|Home | About | Journals | Submit | Contact Us | Français|
There is considerable experimental evidence that hypothermia is neuroprotective and can reduce the severity of brain damage after global or focal cerebral ischaemia. However, despite successful clinical trials for cardiac arrest and perinatal hypoxia-ischaemia and a number of trials demonstrating the safety of moderate and mild hypothermia in stroke, there are still no established guidelines for its use clinically. Based upon a review of the experimental studies we discuss the clinical implications for the use of hypothermia as an adjunctive therapy in global cerebral ischaemia and stroke and make some suggestions for its use in these situations.
Renewed interest in the use of hypothermia to reduce brain injury after ischaemia dates back to the late 1980s when experimental studies in dogs and rats showed that moderate levels of hypothermia (32-34°C) during or following ischaemia could be neuroprotective and improve neurological outcome [Leonov et al. 1990ab; Busto et al. 1987]. In addition, the use of small animal models of cerebral ischaemia, in which the effects of hypothermia can be easily assessed, has since become increasingly popular. Such studies have provided ample evidence that hypothermia can reduce neuronal loss after global or focal ischaemia, both in the mature and immature brain, even when applied after the ischaemic event. Indeed, a number of subsequent clinical trials evaluating hypothermia as a neuroprotective treatment following cardiac arrest and perinatal hypoxiaischaemia have been successful (see Table 1; Shah et al. ). On the other hand, fewer studies have investigated the effectiveness of hypothermia in ischaemic stroke and those that have been performed have involved relatively small numbers of patients and have mainly addressed safety issues (Table 2).
In spite of the substantial experimental evidence that hypothermia is neuroprotective, and the positive results of clinical studies in global cerebral ischaemia, there are still no established guidelines for its use in different clinical settings and there are still outstanding questions as to the most effective and safest levels of hypothermia, when hypothermia should be commenced and how long it should be continued, and the most appropriate methods for inducing hypothermia.
In this review we evaluate the experimental evidence that hypothermia is neuroprotective after ischaemia in the adult brain, and discuss the implications of the experimental findings for further clinical trials in global cerebral ischaemia and stroke.
We have summarized previous experimental studies of hypothermia following transient global cerebral ischaemia (global/forebrain ischaemia, cardiac arrest) in Table 3. These studies have largely been performed in small animal models such as rats and gerbils using bilateral carotid artery occlusion (alone or combined with systemic hypotension; 2VO or vertebral artery occlusion; 4 VO), or using cardiac arrest to induce cerebral ischaemia for periods ranging from 5 to 30 min. Most studies have assessed histological changes, mainly neuronal loss in the CA1 region of the hippocampus and many studies have also evaluated neurological outcomes. The majority of studies have used moderate levels of hypothermia (≥30 to ≤34.5°C) maintained for relatively short periods (5 min to ≤6 h) and commenced during the ischaemic period or within the first hour after ischaemia. These studies have shown that the degree of neuroprotection with such protocols is very variable, ranging from no protection to almost complete protection (Table 3). However, if the duration of hypothermia is extended beyond 6 h (e.g., to 12-48h) the level of neuroprotection is usually greater and more robust (Table 3). A therapeutic time-window study in rats [Coimbra and Wieloch, 1994] showed that hypothermia at 33°C maintained for 5 h was neuroprotective when commenced 2, 6, or 12 h after global cerebral ischaemia but not if it was delayed for 24 h. Similarly, in another study in rats hypothermia commenced 6 h after ischaemia and maintained at 34-35°C for 48 h was found to be highly neuroprotective [Colbourne et al. 1999a].
The neuroprotective effects of milder degrees of hypothermia (>34.5-36.5°C) in global ischaemia models have revealed that this level of hypothermia is effective during but not if induced after ischaemia and maintained for periods of <6 h (Table 3). However, it has been shown that longer durations of hypothermia in this temperature range can be effective even when commenced several hours after ischaemia. For example, our own studies in a rat model of global cerebral ischaemia have shown that while a 6 or 12 h period of hypothermia at 35°C is ineffective when commenced 2 h after ischaemia, 24 h of hypothermia significantly reduces neuronal loss and increases hippocampal CA1 neuronal survival from 5 to 43% at 7 days (Figure 1; Zhu et al. ).
