Stroke-Related Animal Models: Methodologies, Applications, and Translational Perspectives

Chenglong Shi; Mengyuan Xu; Yanmei Li; Lei Zhao; Haitao Xu; Lei Xian; Zhenyu Wang; Longjie Qin; Xiao-li Min

doi:10.3791/70147

Review Article

Stroke-Related Animal Models: Methodologies, Applications, and Translational Perspectives

DOI:

10.3791/70147

⸱

May 19th, 2026

Chenglong Shi^*¹^,² , Mengyuan Xu^*³ , Yanmei Li^*⁴ , Lei Zhao¹^,² , Haitao Xu¹^,² , Lei Xian¹^,² , Zhenyu Wang¹^,² , Longjie Qin¹^,² , Xiao-li Min¹^,²

¹Department of Psychiatric Sleep Medicine Center, the Second Affiliated Hospital of Kunming Medical University, ²Department of Neurosurgery, the Second Affiliated Hospital of Kunming Medical University, ³Department of Neurology, the Second Affiliated Hospital of Kunming Medical University, ⁴Department of Rehabilitation, the Second Affiliated Hospital of Kunming Medical University

^* These authors contributed equally

Summary

$$\rightleftharpoonup{xx}$$ $$\longleftharp{xx}$$, $$\longrightharp{xx}$$,

This review surveys major stroke animal models, evaluating their technical principles, strengths, and limitations. Emphasis is placed on rigorous, clinically aligned design and translational relevance. Advancing comorbidity-integrated platforms, biomarker-guided assessment, and cross-species validation will be essential to improve preclinical-to-clinical success in stroke therapy development.

Abstract

$$\rightleftharpoonup{xx}$$ $$\longleftharp{xx}$$, $$\longrightharp{xx}$$,

Stroke remains a leading cause of death and disability worldwide. Despite hundreds of neuroprotective agents showing promise in preclinical studies, nearly all have failed in clinical trials, largely due to inconsistencies between animal models and the human condition. This review provides a comprehensive overview of experimental stroke models, encompassing ischemic stroke (middle cerebral artery occlusion, photothrombosis, thromboembolic models, endothelin-1–induced vasoconstriction, global ischemia), hemorrhagic stroke (intracerebral hemorrhage, subarachnoid hemorrhage, epidural hematoma), cerebrovascular disease-related models (intracranial aneurysm, arteriovenous malformation), spontaneous stroke-prone models (hypertension, cerebral amyloid angiopathy), and special-condition models (neonatal hypoxic–ischemic encephalopathy, chronic cerebral hypoperfusion, post-stroke complications such as epilepsy, depression, dysphagia, and cognitive impairment). We systematically summarize the technical approaches, critical parameters, advantages, and limitations of each model, highlighting their applications in studying neuroprotection, reperfusion, hematoma expansion, inflammation, blood–brain barrier disruption, and rehabilitation strategies. In line with international guidelines such as STAIR and ARRIVE, we emphasize the importance of rigorous study design, including control of species, strain, sex, age, and comorbidities, as well as outcome measures aligned with clinical scenarios. Future directions include developing hybrid and comorbidity-integrated models to reflect the heterogeneity of human stroke, implementing cross-species validation, incorporating imaging and blood biomarkers into preclinical workflows, and advancing translational platforms for novel drugs and devices. Bridging the gap between animal modeling and clinical trial design will be crucial for accelerating the discovery of effective stroke therapies.

Introduction

$$\rightleftharpoonup{xx}$$ $$\longleftharp{xx}$$, $$\longrightharp{xx}$$,

Stroke is a leading cause of death and long-term disability worldwide, imposing a substantial social and healthcare burden. The latest Global Burden of Disease results indicate that in 2021, there were approximately 11.9 million incident strokes; of these, 65.3% were ischemic, 28.8% intracerebral hemorrhage (ICH), and 5.8% subarachnoid hemorrhage (SAH)¹^,². Although reperfusion therapies—intravenous thrombolysis with recombinant tissue plasminogen activator (rt-PA) and mechanical thrombectomy—have transformed acute care, only a minority of patients currently receive or benefit from these treatments³^,⁴. Moreover, many neuroprotective approaches effective in animal studies have failed in clinical trials, highlighting a persistent translational gap⁵. A key contributor is that success is measured differently across studies and settings, ranging from acute histological injury to long-term functional recovery, and increasingly incorporating longitudinal and multimodal phenotyping.

Animal models are indispensable in this translational continuum. Since the mid-20^th century, diverse preparations have advanced understanding of cerebrovascular pathophysiology, the ischemic cascade, hemorrhagic dynamics, and neuroinflammation and repair. They serve three principal purposes: (i) to delineate cellular and molecular mechanisms of ischemic and hemorrhagic injury; (ii) to screen and optimize neuroprotective and neurorestorative strategies; and (iii) to provide bridging platforms for endovascular, rehabilitative, and systems-of-care interventions⁶^,⁷. Importantly, model choice should be driven by the clinical scenario and the primary endpoint of interest (e.g., reperfusion biology, hematoma-related toxicity, early brain injury, or long-term behavioral outcome), rather than procedural convenience. Given the marked heterogeneity of human stroke—shaped by age, sex, comorbidities, vascular anatomy, and etiologic subtype—no single model can fully recapitulate clinical reality; complementary models are required to emulate specific scenarios⁶^,⁷.

The evolution of stroke models has closely tracked clinical needs. Early global-ischemia paradigms (e.g., four-vessel occlusion) were developed to study hypoxic–ischemic encephalopathy (HIE) and post-cardiac-arrest injury⁸. The intraluminal filament middle cerebral artery occlusion (MCAO) model introduced by Koizumi and refined by Longa enabled reproducible permanent or transient focal ischemia without craniectomy, greatly accelerating research aligned with large-vessel occlusion (LVO) and reperfusion⁹^,¹⁰. Subsequent focal ischemia paradigms diversified to balance experimental control versus thromboembolic realism: thromboembolic models facilitate evaluation of thrombolysis and antithrombotics¹¹; photothrombotic methods generate sharply demarcated cortical infarcts suited to circuit-level and rehabilitation studies¹²; and endothelin-1 (ET-1) microinjection produces spatially targeted, reversible vasoconstriction for modelling small-vessel disease and deep lesions¹³. In hemorrhagic stroke, collagenase-induced and autologous-blood ICH models, together with endovascular perforation (EVP) models of SAH, have advanced understanding of hematoma expansion, blood-product neurotoxicity, early brain injury (EBI), and vasospasm¹⁴^,¹⁵^,¹⁶^,¹⁷. More recently, special-condition and complication-focused paradigms—including neonatal HIE, chronic hypoperfusion/vascular cognitive impairment (VCI), and post-stroke systemic complications—have been used to better reflect clinical diversity and recovery trajectories¹⁸^,¹⁹.

Still, challenges remain. Anatomical and physiological differences between rodents and humans limit direct translation and motivate complementary studies in large animals and, selectively, non-human primates (NHPs)²⁰. Between-laboratory variability in surgical technique, occlusion duration, physiological monitoring, and outcome measures undermines reproducibility; meta-research highlights deficits in randomization, blinding, sample-size planning, and transparent reporting²¹^,²². In addition, the predominance of young, otherwise healthy animals and short-term readouts in preclinical studies contrasts with the clinical reality of aging, sex-specific biology, comorbidity burden, and immune–brain interactions that shape both acute injury and long-term recovery. To improve rigor and translational alignment, the Stroke Therapy Academic Industry Roundtable (STAIR) recommendations emphasize prespecified power, randomization/blinding, clinically meaningful endpoints, incorporation of comorbidities, multi-model validation, and cross-laboratory replication; ARRIVE 2.0 provides detailed reporting standards. Consistent implementation of these practices is essential to strengthen reproducibility and translational validity²³^,²⁴^,²⁵. Against this background, a systematic synthesis of stroke-related animal models is warranted. In this review, we categorize ischemic, hemorrhagic, special-condition, and bridging models; distill cross-model trade-offs (construct validity, reproducibility, and endpoint alignment); and provide a concise model-selection framework—supported by a structured comparative table—to guide model choice according to research objectives and translational intent.

Access restricted. Please log in or start a trial to view this content.

Review and Perspective

$$\rightleftharpoonup{xx}$$ $$\longleftharp{xx}$$, $$\longrightharp{xx}$$,

Methodological approaches to stroke modeling

The methodological development of stroke animal models has spanned more than half a century. The core objective is to reproduce, as faithfully as possible, the pathophysiological processes of human stroke under experimental conditions while making explicit the trade-offs between construct validity, experimental control, and endpoint alignment. In practice, model selection should start from the clinical scenario and the primary readout, then balance feasibility and reproducibility. According to stroke type and clinical scenario, commonly used models can be grouped into four categories: ischemic models, hemorrhagic models, special-condition models, and large-animal and NHP models. Table 1 summarizes decision-relevant attributes (mechanistic focus, typical lesion profile, variability, and best-fit use cases) to support goal-directed model choice.

Ischemic stroke models

Ischemic stroke models aim to reproduce the loss of cerebral perfusion caused by LVO, distal emboli, microvascular occlusion, or systemic hypoperfusion. They are commonly used to interrogate infarct core–penumbra evolution, reperfusion injury, neuroprotection, and recovery. Because no single paradigm captures all clinical features, the key practical distinction is whether the model prioritizes experimental control (e.g., lesion topography and timing) or thromboembolic realism (e.g., clot biology and recanalization). Principal paradigms include intraluminal filament MCAO, thromboembolic models, photothrombosis, ET-1 vasoconstriction, and global ischemia/hypoxia–ischemia (HI).

Middle cerebral artery occlusion

Among ischemic paradigms, the intraluminal filament MCAO remains the most widely used and versatile small-animal technique. Originally described by Koizumi and refined by Longa, it enables permanent or transient focal ischemia without craniectomy, making it a practical workhorse for mechanistic and interventional studies⁹^,¹⁰. In essence, a filament transiently interrupts MCA inflow, and occlusion duration serves as the main lever to tune injury severity and to model reperfusion²⁶^,²⁷. Consequently, MCAO has been pivotal for delineating infarct core–penumbra dynamics and reperfusion injury biology, and it provides a commonly used experimental analogue of acute LVO followed by vessel reopening²⁸.

Over time, refinements have focused less on procedural novelty and more on quality control and reproducibility. Physiological monitoring and confirmation of a robust cerebral blood flow reduction during occlusion (and recovery on reperfusion) are widely used to detect technical failure and reduce between-animal variability²⁹^,³⁰^,³¹. Across laboratories, transparent reporting of key parameters (animal characteristics, anesthetic/temperature control, occlusion duration, and exclusion criteria) is often more important for comparability than minor procedural variants²⁹^,³⁰^,³¹.

