A Review of Clinical Trials of Hypothermia Treatment for Severe Traumatic Brain Injury

Abstract

Clinical trials of hypothermia treatment of traumatic brain injury can be divided into (1) trials designed to abort the biochemical cascade after injury—neuroprotection, (2) trials primarily designed to test the effect of hypothermia in reducing elevated intracranial pressure (ICP), and (3) trials with features of both neuroprotection and elevated ICP control. Three of the four clinical trials testing hypothermia induction after failure of conventional means of ICP control showed decreased mortality rate, though sample sizes were small and findings were not always statistically significant. Nine randomized trials have tested hypothermia as a neuroprotectant, inducing it from 2.5 to 15 hours after injury and continuing it for a predetermined period of time regardless of ICP. Eight of these nine trials have been negative with three finding an effect in patients with evacuated hematomas, two of these if hypothermia is rendered before or soon after craniotomy. Despite extensive clinical testing over a range of treatment windows after injury, there is no evidence for the use of hypothermia as a neuroprotectant in patients with diffuse brain injury. Four randomized trials have features of neuroprotection and ICP control, randomizing and initiating hypothermia within 15 hours of injury and continuing hypothermia for the duration of ICP elevation. All found improved outcome and reduced ICP. Based on these findings and the negative results of neuroprotection trials that extended hypothermia for a defined period of time, it is likely that the mechanism of protection in these combined mechanism trials was early control of ICP. This literature suggests the need for clinical trials with two distinct objectives—(1) testing hypothermia for ICP control when conventional means (sedation and paralysis, mannitol, hyperventilation, and cerebrospinal fluid drainage) fail and (2) testing early induction of hypothermia before hematoma evacuation individualizing the duration of hypothermia to the patient's ICP responses.

Introduction

Clinical investigations of hypothermia in treatment of severe brain injury began as early as 1958 with 121 patients reported in the medical literature by 1989. Patients were treated at temperatures of 28°C–34°C for durations of 2–10 days (Lazorthes and Campan, 1958; Sedzimir, 1959; Drake and Jory, 1962; Shapiro et al., 1974; Strachan et al., 1989). Given the methodological limitation of these early studies, about all that can be said about them is that cardiac arrhythmias, a major risk of hypothermia, were infrequent. Investigations began anew when investigators found that only a few degrees of hypothermia were protective if rendered after experimental global ischemia and fluid percussion brain injury (Busto et al., 1987; Clifton et al., 1991). In 1993, two phase II clinical trials inducing hypothermia by surface cooling, reaching 32°C at 8 and 10 hours after injury and maintaining it for 48 and 24 hours, respectively, showed no clear hypothermia-related toxicity with nonsignificant trends toward improved outcome. A period of widespread clinical testing of hypothermia treatment for severe brain injury ensued.

The thesis of this article is that controlled, randomized studies of hypothermia in severe traumatic brain injury can be divided into (1) trials primarily designed to test the effect of hypothermia in reducing elevated intracranial pressure (ICP), (2) trials designed to abort the biochemical cascade after injury, and (3) trials with features of both neuroprotection and elevated ICP control.

In this article, the classification of a trial as being targeted to neuroprotection is based on relatively early induction of hypothermia (induction reaching 33°C within 15 hours of injury) and maintenance of hypothermia for a predetermined period of time, irrespective of ICP. In this article, a trial is considered to be designed to test hypothermia as a means of managing elevated ICP if cooling was initiated when other means of ICP control had failed. Trials classified in this article as having features of both are those that induced hypothermia within 15 hours of injury and continued it until ICP normalized.

The clinical trials from which this article was constructed were randomized trials of hypothermia treatment (1587 patients) that reported death or severe disability in a recent Cochrane Database System Review article (Sydenham et al., 2009). Five trials included in the Cochrane Review are not included in this analysis. One randomized clinical trial not detected by that review (Jiang et al., 2006), another not published at the time of the review (Clifton et al., 2011), and two nonrandomized trials of good quality are included in this article (Polderman et al., 2002; Shiozaki et al., 2003). In the current article, each of the studies are briefly described chronologically in three groups enumerating the age range of study subjects, findings on outcome, time to target temperature if known, rewarming time if known, effect of hypothermia on ICP if known, complications, and subgroup analyses.

