Hypothermic Total Liquid Ventilation Is Highly Protective Through Cerebral Hemodynamic Preservation and Sepsis-Like Mitigation After Asphyxial Cardiac Arrest*

Objectives:Total liquid ventilation provides ultrafast and potently neuro- and cardioprotective cooling after shockable cardiac arrest and myocardial infarction in animals. Our goal was to decipher the effect of hypothermic total liquid ventilation on the systemic and cerebral response to asphyxial cardiac arrest using an original pressure- and volume-controlled ventilation strategy in rabbits. Design:Randomized animal study. Setting:Academic research laboratory. Subjects:New Zealand Rabbits. Interventions:Thirty-six rabbits were submitted to 13 minutes of asphyxia, leading to cardiac arrest. After resumption of spontaneous circulation, they underwent either normothermic life support (control group, n = 12) or hypothermia induced by either 30 minutes of total liquid ventilation (total liquid ventilation group, n = 12) or IV cold saline (conventional cooling group, n = 12). Measurements and Main Results:Ultrafast cooling with total liquid ventilation (32°C within 5 min in the esophagus) dramatically attenuated the post–cardiac arrest syndrome regarding survival, neurologic dysfunction, and histologic lesions (brain, heart, kidneys, liver, and lungs). Final survival rate achieved 58% versus 0% and 8% in total liquid ventilation, control, and conventional cooling groups (p < 0.05), respectively. This was accompanied by an early preservation of the blood-brain barrier integrity and cerebral hemodynamics as well as reduction in the immediate reactive oxygen species production in the brain, heart, and kidneys after cardiac arrest. Later on, total liquid ventilation also mitigated the systemic inflammatory response through alteration of monocyte chemoattractant protein-1, interleukin-1&bgr;, and interleukin-8 transcripts levels compared with control. In the conventional cooling group, cooling was achieved more slowly (32°C within 90–120 min in the esophagus), providing none of the above-mentioned systemic or organ protection. Conclusions:Ultrafast cooling by total liquid ventilation limits the post–cardiac arrest syndrome after asphyxial cardiac arrest in rabbits. This protection involves an early limitation in reactive oxidative species production, blood-brain barrier disruption, and delayed preservation against the systemic inflammatory response.