Limitations are equally well recognized. MCAO is operator-dependent and may lead to vascular injury and hemorrhagic complications, and off-target damage or dysregulated thermophysiology can confound outcomes if not monitored³²^,³³. Moreover, collateral anatomy and strain differences contribute to variability in infarct size even under nominally identical conditions³⁴. Finally, because MCAO is a mechanical occlusion rather than a thrombus-driven event, it is not optimal for directly testing thrombolysis or clot-targeting strategies; thromboembolic paradigms are better suited when clot biology and recanalization efficacy are central³⁵^,³⁶. Taken together, MCAO is best positioned for: (i) controlled studies of focal ischemic pathobiology, and (ii) evaluation of adjunctive neuroprotection in the reperfusion era—provided that monitoring, prespecified criteria, and reporting rigor are in place.

Thromboembolic models

Thromboembolic paradigms reproduce the clinically relevant sequence of intravascular clot occlusion and spontaneous or therapy-induced recanalization, thereby offering high face validity for cardioembolic and selected atherothrombotic strokes. Their distinctive value is that clot biology and recanalization dynamics are built into the injury model, making them the preferred choice when the research question centers on thrombolysis, antithrombotic therapy, distal embolization, no-reflow, or hemorrhagic transformation. Contemporary preparations include (i) injection of preformed autologous clots into the internal carotid artery or directly into the MCA to generate large-artery embolic occlusion in rodents; (ii) in situ thrombin-triggered clot formation at the MCA origin, which enables tight control of occlusion timing and systematic testing of rt-PA; and (iii) endovascular, imaging-verified variants compatible with angiography, LDF, and MRI readouts³⁵^,³⁷^,³⁸^,³⁹^,⁴⁰. Across these variants, confirmation of occlusion/recanalization (flow- or image-based) is central to interpreting efficacy endpoints and adverse events³⁵^,³⁷^,³⁸^,³⁹^,⁴⁰^,⁴¹.

Model choice within this class should be driven by the desired balance between thromboembolic realism and experimental controllability. Clot composition and delivery strategy influence occlusion stability, lesion topography, and lytic susceptibility³⁵^,³⁸^,³⁹^,⁴⁰. Rather than enumerating all preparation parameters, a practical decision point is whether the study requires recanalization responsiveness: autologous or fibrin-rich clots enable thrombolysis studies, whereas inert embolic agents (e.g., microspheres) provide dose-defined shower embolization suited to modelling microinfarct burden and delayed lesion evolution, but are largely irreversible and therefore less informative for thrombolysis efficacy⁴²^,⁴³^,⁴⁴^,⁴⁵^,⁴⁶. Upstream carotid injury models (e.g., FeCl₃-triggered thrombosis with downstream embolization) can support antithrombotic screening but typically yield greater heterogeneity in intracranial target vessels and lesion distribution⁴⁷^,⁴⁸. For device development and peri-procedural biology, large-vessel embolic models in rabbits and swine uniquely enable fluoroscopic guidance and thrombectomy using human-scale catheters, complementing rodent thromboembolism for pharmacologic testing and systems-of-care questions⁴¹^,⁴⁹.

The chief limitations of thromboembolic models are variability and control. Compared with intraluminal filament MCAO, heterogeneity in clot lodging site, fragmentation, spontaneous lysis, and off-target embolization increases dispersion in infarct volume and neurological outcomes and can introduce multifocal lesions that complicate mechanistic inference³⁷^,³⁸^,³⁹^,⁴². Accordingly, studies benefit from prespecified inclusion/exclusion criteria (e.g., flow drop thresholds, angiographic occlusion confirmation), defined therapeutic windows, and harmonized rt-PA dosing and monitoring across species/strains, together with transparent reporting aligned with ARRIVE/STAIR principles³⁷^,³⁹. When rigorously standardized, thromboembolic paradigms provide one of the most clinically concordant platforms for testing reperfusion pharmacotherapy and for dissecting recanalization-related injury mechanisms.

Photothrombosis

Photothrombosis combines systemic delivery of a photosensitizer (classically Rose Bengal) with focal cortical illumination to induce rapid, spatially confined microvascular thrombosis and circumscribed infarction. Introduced by Watson and colleagues, the method is minimally invasive (often transcranial), technically accessible, and yields highly reproducible cortical lesions without craniectomy, making it widely used for studies that require precise lesion targeting and low between-animal variability¹². Within the broader model landscape, photothrombosis represents a high-control paradigm: it prioritizes spatial reproducibility and lesion geometry over thromboembolic realism and reperfusion physiology. Contemporary refinements—including standardized illumination conditions with strain-specific calibration and flow/imaging confirmation—further improve control over lesion size and location⁵⁰. Spatially refined variants extend their utility, ranging from vessel-targeted photothrombotic injuries to single-vessel occlusion approaches that model microinfarcts with near cellular-scale resolution⁵¹^,⁵²^,⁵³^,⁵⁴. Collectively, these features confer excellent reproducibility and spatial targeting, with clinical correspondence to cortical microinfarcts and small, circumscribed infarcts.

Practically, the primary design choice is the desired spatial scale and depth of injury (territorial cortical lesion vs. microinfarct vs. single-penetrator occlusion), rather than procedural fine-tuning. Protocol parameters (photosensitizer dose, illumination geometry/exposure, skull preparation, and real-time flow confirmation) modulate infarct size and topography⁵⁰, but for most applications, it is sufficient to report these variables transparently and to verify lesion placement/extent with standardized readouts. Several adaptations have been introduced to better approximate evolving or heterogeneous lesions, including paradigms that enlarge the peri-lesional vulnerable zone and ring-shaped illumination designs that generate a structured stroke-in-evolution geometry⁵⁵^,⁵⁶. Depth-targeted photothrombosis (e.g., two-photon–guided vessel occlusion) enables mechanistic dissection of microvascular failure, collateral limits, perivascular inflammation, and blood–brain barrier (BBB) disruption with high spatial fidelity⁵³^,⁵⁷.

Intrinsic limitations temper translational scope. Because photothrombosis drives immediate platelet-rich microvascular occlusion with early BBB disruption and minimal collateral recruitment, it typically yields little or no physiologic penumbra and responds poorly to rt-PA, limiting its utility for thrombolysis or reperfusion studies⁵⁰^,⁵⁸. Lesions are largely cortical, with limited flexibility for deep territories unless specialized approaches are used⁵²^,⁵³^,⁵⁴. Accordingly, photothrombosis is best positioned for studies where reproducible cortical topography is essential—such as cortical microinfarcts, circuit plasticity, rehabilitation-relevant remodeling, and BBB pathology—rather than for paradigms predicated on salvageable penumbra, large-vessel recanalization, or device-based reperfusion.

Endothelin-1 vasoconstriction

The ET-1 vasoconstriction model induces focal ischemia by stereotaxic microinjection of ET-1 adjacent to target cerebral vessels or into defined parenchymal regions (e.g., cortex, striatum, internal capsule)²⁰^,⁵⁹^,⁶⁰^,⁶¹. ET-1 activates ETA/ETB receptors on vascular smooth muscle and endothelium, causing rapid, dose-dependent luminal narrowing and a marked fall in rCBF, typically followed by gradual, spontaneous reperfusion as the peptide dissipates²⁰^,⁵⁹^,⁶⁰^,⁶². Within the ischemic-model spectrum, ET-1 is a precision targeting paradigm: it is most useful when lesion location (including subcortical/white-matter–predominant injury) and reproducible behavioral phenotypes are the primary objectives. Accordingly, ET-1 approaches can generate subcortical or internal-capsule–centered injuries that yield consistent sensorimotor deficits and are often used as experimental analogues of lacunar syndromes⁶¹^,⁶³^,⁶⁴. Compared with intraluminal filament MCAO, ET-1 modeling is technically straightforward and minimally invasive (typically requiring only a small burr hole), and it can be implemented in awake or lightly anesthetized paradigms to reduce anesthesia-related confounds when behavioral recovery is a key endpoint²⁰^,⁵⁹^,⁶⁰.

For most study designs, the major practical design choice is the intended territory (cortical–striatal vs. deep motor pathways) rather than the specific delivery variant. ET-1 can be targeted perivascularly in the MCA territory to approximate cortical–striatal injury or delivered to deep structures (e.g., internal capsule) to model focal motor pathway damage with persistent deficits⁶⁰^,⁶¹^,⁶²^,⁶⁵. While injection dose and volume modulate the depth and duration of vasoconstriction, aggressive dosing can increase variability and mortality, so titration and prespecified inclusion/exclusion criteria are critical for reproducibility⁶². Species and strain differences also affect responsiveness, with mice often showing attenuated vasoconstriction relative to rats, underscoring the need for pilot calibration and transparent reporting of coordinates, dose/volume, anesthesia/temperature control, and physiological monitoring²⁰^,⁵⁹^,⁶⁰^,⁶⁶^,⁶⁷. Where feasible, flow confirmation (or standardized imaging-based lesion verification) strengthens interpretability and cross-study comparability²⁰^,⁵⁹^,⁶⁰^,⁶⁸.

Limitations reflect the model’s underlying pathophysiology. ET-1 induces vasospasm rather than intraluminal thrombosis, and therefore does not capture clot-driven occlusion, thrombolytic responsiveness, or device-mediated recanalization²⁰^,⁵⁹^,⁶⁰. Peptide diffusion can contribute to off-target vasoconstriction and non-ischemic effects, and the resulting ischemic territory may exhibit a restricted or atypical penumbra compared with embolic large-vessel occlusion²⁰^,³⁷^,⁵⁹^,⁶⁰. Taken together, ET-1 is best positioned for mechanistic and rehabilitation-oriented studies that require anatomically specific cortical or deep lesions (including white-matter–predominant injury), while being less suitable for studies centered on clot biology, thrombolysis, or endovascular reperfusion. When paired with rigorous monitoring and transparent reporting aligned with STAIR/ARRIVE principles, ET-1 remains a valuable platform for modelling small-vessel disease–relevant phenotypes and circuit-specific recovery²⁰^,⁶⁰^,⁶¹^,⁶⁴.