While few studies have assessed the efficacy of deeper levels of hypothermia (23 to <30°C) they have generally reported positive outcomes, although in two studies using either a 2 h or a 20 min duration of hypothermia at 23 or 29-31°C commenced immediately after ischaemia, treatment was not effective [Hara et al. 1995; Welsh and Harris, 1991; Table 3].
The experimental hypothermia studies in animal models of global cerebral ischaemia paved the way for several clinical trials conducted from late 1990 to 2006 (Table 1). Two medium sized phase III trials using surface cooling showed a similar degree of improvement in outcome (55% vs 39%) and reduction in mortality (41% vs 55%) when moderate hypothermia (33°C) was commenced within 4 h of restoration of spontaneous circulation and maintained for 24 h, [Hypothermia After Cardiac Arrest Trial, 2002] or when commenced within 2 h of restoring spontaneous circulation and maintained for 12 h [Bernard et al. 2002]. Subsequent similar sized trials, [e.g., Holzer et al. 2006] using intra-vascular or surface cooling to induce hypothermia at 33°C commencing several hours after the restoration of spontaneous circulation and maintained for 24 h, also reported improved neurological outcomes. A recent trial has evaluated the safety and feasibility of inducing moderate hypothermia with rapid infusion of cold saline in out-of-hospital cardiac arrest patients [Kim et al. 2007]. While it is not possible to conclude from these trials if different cooling procedures are more or less efficacious, it appears that all methods are safe and have a low risk of adverse events. Other factors that may be associated with improved efficacy of hypothermia will be discussed later.
Animal experimental studies using hypothermia following transient or permanent focal cerebral ischaemia are summarized in Tables 4 and 5, respectively. These studies have been performed almost exclusively in rats and have mostly used mechanical methods to occlude the middle cerebral artery (e.g., intraluminal suture, clips, or ligatures). The outcomes in different models were very variable and were influenced by a variety of factors such as the age and strain of rat, the severity and duration of ischaemia, and the experimental endpoints chosen. The majority of studies have used the transient focal model with moderate levels of hypothermia (≥30 to ≤34.5°C) maintained for periods of 0.5 to ≤6 h, commencing during or within 2 h of ischaemia, and have had short-term endpoints (1-3 days). The results from these studies have been highly variable, some studies showing no protection, while others showed substantial levels of neuroprotection (Table 4). The variability in these studies may be related to the differences in study design as mentioned above, in addition to other factors that will be discussed below.
Of particular interest are the recent studies by Kollmar et al.  and Ohta et al. . The first study assessed different levels of hypothermia (32, 33, 34, 35, 36°C) induced immediately following a 90 min period of middle cerebral artery occlusion (MCAO) in rats and maintained for 4h. When examined 2 days after ischaemia, infarct volume was significantly reduced (by 22-53%) only with hypothermia between 32 and 34°C, while the other levels of hypothermia were associated with only a nonsignificant reduction in infarct volume (22%). The study by Ohta et al.  examined the effect of a 42-48 h period of moderate hypothermia at 34-35°C commencing 2, 4, 6, or 8h after the commencement of 120 min of MCAO. Infarct volume assessed 2 days after ischaemia was found to be significantly reduced (by 27-58%) when hypothermia was induced within the first 6 h after ischaemia, but not after 8 h. Taken together these studies indicate that for moderate/mild hypothermia (32-36°C) to be effective it must be induced within the first 6 h after the onset of ischaemia and maintained for periods longer than 4h.
Interestingly, a study by Inamasu et al.  showed that moderate/mild hypothermia at 33.5-35.5°C for 16h induced immediately following 60 min of transient MCAO significantly reduced infarct volume when measured 1-2 days postischaemia, but the protection was lost when examined 3, 5, or 7 days postischaemia. In contrast, studies from Colbourne's laboratory [Corbett et al. 2000; Colbourne et al. 2000] have shown that the neuroprotective effect of 48 h of moderate hypothermia in the 33-35°C range commencing 0.5-1 h after transient focal ischaemia is still preserved even 60 days after ischaemia. These findings indicate that medium durations of hypothermia in the range of 12-24 h may only delay the progression of postischaemic brain injury and that for the neuroprotection to be long-lasting periods of hypothermia ≥24h may be necessary.
