Address correspondence and reprint requests to John H. Zhang, MD, PhD, Department of Neurosurgery, Loma Linda University Medical Center, 11234 Anderson Street, Room 2562B, Loma Linda, CA 92354, USA. E-mail: firstname.lastname@example.org
Subarachnoid hemorrhage is a devastating disease that can be difficult to manage. Not only is the initial bleeding and rebleeding associated with high mortality, but a large fraction of patients also develop a delayed neurological deficit even when the aneurysm was successfully secured with clipping or coiling. Past research effort has traditionally been focused on vasospasm, which was conceived to be the sole factor for delayed neurological deficit. The failure of anti-vasospastic drugs to improve outcome in clinical trials has brought into focus the significance of early brain injury. The immediate events associated with subarachnoid hemorrhage, including increased intracranial pressure, decreased cerebral blood flow and global ischemia initiate a cascade of pathological changes that occur before the onset of delayed vasospasm. These pathological changes in the very early stage of the hemorrhage propagate and cause blood–brain barrier disruption, inflammation, oxidative stress and cell death. Focusing only on the treatment of vasospasm with complete disregard for early brain injury is insufficient for the management of subarachnoid hemorrhage. Instead, a therapeutic intervention has to aim at stopping the molecular cascades of early brain injury that may lead to long-term deficits in addition to vasospasm. We review the pathological mechanisms of early brain injury, which may reveal new therapeutic avenues that can be exploited to serve as combination therapy with anti-vasospasm medications in the future.
Aneurysmal subarachnoid hemorrhage (SAH) represents 5–7% of all stroke types (Becker 1998; Kaptain et al. 2000) and affects about 10 per 100 000 individuals annually (Pluta et al. 2009). Following aneurysmal hemorrhage, 30% of patients die within the first days of initial bleeding (Broderick et al. 1994) and the overall mortality and combined morbidity is greater than 50% (King1997).
The most important cause of delayed deterioration, that is, delayed ischemic neurological deficits (DIND) (Vergouwen et al. 2010) and morbidity has been identified for decades as vasospasm. Cerebral vasospasm (CVS) as a syndrome was first reported in 1951 (Ecker and Riemenschneider 1951), and subsequent research showed that arterial narrowing after SAH leads to ischemia, and ultimately infarctions and poor outcome (Sehba et al. 2011). Since then tremendous research effort has been made to determine the responsible mechanisms of vasospasm and to develop treatment modalities to reduce the ischemic deficits attributed to CVS. These treatments, however, failed to show improved outcome (Dumont et al. 2003; Nishizawa and Laher 2005; Rothoerl and Ringel 2007; Pluta et al. 2009; Velat et al. 2011).
CONSCIOUS-1, a randomized, double-blind, placebo-controlled phase 2 trial, showed that Clazosentan, an endotelin (ET)-1 receptor antagonist administered 56 h after the initial bleeding was able to reduce moderate and severe vasospasm from 66% to 23% in a dose-dependent manner (Macdonald 2008). However, CONSCIOUS-2 trial failed to show improved outcome, mortality and vasospasm-related morbidity (Macdonald et al. 2011). The fact that among the SAH survivors 21% developed delayed ischemic injury without showing angiographic vasospasm and only 20–30% who did develop vasospasm suffered from delayed ischemic injury (Alaraj et al. 2009) has led researchers to cast doubt on the importance of vasospasm and to seek new concepts in the treatment of patients suffering from SAH.
The term early brain injury (EBI) was first coined in 2004 (Kusaka et al. 2004) and refers to immediate injury to the brain before the development of vasospasm. In recent years, increasing research efforts are directed at elucidating the mechanisms of EBI. Mounting evidence that the pathological derangements of EBI starts immediately after ictus at the first impact and lasts 72 h until the development of vasospasm (Pluta et al. 2009). It can be suggested that the etiology of vasospasm may be linked to EBI, because they share many of the same pathogenic factors (Cahill and Zhang 2009; Ayer and Zhang 2010). The purpose of this review is to show the the mechanisms of EBI pointing out the common pathway changes seen both in EBI and vasospasm. For this purpose first, we briefly provide insight into the pathophysiology of vasospasm to illustrate that the two entities are inherently connected and more importantly vasospasm can be a late manifestation of EBI.
