Correspondence: Z. Zheng, Department of Neurosurgery, Heidelberg University Hospital, Im Neuenheimer Feld 400, D-69120 Heidelberg, Germany (tel.: +49 6221 56 36306; fax: +49 6221 56 8342; e-mail: email@example.com).
Subarachnoid hemorrhage (SAH) is a devastating disease associated with death and poor functional outcome. Despite decades of intense research and improvements in clinical management, delayed cerebral ischaemia (DCI) remains the most important cause of morbidity and mortality after SAH. The key role of angiographic cerebral vasospasm, thought to be the main cause of DCI, has been questioned. Emerging evidence suggests that DCI is likely to have a multifactorial etiology. Over the last few years, spreading depolarization (SD) has been identified as a potential pathophysiological mechanism contributing to DCI. The presence of cortical spreading ischaemia, due to an inverse hemodynamic response to SD, offers a possible explanation for DCI and requires more intensive research. Understanding the role of SD as another mechanism inducing DCI and its relationship with other pathological factors could instigate the development of new approaches to the diagnosis and treatment of DCI in order to improve the clinical outcome.
Subarachnoid hemorrhage (SAH) resulting from aneurysmal rupture is a delicate condition frequently leading to death and poor outcome. Its major complication, delayed cerebral ischaemia (DCI), is a serious and poorly understood entity, and remains the main cause of high mortality due to its progression into cerebral infarction. DCI refers to a neurological syndrome of focal and cognitive deficits with a peak on the 10th day, occurring in about 20–40% of patients who survive the initial hemorrhage. For many years it was thought that the primary mechanism of DCI development was an arterial vasoconstriction that leads to ischaemia, known as cerebral vasospasm (CVS) . However, emerging evidence suggests that CVS has only a limited role in DCI development. This is supported by a series of inconsistencies. In general, the incidence of angiographic CVS is about 70%, and that of DCI is about 30% and, as a result, not all patients with CVS develop DCI . Although severe angiographic CVS is associated with reductions in regional cerebral blood flow (rCBF), regional hypoperfusion and oligemia frequently occurred in regions and patients without CVS . To date, nimodipine, a calcium channel blocker, is the only effective drug that can reduce the incidence of DCI and improve patients’ outcome. This is, however, without a clear effect on CVS . Recently, clazosentan, an endothelin receptor antagonist, has been shown to be efficacious in reducing CVS. Nonetheless, the Clazosentan to Overcome Neurological Ischaemia and Infarct Occuring After Subarachnoid Hemorrhage studies failed to demonstrate an improvement in outcome . Therefore, effectively decreasing angiographic CVS does not translate into a reduction of DCI-related morbidity or mortality. This, again, indirectly questions the role of CVS in DCI development per se. As a result, several additional hypotheses besides CVS have been proposed to explain its occurrence; one of those corresponds to spreading depolarization (SD) [1, 6]. In this article we briefly review the evidence that points to SD as an etiological factor for DCI induction following SAH.
Spreading depolarization corresponds to a self-propagating front of depolarization waves of the central nervous system with neuronal and glial cell participation , and can be accompanied by depression of the electrocorticographic (ECoG) activity (spreading depression, Fig. 1). It usually spreads out in all directions at a velocity of 2–5 mm min−1 and resolves after 5–15 min [7, 8]. Underlying this depolarization, a remarkable breakdown of ion homeostasis between extra- and intracellular spaces exists . This imbalance favors the osmotic movement of water causing the neuron to swell and dendrites to distort. Simultaneously, extracellular pH becomes first alkaline and then acidic up to approximately 6.9 . This acidosis is related to the excessive production of CO2 and lactic acid by a pronounced utilization of glucose and oxygen consumption due to the increased metabolic activity. The augmentation in metabolic rate is required to restore ion homeostasis by activation of Na+/K+-ATPase and Ca2+ pumps . SD induces the release of neurotransmitters into the extracellular space . In the particular case of glutamate, it can generate excitotoxicity through the overactivation of N-methyl-d-aspartate (NMDA), α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) or kainate receptors, which allow a calcium overload. The glutamate–calcium cascade can lead to necrosis, thus having an important role in neuronal death .
Spreading depolarization also triggers important changes in cerebral perfusion; in this regard rCBF is affected. It is observed as an increment of more than 100% of blood flow in response to depolarizations, usually sustained for up to 2 min (spreading hyperemia)  (Fig. 2). A low rise in tissue partial pressure of oxygen and decrease in cerebral metabolic rate of oxygen are consistent with the rise in rCBF and onset of depolarizations . However, following this pattern, a reduction of approximately 30% in blood flow develops for up to 2 h (spreading oligemia) [7, 12]. Nevertheless, under pathological conditions, such as hypoxia and ischaemia, this hypoperfusion is accentuated, where SD, instead of inducing spreading oligemia, induces a severe spreading ischaemia (SI), as a result of an inverse neurovascular response that consists of a prolonged hypoperfusion originated by severe arteriolar vasoconstriction [12-14]. In such cases, severe microvascular spasm instead of vasodilatation is coupled to the depolarization phase of SD, rendering neural tissue vulnerable to damage up to the development of widespread necrosis [12, 14]. In addition, according to the degree of ischaemia, SD has been associated with different patterns. A single short-lasting event of SD is usually present in very mild cases of ischaemia, whilst in mild ischaemia a cluster of SDs with intermediate duration can be observed. This type of SDs can cause a cascade of events leading to cell death. Finally, terminal SD is present in severe ischaemia and it is always associated with neuronal death .
