Hypothermia Therapy for Traumatic Spinal Cord Injury: An Updated Review

Although hypothermia has shown to protect against ischemic and traumatic neuronal death, its potential role in neurologic recovery following traumatic spinal cord injury (TSCI) remains incompletely understood. Herein, we systematically review the safety and efficacy of hypothermia therapy for TSCI. The English medical literature was reviewed using PRISMA guidelines to identify preclinical and clinical studies examining the safety and efficacy of hypothermia following TSCI. Fifty-seven articles met full-text review criteria, of which twenty-eight were included. The main outcomes of interest were neurological recovery and postoperative complications. Among the 24 preclinical studies, both systemic and local hypothermia significantly improved neurologic recovery. In aggregate, the 4 clinical studies enrolled 60 patients for treatment, with 35 receiving systemic hypothermia and 25 local hypothermia. The most frequent complications were respiratory in nature. No patients suffered neurologic deterioration because of hypothermia treatment. Rates of American Spinal Injury Association (AIS) grade conversion after systemic hypothermia (35.5%) were higher when compared to multiple SCI database control studies (26.1%). However, no statistical conclusions could be drawn regarding the efficacy of hypothermia in humans. These limited clinical trials show promise and suggest therapeutic hypothermia to be safe in TSCI patients, though its effect on neurological recovery remains unclear. The preclinical literature supports the efficacy of hypothermia after TSCI. Further clinical trials are warranted to conclusively determine the effects of hypothermia on neurological recovery as well as the ideal means of administration necessary for achieving efficacy in TSCI.


Introduction
Traumatic spinal cord injury (TSCI) affects 17,800 Americans annually, an incidence that has remained relatively stable over the past three decades [1][2][3]. TSCI is associated with significant increases in care costs, reductions in life expectancy, and reductions in quality of life. Notably, <1% of TSCI victims will be discharged from the hospital with normal neurological function, with approximately 1/3 having complete tetraplegia or paraplegia [1].
Acute SCI pathogenesis is characterized by primary and secondary injury. Primary injury is that which occurs from the initial physical forces of the traumatic event and is often irreversible [4,5]. Secondary injury, in contrast, results from subsequent expansion of the damage [6,7]. It is thought to be, at least in part, preventable, and to stem from multiple mechanisms, including oxidative damage, blood-brain barrier disruption, intracellular hypercalcemia, and neurotransmitter excitotoxicity, which cumulatively drive further tissue destruction [8].

Methods
The English language medical literature was reviewed using Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) guidelines to identify all published preclinical and clinical studies published by 17 August 2021 that have assessed the benefits and safety of hypothermia therapy after TSCI. Articles were identified by querying the PubMed, OVID EMBASE, OVID Medline, Web of Science, and Scopus databases using database-specific queries. The query for PubMed was: (hypothermia OR cooling OR "body temperature" OR "temperature control" OR "body temperature control") AND ("spine trauma" OR "spinal trauma" OR "spinal cord injury" OR "spine injury" OR "spinal injury" OR "vertebral column injury"). Articles yielded from the search were uploaded into the COVIDENCE (Covidence, Melbourne, Australia) system, and these studies underwent title and abstract screening by two independent reviewers, with points of disagreement being resolved by a third reviewer.
Articles were included if they met the following criteria: (1) Study compared subjects treated with hypothermia (cooling to temperature < 36 • C) as a monotherapy or part of a bundled therapy for patients with spinal cord injury. (2) Study employed a human or animal model of traumatic spinal cord injury at any level. (3) Study evaluated one of the following outcomes: mortality, new-onset neurological deficits, neurological deterioration/improvement, or complications. Articles were excluded on full-text review if the study: (1) did not include primary data (i.e., the article was a commentary, letter to the editor, review article, protocol), (2) described fewer than five patients/subjects, (3) a full-text English translation was not available, or (4) the included patient cohort was assessed more thoroughly in a later study by the same authors. Data were extracted using a custom Covidence template. Full-text review and data extraction were performed by two independent reviewers, with a third reviewer serving as a referee in cases of disagreement. Significant heterogeneity in the protocols implemented prevented quantitative analysis.

