Rodent ischemic stroke models and their relevance in preclinical research

Abstract

Stroke is a leading cause of death and disability worldwide, with ischemic stroke caused by an occluded vessel accounting for the majority of cases. Current treatments are limited to recanalization, either through thrombectomy or thrombolysis. Approved pharmacological interventions to suppress stroke‐associated excitotoxicity and neuroinflammatory events, leading to brain tissue death, are still lacking. Although numerous preclinical studies have been performed, they have yet to be translated into clinically relevant interventions. First‐line preclinical in vivo studies include the use of rodent ischemic stroke models, which vary in terms of how well they replicate human stroke pathophysiology and phenotype (including the formation of blood clot, blood–brain barrier disruption, neuroinflammation, and edema generation). Thus, rodent ischemic stroke models must be carefully chosen according to the specific pharmacological intervention to be tested. In this review, we aimed to provide an overview of the five most commonly used rodent ischemic stroke models and critically assess their advantages and limitations, with a primary focus on the acute phases of stroke.

Ischemic stroke is a leading cause of death and disability, with current treatments limited to recanalization techniques and no approved pharmacological interventions for excitotoxicity and neuroinflammation. This review critically examines the five most commonly used rodent ischemic stroke models, evaluating their ability to replicate human stroke pathophysiology and phenotype, particularly in the acute phases.

Highlights

• Rodent ischemic stroke models simulate key aspects of human ischemic stroke pathophysiology, but they do so to different degrees. Therefore, the choice of an appropriate model depends on the specific research goal at hand.

• Translational success in clinical trials has been limited, underscoring the need for improved model fidelity, standardized methodologies, and rigorous behavioral assessments.

• This review will guide researchers in the field towards the best‐suited model for a specific research question to improve the predictive value of their preclinical studies, and ultimately increase the likelihood of positive clinical outcomes.

• Future research should consider the impact of comorbidities on stroke outcomes to better mimic human stroke variability and pathophysiology in rodent ischemic stroke models.

1. INTRODUCTION

Stroke is the third most common cause of death and disability worldwide and poses a substantial societal burden. ¹ Globally, approximately five million stroke‐related deaths occur annually, with an additional five million surviving stroke patients suffering from long‐term disabilities. ² Stroke can broadly be classified into two different categories, namely as hemorrhagic or ischemic. Hemorrhagic stroke is caused by the rupture of a blood vessel, resulting in bleeding into the brain parenchyma, whereas acute ischemic stroke occurs when a blood vessel becomes occluded, thus compromising blood flow to tissue regions beyond the occlusion. Decreased blood flow leads to a lack of nutrients and oxygen in the affected brain area, causing energy failure and, ultimately, cell death and infarct formation. ³ With the ischemic stroke accounting for approximately 87% of stroke incidences, this review focus on this predominant category. ¹

Current stroke treatments during the hyperacute phase are limited to recanalization via thrombolysis (with human recombinant tissue plasminogen activator (rh‐tPA)‐alteplase or tenecteplase) or endovascular thrombectomy (surgical clot removal), both of which require early (within hours) intervention. ⁴ To date, despite decades of research, pharmacological interventions to curb stroke‐associated excitotoxic events leading to brain tissue death and ultimately improve motor and cognitive skills are nonexistent. Despite tremendous efforts to understand the mechanisms underlying acute cell death after stroke and to promote long‐term recovery of the affected brain tissue, ⁵ findings are yet to be translated into pharmacological interventions. Most recently, a series of clinical phase III trials (referred to as ESCAPE‐NA1, ESCAPE‐NEXT, and FRONTIER) with the peptide NA‐1 (also known as nerinetide) ⁶ , ⁷ have placed the field at an exciting crossroads. Thus, the trials have ushered in both innovative trial design and new approaches for combination therapy, namely thrombolysis in combination with a neuroprotective agent. ⁸ Moreover, although significant work has been undertaken to establish the mechanisms associated with acute cell death, there is an urgent need to understand the mechanisms underlying recovery to promote the discovery of regenerative therapies. Pharmacological therapies such as the chemokine receptor 5 antagonist, maraviroc are now being investigated in the clinic to improve recovery. ⁹

To study the mechanisms underlying ischemic stroke and identify relevant targets for human stroke pathology, rodent models, in conjugation with modeling systems such as human cell models, are generally useful and relevant for modeling complex pathologies, despite their nonhuman origin. The advantages of rodent models include the following: (1) Ethical approval for rodent models is easily obtained compared to that for nonhuman primates; (2) Rodents have a rapid reproduction rate, ensuring an adequate supply and allowing for the quick development of transgenic models; (3) Ischemic infarct formation is more consistent in rodents than in humans because there is less variation in genetics, anatomical location, and manifestation. This allows for the precise analysis of the induced stroke pathophysiology and for the study of the possible neuroprotective effects of drugs; (4) Human and rodent brain vasculatures contain architectural features similar to those of other commonly used laboratory animals. However, rodents lack the complex cortical patterning and architecture of humans; therefore, some researchers have established models for large animals; and (5) molecular, genetic, and biochemical investigations often require invasive access to the brain tissue, which can only be accomplished using animal models.