30 minutes of total liquid ventilation (total liquid ventilation group, n = 12) or IV cold saline (conventional cooling group, n = 12). Measurements and Main Results: Ultrafast cooling with total liquid ventilation (32°C within 5 min in the esophagus) dramatically attenuated the post-cardiac arrest syndrome regarding survival, neurologic dysfunction, and histologic lesions (brain, heart, kidneys, liver, and lungs). Final survival rate achieved 58% versus 0% and 8% in total liquid ventilation, control, and conventional cooling groups (p < 0.05), respectively. This was accompanied by an early preservation of the blood-brain barrier integrity and cerebral hemodynamics as well as reduction in the immediate reactive oxygen species production in the brain, heart, and kidneys after cardiac arrest. Later on, total liquid ventilation also mitigated the systemic inflammatory response through alteration of monocyte chemoattractant protein-1, interleukin-1β, and interleukin-8 transcripts levels compared with control. In the conventional cooling group, cooling was achieved more slowly (32°C within 90-120 min in the esophagus), providing none of the above-mentioned systemic or organ protection. Conclusions: Ultrafast cooling by total liquid ventilation limits the post-cardiac arrest syndrome after asphyxial cardiac arrest in rabbits. This protection involves an early limitation in reactive oxidative species production, blood-brain barrier disruption, and delayed preservation against the systemic inflammatory response. (Crit Care Med 2015; 43:e420-e430) Key Words: asphyxia; cardiac arrest; cardiopulmonary resuscitation; cerebral ischemia; hypothermia; liquid ventilation B eyond neurologic and hemodynamic disorders (1), multivisceral dysfunction and respiratory complications (2,3) participate to the dramatic outcome of resuscitated patients after cardiac arrest. Temperature management has been proposed to improve this outcome (4,5), but the ideal target temperature is still debated between 32°C and 34°C (mild therapeutic hypothermia) or 36°C (6). This is in apparent contradiction with many animal studies clearly demonstrating potent benefits with therapeutic hypothermia at 32-34°C (7)(8)(9)(10)(11)(12). Such discrepancy could be a consequence of the differential ability to achieve rapid cooling in laboratory animals compared with humans. Indeed, if rodents can be externally cooled within a couple of minutes, humans typically required couple of hours (13). This can, in part, explain the potent protection observed with intra-arrest experimental hypothermia in laboratory animals (7) compared with the lack of benefit with prehospital cooling using current methods in the clinical arena (13). Thus, a key challenge is to determine whether it is possible to cool a human body as rapidly as in animals and to decipher the mechanisms supporting the benefit of such rapid cooling. Several techniques are currently proposed to achieve this goal, including peritoneal lavage (14), nasal evaporative cooling (15), and partial (16) or total liquid ventilation (TLV) (8,11,12). The latter technique can use the lungs as a heat exchanger and provides rapid and systemic cooling independently of body weight (8,17,18). It can for example cool down the entire body at 32°C within 10-20 minutes in rabbits, lambs, or swine (8,(17)(18)(19). After shockable cardiac arrest, hypothermic TLV improved survival (8,17) as well as neurologic (8), cardiac (17), and renal outcomes (20) in rabbits while conventional cooling (CONV) was not protective (8,17). Until now, the clinical translation of this concept was limited by the previous demonstration of lung injuries in early trials with partial liquid ventilation in patients with acute respiratory distress syndrome (21). By contrast, recent experimental studies suggested that TLV could be better tolerated and beneficial in this context of respiratory care (22,23). Here, we propose to use a new and original specifically dedicated liquid ventilator accurately controlling both tidal volume and pulmonary pressure (24,25) in order to prevent lung trauma and to provide systemic and rapid hypothermia in the context of nonshockable cardiac arrest.
For these purposes, we investigated TLV in an experimental model of nonshockable and asphyxial cardiac arrest, as the effect of hypothermia is especially debated after nonshockable rhythm (26). Neurologic, multivisceral, and inflammatory outcomes were therefore assessed after 13 minutes of asphyxia in rabbits. The specific hypothesis was that hypothermic TLV could be highly neuroprotective through mitigation of early reperfusion alterations after cardiac arrest, that is, cerebral blood flow disturbances, blood-brain barrier (BBB) permeability, and reactive oxygen species (ROS) production. Importantly, the effects of TLV were compared with CONV with external method and cold saline infusion.

MATERIALS AND METHODS
Animal instrumentation and the ensuing experiments were conducted in accordance with French official regulations after approval by the local ethical committee (ComEth AnSES/ ENVA/UPEC no. 16).