Global ischemia/Hypoxia-ischemia

Global ischemia models interrogate brain injury resulting from systemic hypoperfusion or oxygen deprivation and are foundational for studies of post–cardiac arrest encephalopathy and neonatal HIE. Within the stroke-model landscape, their defining feature is a whole-brain (or forebrain-predominant) ischemic insult with reperfusion, making them best suited to questions of selective neuronal vulnerability, global reperfusion injury, and system-level neuroprotection (e.g., targeted temperature management), rather than territory-specific large-vessel occlusion biology. In adult rodents, transient forebrain ischemia is commonly induced by vertebral/carotid occlusion paradigms or by carotid occlusion combined with controlled hypotension, which produces delayed neuronal death in hippocampal CA1 and other vulnerable regions and recapitulates key features of cardiac arrest–reperfusion injury⁸^,⁶⁹^,⁷⁰^,⁷¹^,⁷². Mouse adaptations of forebrain ischemia and standardized protocols have improved feasibility and monitoring in small animals⁷³^,⁷⁴^,⁷⁵. Because collateral anatomy strongly shapes injury patterns, species and strain selection is not a mere logistical choice but a determinant of lesion distribution and reproducibility⁷⁴^,⁷⁶. For perinatal injury, the Rice–Vannucci paradigm (unilateral carotid ligation followed by systemic hypoxia in early postnatal life) remains the most widely used small-animal model of neonatal HIE and supports mechanistic and interventional studies across developmental stages¹⁸^,⁷⁷^,⁷⁸.

Methodological refinements have expanded versatility and rigor by emphasizing physiological control and standardized readouts over procedural variation. In adult global ischemia, consistency in ischemia duration, blood pressure targets, temperature control, and verification of cerebral hypoperfusion (flow- or electrophysiology-based) reduces variability and mortality⁸^,⁷⁰^,⁷⁵. Cardiac arrest/cardiopulmonary resuscitation (CA/CPR) models introduce defined no-flow and low-flow intervals to interrogate postarrest pathophysiology and temperature-based neuroprotection, and simplified induction approaches have lowered technical barriers in mice⁷⁹^,⁸⁰^,⁸¹. In neonatal HI, injury severity is commonly titrated by hypoxia intensity and exposure duration (with careful control of temperature), enabling systematic study of white-matter vulnerability, inflammation, and long-term neurobehavioral trajectories that align with clinically relevant HIE outcomes⁷⁷^,⁷⁸.

Limitations reflect both biology and technique. Mortality can be substantial in severe CA/CPR or prolonged global ischemia, and injury topography varies with collateral anatomy, systemic physiology, anesthesia, and temperature control⁸^,⁷¹^,⁷²^,⁷³^,⁷⁴^,⁷⁵^,⁷⁶. Accordingly, global ischemia and neonatal HI studies are particularly sensitive to methodological drift across laboratories, reinforcing the need for prespecified physiological targets, standardized monitoring, and transparent reporting. By design, these models lack large-artery thrombus biology and are therefore not suitable for evaluating thrombolysis or device-mediated recanalization; instead, they are best aligned with mechanistic studies of reperfusion injury and selective vulnerability, and with interventions such as hypothermia/targeted temperature management in postarrest encephalopathy and neonatal HIE, where hypothermia remains a clinical standard for moderate–severe cases⁸¹^,⁸². In sum, global ischemia and neonatal HI paradigms complement focal LVO and embolic models by addressing distinct, clinically important scenarios that are otherwise poorly captured in territory-based stroke preparations.

Hemorrhagic stroke models

Hemorrhagic stroke encompasses ICH, SAH, and epidural hemorrhage (EDH, less commonly studied in stroke research), which pose distinct biological and translational challenges compared with ischemic stroke, including hematoma-related mass effect, blood-product toxicity, and secondary inflammatory injury. ICH and SAH, therefore, represent the principal focus of preclinical hemorrhagic modeling.

Intracerebral hemorrhage models

Among hemorrhagic stroke paradigms, ICH models are central for dissecting hematoma formation/expansion, mass effect, perihematomal edema, and blood-derived neurotoxicity. Accordingly, the most important design decision is whether the study aims to model a fixed-volume clot, progressive bleeding, or a specific coagulation-linked pathway. The two workhorse preparations are (i) autologous blood injection, which stereotaxically introduces whole blood—most commonly into the striatum—to generate a space-occupying hematoma, and (ii) bacterial collagenase injection, which enzymatically degrades the microvascular basal lamina to provoke progressive bleeding that better captures dynamic hematoma growth¹⁴^,¹⁵^,⁸³. Autologous blood injection is technically accessible and offers direct control of hematoma volume and composition, making it well aligned with questions centered on mass effect, clot burden, and blood-product–driven secondary injury⁸⁴^,⁸⁵^,⁸⁶. In contrast, collagenase models better approximate ongoing bleeding and evolving edema, which are clinically relevant to hematoma expansion and secondary injury cascades, but the enzymatic component can introduce confounds (matrix proteolysis and exaggerated BBB disruption) and may amplify inflammation and mortality in a dose-dependent manner¹⁴^,¹⁵^,⁸³^,⁸⁷^,⁸⁸^,⁸⁹.

A third, more reductionist approach is thrombin injection, which reliably elicits edema, microglial activation, and protease-activated receptor (PAR) signaling; it is well suited to probing coagulation-linked pathways but does not reproduce vessel rupture, hematoma mechanics, or expansion dynamics and therefore has limited fidelity to spontaneous ICH⁹⁰^,⁹¹. Across ICH models, pathology and readouts depend on hemorrhage location and burden and on how secondary injury is captured over time¹⁵^,⁸⁵^,⁸⁶^,⁸⁷^,⁸⁸. Contemporary guidance therefore emphasizes a priori alignment of model and endpoint—for example: autologous blood injection for mass effect and blood-product toxicity; collagenase for hematoma expansion and edema evolution; and thrombin for PAR/coagulation signaling—together with ARRIVE/STAIR-aligned reporting to improve rigor and reproducibility⁸³^,⁸⁹^,⁹².

No single ICH model recapitulates the full human spectrum. Autologous blood lacks ongoing vessel rupture and may exhibit variable clot resorption; collagenase may overrepresent enzymatic injury; and thrombin isolates a single mediator. Nevertheless, when rigorously standardized—and paired with appropriate endpoints (edema, iron/hemoglobin–heme toxicity, neuroinflammation, functional deficits)—these complementary paradigms provide robust platforms for mechanistic discovery and preclinical testing of hemostatic agents, iron chelation, anti-edema strategies, and neurorestorative interventions⁹³.

Subarachnoid hemorrhage models

Preclinical SAH research relies on three principal rodent paradigms that balance construct validity, technical feasibility, and outcome reproducibility. A practical starting point is the dominant biological question—EBI after rupture versus blood-exposure–driven vasospasm/delayed cerebral ischemia (DCI)—because no single model optimally captures both. EVP uses an intraluminal filament to rupture an intracranial artery, producing an abrupt rise in intracranial pressure (ICP), a sharp fall in cerebral blood flow (CBF), diffuse basal cisternal clot, and high early mortality—features that most closely mimic aneurysmal rupture and EBI in patients¹⁶^,¹⁷. Accordingly, EVP is best positioned for rupture-proximal mechanisms (ICP/CBF crisis, global ischemia components, acute neuroinflammation), while acknowledging higher variability in hemorrhage burden and early survival. Outcome variability arises from differences in perforation depth, filament tip geometry, vascular target, and physiologic control; severity grading and standardized confirmation of hemorrhage burden (e.g., basal cistern clot scoring and imaging-based scales) improve interpretability and support downstream analyses of vasospasm and DCI⁹⁴^,⁹⁵.

In contrast, blood-injection models prioritize controlled exposure to subarachnoid blood. Cisterna magna (CM) blood injection provides technical simplicity and high reproducibility by delivering a defined volume of autologous blood into the subarachnoid space, facilitating controlled investigations of blood-product neurotoxicity, inflammation, and large-artery narrowing; however, it lacks rupture dynamics and EBI physiology⁹⁶^,⁹⁷. Prechiasmatic cistern injection (including modern double-injection mouse variants) offers an intermediate option that deposits clot around the circle of Willis and anterior circulation with manageable mortality and consistent vasospasm, though it still omits the rupture event⁹⁶^,⁹⁷. Thus, CM and prechiasmatic paradigms are most useful when vasospasm/DCI signaling, blood-breakdown products, and inflammatory trajectories are primary endpoints, whereas EVP is preferred when rupture realism and acute EBI are central.

Across paradigms, contemporary best practices emphasize physiological control and endpoint verification rather than incremental procedural variants: perioperative temperature and blood-pressure control, CBF/ICP monitoring when feasible, and imaging-based confirmation of hemorrhage location and burden to reduce misclassification¹⁷^,⁹⁸. Given the field’s translational challenges, incorporating multimodal and longitudinal endpoints (behavioral recovery, imaging markers, vascular narrowing, and inflammatory readouts) can better align preclinical SAH work with clinically meaningful outcomes. Procedural refinements that stabilize injury severity (e.g., standardized perforation approaches and strain/sex-aware design) may reduce early mortality and improve comparability across laboratories¹⁷^,⁹⁹. Large-animal models remain valuable for device testing and angiographic vasospasm endpoints, but are used selectively due to cost and ethics¹⁰⁰. Overall, EVP captures aneurysmal rupture and acute EBI most directly, whereas CM and prechiasmatic injections provide controlled blood-exposure platforms for vasospasm/DCI-focused studies; in all cases, rigorously standardized severity grading and transparent reporting are essential to narrow translational gaps, particularly for DCI biology and long-term outcomes.

Epidural hemorrhage models

EDH models are infrequently used in stroke research because EDH is predominantly traumatic rather than spontaneous; nevertheless, they provide experimentally controlled platforms to study space-occupying mass lesions, ICP–cerebral perfusion pressure (CPP) coupling, and monitoring/evacuation strategies that overlap with critical-care management relevant to severe hemorrhagic brain injury. Two principal approaches are commonly used. First, balloon epidural expansion uses an epidural balloon that is incrementally inflated to achieve prespecified ICP levels, enabling highly controllable and reproducible titration of mass effect and facilitating standardized physiological and imaging endpoints¹⁰¹^,¹⁰². Second, autologous blood injection introduces a defined volume of fresh blood into the epidural space to form a hematoma, incorporating clot biomechanics and blood-related secondary injury signals but typically with greater variability in hematoma geometry and ICP trajectories¹⁰³^,¹⁰⁴. Large-animal adaptations—most recently a swine EDH model that mirrors clinical progression and facilitates translational device testing—extend compatibility with neuromonitoring and image-guided evacuation trials¹⁰⁵.

