The dual role of the neuroinflammatory response after ischemic stroke: modulatory effects of hypothermia

Inflammation is an essential tool to defend oneself against infectious organisms. However, it becomes detrimental when it is prolonged or attacks self antigens [1]. Stroke and neurodegenerative diseases such as Alzheimer's disease, multiple sclerosis and Parkinson's disease are associated with a chronic inflammatory response [2, 3]. After ischemic stroke, the death of ischemic neurons and especially the release of necrotic cell debris triggers inflammation resulting in strong activation of phagocytic cells [1, 4]. Over the past two decades, our understanding of the inflammatory response after stroke and in other diseases has increased due to extensive research. Previously, it was thought that the inflammatory response in brain was beneficial and necessary for repair. Later, it became clear that this "neuro" inflammatory response could be detrimental too, and that even peripheral immune responses can be regulated by the brain [5]. Furthermore, injury to the brain can make the body more vulnerable to systemic infections. For example, a central nervous system injury-induced immunodepression syndrome has been identified in experimental stroke models leading to spontaneous systemic bacterial infections within 3 days after stroke [6‐8].

Stroke is a broad term that includes conditions caused by occlusion of or hemorrhage from a blood vessel supplying the brain [7, 9]. Its incidence remains high and the number of approved therapies low. As our society ages, the number of stroke patients continues to increase, and will become an important socio-economic burden, as 80% of patients who survive their stroke remain permanently disabled. Ischemic strokes represent more than 80% of all cases and are characterized by the occlusion of a blood vessel due to a thrombus or embolus [10, 11]. The location and the size of the ischemic area of the brain varies, depending on which artery is occluded, thereby causing metabolic and functional dysregulations [12, 13]. Furthermore, the occlusion can be permanent or transient, the latter meaning that reperfusion will occur. After ischemic stroke, two main regions of damage can be defined according to the remaining blood supply. The area in the brain where complete abolishment of blood supply occurred (blood flow reduced to less than 12 ml/100g/min), is called the core of the insult. Complete, or almost complete, energetic failure resulting in necrosis, defines this region. The penumbra is the area surrounding the core which is hypoperfused during the occlusion period. Collateral blood supply from surrounding arteries ensures that a flow of approximately 30 ml/100g/min is maintained. Some energetic metabolism persists in this region and if reperfusion can be restored quickly, this tissue can be salvaged [14‐17]. Consequently, the penumbra is an attractive target to rescue brain tissue as this region can remain potentially viable for 16 to 48 hours, enabling clinicians to intervene and reduce post-stroke disability [7, 15].