Two laboratory studies using Spargue Dawley [SD; Huh et al. 2000] or spontaneously hypertensive rats [SHR; Kurasako et al. 2007], but similar levels and durations of marked hypothermia (27-28° C/2-3h) induced immediately after transient focal ischaemia produced a positive neuroprotective outcome or a trend towards neuroprotection. In addition, to the strain of rat used in the two studies, the shorter period of ischaemia in the positive SHR rat study (90 min) compared to the negative SD rat study (120 min) may have accounted for these results. In addition to the Kollmar et al.  study (35 or 36°C; 4h), two other studies assessing mild hypothermia (>34.5 to 36.5) produced conflicting findings. Nito et al.  maintained rats mildly hypothermic (35°C) only during a 120 min period of MCAO and like Kollmar et al.  did not observe any significant reduction in infarct volume after 1 day. In contrast, Aronowski et al.  induced mild hypothermia at 35° C for 4h, starting 1 h after the commencement of a 3 h period of transient MCAO and observed a 48% reduction in infarct volume after 1 day. The contrasting results in these studies further highlight how study parameters can influence experimental outcomes and that the timing and duration of hypothermia is critical.
Moderate and marked hypothermia has also been shown to be protective in permanent MCAO models (Table 5). Although, it is difficult to make direct comparisons, the level of protection appears to be comparable to that obtained in the transient MCAO models. For example, a number of studies have shown that marked hypothermia (24° C) for durations of 1 or 6 h commencing within 1 h of ischaemia can result in infarct volume reductions ranging from 27 to 84% (Table 5). However, with such short periods of marked hypothermia the neuroprotective effect is lost if the induction of hypothermia is delayed beyond 2 h after ischaemia [Baker, 1992].
Studies using moderate hypothermia in permanent MCAO models have produced mixed results, but have generally shown that durations of hypothermia between 2 and 24 h induced within 1 h of ischaemia can significantly reduce infarct volume by 20-77% when assessed within 2 days. Of note is the study by Ren et al.  who observed that a 2 h duration of hypothermia at 33°C commenced 30 min after MCAO reduced infarct volume at 1 day in Wistar rats, but not in SHR rats. It also appears that for short durations of hypothermia (1–4h) the neuroprotective effects are reduced or lost if extended endpoints are used [Morikawa et al. 1992; Ridenour et al. 1992]. In addition, we have recently shown that a 24 h duration of mild hypothermia when commenced 2 h after MCAO is ineffective [Campbell et al. 2008].
Finally, although less commonly assessed than in global ischaemia experiments, the neuroprotective effects of hypothermia observed in focal ischaemia experiments have usually been associated with reductions in neurological deficits [Kollmar et al. 2002; Maier et al. 2001; Huh et al. 2000].
Despite the lack of specific treatments to minimize progressive brain damage following acute ischaemic stroke, the assessment of hypothermia as a neuroprotective strategy has been relatively under explored. Treatments that are available aim to improve cerebral blood flow and consist of thrombolytic therapy (plasminogen activator; tPA) and decompressive hemicraniectomy (Dhamija and Donnan, 2007). However, thrombolysis is only useful within the first 3 h after a thromboembolic stroke and carries a significant risk of intracerebral haemorrhage after 3 h. The benefit of hemicraniectomy is limited to a small number of patients who develop rapid cerebral oedema and raised intracranial pressure 24-72 h post-stroke, but is highly invasive and requires a specialized stroke and neurosurgical unit. Hence, if the full potential of therapeutic hypothermia could be exploited, it could provide an effective widely applicable treatment when used alone or when combined with current treatments to reduce brain damage following stroke.