Results
The search query identified 1142 articles, of which 57 met the criteria for full-text review. Of these, 28 articles met the criteria for inclusion in the final qualitative analysis: 24 examined preclinical TSCI models and 4 described human clinical series. Of the 29 excluded studies, the most common reasons were ( Figure 1): full text was not available online in English language (n = 12) and none of the outlined endpoints were examined (n = 9). One clinical study was excluded due to the authors reporting more extensive outcomes on the same patient cohort in a later included study.

Results
The search query identified 1142 articles, of which 57 met the criteria for full-text review. Of these, 28 articles met the criteria for inclusion in the final qualitative analysis: 24 examined preclinical TSCI models and 4 described human clinical series. Of the 29 excluded studies, the most common reasons were ( Figure 1): full text was not available online in English language (n = 12) and none of the outlined endpoints were examined (n = 9). One clinical study was excluded due to the authors reporting more extensive outcomes on the same patient cohort in a later included study.

Preclinical Data
Of the 24 preclinical studies (Table 1), most (n = 21) employed a thoracic spinal cord model [5,. Twelve studies employed local cooling, eleven used systemic hypothermia, and one evaluated both application modalities. In 12 studies, Sprague-Dawley rats

Preclinical Data
Of the 24 preclinical studies (Table 1), most (n = 21) employed a thoracic spinal cord model [5,. Twelve studies employed local cooling, eleven used systemic hypothermia, and one evaluated both application modalities. In 12 studies, Sprague-Dawley rats were used as the model of interest. Among the 12 studies using exclusively local hypothermia, 8 saw improvement in motor function. The most common means of inducing local hypothermia was epidural cold perfusion. Ha et al. compared neurological outcomes among Sprague-Dawley rats following T9 spinal cord injury that were treated with either 18 • C saline instilled into the epidural space or underwent contusion alone [26]. Animals in the experimental arm were treated for 48 h with a target epidural temperature of 30 • C. At seven days post-injury, treated animals showed significantly better recovery in neurological function, as assessed on the Gale and inclined plane scores. In contrast, Casas et al. failed to demonstrate a significant difference in neurological recovery following T10 injury in female Sprague-Dawley rats treated with 4 • C epidural saline [23]. Animals in the treatment and control arms showed no difference in Basso-Beatie-Bresnahan (BBB) scores at 6 weeks post-injury. Though the reason for the difference in these two studies employing similar injuries is unknown, differences in the injury severity ( [38]. However, unlike the aforementioned study, Teh et al. found significant improvement in neurological function on BBB testing at the 6-week follow-up. Importantly, this study employed delayed initiation of therapy, as animals did not start treatment until 2 h post-injury. Across all included studies, none found a significant difference in the rate of neurological deterioration between treatment and control animals. Of the 11 studies employing systemic hypothermia, 10 employed surface cooling, and the remaining study used a cold air chamber [36]. Interestingly, every preclinical study using surface cooling monotherapy detected improvement in at least one aspect of neurologic function. Seo [36,37]. Each study found that hypothermiatreated animals had significantly (p < 0.05) higher BBB scores at six weeks post-injury. Batchelor et al. studied systemic hypothermia as an adjuvant to surgical decompression in TSCI [22]. Injured animals first underwent systemic hypothermia (target temperature) for 7.5 h, after which surgical decompression was performed, while control animals underwent decompression alone. Animals in the treatment arm had significantly better BBB and ladder-stepping scores as compared to animals that underwent decompression alone.

