Edited by: Barak Bar, Loyola University Medical Center, United States
Reviewed by: M. Kamran Athar, Thomas Jefferson University, United States; RajaNandini Muralidharan, Winthrop University Hospital, United States
This article was submitted to Neurocritical and Neurohospitalist Care, a section of the journal Frontiers in Neurology
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
Large hemispheric infarction (LHI) is commonly associated with arterial occlusion of the proximal middle cerebral artery (MCA) or internal carotid artery (ICA) (
Hypothermia is an effective intervention for neuroprotection available to acute brain injury patients (
Studies have investigated the efficacy and safety of using hypothermia for LHI patients. Nevertheless, its role in managing the patients suffering from LHI remains debatable (
We followed the Cochrane Handbook for Systematic Reviews of Interventions to conduct this systematic review and meta-analysis (
The eligibility criteria for this systematic review were as follows. (1) Type of studies: RCTs of hypothermia vs. normothermia. (2) Type of participants: clinical signs of a large supratentorial infarction, infarction involving not <2/3 of the MCA territory confirmed by cranial computed tomography (CT) or magnetic resonance imaging (MRI), with or without the territories of anterior and/or posterior cerebral artery additionally affected (
The MEDLINE (OvidSP) database, Embase (OvidSP), Cochrane Central Register of Controlled Trials (CENTRAL, OvidSP), China Biological Medicine Database (CBM), and clinical trials registers were searched comprehensively. The final search was conducted on September 21, 2018. Additionally, we searched the reference lists of selected publications, and contacted authors for more information if necessary. Search terms that we used to identify publications included “stroke OR brain ischemia OR cerebrovascular disorders OR ischemia apoplexia OR cerebral occlusion OR cerebral infarction” AND “hypothermia OR cryotherapy OR cooling OR temperature OR normothermia” AND “randomized controlled trial OR randomized OR controlled clinical trial.” The search strategies we used in MEDLINE (OvidSP) are shown in
Two reviewers (JL and YHG) carried out the study selection and data extraction independently throughout the process. In the first sift-prescreening after pretest, two reviewers screened the titles and abstracts independently to identify studies that could be included in the next section. In the second sift-selection after pretest, the two reviewers went through the full texts for studies that met the eligibility criteria. Similar pretests were conducted throughout the process. Next, the two reviewers extracted data from every report using a pre-designed electronic form for data collection. The differences were resolved by discussion, and a third reviewer (LXW) acted as a mediator if an agreement could not be reached. We also established contact with the corresponding authors to get extra information if required.
The data were extracted as follows: (1) first author, publication year, country of origin, and publication type; (2) participant characteristics: sex, age, and baseline National Institutes of Health Stroke Scale (NIHSS) score; (3) type of intervention: the method of inducing hypothermia, target temperature, duration of hypothermia, rewarming rate, whether the patient underwent DHC or did not undergo DHC in the hypothermia group, whether normothermia was maintained, and whether patients underwent DHC in the control group; (4) type of outcome measures: mortality, neurological outcome [assessed by the modified Rankin scale (mRS), Barthel index (BI), or NIHSS], adverse events during treatment (incidence of intracranial hypertension or developing herniation in the rewarming process), adverse events related to hypothermia (such as bleeding events, infections, hyperglycemia, cardiac arrhythmia, hypotension, and shivering), and time points of measurement.
The study quality was assessed using the Cochrane Handbook as a reference for risk of bias (
For dichotomous outcomes, mortality, neurological outcome of survivors as assessed by mRS (0–3 vs. 4–5), and adverse events during treatment, risk ratios (RRs) and 95% confidence intervals (CIs) were calculated for the meta-analysis. For continuous outcomes and neurological outcome assessed by mRS, standardized mean differences (SMDs) and 95% CI were used. The chi-squared test and
There were 3,699 records identified, including 3,696 records from the five databases and three records obtained from searching the reference lists. After removing duplicates, 3,382 records were screened for their titles and abstracts. A total of 75 articles required reading of the full texts, and 70 of them were excluded. Two studies were added to ongoing studies (
Flow diagram of study selection for the meta-analysis. CENTRAL, Cochrane Central Register of Controlled Trials; CBM, China Biological Medicine Database; LHI, large hemispheric infarction.
Three RCTs published between 2006 and 2018 with 131 participants were included. Two studies were conducted in China, while one was conducted in Germany. All of the three studies had reported the age, gender, and baseline NIHSS of the two groups, and two of these studies had elaborated on patients' underlying diseases. For baseline of the three studies, no statistical significance was found between the hypothermia group and the control group, apart from one study in which age and gender were not comparable (
Characteristics of the included studies and participants.
