We performed a comprehensive literature search using the PubMed (https://pubmed.ncbi.nlm.nih.gov/), Web of Science (https://www.webofscience.com/), and Scopus (https://www.scopus.com/) databases. Search terms were combined using Boolean operators and included: “global cerebral ischemia”, “forebrain ischemia”, “four-vessel occlusion”, “cardiac arrest”, “return of spontaneous circulation”, “CA/ROSC”, “neuronal death”, “delayed neuronal death”, “ischemic duration”, “mitochondrial dysfunction”, “reperfusion injury”, and “rat model”. To ensure coverage of both foundational concepts and contemporary mechanistic advances, we considered publications spanning from the early 1980s through 2025. Seminal studies defining selective neuronal vulnerability and ischemic thresholds were intentionally included alongside recent experimental and translational reports addressing mitochondrial biology, proteostasis, reperfusion dynamics, and systemic inflammation. In general, studies were selected based on their relevance to mechanistic insights, experimental reproducibility, and their contribution to understanding context-dependent determinants of neuronal injury beyond ischemic duration alone.

Studies were included when they met the following criteria: (1) Experimental studies conducted in rats, focusing on global brain IR or CA/ROSC. (2) Use of defined ischemic durations, particularly within the commonly employed range of approximately 5–10 minutes, allows conceptual comparison across models. And (3) Reporting of histopathological, molecular, physiological, or functional outcomes related to neuronal injury or survival. Studies were excluded if they: (1) Primarily involved focal ischemia models (e.g., middle cerebral artery occlusion), (2) Used non-rodent species as the primary experimental system, or (3) Focused exclusively on anesthetic, pharmacokinetic, or technical outcomes without relevance to neuronal injury mechanisms. Model Stratification and Conceptual Framework: Included studies were stratified into two principal experimental paradigms: (1) brain-restricted global IR models, including four-vessel occlusion and related forebrain ischemia paradigms without complete systemic circulatory arrest; and (2) CA/ROSC models, encompassing asphyxial or ventricular fibrillation–induced cardiac arrest followed by standardized resuscitation protocols. Rather than comparing outcomes solely on the basis of ischemic duration, we evaluated studies within a conceptual framework that emphasizes ischemic quality, reperfusion biology, and systemic involvement. Particular attention was given to differences in energy failure depth, mitochondrial integrity, oxidative and calcium-mediated injury, proteostasis disruption, and brain–body interactions.

From each study, we extracted qualitative information regarding: (1) ischemic duration and experimental conditions, (2) regional and temporal patterns of neuronal injury, (3) molecular and cellular injury mechanisms, and (4) reported implications for neuroprotection or recovery. Rather than performing quantitative aggregation, we synthesized evidence thematically, integrating findings from both classical and contemporary literature to construct a mechanistic narrative explaining divergent neuronal outcomes despite comparable ischemic durations.

As a narrative review, this work does not include a formal quantitative assessment of publication bias or effect size. However, we placed particular emphasis on reproducible experimental paradigms, consistency of reported injury patterns, and convergence of mechanistic evidence across independent laboratories. By focusing on rat models exclusively, species-related variability was minimized, allowing clearer conceptual comparison between global IR and CA/ROSC paradigms.

Global brain IR models in rats have been extensively used to investigate the temporal thresholds and regional vulnerability of neurons under conditions of transient, brain-wide hypoperfusion without complete circulatory arrest [1, 5]. The most widely adopted paradigm is the four-vessel occlusion (4VO) model, originally introduced by Pulsinelli and Brierley [5]. This model involves permanent occlusion of the vertebral arteries followed by transient bilateral common carotid artery occlusion, resulting in forebrain ischemia while maintaining systemic circulation. Using this approach, early studies demonstrated that the duration of ischemia is a critical determinant of neuronal outcome, with approximately 5–10 minutes of ischemia sufficient to induce selective neuronal injury in vulnerable regions [5, 6]. Notably, neuronal damage in this model is not distributed uniformly across the brain. Instead, pyramidal neurons in the hippocampal CA1 region exhibit a characteristic pattern of delayed neuronal death, typically becoming evident 48–72 hours after reperfusion [4, 11]. The delayed nature of CA1 neuronal loss has been a defining feature of global IR models and has contributed substantially to the concept of selective neuronal vulnerability. Histopathologically, affected CA1 pyramidal neurons display cytoplasmic eosinophilia, nuclear pyknosis, and progressive neuronal loss over several days following reperfusion [12]. These pathological features occur in the relative absence of widespread necrosis in other brain regions, indicating that ischemic duration alone does not uniformly determine neuronal fate.