With respect to clinical studies using hypothermia in acute stroke there have been a number of small Phase I and Phase II trials, but as yet no Phase III trials (Table 2). These trials have been designed largely to assess the safety and feasibility of using mainly surface cooling techniques (e.g., cooling blankets/helmets, ice packs, cool air), and also endovascular cooling. Moderate hypothermia (32-33°C) has been trialled in 11 studies and mild hypothermia (35-35.5°C) in 4 studies with durations of between 6-72 h most commonly tested, usually commenced within the first 24 h after stroke. Of note were the longer durations (72-120 h) of moderate hypothermia (33°C) used by Naritomi et al. . However, as these trials were not designed to assess patient outcomes no conclusions can be drawn regarding the neuroprotective efficacy of hypothermia. However, there does appear to be general consensus that hypothermia, especially mild hypothermia appears, to be safe, well tolerated and relatively easy to administer for periods of up to 48 h. Despite this optimistic outlook many questions and issues regarding the use of hypothermia therapeutically in stroke require investigation and these will be discussed below.
Based on the results of both the experimental and clinical studies to date it is clear that hypothermia has the potential to reduce neuronal loss and improve functional outcomes when used as adjunctive therapy after cerebral ischaemia. Given its clinical effectiveness following cardiac arrest, its application in other forms of global cerebral ischaemia associated with post-traumatic brain swelling, vasospasm following subarachnoid haemorrhage, systemic shock and hypotension, heart and carotid artery surgery and near drowning warrants further evaluation. With respect to ischaemic stroke the safety and feasibility of using hypothermia have been evaluated, but determining its true clinical potential awaits a large randomized Phase III trial. Finally, despite the proven neuroprotective effects of hypothermia, many issues concerning its current and future use in stroke and global cerebral ischaemia require further evaluation experimentally and clinically.
Global ischaemia: While it is well established that the earlier hypothermia is commenced after ischaemia the more efficacious it is likely to be [Nozari et al. 2006; Zanten and Polderman, 2005], a defined therapeutic time window has not been established. This is largely due to the lack of studies that have simultaneously evaluated therapeutic time windows for hypothermia induction, and different depths and durations of hypothermia. An added complication is the period and severity of ischaemia, which is likely to reduce the therapeutic window and efficacy of hypothermia. For example, it is emerging that patients with prolonged interruption of cerebral circulation following cardiac arrest are poor candidates for hypothermic therapy [Green and Howes, 2007; Broccard, 2006]. With respect to the depth of hypothermia, it appears that moderate hypothermia (≈ 330C) and mild hypothermia (≈35°C) protocols are preferable for use in clinical practice because deeper levels of hypothermia (<32°C) are more likely to be associated with adverse effects (such as cardiac arrhythmias and impaired myocardial function, coagulopathy, immune suppression and infection) without necessarily offering substantial additional neuroprotection [Kollmar et al. 2007].
Given that moderate hypothermia (33-34°C) has already been shown to be beneficial clinically, two important questions arise: (1) What is the optimal duration for which hypothermia should be maintained? and (2) Is mild hypothermia (≈34.5-35.5°C) more, less or equally effective than moderate hypothermia? With regard to the second question, we believe that mild hypothermia has relative advantages over moderate hypothermia, with respect to safety, reduced side effects and importantly ease of induction (see below). The potential downside is reduced efficacy, but this dosen't appear to be the case [Logue et al. 2007], and in fact, it is possible that long durations of moderate (and marked) hypothermia may be more inhibitory to endogenous neurosurvival and neuroregenerative pathways in the brain and may also adversely affect myocardial function [Leonov et al. 1990a].
While the cardiac arrest trials have shown that durations of moderate hypothermia of between 12 and 24 h are effective when commenced within 4 h of restoration of spontaneous circulation, we believe the 24 h duration should become standard if commenced within 4 h of the ischaemic event. To this end, several experimental studies [Logue et al. 2007; Zhu et al. 2005; Colbourne et al. 1999ab; Nurse and Corbett, 1996; Colbourne and Corbett, 1994, 1995] have confirmed the effectiveness of long durations of mild-moderate hypothermia when commenced several hours after global cerebral ischaemia. For example, we have shown that with mild hypothermia (35°C) a 24 h duration is required to obtain significant neuroprotection when commenced 2 h after global ischaemia in the rat [Zhu et al. 2005; Figure 1]. Therefore, we propose that moderate/mild hypothermia should be maintained for at least 24 h when commenced within the first 4 h after a global cerebral ischaemic event and that longer durations (e.g., 36-48 h) should be considered if hypothermia is commenced later (4-12 h; Colbourne et al. 1999ab). Further experimental studies and clinical trials are, however, needed to define more precisely the therapeutic time windows and optimal duration of moderate/mild hypothermia when it is commenced later than 4 h.