Among trials included in the Sydenham report and not included in this article is a small trial that induced cooling with a cooling cap and reported that effective cerebral hypothermia was not delivered (Harris et al., 2009). Other reports were published in journals that could not be accessed (Zhang and Wang, 2000) or contained insufficient information to classify the study (Hirayama et al., 1994; Yan et al., 2010). One small study of the endocrine response to injury only examined mortality rate and not functional outcome (Meissner et al., 2003).

A total of 15 randomized and two nonrandomized trials are included in this article. Unless specifically stated, the entry criteria are patients in coma from traumatic brain injury without serious associated injuries. Patients were randomized to treatment at hypothermia versus normothermia. Outcomes, unless otherwise stated, are classified as the percent of poor outcomes on the dichotomized Glasgow Outcome Scale (good outcome=good recovery and moderate disability; poor outcome=severe disability, vegetative state, and death). All the studies reviewed here used surface cooling methods. The reliability of clinical trials in which allocation is not concealed is doubtful (Higgins and Green, 2008). We note whether the Cochrane reviewers or this reviewer judged allocation concealment to be clear or unclear.

Studies Designed to Deliver Hypothermia as a Neuroprotectant

1. In a single-center study of 82 patients of ages 16–75 years reported in 1997 by Marion et al., patients assigned to hypothermia were cooled to 33°C a mean of 10 hours after injury, kept at 32 to 33°C for 24 hours, and rewarmed at a rate no >1°C per hour. At 12 months, 38 percent of the patients in the hypothermia group and 62 percent of those in the normothermia group had poor outcomes (p=0.05). Mean ICP was lower in the hypothermia group during the period of hypothermia. The reported effect was exclusively in patients with admitting GCS 5–7; however, patients with GCS 5–7 treated at normothermia had unusually poor outcomes (80% as compared with a more usual level of about 50%) (Clifton et al., 2001). The trial reported allocation concealment (Table 1).

2. In 2001 Clifton et al. reported a multicenter study of 392 patients, ages 16–65 years with target temperature of 33°C reached by 8.4±3.0 hours after injury and maintained for 48 hours. Rewarming was 0.5°C every 2 hours. Outcome at 6 months was poor in 57% of patients in both groups. Ten percent of the patients in the hypothermia group and 3 percent of those in the normothermia group had severe hypotension (p=0.01). The incidence of severely elevated ICP (>30 mmHg) was lower in the hypothermia group during the period of hypothermia. In subgroup analyses, adverse outcome was associated with hypothermia induction in patients older than 45 years of age, and better outcome was associated with maintenance of hypothermia in patients who were already hypothermic (≤35°C) on admission. Allocation was concealed.

3. In a multicenter trial reported in 2001, Shiozaki et al. compared the effect of 48 hours of hypothermia with treatment at normothermia in 91 patients, ages 9 to 70 years, who did not have elevated ICP. Patients in the hypothermia group reached 34°C at 11.1±6.4 hours after admission (not injury) and were rewarmed at 1°C per day over 3 days. The incidences of pneumonia, meningitis, leukocytopenia, thrombocytopenia, hypernatremia, hypokalemia, and hyperamylasemia were significantly higher in the hypothermia than in the normothermia group (p<0.05). There was no difference in outcome, with 53% of patients in the hypothermia group and 51% of patients in the normothermia group having poor outcomes. Allocation concealment is unclear.

4. In 2002, Biswas et al. reported a clinical trial of 21 children <18 years of age in which cooling was initiated in the treatment group within 6 hours of injury and maintained for 48 hours. Patients were rerewarmed over 24 hours. The investigators found no difference in mean ICP between groups but found that hypothermia decreased the severity of intracranial hypertension. There were no differences in complications between groups and no difference in functional outcomes at 3 and 12 months. It is unclear as to whether allocation was concealed.

5. In 2005, Adelson et al. reported a multicenter phase II study of 48 children <13 years of age stratified by age with cooling to 32°C to 33°C for 48 hours. Patients were rewarmed at 1°C every 3–4 hours. The average time to reach the target temperature of 32°C to 33°C from the time of injury was 13.78±7.23 hours. Patients with hypothermia had lower mean ICP during the first 24 hours. The mortality rate of patients with hypothermia was 8% and of patients in the normothermia group was 16% (p=0.44). Patients with hypothermia had an increased incidence of arrythmias. Allocation was concealed.