The key trade-off mirrors that seen across hemorrhagic paradigms: control versus biological realism. Balloon expansion provides unmatched repeatability for studying ICP/CPP dynamics and the timing of decompression, but it lacks the hemostatic and inflammatory milieu associated with real blood, whereas blood-injection EDH models capture clot-related effects while sacrificing stability and precision¹⁰¹^,¹⁰²^,¹⁰³^,¹⁰⁴^,¹⁰⁵. Accordingly, contemporary reviews most often position EDH paradigms as tools for interrogating mass-lesion pathophysiology, neuromonitoring calibration, and surgical timing rather than mechanisms specific to spontaneous cerebrovascular stroke¹⁰⁶. Emerging therapeutics (e.g., anti-HMGB1 monoclonal antibodies) have been explored in rat EDH, underscoring the utility for testing anti-inflammatory interventions under defined mass-effect conditions¹⁰⁷. Overall, EDH models have limited translational relevance to stroke per se, but they remain useful for questions centered on space-occupying extra-axial hematomas, ICP-CPP physiology, and surgical/monitoring strategies.

Special-condition models

Beyond classical ischemic and hemorrhagic paradigms, several special condition models address clinically important, less common scenarios in cerebrovascular disease. These models are most valuable when the clinical question is not territorial infarction size, but rather vascular lesion evolution (aneurysm/AVM), chronic hypoperfusion and white-matter injury, developmental vulnerability, or post-stroke sequelae—often requiring longitudinal and multimodal endpoints.

Cerebral aneurysm induction

Intracranial aneurysm (IA) models typically combine hemodynamic and wall-weakening stimuli—most often pharmacologic hypertension with stereotaxic elastase delivered to the basal cistern or circle of Willis—to induce aneurysm initiation and, in some variants, rupture; related approaches add unilateral carotid flow alteration or carotid–branch ligations to augment wall shear stress¹⁰⁸^,¹⁰⁹^,¹¹⁰^,¹¹¹^,¹¹². Conceptually, these paradigms are best positioned for mechanistic studies of aneurysm formation/instability (inflammation, extracellular-matrix remodeling) and for preclinical evaluation of endovascular or anti-inflammatory strategies using longitudinal imaging readouts. Because rupture is variably achieved and sensitive to background physiology, standardized severity metrics and imaging-based follow-up are critical for interpretability and cross-study comparison.

Arteriovenous malformation models

Brain arteriovenous malformation (bAVM) models fall into two broad families with distinct strengths. Genetic/angiogenic paradigms delete Alk1 or Eng in endothelium and apply focal VEGF to trigger nidus formation and high-flow shunting that recapitulate many human features; newer inducible/localized systems improve survival and lesion control for longitudinal studies¹¹³^,¹¹⁴^,¹¹⁵^,¹¹⁶. Surgical shunt/anastomosis constructs (artery-to-vein) remain useful where anatomy or device testing is the priority, albeit with lower construct validity for spontaneous bAVM pathogenesis. Therefore, genetic/angiogenic models are preferred for pathobiology and medical therapy questions, whereas shunt models are best used for workflow/device evaluation and controlled hemodynamic studies.

Chronic hypoperfusion

Chronic cerebral hypoperfusion models, used to study subcortical VCI and white-matter injury, include bilateral CCA stenosis (BCAS) in mice, bilateral carotid occlusion/stenosis in rats, and device-assisted stenosis or balloon approaches in larger species. BCAS reliably produces progressive white-matter rarefaction, gliosis, and executive/working-memory deficits; asymmetric or low-cost BCAS variants improve MRI compatibility and titration of severity¹¹⁷^,¹¹⁸^,¹¹⁹^,¹²⁰^,¹²¹. Large-animal adaptations permit invasive monitoring and clinical-style imaging, but require continued standardization of stenosis severity, longitudinal endpoints, and reporting practices.

Developmental stroke

Developmental stroke models address the unique vulnerability of the immature brain and the distinct trajectories of injury and recovery. The Rice–Vannucci neonatal HI paradigm remains the workhorse in rodents and supports mechanistic and interventional studies across developmental stages; complementary large-animal neonatal HI models (e.g., piglet) offer gyrencephalic anatomy, clinically relevant neurophysiology (including seizures), and compatibility with longitudinal imaging, supporting translational evaluation of therapeutic hypothermia and adjunct strategies⁷⁷^,¹²²^,¹²³. In this context, long-term neurobehavioral outcomes and developmental milestones are essential endpoints, not optional add-ons.

Post-stroke complications

Post-stroke complications are modeled to reflect long-term outcomes beyond acute infarction. Post-stroke epilepsy has been reproduced after focal ischemia (e.g., MCAO, photothrombosis) with late epileptiform activity and biomarker readouts; post-stroke depression is often modeled by combining MCAO with chronic stress to enhance face validity; post-stroke dysphagia paradigms leverage focal cortical/brainstem lesions with quantitative swallowing assays; and networks underlying post-stroke cognitive impairment are increasingly assessed with multimodal behavior and systems-neuroscience endpoints¹²⁴^,¹²⁵^,¹²⁶^,¹²⁷^,¹²⁸. Their major strength is endpoint relevance (functional recovery and chronic symptoms), whereas their major limitation is heterogeneity and sensitivity to background factors (age, sex, stress exposure, immune state, and baseline cognition), which complicates standardization. Consequently, these models benefit from longitudinal, multimodal phenotyping (behavior, imaging, electrophysiology, and inflammatory markers) and from transparent reporting of inclusion criteria and attrition. More broadly, incorporating aging, sex as a biological variable, and comorbidity-enriched backgrounds can improve translational fidelity for post-stroke sequelae, particularly where immune–brain interactions and systemic illness shape recovery trajectories.

Large-animal and non-human primate models

Although rodent paradigms dominate preclinical stroke research, large animal and NHP models are essential for bridging to clinical practice. Dogs, pigs, and sheep offer gyrencephalic brains, higher white-matter content, human-scale vessels, and compatibility with clinical MRI/CT, angiography, and endovascular workflows—features that enable realistic LVO modeling and device testing⁴³^,¹²⁹^,¹³⁰. In practice, large animals are most useful for late-stage validation, catheter/device feasibility, and clinically realistic peri-procedural physiology, rather than early mechanism screening. In sheep, reproducible permanent or transient MCA occlusion can be produced through microsurgical exposure with clip application or coagulation, permitting longitudinal neuroimaging, ICP monitoring, and assessment of malignant edema¹³¹^,¹³²^,¹³³^,¹³⁴. Swine models increasingly use minimally invasive or endovascular approaches to recapitulate clinical imaging signatures and facilitate intra-arterial therapy studies, addressing historical limitations imposed by the rete mirabile³⁷^,⁴³^,¹³⁴^,¹³⁵. Large animals also support hemorrhagic paradigms in anatomically relevant contexts; for example, the canine double-hemorrhage SAH model remains a workhorse for delayed vasospasm and imaging-based biomarker work¹³⁶.

NHP models (macaques, baboons) provide the closest anatomical, physiological, immunogenomic, and behavioral correspondence to humans. Reversible or permanent endovascular MCAO, autologous-clot embolization, and targeted cortical infarction have been standardized with multimodal MRI, yielding clinically familiar diffusion–perfusion dynamics and enabling study of complex endpoints such as fine motor control, network reorganization, and higher cognition¹³⁷^,¹³⁸^,¹³⁹^,¹⁴⁰. Recent NHP studies integrate neuroimaging, quantitative behavior, and serum proteomics, and they reveal mesoscale network remodeling after stroke—outcomes that are difficult to capture outside primate systems¹³⁹^,¹⁴⁰. These strengths, however, come with constraints: cost, colony availability, and stringent ethical/regulatory oversight (3Rs, refinement of anesthesia/analgesia, and justification of translational value) typically reserve NHP and large-animal work for late-stage validation, comparative efficacy, and device/procedural optimization rather than early mechanism screening¹⁴¹^,¹⁴². In sum, when used judiciously and reported rigorously, large-animal and NHP models function as a critical translational bridge linking rodent discoveries to human stroke interventions⁴³^,¹²⁹^,¹³⁰.

Variable control and methodological quality

Outcomes in experimental stroke are critically shaped by biological variables that mirror clinical heterogeneity. Age, sex, strain, and common comorbidities (e.g., hypertension, diabetes, dyslipidemia, smoking exposure) alter cerebrovascular anatomy, inflammatory tone, and ischemic tolerance, thereby shifting infarct size, edema, hemorrhagic transformation, and recovery trajectories³⁴^,¹⁴³^,¹⁴⁴. Strain-dependent differences in the circle of Willis and collateralization introduce systematic variance even under standardized protocols³⁴. Because these factors influence not only acute injury but also longitudinal recovery, they should be treated as design variables rather than post hoc explanations. Sex is a major modifier across the life span; aged females and males display distinct immune and barrier responses after stroke, supporting routine inclusion of both sexes and sex-specific analyses consistent with NIH SABV policy¹⁴⁴^,¹⁴⁵. Where feasible, factorial or stratified designs and prespecified covariate strategies can improve external validity and clarify effect modification¹⁴³.

Methodological rigor should follow contemporary guidance (ARRIVE 2.0, STAIR, SRRR). Core elements include randomization, allocation concealment, blinded outcome assessment, prespecified inclusion/exclusion criteria, and justified sample-size calculations/power analyses; failure in these domains inflates effect sizes and undermines reproducibility¹⁴⁶^,¹⁴⁷^,¹⁴⁸. To support cross-study synthesis and translation, reporting should prioritize decision-critical parameters—for example, injury induction timing/duration, physiological targets and deviations, attrition and exclusions, and prespecified primary endpoints—together with transparent documentation of analytic choices. Multi-model validation and, when possible, cross-laboratory replication provide a practical stress test for robustness before clinical trial escalation²⁴^,¹⁴⁸.

Perioperative care is a major source of variance. Continuous temperature monitoring with feedback control is essential to avoid inadvertent hypothermia/hyperthermia, which confounds infarct metrics and behavior¹⁴⁹. Anesthetic regimens can carry intrinsic neurovascular effects (e.g., isoflurane-mediated cerebroprotection or hemodynamic shifts); protocols should standardize agent, dose, timing, and ventilation/oxygenation and justify deviations¹⁵⁰^,¹⁵¹. Physiologic monitoring of blood gases, blood pressure, glucose, and rCBF is encouraged; hyperglycemia worsens infarction and BBB leakage and should be prevented or accounted for analytically¹⁵²^,¹⁵³^,¹⁵⁴. Humane analgesia is mandatory; because analgesics can modulate inflammatory tone, investigators should report drug choice and timing explicitly and consider pilot checks for model neutrality¹⁵¹^,¹⁵⁵.

Finally, outcome selection should privilege clinically interpretable recovery alongside tissue metrics. A staged battery spanning sensorimotor, cognitive/affective domains when relevant, and longitudinal time points better reflects clinical endpoints and SRRR recommendations¹⁴⁸^,¹⁵⁶^,¹⁵⁷. Increasingly, multimodal phenotyping (behavioral batteries paired with imaging, electrophysiology, and inflammatory/vascular biomarkers) provides a more mechanistically anchored and translationally aligned endpoint framework than infarct volume alone. Where feasible, multicenter preclinical studies and harmonized core outcomes (with shared SOPs and blinded central analysis) can further improve reproducibility and raise the ceiling for clinical translation²⁴^,¹⁵⁸^,¹⁵⁹.