Histological and immunohistochemical analyses have further revealed that neuronal degeneration in global IR models is accompanied by robust glial responses. Astrocytic activation, reflected by increased expression of glial fibrillary acidic protein (GFAP), and microglial activation are consistently observed in the hippocampus during the delayed injury phase [13, 14]. In addition, white matter alterations and myelin loss have been reported, indicating that global IR engages both neuronal and non-neuronal compartments [13]. A study has extended classical histopathological observations by incorporating molecular and proteomic approaches. Proteome profiling of the hippocampus after global IR has shown that IR alters protein networks related to mitochondrial function, stress responses, and synaptic integrity [15]. Furthermore, experimental evidence suggests that delayed neuronal death in global IR involves overlapping cell death programs. Zakharova et al. [16] demonstrated that both autophagy- and apoptosis-related pathways contribute to neuronal loss in the hippocampal CA1 region, supporting the interpretation that delayed neuronal death reflects multifactorial disruption of cellular homeostasis rather than a single dominant death mechanism.

Collectively, global brain IR models provide a controlled framework for studying delayed neuronal injury and regional vulnerability under conditions of transient cerebral hypoperfusion. However, preservation of systemic circulation in these models limits the contribution of extracerebral physiological stressors, a distinction that becomes particularly relevant when comparing global IR with cardiac arrest–induced ischemia.

In contrast to brain-restricted ischemia, CA/ROSC models induce complete cessation of systemic blood flow, resulting in simultaneous ischemia of the brain and peripheral organs [7, 17]. In rats, CA is most commonly induced by asphyxia or ventricular fibrillation, followed by standardized cardiopulmonary resuscitation protocols to achieve ROSC [17, 18]. Typical ischemic durations range from 5 to 8 minutes, overlapping with those used in global IR models. Despite these nominal similarities in ischemic duration, neuronal outcomes in CA/ROSC models differ markedly from those observed in global IR. The abrupt loss of cerebral perfusion during CA leads to rapid depletion of adenosine triphosphate, severe metabolic acidosis, and early mitochondrial dysfunction. Consequently, neuronal injury in CA/ROSC models often emerges earlier and extends across broader brain regions, including the hippocampus, cerebral cortex, striatum, and thalamus [7, 19].

Experimental studies have shown that neuronal death after CA/ROSC is accompanied by pronounced blood–brain barrier disruption, cerebral edema, and neuroinflammatory responses during the early reperfusion period [20, 21]. These pathological features reflect the contribution of systemic ischemia–reperfusion injury, which is largely absent in brain-restricted global IR models [22]. Moreover, mitochondrial dysfunction and impaired energy recovery persist after ROSC, further exacerbating neuronal vulnerability [23]. Importantly, CA/ROSC models also underscore the central role of brain–body interactions in post-ischemic brain injury. Peripheral organ dysfunction, systemic inflammation, and metabolic disturbances can secondarily influence neuroinflammation and neuronal survival, emphasizing that ischemic duration cannot be interpreted independently of systemic context [22, 24].

Thus, while CA/ROSC models more closely approximate the clinical scenario of cardiac arrest, they introduce additional layers of systemic and physiological complexity that fundamentally distinguish them from global brain IR models, even when ischemic durations are matched.

Taken together, the experimental evidence reviewed in Sections 3.1 and 3.2 demonstrates that brain-restricted global IR and CA/ROSC models represent fundamentally distinct biological paradigms, despite frequent use of comparable ischemic durations. As summarized in Table 1 (Ref. [4, 5, 6, 7, 11, 13, 14, 16, 17, 18, 19, 20, 21, 22, 23, 24]), global IR models preserve systemic circulation and are characterized by delayed, region-restricted neuronal injury—most prominently delayed neuronal death in hippocampal CA1 pyramidal neurons—allowing mechanistic interrogation of selective vulnerability under relatively controlled reperfusion conditions. In contrast, CA/ROSC models involve complete systemic circulatory arrest followed by biologically hostile reperfusion, resulting in earlier and spatially expanded neuronal injury accompanied by blood–brain barrier disruption, neuroinflammation, and brain–body interactions that substantially modify post-ischemic outcomes. Collectively, these model-specific differences indicate that ischemic duration alone is insufficient to predict neuronal fate and must be interpreted within the broader context of hemodynamic conditions, systemic involvement, and reperfusion biology. Recognition of these distinctions provides a critical foundation for understanding divergent ischemic thresholds and mechanistic trajectories across models, which are explored in the subsequent sections.