6. In 2005, Smrcka et al. reported a study of 72 adult patients <60 years of age in whom hypothermia to 34°C was administered to the treatment group within 15 hours of injury and maintained for 72 hours. Patients were passively rewarmed. ICP was significantly lower in the hypothermia-treated group. Patients treated with hypothermia with extracerebral hematomas but not diffuse brain injury had a significantly better mean Glasgow Outcome Score at 6 months than patients treated at normothermia. It is unclear as to whether allocation was concealed.

7. In 2008, Hutchison et al. reported a multicenter study of 225 children aged 1 to 17 years in the hypothermia group cooled to 32.5°C for 24 hours. Rewarming was at 0.5°C every 2 hours. Patients with hypothermia reached target temperature at 3.9±2.6 hours after admission. Unfavorable outcomes at 6 months were nonsignificantly increased in patients with hypothermia (hypothermia—31%, normothermia—22% (p=0.14). Mortality rate was increased in the hypothermia group (hypothermia—21%, normothermia—14% (p=0.06). ICPs in the hypothermia group were lower during the cooling period but higher during the rewarming period. The incidence of hypotension was increased in the hypothermia group. Allocation was concealed.

8. In a 2011 follow-up to their initial study, Clifton et al. induced hypothermia in transit to or on admission to the emergency department in a randomized study of 97 patients aged 16 to 45 years. Patients with hypothermia reached 35°C at 2.6±1.2 hours after injury and 33°C at 4.4±1.5 hours after injury. Hypothermia to 33°C was maintained for 48 hours and patients rewarmed at 0.5°C every 2 hours (about 18 hours). A protocol of aggressive fluid expansion during rewarming and low-dose morphine was used to prevent the hypotension that had complicated use of hypothermia in the group's first study. Overall, there was no improvement in outcome at 6 months, but hypothermia had opposite effects on outcomes of patients with diffuse brain injury and on those with evacuated hematomas (p=0.001). Fewer patients treated with hypothermia had poor outcomes (hypothermia—33%, normothermia—69%, p=0.02), whereas more patients with diffuse brain injury treated with hypothermia had poor outcomes (hypothermia—70%, normothermia—50%, p=0.09). Patients treated with hypothermia had a higher number of total episodes of elevated ICP, especially during rewarming. Patient allocation was concealed.

9. Clifton et al. in a post hoc analysis (G.L. Clifton, unpublished data) examined the relationship of outcomes to time of hypothermia induction relative to the time of surgery in the group's two clinical trials—The National Acute Brain Injury Study: Hypothermia I (NABIS:H I) and The National Acute Brain Injury Study: Hypothermia II (NABIS:H II). In NABIS:H II, all subjects with evacuated hematomas randomized to hypothermia were below 35°C before start of surgery or within 1.5 hours of surgery and reached 33°C within 5.55 hours of surgery. Outcome was poor (severe disability, vegetative state, or death) in 5 of 15 patients in the hypothermia group and 9 of 13 patients in the normthermia group (p=0.02, RR 0.44, 95% confidence interval [CI] 0.22, 0.88). In NABIS:H I, 31 of the 54 patients treated for hypothermia reached a temperature of 35°C or less within 1.5 hours after surgery start time, and the remaining 21 patients reached 35°C later. Outcome was poor in 14 of 31 patients (45%) reaching 35°C within 1.5 hours of surgery, 14 of 23 patients (61%) reaching 35°C>1.5 hours of surgery, and 35 of 58 patients (60%) in the normothermia group (p=0.16, RR 0.74, 95%, CI 0.49, 1.13). A meta-analysis of 46 patients with hematomas who reached 35°C within 1.5 hours of surgery start time in NABIS:H I and II showed a significantly reduced rate of poor outcomes compared with 94 patients treated with hypothermia who failed to reach 35°C within that time and patients treated at normothermia (p=0.009).

Table 1.