Translational relevance and guidelines

The persistent bench-to-bedside gap in stroke has driven consensus frameworks that couple methodological rigor to translational alignment. The STAIR initiative (latest comprehensive update: STAIR XI) prioritizes dose–response and therapeutic time-window studies, inclusion of both permanent and transient ischemia with reperfusion paradigms, multi-species/multi-model replication, and—critically—long-term functional outcomes as primary endpoints to mirror clinical trials²⁴. ARRIVE 2.0 complements STAIR by specifying design and reporting standards that reduce bias and improve reproducibility (sample-size justification, randomization, blinding, statistical transparency, prespecified inclusion/exclusion criteria)¹⁴⁶. SRRR recommendations further emphasize recovery-oriented outcomes and harmonized assessments that can be mapped to clinical function¹⁴⁸^,¹⁵⁶^,¹⁵⁷. Collectively, these frameworks converge on a practical translational logic: prespecify what success means, test robustness across plausible clinical conditions, and report in a way that enables replication and meta-synthesis. Despite progress, audits indicate variable adherence across the literature, motivating explicit protocol preregistration and transparent reporting of negative findings to curb publication bias¹⁴⁷^,¹⁵⁴.

Clinical relevance improves when preclinical cohorts model patient realities. Priorities include systematic incorporation of aging, hypertension, diabetes/metabolic dysfunction, and sex as a biological variable—each shaping cerebrovascular physiology, infarct evolution, immune–brain interactions, and treatment response³⁴^,¹⁴³^,¹⁴⁴. Rather than treating comorbidity-enriched designs as optional add-ons, translational pipelines can use staged validation: begin with controlled cohorts for mechanistic clarity, then escalate to aged/comorbid and both-sex cohorts to evaluate effect modification and safety. Translational pathways also benefit from bridging across species: rodents for mechanistic discovery and early screening; large animals (e.g., swine, ovine) for gyrencephalic anatomy, human-scale vasculature, and interventional workflows; and NHP for higher-order behavior and network-level recovery, typically reserved for late-stage validation and device/procedural optimization given cost and ethics³⁷^,⁴³^,¹²⁹^,¹³⁸^,¹⁴⁰.

Standardization of multimodal endpoints strengthens alignment with clinical trials. Combining neuroimaging (e.g., structural/DTI, perfusion/diffusion, fMRI), quantitative neurobehavior, and histopathology—ideally within multicenter preclinical networks—enables robust go/no-go decisions before first-in-human studies²⁴^,¹⁴⁸^,¹⁵⁶^,¹⁵⁷^,¹⁵⁸^,¹⁵⁹. Harmonized core outcome sets and shared SOPs can reduce interpretive flexibility and enable cross-laboratory comparability, especially when paired with blinded analysis. Randomized, blinded multicenter designs in animals have already demonstrated their value in stress-testing candidate therapies and exposing lab-specific effects that single-center studies may miss¹⁵⁸^,¹⁵⁹. In practice, adherence to STAIR/ARRIVE/SRRR, deliberate modeling of age/sex/comorbidity, scenario-matched model selection, and multimodal, function-centered endpoints together provide a credible route to reduce late-stage translational failure.

Challenges and future directions

Despite decades of refinement, preclinical stroke models still face three structural limitations that blunt translational impact: (i) heterogeneity and reproducibility, (ii) under-modeling of clinical complexity, and (iii) short-term, tissue-centric endpoints. Inter-laboratory variance arises from surgical expertise, strain choice, occlusion technique, perioperative care, and outcome readouts; even within widely used paradigms such as filament MCAO, small procedural differences interact with collateral anatomy to produce materially different lesion distributions, attrition, and treatment effects, complicating multicenter replication and go/no-go decisions³⁴^,¹⁴⁶^,¹⁴⁷^,¹⁵⁴^,¹⁵⁸. Clinical complexity remains underrepresented because most studies rely on young, otherwise healthy rodents; yet aging, hypertension, diabetes/metabolic dysfunction, and sex shape cerebrovascular physiology, immune–brain interactions, and treatment response. These variables are still inconsistently modeled, diminishing external validity²⁴^,¹⁴⁶^,¹⁵⁸. Finally, endpoints often emphasize acute lesion metrics over months-long functional recovery, rehabilitation responsiveness, and neuropsychiatric sequelae, in contrast to clinical trials²⁴^,¹⁴⁶^,¹⁵⁸. Taken together, the field’s central challenge is not the absence of models, but the absence of scenario-matched, endpoint-aligned validation pipelines.

Standardization and harmonization remain priorities. Consensus frameworks (ARRIVE 2.0; STAIR XI; emerging multicenter PRISM guidance) call for randomization, allocation concealment, blinded assessment, a priori sample-size justification, transparent reporting of inclusion/exclusion criteria, prespecified endpoints, and cross-model/cross-laboratory replication—practices that reduce inflated effect sizes and facilitate credible translation²⁴^,¹⁴⁶^,¹⁴⁷^,¹⁵⁸. A practical next step is broader adoption of harmonized common data elements and core outcome sets, enabling cross-study synthesis and reducing interpretive flexibility. Multicenter preclinical consortia and common data elements are increasingly advocated to stress-test robustness before first-in-human studies¹⁵⁸.

Emerging technologies provide credible routes to address the above constraints when integrated into coherent validation strategies. First, comorbidity-enriched, genetic, and humanized models can improve construct validity for specific therapeutic classes. Gene-editing approaches enable modeling of vascular risk and small-vessel disease biology (e.g., NOTCH3; COL4A1/A2), and humanized immune systems can support mechanistic evaluation of immunomodulatory interventions in vivo³⁴^,¹⁶⁰^,¹⁶¹^,¹⁶²^,¹⁶³^,¹⁶⁴^,¹⁶⁵^,¹⁶⁶. Second, human-derived experimental systems can serve as scalable translation bridges, not replacements. Vascularized brain organoids, microphysiological platforms, and organotypic/ex vivo preparations subjected to oxygen–glucose deprivation enable human-cell interrogation and screening; when used bidirectionally with in vivo models, they can confirm conserved mechanisms and de-risk targets that fail to generalize across systems¹⁶¹^,¹⁶²^,¹⁶³. Third, multi-omics and systems biology approaches—particularly cross-species profiling anchored to patient cohorts—can identify conserved injury–repair programs and candidate biomarkers that better connect animal results to clinical phenotypes¹⁶⁷. Fourth, AI/ML methods are increasingly practical in preclinical pipelines, enabling automated behavioral quantification, harmonized imaging readouts, and prediction of functional outcomes, thereby reducing measurement noise and supporting richer longitudinal endpoints¹⁶⁸^,¹⁶⁹^,¹⁷⁰. Critically, these tools add translational value only when paired with prespecified endpoints and standardized acquisition/analysis pipelines.

Ethics and the 3Rs (Replacement, Reduction, Refinement) remain foundational. Replacement via organoids, microphysiological systems, and in silico models can triage candidates before in vivo testing; reduction and refinement through optimized analgesia/anesthesia, telemetry-guided homeostasis, and standardized monitoring improve welfare and data quality. Operationalizing the 3Rs should be viewed as a rigor strategy as well as an ethical obligation, because improved welfare often reduces variance and attrition.

Progress will depend on integrated, multi-model, multi-species, multimodal roadmaps. One pragmatic pipeline is staged validation: discovery and mechanism in controlled rodent cohorts; replication across complementary models (e.g., controlled focal ischemia plus thromboembolic realism, or fixed-volume ICH plus expansion-prone ICH); escalation to aged/comorbid and both-sex cohorts; and targeted bridging in large animals and, when justified, NHPs for clinical workflow/device realism and higher-order recovery endpoints. Parallel human-cell systems (organoids/slices), omics-anchored target selection, and AI-assisted outcome analytics can provide mechanistic depth and endpoint richness across stages. Coupled with ARRIVE/STAIR-aligned rigor and explicit model-selection logic (Table 1), this staged, scenario-matched strategy offers a credible path to narrowing the translational gap and accelerating effective therapies for stroke.

Access restricted. Please log in or start a trial to view this content.

Conclusions

$$\rightleftharpoonup{xx}$$ $$\longleftharp{xx}$$, $$\longrightharp{xx}$$,

No single animal model can recapitulate the full complexity and heterogeneity of human stroke. Instead, each paradigm offers distinct trade-offs among construct validity, reproducibility, feasibility, and ethical burden. Accordingly, the central principle is scenario-matched, endpoint-aligned model selection: investigators should start from the clinical scenario and the primary endpoint (acute tissue injury, reperfusion biology, hematoma expansion/toxicity, vasospasm/DCI, or long-term functional recovery), then choose th...

Access restricted. Please log in or start a trial to view this content.

Disclosures

$$\rightleftharpoonup{xx}$$ $$\longleftharp{xx}$$, $$\longrightharp{xx}$$,

The authors declare that they have no conflict of interest.

Acknowledgements

$$\rightleftharpoonup{xx}$$ $$\longleftharp{xx}$$, $$\longrightharp{xx}$$,

This work was supported by the National Natural Science Foundation of China (No. 82260246), the Special Funding for the Training of High-Level Health Technology Talents in Yunnan Province (No. L-2025004), and the Special Funding for the Talent Team Project of the Second Affiliated Hospital of Kunming Medical University (No. RCTDXS-202307).

Access restricted. Please log in or start a trial to view this content.