But although rodent models are useful, to date, no preclinically tested drug candidates, except for rh‐tPA, ¹⁰ have been successfully translated into approved stroke treatments. This is regardless of whether the drug is administered as a single therapy or combination therapy. ¹¹ , ¹² This poor preclinical to clinical translation may, to some extent, rely on the choice of the specific animal model and/or the study setup, as well as the choice of behavioral studies to capture potential sensory, motor, and/or cognitive improvements resulting from the administration of a drug candidate. To improve the translational value and quality of preclinical stroke research, the Stroke Therapy Academic Industry Roundtable (STAIR) and the Stem Cell Therapies as an Emerging Paradigm in Stroke (STEPS) recommendations for experimental studies were devised. ¹³ These recommendations from leading stroke experts include the elimination of assessment and randomization bias in studies and basing sample sizes on power analyses. Furthermore, while initial evaluation can be conducted in young healthy male animals, the findings should be validated in females, aged animals, and animals with comorbidities. ¹³ , ¹⁴ Together with the Animal Research: Reporting of in vivo Experiment (ARRIVE) guidelines, which encourage scientists to conduct transparent preclinical research and emphasizes rigorous reporting of results, the STAIR and STEPS recommendations should be considered when performing preclinical stroke studies. ¹³ , ¹⁵ However, although the STAIR and STEPS recommendations offer strict criteria to improve the cross‐national potential of hyperacute stroke neuroprotection research, not all recommendations are suitable for recovery research. Therefore, the Stroke Recovery and Rehabilitation Roundtables (SRRR) and the International Stroke Recovery and Rehabilitation Alliance (ISRRA) were formed. ¹⁶ , ¹⁷

In this review, we provide detailed accounts of the pathophysiological phenotypes of the five most commonly used experimental rodent models of ischemic stroke and, when feasible, relate each model event to the human pathophysiological phenotype. While our focus is primarily on the hyperacute and acute phases of ischemic stroke, where appropriate, stroke recovery is also considered. Moreover, for each of the models, we delve into their respective advantages and limitations, thereby underlining that there is no “one‐size‐fits‐all” animal model for preclinical stroke research. In doing so, we hope to provide benchmark information and guidelines for researchers in the field. For new researchers, this review can be used as a guide to select the appropriate animal model for a specific research question, taking into account the many different molecular aspects and lessons learned from previous studies. For more experienced researchers, it provides an updated and informative overview of commonly used methods. Using the review as a reference will not only improve the predictive value of research conducted in the field but may also increase the likelihood of positive clinical outcomes.

2. SETTING THE STAGE: ISCHEMIC STROKE PATHOPHYSIOLOGY AND PHENOTYPE DURING THE HYPERACUTE AND ACUTE PHASE

An ischemic stroke occurs when a vessel supplying blood to the brain is obstructed. It is a complex pathological condition that causes injury to the affected brain tissue, with associated cell death, inflammation, and ultimately functional and cognitive dysfunction. ¹⁸ The immediate area affected by the occluded vessel, the ischemic core (Figure 1A), experiences irreversible tissue damage minutes after the initial injury. ¹⁸ The adjacent tissue surrounding the ischemic core, the penumbra, is less affected by the occluded vessel as the blood supply is maintained by collateral vessels. ¹⁹ The penumbra is considered a salvageable tissue; but, it relies on timely interventions to restore blood flow (via thrombolysis or thrombectomy) and minimize the extent of cell death (with neuroprotectants). ⁸ However, if the penumbra is left untreated, potentially salvageable tissue becomes part of the infarct core, increasing the risk of further functional and cognitive dysfunction. ²⁰ Further distinction between the penumbra and peri‐infarct regions should be noted. Thus, the penumbra is electrophysiologically silent, whereas the peri‐infarct region is electrophysiologically active. ²¹ While this is highly relevant information when designing experiments and interpreting observations, the differences between the terms lie outside the scope of this review.

Figure 1.

Classification of the penumbra and ischemic core, the main pathophysiological phenotypes, and the ischemic stroke phases. (A) Overview of the ischemic core and penumbra in human ischemic stroke with common pathophysiological mechanisms, namely excitotoxicity, blood–brain barrier (BBB) leakage and edema, and neuroinflammation. (B) The initial stage following obstruction of blood flow, that is, the hyperacute phase, is characterized by excitotoxity, edema, and rapid cell death, ultimately forming the ischemic core, which expands into surrounding penumbral tissue regions. The hyperacute phase is quickly replaced by the acute phase, a phase characterized by neuroinflammation and blood–brain barrier breakdown. The subacute phase marks a change in the phenotype where the glial scar has fully formed and corresponds to a period of time during which the use of regenerative and repair interventions is the most promising as endogenous plasticity mechanisms are heightened. Finally, in the chronic phase of ischemic stroke, neurogenesis and angiogenesis are pronounced.