The physiopathology of cerebral vasospasm and clinical trials
Cerebral vasospasm is a multifactorial process and involves different pathological changes at different time points (Zubkov et al. 2003). Altered gene expression was noted in experimental SAH that includes membrane receptors, inflammation-related cytokines, extracellular matrix, cell adhesion molecules, pro-apoptotic proteins and metabolic enzymes (Kasuya et al. 1993; Onda et al. 1999; Aihara et al. 2001; Macdonald et al. 2002; Vikman et al. 2006).
Among these, the vasoactive peptides and amines released into the subarachnoid space after the rupture of an aneurysm include neurogenic factors, biogenic amines (such as histamine and norepinephrine), 5-hydroxytryptamine (5-HT) reductase, eicosanoids (such as prostaglandins, thromboxans and leukotrienes), endothelin, hemoglobin, nitric oxide (NO) and free radicals. The effects of these vasoactive factors have long been studied in regard to the pathogenesis of CVS (Nishizawa and Laher 2005). The main pathological changes responsible for cerebral vasospasm can be summarized under (i) smooth muscle contraction, (ii) endothelial dysfunction, (iii) inflammatory changes and (iv) gene expression (Rothoerl and Ringel 2007).
The central event in vascular smooth muscle contraction is the increase in intracellular calcium concentration, which activates myosin light chain when combined with calmodulin. myosin light chain further interacts with actin filaments to cause contraction and CVS (Bulter et al. 1996; Sato et al. 2000). Protein kinase C and RhoA are the main mediators in the vascular signal transduction pathway (Nishizawa and Laher 2005). Protein kinase C alters several other intracellular signaling pathways including mitogen-activated protein kinase, protein tyrosine kinase and Rho kinase (Zubkov et al. 2003). The dysfunction of endothelial integrity and the imbalance between vasospastic amines like ET and prostaglandins and vasodilators like superoxide anion radicals and endothelium-derived hyperpolarizing factor released by endothelial cells may lead to CVS (Faraci and Heistad 1998).
The ET system has been widely studied in regard to the pathogenesis of CVS. Astrocytes and leukocytes release ET-1, a potent vasoconstrictor in response to inflammation and ischemia after SAH (Pluta et al. 1997; Fassbender et al. 2000). In an attempt to block ET-1-mediated constriction of cerebral arteries, different approaches have been studied in experimental models. These approaches include blocking ET-1 biosynthesis (Caner et al. 1996; Onoda et al. 1996), reducing extracellular ET-1 levels (Yamaura et al. 1992), blocking ET-1 receptors (Clozel and Watanabe 1993; Macdonald et al. 2008) and inhibiting the up-regulation of endothelin receptors (Beg et al. 2006). As mentioned above, the clinical trials on Clazosentan, a potent ET-1 antagonist, has failed to improve outcome (Macdonald et al. 2008, 2011).
Nitric oxide is synthesized by endothelial nitric oxide synthase (e-NOS) in the endothelium and perivascular nerves (Pluta 2005). It is speculated that after SAH, expression of e-NOS is increased but uncoupling of electron transport chain occurs and this results in a decrease in NO levels but increase in peroxynitrite (Sabri et al. 2011). An important enzyme for NO production is inducible NO synthase, which is inactive under physiological condition (Xie et al. 1992). Experimental studies indicate an increase in inducible nitric oxide synthase (i-NOS) mRNA after hemorrhage from day 1 to 7, which was postulated to cause CVS (Sayama et al. 1999); however an i-NOS inhibitor failed to reduce blood–brain barrier (BBB) disruption and improve neurological scores in experimental subarachnoid hemorrhage (Yatsushige et al. 2006).
Statins induce endothelial nitric oxidase synthase and increase nitric oxide production in addition to their salutary effects on endothelial and neuronal apoptosis, inflammation and angiogenesis (McGirt et al. 2006). Although several randomized clinical trials demonstrated detrimental results (McGirt et al. 2002; Lynch et al. 2005; Chou et al. 2008), meta-analysis of pooled data revealed that statin treatment significantly reduce the risk of vasospasm and the incidence of DINDs in SAH patients (Sillberg et al. 2008).
Post-SAH inflammation and injury includes increased expression of cell adhesion molecules, cytokines, leukocyte adhesion and production of complements (Rothoerl and Ringel 2007). Adhesion molecules, such as vascular cell adhesion molecule-1, intercellular adhesion molecule-1 and E-selectin facilitate leukocyte migration and consequently cause inflammation (Handa et al. 1995; Bavbek et al. 1998; Polin et al. 1998; Lin et al. 2005). Similarly these molecules were found elevated in SAH patients and were related to the occurrence of delayed neurological deficits (Mack et al. 2002; Mocco et al. 2002). C-reactive protein was elevated in patients after SAH, and significantly higher values were detected in patients who developed DIND (Dumont et al. 2003; Rothoerl et al. 2006). Anti-inflammatory therapy, such as in the clinical trials of tirilazad, a non-glucorticoid aminosteroid inhibitor of lipid peroxidation, was shown to improve outcome and decrease mortality in poor grade male patients (Kassell et al. 1996). However, a meta-analysis including five randomized placebo-controlled trials found no evidence that tirilazad reduces the risk of death and disability after SAH (Zhang et al. 2010).