Els et al. ( |
Germany | 6/12 | 4/13 | 49 ± 12 | 49 ± 6 | 18 ± 2 | 19 ± 2 | 10 patients via endovascular cooling; 2 patients via surface cooling | 35°C | 48 h | 1°C/24 h | Underwent | Maintained normothermia | Underwent | Within the first week after the onset | At 6 months of follow-up |
Su et al. ( |
China | 1/16 | 7/17 | 59.8 ± 8.6 | 68.5 ± 8.5 | 19.7 ± 2.9 | 20.4 ± 3.8 | Endovascular cooling | 33 or 34°C | 24–72 h | Raised 0.5°C every 12 h; rewarming rate was 0.1°C/h | Did not undergo | Maintained normothermia | Did not undergo | At 6 months of follow-up | At 6 months of follow-up |
Liang et al. ( |
China | 7/37 | 11/36 | 61.2 ± 6.2 | 60.6 ± 5.8 | 19.6 ± 2.4 | 19.7 ± 2.6 | Surface cooling | 32–35°C | 5–7 days | Duration of more than 8 h from 34 to 36°C; rewarming phase, more than 12 h | Did not undergo | Maintained normothermia | Did not undergo | At 3 months of follow-up | At 3 months of follow-up |
The seven items for risk of bias are shown graphically (
All of the three studies reported mortality (
Meta-analysis of association between hypothermia and the mortality of large hemispheric infarction (LHI).
Three studies reported neurological outcome as assessed by mRS. Two of them presented it as a dichotomous variable of good neurological outcome for survivors (
Meta-analysis of association between hypothermia and a good neurological outcome for survivors.
Meta-analysis of association between hypothermia and a low level of mRS for large hemispheric infarction (LHI).
Only one study with 73 participants reported data in which herniation developed in the rewarming process. Herniation during the rewarming process following hypothermia was not significantly higher than after normothermia (RR, 1.46; 95% CI, 0.45–4.74). No study had details for incidence of intracranial hypertension.
All of the three studies reported the adverse events related to hypothermia, while one study had no details for the hypothermia group or the control group, and thus was excluded. In total, 106 participants were included in the analysis. There were significantly higher risks of gastrointestinal bleeding (RR, 3.24; 95% CI, 1.62–6.49), gastric retention (RR, 2.43; 95% CI, 1.47–3.99), and electrolyte derangement (RR, 1.59; 95% CI, 1.12–2.28), following hypothermia compared with following normothermia. There were no significant differences found in the association between hypothermia and incidences of pneumonia (RR, 1.40; 95% CI, 0.87–2.25), cardiac arrhythmia (RR, 1.20; 95% CI, 0.75–1.91), hemorrhagic transformation (RR, 0.67; 95% CI, 0.20–2.23), hyperglycemia (RR, 1.36; 95% CI, 0.68–2.69), hypotension (RR, 1.87; 95% CI, 0.76–4.60), and acute kidney injury (RR, 5.00; 95% CI, 0.90–27.71) compared with normothermia. Significant associations were found between hypothermia and shivering (RR, 5.84; 95% CI, 1.88–18.12) in one study with 73 participants. Only one study with 33 participants looked at venous thrombotic events, and no significant associations were found (RR, 2.13; 95% CI, 0.45–10.05). These findings are shown in
There was low heterogeneity detected between studies for adverse events related to hypothermia, with the exception of substantial heterogeneity for hyperglycemia (
The results of meta-analysis for adverse events during treatment.
Developed herniation in the rewarming process | 1 | 6 |
4/36 | – |
– | 0.63 | 0.53 | Fixed effect | 1.46 | (0.45, 4.74) |
Pneumonia | 2 | 25/53 | 18/53 | 39% | 0.20 | 1.37 | 0.17 | Fixed effect | 1.40 | (0.87, 2.25) |
Cardiac arrhythmia | 2 | 22/53 | 18/53 | 33% | 0.22 | 0.77 | 0.44 | Fixed effect | 1.20 | (0.75,1.91) |
Hemorrhagic transformation | 2 | 4/53 | 6/53 | 0% | 0.82 | 0.66 | 0.51 | Fixed effect | 0.67 | (0.20, 2.23) |
Gastrointestinal bleeding | 2 | 26/53 | 8/53 | 0% | 0.65 | 3.32 | 0.0009 | Fixed effect | 3.24 | (1.62, 6.49) |
Gastric retention | 2 | 31/53 | 13/53 | 0% | 0.79 | 3.49 | 0.0005 | Fixed effect | 2.43 | (1.47, 3.99) |
Hyperglycemia | 2 | 15/53 | 11/53 | 72% | 0.06 | 0.87 | 0.38 | Fixed effect | 1.36 | (0.68, 2.69) |
Hypotension | 2 | 11/53 | 6/53 | 0% | 0.77 | 1.36 | 0.17 | Fixed effect | 1.87 | (0.76, 4.60) |
Electrolyte derangement | 2 | 35/53 | 22/53 | 0% | 0.45 | 2.56 | 0.01 | Fixed effect | 1.59 | (1.12, 2.28) |
Shivering | 1 | 18/37 | 3/36 | – | – | 3.05 | 0.002 | Fixed effect | 5.84 | (1.88, 18.12) |
Acute kidney injury | 2 | 7/53 | 1/53 | 0% | 0.96 | 1.84 | 0.07 | Fixed effect | 5.00 | (0.90, 27.71) |
Venous thrombotic events | 1 | 4/16 | 2/17 | – | – | 0.95 | 0.34 | Fixed effect | 2.13 | (0.45, 10.05) |
For mortality, subgroup analyses were performed according to whether the participants underwent DHC or did not undergo it, and at different measurement time points during hospitalization or at the end of follow up. Neither participants with DHC nor without DHC showed a statistically significant association with mortality (RR: 0.54, 95% CI, 0.06–5.24; RR: 1.16, 95% CI, 0.79–1.72, respectively). No significant associations were found between hypothermia and mortality compared with normothermia, neither during hospitalization nor at the end of follow-up (RR: 0.54, 95% CI, 0.06–5.24; RR: 1.16, 95% CI, 0.79–1.72, respectively). Both subgroup analyses are shown in