The concept of ischemic thresholds emerged from early experimental studies demonstrating that neuronal injury evolves in a stepwise manner as cerebral blood flow and energy availability decline. Thresholds have been proposed for reversible electrical failure, loss of ionic homeostasis, and irreversible structural injury [1, 25, 36]. Importantly, these thresholds should not be regarded as fixed values, but instead vary according to multiple factors, including baseline metabolic demand, temperature, and the speed and completeness of reperfusion. At the cellular level, ischemic neuronal death reflects progressive failure of bioenergetic homeostasis, excitotoxic signaling, and mitochondrial integrity rather than a simple function of ischemic duration alone [26]. Recent experimental work further supports this dynamic framework by demonstrating that molecular stress responses and mitochondrial resilience can shift effective injury thresholds even under nominally comparable ischemic durations [9].

In forebrain ischemia models, partial reductions in cerebral perfusion may permit short-term neuronal survival despite suppressed synaptic activity, whereas more severe or prolonged ischemia leads to delayed or immediate neuronal death [37, 38]. This framework has been particularly influential in interpreting delayed neuronal death in the hippocampal CA1 region, where neurons survive the initial ischemic insult but succumb during the reperfusion phase [4, 11, 39]. More recent evidence indicates that delayed neuronal death reflects progressive failure of cellular homeostasis during reperfusion, including disturbances in proteostasis and mitochondrial quality control, rather than representing a direct consequence of ischemic duration alone [27]. Thus, ischemic duration should be viewed as one component within a broader pathophysiological continuum rather than as an isolated determinant.

In global brain IR models, ischemic durations of approximately 5–10 minutes are sufficient to induce selective neuronal injury, most prominently in the hippocampal CA1 region [5, 6, 39]. Despite comparable durations, widespread cortical or subcortical necrosis is generally absent, and neuronal injury evolves gradually over 48–72 hours after reperfusion [12]. Notably, recent analyses further confirm that even under standardized ischemic durations, neuronal injury remains spatially restricted in global IR models, reinforcing the concept of region-specific vulnerability rather than uniform ischemic damage [28]. This delayed and region-restricted pattern of injury reflects several features of global IR models [11, 29]. First, partial preservation of systemic circulation limits the severity of metabolic derangements during ischemia. Second, reperfusion restores oxygen and glucose delivery relatively rapidly, thereby shifting the dominant injury process toward delayed mechanisms such as oxidative stress, impaired protein homeostasis, and dysregulated calcium signaling. At the cellular level, delayed neuronal death in CA1 pyramidal neurons has been shown to involve apoptotic pathways rather than immediate necrosis, highlighting the prolonged vulnerability of these neurons after transient ischemia [30]. Proteomic and molecular studies further demonstrate that reperfusion in global IR preferentially activates stress-response pathways, including proteasomal dysfunction and endoplasmic reticulum stress, rather than immediate necrotic cascades [27]. Histopathological findings—including cytoplasmic eosinophilia, nuclear pyknosis, and progressive neuronal loss—are therefore interpreted as consequences of post-ischemic cellular dysfunction rather than immediate energy collapse [11, 40]. Taken together, these observations indicate that in global IR models, ischemic duration defines susceptibility rather than inevitability, allowing delayed neuronal death to serve as a window for mechanistic investigation.

In CA/ROSC models, ischemic durations often overlap with those used in global IR experiments, typically ranging from 5 to 8 minutes [17, 18]. Nevertheless, despite these nominal similarities, the temporal onset, spatial distribution, and mechanistic drivers of neuronal injury differ markedly from those observed in brain-restricted IR models. Neuronal injury in CA/ROSC models frequently manifests earlier and involves broader brain regions, including the hippocampus, cerebral cortex, striatum, thalamus, and brainstem [7, 19]. A defining feature of CA/ROSC models is the complete cessation of systemic circulation, resulting in abrupt depletion of cerebral adenosine triphosphate, rapid intracellular acidosis, loss of mitochondrial membrane potential, and collapse of ion homeostasis across multiple brain regions [2, 24]. Unlike global IR, where residual perfusion and rapid reperfusion limit the depth of energy failure, CA produces a more uniform and severe ischemic insult at the whole-body level, a concept emphasized early in cerebral resuscitation research [31]. The reperfusion phase following ROSC further amplifies neuronal injury through systemic mechanisms. Experimental studies consistently demonstrate early blood–brain barrier disruption, cerebral edema, and robust neuroinflammatory activation within hours after ROSC [20, 21]. These processes promote secondary neuronal injury by facilitating immune cell infiltration, cytokine release, and oxidative stress, thereby accelerating neuronal loss beyond the delayed paradigm typically observed in global IR models [32]. At the cellular level, neuronal death in CA/ROSC models reflects a convergence of injury pathways rather than a predominantly delayed program. Early necrotic injury driven by profound energy failure coexists with delayed apoptotic and autophagic processes during reperfusion, as evidenced by caspase activation, mitochondrial dysfunction, and dysregulated autophagy signaling in vulnerable regions [23, 33]. This convergence of injury mechanisms shortens the temporal interval between ischemia and irreversible neuronal damage, effectively shifting ischemic thresholds toward earlier time points. Importantly, CA/ROSC models highlight the critical role of brain–body interactions in post-ischemic neuronal injury. Systemic inflammatory responses, metabolic derangements, and multi-organ dysfunction contribute secondarily to neuroinflammation and impaired recovery, reinforcing the interpretation that neuronal fate cannot be dissociated from extracerebral physiology in this setting [22, 34].