Other notable clinical reports include the most recent meta-analysis of 16 randomized controlled trials on calcium canal blockers after SAH and concluded that nimodipine reduces vasospasm-associated DIND and improves overall clinical outcomes following aneurysmal SAH (Dorhout Mees et al. 2007; Velat et al. 2011). A Rho-Kinase inhibitor Fasudil has been proven to be effective in reducing angiographic vasospasm, however exhibited no significant benefits on the incidence of DIND or clinical outcome (Shibuya et al. 1992; Zhao et al. 2006).
Early brain injury
The cascade of EBI starts immediately after aneurysm rupture, which not only is responsible for the initial signs and symptoms of SAH, but also contributes to the delayed neurological deterioration and poor long-term outcome traditionally linked to vasospasm (Suzuki et al. 2010b; Sherchan et al. 2011).
With the paradigm shift from delayed vasospasm to EBI, the research focus has changed from large arteries to brain parenchyma, and thus previously commonly used experimental models such as the double blood injection model (Titova et al. 2009) has fallen out of vogue in recent years in favor of the endovascular perforation model (Bederson et al. 1995). In comparison, for translational research, the double injection model was found more suitable for the study of vasospasm (Gules et al. 2002), whereas the endovascular perforation model served as a better simulation of clinical subarachnoid hemorrhage for the study of EBI (Lee et al. 2009b). This is because of the more natural interplay in response to altered hemodynamics between the systemic circulation (Cushing's response) and the intracranial circulation seen in the endovascular perforation model. The mechanisms responsible for the EBI are discussed below (Figs 1 and 2).
Intracranial pressure and cerebral perfusion pressure
It has been shown both in clinical and experimental studies that there is an acute rise in the intracranial pressure (ICP) and a resultant decrease in the cerebral perfusion pressure (CPP) after the onset of subarachnoid hemorrhage (Voldby and Enevoldsen 1982; Bederson et al. 1995). Increase in ICP was attributed to the blood filling of the subarachnoid space, which also obstructs cerebrospinal fluid outflow, causes vasoparalysis and consequently distal arteriolar vasodilation (Grote and Hassler 1988; Brinker et al. 1990). Thus, the magnitude of ICP increase was directly correlated with outcome (Heuer et al. 2004; Westermaier et al. 2009). Decrease in CPP negatively correlated with ICP but was not by itself sufficient to cause perfusion arrest (Kuyama et al. 1984). Interestingly, CPP reduction was not correlated with clinical outcome (Heuer et al. 2004).
Cerebral blood flow
Cerebral blood flow (CBF) was shown to decrease to almost zero after the initial impact of SAH in experimental models, which can explain the sometimes manifested syncope in patients. Perfusion arrest may or may not recover depending on the severity of the hemorrhage (Bederson et al. 1995; Alkan et al. 2001; Ostrowski et al. 2005). The mortality rate was 100% when CBF was reduced to less than 40% of baseline for 60 min after SAH, while less CBF reduction resulted in 19% mortality (Bederson et al. 1998).
A very recent study showed that more than 70% arterioles constricted by 22% to 33% up to 3 days after the hemorrhage and approximately 30% of constricted arterioles were occluded by microthrombi suggesting that micro-arteriolar constrictions and microthrombosis may explain the early CPP-independent decrease in CBF and early perfusion deficits after SAH (Friedrich et al. 2012). The early CBF reduction was accompanied by reduced cerebral metabolic rate of oxygen (Frykholm et al. 2004).
Blood–brain barrier disruption and brain edema
After SAH, global brain edema was present on admission CT in 8% of patients and developed secondarily in another 12% and was an independent risk factor for mortality and poor outcome (Claassen et al. 2002). Brain edema was distributed in gray and white matter as well as the cortex, the deep nuclei and the parasagittal watershed areas (Shigeno et al. 1982; Kamiya et al. 1983). Both primary vasogenic and secondary cytotoxic components were implicated (Doczi 1985; Orakcioglu et al. 2005).