SD in experimental SAH models
Several experimental animal studies replicating SAH models have shown the occurrence of SDs. Hubschmann et al. , for the first time, detected neuronal depolarization waves using ion-specific microelectrodes and ECoG in a cat SAH model. This experiment showed that SAH generates a primary cellular dysfunction capable of inducing depolarizations. They found that the cellular response after SAH is characterized by a depolarization accompanied by alterations in ion homeostasis . However, these studies were restricted only to the initial hours after the hemorrhage, and the relationship between SDs and SAH was not clear. Replicating the delayed conditions after subarachnoid hemolysis in a rat SAH model, Dreier et al.  found that the combination of decreased nitric oxide (NO) concentration and increased subarachnoid K+ concentration were able to trigger SDs accompanied by an ischaemic response. This ischaemia, resulting from an inverse hemodynamic response, was coupled to the depolarizations. Later on, another study by the same group showed that hemoglobin combined with either a high concentration of K+ or low glucose in the subarachnoid space causes SI, which led to widespread cortical necrosis in rats .
Similar results were found in models of ischaemic stroke, where depolarizations were linked to microvascular dysfunction (Fig. 3). In such cases, SDs were able to generate vasoconstriction and reduction of the rCBF in the ischaemic area, and the presence of SDs in clusters correlated with infarct growth in mice and cats’ brains [19, 20]. Even though ischaemic stroke is a different etiology, it is possible that SDs may share similar pathways of intervention under hypoxic conditions in these pathologies, changing the normal hemodynamic response to vasoconstriction and causing lasting neuronal damage. However, more research is necessary to discern epiphenomenal depolarization phenomena from injurious SI caused by SD.
Detection of SD and its relationship with DCI in patients with SAH
Based on the finding that SD occurs in SAH models and may cause SI, which is sufficient to induce progressive ischaemic damage in rats, it was proposed that SI is involved in the clinical development of DCI as well . During the past decade, several studies using perifocal ECoG with subdural electrode strip electrodes have confirmed the presence of such depolarizations propagating across the cortex in patients with SAH, as well as the occurrence of depolarizations in clusters in patients who developed DCI (Table 1). The first study assessed the incidence and timing of SDs and delayed ischaemic neurological deficit (DIND) after SAH in humans. This was a prospective, multicenter study by the Co-Operative Study on Brain Injury Depolarizations (COSBID) group with 18 patients . SDs were recorded for up to 10 days by ECoG, and delayed infarcts were verified by serial computed tomography (CT) and/or magnetic resonance imaging scans. A total of 298 SDs was found in 72% of patients. In seven patients DINDs were observed 7.8 days after the hemorrhage and time-locked to new SDs in clusters. A positive predictive value of 86% for a new cluster of SD for DIND was found compared with a negative predictive value of 100% if no cluster occurred. The number of SDs per day was higher in patients with DCI between days 7 and 9 (P = 0.006), and it was observed that SDs with depression periods beyond 60 min were always spatially confined to ECoG channels of cortical areas showing ischaemia on imaging studies. In addition, patients who presented ECoG depression periods longer than 10 min developed a worse outcome at time of discharge (P = 0.008) . This study shows for the first time that SDs have a high incidence in SAH, and that SDs with prolonged depression are an indicator of progressive neuronal damage.
Table 1. Studies that analyze the relationship between SAH, SD and DCI in patients
Aim of the study
SAH patients, n
Patients with SDs, n (%)
Total of SDs
Patients with DCI, n (%)
SDs observed in patients with DCI, n (%)
COSBID, Co-Operative Study on Brain Injury Depolarizations; CVS, cerebral vasospasm; DCI, delayed cerebral ischaemia; DIND, delayed ischaemic neurological deficit; ECoG, electrocorticography; EEG, electroencephalography; NA, not available; NPRI, nicardipine prolonged-release implants; SAH, subarachnoid hemorrhage; SDs, spreading depolarizations. aIsolated SDs. bClusters of SDs. cNumber of SDs using ECoG. dNumber of SDs using simultaneous ECoG/EEG recordings.
Pilot study to determine the duration of depolarizations as measured by the negative direct current shifts in ECoG.
The second study investigated if inverse hemodynamic response occurs in the human brain in order to examine the presence of SI after SAH and its possible relationship with DCI . For this purpose, a prospective, multicenter study with 13 patients with SAH was performed using subdural opto-electrode technology for simultaneous laser-Doppler flowmetry and direct current-ECoG, combined with recording of tissue partial pressure of oxygen. In total 603 SDs were observed. Isolated SDs were recorded in 92% of the patients, and were associated with either physiological, absent or inverse rCBF responses. Hyperemic SDs led to brain hyperoxia, whereas oligemic led to brain hypoxia. Clusters of prolonged SDs were measured in five patients. These clusters were associated with depression periods of cortical activity for 2–3 h, and in one case up to 60 h. Such clusters were associated with SI and structural brain damage as observed in the imaging studies . These studies may lead to the theory that hypoperfusion, as a result of SI, may contribute to the pathogenesis of DCI even in the absence of CVS. These studies are important despite the limited number of patients. They demonstrate for the first time that SDs and SI both occur in humans and are associated with DCI formation.