Taken together, findings from global IR and CA/ROSC models indicate that ischemic thresholds are model-dependent and context-sensitive. In brain-restricted IR models, thresholds for irreversible neuronal injury are reached gradually and selectively, permitting delayed neuronal death. By contrast, CA/ROSC models rapidly exceed multiple injury thresholds as a result of systemic ischemia and reperfusion-associated insults, resulting in accelerated and spatially expanded neuronal damage. These divergent threshold dynamics are consistent with contemporary interpretations that distinguish localized cerebral ischemia from whole-body ischemia–reperfusion as fundamentally different pathophysiological entities [29]. These distinctions carry important implications for experimental interpretation. Matching ischemic duration across models does not ensure equivalence of injury severity or mechanism. Instead, ischemic duration must be interpreted in conjunction with hemodynamic conditions, systemic responses, and reperfusion biology. Failure to account for these contextual factors may contribute to inconsistencies in experimental outcomes and limit the translational relevance of neuroprotective strategies, particularly when extrapolating from global IR paradigms to clinical cardiac arrest settings, where prolonged post-resuscitation care and temperature management further modulate injury trajectories [35].

The hippocampus, particularly pyramidal neurons in the CA1 region, represents the most extensively characterized example of selective neuronal vulnerability following transient global ischemia. In global IR models, CA1 neurons typically survive the ischemic episode itself but undergo delayed neuronal death that becomes histologically apparent 48–72 h after reperfusion [4, 11]. Histopathologically, CA1 injury is characterized by cytoplasmic eosinophilia, nuclear pyknosis, and progressive neuronal loss, often accompanied by pronounced astrocytic reactivity marked by increased GFAP expression [12, 41]. These changes occur despite relative preservation of adjacent hippocampal subfields, reinforcing the concept of region-specific vulnerability. More recent studies have further refined this paradigm by indicating that CA1 vulnerability reflects the accumulation of post-ischemic disturbances, including mitochondrial dysfunction, impaired proteostasis, endoplasmic reticulum stress, and dysregulated calcium signaling [27, 40]. Notably, accumulating evidence suggests that CA3 neurons, although more resistant, are not uniformly protected and may exhibit injury under more severe or prolonged ischemic conditions, highlighting a graded rather than absolute vulnerability gradient within the hippocampus [28].

Mechanistic analyses additionally demonstrate that delayed CA1 neuronal death involves overlapping caspase-dependent and caspase-independent pathways together with impaired protein homeostasis, rather than a single dominant process [27, 40]. Consistent with this gradient concept, CA3 pyramidal neurons—while relatively resistant—can undergo injury under conditions of increased ischemic severity or altered reperfusion dynamics [28]. In CA/ROSC models, hippocampal injury often manifests earlier and is less restricted to CA1. Experimental evidence indicates that CA1 neuronal loss may occur sooner after reperfusion and frequently coexists with injury to CA3 and dentate regions, reflecting the deeper metabolic collapse associated with systemic ischemia [7, 19].

Cortical vulnerability differs markedly between global IR and CA/ROSC models. In classical global IR paradigms, neocortical neurons are relatively spared following short ischemic durations, with neuronal injury largely confined to selectively vulnerable regions such as the hippocampal CA1 sector [6]. When cortical injury occurs under these conditions, it is typically delayed and regionally heterogeneous, reflecting partial preservation of cerebral perfusion and rapid restoration of metabolic substrates during reperfusion.

By contrast, CA/ROSC models more frequently produce early cortical neuronal injury, particularly within frontal and parietal cortices. This early cortical involvement reflects the abrupt global energy failure associated with complete circulatory arrest, followed by impaired reperfusion and systemic inflammatory amplification after ROSC [7, 19]. Experimental studies have demonstrated that cortical edema and neuronal degeneration can be detected within days after resuscitation, indicating that cortical ischemic thresholds are exceeded more rapidly in CA/ROSC than in brain-restricted global IR paradigms [19, 20, 22].

