Impact of Therapeutic Hypothermia During Cardiopulmonary Resuscitation on Neurologic Outcome: A Systematic Review and Meta-analysis

Background: Therapeutic cooling initiated during cardiopulmonary resuscitation (intra arrest therapeutic hypothermia, IATH) provided diverging effect on neurological outcome of out-of-hospital cardiac arrest (OHCA) patients depending on the initial cardiac rhythm and the cooling methods used. Methods: We performed a systematic search of PubMed, EMBASE and the CENTRAL databases using established Medical Subject Headings (MeSH) terms for IATH and OHCA. Only studies comparing IATH to standard in-hospital targeted temperature management (TTM – control group) were selected. We used the revised Cochrane RoB-2 and the Newcastle-Ottawa scale tool to assess risk of bias of each study. Primary outcome was favorable neurological outcome (FO); secondary outcomes included return of spontaneous circulation (ROSC) rate and overall survival to hospital discharge. Two authors independently assessed the validity of included human studies and extracted data regarding characteristics of the studied cohorts and main outcomes. Results: Out of 20950 studies, 8 studies (n=3493 patients, including 4 randomized trials, RCTs) were included in the nal analysis. When compared to controls, the use of IATH was not associated with improved favorable neurological outcome (OR 0.96 [95% CIs 0.68-1.37]; p= 0.84), increased ROSC rate (OR 1.11 [95% CIs 0.83-1.49]; p= 0.46) or survival to hospital discharge (OR 0.91 [95% CIs 0.73-1.14]; p= 0.43). Signicant heterogeneity among studies was observed only for the analysis of ROSC rate (I 2 =69%). Trans-nasal evaporative cooling and cold uids were explored in two RCTs each and no signicant differences were observed on neurological outcome. However, trans-nasal evaporative cooling was associated with a higher probability of favorable neurological outcome when compared to controls in patients with an initial shockable rhythm (OR 1.62 [95% CI 1.00-2.64]; p=0.05]. Conclusions: In this meta-analysis, IATH was not associated with improved neurological outcome when compared to standard in-hospital TTM. However, there are considerable outcome differences depending on the methods used and the studied population that need to be explored in future trials. three components for each


Introduction
Anoxic brain injury frequently causes signi cant neurological sequelae and is the main cause of poor outcome in resuscitated cardiac arrest patients (CA) [1,2]. At present, there are few established treatments that in randomized trials have shown to limit the magnitude of such injury. Thus, these patients in general receive supportive therapy and admission to the intensive care units (ICUs) to control the systemic causes of secondary brain damage (i.e. fever, anemia, hypocapnia, hyponatremia, hyperoxemia and dysglycemia).
The use of targeted temperature management (TTM) with a core body temperature of 33-36° C for at least 24 hours is recommended to mitigate brain injury. [3]. In particular, rapid brain cooling has been considered to have the greatest potential to reduce the extent of post-anoxic brain injury in the immediate post-CA phase, through different mechanisms, including reduction of cerebral metabolic rate, excitatory amino acids release, reduction in in ammation and apoptotic signals and free radical production [4][5][6][7].
However, diverse aspects related to the application of TTM remain controversial, including the optimal target temperature, patient selection, timing of TTM initiation, the most effective methods to induce and maintain TTM, the duration of the cooling phase and the rewarming rate/fever control strategies [8]. In particular, contrary to cardiac arrest laboratory investigations, initiation of TTM is often delayed by several hours in clinical practice because of several factors, such as the need for transportation or diagnostic procedures, the availability and performance of different cooling devices, thus potentially limiting its bene cial effects. To reduce the delay between the onset of the cardiac arrest and TTM initiation, several trials has focused on the early pre-hospital period and provision of TTM in the eld, i.e. either during cardiopulmonary resuscitation (CPR) or immediately after ROSC prior to hospital arrival. This very early approach, which has been identi ed as "intra-arrest therapeutic hypothermia" (IATH), has been administered by different cooling methods, including selective brain cooling, rapid infusion of cold uids and trans-nasal evaporative cooling (TNEC) [9][10][11][12][13][14][15][16]. However, despite some preliminary encouraging ndings [10], variable results in RCTs have suggested either a potential bene t (i.e. in OHCA patients with an initial shockable rhythm) or deleterious effects, in particular when large volumes of cold uids have been used [14,[16][17].
The aim of this systematic review and meta-analysis was therefore to investigate whether the application of IATH could have an impact on clinically relevant outcomes, including neurological outcome (FO), return of spontaneous circulation (ROSC) and overall survival, in patients suffering from out-of-hospital cardiac arrest (OHCA).