The effects of ischemic stroke stretch over the course of several months, with cellular changes occurring simultaneously or at different time points after an ischemic event (Figure 1B). ²² Immediately following the ischemic insult (approximately 6 h), the hyperacute phase is marked by a series of neuronal changes and pathological processes involving all cells of the neurovascular unit. ¹⁸ , ²⁰ The initial hyperacute phase (the immediate hours post‐insult) is followed by the acute phase, which is characterized by severe neuroinflammation and blood–brain barrier (BBB) breakdown. ²³ This phase is then replaced by the subacute phase approximately 1 week after the initial ischemic insult. ²³ During this period, a change in neuroinflammation is observed, with a switch to more neuroprotective inflammatory events. ²³

How well rodent models of ischemic stroke mimic human processes highly depends on the phase of injury. In general, the hyperacute phase in both rodents and humans is highly correlated, as evidenced by the translation of rh‐tPA into clinical use. However, as we gradually progress through the phases, the temporal time window for when various biological and physiological changes occur differs markedly, especially during the chronic phase. ²² For instance, in humans, 6 months post‐injury is equivalent to approximately 1–2 months in rats and slightly less in mice. However, to improve the predictive nature of our preclinical studies, correlative studies must be conducted to better define the phases of stroke injury based on pathophysiological changes and biomarker readouts. Temporal differences in stroke progression need to be considered when designing and interpreting experiments, or we risk continuing to misinform the clinical trial design.

3. EXCITOTOXICITY

Obstruction of blood flow to the brain acutely initiates several detrimental events at the molecular level, leading to excitotoxicity and culminating in ischemic cell death and infarct formation. This involves disturbances in key events centered on the overstimulation of glutamate receptors, Ca²⁺ influx, and subsequent downstream intracellular events involving kinases, phospholipases, proteases, and reactive oxygen and nitrogen species (Figure 2). A detailed overview of the cellular mechanisms involved have been offered in previous reviews. ²⁵ Here, it is important to highlight how central excitotoxic events have been studied in rodent models of ischemic stroke. Thus, animal studies have provided important validation of the mechanisms involved and have helped focus on drug target selection, for example, with the identification of central players such as glutamate receptor ionotropic N‐methyl d‐aspartate 2B, postsynaptic density protein 95, calcium/calmodulin‐dependent protein kinase II, and reactive nitrogen species scavengers. ⁷ , ²⁴ , ²⁶

Figure 2.

Simplified overview of excitotoxic damage after ischemic stroke. Obstruction of blood flow results in lack of cellular energy and increased neuronal depolarization. This, in turn, leads to a buildup of glutamate in the synaptic cleft, which overstimulates glutamate receptors, like the N‐methyl‐d‐aspartate acid (NMDA) and the α‐amino‐3‐hydroxy‐5‐methyl‐4‐isoxazolepropionic acid (AMPA) receptor. This overstimulation causes an influx of calcium (Ca²⁺) and sodium (Na⁺) ions, creating an osmotic gradient resulting in cell swelling and edema. Enhanced concentrations of Ca²⁺ will disrupt Ca²⁺‐dependent pathways, affecting enzymes such as phospholipases, proteases, and kinases, including CaMKII. Overactivation of phospholipases and proteases will impact the structural integrity of the cell, as well as overstimulate enzymes responsible for producing reactive oxygen species. The elevated levels of Ca²⁺ and reactive oxygen species cause mitochondrial damage, leading to apoptosis. Together, excitotoxicity, oxidative stress, and apoptosis promote ischemic cell death and culminate into the formation and expansion of the infarct that occurs in conjunction with inflammation involving microglia. Figure modified from previous work by Griem‐Krey et al. ²⁴

While increased neuronal activity (excitotoxicity) is detrimental during the hyperacute phase of stroke, it should be noted that increased neuronal activity is required to promote functional recovery at later time points. Rehabilitative modalities (e.g., physical therapy, transcranial magnetic stimulation, and pharmacological intervention) that enhance sensory and motor maps and stimulate changes in the axonal structure that underpin recovery are dependent on patterned neuronal activity. ²⁷ Since post‐stroke neural repair and recovery involve neuronal sprouting and remapping of sensory, motor, and language centers, it has been shown preclinically that changes in neuronal excitability underlie neural repair. ²⁸ , ²⁹ Understanding how rehabilitative interventions and associated mechanisms work is not possible without the aid of preclinical focal stroke models, which highlight the importance of choosing the correct model for assessing a specific pathological event.

4. NEUROINFLAMMATION

During the hours immediately following the ischemic insult, the hyperacute phase, dying neurons and glial cells release damage‐associated molecular patterns, which are quickly recognized by the brain‐resident immune cells and microglia, causing them to become activated. ³⁰ Damage‐associated molecular patterns also activate pericytes and vascular endothelial cells, leading to BBB disruption. ³¹ Activated microglia release reactive oxygen species (ROS), cytokines, and chemokines, which act on adjacent astrocytes, pericytes, and vascular endothelial cells. ³² Activated astrocytes also release cytokines and chemokines that trigger vascular endothelial cell activation and compromise BBB integrity. ³³ At the same time, ischemia causes astrocytic death, contributing to further BBB disruption through the downregulation of the tight junction proteins occludin, claudin‐5, and zonula occludens‐1, ³⁴ with all of them exacerbating neuronal injury and contributing to the development and expansion of the ischemic core. After the initial occlusion, an acute phase occurs during which the infarct core is fully established and expands into the penumbra. The infiltration of circulating leukocytes (neutrophils and monocytes/macrophages) then begins to play a more prominent role. ³⁰ , ³⁵ Microglia and astrocytes continue their release of cytokines and chemokines, leading to further recruitment and infiltration of neutrophils and macrophages. ³⁰ , ³⁶