Taken together, these observations indicate that cortical vulnerability is strongly shaped by the hemodynamic and systemic context of ischemia, rather than ischemic duration alone.

Subcortical structures, particularly the striatum and thalamus, also show clear model-dependent differences in vulnerability. In brain-restricted global IR paradigms, injury is often concentrated in selectively vulnerable regions such as the hippocampal CA1 sector, whereas striatal and thalamic neuronal loss is comparatively limited or absent at commonly used ischemic durations. This pattern is consistent with classic mapping studies demonstrating non-uniform regional injury distributions after brief forebrain ischemia in rats [6].

In contrast, following CA/ROSC, subcortical injury becomes more prominent, reflecting the combined impact of systemic no-flow ischemia, post-resuscitation hypoperfusion, metabolic acidosis, and reperfusion-associated inflammation. In a multimodal rat CA model, cortical and striatal edema were detectable within several days after resuscitation, followed by evolving white-matter injury over 1–2 weeks, supporting the presence of a broader spatiotemporal injury footprint than that typically observed in brain-restricted global IR models [42]. Moreover, experimental CA paradigms have reported early thalamic injury accompanied by microglial activation within 24 h, particularly under severe asphyxial conditions, indicating that thalamic vulnerability can emerge rapidly when systemic and reperfusion-related stressors are substantial [43]. Contemporary syntheses of preclinical post–cardiac arrest brain injury further emphasize that degenerating neurons can be detected in both cortex and striatum in rat CA models, often in association with blood–brain barrier disruption and inflammatory signatures [44].

Taken together, these findings indicate that striatal and thalamic vulnerability is not simply determined by ischemic duration, but instead reflects the quality of ischemia (brain-restricted versus systemic), the post-resuscitation hemodynamic milieu, and reperfusion biology, which together shift regional injury thresholds in CA/ROSC relative to global IR.

The brainstem represents a critical but often underappreciated target of ischemic injury. In global brain IR models, brainstem nuclei are typically preserved, consistent with partial maintenance of systemic circulation and relatively rapid reperfusion, even when selectively vulnerable forebrain regions such as the hippocampal CA1 sector undergo delayed neuronal death [4, 6]. Recent comparative analyses of experimental ischemia models further support that brainstem structures remain largely resistant to injury in brain-restricted global IR paradigms, unless ischemia is prolonged or compounded by systemic instability [28]. Under these experimental conditions, ischemic thresholds in the brainstem are rarely exceeded at commonly used ischemic durations, and overt histopathological damage is uncommon.

In contrast, CA/ROSC models consistently demonstrate vulnerability of brainstem structures, including respiratory and autonomic nuclei, even after relatively brief ischemic durations [7, 24]. This heightened susceptibility likely reflects the high metabolic demand of brainstem nuclei, their continuous role in cardiorespiratory regulation, and their limited tolerance for systemic hypoxia, hypercapnia, and metabolic acidosis during no-flow ischemia. Experimental studies have shown that CA/ROSC is associated with persistent impairment of brainstem-mediated functions, including blunted hypoxic ventilatory responses and disrupted autonomic regulation, which correlate with reduced long-term survival despite restoration of spontaneous circulation [24, 45]. These observations indicate that brainstem dysfunction is not merely a secondary consequence of supratentorial injury, but instead represents a primary and clinically relevant component of post–cardiac arrest brain injury [2, 34].

The cerebellum represents another region in which vulnerability is strongly influenced by ischemic context. In global IR models, cerebellar neurons—particularly Purkinje cells—are relatively resistant to brief forebrain ischemia, and overt degeneration is uncommon unless ischemic durations are prolonged or compounded by additional stressors such as hyperthermia or hypotension [1, 6]. However, following CA/ROSC, cerebellar injury has been reported in several experimental paradigms, often emerging in a delayed fashion during the post-resuscitation period. Recent studies suggest that cerebellar vulnerability after CA/ROSC is linked to impaired microcirculatory reperfusion, mitochondrial dysfunction, and neuroinflammatory amplification rather than ischemic duration alone [22, 34]. Although cerebellar injury is less consistent than hippocampal or brainstem damage, its occurrence highlights that systemic ischemia and reperfusion-associated disturbances can extend neuronal injury beyond classically defined vulnerable regions, particularly under conditions of severe or prolonged circulatory arrest.

Taken together, these observations indicate that the brainstem and cerebellum are preferentially affected by systemic ischemia and reperfusion stress rather than brain-restricted hypoperfusion, reinforcing the importance of ischemic quality and systemic context in shaping regional patterns of neuronal vulnerability.