Animal Preparation and Experimental Protocol
Male New Zealand rabbits (2.5-3.0 kg) were anesthetized using zolazepam, tiletamine, and pentobarbital (all 20-30 mg/kg, IV). After intubation and mechanical ventilation (Fio 2 = 30%), rabbits were paralyzed by cisatracurium besylate (0.2 mg/kg, IV). Asphyxial cardiac arrest was then induced by disconnecting the endotracheal tube from the ventilator during 13 minutes. Preliminary experiments demonstrated that cardiac arrest, that is, null cardiac output, was obtained within 4.5-5.5 minutes (data not shown). After 13 minutes of asphyxia, animals were resuscitated using external cardiac massage (200 compressions/min) and epinephrine administration (15 μg/kg, IV). After resumption of spontaneous circulation (ROSC), animals were randomly assigned to one experimental group (control, CONV, or TLV) (Fig. 1A). The administration of epinephrine was further permitted during 8 hours (target mean blood pressure = 80 mm Hg). The control group did not receive any additional procedure and was maintained under normothermic condition by the use of thermal pads. In the CONV group, hypothermia was induced by CONV through the IV administration of 30 mL/kg of cold saline (NaCl, 0.9%, 4°C) during the first 30 minutes after ROSC. It was combined with the application of cold blankets (0-4°C) upon the skin with no prior shaving. In the TLV group, ultrafast cooling was induced by TLV started 5 minutes after ROSC. The lungs were filled by 13 mL/kg of perfluorooctane (F2 Chemicals, Preston, United Kingdom), with an initial temperature of 20°C and a progressive increase to 33°C. The liquid ventilator was initially set to a tidal volume of 10 mL/kg, a respiratory frequency of 8 cycles/min, and a positive end-expiratory pressure (PEEP) of 2 cm H 2 O (Fig. 1B). We used a previously described algorithm with a volume-and pressure-controlled liquid ventilation mode monitoring PEEP and positive end-inspiratory pressure (PEIP) (Fig. 1C). Static lung compliance was calculated by dividing tidal volume by PEIP -PEEP. After 30 minutes of TLV, animals were weaned from TLV by prolonged exhalation at -15 cm H 2 O.
The liquid ventilator was then disconnected and animals were shifted to conventional mechanical ventilation. In the two hypothermic groups (CONV and TLV), the target temperature of 32°C was maintained by cold blankets until the 4th hour after ROSC. Animals were then slowly rewarmed with infrared lights and thermal pads during 4 hours before weaning from conventional ventilation and awakening. Mechanical ventilation was continued until weaning from the mechanical ventilation and awakening of the animals. Rabbits were subsequently returned to cages for survival follow-up. They were housed in a closed cage enriched in oxygen for 24 hours. They received antibiotics (enrofloxacin, 5 mg/kg, IM) and analgesics (buprenorphine, 30 μg/kg, SC) every day for 3 days.   (8,17,27). Survival was monitored during 3 days before euthanasia and organs sampling. In accordance with our ethical committee, animals eliciting a neurologic dysfunction score above 80% at 24 hours or 60% at 48 hours were prematurely euthanized. he heart, kidneys, liver, lungs, and brain were examined by histopathology, and lesions were quantified using a 0-3 score system (0 = normal; 3 = very severe; Supplemental Table 2, Supplemental Digital Content 1, http://links.lww.com/ CCM/B353; Supplemental Table 3, Supplemental Digital Content 1, http://links.lww.com/CCM/B353), as previously described (8,17,27). For lungs, two different score systems were used for infectious or cardiogenic congestive lesions, respectively. In the brain, the overall score was obtained by the average of the right and left cortex, hippocampus, and cerebellum (8,17,27). Blood was also sampled at different time points for quantification of messenger RNA expression of inflammation and/ or hypoxia markers, including interleukins (IL) 1β, 8, and 10, interferon (INF)-γ, tumor necrosis factor (TNF)-α, hypoxiainducible factor 1 (HIF1)-α, and heme oxygenase (HO)-1, as previously described (20) (Supplemental Table 4, Supplemental Digital Content 1, http://links.lww.com/CCM/B353).

Assessment of ROS Production, BBB Integrity, and Cerebral Blood Volume
Additional rabbits were submitted to the previously described procedure for early organ sampling and fixation, that is, 30 minutes after cardiac arrest. Brain cortex and hippocampus, heart, kidneys, lungs, and liver were frozen in liquid nitrogen.
ROS production was evaluated by electron paramagnetic resonance spectroscopy (detailed protocol in Supplemental Method, Supplemental Digital Content 1, http://links.lww. com/CCM/B353) as previously described (28)(29)(30). Other rabbits were submitted to the same procedures of cardiac arrest for BBB integrity evaluation using the Evans Blue Dye leakage   ). In a last set of experiments, cerebral hemodynamics was assessed during cardiac arrest and after resuscitation using ultrafast ultrasound Doppler imaging, as previously described (31,32).