Randomized Clinical Trials Designed to Deliver Hypothermia as a Neuroprotectant

Principal investigator	Study publication date	Single or multicenter	n	Hours from injury to 32°C–34°C	Duration of hypothermia (hours)	Hypothermia group outcome	Hypothermia subgroups showing positive effects
Marion	1997	Single	82	10.0	24	Improved outcome	GCS 5–7 on admission
Clifton	2001	Multi	392	8.4	48	No outcome improvement	Hypothermic when admitted
Shiozaki	2001	Single	91	11.1	48	No outcome differences	Unknown
Biswas	2002	Single	21	6.0	48	No improvement in functional outcomes	Unknown
Adelson	2005	Multi	48	13.8	48	No outcome differences	Unknown
Smrcka	2005	Single	72	15.0	72	No outcome differences	Evacuated hematomas
Hutchison	2008	Multi	225	3.9	24	No improvement	Unknown
Clifton	2011	Multi	97	4.4	48	No improvement	Evacuated hematomas

GCS, Glasgow Coma Score.

Randomized Studies Designed to Deliver Hypothermia for ICP Control

1. In 1993, Shiozaki et al. reported a randomized study of the use of hypothermia to 34°C in 33 patients whose ICP remained higher than 20 mmHg at 5 to 6 hours after induction of high-dose barbiturate therapy. In the group randomized to hypothermia, cooling was continued for 2 days or until it was considered ineffective. Patients were initially rewarmed to 35.5°C to 36.5°C for 24 hours and then recooled if ICP increased above 20 mmHg. ICP was significantly lower in the hypothermia treatment group than in the group with continued management at normothermia. Fourteen of 17 patients in the normothermia group died versus 8 of 16 patients in the hypothermia group (p<0.05). Ten of 16 patients with hypothermia had poor outcomes versus 16 of 17 patients with normothermia (Table 2).

2. In 2003, Hashiguchi et al. reported a study in which 17 patients, ages 20–63 years, whose ICP was controlled to 20 mmHg or less by conventional means (mannitol, drainage of cerebrospinal fluid, hyperventilation with sedation and paralysis, and barbiturates), were randomized to further management at normothermia or mild hypothermia (34°C) for 5 days. Six of nine patients (67%) treated with hypothermia and 7 of 8 patients (88%) treated at normothermia had good outcomes. The pneumonia rate was significantly increased in patients treated for hypothermia.

Table 2.

Randomized Clinical Studies Designed to Deliver Hypothermia for Intracranial Pressure Control Only

Principal investigator	Date published	Single or multicenter	n	Initiation of cooling	Length of time cooled	Hypothermia group outcome results
Shiozaki	1993	Single	33	After administration of barbiturates	Duration of intracranial pressure elevation	Positive
Hashaguchi	2003	Single	17	After administration of barbiturates	5 days	Negative

Nonrandomized Studies of Hypothermia for ICP Control

1. In 2003, Shiozaki et al. studied 22 patients, ages 16–72 years, to determine whether hypothermia treatment lowered to 31°C controlled severely elevated ICP that was intractable to conventional therapy at mild hypothermia (34°C). Reducing body temperature to 31°C controlled severely elevated ICP in 3 of the 22 patients; however, all 22 patients died (Table 3).

2. Polderman et al. (2002) treated 136 consecutive patients who had elevated ICP after conventional management including barbiturates. If ICP responded to barbiturates, then patients were continued on them and served as the control group. If ICP was not reduced by barbiturates, then patients were cooled to 32°C–34°C until ICP was reduced to 20 mmHg or less for an average of 4.8 days. The rate of rewarming was not reported, but if ICP increased after 24 hours of hypothermia during rewarming, then hypothermia was induced again. Cooling reduced ICP; there was no difference in the rate of complications between the two groups. The rate of poor outcomes in patients with elevated ICP responding to barbiturates was 90% as compared with an 84% poor outcome rate in patients who did not respond to barbiturates and were then treated with hypothermia. Mortality rate was 62% in the hypothermia group and 72% in those who responded to barbiturates without hypothermia (p>0.05).

Table 3.

Nonrandomized Studies of Hypothermia for Intracranial Pressure Control

Principal investigator	Date published	Single or multicenter	n	Cooling protocol	Hours after injury to 32°C–34°C	Length of time cooled	Hypothermia group outcome results
Shiozoki	2003	Single	22	Cooled to 34°C until ICP 40 mmHg and then lowered temp to 31°C	Unknown	Until ICP controlled or death	Negative
Polderman	2002	Single	136	Patients with ICP intractable to barbiturate treatment cooled to 32°C–34°C	5.2 hours after admission	4.8 days	Positive

ICP, intracranial pressure.