Across brain regions, temporal patterns of neuronal injury differ substantially between global IR and CA/ROSC models. Global IR is characterized by delayed and region-restricted neuronal death, with hippocampal CA1 injury typically emerging over several days after reperfusion and remaining the dominant lesion during the subacute phase [4, 11, 12]. This delayed temporal profile reflects neuronal survival through the ischemic episode itself, followed by progressive disruption of cellular homeostasis during reperfusion, rather than immediate energy collapse.

By contrast, CA/ROSC models exhibit a more compressed temporal profile, with early neuronal injury, rapid propagation across multiple brain regions, and overlapping necrotic, apoptotic, and autophagic processes during reperfusion [2, 34]. Experimental CA studies consistently demonstrate that neuronal degeneration, blood–brain barrier disruption, and neuroinflammatory activation can be detected within hours to days after resuscitation, indicating that ischemic thresholds are exceeded earlier and across a broader spatial extent than in brain-restricted global IR paradigms [7, 19].

Taken together, these observations indicate that ischemic thresholds are dynamically shifted by systemic factors and reperfusion biology. Regions that tolerate brief ischemia under brain-restricted conditions may exceed injury thresholds rapidly when systemic ischemia, metabolic derangement, and inflammatory amplification are superimposed. Accordingly, the temporal evolution of injury should be regarded as a model-defining feature rather than a simple function of ischemic duration, with important implications for interpreting experimental outcomes and designing time-sensitive neuroprotective interventions.

Overall, regional and temporal patterns of neuronal vulnerability indicate that ischemic injury cannot be predicted solely by ischemic duration. Instead, neuronal fate reflects a dynamic interplay between region-specific metabolic demands, network connectivity, and intrinsic cellular resilience, as well as the broader physiological context in which ischemia and reperfusion occur. Comparisons between brain-restricted global IR and cardiac arrest–induced ischemia further demonstrate that identical ischemic durations can yield fundamentally different injury phenotypes depending on systemic involvement and reperfusion biology (Table 3, Ref. [1, 2, 4, 6, 7, 11, 19, 20, 22, 24, 28, 34, 42, 43, 44, 45]). Recognition of these spatiotemporal and model-dependent patterns is essential for interpreting experimental outcomes and for designing neuroprotective strategies that are appropriately matched to the underlying ischemic context. If these distinctions are not adequately considered, key mechanisms of selective vulnerability may be obscured and the translational relevance of preclinical neuroprotection studies may be reduced.

A fundamental distinction between global IR and CA/ROSC models lies in the depth and completeness of energy failure during ischemia. In brain-restricted global IR, partial preservation of systemic circulation allows residual oxygen and substrate delivery, resulting in profound suppression of synaptic activity but incomplete depletion of adenosine triphosphate (ATP) in many brain regions [1]. Recent experimental work further suggests that, under these conditions, mitochondrial respiration and membrane potential can remain partially preserved during ischemia, enabling neurons to engage delayed recovery or stress-adaptive responses during reperfusion [9, 38].

By contrast, CA/ROSC is associated with abrupt and complete cessation of systemic circulation, leading to rapid and near-total ATP collapse across the brain. This profound energy failure promotes early mitochondrial depolarization and increases the likelihood of mitochondrial permeability transition pore opening, irreversibly compromising oxidative phosphorylation and priming neurons for early death [2]. Experimental studies further demonstrate that mitochondrial functional recovery during reperfusion is markedly impaired after CA/ROSC, even when cerebral blood flow is restored, indicating that ischemic duration alone does not adequately reflect the depth of mitochondrial injury [23, 47, 48].

In global IR models, reperfusion typically restores oxygen and glucose delivery relatively rapidly and with limited systemic disturbance. Under these conditions, neurons transition from ischemic suppression to delayed injury pathways in which ROS generation and calcium dysregulation evolve over hours to days rather than minutes [12]. Recent evidence further suggests that reperfusion after brain-restricted ischemia preferentially activates mitochondrial and cytosolic oxidative stress pathways without immediately overwhelming endogenous antioxidant defenses, permitting delayed neuronal injury rather than rapid necrosis [15, 27].

By contrast, reperfusion following CA/ROSC is frequently heterogeneous and biologically hostile. Systemic inflammatory activation, endothelial dysfunction, and impaired microcirculatory flow amplify the early ROS burst and exacerbate calcium influx through compromised membrane integrity and excitotoxic signaling [7]. Excessive ROS generation and calcium overload destabilize mitochondria and activate calcium-dependent proteases, accelerating neuronal injury [20, 49].