In vitro studies provided evidence that the degradation products of blood such as oxyhemoglobin (Meguro et al. 2000), inflammatory mediators and vasoactive substances that were detected in patient after SAH, among others tumor necrosis factor-alpha (TNF-α) (Jakobsen et al. 1990), thromboxane A2 (TXA2) (Gao et al. 2000) and natriuretic peptides (Suenobu et al. 1999) can cause endothelial apoptosis. A non-specific caspase inhibitor benzyloxycarbonyl-Val-Ala-Asp (OMe) fluoremethylketone (Z-VAD-FMK) was shown to reduce endothelial apoptosis (Park et al. 2004) and thereby attenuated BBB disruption and brain edema. In a recent study of experimental SAH, the inhibition of p53 up-regulated modulator of apoptosis attenuated endothelial apoptosis and brain edema and it was postulated that p53 up-regulated modulator of apoptosis can play an orchestrating role in the induction of endothelial apoptosis (Yan et al. 2011), whereas p53-related endothelial apoptosis in major cerebral arteries was associated with vasospasm (Cahill et al. 2006a, 2007).
Accumulating evidence suggests a role of matrix metallopeptidase 9 (MMP-9) in the early disruption of BBB after subarachnoid hemorrhage (Guo et al. 2010). MMP-9 degrades extracellular matrix of cerebral microvessel basal lamina, including collagen IV, laminin, fibronectin and inter-endothelial tight junction proteins such as zona occludens-1 (Sehba et al. 2004; Yatsushige et al. 2007; Suzuki et al. 2010a). Basal lamina degradation starts as early as 6 h and peaks to a maximum level at 48 h after experimental SAH (Scholler et al. 2007).
An ischemic insult to the brain may trigger complex cellular events which can lead to apoptotic and necrotic cell death (Yuan 2009). Necrosis is an energy-independent process, whereas apoptosis is energy-dependent. As the initial bleeding and the ischemia are not long enough to cause large ischemic tissues necrosis, it is not a major event in SAH (Cahill et al. 2006b).
Apoptosis, especially neuronal apoptosis, however, has been the main focus of several studies in the last few years on EBI after SAH (Endo et al. 2006; Duris et al. 2011; Hasegawa et al. 2011a; Altay et al. 2012). Apoptosis was demonstrated in the cerebral cortex as well as the hippocampus and was linked to late cognitive and memory dysfunctions after SAH (Kreiter et al. 2002; Sherchan et al. 2011). Apoptosis occurs not only in endothelial cells and neurons, but also astrocytes and oligodendrocytes 24 h after SAH (Prunell et al. 2005).
There are a number of apoptotic pathways that are believed to play a role in SAH: the death receptor pathway, caspase-dependent and -independent pathways, as well as the mitochondrial pathway (Hasegawa et al. 2011b). Intrinsic mechanisms appear to be mainly caspase-dependent (Cheng et al. 2009); however, caspase-independent intrinsic mechanisms involving free radicals also exists after SAH (Endo et al. 2007).
Autophagy is a physiological process for the removal of aged proteins and organelles through an autophagosomal–lysosomal pathway that is essential for cell homeostasis and survival but can also promote cell death (Wang and Klionsky 2003). Therefore, the role of autophagy in acute cerebral vascular accidents is unclear and there is controversy over whether activated autophagic pathway represents a mechanism of cell death or a rescue mechanism of endogenous neuroprotective response (Carloni et al. 2008). Autophagy pathway was found activated in SAH and it was speculated to have a beneficial neuroprotective role in subarachnoid hemorrhage (Lee et al. 2009a; Jing et al. 2012; Wang et al. 2012).
It has been shown that excess stress to the endoplasmic reticulum (ER) triggers apoptosis (Yorimitsu et al. 2006) but the role of ER stress after SAH was unclear until recently. Silencing the C/EBP-homologous protein, a critical step for ER stress-induced apoptosis, attenuated EBI after experimental SAH (He et al. 2012).
Changes in ion hemostasis
As mentioned above, intracellular calcium increase in endothelial and smooth muscle cells is one of the most attributed factors for cerebral vasospasm. Experimental studies suggest that this pathological rise starts at the very early stage as soon as 15 min after SAH (Kohno et al. 1991; Meguro et al. 2001; Ishiguro et al. 2008).
