecutive damage of dependent tissue. One the other hand, however, ischemic stroke is a very complex and heterogeneous disorder because of a multitude of modifying factors such as the duration and severity of ischemia, presence or absence of functioning collateral systems and an adequate systemic blood pressure, etiology and localization of the infarct as well as age, sex, comorbidities with multimedications and genetic background. Strictly speaking, stroke is not a single neurological disease but the manifestation of an underlying systemic problem such as atherosclerosis, inflammation or infection, which may cause infarcts just as well in other organs and manifest, e.g., as a heart attack. Experimental models of ischemic stroke are valuable tools to analyze specific facets of stroke more or less close to human stroke. Being aware of the limitations of the individual models and ideally having support from studies using human tissues are of utmost importance before drawing conclusions concerning stroke in humans. This issue is illustrated by the still persistent failure of successful translation of effective experimental neuroprotective strategies into patients suffering from this devastating disease.