Ischaemic stroke: Many of the issues concerning timing/therapeutic window, depth and duration of hypothermia following global ischaemic also apply to focal ischaemia. However, the therapeutic time window for moderate/mild hypothermia in ischaemic stroke is likely to be more variable and to be influenced by factors such as the size of the occluded artery and infarct, whether blood flow is restored by tPA-induced or spontaneous thrombolysis, the degree of swelling and raised intracranial pressure and the presence of co-morbidities such as old age, diabetes, vascular disease and occurrence of early hyperthermia. For these reasons it has been suggested [Fisher, 2006] that patient selection should be made on the grounds of the presence of salvageable brain tissue in the ischaemic penumbra as demonstrated by diffusion-perfusion mismatch on MRI imaging studies. However, as such studies may not always be readily available and in view of the minimal side effects associated with inducing moderate/mild hypothermia it would seem reasonable to consider the use of hypothermia routinely, particularly in patients with large artery occlusions who present within the first 6 h of stroke onset. While this is the optimal therapeutic time window demonstrated in experimental animal studies, the possibility that the time window may be longer in patients cannot be excluded and needs to be investigated. Moreover, as hypothermia is not contraindicated in haemorrhagic stroke, the requirement to identify stroke subtype before hypothermia induction would not be essential.
Although the question of the optimal duration of hypothermia remains uncertain [Van der Worp et al. 2007] and must await further experimental and clinical trials, for the present it would seem reasonable to follow the same recommendations as proposed above for global cerebral ischaemia. It should be mentioned however that studies by Naritomi et al. [1996, 2002; Table 2] have used durations of moderate hypothermia (33°C) for between 3 and 5 days in ischaemic stroke patients without serious complications and with good outcomes.
For individuals who have experienced an out-of-hospital cerebral ischaemic episode such as cardiac arrest, stroke or closed head injury, initial steps by bystanders, friends or family members should avoid active measures to keep the patient warm. Once attended by trained medical personnel hypothermic induction can be commenced en route to hospital, using a variety of techniques including cooled intravenous saline infusion, ice packs, alcohol sponging and portable cooling helmet/neck devices. In addition, to minimize discomfort and shivering in awake patients, and to assist hypothermia induction, pethidine (meperidine) alone or in combination with the anxiolytic buspirone can be administered. These agents have been trialled in stroke patients and have been found to reduce the shivering threshold by 2-4° C [Lyden et al. 2005; Mokhtarani et al. 2001]. On hospital arrival, following further patient assessment controlled hypothermia can then be continued with more precise temperature control.
In the clinical setting, in addition to the procedures described above hypothermia can also be induced by other external or internal cooling devices such as mattresses, blankets, helmets, body pads, neck braces, cold air blowers and endovascular and nasal catheters. [Wolfson et al. 2008; Jordan and Carhuapoma, 2007]. Each system has its advantages and disadvantages [see review by Jordan and Carhuapoma, 2007]. However for routine application, initial induction can be carried out by intravenous infusion of 1-2 L of cold saline in combination with alcohol sponging and pethidine and/or buspirone administration. These cooling procedures are relatively straightforward, safe and enable the initial rapid induction of hypothermia. In order to maintain hypothermia, a surface-cooling procedure consisting of head, neck and body pads that allow good heat exchange of circulating temperature controlled solutions, and which also allows for controlled re-warming is required. Importantly, mild hypothermia can be achieved in awake patients using these surface cooling techniques [Wang et al. 2004; Zweifler et al. 2003; Kammersgaard et al. 2000]. The cooling system should preferably incorporate a computer-controlled feedback system to allow precise control of cooling and re-warming. With respect to re-warming, although an optimal rate has not been determined, slow warming in the order of 1°C/4-8h and 1°C/8-12h for hypothermia durations of 24-48 h and >48h, respectively, appears to be beneficial and can avoid re-bound hyperthermia [Bardutzky and Schwab, 2007; Bernard and Buist, 2003].