Repair mechanisms are initiated a few hours after the onset of stroke; however, the brain undergoes the most significant improvements from days to months after the insult during the subacute and chronic phases, respectively. ³⁷ This includes the stabilization of excitatory and inhibitory connections, suppression of neuroinflammation, dendritic growth, BBB repair, axonal sprouting, angiogenesis, and formation of new synapses. ³⁸ Microglia may play a dual role as they phagocytose dead cells and debris, ³⁹ as well as viable ischemic neurons. ⁴⁰ In addition, microglia may increase the production of anti‐inflammatory cytokines and promote the recruitment of oligodendrocyte progenitor cells for myelin preservation and axonal remyelination. ⁴¹ The multiple roles of microglia during neuroinflammation make them attractive drug targets, although the other cellular players, the astrocytes and neutrophils, could hold just as relevant a therapeutic potential. ⁴² , ⁴³ , ⁴⁴ Secondary cell death also occurs during this period, particularly in white matter tracts where a progressive loss of axons occurs in response to reactive astrocytes, thereby playing a major role in the development of cognitive impairment. ⁴⁵ , ⁴⁶ This highlights the need to target astrocytes and inflammation during the hyperacute and acute phases as well as the subacute and chronic phases.

5. BRAIN EDEMA

Brain edema is often a part of the pathophysiology of ischemic stroke and is associated with high mortality rates. ⁴⁷ The main cause of brain edema is alterations in the BBB, which lead to an influx of water. Severe brain edema typically results from large vessel occlusion. ⁴⁸ Excitotoxicity and neuroinflammation contribute to BBB alterations, resulting in brain edema. Other contributors are matrix metalloproteinases (MMP) released from pericytes and neutrophils, particularly MMP2 and MMP9, which mediate the degradation of the basement membrane and tight junction proteins. ⁴⁹ Moreover, astrocytic aquaporin‐4 expression increases and interacts with sulfonylurea receptor 1 and the transient receptor potential melastatin 4 (SUR1‐TRPM4) complex, which regulates sodium flow, thereby resulting in further exacerbation of brain swelling. ⁵⁰ In addition, reperfusion initiated by rh‐tPa or thrombectomy may cause or exacerbate brain edema because the restoration of blood flow initiates the transport of blood immune cells to the site, aggravating BBB breakdown. Brain edema is treated with hypertonic mannitol infusion, which may be associated with severe complications. ⁵¹ However, ongoing clinical trials are investigating inhibitors of aquaporin‐4 and SUR1‐TRPM4 as treatment options for brain edema. ⁵² , ⁵³ A more comprehensive description of brain edema following ischemic stroke can be found in other reviews, for example by Han et al. ⁴⁹

6. RODENT ISCHEMIC STROKE MODELS

The commonly used rodent models of ischemic stroke include intraluminal or transcranial middle cerebral artery occlusion (MCAO), thromboembolic stroke, photothrombotic stroke (PTS), and endothelin‐1 (ET‐1)‐induced stroke. The middle cerebral artery (MCA) and its subsequent branches are the most prevalent sites for ischemic stroke, accounting for 70% of all clinically diagnosed infarcts. ⁵⁴ Correspondingly, modeling the occlusion of the MCA is the most widely explored approach in preclinical ischemic stroke studies. ⁵⁵ Different rodent models and their advantages and limitations in representing human ischemic stroke pathophysiology during the hyperacute and acute phases are reviewed in the subsequent sections. Moreover, the pathophysiologies and phenotypes are summarized for all models in Table 1.

Table 1.

Pathophysiologies and phenotypes of commonly used rodent ischemic stroke models.

Rodeent stroke models	Stroke pathophysiology and phenotype
	Physiological clot	Penumbra	Blood‐brain barrier permeability	Neuroinflammation	Brain edema
Intraluminal MCAO	×	√	√	√	√
Transcrianial MCAO	×	√	?	√	?
Thromboembolic stroke	√	√	√	?	√
Photothrombotic stroke	√	?	√	√	√
Endothelin‐1 induced stroke	×	√	?	?	?

×, not present; √, results indicating its presence; ?, inconclusive or inconsistent results; MCAO, middle cerebral artery occlusion.

7. INTRALUMINAL MIDDLE CEREBRAL ARTERY OCCLUSION

The intraluminal MCAO model can mimic either permanent or transient ischemia; the latter is followed by reperfusion. ⁵⁵ The technique involves temporary or permanent occlusion of the common carotid artery by introducing a suture directly into the internal carotid artery and advancing the suture until it disrupts the blood supply to the MCA. ⁵⁶ However, close care should be taken, as vasospasms can occur, which may lead researchers to believe that complete bifurcation has been achieved when only partial occlusion has been performed. Common durations of transient MCAO range from 30 to 120 min with longer occlusion times, resulting in larger lesion volumes. ⁵⁵ Intraluminal MCAO in rodents leads to well‐defined infarcts in a substantial proportion of the hemisphere, involving the striatum, frontoparietal, and temporal cortices, whereas infarcts in the thalamus and hypothalamus are rarely obtained. ⁵⁷ However, longer occlusion times can result in lesions that may affect both the thalamic and hypothalamic functions. Hypothalamic ischemia in rats results in spontaneous hyperthermia, which should be taken into consideration, as hyperthermia itself may influence infarct size ⁵⁸ and, more importantly, the ability to thermoregulate, which impacts mobility, feeding, and ultimately survival. However, in human MCA strokes, infarcts are rarely observed in the hypothalamus. Moreover, hemorrhagic transformation, which exacerbates injury, is common in this model. The resulting infarct size may also differ according to the strain, as observed when comparing the ischemic lesion volume and edema in Wistar and Sprague–Dawley rats subjected to MCAO. ⁵⁹ Moreover, the gut microbiota may influence ischemic lesion volume and cause species differences in inflammation and stroke outcome. ⁶⁰ Such species differences may be relevant when considering commonly used preclinical ischemic stroke models.