Randomized Studies with Features of Neuroprotection and ICP Control

1. In 2000, Aibiki et al. reported a single-center study of 26 patients aged 9 to 76 years. Hypothermia to 32°C–33°C was induced within 3–4 hours of injury. Hypothermia continued until there were no signs of brain swelling on CT scan and ICP was normal, usually 3–5 days. Rewarming rate and ICPs were not reported. Complication rates were not different between groups. Three of 15 patients in the hypothermia group had poor outcomes and 5 of 11 in the normothermia group. Whether allocation was concealed is unclear (Table 4).

2. Jiang et al. in 2000 randomized 87 adult patients to treatment at normothermia or hypothermia to 33°C reaching target temperature at a mean of 15 hours after injury. Hypothermia was maintained for 3 to 14 days until the patient's ICP was <15 mmHg. Patients were rewarmed at a rate no >1°C per hour. At 1 year after injury, 53.5% of patients treated with hypothermia and 73% of patients treated at normothermia had poor outcomes (p<0.05). Mean ICPs were significantly lower in patients treated with hypothermia. Allocation concealment was unclear.

3. In 2006, Jiang et al. reported the results of a randomized, multicenter trial of 215 patients, ages 18 to 45 years. Patients were randomized within 4 hours of injury to hypothermia (35°C–33°C) for either 5 or 2 days and then rewarmed at 1°C per hour. ICP was significantly lower in the long-term hypothermia group than the short-term hypothermia group as was the percentage of poor outcomes (long-term hypothermia—56.5% poor outcomes, short-term hypothermia—71% poor outcomes, p>0.05). There were no differences in rates of complications between the two treatment groups. Allocation concealment was unclear.

4. In 2007, Qui et al. reported 80 patients aged 19 to 65 years randomized in a single-center study. Entry criteria were GCS<8 and craniotomy for intracranial hematomas after nonpenetrating trauma. Hypothermia with systemic temperatures of 34.5°C to 36°C was maintained for 4 days with rewarming over 12 to 13 hours. The time after injury of randomization in which target temperature was reached was not reported, but the authors report that target temperatures were reached within 2.5 hours of craniotomy, which was performed at a mean time of 4.6 hours after injury. ICP during the period of hypothermia was significantly lower in the hypothermia group. Poor outcome rate at 1 year in the hypothermia group was 30% versus 53% in the normothermia group (p=0.041); however, data were not shown. Complication rates were not different in the two groups. Allocation was concealed.

Table 4.

Randomized Studies Designed for Both Neuroprotection and Intracranial Pressure Control

Principal investigator	Date published	Design	n	Hours after injury to 32°C–35°C	Length of time cooled	Hypothermia group outcome results
Abiki	2000	Single center	26	3–4 hours	3–5 days	Positive
Jiang	2000	Single	87	15 hours	3–14 days depending upon ICP	Positive
Jiang	2006	Multicenter	215	4 hours	5 or 2 days	Positive
Qui	2007	Single center	80	Initiated directly after craniotomy	4 days	Positive

Discussion

There is little question that hypothermia lowers ICP during the period of hypothermia as compared with normothermia; the question is under what conditions that effect improves outcome. Nine of 13 randomized trials, whether designed to deliver hypothermia as a neuroprotectant or as a means of controlling elevated ICP or both, report a significant reduction in ICP during the period of hypothermia, but not after hypothermia was discontinued. (Shiozaki et al., 1993; Marion et al., 1997; Jiang et al., 2000, 2006; Clifton et al., 2001; Adelson et al., 2005; Qui et al., 2007; Hutchison et al., 2008; Clifton et al., 2011).

There is reasonable evidence that the use of hypothermia to control elevated ICP results in improved outcomes, but the effect is likely the greatest if hypothermia is administered before ICP becomes refractory to conventional management including barbiturates. Two randomized and two nonrandomized studies here classified as designed to improve outcome by reducing ICP employed hypothermia only when ICP elevation was refractory to conventional means including barbiturates. Three of the four studies reported a trend or statistically significant improvement in mortality rate, though because of the selection criteria, mortality rates were high in both treatment groups. The validity of this data is weak, because the two randomized studies were single-center studies of small size, and the other two trials were not randomized using a cross-over design.