Clinical Series
Four articles reporting primary data from hypothermia clinical trials were included in this review. An overview of experimental design, inclusion/exclusion criteria, and patient parameters from each study are presented in Table 2. In total, 60 human patients were treated with hypothermia after TSCI, of which 35 received systemic hypothermia and 25 received local hypothermia. Each of the two studies utilizing systemic hypothermia treatment required patients to have an AIS grade A (complete) injury to be eligible for hypothermia therapy [43,44]. Three studies examined hypothermia as an adjuvant to standard care, while the study from Hansebout and Hansebout examined the synergistic effect of local hypothermia and dexamethasone treatment [45]. In all respective cases, intravascular heat exchange catheters were used to induce systemic hypothermia and epidural cooling units were used for local hypothermia. All patients were intubated and sedated based on their clinical condition [43][44][45][46], and those receiving systemic hypothermia were given additional sedation as needed to control shivering [43,44].       Cumulatively, 43% of patients (n = 15) with AIS grade A converted to grade B or better following systemic hypothermia therapy and 58% of individuals (n = 14) with grade A injury converted following local hypothermia (Figure 2). Of note, the AIS A conversion rates of the 31 patients with cervical injuries who received systemic hypothermia (35.5%) were higher than those from 3 major SCI database control studies of patients with cervical injuries who received standard care (26.1%) [44,[47][48][49]. Furthermore, Hansebout and Hansebout reported that 9 out of 14 (64%) patients from their study with cervical TSCI converted from AIS grade A to B or better [45]. None of the patients reported by Gallagher et al. experienced a cervical injury [46]. All clinical studies reported that no neurologic deterioration occurred in any patient as a result of hypothermia treatment.
for hypothermia therapy [43,44]. Three studies examined hypothermia as an adjuvant to standard care, while the study from Hansebout and Hansebout examined the synergistic effect of local hypothermia and dexamethasone treatment [45]. In all respective cases, intravascular heat exchange catheters were used to induce systemic hypothermia and epidural cooling units were used for local hypothermia. All patients were intubated and sedated based on their clinical condition [43][44][45][46], and those receiving systemic hypothermia were given additional sedation as needed to control shivering [43,44].
Cumulatively, 43% of patients (n = 15) with AIS grade A converted to grade B or better following systemic hypothermia therapy and 58% of individuals (n = 14) with grade A injury converted following local hypothermia (Figure 2). Of note, the AIS A conversion rates of the 31 patients with cervical injuries who received systemic hypothermia (35.5%) were higher than those from 3 major SCI database control studies of patients with cervical injuries who received standard care (26.1%) [44,[47][48][49]. Furthermore, Hansebout and Hansebout reported that 9 out of 14 (64%) patients from their study with cervical TSCI converted from AIS grade A to B or better [45]. None of the patients reported by Gallagher et al. experienced a cervical injury [46]. All clinical studies reported that no neurologic deterioration occurred in any patient as a result of hypothermia treatment.  The most frequent complications were pneumonia (60% following systemic and 28% following local hypothermia) and atelectasis (83% following systemic and 36% following local hypothermia) (Figure 3). No thromboembolic events were seen in the retrospective cohort of 14 patients receiving systemic hypothermia from Levi et al. [44]. However, Dididze et al. reported that 24% (n = 5) of patients in their prospective cohort experienced thromboembolic complications: two PEs, one inferior vena cava clot, one DVT of femoral vein, and one clot seen in a subclavian vein [43]. It was noted that the exact same treatment protocols were used for each study, with the exception that Fragmin was used for DVT prophylaxis in the prospective cohort and Lovenox was used in the retrospective cohort [43,44]. The authors also report that one case of ARDS was likely due to aspiration of salt water from drowning and another two were due to aspiration of orogastric fluid during resuscitation [43]. Gallagher et al. decided to terminate their study after 3 out of 5 (60%) patients developed culture-positive wound infections [46]. In this trial, multiple catheters were implanted in the surgical incision site of patients for both local cooling and pathophysiologic monitoring of chemokines/cytokines. On the other hand, only 1 postoperative infection occurred (5%) out of the 20 patients from Hansebout and Hansebout's study who received local hypothermia [45].
water from drowning and another two were due to aspiration of orogastric fluid during resuscitation [43]. Gallagher et al. decided to terminate their study after 3 out of 5 (60%) patients developed culture-positive wound infections [46]. In this trial, multiple catheters were implanted in the surgical incision site of patients for both local cooling and pathophysiologic monitoring of chemokines/cytokines. On the other hand, only 1 postoperative infection occurred (5%) out of the 20 patients from Hansebout and Hansebout's study who received local hypothermia [45].