Statistical Analysis
Data were expressed as mean ± sem. Hemodynamic and biochemical variables were compared between different groups using a two-way analysis of variance for repeated measures. If necessary, post hoc analyses were performed at each time point using a Student t test with Bonferroni correction. Values were not compared between the different time points to avoid multiple comparisons. Neurologic dysfunction and histologic scores were compared between groups using a Mann-Whitney nonparametric test. Survival curves were obtained with a Kaplan-Meier analysis and compared between groups with a log-rank test. Significant differences were determined at p value less than or equal to 0.05.

TLV Can Be Instituted Safely After Resuscitation
As shown in Table 1 Table 2) and static lung compliance (Fig. 1D) in the TLV group. The compliance impairment after cardiac massage was even partially reversed after weaning in the TLV group compared with control and CONV groups, respectively.

TLV Induces Rapid Cooling Without Hemodynamic Disorders
In the TLV group, body temperatures decreased to 32°C within 5-15 minutes as illustrated in Figure 2. In comparison, CONV required 90-120 minutes to achieve the same target (CONV group). As shown in Table 1, heart rate decreased throughout hypothermia in the CONV and TLV groups while mean arterial pressure, blood gases, and biochemical variables were not different between groups. The amount of epinephrine administered to prevent hypotension was however significantly lower in the TLV group compared with control.

TLV Offers Neurologic and Lung Protection and Improves Survival
As illustrated in Figure 3A, ultrafast cooling by TLV was associated with a dramatic improvement of the neurologic status compared with control and CONV groups. For example, neurologic dysfunction score was 48% at day 1 in the TLV  group compared with 89% and 90% in the CONV and control groups (median values), respectively. This reduction was significant when all animals were took into account (including dead animals with 100% score) but also when dead animals were excluded from this analysis (scores of alive animals only). The final survival rate achieved 58% in the TLV group versus 0% and 8% in the control and CONV groups, respectively (p < 0.05) (Fig. 3B). The neuroprotective effect of TLV was associated with a significant limitation of brain ischemic lesions compared with the control and CONV groups ( Fig. 4 and Supplemental Fig. 1A-D Beyond cerebral lesions, lung congestive and infectious complications were the most severe consequences of cardiac arrest after pathologic examinations (Fig. 4). Congestive lesions were attenuated in TLV (Supplemental Fig. 1, E and F Fig. 1H, Supplemental Digital Content 1, http://links.lww.com/CCM/ B353). A trend toward reduced pulmonary infection was also observed in the TLV versus control and CONV groups (Fig. 4).

TLV Limits Early BBB Disruption and Cerebral Hyperemia
We then hypothesized that the multiple organ protection afforded by TLV could be linked to mitigation of early reperfusion injury. For this purpose, we investigated ROS production in additional organ samples withdrawn 30 minutes after cardiac arrest in rabbits (n = 8 in each group). As illustrated in Figure 5, this production was decreased in the brain cortex, heart, and kidneys in the TLV group compared with control (p < 0.05). A nonsignificant tendency was also observed in the hippocampus and liver. Interestingly, slow cooling did not reduce ROS production in the CONV group. We then hypothesized that the decrease in ROS production could be a major trigger of neuroprotection through BBB integrity and cerebral hemodynamics protection. Indeed, BBB disruption was observed very early after cardiac arrest through Evans blue dye leakage (Fig. 6, A and B; n = 4 in each group). TLV significantly limited this disruption compared with control and CONV groups when assessed 30 minutes after cardiac arrest (n = 4 in each group). In order to corroborate this vascular effect, cerebral blood volume was further evaluated by ultrafast ultrasound Doppler in other rabbits (Fig. 6C). As shown in Figure 6D, cerebral hyperemia lasted approximately 30 minutes in the control group (n = 2) and was attenuated in the TLV group (n = 2).