There are excellent reviews elaborating extensively on the complex ischemic cascade after stroke [5, 7, 14, 17, 18]. Briefly and much simplified, three phases can be characterized in infarct progression. The acute phase starts within minutes to a few hours after stroke onset in which the decrease in cerebral blood flow perturbs the ionic homeostasis. This leads to increased intracellular calcium concentrations and stimulation of glutamate release, causing excitotoxicity and a spreading depression throughout the ischemic region [14, 18, 19]. Water shifts to the intracellular space due to osmotic gradients and cells swell. The resulting vasogenic edema can influence reperfusion negatively and causes intracranial pressure, vascular compression and herniation [18]. Furthermore, generation of reactive oxygen species (ROS), especially if reperfusion takes place, can damage membranes, mitochondria and DNA, leading to misfolding of proteins and enzyme dysfunctions. In the second, subacute phase (a few hours to a few days after ischemia), an apoptotic and neuroinflammatory response develops as a result of the stimulatory influences of the acute phase [14, 17, 18]. The high intracellular calcium concentrations built up during the acute phase lead to an overactivation of several proteolytic enzyme systems. Finally, in the chronic phase, which can last up to some months after the actual ischemic stroke, repair and regeneration will determine the ultimate extent of damage [19]. As contradictory as it may appear, reperfusion following an occlusion may exacerbate neuronal injury through increased production of ROS and a stronger inflammatory response [16, 20]. On the other hand, quickly re-establishing the blood supply is vital as it may allow salvation of neurons in the penumbra from irreversible damage, as energy metabolites and cellular membrane ionic gradients would be restored to normal levels [10]. Recanalization is the only approved and effective therapy for stroke to date. Recombinant tissue plasminogen-activator (rt-PA) must be administered within 4.5 hours after stroke onset to restore reperfusion. Unfortunately, due to the short time window and several risk factors, this treatment can only be administered in 5-10% of cases [11, 15]. This emphasizes the need for other neuroprotective therapies not influencing revascularisation, but a strategy that antagonizes, interrupts or slows down injurious biochemical and molecular events in cases treatment with rt-PA is not indicated [21]. Such neuroprotective strategies would primarily focus on reducing the extent of damage in the penumbral region and thus the outcome after stroke [15]. Combining quick reperfusion with the right neuroprotective strategy to address the downstream damaging cascades induced by the occlusion could lead to a better outcome. Nowadays, hypothermia seems to be the most promising neuroprotective therapy that has been translated to clinical practice [22, 23]. Many other proven neuroprotectants in experimental models have not improved outcome after ischemic stroke in clinical settings. Several reasons may explain this discrepancy. Firstly, research is performed in many different stroke models ranging from permanent occlusion models to transient ones. Furthermore, the duration of transient occlusion can range between 30 minutes to 3 hours. Therefore, as functional and structural damage after ischemia depends on its severity and duration, it is important to stress which experimental model was used [9]. Secondly, experiments are generally performed in healthy young rodents. But in reality, most patients who suffer stroke have several risk factors (e.g. age, diabetes and hypertension) which negatively interfere with the ischemic cascade. Thirdly, the adaptation mechanism of high phylogenic species such as human beings is expected to be different from lower phylogenic ones. Most experimental approaches involve occlusion of the middle cerebral artery (MCA), the same artery that is most commonly occluded in clinical stroke [11]. In order to improve the relevance of preclinical studies to clinical practice, STAIR (Stroke Therapy Academic Industry Roundtable) criteria were set up [10, 22]. Neuroprotective agents that target the acute effects of the ischemic cascade (e.g. glutamate release) do not produce any significant improvement in outcome in a clinical setting, partly due to the (too) short time window to start such therapies. Modulation of later events, like inflammation, could lead to successful neuroprotective therapy [18]. However, in clinical trials, anti-inflammatory drugs have not yet shown the expected results [20, 21]. It is hypothesized that such discrepancies indicate that the neuroinflammatory response after stroke is not entirely detrimental and that, without this response, the damage would be even greater. Inflammation causes cell death acutely, but in a later stage, it may help restore body balance [24, 25]. Or, vice versa, chronic inflammation can cause tissue damage as well. Therefore, it is necessary to understand the detrimental and neuroprotective effects of each component of the neuroinflammatory response at different time points after ischemic stroke. This will allow a more specific approach to develop therapies targeting different aspects of neuroinflammation (see Table 1).

As stroke is a multi-faceted phenomenon, a neuroprotective approach of choice should act on several levels of the ischemic cascade. With respect to inflammation, toxic effects should be inhibited while the associated repair mechanisms should be stimulated. Moderate (30-32°C) to mild (32-34°C) hypothermia could be a good option, as it acts on several pathways of the ischemic cascade and exerts differential effects on the inflammatory response at different time points [23]. Indeed, data from the literature indicate that stimulating and inhibiting effects may be involved in the neuroprotective effects of hypothermia. In cardiac arrest, hypothermia has already proven to be neuroprotective in a clinical setting [155]. Furthermore, in experimental models of transient cerebral ischemia, moderate-to-mild hypothermia can reduce infarct volume by more than 50% [156, 157]. However, in permanent MCAo models the effect is less clear.