Four randomized clinical trials have features of both neuroprotection and ICP control, though none of them reported time to target temperature. Aibiki instituted hypothermia within 3–4 hours of injury and continued it until ICP levels were normal. The sample size was too small for the effect to be statistically significant, but there was a trend toward better outcomes in patients treatedwith hypothermia (Aibiki et al., 2000). In his group's first study in 2000, Jiang tested hypothermia induced within 15 hours after injury and maintained throughout the duration of ICP elevation with a statistically significant improvement in outcome with hypothermia (Jiang et al., 2000). In 2006, Jiang et al. sought to answer the question of the appropriate duration of hypothermia, inducing hypothermia at 3 hours after injury and maintaining it for 3 versus 5 days after 33°C was reached. This investigator found that 5 days of hypothermia resulted in statistically better outcomes and lower ICP than 3 days of hypothermia. Qui induced hypothermia within the first hours after injury, maintained it for 4 days, and reported significantly improved outcome and reduced ICPs (Qiu et al., 2007).

Trials here classified as designed for neuroprotection with early administration of hypothermia for a defined period of time irrespective of ICP have almost uniformly been negative with the exception of patients undergoing craniotomy. The time of the treatment window has been thoroughly tested. The mean times to 33°C in studies specifically designed for neuroprotection are as follows: Marion, 1997—10 hours, Clifton, 2001—8.4 hours, Shiozaki et al., 2001—11.6 hours, Biswas—about 10 hours, Adelson—13.78 hours, and Hutchison—8 hours. To explore the question of whether these trials were negative because the treatment window was missed, Clifton et al., 2011 instituted a randomized multicenter trial in which cooling to 35°C was achieved at 2.6±1.2 (standard deviation [SD]) hours after injury and 33°C at 4.4±1.5 (SD) hours of injury. The overall result of that trial was also negative. Further, a nonsignificant trend toward a worse outcome was found with hypothermia treatment correlated with elevated ICP upon rewarming (Clifton et al., 2011).

On balance, there is reasonable, though not conclusive, evidence that early induction of hypothermia with continuation throughout the period of ICP control improves outcome. Considering the lack of evidence for improved outcome from trials that induce hypothermia early and continue it for a predetermined period of time, here classified as neuroprotection trials, the most likely mechanism of improvement in studies that induce hypothermia early and continue it until ICP is low, is early treatment of elevated ICP rather than neuroprotection. Elevated ICP affects as many as one-third of patients with diffuse injury and half of patients with evacuated hematomas (Miller et al., 1977). Although not all patients in a large cohort are sensitive to hypothermia used in this manner, a high percentage is.

Although the overall results of trials of early induction of hypothermia for a 24- to 72-hour period regardless of ICP have been negative, three studies here classified as neuroprotection trials have each found that patients with evacuated hematomas were sensitive to treatment. Smrcka et al. (2005) found that patients treated with hypothermia with extracerebral hematomas but not diffuse brain injury had a significantly better mean Glasgow Outcome Score at 6 months than patients treated at normothermia. Similarly, Clifton, in an analysis that was specified before the study blind was broken, found a statistically significant improvement in outcome of patients with evacuated hematomas but not those with diffuse brain injury (Clifton et al., 2011). In this study, patients reached 35°C within 1.5 hours of craniotomy and 33°C within 5.5 hours of craniotomy. In re-evaluating the earlier 2001 study of 392 patients randomized to normothermia or hypothermia to 33°C within 8 hours of injury, those patients with evacuated hematomas who reached 35°C within 1.5 hours of injury and 33°C within 5.5 hours of injury showed a trend toward improved outcomes as compared with patients with evacuated hematomas cooled later and those treated at normothermia (Clifton G.L., unpublished data).

This literature indicates the need for two types of clinical trials—one testing ICP control and the other, neuroprotection. For ICP control, there are two options—(1) induction of moderate hypothermia in patients who fail conservative management but before administration of barbiturates or (2) when barbiturate coma fails to control ICP. Based on the results of four studies that induced hypothermia early and continued it throughout the period of ICP elevation (Aibiki et al., 2000; Jiang et al., 2000, 2006; Qiu et al. 2007) and the results of studies that tested hypothermia when all other measures failed (Polderman et al., 2002; Hashaguchi et al., 2003; Shiozaki et al., 2003), the former design holds the greater possibility for improvement in outcome.