TLV Mitigates the Systemic Responses to Cardiac Arrest
In order to assess the subsequent systemic responses to cardiac arrest, we then investigated the effect of TLV on the Reactive oxygen species (ROS) production. The ROS production was assessed in the brain cortex, hippocampus, heart, kidneys, lungs, and liver at t = 30 min after cardiac arrest (n = 8 in each group) by electron paramagnetic resonance spectroscopy. Data are expressed in arbitrary unit (AU) per gram of protein per minute. *p < 0.05 versus control; †p < 0.05 versus conventional cooling (CONV). TLV = total liquid ventilation.
Critical Care Medicine www.ccmjournal.org e427 sepsis-like syndrome (2). The severity of hypoxic injury was evidenced by a potent up-regulation of HIF1-α and HO-1 (Fig. 7A). Interestingly, early cooling with TLV amplified HO-1 up-regulation from the 8th hour after cardiac arrest. As illustrated in Figure 7B, an early proinflammatory response was also observed through an increase in IL-1β, IL-8, and monocyte chemoattractant protein (MCP)-1 transcripts. This response was significantly reduced by TLV versus control at the 8th hour after cardiac arrest. The anti-inflammatory IL-10 cytokine was simultaneously upregulated in the control group and tended to be attenuated by TLV at the 24th hour after cardiac arrest. These data suggest that early cooling with TLV mitigates both the early proinflammatory and delayed anti-inflammatory responses induced by cardiac arrest. This was also associated with a down-regulation of INF-γ and TNF-α at the 24th hour after cardiac arrest in all groups.