Fever occurs often after ischemic stroke and increases cell damage and cell death [158]. If hypothermia is induced, more neurons survive the ischemic period [159]. However, the neuroprotective effect of hypothermia is not yet fully elucidated. Hypothermia has no effect on the cerebral blood flow in the penumbra during ischemia, but can block hyperperfusion after ischemic insult [23, 160]. Furthermore, the neuroprotective effect of hypothermia cannot be explained by solely a reduction in cerebral metabolic demand or glutamate release as hypothermia still reduces infarct volume when initiated in the subacute phase of the ischemic cascade, when the initial increase in glutamate has already passed [156, 161, 162]. Consequently, hypothermia affects several pathways.

Hypothermia has effects on the neuroinflammatory response. Unlike anti-inflammatory drugs, hypothermia exerts both stimulating and inhibiting effects on the inflammatory response (see table 1). Hypothermia reduces the increase of IL-1β after stroke [153, 162‐164]. It would be interesting to know what a combination of hypothermia with IL-1ra could accomplish. While there is no consensus on the role of TNF-α after stroke, there is also as of yet no consensus on the effect of hypothermia on TNF-α. Most studies show a reduction of TNF-α after hypothermic treatment. More interesting is that mild hypothermia can also influence the expression of TNF-α receptors and the activation of NF-κB [23, 165]. Hypothermia can reduce the increased expression of TNF receptor 1 as early as 15 minutes after traumatic brain injury [165]. This receptor could be associated with detrimental effects after brain injury. It would be interesting to know if hypothermia also decreases the expression of TNF receptor 2. A recent study using a transient MCAo model showed that pre- as well as post-ischemic hypothermia is able to attenuate HMGB1 protein levels in brain and plasma [166]. However, little literature exists regarding the influence of hypothermia on the expression of the main receptors of HMGB1. TNF receptors and TLRs are known to primarily mediate activation of NF-κB. Indeed, hypothermia has anti-apoptotic effects that could also be mediated by affecting NF-κB [96, 153, 157]. In normal conditions, NF-κB is present in the cytoplasm, but is bound to a family of inhibiting proteins. To become activated, IκB kinase must phosphorylate these inhibiting proteins to liberate NF-κB and let it enter the nucleus and induce gene expression [153, 167]. In a 2-hour MCAo model, increased NF-κB binding was observed as early as 2 hours after ischemia. Intra- and post-ischemic hypothermia was able to reduce IκB kinase expression, resulting in less NFκB translocation which could be the result of two pathways [153, 167, 168]: (1) hypothermia may directly influence the expression of IκB kinase, or (2) cells that are protected by hypothermia may influence surrounding cells by releasing fewer inflammatory stimuli [167]. Inhibition of NF-κB activation could be key to the neuroprotective effects of hypothermia, but it should be taken into account that NF-κB also exerts anti-apoptotic effects. Indeed, there are conflicting reports on this subject as both increases and decreases in ischemic injury have been observed when inhibiting the activity of NF-κB [167‐169]. Ultimately, it will be the balance between type I and type II TNF-α receptor expression and the signaling pathways involved that will determine the outcome [170]. Chemokines are reduced by hypothermic treatment. Recently, a study showed downregulation of MCP-1 after application of mild hypothermia in vitro [92]. However, these results need to be confirmed in vivo. Hypothermia can suppress oxidative stress after an insult. Superoxide anion radical production is attenuated by a post-ischemic moderate hypothermic treatment [166]. These results also confirm earlier findings that hypothermia reduces the production of hydroxyl radicals after a transient MCAo in rats [96]. Further research has shown that post- but not intra-ischemic hypothermia reduces the release of superoxide anions, as most ROS are only produced in the reperfusion phase [171]. Concerning NOS, it has been shown that hypothermia decreases nNOS by 50% in the penumbra, while no effect is seen on iNOS [96]. Intra-ischemic mild hypothermia reduces expression of MMP-9 in rats subjected to 2 hours of MCAo [104]. There is no effect on MMP-2, neither after stroke nor after hypothermic treatment up to 24 hours after the insult [104]. Also, MMP-9 expression is low in patients after hypothermia, which could suggest a temperature sensitivity of MMPs [172]. In vitro studies show that hypothermia can reversibly inhibit E-selectin, without affecting L-selectin [173, 174]. Furthermore, 2 hours of mild post-ischemic hypothermia reduces ICAM-1 levels in brain and plasma for up to 7 days after 2 hours of MCAo [162, 166, 175]. In vitro, as well as in patients with cardiopulmonary bypass, hypothermia influences integrin expression. Hypothermia (27°C) delays upregulation of Mac-1 and CD11c, but has no effect on LFA-1 [174, 176]. Hypothermia can inhibit but possibly only delay glial activation [177, 178]. Hypothermia appears to decrease tissue density of microglia/macrophages and to inhibit microglial activation. This inhibition leads to a 54% reduction in microglial activation after 3 days if intra- or early post-ischemic hypothermic treatment in rats (subjected to a 2-hour MCAo) is maintained for 2 hours [153, 162, 168, 179]. However, as the phagocytic activity of activated microglia and macrophages exerts beneficial effects in the long term, such parameters should be studied at later time points after the induction of hypothermia. Hypothermia also reduces BBB disruption occurring after stroke, which especially has implications for the passage of large molecules. Permeability to small molecules remains for at least 4 days post-injury [180]. Infiltration of neutrophils and monocytes can be reduced after transient MCAo with post-ischemic mild hypothermic treatment at 3 and 7 days post-injury [162]. This reduction can increase to approximately 75% if the hypothermic treatment is started 1-2 hours after a MCAo of 2 hours [157]. Similar results have been observed in other traumatic injury models [181].