Discussion
To date, there have been four studies performed which describe clinical outcomes of 60 total patients treated with therapeutic hypothermia. None of the clinical trials reported that a worsening in neurological function occurred because of hypothermia treatment. Overall, these studies found higher AIS conversion rates when adding hypothermia to the TSCI treatment regimen compared to major TSCI database controls. Despite these results, further studies are needed before any definitive conclusions regarding the efficacy of hypothermia in TSCI can be made. Additional investigation into the utility of hypothermia for TSCI has been somewhat tempered by poor outcomes in the literature describing its use for traumatic brain injury (TBI) [50,51]. Like TSCI, TBI was thought to be amenable to hypothermia based on early preclinical and clinical experiences [52][53][54][55]. Nevertheless, despite initial enthusiasm, a recent meta-analysis of the available data found early

Discussion
To date, there have been four studies performed which describe clinical outcomes of 60 total patients treated with therapeutic hypothermia. None of the clinical trials reported that a worsening in neurological function occurred because of hypothermia treatment. Overall, these studies found higher AIS conversion rates when adding hypothermia to the TSCI treatment regimen compared to major TSCI database controls. Despite these results, further studies are needed before any definitive conclusions regarding the efficacy of hypothermia in TSCI can be made. Additional investigation into the utility of hypothermia for TSCI has been somewhat tempered by poor outcomes in the literature describing its use for traumatic brain injury (TBI) [50,51]. Like TSCI, TBI was thought to be amenable to hypothermia based on early preclinical and clinical experiences [52][53][54][55]. Nevertheless, despite initial enthusiasm, a recent meta-analysis of the available data found early prophylactic hypothermia to offer no benefit in terms of neurological recovery [51]. Additionally, it was associated with higher rates of complication as compared to standard of care with normothermic temperature goals. For this reason, the most recent version of the Brain Trauma Foundation guidelines recommends against the use of prophylactic hypothermia [56].

Neuroprotective Effects of Hypothermia in Animals
Initial studies performed concerning the neuroprotective role of hypothermia in animals indicated that hypothermia reduces cerebral metabolic oxygen consumption in a canine model [57]. These reports led to the hypothesis that lower temperatures decrease the metabolic demand of the central nervous system, thus protecting it from ischemic damage. Both moderate systemic [25,37,58] and local hypothermia [26,36] treatments exhibit antiapoptotic and anti-inflammatory effects in preclinical models of TSCI. Some reports indicate that systemic application may suppress extrinsic apoptosis signaling more effectively than localized cooling [36]. Additionally, systemic hypothermia diminishes autophagy and oxidative stress in animal spinal cord contusion models [30,37,59]. Assessment of electrophysiologic, histologic, and functional outcomes following spinal cord injury in rats has demonstrated that those under hypothermic conditions exhibit fewer sequelae compared to those under normothermic conditions following compressive TSCI [60]. Additionally, hy-pothermia following TSCI in the rat model has been shown to increase normal white matter and gray matter by up to 31% and 38%, respectively [32]. Other observed histologic effects of hypothermia include a four-fold preservation of neurons adjacent to the TSCI epicenter as well as 127% greater sparing of axonal connections supplied by reticulospinal neurons.