For neuroprotection, hypothermia should be tested in patients who require hematoma evacuation, inducing hypothermia to 35°C before craniotomy and continuing hypothermia to 33°C until ICP is normal. It can be concluded from the rebound increase in ICP during and after rewarming in the study of Clifton et al., 2011) and from three positive randomized studies that induced hypothermia early and continued it throughout the period of ICP elevation (Aibiki et al., 2000; Jiang et al., 2000, 2006; Qiu et al., 2007) that individualizing the duration of hypothermia to fit a patient's ICP in future trials is a better strategy than a predetermined period of hypothermia regardless of ICP. Design of future trials of hypothermia in severe brain injury should consider both the mechanism being tested and the difference between the pathophysiology of patients with evacuated hematomas and those with diffuse brain injury.

Footnotes

Disclosure Statement

No competing financial interests exist.

References

Aibiki

, Maekawa

, Yokono

. Moderate hypothermia improves imbalances of thromboxane A2 and prostaglandin I2 production after traumatic brain injury in humans. Crit Care Med, 2000; 28:3902–3906.

Adelson

, Ragheb

, Muizelaar

, Kanev

, Brockmeyer

, Beers

, Brown

, Cassidy

, Chang

, Levin

. Phase II clinical trial of moderate hypothermia after severe traumatic brain injury in children. Neurosurgery, 2005; 56:740–754.

Biswas

, Bruce

, Sklar

, Bokovoy

, Sommerauer

. Treatment of acute traumatic brain injury in children with moderate hypothermia improves intracranial hypertension. J Crit Care Med, 2002; 30:2742–2751.

Busto

, Dietrich

, Globus

MYT

et al. Small differences in intraischemic brain temperature critically determine the extent of ischemic neuronal injury. J Cereb Blood Flow Metab, 1987; 7:141–148.

Clifton

. Hypothermia in patients with brain injury: the way forward?–Author's reply. Lancet Neurol, 2011; 10:406–407.

Clifton

, Jiang

, Lyeth

, Jenkins

, Hamm

, Hayes

. Marked protection by moderate hypothermia after experimental traumatic brain injury. Cereb Blood Flow Metab, 1991; 11:114–121.

Clifton

, Miller

, Choi

, Levin

, McCauley

, Smith

, Muizelaar

, Wagner

, Marion

, Luerssen

, Chesnut

, Schwartz

. Lack of effect of induction of hypothermia after acute brain injury. N Engl J Med, 2001; 344:556–563.

Clifton

, Valadka

, Zygun

, Coffey

, Drever

, Fourwinds

, Janis

, Wilde

, Taylor

, Harshman

, Conley

, Puccio

, Levin

, McCauley

, Bucholz

, Smith

, Schmidt

, Scott

, Yonas

, Okonkwo

. Very early hypothermia induction in patients with severe brain injury (the National Acute Brain Injury Study: Hypothermia II): a randomised trial. Lancet Neurol, 2011; 10:131–139.

Drake

, Jory

. Hypothermia in the treatment of critical head injury. CMAJ, 1962; 87:887–891.

10.

Harris

, Muh

, Suries

, Pan

, Rozycki

, Macleod

, Easley

. Discrete cerebral hypothermia in the management of traumatic brain injury: a randomized controlled trial. J Neurosurg, 2009; 110:1256–1264.

11.

Hashaguchi

, Shiozaki

, Ogura

, Tanaka

, Koh

, Noborio

, Fugita

, Akimau

, Kuwagata

, Shimazu

, Sugimoto

. Mild hypothermia reduces expression of heat shock protein 60 in leukocytes from severely head-injured patients. J Trauma, 2003; 55:1054–1060.

12.

Higgins

JPT

, Green

. The Cochrane Handbook of Systematic Reviews. The Cochrane Collaboration. 2008Vol. Version 5.0.0 [updated February 2008]www.cochrane-handbook.org.

13.

Hirayama

, Katayama

, Kano

, Hayashi

, Tsubokawa

. Impact of moderate hypothermia on therapies for intracranial pressure control in severe traumatic brain injury. Intracranial Pressure IX. Nagai

, Ishii

, Maeda

. Tokyo: Springer-Verlag, 1994; 233–236.