DISCUSSION
In the present study, ultrafast cooling with TLV potently limited the post-cardiac arrest syndrome after nonshockable and asphyxial cardiac arrest in rabbits while CONV was inefficient. The protection was initiated as early as the first minutes following resuscitation. It was associated with inhibition of ROS production, limitation of BBB disruption, and prevention of cerebral hyperemia. An ultimate attenuation of the so-called sepsis-like syndrome was also observed, showing that short, rapid, and systemic hypothermia could attenuate the entire systemic response to anoxia. Figure 6. Assessment of blood-brain barrier integrity and cerebral blood volume after cardiac arrest. A, Bloodbrain barrier permeability assessed by cerebral Evans blue dye concentration. Evans blue was distributed in the cerebral tissue through vascular leakage after in vivo administration at t = 30 min after cardiac arrest (n = 4 in each group). B, Typical pictures of brain slices in the different groups. Blue areas represent Evans blue dye vascular leakage. C, Ultrafast ultrasound Doppler imaging after cardiac arrest. The echographic frontal views show the visual repartition of cerebral blood volume at the maximum intensity in representative rabbits of the control and total liquid ventilation (TLV) groups. The corresponding hippocampus and cortical blood volumes were calculated in one control and one TLV rabbit. D, Overall cortical blood volume from ultrafast ultrasound Doppler imaging in one control and one TLV rabbit. *p < 0.05 versus control; †p < 0.05 versus conventional cooling (CONV). A.U. = arbitrary unit, CPR = cardiopulmonary resuscitation.
Importantly, our experimental conditions of cardiac arrest were particularly severe as mortality and neurologic dysfunctions were maximal in the control group. CONV was even unable to provide any benefit compared with TLV, supporting the concept that the rapidity of cooling plays a decisive protective role in such very severe conditions. Previous studies also demonstrated that immediate hypothermia was dramatically protective in mice after 8 minutes of asystole but not after a 20-minute delay after resuscitation (7). Interestingly, the benefit of conventional hypothermia is also highly challenged in humans after nonshockable cardiac arrest (26).
From a biochemical point of view, the severity of the ischemic injury is associated with a potent up-regulation of HIF-1α in all groups in the present study. This transcription factor is well known to be enhanced during hypoxia (33) and could lead to various consequences such as vascular endothelium growth factor secretion, apoptosis, or oxidative stress (34). Its up-regulation could be linked to the activation of HO-1 protective pathway (35), which occurred earlier in the TLV group compared with control or CONV groups. Such an earlier activation could promote ubiquitous benefits regarding ROS production and inflammation in the TLV group (36). In addition, ultrafast cooling with TLV potently limited multiple organ ROS production as soon as 30 minutes after cardiopulmonary resuscitation initiation. As this was observed not only in the brain but also in the heart and kidneys, it emphasizes the importance of systemic hypothermia by TLV.
In order to further investigate the neuroprotective effect of TLV, we also investigated BBB permeability and showed very rapid impairment in control conditions, that is, 30 minutes after cardiac arrest. This impairment was totally prevented by TLV, in agreement with the previously shown limitation in ROS production and reduced hyperemia observed with ultrafast ultrasound imaging. The lack of benefit of CONV in these conditions also suggests that hypothermia should be achieved at least before 90 minutes after cardiac arrest to efficiently inhibit the triggering events of neurologic injury.
Since one major interest of TLV is to provide rapid and systemic cooling, we also investigated the systemic responses to cardiac arrest and the "sepsis-like" biochemical disorders (2). Importantly, the balance between the early proinflammatory and the later anti-inflammatory responses was modified by TLV but not through CONV. This is consistent with previous studies showing a lack of benefit with delayed hypothermia in rats submitted to asphyxial cardiac arrest (37). The benefit of TLV persisted far into the rewarming phase (i.e., 8 hr after cardiac arrest), demonstrating that mitigation of early resuscitation disorders (e.g., ROS generation) resulted in delayed benefits during the entire post-cardiac arrest syndrome. The alterations in MCP-1 levels are especially of interest in this regard, as it is associated with brain injury during stroke (38). Conversely, INF-γ and TNF-α were not really modified in our conditions, which is likely related to a selective regional activation and/or different time courses of activation (39).
In agreement with our previous studies (8,17,20), we confirmed here that TLV was well tolerated regarding lung function and histology. For this purpose, we used an original and advanced liquid ventilator that accurately controls both liquid filling pressures and volumes within the lungs (25). Beyond its capacity to induce an ultrafast and systemic hypothermia, such liquid ventilator was also able to preserve lung function and ultrastructure at difference with conventional hypothermia with cold fluid which promoted lung congestion (13). Alteration in lung static compliance after cardiac massage was even reversed by TLV, several hours after cardiac arrest compared with other groups. This impairment in control conditions could be consequence of pulmonary ischemia-perfusion and/or direct effect of the cardiac massage. TLV and perfluorocarbons may directly and independently reduce the congestive lesions and pulmonary infection. This shows that TLV could be well tolerated with such sophisticated and dedicated liquid ventilator compared with previous studies using partial liquid ventilation which were associated with lung injury (21). This new ventilator could therefore open promising perspectives for the applications of TLV and for the future management of cardiac arrest.
Our study presents, however, several limitations. As previously discussed, we only investigated some time points for biochemical, ROS, and imaging studies. Other experiments with additional time points could be highly relevant to better understand the kinetic of alterations. It could also be relevant to investigate normothermic TLV or hypothermic TLV with a different schedule of institution (e.g., delayed initiation and/or duration). However, normothermic TLV did not provide any benefit after shockable arrest (8), supporting that TLV is only protective through its cooling properties. Longer episode of hypothermia could also be tested, as currently done in the clinical setting. However, if hypothermia needs to be prolonged to provide benefits after experimental stroke (> 12 hr) (40), it is well known that shorter episodes can be maximally protective after cardiac arrest (7,41).
In conclusion, ultrafast cooling with TLV limits the postcardiac arrest syndrome after asphyxial cardiac arrest in rabbits with potent neurologic and survival benefits. This protection involves an early and global limitation of ROS production, a rapid BBB preservation, and an ultimate mitigation of the systemic inflammatory response. This was achieved using a "latest generation" of liquid ventilator providing TLV safely and using the lungs as a unique medium for therapeutic hypothermia.