New Avenues to Monitor Neurologic Injury
To assess a therapeutic's efficacy in TSCI remains an extremely difficult task. To date, no statistical conclusions can be drawn on the ability of hypothermia therapy to improve neurologic recovery despite higher AIS conversion rates reported for treated individuals when compared to established control studies from the literature. The broad variation in traumatic injury mechanisms between cases effects neurologic outcomes beyond what can be detected by current imaging standards alone. Indeed, no two TSCI cases are the same. However, novel techniques designed to monitor the inflammatory biomarkers via intrathecal catheter may aid in predicting neurologic recovery after TSCI [61]. Studies from Dalkilic et al. reveal that MRI measures in combination with cerebrospinal fluid (CSF) biomarkers provide the strongest model for classifying baseline AIS grade. However, when predicting AIS conversion, a logistic regression model of CSF biomarkers alone showed 91.2% accuracy and was not aided by the addition of MRI and/or baseline AIS data. Gallagher et al. utilized these techniques in parallel with local hypothermia but stopped the trial due to high infection rates (60%) [46]. Hansebout and Hansebout reported only 1 infection in 20 patients (5%) but initiated treatment sooner than Gallagher and for a shorter duration [45,46]. No studies to date have attempted to monitor CSF biomarkers in patients with TSCI who received systemic hypothermia. However, systemic hypothermia has been unsuccessful in attenuating CSF cytokines after TBI in pediatric patients [62]. Further clinical studies are needed to delineate the relationship between CSF inflammatory markers and neurologic recovery after TSCI.

Safety of Hypothermia Therapy
Current data suggest that hypothermia is not a benign therapy. While most of the complications in the clinical series identified here were attributed to the neurological injury as opposed to hypothermia treatment, patients receiving systemic hypothermia are potentially at higher risk for coagulopathy due to decreased body temperature and presence of an intravenous catheter [63,64]. Notably, a small study (n = 10) showed that 50% of patients treated with hypothermia for TBI experienced DVT [65]. However, patients whose catheters were removed within 4 days of insertion had a 33% incidence rate compared to 75% for those removed at later time points. Such risks may potentially be avoidable by implementing a local as opposed to systemic hypothermia regimen. To this end, Levi et al. found similar rates of complications among TSCI patients treated with systemic hypothermia and case-matched controls who did not receive hypothermia therapy [44]. In aggregate, the clinical studies in this review identified venous thromboembolic events in 14% of patients who received systemic hypothermia after TSCI as compared to 24% who were treated with local hypothermia, further supporting the safety of systemic treatment [43]. Nevertheless, local hypothermia confers its own risks, which are in part related to the means in which local cooling is implemented. Gallagher et al. used a cooling catheter placed in the epidural space to effect local hypothermia [46]. Three of their five patients suffered surgical site infection, leading them to prematurely terminate the study. However, in a previous report by the group looking at intradural monitoring for TSCI, no infections were experienced [66]. Consequently, it is unclear if the risk stems from sampling bias or the hypothermia intervention itself.

Future Directions
As demonstrated by the data reviewed here, the limited human studies demonstrate that hypothermia was not associated with any increased risk of new neurological injury or deterioration, though there were increased respiratory risks. Additionally, these data suggest that hypothermia may increase the odds of neurological improvement, by at least one level on the AIS grade, though this data is relatively limited. This by-in-large agrees with the data that are reported for the preclinical, animal literature. Larger, multi-center randomized controlled trials are necessary to draw definitive conclusions on the efficacy of and complication rates associated with hypothermia therapy for TSCI.

Limitations
There are several limitations to the present study. Preclinical literature has been previously reported to have high rates of non-publication of negative findings [67]. Consequently, the preclinical data presented here may reflect a systemic bias. Indeed, there is a notable degree of heterogeneity present within the studies included in this review. For example, among the animal studies we described, there were significant differences with respect to specific techniques used to model TSCI, animals used, and outcomes measured, among other factors. Additionally, the conclusions drawn from human trials in this review were limited in part by the patient sample size, the retrospective nature of data, and the heterogeneity of patient treatment timelines, as well as the means by which hypothermia therapy was delivered.

Conclusions
Hypothermia therapies represent a heretofore underexplored therapeutic option for patients with traumatic spinal cord injury. Though the clinical studies at present are extremely limited, they suggest that hypothermia may lead to a small improvement in longterm neurological outcomes, though with some increased respiratory risk. Based on this analysis, hypothermia was found to be a relatively safe therapeutic option with potential to enhance the neurological recovery of individuals who experience a TSCI. Further research is warranted to determine the ideal means of delivering hypothermia as well as the optimum target temperature and treatment duration necessary for achieving efficacy after TSCI in humans.

Conflicts of Interest:
The authors declare no conflict of interest.