The intraluminal MCAO model is robust and mimics human ischemic stroke in terms of infarct core and penumbra. ⁶¹ , ⁶² In both humans and MCAO rodent models, the ischemic core and penumbra are formed immediately following the acute ischemic stroke and continue to develop over time due to excitotoxicity and inflammation, even after reperfusion. ⁶³ , ⁶⁴ The core and penumbra can be distinctively defined using molecular biology methods such as immunohistochemistry and imaging, including computed tomography and magnetic resonance imaging. ⁶² , ⁶⁵ , ⁶⁶

8. TRANSCRANIAL MIDDLE CEREBRAL ARTERY OCCLUSION

The transcranial MCAO model differs from the intraluminal MCAO model in that the MCA is accessed and occluded through a cranial window. This is achieved by separating the parotid gland and the temporalis muscle and transecting the zygomatic arch before performing a small burr hole craniotomy to expose the MCA. The MCA can then be permanently occluded using electrocoagulation or transiently occluded using a suture or microaneurysm clip. ⁶⁷ Transcranial MCAO produces smaller infarcts when compared to intraluminal MCAO. Depending on the size of the lesion, it will affect somatosensory regions (the temporal cortex, preferentially), although it can extend and encompass the frontal, parietal, and rostral occipital regions as well as the underlying white matter. ⁶⁸ An additional advantage of this model over the intraluminal model is that infarcts do not affect the thalamus or the hypothalamus. ⁵⁵ Other advantages include the reproducibility of infarct size, neurological deficits, and a low mortality rate. Long‐lasting neurological deficits may be difficult to assess after MCAO; however, it is suitable for investigating biological processes, such as neurogenesis and axonal sprouting. ⁶⁹ , ⁷⁰

A notable disadvantage of the transcranial MCAO model is the need for a craniotomy, which carries the risk of injuring the underlying cortex or causing vessel rupture during drilling or electrocoagulation, which is generally laborious and requires expert skills. Drilling and electrocoagulation may also affect subsequent neuroinflammation and edema. Moreover, this procedure may affect the BBB function, as it has been shown that craniotomy itself may be the main cause of brain edema as well as increased BBB permeability to albumin and Evans Blue. ⁷¹

In summary, intraluminal and transcranial MCAO models closely resemble the ischemic stroke pathogenesis reported in human patients in terms of reproducing the core and penumbra regions, with neuronal death caused by excitotoxicity as well as cerebral inflammation, BBB damage, and subsequent edema, which collectively result in behavioral deficits. ⁷² However, distinct differences are notable between the two MCAO models. Transcranial MCAO usually results in a larger infarct core with a smaller penumbra, making it less suitable for investigating neuroprotective interventions. Meanwhile, the intraluminal MCAO model is commonly used to assess neuroprotective agents. However, given the size of the lesions produced by the intraluminal MCAO model, multiple cortical and subcortical regions are affected, making it difficult to truly evaluate lasting impairments and repair.

A significant drawback of MCAO models is that they do not mimic actual blood clot formation, thereby excluding studies investigating the effect of rh‐tPA. Cognitive, motor, and sensorimotor deficits are captured in MCAO models, with intraluminal MCAO generally resulting in larger lesions and more pronounced impairments. The transcranial MCAO model results in well‐established sensorimotor impairments but milder motor impairments than those in the intraluminal model. These impairments can be studied using the versatile toolbox of behavioral tests, such as the novel object recognition test to assess cognition, the foot fault test for motor function, and the adhesion removal test for sensorimotor function. ⁷³ , ⁷⁴ It is important to keep in mind that some behavioral tests are better suited for certain rodent species.

The disease phenotype described in MCAO animal models contributes to neuroinflammation and is further influenced by inflammatory responses. ²² , ⁷⁵ Neuroinflammation has been extensively investigated in rodent models. ⁷⁵ , ⁷⁶ , ⁷⁷ However, patient data are limited. ⁷⁶ The inflammatory profile in humans has been studied primarily using serum samples from patients to quantify cytokines ⁷⁶ , ⁷⁸ and infiltrating immune cells ⁷⁹ , ⁸⁰ , ⁸¹ or by gene expression studies. ⁸⁰ , ⁸² , ⁸³ Although human studies have provided valuable insights into post‐stroke neuroinflammation, the information is primarily limited to the hyperacute phase, with patients receiving treatment within the first 4 h after hospital admission. Although similarities have been observed between rodent MCAO models and human patients, the exact degree of similarity is debated owing to uncertainties in the temporal dimension of neuroinflammation as well as rodent‐to‐human discrepancies. ⁷⁷

9. THROMBOEMBOLIC STROKE

The thromboembolic stroke model is surgically challenging to perform. It is based on the principle of spontaneously formed or thrombin‐induced clots from autologous or allogenic blood. ⁸⁴ , ⁸⁵ The reported clot morphology differs depending on the type of the thromboembolic stroke. When thrombin is injected into the MCA, a fibrin‐rich clot with limited cell and platelet content is obtained. ⁸⁶ However, when collagen is injected into the MCA, a clot rich in fibrin and platelets is obtained. ⁸⁷ When autologous or allogenic blood clots are introduced into the MCA, the clot contains a rich mixture of cells, platelets, and fibrin. ⁸⁸ , ⁸⁹ This clot composition is highly comparable to that observed in human patients with ischemic stroke ⁹⁰ and reflects interpatient variance. ⁹¹ Blood clots may also be modified with thrombin and fibrinogen to produce more homogeneous clots. ⁹² However, as the model targets the MCA, it affects somatosensory regions and not strictly motor function and cognition, similar to the intraluminal and transcranial MCAO models, making it difficult to assess long‐lasting functional changes.