It is possible that the combined use of hypothermia with tPA within the optimal 3 h therapeutic time window for thrombolysis could provide added benefit by reducing the severity of brain injury associated with ischaemia and reperfusion. In addition, it has been suggested [Feigin et al. 2002] that hypothermia could increase the therapeutic time window during which tPA can be safely administered (e.g., to 3-9h). There have however been some concerns that hypothermia might reduce the efficacy and safety of tPA, related to reduced thrombolysis, and impaired platelet function and coagulation at lower temperatures [Hemmen and Lyden, 2007]. One strategy to minimize these potential drawbacks would be to use only mild hypothermia (35°C), at least during the initial stages of tPA treatment, which is less likely to have a negative impact on thrombolysis, platelet function and coagulation.
Hypothermia could also be a beneficial intervention in patients with large supratentorial infarcts requiring decompressive hemicraniectomy as it appears that hypothermia reduces brain oedema [Bardutzky and Schwab, 2007; Schwab et al. 1998]. In addition, an experimental study of MCAO in rats has shown that combined craniectomy and hypothermia (32°C/5h) leads to additional reduction in infarct volume than craniectomy alone [Doerfler et al. 2001]. As patients are anaesthetized during hemicraniectomy and are in a surgical environment, hypothermia using an endovascular cooling technique would be relatively easy to implement in this situation.
There is also potential for combining hypothermia with other neuroprotective strategies targeting different pathways of postischaemic neuronal injury and death. It is possible that hypothermia may act synergistically with other neuroprotective agents such as glutamate antagonists, antioxidants, neurotrophic factors and magnesium when used in combination after ischaemia (see review by Campbell et al. 2007).
Although the experimental evidence that hypothermia is neuroprotective after cerebral ischaemia is compelling, it is difficult to extrapolate directly from the results of experimental animal studies to clinical situations and it cannot be assumed that the optimal levels and duration of hypothermia are the same in humans who have experienced a global or focal ischaemic insult as in small animal models. Moreover, because of the variability in the results of the experimental studies there is still some uncertainty as to the long-term effectiveness of hypothermia. There is a need for further carefully planned experimental studies to define more precisely the therapeutic time windows and optimal durations of hypothermia using longer end-points and functional methods of assessment to determine whether the neuroprotective effects which have been demonstrated translate into worthwhile functional gains.
However, based upon the available experimental data and the results of clinical studies performed to date it does appear that the use of mild or moderate hypothermia protocols is feasible and safe to implement in clinical situations associated with global cerebral ischaemia and in patients with ischaemic stroke. Based upon the results of the experimental studies it can be predicted that if hypothermia is used in these situations it should be commenced as soon as possible after the ischaemic event and that a duration of at least 24h, and possibly 48h, would be necessary to achieve a sustained benefit in terms of neuronal recovery and survival and functional benefits. On the basis of the experimental studies in focal ischaemia models and the results of the early-phase clinical studies in stroke there would seem to be sufficient justification for a Phase III randomized controlled trial to further evaluate the effectiveness of hypothermia alone or in combination with thrombolysis and other potential neuroprotective agents such as magnesium, glutamate antagonists and antioxidants.
The authors would like to thank Tegan Phillips for assistance in compiling the reference list.
Bruno P. Meloni, Australian Neuromuscular Research Institute A Block, 1st Floor QEII Medical Centre Nedlands, Western Australia, Australia 6009 ; Email: ua.ude.awu.enellyc@inolem.
Frank L. Mastaglia, Centre for Neuromuscular and Neurological Disorders, Australian Neuromuscular Research Institute, University of Western Australia Australia.
Neville W. Knuckey, Centre for Neuromuscular and Neurological Disorders, Australian Neuromuscular Research Institute, University of Western Australia, and Department of Neurosurgery, Sir Charles Gairdner Hospital Nedlands, Western Australia, Australia.