Methods
We adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis -Protocols (PRISMA-P) guidelines [18]. The protocol of this study was registered with the International Prospective Register of Systematic Reviews (PROSPERO) and last edited on the 2nd of June 2020 (CRD42019130322).

Data sources and Search strategies
A systematic literature search was performed up to the 28th of May 2020 in the PubMed, EMBASE and CENTRAL databases using the following Medical Subject Headings (MeSH) terms: "heart arrest" AND/OR "hypothermia" AND/OR "hypothermia, induced" AND/OR cryotherapy" AND/OR cardiopulmonary resuscitation" AND/OR critical care outcomes" AND/OR "patient outcome assessment" AND/OR "morbidity" AND/OR "mortality". Only studies using IATH and having a control group treated with standard in-hospital care including TTM were selected for the nal analysis.
The search included only original studies published in English in peer-reviewed journals. The complete research string for each database as well as the main research questions, with reference to participants, interventions, comparisons, outcomes and study design (PICOS), are reported in ESM (Appendix 1 and Table S1). In addition, we searched the reference lists of eligible studies and relevant reviews for additional published and unpublished data, searched by contacting experts, and used a web search for abstracts, proceedings, and unpublished studies.

Study screening and selection
Two authors (FA and LP) independently screened study titles and abstracts for potential eligibility and assessed their validity. Disagreement between authors was assessed and resolved through a third reviewer (FST), who reviewed the original text of the article. FA and MF extracted data regarding characteristics of the studied cohort and the main outcomes related to the use of IATH including neurological outcome, ROSC rate and survival.
In the analysis, we included only studies evaluating IATH in adult (> 18 years of age) CA patients in either retrospective, prospective or randomized controlled trials (RCTs). Studies conducted in healthy volunteers or in animal models were excluded. Editorials, commentaries, letters to editor, opinion articles, reviews, meeting abstracts, case reports, and studies published in other languages were also excluded, as well as original articles lacking abstract and/or quantitative details on neurological outcome and survival. When multiple publications of the same research group/center described case series with potential overlap, the more recent publication, if eligible, was considered. None of the authors of the original studies was contacted to obtain further information, which were not available in the published manuscript. Only studies that met all the above criteria were incorporated for quantitative synthesis.

Appraisal of study quality
The level of evidence (LOE) of each study was assessed according to the Grading of Recommendations, Assessment, Development and Evaluations (GRADE) evidence system [19]. For RCTs, the risk of bias (ROB) was assessed using the Revised Cochrane RoB-2 tool that classi es the ROB as "low," "probably low," "probably high," or "high" for each of the following domains: randomization process, deviation from the intended interventions, missing outcomes data, measurement of the outcome and selection of the reported results. [20] The ROB for non-RCTs was assessed using the Newcastle-Ottawa Scale [21]; three different components were evaluated for each study: (a) selection of cases: studies were considered as "low" ROB if case de nition was adequate, cases were representative and outcome of interest was not present at the beginning of the study; (b) comparability of cohorts: studies were considered as "low" ROB if adjustment was made for usual prognostic factors (i.e., Utstein variables); (c) exposure and outcome: studies were considered as "low" ROB if assessment of outcome and follow-up were appropriate. Overall, a study was considered as "low" ROB if each single component was classi ed as "low." LOE was further analyzed by two experts (FST, PN) and one independent statistician. Disagreement was resolved by consensus.
The primary outcome of the meta-analysis was the occurrence of favorable neurological outcome (FO), whenever it was recorded. Favorable neurological outcome was de ned as a Cerebral Performance Category of 1-2. Secondary outcomes were ROSC rate and survival, whenever this was collected. Prede ned analyses were performed in subgroups of patients: a) IATH using TNEC vs. IATH using other methods; b) patients presenting with an initial shockable rhythm.