Notably, clinical trials examining post–cardiac arrest temperature management indicate that modulation of reperfusion-associated metabolic stress can influence neurological outcome, underscoring the biological severity of early reperfusion injury in CA/ROSC [50, 51]. Consistent with these observations, blood–brain barrier (BBB) disruption, cerebral edema, and oxidative damage emerge early after CA/ROSC, indicating that reperfusion-associated stress rapidly exceeds neuronal injury thresholds [20, 21].

In global IR models, delayed neuronal death is closely linked to progressive failure of cellular proteostasis rather than immediate structural collapse. Following transient ischemia, neurons activate adaptive stress responses, including the unfolded protein response (UPR), proteasomal degradation, and autophagy, to manage ischemia-induced protein misfolding and organelle damage [9]. In selectively vulnerable populations, such as hippocampal CA1 pyramidal neurons, these protective mechanisms initially preserve cellular integrity but can become maladaptive under sustained stress, leading to delayed neuronal death [4, 27, 52].

By contrast, CA/ROSC imposes a qualitatively different proteostatic burden on neurons. Profound energy depletion rapidly impairs ATP-dependent protein quality control, resulting in early collapse of proteostasis during reperfusion. Experimental CA studies demonstrate robust engagement of apoptotic cascades alongside necrotic features, reflecting the limited capacity of neurons to sustain controlled stress responses under conditions of severe metabolic failure [16, 40]. Autophagy, which may serve adaptive roles in global IR, appears to be insufficiently activated or dysregulated after CA/ROSC, contributing to mixed necrotic–apoptotic neuronal loss.

In brain-restricted global IR models, neuronal injury develops largely within the central nervous system. Although transient ischemia activates local inflammatory and glial responses, systemic physiology remains relatively preserved, limiting extracerebral contributions to post-ischemic brain injury [14]. Under these experimental conditions, neuroinflammation is driven predominantly by intrinsic mechanisms within selectively vulnerable regions, such as the hippocampal CA1 [28].

By contrast, CA/ROSC induces a systemic ischemia–reperfusion syndrome that profoundly reshapes the cerebral injury landscape. Complete circulatory arrest results in simultaneous ischemia of the brain and peripheral organs, followed by reperfusion accompanied by systemic inflammation and immune activation—collectively described as post–cardiac arrest syndrome [2, 46]. Peripheral organ injury leads to the release of circulating cytokines and damage-associated molecular patterns that amplify neuroinflammation and endothelial dysfunction after ROSC [22, 34].

Beyond inflammatory signaling, persistent systemic hypoperfusion and metabolic instability can further compromise cerebral energy recovery, even after apparent restoration of cerebral blood flow. Experimental and clinical evidence indicate that such systemic disturbances contribute to delayed neurological deterioration and heterogeneous recovery trajectories after CA/ROSC [24, 53]. This biological heterogeneity has, in turn, motivated the development of multimodal neuroprognostication strategies incorporating electrophysiology and neuroimaging to improve outcome prediction [54, 55, 56].

Overall, these mechanisms indicate that ischemic duration is a necessary but insufficient predictor of neuronal outcome. Global IR models primarily engage delayed, region-restricted injury governed by partial energy failure, adaptive stress responses, and controlled reperfusion. By contrast, CA/ROSC is characterized by complete systemic ischemia, severe mitochondrial collapse, intensified reperfusion injury, proteostasis failure, and systemic inflammatory amplification. As a result, nominally identical ischemic durations occupy fundamentally different positions along the injury continuum across experimental models [46].

A major challenge in translational neuroprotection research is the persistent extrapolation of findings from brain-restricted global IR models to CA/ROSC. Although both paradigms are often implemented using comparable ischemic durations, accumulating experimental and clinical evidence indicates that they occupy fundamentally different positions along the ischemic injury continuum [1, 2]. Recent international guidelines and consensus statements increasingly frame post–cardiac arrest brain injury as a dynamic, multisystem syndrome rather than a simple extension of transient global ischemia, underscoring why translation from global IR models often fails [57, 58, 59]. One major explanation for this translational gap lies in qualitative differences in injury biology, rather than ischemic duration alone. In global IR models, partial preservation of systemic circulation allows neurons to retain residual metabolic capacity, engage delayed stress responses, and progress toward injury over hours to days. Under these conditions, therapeutic interventions targeting oxidative stress, excitotoxicity, or delayed apoptotic signaling can meaningfully influence neuronal fate when administered within an extended post-ischemic window [1, 12].

By contrast, CA/ROSC is characterized by abrupt and complete systemic ischemia followed by a biologically hostile reperfusion phase, marked by profound mitochondrial dysfunction, systemic inflammation, and microcirculatory impairment—features collectively conceptualized as post–cardiac arrest syndrome (PCAS) [60, 61, 62]. These systemic modifiers fundamentally reshape the temporal and mechanistic landscape of neuronal injury compared with brain-restricted ischemia.