The stability and composition of clots are key factors influencing stroke outcomes. ⁹³ Therefore, the thromboembolic stroke model allows for the study of different clinically relevant clot subtypes and closely resembles the vascular occlusion observed in the majority of human ischemic strokes. ⁹³ This further allows for detailed studies of thrombolytic compounds, for example, in combination with neuroprotective drug candidates, such as in a recent nerinetide clinical trial. ⁶ However, the volume and exact location of the infarct largely depend on the elasticity of the clot, ⁹² which decreases reproducibility when compared to the MCAO and the PTS models. Furthermore, the infarct volume is smaller and more variable in the thromboembolic model than in the MCAO and PTS models. ⁹⁴ In the thromboembolic stroke model, unaided reperfusion of microvessels is typically observed within 1 h of clot formation, with cortical vessels opening after 3 h and striatal vessels reperfusing within 24 h. ⁹⁵ This reperfusion somewhat resembles that in human patients, albeit occurring much faster. ⁹⁶

The ischemic core and penumbra can be readily defined using thromboembolic stroke models. The size and temporal characteristics of the core and penumbra differ from those observed in intraluminal MCAO models when investigating cerebral blood flow and diffusion using magnetic resonance imaging ⁸⁸ ; however, they are still comparable to the phenotype reported in patients. ⁹⁷ BBB breakdown has been observed to a similar extent in thromboembolic models such as intraluminal MCAO models and in patients. ⁴⁸ , ⁹⁸ , ⁹⁹ Some evidence suggests that BBB breakdown is caused by alterations in tight junction protein composition, ¹⁰⁰ whereas other studies suggest that mechanical endothelial disruption is the main cause of BBB breakdown. ⁹⁸ , ¹⁰¹ BBB breakdown mechanisms have also been reported in PTS models. ⁹⁸ , ¹⁰⁰ , ¹⁰¹ However, the exact cause of BBB breakdown in patients with ischemic stroke remains unclear. Increased levels of tight junction proteins have been observed in the serum of patients, pointing towards the disruption of tight junction protein composition during BBB breakdown. ¹⁰²

Brain edema is a detrimental consequence of BBB breakdown. The severity of brain edema has been classified in patients using data extracted from the international register Safe Implementation of Treatments in Stroke. ¹⁰³ Severity can also be mimicked in a thromboembolic stroke model by modifying the blood clot dimensions, with an increased length resulting in increased water content in the brain. ¹⁰⁴ This classification of brain edema cannot be modeled using either the MCAO or PTS model. ¹⁰⁵ , ¹⁰⁶ Moreover, the duration of post‐embolization anesthesia may affect brain water content in the thromboembolic stroke model. ¹⁰⁷ Excitotoxicity and neuroinflammation have not been comprehensively studied in thromboembolic stroke models, in contrast to the MCAO and PTS models. Limited published data support the presence of ROS, downstream mediators of glutamate excitotoxicity, and the activation of inflammatory cells following thromboembolic stroke. ¹⁰⁸ , ¹⁰⁹

10. PHOTOTHROMBOTIC STROKE

In the PTS model, cerebral blood vessels are blocked by clot formation, which occurs via platelet aggregation in response to photosensitive dye activation. ¹¹⁰ Commonly used photosensitive dyes include Rose Bengal and erythrosin B, which are injected intraperitoneally in mice ¹¹¹ or intravenously in rats. ¹¹⁰ After a few minutes of brain exposure, green spectrum light is delivered via either a cold light source or a laser shone through either the intact skull in mice or a thinned skull in rats over the cortical regions of interest (commonly, the sensorimotor and prefrontal regions). ¹¹⁰ The production of oxygen radicals leads to endothelial damage, platelet activation, and platelet aggregation in the pial and intraparenchymal vessels. ¹¹² Rapid progression of ischemic cell death is observed, resulting in well‐defined ischemic lesions in highly specific cortical regions. Compared to both MCAO models, the PTS model requires minimal surgical intervention. Depending on the light source used to induce the stroke (cold light vs. laser), the stroke is classified as either a multi‐vessel or single‐vessel occlusion. This method is particularly useful in investigating small lacunar strokes. ¹¹³

The PTS model primarily serves as a permanent‐occlusion model. However, recent advances have resulted in the establishment of a reperfusion model that is responsive to rh‐tPA. ¹¹⁴ The clot originating from the introduction of Rose Bengal stain alone is not fibrin‐rich and is therefore unresponsive to standard thrombolytic agents. To this end, research groups have established a PTS model in which thrombin is mixed with Rose Bengal before administration, resulting in a fibrin‐rich clot that responds to rh‐tPA administration. ¹¹⁴ The PTS model has also been modified to induce stroke in awake animals. ¹¹⁵