Statistical analysis
Means of survival and favorable neurological outcomes probabilities were obtained by weighting each study by the inverse of variance. The Mantel-Haenszel method was chosen as the reference method for xed effects analysis. The Mantel-Haenszel formula is applied to calculate an overall, unconfounded, effect estimate of a given exposure for a speci c outcome by combining stratum-speci c odds-ratios (OR). Stratum-speci c ORs are calculated within each stratum of the confounding variable and compared with the corresponding effect estimates in the whole group. A Z test was carried out to assess the signi cance of the risk differences. The I 2 was calculated by χ 2 test to assess variability due to heterogeneity rather than chance. A substantial heterogeneity was assumed with I 2 > 50%. 95% con dence intervals (CIs) for neurologic outcome and survival were calculated with the Wilson method and placed in forest plots and statistical signi cance was assumed for p < 0.05. The presence of publication bias was evaluated by trim and ll. The trim and ll method estimates the number of missing studies from a meta-analysis due to the suppression of the most extreme results on one side of the funnel plot. Then, this method augments the observed data and recomputes the summary estimate based on the complete data. The trim and ll outputs were obtained with iterations. Analyses were performed for all the selected studies, as well as grouped by RCT vs. observational trials. Statistical analysis was conducted by Review Manager 5.3 software and funnel and forest plots were developed.

Study selection
A total of 20950 records were identi ed after the initial search. After the rst screening procedure, 43 studies were assessed for eligibility. Of those, 35 did not t inclusion criteria, as reported in Figure 1; a total of 8 studies [9][10][11][12][13][14][15][16], including 3493 patients, were then included for meta-analysis.

Study characteristics
The characteristics of the selected studies are summarized in Table 1. We identi ed 4 RCTs (high quality of evidence for two [12,16] and moderate for two [10,14]), two prospective studies (low level of evidence in one [9] and very low quality of evidence [15] in the other, which compared a prospective cohort with historical controls) and two retrospective studies (very low quality of evidence [11,13]). The ROB for RCTs was low in three studies [10,12,16] and showed some concerns in one [14] ( Table 2). For all non-RCTs, the ROB was high [9,11,13,15] (Table 3). All selected studies included only patients after OHCA. In ve studies, IATH was induced using intravenous cold uids [11][12][13][14][15], whereas two studies used TNEC [10,16] and one selective cranial cooling [9].

Analysis of outcomes according to initial rhythm
Speci c data on neurological outcome according to the initial rhythm were reported only in two studies [10,16]. Among patients with an initial shockable rhythm, FO occurred more frequently in the IATH than in the control group (OR=1.62 [95% CIs 1.00-2.64]; p= 0.05, Figure 4); no differences in patients with an initial non-shockable rhythm were observed. Three studies reported data on survival rate according to the initial rhythm [10,14,16]; no differences in ROSC rate and survival were observed between IATH and controls, regardless of the initial rhythm ( Figure S3).

Discussion
This systematic review and meta-analysis analyzes the application of IATH in the clinical setting, including 8 studies and 3493 patients. Our results showed that IATH does not improve the occurrence of favorable neurological outcome, ROSC rate or survival in the entire population when compared to standard in-hospital care. Nevertheless, it also appeared that speci c subgroups of patients, such as the ones presenting with an initial shockable rhythm, might bene t the most from such intervention. Moreover, the method selected to induce IATH might also be relevant, suggesting a possible greater bene t from the use of TNEC compared to other methods.
Previously, other reviews and meta-analysis have been conducted to explore the impact of pre-hospital cooling after cardiac arrest [22][23][24][25][26][27], but none of them was speci cally designed to investigate the application of only IATH. To our knowledge, only a previous review conducted in 2012 [28] speci cally addressed this topic, but human data were scarce at the time, with the two largest published trials missing [14,16], and also experimental studies were included. Two additional studies investigating IATH using cold intravenous uids in the human setting, for a total of 50 patients, were also identi ed during the research process but were eventually excluded because of the lack of a control group [29,30]. The two largest studies included, despite not showing signi cant effects of IATH, suggested opposite effects of the treatment regarding both ROSC and survival rate. In the study by Bernard et al. [14], IATH was induced using rapid infusion of cold intravenous uids and the trial was stopped after enrolling 48% of the previewed recruiting target due to changes in TTM target in most Australian sites. In this trial, patients treated with IATH required longer CPR, additional epinephrine, experienced more frequently pulmonary edema on admission and, among patients with an initial shockable rhythm, a signi cantly lower ROSC rate was reported. Interestingly, in the study by Nordberg et al. [16], which used TNEC to induce IATH, no differences were observed between groups related to the ROSC rate, adrenaline requirements and CPR duration. However, a higher proportion of patients achieving FO, in particular complete neurological recovery (i.e. CPC 1) was observed in the IATH group when patients with an initial shockable rhythm were analyzed. Whether the method used to induce IATH could explain those differences remains debatable, but our ndings suggest that possibly TNEC might be more effective in patients presenting with shockable rhythm, probably because of selected and rapid brain cooling and the lack of adverse effects on the coronary perfusion pressure, which can be reduced by cold uids [31].
Our study has several limitations. First, despite not being signi cant in most of the performed analyses, the heterogeneity between studies regarding differences in IATH protocol, including the timing to assess the outcome, have probably reduced the possibilities to identify an effect of the treatment. Second, we could not exclude that differences in managing neuro-prognostication and withdrawal of life sustaining therapies would result in diverse end-of life procedures among the selected trials, possibly in uencing the pooled results. Third, the number of patients that could be included for subgroup analysis was limited and our results are only hypothesis-generating. Forth, no patient data were available, thus introducing potential limitations to adjust for confounders. Fifth, TTM protocols may vary between hospitals and along the time, possibly modifying the post-resuscitation care and thus having an impact on the neurological outcome of the patients included. Finally, several studies presented with a high risk of bias or raising some concerns, thereby reducing the robustness of our ndings.

Conclusions
This systematic review and meta-analysis showed that IATH does not improve the occurrence of favorable neurological outcome, ROSC rate or survival when compared to standard in-hospital TTM in the global population of OHCA patients, but it could possibly be bene cial in the subgroup of patients presenting with a shockable rhythm if treated with TNEC. Further studies should then assess how to identify the population of patients that might bene t the most from such intervention.   Forest plot for favorable neurological outcome (2A), return of spontaneous circulation (ROSC) rate (2B) and survival (2C) including randomized clinical trials (RCTs) and non-RCTs: intra arrest therapeutic hypothermia (IATH) vs standard in-hospital targeted temperature management (Control). legend: The size of the squares for the risk ratio re ects the weight of the trial in the pooled analysis. The horizontal bars represent 95% con dence intervals (CIs).
standard in-hospital targeted temperature management (Control). legend: The size of the squares for the risk ratio re ects the weight of the trial in the pooled analysis. The horizontal bars represent 95% con dence intervals (CIs).

Figure 4
Forest plot for favorable neurological outcome (FO) in patients presenting with shockable rhythms vs non-shockable rhythm: intra arrest therapeutic hypothermia (IATH) vs. standard of care (Control). legend: The size of the squares for the risk ratio re ects the weight of the trial in the pooled analysis. The horizontal bars represent 95% con dence intervals (CIs).

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.