References

$$\rightleftharpoonup{xx}$$ $$\longleftharp{xx}$$, $$\longrightharp{xx}$$,

Feigin, V. L., et al. Global, regional, and national burden of stroke and its risk factors, 1990-2021: A systematic analysis for the global burden of disease study 2021. Lancet Neurol. 23 (10), 973-1003 (2021).
Feigin, V. L., et al. World stroke organization: Global stroke fact sheet 2025. Int J Stroke. 20 (2), 132-144 (2025).
National Institute of Neurological Disorders and Stroke rt-PA Stroke Study Group. Tissue plasminogen activator for acute ischemic stroke. N Engl J Med. 333 (24), 1581-1587 (1995).
Goyal, M., et al. Endovascular thrombectomy after large-vessel ischaemic stroke: A meta-analysis of individual patient data from five randomised trials. Lancet. 387 (10029), 1723-1731 (2016).
O'collins, V. E., et al. 1,026 experimental treatments in acute stroke. Ann Neurol. 59 (3), 467-477 (2006).
Fluri, F., Schuhmann, M. K., Kleinschnitz, C. Animal models of ischemic stroke and their application in clinical research. Drug Des Devel Ther. 9, 3445-3454 (2015).
Macrae, I. M. Preclinical stroke research--advantages and disadvantages of the most common rodent models of focal ischaemia. Br J Pharmacol. 164 (4), 1062-1078 (2011).
Pulsinelli, W. A., Buchan, A. M. The four-vessel occlusion rat model: Method for complete occlusion of vertebral arteries and control of collateral circulation. Stroke. 19 (7), 913-914 (1988).
Koizumi, J. I., Yoshida, Y., Nakazawa, T., Ooneda, G. Experimental studies of ischemic brain edema. Jpn J Stroke. 8 (1), 1-8 (1986).
Longa, E. Z., Weinstein, P. R., Carlson, S., Cummins, R. Reversible middle cerebral artery occlusion without craniectomy in rats. Stroke. 20 (1), 84-91 (1989).
Zhang, R. L., Chopp, M., Zhang, Z. G., Jiang, Q., Ewing, J. R. A rat model of focal embolic cerebral ischemia. Brain Res. 766 (1-2), 83-92 (1997).
Watson, B. D., Dietrich, W. D., Busto, R., Wachtel, M. S., Ginsberg, M. D. Induction of reproducible brain infarction by photochemically initiated thrombosis. Ann Neurol. 17 (5), 497-504 (1985).
Tatlisumak, T., et al. A novel endothelin antagonist, a-127722, attenuates ischemic lesion size in rats with temporary middle cerebral artery occlusion. Stroke. 29 (4), 850-858 (1998).
Rosenberg, G. A., Mun-Bryce, S., Wesley, M., Kornfeld, M. Collagenase-induced intracerebral hemorrhage in rats. Stroke. 21 (5), 801-807 (1990).
Krafft, P. R., et al. Modeling intracerebral hemorrhage in mice: Injection of autologous blood or bacterial collagenase. J Vis Exp. (67), e4289(2012).
Bederson, J. B., Germano, I. M., Guarino, L. Cortical blood flow and cerebral perfusion pressure in a new noncraniotomy model of subarachnoid hemorrhage in the rat. Stroke. 26 (6), 1086-1092 (1995).
Sehba, F. A. Rat endovascular perforation model. Transl Stroke Res. 5 (6), 660-668 (2014).
Rice, J. E., Vannucci, R. C. 3rd, Brierley, J. B. The influence of immaturity on hypoxic-ischemic brain damage in the rat. Ann Neurol. 9 (2), 131-141 (1981).
Shibata, M., Ohtani, R., Ihara, M., Tomimoto, H. White matter lesions and glial activation in a novel mouse model of chronic cerebral hypoperfusion. Stroke. 35 (11), 2598-2603 (2004).
Nowak, B., et al. Animal models of focal ischemic stroke: Brain size matters. Front Stroke. 2, 1165231(2023).
Friedrich, J., Lindauer, U., Höllig, A. Procedural and methodological quality in preclinical stroke research-a cohort analysis of the rat mcao model comparing periods before and after the publication of stair/arrive. Front Neurol. 13, 834003(2022).
Sena, E. S., Currie, G. L., McCann, S. K., Macleod, M. R., Howells, D. W. Systematic reviews and meta-analysis of preclinical studies: Why perform them and how to appraise them critically. J Cerebral Blood Flow Metab. 34 (5), 737-742 (2014).
Fisher, M., et al. Update of the stroke therapy academic industry roundtable preclinical recommendations. Stroke. 40 (6), 2244-2250 (2009).
Lyden, P., Buchan, A., Boltze, J., Fisher, M. Top priorities for cerebroprotective studies-a paradigm shift: Report from stair xi. Stroke. 52 (9), 3063-3071 (2021).
Du Sert, P. N., et al. The arrive guidelines 2.0: Updated guidelines for reporting animal research. Plos Biol. 18 (7), e3000410(2020).
Kuge, Y., Minematsu, K., Yamaguchi, T., Miyake, Y. Nylon monofilament for intraluminal middle cerebral artery occlusion in rats. Stroke. 26 (9), 1655-1657 (1995).
Takano, K., Tatlisumak, T., Bergmann, A. G., Gibson, D. G., Fisher, M. 3rd Reproducibility and reliability of middle cerebral artery occlusion using a silicone-coated suture (koizumi) in rats. J Neurol Sci. 153 (1), 8-11 (1997).
Li, Y., et al. Distinctions between the koizumi and zea longa methods for middle cerebral artery occlusion (mcao) model: A systematic review and meta-analysis of rodent data. Sci Rep. 13 (1), 10247(2023).
Morris, G. P., et al. A comparative study of variables influencing ischemic injury in the longa and koizumi methods of intraluminal filament middle cerebral artery occlusion in mice. PLoS One. 11 (2), e0148503(2016).
Ingberg, E., Dock, H., Theodorsson, E., Theodorsson, A., Ström, J. O. Effect of laser doppler flowmetry and occlusion time on outcome variability and mortality in rat middle cerebral artery occlusion: Inconclusive results. BMC Neurosci. 19 (1), 24(2018).
Hedna, V. S., et al. Validity of laser doppler flowmetry in predicting outcome in murine intraluminal middle cerebral artery occlusion stroke. J Vasc Interv Neurol. 8 (3), 74-82 (2015).
Schmid-Elsaesser, R., Zausinger, S., Hungerhuber, E., Baethmann, A., Reulen, H. J. A critical reevaluation of the intraluminal thread model of focal cerebral ischemia: Evidence of inadvertent premature reperfusion and subarachnoid hemorrhage in rats by laser-doppler flowmetry. Stroke. 29 (10), 2162-2170 (1998).
Li, F., Omae, T., Fisher, M. Spontaneous hyperthermia and its mechanism in the intraluminal suture middle cerebral artery occlusion model of rats. Stroke. 30 (11), 2464-2470 (1999).
Qian, B., Rudy, R. F., Cai, T., Du, R. Cerebral artery diameter in inbred mice varies as a function of strain. Front Neuroanat. 12, 10(2018).
Chen, Y., et al. A novel mouse model of thromboembolic stroke. J Neurosci Meth. 256, 203-211 (2015).
Sutherland, B. A., et al. The transient intraluminal filament middle cerebral artery occlusion model as a model of endovascular thrombectomy in stroke. J Cereb Blood Flow Metab. 36 (2), 363-369 (2016).
Keister, A., et al. Review: Preclinical models of large-vessel occlusion stroke. Stroke Vasc Intervent Neurol. 4 (4), e000604(2024).
Orset, C., et al. Mouse model of in situ thromboembolic stroke and reperfusion. Stroke. 38 (10), 2771-2778 (2007).
Arkelius, K., Vivien, D., Orset, C., Ansar, S. Validation of a stroke model in rat compatible with rt-pa-induced thrombolysis: New hope for successful translation to the clinic. Sci Rep. 10 (1), 12191(2020).
Ostrova, I. V., Kalabushev, S. N., Ryzhkov, I. A., Tsokolaeva, Z. I. A novel thromboplastin-based rat model of ischemic stroke. Brain Sci. 11 (11), 1475(2021).
Lapchak, P. A. Translational stroke research using a rabbit embolic stroke model: A correlative analysis hypothesis for novel therapy development. Transl Stroke Res. 1 (2), 96-107 (2010).
Yuki, I., et al. Impact of target artery size on the performance of aspiration thrombectomy: Insights from a swine model with real-time visualization. Am J Neuroradiol. 45 (6), 727-730 (2024).
Golubczyk, D., et al. Endovascular model of ischemic stroke in swine guided by real-time mri. Sci Rep. 10 (1), 17318(2020).
Van Der Wijk, A. E., et al. Extravasation of biodegradable microspheres in the rat brain. Drug Delivery. 30 (1), 2194579(2023).
Shen, Y., et al. Histochemistry of microinfarcts in the mouse brain after injection of fluorescent microspheres into the common carotid artery. Neural Regenerat Res. 17 (4), 832-837 (2022).
Xue, Y., et al. Quantification of hypoxic regions distant from occlusions in cerebral penetrating arteriole trees. PLOS Computat Biol. 18 (8), e1010166(2022).
Joshi, S., et al. Ferric chloride-induced arterial thrombosis and sample collection for 3d electron microscopy analysis. J Vis Exp. (193), e64985(2023).
Shim, Y., et al. Characterization of ferric chloride-induced arterial thrombosis model of mice and the role of red blood cells in thrombosis acceleration. Yonsei Med J. 62 (11), 1032-1041 (2021).
Anagnostakou, V., et al. Preclinical modeling of mechanical thrombectomy. J Biomech. 130, 110894(2022).
Knezic, A., Broughton, B. R. S., Widdop, R. E., Mccarthy, C. A. Optimising the photothrombotic model of stroke in the c57bi/6 and fvb/n strains of mouse. Sci Rep. 12 (1), 7598(2022).
Conti, E., Carlini, N., Piccardi, B., Allegra Mascaro, A. L., Pavone, F. S. Photothrombotic middle cerebral artery occlusion in mice: A novel model of ischemic stroke. eneuro. 10 (2), 0244(2023).
Taylor, Z. J., Shih, A. Y. Targeted occlusion of individual pial vessels of mouse cortex. Bio Protoc. 3 (17), e897(2013).
Nishimura, N., Schaffer, C. B., Friedman, B., Lyden, P. D., Kleinfeld, D. Penetrating arterioles are a bottleneck in the perfusion of neocortex. Proc Natl Acad Sci. 104 (1), 365-370 (2007).
Fukuda, M., Matsumura, T., Suda, T., Hirase, H. Depth-targeted intracortical microstroke by two-photon photothrombosis in rodent brain. Neurophotonics. 9 (2), 021910(2022).
Wester, P., Watson, B. D., Prado, R., Dietrich, W. D. A photothrombotic ‘ring’ model of rat stroke-in-evolution displaying putative penumbral inversion. Stroke. 26 (3), 444-450 (1995).
Qian, C., et al. Precise characterization of the penumbra revealed by mri: A modified photothrombotic stroke model study. PLOS ONE. 11 (4), e0153756(2016).