Following the induction of PTS, rapidly progressing ischemic damage and subsequent endothelial injury lead to the early development of both intracellular (cytotoxic) and extracellular (vasogenic) edema. ¹¹⁶ This rapidly evolving ischemic lesion renders the PTS model suboptimal for screening neuroprotective compounds. However, the power output and light intensity can be changed, which slows the progression of stroke and vasogenic edema, thereby allowing the screening of neuroprotective compounds. In contrast, primary cytotoxic edema is followed by vasogenic edema in stroke patients. ⁵⁵ , ¹¹⁷

Induction of PTS results in a well‐defined ischemic core; however, some studies have reported a lack of or minimal penumbra with limited local collateral flow. ¹¹⁵ , ¹¹⁷ Consequently, recent modifications to the PTS model have included the use of a circular laser beam that can produce a ring zone surrounding the ischemic core. ¹¹⁸ The ring zone is characterized by local hypoperfusion mimicking the penumbra. ¹¹⁹ Less intense photoirradiation for a prolonged time is also a strategy to induce a penumbra. ¹²⁰ Depending on the methodological setup, this model may not be appropriate for testing neuroprotective interventions that solely target otherwise potentially salvageable penumbral tissue. ¹²¹ Although usually, this model is not used to assess protection, it is widely applied to investigate mechanistic changes occurring during the repair phases (over days, weeks, and months) after stroke.

BBB breakdown has been observed in both the ischemic core and the penumbra following PTS; ¹²² however, the underlying cause of BBB breakdown is not fully understood. One study reported that changes in pericytes and tight junction coverage of brain capillaries caused breakdown, ¹²³ whereas another study suggested that upregulated transcytosis was the cause of greater BBB permeability. ¹²²

Distinct differences in post‐stroke inflammatory components are observed between the PTS and MCAO models. Microglial activation and phagocytosis are present at far earlier stages following MCAO compared to PTS. ¹²⁴ Similar results were observed for astrocytic gliosis, with reactive astrocytes being present 1 day after MCAO but absent after PTS. ¹²⁴ In contrast, the recruitment of peripheral immune cells appears to be more pronounced after PTS than after MCAO. ¹²⁴ , ¹²⁵ Lymphocytes were found to be present to a much larger extent following PTS. ¹²⁵ The production of pro‐inflammatory cytokines was found to be comparable in the MCAO and PTS models overall, or slightly higher in the PTS model. ¹²⁴ , ¹²⁵ As such, there are similarities between the PTS and MCAO models, both of which replicate the human neuroinflammatory phenotype to a certain extent, particularly with respect to lymphocyte infiltration and cytokine production.

In summary, the PTS model represents a far less invasive alternative to the MCAO models, yet resembles human ischemic stroke in terms of ischemic core, BBB breakdown, and inflammation. PTS is a beneficial model for studying platelet aggregation, neuroinflammation, and behavioral deficits, which can be cognitive, motor, and/or sensorimotor deficits, depending on the cortical area to which the infarct is introduced. ¹²⁶ , ¹²⁷ Human ischemic stroke is commonly associated with MCA occlusion, which is not observed in the PTS‐ or ET‐1‐induced models. However, in humans, the MCA feeds on specific motor nuclei and striatal motor aspects, whereas an occluded MCA in rodents first hits the somatosensory system. Thus, to target specific motor nuclei, the PTS and ET‐1‐induced models may, in some cases, be preferred over the MCAO models.

11. ENDOTHELIN‐1‐INDUCED STROKE MODEL

In the ET‐1‐induced stroke model, vasoconstriction of blood vessels in the brain occurs because of the administration of ET‐1, an endogenous compound produced by endothelial cells. ¹²⁸ It has been observed that the plasma level of ET‐1 is significantly higher in acute ischemic stroke patients compared to healthy controls and is correlated with severe cerebral edema. ¹²⁹ , ¹³⁰ However, in humans, endogenous ET‐1 is not the actual cause of ischemic stroke, as observed in an ET‐1‐induced stroke model. Similar to the PTS model, the ET‐1 model is versatile, as ET‐1 can be administered to different brain regions, such as the perivascular, intracortical, or intraventricular space, thereby modeling different affected areas. ¹³¹ , ¹³² , ¹³³ , ¹³⁴ It can even be administered in conscious animals if a guide cannula is surgically implanted, thereby eliminating any confounding effects of anesthesia. ¹³⁴ A dose‐dependent increase in infarct size was observed after ET‐1 administration; however, the infarct volume was highly variable within the study groups. ⁷² , ¹³⁴ , ¹³⁵ Additionally, the infarct volume must be corrected for tissue swelling caused by edema following ET‐1 administration, which makes the assessment of brain edema post‐stroke in this model challenging. ¹³⁶ Moreover, reperfusion subsequently occurs in this rodent model with the natural removal of ET‐1, thereby mimicking reperfusion in patients with stroke, where spontaneous reperfusion occurs in 80% of cases within 2 weeks after stroke onset. ⁹⁶ However, as the ET‐1 model is a multivessel occlusion model, like the PTS model, it is difficult to obtain a deep understanding of how reperfusion occurs.

Due to the lack of a physiological clot, the effects of rh‐tPa cannot be investigated in this rodent model, whereas neuroprotective drug candidates have been successfully investigated in the ischemic penumbra. ¹³² , ¹³⁴ , ¹³⁷ A significant drawback of this model is that ET‐1 has direct paracrine effects on oligodendrocyte differentiation and maturation, thereby complicating the results of post‐stroke white matter biology studies. ¹³⁸ , ¹³⁹ Since then, several groups have modified the model using either a mixture of ET‐1 and N(5)‐(1)‐iminoethyl‐l‐ornithine HCl (l‐Nio) or only l‐Nio, as this compound has no effect on the white matter. ¹⁴⁰ , ¹⁴¹ Therefore, the ET‐1 model remains a valuable tool for assessing post‐stroke repair and recovery.

The temporal profile of BBB breakdown in the ET‐1 model differs from that in both the MCAO and PTS rodent models, as BBB breakdown is slightly delayed in the ET‐1 model. ¹²² , ¹²³ , ¹³¹ , ¹³² Ischemic stroke strongly correlates with BBB breakdown. ¹⁴² In addition, similar to the transcranial MCAO model, the procedure in the ET‐1‐induced stroke model may be associated with BBB damage caused by craniotomy and cannula insertion, ⁷² as marked BBB breakdown has been observed immediately following ET‐1 administration. ¹³¹ , ¹³² This can be avoided by careful surgical procedures and the delivery of ET‐1 using a fine glass capillary pipette to avoid disturbing the dura. ¹⁴³ Despite the inconsistent results regarding BBB breakdown, evidence suggests the presence of certain inflammatory cells in the ET‐1‐induced stroke model. Twenty‐four hours after stroke, an increase in the number of reactive astrocytes, microglia/macrophages, and infiltrating neutrophils is observed, with the number of neutrophils being proportional to the infarct size, ¹⁴⁴ similar to what is observed in the MCAO models but not in the PTS model. ¹²⁴ , ¹³² However, there are still uncertainties related to the inflammatory response in the ET‐1‐stroke model, rendering further investigations essential before comparison with human pathology can be made. ¹⁴⁵

12. CONCLUDING REMARKS

Ischemic stroke is a detrimental disease with no pharmacological treatment options, except for the thrombolytic agent rh‐tPa. Rodent models that mimic human diseases are crucial for investigating novel drug candidates. However, specific models must be carefully selected to ensure they sufficiently capture the pathophysiology and underlying mechanisms relevant to the drug under test. However, preclinical ischemic stroke research does not always result in high predictive validity, as exemplified by the poor translation of preclinical NA‐1/nerinetide studies into clinical trials. ⁶ In this review, five different rodent ischemic stroke models and their pathophysiologies are described and compared to the human ischemic stroke phenotype, which displays a high degree of interindividual differences. Depending on the stage of drug development, different guidelines may be applied to ensure good animal research. The STAIR recommendations and ARRIVE guidelines apply in stroke research when moving from bench to bedside. ¹⁴⁶ The STAIR framework recommends including comorbidities in preclinical investigations of animals with ischemic stroke because of the highly heterogeneous pathophysiology often associated with comorbidities. ¹³ In fact, the prevalence of comorbidities in patients with ischemic stroke patients is 75%–99%. ¹⁴⁷ , ¹⁴⁸ In particular, comorbidities such as diabetes and hypertension increase the risk of ischemic stroke, as do risk factors such as age, Westernized diet, and lack of physical activity. ¹⁴⁹ This highlights the importance of conducting stroke research in animals with comorbidities to better model human diseases. Additionally, the STAIR recommendations highlight the importance of behavioral testing to investigate physiological outcomes after a stroke. Thus, when assessing behavior in rodents, it is important to choose an appropriate behavioral model to study functional outcomes following ischemic stroke. This is largely dependent on the model used to mimic ischemic stroke, as recently analyzed in detail by Ruan and Yao. ¹⁵⁰

Preclinical stroke research is complex; thus, so far, its translational value has been inadequate. In this review, we provide an overview of the advantages and limitations of the most commonly used preclinical stroke models to aid in choosing the best fit for a specific drug candidate. This may improve future preclinical research, with the potential for translation into novel treatment options for patients with ischemic stroke that could ultimately limit their post‐stroke cognitive and functional disabilities.

AUTHOR CONTRIBUTIONS

Maria Thaysen: Conceptualization; visualization (lead); writing—original draft (lead); writing—review and editing (equal). Emil Westi: Conceptualization; visualization (lead); writing—original draft (lead); writing—review and editing (equal). Andrew N. Clarkson: Writing—original draft; writing—review and editing (equal). Petrine Wellendorph: Funding acquisition (lead); conceptualization (lead); visualization; writing—original draft; writing—review and editing (equal). Mie Kristensen: Funding acquisition (lead); conceptualization (lead); visualization; writing—original draft; writing—review and editing (equal).

CONFLICT OF INTEREST STATEMENT

Consultant Andrew N. Clarkson is a member of Neuroprotection editorial board and is not involved in the peer review process of this article.

ETHICS STATEMENT

Not applicable.

Thaysen M, Westi E, Clarkson AN, Wellendorph P, Kristensen M. Rodent ischemic stroke models and their relevance in preclinical research. Neuroprotection. 2024;2:296‐309. 10.1002/nep3.62

DATA AVAILABILITY STATEMENT

Data sharing is not applicable to this article as no new data were created or analyzed in this study.