Mild-to-moderate intra-ischemic hypothermia has a pan-inhibiting effect [23, 182]. It reduces ATP depletion, anoxic depolarization, glutamate release, and apoptosis; maintains BBB integrity, inhibits white matter injury and blocks necrosis (if started during ischemia itself) [23, 160, 183]. The effects of hypothermia on inflammation show a more modulating response. However, hypothermia has been suggested to only delay neuronal damage rather than to provide permanent protection. Seven hours after a hypothermic treatment, a hyperthermic episode has been described which may revive or re-initiate pathophysiological mediators which had been quiet [159]. The challenge in hypothermic treatment is thus to know when to start cooling, how long to maintain it and how deep to cool. Deep cooling has no advantage over mild or moderate hypothermia, and actually causes more side effects [157]. Furthermore, a distinction has to be made between short and long cooling periods. It has been suggested that the disadvantages of delayed cooling could be overcome by performing a prolonged hypothermic protocol [22, 182]. However, the modulating influence of hypothermia on neuroinflammation could also differ depending on these parameters and should be further investigated [184]. Depth, start of cooling and duration all influence outcome in both experimental and clinical settings, and make translation to the clinic difficult [185].

The neuroinflammatory response after ischemic stroke involves several parameters, all connected with each other and connected to other pathways of the ischemic cascade. This makes it difficult to draw conclusions from research observations and to extrapolate them to clinical settings, because a limited number of differences between man and animal models may affect multiple levels of the inflammatory response. Many inflammatory components have neurotoxic as well as neuroprotective effects that may either stimulate or reduce cell damage after ischemic stroke. Inhibition of only one part of the neuroinflammatory response after ischemic stroke does not induce sufficient protection to improve recovery in patients. There is sufficient data that hypothermia acts at multiple levels of the ischemic cascade and of the neuroinflammatory response, and does not simply inhibit all processes induced by cerebral ischemia. It should be noted that hypothermia influences the expression of several inflammatory parameters at certain time points, but more research is required to discern which positive or negative effects contribute to neuroprotection. The ability of hypothermia to modulate many aspects of the inflammatory response may render translation to the clinic feasible. Another advantage of hypothermia could be the creation of a larger therapeutic time window to administer other neuroprotective agents and thus improve outcome after transient focal cerebral ischemia.