14.

Hutchison

, Ward

, Lacroix

, Hébert

, Barnes

, Bohn

, Dirks

, Doucette

, Fergusson

, Gottesman

, Joffe

, Kirpalani

, Meyer

, Morris

, Moher

, Singh

, Skippen

. Hypothermia Pediatric Head Injury Trial Investigators and the Canadian Critical Care Trials Group. Hypothermia therapy after traumatic brain injury in children. N Engl J Med, 2008; 358:2447–2456.

15.

Jiang

, Xu

, Li

et al. Effect of long-term mild hypothermia or short-term mild hypothermia on outcome of patients with severe traumatic brain injury. J Cereb Blood Flow Metab, 2006; 26:771–776.

16.

Jiang

, Yu

, Zhu

. Effect of long-term mild hypothermia therapy in patients with severe traumatic brain injury: 1-year follow-up review of 87 cases. J Neurosurg, 2000; 93:546–549.

17.

Lazorthes

, Campan

. Hypothermia in the treatment of craniocerebral traumatism. J Neurosurg, 1958; 15:162–167.

18.

Marion

, Penrod

, Kesey

, Obrist

, Kochanek

, Palmer

, Wisniewski

, DeKosky

. Treatment of traumatic brain injury with moderate hypothermia. N Engl J Med, 1997; 336:540–546.

19.

Meissner

, Krapp

, Kauf

, Dohrn

, Reinhart

. Thyroid hormone response to moderate hypothermia in severe brain injury. Intensive Care Med, 2003; 29:44–48.

20.

Miller

, Becker

, Ward

, Sullivan

, Adams

, Rosner

. Significance of intracranial hypertension in severe head injury. J Neurosurg, 1977; 47:503–516.

21.

Polderman

, Tjong Tjin Joe

, Peerdeman

, Vandertop

, Girbes

. Effects of therapeutic hypothermia on intracranial pressure and outcome in patients with severe head injury. Intensive Care Med, 2002; 28:1563–1573.

22.

Qui

, Zhang

, Sheng

, Zhang

, Wang

, Liu

, Chen

, Zhou

, Xu

. Effects of therapeutic mild hypothermia on patients with severe traumatic brain injury after craniotomy. J Crit Care, 2007; 22:229–235.

23.

Sedzimir

. Therapeutic hypothermia in cases of head injury. J Neurosurg, 1959; 16:407–414.

24.

Shapiro

, Syte

, Loesser

. Barbiturate-augmented hypothermia for reduction of persistent intracranial hypertension. J Neurosurg, 1974; 40:224–230.

25.

Shiozaki

, Hayakata

, Taneda

et al. A multicenter prospective randomized controlled trial of the efficacy of mild hypothermia for severely head injured patients with low intracranial pressure. J Neurosurg, 2001; 94:50–54.

26.

Shiozaki

, Nakajima

, Taneda

et al. Efficacy of moderate hypothermia in patients with severe head injury and intracranial hypertension refractory to mild hypothermia. J Neurosurg, 2003; 99:47–51.

27.

Shiozaki

, Sugimoto

, Taneda

, Yoshida

, Iwai

, Yoshioka

, Sugimoto

. Effect of mild hypothermia on uncontrollable intracranial hypertension after severe head injury. J Neurosurg, 1993; 79:363–368.

28.

Smrcka

, Vidlák

, Máca

, Smrcka

, Gál

. The influence of mild hypothermia on ICP, CPP and outcome in patients with primary and secondary brain injury. Acta Neurochir Suppl, 2005; 95:273–275.

29.

Strachan

, Whittle

, Miller

. Hypothermia and severe head injury. Brain Inj, 1989; 3:51–55.

30.

Sydenham

, Roberts

, Alderson

. Hypothermia for traumatic head injury. Cochrane Database Syst Rev, 2009; 4:1–49.

31.

Yan

, Tang

, Den

et al. Cerebral oxygen metabolism and neuroeletrophysiology in a clinical study of severe brain injury and mild hypothermia. J Clin Neurosci, 2010; 17:196–200.

32.

Zhang

, Wang

. Comparative study on mild hypothermia in patients with severe head injury and the most severe head injury. Inner Mongolia Med J, 2000; 32:4–6.