Magnesium (Mg) is a physiological antagonist of calcium and plays an important role in maintaining intracellular calcium concentration (Sehba et al. 2012). The total serum Mg levels remain unchanged and the biologically active free ionized form of Mg falls after injury (Memon et al. 1995). The effects of Mg include inhibition of platelet aggregation, inhibition of ET-1 synthesis and excitatory aminoacid (EAA) release, vasodilation through the release of endothelial NO and increased synthesis of prostacyclins (Nadler et al. 1987; McLean 1994; Yang et al. 2000; Berthon et al. 2003; van den Bergh et al. 2004). Magnesium pre-treatment decreased the duration of ischemic depolarization and reduced ischemic brain lesions after experimental SAH (van den Bergh et al. 2002). However, a randomized double-blinded, placebo-controlled, multicenter study failed to show clinical benefit of intravenously administered magnesium sulphate in SAH patients (Wong et al. 2010).
Nitric oxide is a potent vasodilator and pathological alterations in NO has been extensively studied for the pathophysiology of vasospasm (Afshar et al. 1995; Pluta 2005; Durmaz et al. 2008) and EBI (Schwartz et al. 2000; Sehba et al. 2000). Within 10 min after SAH, blood extravasation leads to acute vasoconstriction caused by scavenging of NO (Sehba et al. 2000). NO levels subsequently increase above basal levels at 24 h after SAH (Yatsushige et al. 2006). Pathological rise in cerebral NO levels after 24 h has detrimental effects (Ayer and Zhang 2008) because after this point NO acts as a free radical in itself and in the form of peroxynitrite, to damage the cell membrane of endothelial and smooth muscle cells (Beckman et al. 1990).
Early generation of oxygen free radicals followed by oxidative stress is one of the main pathological changes after SAH (Gaetani et al. 1990; Ersahin et al. 2009). Free radicals generated after SAH involve superoxide anion, hydroxyl radical, hydrogen peroxide, nitric oxide and peroxynitrate (Gaetani et al. 1990, 1994; Marzatico et al. 1993; Asano and Matsui 1999; Ersahin et al. 2010a). Oxygen-free radicals (ROS) are generated early after SAH and cause lipid peroxidation and hemoglobin auto-oxidation and induce oxidative stress, which together contributes to early and also delayed brain injury in rats as well as in humans (Gaetani et al. 1990; Marzatico et al. 1990; Lin et al. 2006). NADPH oxidase, one of the major producers of superoxide anion in the brain, showed increased enzymatic activity at 24 h after SAH. Lipid peroxidation was evident at 6 h and escalated at 24 h after SAH (Ostrowski et al. 2006). After experimental SAH in transgenic mice over-expressing CuZn-superoxide dismutase, vasospasm and EBI were attenuated because of reduction in oxidative stress (Kamii et al. 1999; Endo et al. 2007). Unfortunately, antioxidants have failed to improve outcome in clinical trials after SAH (Gomis et al. 2010).
Intracerebral microdialysis monitoring of patients with SAH has shown signs of delayed ischemia indicated by dramatic changes in extracellular concentrations of EAA including glucose, pyruvate, lactate, glycerol and glutamate (Unterberg et al. 2001). A NMDA receptor antagonist felbamate, aimed at reducing excitatory neurotoxicity from abundant extracellular glutamate, has attenuated BBB disruption 48 h after SAH suggesting the role of EAA in the pathogenesis of EBI (Germano et al. 2007).
The proinflammatory cytokines TNF-α, IL1-β and IL6 were found increased in experimental studies in the early period and were proposed to lead to BBB disruption (Sozen et al. 2009; Sugawara et al. 2009; Ersahin et al. 2010b; Suzuki et al. 2010c). Inhibition of IL1-β attenuated EBI after SAH (Sozen et al. 2009).
Management of subarachnoid hemorrhage remains a challenge. The lack of a single mechanism that sustains EBI or vasospasm is a significant impediment to the design of a defined and standard therapeutic approach. With advances in understanding the physiology and pathology of EBI, it becomes clear that the mechanisms leading to vasospasm and EBI are not mutually exclusive. In fact, many of the pathogenic factors of vasospasm are also implicated in EBI, and thus it can be suggested that vasospasm and EBI are not different entities but rather that vasospasm is a late manifestation of EBI after subarachnoid hemorrhage. The future of SAH research is at a turning point, from a tunnel-visioned search for a ‘silver-bullet’ against vasospasm to a revised multipronged approach to counter a multitude of interrelated pathological pathways for improved outcome.
Conflicts of interests
The authors have no potential conflicts, nor have they received grants, speakers fees, etc from any commercial body within the past 2 years.