As a result, neuroprotective agents optimized in global IR paradigms may be administered outside their effective temporal or mechanistic window when tested in CA/ROSC models or in clinical cardiac arrest settings. Experimental and clinical studies demonstrate that mitochondrial failure, blood–brain barrier disruption, and inflammatory amplification occur early after resuscitation, often preceding the therapeutic windows inferred from brain-restricted ischemia models [7, 21, 63]. Accordingly, translational failure should be interpreted primarily as a mismatch between model-specific injury mechanisms and therapeutic targeting, rather than as evidence that candidate interventions are intrinsically ineffective.

The distinction between global IR and CA/ROSC has direct implications for experimental design and interpretation. First, simply matching ischemic duration across models does not ensure equivalence of injury severity or underlying mechanism. Identical ischemic durations can produce delayed, region-restricted neuronal death in global IR but early and widespread neuronal dysfunction in CA/ROSC, reflecting differences in systemic involvement and reperfusion biology [1, 7, 62].

Second, the selection of outcome measures critically shapes translational relevance. Global IR studies have traditionally emphasized histopathological endpoints—particularly delayed neuronal death in the hippocampal CA1 region—as primary indicators of injury. While such measures are well suited for studying selective vulnerability, they may fail to capture early, diffuse, or functionally relevant impairments that dominate CA/ROSC pathology [4, 6]. In CA/ROSC models and clinical cohorts, functional outcomes, brainstem integrity, cerebral perfusion, and systemic recovery parameters are increasingly recognized as more clinically meaningful endpoints than delayed forebrain histology alone [24, 64].

Third, the temporal alignment between injury mechanisms and therapeutic intervention windows requires careful reconsideration. In global IR, delayed neuronal death provides an extended window for mechanistic interrogation and therapeutic intervention. In CA/ROSC, by contrast, injury mechanisms unfold rapidly and overlap during early reperfusion, resulting in compressed and partially overlapping therapeutic windows [2]. Clinical trials of targeted temperature management illustrate this principle, demonstrating that modulation of systemic physiology and prevention of secondary insults may be more critical than precise temperature targets per se [65, 66]. Moreover, clinical studies conducted within the framework of targeted temperature management indicate that physiological instability, hemodynamic requirements, and the timing of prognostic assessment can substantially influence neurological outcome interpretation, underscoring the limited applicability of delayed-injury paradigms derived from global IR [67, 68].

Overall, these considerations argue for a shift toward integrative experimental strategies that explicitly align model selection with mechanistic targets. Rather than treating global IR and CA/ROSC as interchangeable models distinguished only by ischemic duration, future studies should instead recognize them as complementary paradigms that interrogate distinct dimensions of ischemic brain injury.

Global IR models remain indispensable for dissecting delayed neuronal death, selective vulnerability, and intrinsic neuronal stress-response pathways under controlled conditions [69, 70]. By contrast, CA/ROSC models are essential for capturing the impact of systemic ischemia–reperfusion, brain–body interactions, cerebral hypoperfusion, and early reperfusion-associated injury, features that dominate clinical cardiac arrest [62, 64].

From a translational perspective, this integrative framework is consistent with contemporary clinical paradigms emphasizing early hemodynamic optimization, mitigation of systemic inflammation, and multimodal neuroprognostication rather than exclusive focus on delayed neuronal death pathways [71, 72, 73, 74]. Effective neuroprotection after cardiac arrest is therefore likely to require early mitochondrial stabilization, modulation of reperfusion biology, and control of systemic inflammatory responses, rather than delayed interventions extrapolated from brain-restricted ischemia alone [22, 23, 48]. In addition, translational success should be evaluated across the continuum of recovery, extending from early coma assessment to long-term neurological and functional outcomes, which are increasingly recognized as critical determinants of meaningful survival after cardiac arrest [75].

While most neuroprognostication frameworks have been developed in adult populations, recent evidence indicates that integrated electrophysiological and neuroimaging approaches may further improve prognostic accuracy in pediatric cardiac arrest, where etiologies, injury patterns, and recovery trajectories differ substantially from adults [76]. These observations highlight the need for age- and context-specific prognostication strategies when translating experimental findings or adult-derived paradigms to broader clinical populations.

Finally, it should be recognized that early post-ROSC management decisions, including airway control, circulatory stabilization, and metabolic correction, can substantially influence subsequent neurological injury trajectories. Emergency medicine–oriented analyses emphasize that these early systemic interventions frequently precede any opportunity for neuroprotective targeting derived from experimental ischemia models, thereby highlighting a critical translational gap between controlled laboratory paradigms and real-world cardiac arrest care [77].