A subsequent study by the same group found that SDs are capable of influencing oxygen availability in the cortical microcirculation, in particular the study showed that clusters of SDs are able to reduce oxygen supply and increase its consumption having the capacity to promote DIND. In this study a total of 120 SDs were recorded in 88% of the patients with SAH. More than 45% of the SDs were found in two patients who later developed DIND . Recently, Woitzik et al.  demonstrated the occurrence of SD and DCI in the absence of CVS. Copolymer-based nicardipine prolonged-release implants were placed next to the exposed vasculature in 13 patients with SAH after microsurgical aneurysm clipping. The study found 534 SDs in 77% of the patients. Angiography revealed no CVS in 62% of the patients. Only six patients developed DCI, four of those with proven infarcts by imaging studies. In these patients CVS was excluded as the culprit of the ischaemic damage. Furthermore, five patients developed DCI; these patients presented a higher number of SDs and longer periods of ECoG depression, which correlated with the infarct incidence. These results could potentially explain why DCI may occur without CVS, and why CVS does not necessarily result in DCI.
Additional studies found the frequent occurrence of SDs in patients with SAH (incidence of 66–100%). These studies also describe the development of DCI (44–80%) in patients who presented SDs [25-27]. A common finding is that clusters of SDs with prolonged depression periods are temporally linked to ischaemia. Therefore, this clinical evidence adds a strong support of the role of SDs in DCI formation. However, robust clinical trials are needed to confirm this association of SDs and DCI, and prove their causal linkage.
Pathophysiological relationship between SD and DCI
The mechanisms that explain how SD could be involved in the development of DCI after SAH are not completely understood (Fig. 4). However, it is known that pathological conditions, such as hypoxia, low glucose and/or high extracellular K+ concentrations, NO depletion and free hemoglobin products, have the capacity to trigger SDs [17, 18, 28]. All of them are present in the microenvironment after the hemorrhagic event. Important changes that compromise oxygen availability are established right after SAH, with the vasoconstriction of small cerebral vessels within the lapse of minutes due to endothelial dysfunction . Also, during the first 72 h, increments of intracranial pressure and decrements of cerebral perfusion pressure developed that compromise rCBF stability even further. The theory does not contradict the possibility that development of severe CVS may in some cases be solely sufficient to cause DCI , contributing to the reduction of blood flow and brain energy supply. Another important factor involved in the generation of DCI after SAH may be thromboembolism. The presence of thrombi in small cerebral vessels in patients who developed DCI has been reported . Thrombosis and microemboli can also elicit SDs by producing temporary ischaemia . The above-mentioned circumstances are a clear example of hypoxic states generated after SAH that may be related to SD induction.
The relationship between DCI development triggered by subarachnoid blood and SDs has been thoroughly investigated. Brouwers et al.  showed that the risk for DCI occurrence correlates with the total amount of subarachnoid blood on the initial CT scan. However, in this study the site of ischaemic lesions often did not correspond with the blood distribution, indicating the interaction of other mechanisms. It has been demonstrated that products of hemolysis can cause SI and focal necrosis . These are in contact with the pial surface of the brain and overlying vessels. During the course of hemolysis, K+ and hemoglobin are released into the extracellular space and may be related to depolarizations. Hemoglobin binds to NO with high affinity and, under normal circumstances, NO can counteract the vasoconstriction effects of K+ and strengthen the vasodilatation effects of pH , so the net effect after SD is vasodilatation. However, decreases in NO availability together with increases of K+ have been shown to produce a shift from spreading hyperemia to SI . Therefore, under pathological conditions, the effect of SD can be reversed to vasoconstriction owing to the increase of hemoglobin and K+ and the decrease of NO . Conversely, ischaemia caused by vasoconstriction can render the cortex susceptible to more SDs, which may be the mechanism of self-perpetuating electrocortical activity, such as clustered SDs. If metabolic demands are not met by an adequate neurovascular coupling response, this could result in SI and cortical necrosis contributing in this way to DCI development.
Spreading depolarization is a long known phenomenon that has been extensively explored in basic research. Recent data in human subjects confirm that a complex cascade of interconnected molecular, metabolic, electrophysiological and vascular disturbances may lead to ischaemia, homologous to animal models. The discovery of SDs may lead to a better explanation of DCI in patients with SAH. However, the definite role of SD and its therapeutic implications in SAH are yet unclear. Experimental models of DCI after SAH may open new avenues, mechanistic insights and therapeutic targets that will guide challenging studies in the future. At the bedside, remarkable advances made in monitoring brain function will undoubtedly facilitate the interchange between experiment and clinical approaches.
Disclosure of conflicts of interest
The authors declare no financial or other conflict of interests.