Weber, R. Z., et al. Characterization of the blood brain barrier disruption in the photothrombotic stroke model. Front Physiol. 11, 586226(2020).
Sommer, C. J. Ischemic stroke: Experimental models and reality. Acta Neuropathol. 133 (2), 245-261 (2017).
Trotman-Lucas, M., Gibson, C. A review of experimental models of focal cerebral ischemia focusing on the middle cerebral artery occlusion model [version 2; peer review: 2 approved]. F1000Research. 10, 242(2021).
Ansari, S., et al. Endothelin-1 induced middle cerebral artery occlusion model for ischemic stroke with laser doppler flowmetry guidance in rat. J Vis Exp. (72), e50014(2013).
Sharkey, J., Butcher, S. P. Characterisation of an experimental model of stroke produced by intracerebral microinjection of endothelin-1 adjacent to the rat middle cerebral artery. J Neurosci Meth. 60 (1-2), 125-131 (1995).
Nikolova, S., et al. Endothelin-1 induced mcao: Dose dependency of cerebral blood flow. J Neurosci Meth. 179 (1), 22-28 (2009).
Frost, S. B., Barbay, S., Mumert, M. L., Stowe, A. M., Nudo, R. J. An animal model of capsular infarct: Endothelin-1 injections in the rat. Behav Brain Res. 169 (2), 206-211 (2006).
Blasi, F., Whalen, M. J., Ayata, C. Lasting pure-motor deficits after focal posterior internal capsule white-matter infarcts in rats. J Cerebral Blood Flow Metabol. 35 (6), 977-984 (2015).
Virley, D., et al. A new primate model of focal stroke: Endothelin-1—induced middle cerebral artery occlusion and reperfusion in the common marmoset. J Cerebral Blood Flow Metabol. 24 (1), 24-41 (2004).
Horie, N., et al. Mouse model of focal cerebral ischemia using endothelin-1. J Neurosci Meth. 173 (2), 286-290 (2008).
Roome, R. B., et al. A reproducible endothelin-1 model of forelimb motor cortex stroke in the mouse. J Neurosci Meth. 233, 34-44 (2014).
Mosneag, I. -E., Flaherty, S. M., Wykes, R. C., Allan, S. M. Stroke and translational research – review of experimental models with a focus on awake ischaemic induction and anaesthesia. Neuroscience. 550, 89-101 (2024).
Pulsinelli, W. A., Brierley, J. B. A new model of bilateral hemispheric ischemia in the unanesthetized rat. Stroke. 10 (3), 267-272 (1979).
Deng, P., Xu, Z. C. Four-vessel occlusion model in rats. Springer Protoc Handbooks. , Humana Press. (2009).
Raval, A., Chunli, L., Bingren, H. Rat model of global cerebral ischemia: The two-vessel occlusion (2vo) model of forebrain ischemia. Springer Protoc Handbooks. , Humana Press. (2009).
Sanderson, T. H., Wider, J. M. 2-vessel occlusion/hypotension: A rat model of global brain ischemia. J Vis Exp. (76), e50173(2013).
Onken, M., Berger, S., Kristian, T. Simple model of forebrain ischemia in mouse. J Neurosci Meth. 204 (2), 254-261 (2012).
Kondo, T., et al. Transient forebrain ischemia induces impairment in cognitive performance prior to extensive neuronal cell death in mongolian gerbil (meriones unguiculatus). J Vet Sci. 19 (4), 505-511 (2018).
Kim, H., et al. Protocol for establishing a global ischemia model using a 4-vessel occlusion in rats. STAR Protoc. 4 (4), 102630(2023).
Du, X. Y., et al. Characteristics of circle of willis variations in the mongolian gerbil and a newly established ischemia-prone gerbil group. Ilar J. 52 (1), E1-E7 (2011).
Vannucci, S. J., Back, S. A. The vannucci model of hypoxic-ischemic injury in the neonatal rodent: 40 years later. Dev Neurosci. 44 (4-5), 186-193 (2022).
Lyu, H., et al. A new hypoxic ischemic encephalopathy model in neonatal rats. Heliyon. 7 (12), e08646(2021).
Rutledge, C. A., et al. A novel ultrasound-guided mouse model of sudden cardiac arrest. PLoS One. 15 (12), e0237292(2020).
Liu, H., et al. Novel modification of potassium chloride induced cardiac arrest model for aged mice. Aging Dis. 9 (1), 31-39 (2018).
Perkins, G. D., et al. Improving outcomes after post–cardiac arrest brain injury: A scientific statement from the international liaison committee on resuscitation. Circulation. , (2024).
Arnautovic, T., Sinha, S., Laptook, A. R. Neonatal hypoxic-ischemic encephalopathy and hypothermia treatment. Obstet Gynecol. 143 (1), 67-81 (2024).
Maclellan, C. L., Silasi, G., Auriat, A. M., Colbourne, F. Rodent models of intracerebral hemorrhage. Stroke. 41 (10), S95-S98 (2010).
Sansing, L. H., et al. Autologous blood injection to model spontaneous intracerebral hemorrhage in mice. J Vis Exp. (54), e2618(2011).
Rynkowski, M. A., et al. A mouse model of intracerebral hemorrhage using autologous blood infusion. Nat Protoc. 3 (1), 122-128 (2008).
Zhu, W., et al. Mouse models of intracerebral hemorrhage in ventricle, cortex, and hippocampus by injections of autologous blood or collagenase. PLOS ONE. 9 (5), e97423(2014).
Beray-Berthat, V., et al. Long-term histological and behavioural characterisation of a collagenase-induced model of intracerebral haemorrhage in rats. J Neurosci Meth. 191 (2), 180-190 (2010).
Bai, Q., et al. Intracerebral haemorrhage: From clinical settings to animal models. Stroke Vasc Neurol. 5 (4), 388-395 (2020).
Leonardo, C. C., Robbins, S., Dore, S. Translating basic science research to clinical application: Models and strategies for intracerebral hemorrhage. Front Neurol. 3, 85(2012).
Yang, S., et al. Effects of thrombin on neurogenesis after intracerebral hemorrhage. Stroke. 39 (7), 2079-2084 (2008).
Ziai, W. C. Hematology and inflammatory signaling of intracerebral hemorrhage. Stroke. 44 (6_suppl_1), S74-S78 (2013).
Kim, J. -T., Youn, D. H., Kim, B. J., Rhim, J. K., Jeon, J. P. Recent stem cell research on hemorrhagic stroke: An update. J Korean Neurosurg Soc. 65 (2), 161-172 (2022).
Paiva, W. S., et al. Animal models for the study of intracranial hematomas (review). Exp Ther Med. 25 (1), 20(2023).
Sugawara, T., Ayer, R., Jadhav, V., Zhang, J. H. A new grading system evaluating bleeding scale in filament perforation subarachnoid hemorrhage rat model. J Neurosci Meth. 167 (2), 327-334 (2008).
Egashira, Y., Shishido, H., Hua, Y., Keep, R. F., Xi, G. New grading system based on magnetic resonance imaging in a mouse model of subarachnoid hemorrhage. Stroke. 46 (2), 582-584 (2015).
Prunell, G. F., Mathiesen, T., Diemer, N. H., Svendgaard, N. A. Experimental subarachnoid hemorrhage: Subarachnoid blood volume, mortality rate, neuronal death, cerebral blood flow, and perfusion pressure in three different rat models. Neurosurgery. 52 (1), 165-175 (2003).
Diwan, D., et al. Development and validation of a prechiasmatic mouse model of subarachnoid hemorrhage to measure long-term cognitive deficits. Adv Sci. 11 (46), 2403977(2024).
Liu, S., et al. Endovascular perforation model for subarachnoid hemorrhage combined with magnetic resonance imaging (mri). J Vis Exp. (178), e63150(2021).
Fürstenau, E., Lindauer, U., Koch, H., Höllig, A. Secondary ischemia assessment in murine and rat preclinical subarachnoid hemorrhage models: A systematic review. J Am Heart Assoc. 13 (5), e032694(2024).
Mori, K., et al. Canine double hemorrhage model of experimental subarachnoid hemorrhage. Acta Neurochir Suppl. 120, 347-351 (2015).
Burger, R., Bendszus, M., Vince, G. H., Roosen, K., Marmarou, A. A new reproducible model of an epidural mass lesion in rodents. Part i: Characterization by neurophysiological monitoring, magnetic resonance imaging, and histopathological analysis. J Neurosurg. 97 (6), 1410-1418 (2002).
Bendszus, M., Burger, R., Vince, G. H., Solymosi, L. A reproducible model of an epidural mass lesion in rodents. Part ii: Characterization by in vivo magnetic resonance imaging. J Neurosurg. 97 (6), 1419-1423 (2002).
Pan, A., et al. Experimental epidural hematoma causes cerebral infarction and activates neocortical glial and neuronal genesis in adult guinea pigs. J Neurosci Res. 91 (2), 249-261 (2013).
Balikci, M., et al. Biochemical effects of experimental epidural hematoma on brain parenchyma of rats. Neurol Res. 30 (5), 450-456 (2008).
Donaldson, R. I., Buchanan, O. J., Graham, T. L., Ross, J. D. Development of a novel epidural hemorrhage model in swine. Military Med. 188 (1-2), 20-26 (2021).
Marklund, N., Hillered, L. Animal modelling of traumatic brain injury in preclinical drug development: Where do we go from here. Br J Pharmacol. 164 (4), 1207-1229 (2011).
Gao, S., et al. Anti-hmgb1 mab therapy reduces epidural hematoma injury. Int J Mol Sci. 25 (11), 5889(2024).
Khan, D., et al. Current mouse models of intracranial aneurysms: Analysis of pharmacological agents used to induce aneurysms and their impact on translational research. J Am Heart Assoc. 13 (3), e031811(2024).
Nuki, Y., et al. Elastase-induced intracranial aneurysms in hypertensive mice. Hypertension. 54 (6), 1337-1344 (2009).
Martinez, A. N., et al. Single-cell transcriptome analysis of the circle of willis in a mouse cerebral aneurysm model. Stroke. 53 (8), 2647-2657 (2022).
Thompson, J. W., et al. In vivo cerebral aneurysm models. Neurosurg Focus. 47 (1), E20(2019).
Cayron, A. F., Morel, S., Allémann, E., Bijlenga, P., Kwak, B. R. Imaging of intracranial aneurysms in animals: A systematic review of modalities. Neurosurg Rev. 46 (1), 56(2023).
Han, C., et al. Novel experimental model of brain arteriovenous malformations using conditional alk1 gene deletion in transgenic mice. J Neurosurg. 137 (1), 163-174 (2022).
Arthur, H. M., Roman, B. L. An update on preclinical models of hereditary haemorrhagic telangiectasia: Insights into disease mechanisms. Front Med. 9, 973964(2022).
Jeong, J. Y., et al. Pathophysiology in brain arteriovenous malformations: Focus on endothelial dysfunctions and endothelial-to-mesenchymal transition. Biomedicines. 12 (8), 1795(2024).