This systematic review included three RCTs to evaluate the effects of hypothermia on mortality, neurological outcome, and adverse events. LHI is associated with high mortality and disability rate. The major reasons for it could be the brain swelling that occurs within 1–5 days after the onset of LHI, intracranial hypertension, or even herniation (
Our findings showed that hypothermia was not significantly associated with a reduction in mortality of LHI patients. Similarly, we found that neither a previous meta-analysis with six RCTs involving 252 patients who suffered from acute ischemic stroke (
In our review, neurological outcome was assessed by the mRS, which is widely used to measure neurological outcome for stroke patients in RCTs. The pooled estimate showed that there was statistically significant association between hypothermia and good neurological outcome for survivors. Additionally, there was significant association of hypothermia with improvement in neurological outcome. These positive effects are likely due to the neuroprotection of hypothermia. Our findings were consistent with several clinical studies conducted on LHI patients (
We found that hypothermia was a double-edged sword; it had some benefits in the neurological outcome of LHI patients, but there was an increase in adverse events during treatment as well. The most serious adverse event we noted was the incidence of intracranial hypertension or herniation developing in the rewarming process, which likely results in death of the patient. A rebound increase in ICP during or after rewarming had been observed in several studies (
Subgroup analyses were conducted based on patients who underwent DHC or did not undergo it and at different measurement times. For patients who underwent DHC, statistical significance was detected between hypothermia and normothermia in terms of neurological outcome. A study has shown that LHI patients might achieve more advantages from combining hypothermia with DHC compared with DHC alone (
We note several strengths in our study. First, our review was conducted strictly in accordance with the Cochrane Handbook; our analytic plan was pre-specified in protocol with a rigorous and scientific methodology. Second, our literature search was comprehensive, with neither language restrictions nor publication status limitations. After searching comprehensively, only three RCTs were included in our review. We found that the small sample size was an inherent limitation in this type of clinical trial due to the complexity and variability of LHI patients, clinical studies were too difficult to carry out. Third, the three RCTs were assessed to have a low risk of bias in general, which increased the reliability of the results. Fourth, we conducted subgroup analyses on participants who underwent DHC or did not undergo it, which might act as an alert to the possible advantages of combining hypothermia and DHC for the management of LHI. Finally, we found that hypothermia might contribute to improving neurological outcome after pooling effect estimates, a result which was not obvious in individual studies. To conclude, this is the first systematic review and meta-analysis performed to investigate the effect of hypothermia on patients with LHI.
Several limitations in our study merit careful consideration. The main shortcomings are the very few studies and the small sample size of patients included in our meta-analysis. Unfortunately, the strength and quality of evidence might be restricted by this paucity, which ought to be taken cautiously in clinical practice. However, a small number of studies does not mean the results are meaningless. Many systematic reviews from the Cochrane Library include less than three studies, or even none, but they are still considered clinically important (
This meta-analysis suggested that hypothermia was not associated with the mortality of LHI patients, but it was associated with the improvement of neurological outcome. Meanwhile, we found that hypothermia was associated with higher risk of gastrointestinal bleeding, gastric retention, electrolyte derangement, and shivering. It was, however, not associated with the risk of developing herniation in the rewarming process, pneumonia, cardiac arrhythmia, hemorrhagic transformation, hyperglycemia, hypotension, acute kidney injury, and venous thrombotic events. Given the limited number of studies and low sample size, the evidence should be treated cautiously when applied to clinical practice. Future studies are needed to demonstrate the efficacy and safety of hypothermia for LHI patients.
GL and LW contributed to the conception, design, and data interpretation of this study. JL and YG contributed to the study selection, data extraction, quality assessment, and data analysis. JL drafted the manuscript. XC and MW designed the search strategy and performed the search. XC, MW, and MZ completed the tables and figures. MZ critically revised the manuscript. All authors read and approved the final manuscript.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
The Supplementary Material for this article can be found online at: