Does therapeutic hypothermia during extracorporeal cardiopulmonary resuscitation preserve cardiac function?

Background Extracorporeal cardiopulmonary resuscitation (E-CPR) is increasingly used as a rescue method in the management of cardiac arrest and provides the opportunity to rapidly induce therapeutic hypothermia. The survival after a cardiac arrest is related to post-arrest cardiac function, and the application of therapeutic hypothermia post-arrest is hypothesized to improve cardiac outcome. The present animal study compares normothermic and hypothermic E-CPR considering resuscitation success, post-arrest left ventricular function and magnitude of myocardial injury. Methods After a 15-min untreated ventricular fibrillation, the pigs (n = 20) were randomized to either normothermic (38 °C) or hypothermic (32–33 °C) E-CPR. Defibrillation terminated ventricular fibrillation after 5 min of E-CPR, and extracorporeal support continued for 2 h, followed by warming, weaning and a stabilization period. Magnetic resonance imaging and left ventricle pressure measurements were used to assess left ventricular function pre-arrest and 5 h post-arrest. Myocardial injury was estimated by serum concentrations of cardiac TroponinT and Aspartate transaminase (ASAT). Results E-CPR resuscitated all animals and the hypothermic strategy induced therapeutic hypothermia within minutes without impairment of the resuscitation success rate. All animals suffered a severe global systolic left ventricular dysfunction post-arrest with 50–70% reductions in stroke volume, ejection fraction, wall thickening, strain and mitral annular plane systolic excursion. Serum concentrations of cardiac TroponinT and ASAT increased considerably post-arrest. No significant differences were found between the two groups. Conclusions Two-hour therapeutic hypothermia during E-CPR offers an equal resuscitation success rate, but does not preserve the post-arrest cardiac function nor reduce the magnitude of myocardial injury, compared to normothermic E-CPR. Trial registration FOTS 4611/13 registered 25 October 2012 Electronic supplementary material The online version of this article (doi:10.1186/s12967-016-1099-y) contains supplementary material, which is available to authorized users.


Background
Survival after cardiac arrest is greatly influenced by early post-arrest cardiac function [1,2]. Hence, cardiopulmonary resuscitation (CPR) strategies that preserve post-arrest cardiac function may improve outcome. Extracorporeal CPR (E-CPR) by veno-arterial extracorporeal membrane oxygenation (ECMO) is increasingly used when standard CPR fails. Promising results have been reported by using E-CPR as a rescue method within brief timeframes for selected cases [3][4][5][6]. Therapeutic hypothermia (HT, 32-34 °C) is widely used for patients resuscitated from cardiac arrest as it is believed to exhibit cardiovascular [7,8] and neurological benefits [9][10][11]. Bergan et al. J Transl Med (2016) 14:345 To achieve cardiac benefit from HT the importance of early and rapid cooling has been emphasized in experimental studies [12][13][14]. E-CPR provides the opportunity to rapidly induce HT, but whether hypothermic E-CPR preserves post-arrest cardiac function and hence improves outcome, is not known.
A severe cardiac dysfunction following normothermic E-CPR is recently demonstrated in pigs [15]. The present study aimed to investigate if HT during E-CPR improves cardiac outcome early post-arrest. We hypothesized that hypothermic E-CPR offers an equal resuscitation success rate, but with a better preserved post-arrest left ventricular (LV) function and less myocardial injury compared to normothermic E-CPR.

Animal welfare
The experimental protocol was approved by the Norwegian National Animal Research Authority and the animal experiments were performed in accordance with the European Convention for the Protection of Vertebrate Animals used for Experimental and Other Scientific Purposes (European Council, ETS No. 170). Detailed information according to the ARRIVE guidelines is presented in Additional file 1: Table S1 [16]. With respect to animal welfare`s 3R-principle, eight 38-ECPR animals were included in a separate methodological study demonstrating E-CPR associated post-arrest LV dysfunction by cardiac magnetic resonance imaging (MRI) per se [15].

Animal preparation
The animal preparation included premedication in the pig enclosure and total intravenous anaesthesia in the operating theatre, mechanical ventilation and a succeeding surgical tracheotomy and placements of intravascular catheters and ECMO-cannulas as recently described [15]. Ringer's acetate solution was infused at 10 ml/kg/h.

Baseline assessments
After the preparation and a following 30-min stabilization period, baseline cardiac MRI (Philips Achieva 3 Tesla, Philips Medical Systems, DA Best, Netherland) and haemodynamic measurements of LV function (MPR-500, Millar Instruments, Houston, TX, USA) were obtained. Arterial and mixed venous blood samples were analyzed (ABL 800 Flex, Bergman Diagnostika, Kjeller, Norway) and serum concentrations of cardiac Troponin T (cTnT) and Aspartate aminotransferase (ASAT) were measured.
After an intravenous injection of 2 mg/kg heparin, an electrical stimulator connected to a right ventricular pacing lead (Qstim 5Fr, VascoMed GmbH, Binzen, Germany) induced ventricular fibrillation (VF), confirmed by ECG shape and aortic blood pressure drop.

E-CPR
After 15 min of untreated VF the animals received either normothermic [pulmonary artery blood temperature 38.0 °C (normothermia in the pig)] or hypothermic (32.0-33.0 °C) E-CPR at maximum ECMO blood flow rate with a 100% oxygen sweep gas set at the same flow rate as the ECMO blood flow rate.
HT in the 32-ECPR group was achieved using 20.0 °C priming solution with later adjustments at the heatexchanger. The heat-exchanger thermostat at 38.0 °C ensured normothermia in the 38-ECPR group.
After 5 min of E-CPR 360 Joule monofasic defibrillations (CodeMaster XL + Hewlett Pachard, Lexington, KY, USA) were provided until regain of spontaneous cardiac beating (ROSB) with extracorporeal support continuing at unchanged blood flow rate and temperature target for 120 min. In the 32-ECPR group a 30-min warming period followed, whereas a corresponding 30-min continued run at 38.0 °C was provided in the 38-ECPR group.

Weaning from ECMO
After the 120 + 30-min extracorporal support, with all animals being normothermic, a step-wise separation from ECMO (weaning) was completed during a 60-min period, and the animals were allowed to stabilize after weaning for another 60 min before the post-arrest assessments.

Post-arrest assessments
At 285 min post-arrest LV function was re-assessed by LV pressure measurements and MRI. Finally, a second arterial and mixed venous blood sample were analyzed and blood samples for cTnT and ASAT measurements were collected, before the pig was euthanized.

cTnT and ASAT
The serum concentration measurements of cTnT and ASAT pre-arrest and 6 h post-arrest were performed by an electro-chemiluminescence immunoassay (Troponin T hs, Roche Diagnostics, Rotkreuz, Switzerland) and by an UV-test with pyridoxal phosphate activation (ASAT, Roche Diagnostics) using an automated clinical chemistry analyzer (Modular analytical platform, module E170 and P800, Roche Diagnostics) to estimate myocardial injury.

Statistical analysis
Statistical analyses were made using Graphpad prism 6.04 (GraphPad Software, La Jolla, CA, USA). Data are reported as mean ± standard deviation if not otherwise stated. The statistical significance level α was set to 0.05 and power 1 −β to 0.80.
The study sample size was estimated by a prospective power analysis. The least detectable difference considered as clinically significant between cardiac function variables was 15% of baseline values. Cardiac function variability in range of 5-15% of baseline values in pilot experiments made a calculated sample size of 20 necessary to achieve the desired power.  A paired two-tailed Student's t test (t) was used to compare baseline and post-arrest measurements within each treatment group, and an unpaired two sample t-test (T) was used to compare post-arrest measurements between the two different treatment groups. Alternatively, a two-tailed Mann-Whitney test (MW) of group differences with exact p-value was used for data not normally distributed.
One animal in the 38-ECPR group was euthanized after ROSB, and was thus excluded from further analyses, because MAP could not be sustained as dictated by the protocol, due to ECMO venous cannula malfunctioning.
At initiation of VF the blood temperature was 38.0 ± 0.2 °C and was maintained at this level in the 38-ECPR group (Fig. 1). In the 32-ECPR group the temperature quickly dropped after initiation of hypothermic E-CPR and was 33.3 ± 1.0 °C at the time of defibrillation (Fig. 2). It further dropped to the targeted 32-33 °C within the 1st min after defibrillation (5.25 ± 4.8 min from start E-CPR) and was kept stable at this level until warming.

Defibrillation
All animals were successfully defibrillated by a median of 1 shock (range 1-6) with no significant differences between the groups (MW, p = 0.99). After ROSB three animals in the 38-ECPR group spontaneously had a second VF, and immediately received a median of 1 (range 1-3) additional defibrillations. No additional defibrillations were needed in the 32-ECPR group (MW, p = 0.21).
MAP at the time of defibrillation was 54 ± 9 mmHg in the 38-ECPR group and 60 ± 5 mmHg in the 32-ECPR group (T, p = 0.082) (Fig. 3) with an ECMO blood flow rate of 4.6 ± 0.1 l/min and 4.4 ± 0.2 l/min in the two groups, respectively (T, p = 0.055).

Haemodynamic measurements and blood gas analyses
HR increased similarly post-arrest in both groups (Table 1), limiting the reductions in CO. The increased Ea post-arrest did not differ between the groups, and the dP/dt max , dP/dt min , EDP, EDP/EDV relationship, and tau did not change significantly from baseline values in either group.
Mixed venous oxygen saturation dropped considerably and group-similarly post-arrest (Table 1). A small decrease in haemoglobin content, base excess and pH was also measured post-arrest, with an accompanying increase in arterial lactate.

TTC staining post mortem
The assessment of myocardial infarction by TTC staining post mortem did not demonstrate regional infarctions.

Discussion
In the present controlled animal study comparing hypothermic and normothermic E-CPR considering resuscitation success, post-arrest LV function, and myocardial injury, surprisingly, and contrary to our hypothesis, no beneficial effects of HT during E-CPR could be demonstrated.
A better preserved LV function would be desirable in resuscitated patients as post-arrest cardiac function is related to patient survival. In the present study, HT during E-CPR was hypothesized as being beneficial because HT has demonstrated cardioprotective effects in various animal studies with regional myocardial ischaemia (i.e. coronary occlusion) [21,22].
The best strategy of E-CPR to preserve early postarrest cardiac function is not known and no guidelines exist to assist clinicians deciding on an E-CPR strategy for patients in refractory cardiac arrest. To date HT is not recommended as a cardioprotective intervention in patients with acute myocardial infarction without associated cardiac arrest [22,23]. In resuscitated patients, however, HT is an established treatment due to neurological benefits, irrespective of any cardioprotection [9][10][11]. Whole body cooling targeting 32-36 °C is the latest recommendation (preferably a constant temperature in this range) and a HT induction time-frame of 4-6 h postarrest is usually accepted [24]. A delay of several hours from resuscitation to target temperature may exclude a cardioprotective effect of hypothermia per se. HT may nevertheless be favorable to the cardiovascular function as it may reduce cardiac work load as a consequence of reduced whole body metabolism during HT [25].
Correct timing of HT has been emphasized in recent years as HT is claimed to be cardioprotective only if induced either shortly before or at the time point of myocardial reperfusion [12,13,26,27]. Maeng and coworkers found that HT induced at the time of coronary reperfusion did not reduce myocardial infarct size in pigs [28]. Despite the efficient HT induction, the present study did not demonstrate cardioprotection by hypothermic E-CPR (i.e. reperfusion) of the fibrillating heart. Cooling the myocardium prior to myocardial reperfusion may thus be a crucial procedure, but effort is needed to achieve hypothermia this early in the clinical coronary occlusion scenario [29,30]. Correspondingly, a rationale for cardioprotective intra-arrest HT (i.e. HT induced before ROSB) exists in cardiac arrested patients, and is supported by animal studies [13,31,32], but the suggested post-arrest cardiac function benefits are not specifically investigated in humans. If interventions to reperfuse the myocardium are postponed until HT is established, the harm of prolonged ischaemia may cancel any benefits of HT. This issue could be further investigated experimentally using topical cooling of the arrested heart prior to reperfusion. Two-hour HT duration was investigated as brief cooling has been sufficient to achieve cardioprotective effects in previous animal studies [12] and 3-h cooling has made myocardial damage worse [33]. Whether a longer HT duration would be beneficial in our study is not known, but cannot be excluded. On the other hand, extracorporeal circulatory support by ECMO is not without complications and side effects, and clinical practice is to wean as soon as the heart is capable to independently handle the circulation.
Compared to our study, rewarming of patients after cardiac arrest is slow (0.3-0.5 °C/h), tailored for neuroprotection [24]. In cardiac surgery, however, a quick rewarming is well tolerated by the heart even after

Table 1 Haemodynamic measurements and blood gas analyses
Values are expressed as mean ± standard deviation. Comparison post-arrest to baseline within group by paired student's t-test Post-arrest comparison of groups by unpaired two-sample student's t-test. MD mean difference, CI 95% confidence interval HR heart rate, MAP mean aortic blood pressure, CVP central venous pressure, LVP max systolic left ventricular pressure maximum, EDP end-diastolic pressure, dP/dt max maximum left ventricular pressure first time derivate, dP/dt min

Inotropes
Inotropic support is regularly used during VA-ECMO to sustain aortic ejections with a sufficient pulse-pressure to avoid LV distention and failure. The increased dobutamine requirements observed during HT may be related to altered pharmacologic properties of dobutamine with a reduced effect [34] as supported by the similar requirements in the two groups after rewarming. HT also affects LV function, causing slower LV contraction and relaxation velocities [35], and may thus increase the need for inotropic stimulation to reach preset targets. The optimal MAP target during HT is not known, and could possibly differ from MAP target at normothermia, but for comparison, they were set at the same level.

LV function
In the present study, the 50-70% reductions in SV, EF, wall thickening, strain and MAPSE were consistent, demonstrating a severe global systolic LV dysfunction post-arrest with uniform impairments in all directions of systolic LV motion, and with no differences between the two treatment groups.
The diastolic LV function assessed by dP/dt min , tau and EDP/EDV relationship was preserved in both groups post-arrest, and neither EDV (preload) nor Ea (afterload) differed between the two groups.

Myocardial injury evaluation
The post-arrest LV dysfunction indicated a severe and global myocardial injury as was confirmed by the considerable increase in serum concentrations of cTnT and ASAT in both groups. A global injury without distinguished regional areas was also confirmed by the TTC assessment, as no regional infarctions could be demonstrated, excluding coronary thrombus or embolic complications.
The success rate of resuscitation by hypothermic E-CPR was not inferior to the normothermic strategy, and neither LV function nor myocardial injury was exacerbated. E-CPR initiated by a room tempered ECMO may be convenient, as a normothermic ECMO (or a circuitry heating device) will not always be available in clinical emergency settings that may include emergency rooms, ambulance transfers and even pre-hospital use.

Limitations
The clinical scenario of E-CPR differs from a controlled animal experiment as the period of no-flow is usually Fig. 4 Cardiac MRI measurements. The systolic left ventricle function variables did not differ between the normothermic and the hypothermic E-CPR group at baseline or post-arrest. Strain, peak global systolic left ventricular circumferential strain; Wall thickening, radial mid left ventricular wall thickening; MAPSE, mitral annular plain systolic excursion. Line at mean ± standard deviation (short line) short (preferably <5 min) and the patient is cannulated at ongoing CPR in a low-flow period of varying duration. In the present study, a healthy pig heart suffered an electrically induced VF and the total ischemic insult was prepared to be substantial and consistent to assure a significant post-arrest cardiac dysfunction and injury that could be compared between the two different treatment groups. The duration of no-flow thus exceeded usual clinical limits, and the low-flow period was bypassed.

Conclusions
E-CPR is an effective resuscitation technique for prolonged cardiac arrest that may rapidly induce HT. In the present animal study, 2-h HT during E-CPR offered an equal resuscitation success rate, but did not preserve the post-arrest cardiac function nor reduce the magnitude of myocardial injury, compared to normothermic E-CPR.