Ricciardelli, A. R., et al. From bench to bedside: Murine models of inherited and sporadic brain arteriovenous malformations. Angiogenesis. 28 (2), 15(2025).
Ishikawa, H., et al. A brief overview of a mouse model of cerebral hypoperfusion by bilateral carotid artery stenosis. J Cereb Blood Flow Metab. 43 (2_suppl), 18-36 (2023).
Kakae, M., Kawashita, A., Onogi, H., Nakagawa, T., Shirakawa, H. Bilateral common carotid artery stenosis in mice: A model of chronic cerebral hypoperfusion-induced vascular cognitive impairment. Bio Protoc. 14 (13), e5022(2024).
Weng, Z., et al. A novel needle mouse model of vascular cognitive impairment and dementia. J Neurosci. 43 (44), 7351-7360 (2023).
Washida, K., Hattori, Y., Ihara, M. Animal models of chronic cerebral hypoperfusion: From mouse to primate. Int J Mol Sci. 20 (24), 6176(2019).
Abdou, H., et al. Characterizing brain perfusion in a swine model of raised intracranial pressure. J Surg Res. 278, 64-69 (2022).
Ourednik, J., Ourednik, V., Ghosh, N., Snyder, E. Y. Protocol to optimize the rice-vannucci rat pup model of perinatal asphyxia to ensure predictable hypoxic-ischemic cerebral lesions. STAR Protoc. 5 (2), 103025(2024).
Martinello, K. A., et al. Hypothermia is not therapeutic in a neonatal piglet model of inflammation-sensitized hypoxia–ischemia. Pediat Res. 91 (6), 1416-1427 (2022).
He, M., et al. Animal models of epilepsy after ischemic stroke. Neuroscience. 576, 1-7 (2025).
Mușat, M. I., Cătălin, B., Hadjiargyrou, M., Popa-Wagner, A., Greșiță, A. Advancing post-stroke depression research: Insights from murine models and behavioral analyses. Life (Basel). 14 (9), 1110(2024).
Sasegbon, A., Cheng, I., Hamdy, S. The neurorehabilitation of post-stroke dysphagia: Physiology and pathophysiology. J Physiol. 603 (3), 617-634 (2025).
Bandet, M. V., Winship, I. R. Aberrant cortical activity, functional connectivity, and neural assembly architecture after photothrombotic stroke in mice. eLife. 12, RP90080(2024).
Brezzo, G., et al. Assessing post-stroke cognition in pre-clinical models: Lessons and recommendations from a multi-center study. J Cereb Blood Flow Metab. , (2025).
Kaiser, E. E., West, F. D. Large animal ischemic stroke models: Replicating human stroke pathophysiology. Neural Regen Res. 15 (8), 1377-1387 (2020).
Taha, A., et al. Comparison of large animal models for acute ischemic stroke: Which model to use. Stroke. 53 (4), 1411-1422 (2022).
Herrmann, A. M., et al. Development of a routinely applicable imaging protocol for fast and precise middle cerebral artery occlusion assessment and perfusion deficit measure in an ovine stroke model: A case study. Front Neurol. 10, 1113(2019).
Boltze, J., et al. Permanent middle cerebral artery occlusion in sheep: A novel large animal model of focal cerebral ischemia. J Cereb Blood Flow Metab. 28 (12), 1951-1964 (2008).
Wells, A. J., et al. A surgical model of permanent and transient middle cerebral artery stroke in the sheep. PLoS One. 7 (7), e42157(2012).
Wells, A. J., et al. Elevated intracranial pressure and cerebral edema following permanent mca occlusion in an ovine model. PLoS One. 10 (6), e0130512(2015).
Castaño, C., et al. Establishment of a reproducible and minimally invasive ischemic stroke model in swine. JCI Insight. 8 (8), e163398(2023).
Jadhav, V., Sugawara, T., Zhang, J., Jacobson, P., Obenaus, A. Magnetic resonance imaging detects and predicts early brain injury after subarachnoid hemorrhage in a canine experimental model. J Neurotrauma. 25 (9), 1099-1106 (2008).
Wu, D., et al. Endovascular ischemic stroke models of adult rhesus monkeys: A comparison of two endovascular methods. Sci Rep. 6 (1), 31608(2016).
Lin, X., et al. Nonhuman primate models of ischemic stroke and neurological evaluation after stroke. J Neurosci Meth. 376, 109611(2022).
Li, G., et al. Comprehensive assessment of ischemic stroke in nonhuman primates: Neuroimaging, behavioral, and serum proteomic analysis. ACS Chem Neurosci. 15 (7), 1548-1559 (2024).
Nashed, J. Y., Gale, D. J., Gallivan, J. P., Cook, D. J. Changes in cortical manifold structure following stroke and its relation to behavioral recovery in the male macaque. Nat Comm. 15 (1), 9005(2024).
Lemon, R. N. Applying the 3rs to neuroscience research involving nonhuman primates. Drug Disc Today. 23 (9), 1574-1577 (2018).
Basso, M. A., et al. The future of nonhuman primate neuroscience: Peril or possibilities. J Neurosci. 44 (37), e1458242024(2024).
Cheng, W., Zhao, Q., Li, C., Xu, Y. Neuroinflammation and brain-peripheral interaction in ischemic stroke: A narrative review. Front Immunol. 13, 1080737(2022).
Ahnstedt, H., et al. Sex differences in t cell immune responses, gut permeability and outcome after ischemic stroke in aged mice. Brain Behav Immun. 87, 556-567 (2020).
Kostas-Polston, E. A., et al. Ensuring accountability for consideration of sex as a biological variable in research. Nurs Outlook. 72 (4), 102194(2024).
Percie Du Sert, N., et al. Reporting animal research: Explanation and elaboration for the arrive guidelines 2.0. PLOS Biol. 18 (7), e3000411(2020).
Macleod, M. R., et al. Risk of bias in reports of in vivo research: A focus for improvement. PLOS Biol. 13 (10), e1002273(2015).
Corbett, D., et al. Enhancing the alignment of the preclinical and clinical stroke recovery research pipeline: Consensus-based core recommendations from the stroke recovery and rehabilitation roundtable translational working group. Int J Stroke. 12 (5), 462-471 (2017).
Barber, P. A., Hoyte, L., Colbourne, F., Buchan, A. M. Temperature-regulated model of focal ischemia in the mouse. Stroke. 35 (7), 1720-1725 (2004).
Archer, D. P., Walker, A. M., Mccann, S. K., Moser, J. J., Appireddy, R. M. Anesthetic neuroprotection in experimental stroke in rodents: A systematic review and meta-analysis. Anesthesiology. 126 (4), 653-665 (2017).
Oh, S. S., Narver, H. L. Mouse and rat anesthesia and analgesia. Curr Protoc. 4 (2), e995(2024).
Denorme, F., Portier, I., Kosaka, Y., Campbell, R. A. Hyperglycemia exacerbates ischemic stroke outcome independent of platelet glucose uptake. J Thrombosis Haemostasis. 19 (2), 536-546 (2021).
Liu, S., Zhen, G., Meloni, B. P., Campbell, K., Winn, H. R. Rodent stroke model guidelines for preclinical stroke trials (1st edition). J Exp Stroke Transl Med. 2 (2), 2-27 (2009).
Sena, E. S., van der Worp, H. B., Bath, P. M. W., Howells, D. W., Macleod, M. R. Publication bias in reports of animal stroke studies leads to major overstatement of efficacy. PLoS Biol. 8 (3), e1000344(2010).
Peterson, N. C., Nunamaker, E. A., Turner, P. V. To treat or not to treat: The effects of pain on experimental parameters. Comp Med. 67 (6), 469-482 (2017).
Balkaya, M., Cho, S. Optimizing functional outcome endpoints for stroke recovery studies. J Cereb Blood Flow Metab. 39 (12), 2323-2342 (2019).
Pohl, J., et al. Consensus-based core set of outcome measures for clinical motor rehabilitation after stroke—a delphi study. Front Neurol. 11, 875(2020).
Ayata, C., et al. Preclinical ischemic stroke multicenter (prism) trials collective statement: Opportunities, challenges, and recommendations for a new era. Stroke. 57 (1), e12-e40 (2026).
Llovera, G., et al. Results of a preclinical randomized controlled multicenter trial (prct): Anti-cd49d treatment for acute brain ischemia. Sci Transl Med. 7 (299), 299ra121-299ra121 (2015).
Zhang, T., et al. Humanized mouse model reveals t cell anxa2 as a potential therapeutic target in ischemic stroke. iScience. 28 (5), 112302(2025).
Giorgi, C., Castelli, V., D'angelo, M., Cimini, A. Organoids modeling stroke in a petri dish. Biomedicines. 12 (4), (2024).
Kistemaker, L., Van Bodegraven, E. J., De Vries, H. E., Hol, E. M. Vascularized human brain organoids: Current possibilities and prospects. Trends Biotechnol. 43 (6), 1275-1285 (2025).
Bryniarska-Kubiak, N., et al. Oxygen-glucose deprivation in organotypic hippocampal cultures leads to cytoskeleton rearrangement and immune activation: Link to the potential pathomechanism of ischaemic stroke. Cells. 12 (11), (2023).
Granata, A. Functional genomics in stroke: Current and future applications of ipscs and gene editing to dissect the function of risk variants. BMC Cardiovasc Disord. 23 (1), 223(2023).
Bonowicz, K., et al. Crispr-cas9 in cardiovascular medicine: Unlocking new potential for treatment. Cells. 14 (2), (2025).
Al-Thani, M., et al. A novel human ipsc model of col4a1/a2 small vessel disease unveils a key pathogenic role of matrix metalloproteinases. Stem Cell Rep. 18 (12), 2386-2399 (2023).
Cheng, Z., Zhu, H., Feng, S., Zhang, Y., Xiong, X. Cross-species multi-omics analysis reveals myeloid-driven endothelial oxidative stress in ischemic stroke. Front Biosci (Landmark Ed). 30 (4), 37429(2025).
Koska, İ, Selver, A. Artificial intelligence in stroke imaging: A comprehensive review. Eurasian J Med. 55 (1), 91-97 (2023).
Weber, R. Z., Mulders, G., Kaiser, J., Tackenberg, C., Rust, R. Deep learning-based behavioral profiling of rodent stroke recovery. BMC Biol. 20 (1), 232(2022).
Knab, F., et al. Prediction of stroke outcome in mice based on noninvasive mri and behavioral testing. Stroke. 54 (11), 2895-2905 (2023).

Access restricted. Please log in or start a trial to view this content.

Reprints and Permissions

Request permission to reuse the text or figures of this JoVE article

Request Permission

Stroke-Related Animal Models: Methodologies, Applications, and Translational Perspectives

In This Article

Summary

Abstract

Introduction

Review and Perspective

Conclusions

Disclosures

Acknowledgements

References

Reprints and Permissions

Tags

Related Articles