Kidney Protection by Hypothermic Total Liquid Ventilation after Cardiac Arrest in Rabbits

INSTITUTION of therapeutic hypothermia has been well demonstrated to improve both survival and neurological outcome in patients resuscitated after out-of-hospital cardiac arrest.^1,2  Beyond this neuroprotective effect, it is also important to investigate the effect of hypothermia on the multivisceral dysfunction and the so-called “postcardiac syndrome.”³  As example, acute kidney injury affects approximately 12% of the survivors after cardiac arrest and could worsen the prognosis.⁴  In this setting, Susantitaphong et al.⁵  recently analyzed clinical studies reporting kidney-related outcomes and demonstrated that therapeutic hypothermia prevented neither the development of acute kidney injury nor dialysis requirement. In animal models of cardiac arrest, the benefit afforded by hypothermia however directly depends upon its rapidity of institution after cardiopulmonary resuscitation.⁶  This benefit was investigated regarding cardiac and neurological outcomes,^6–10  whereas renal function was not precisely investigated. Therefore, we hypothesized that hypothermia can also exert a renal-protective effect after cardiac arrest if applied very rapidly after the no-flow episode.

An original strategy providing ultrafast cooling is total liquid ventilation (TLV) of the lungs with temperature-controlled perfluorocarbons.¹¹  TLV can indeed use the lung as a heat exchanger and cool the body while maintaining gas exchanges.^10,12–14  As compared with conventional cooling with combined cold blankets and fluid administration, TLV provided a potent protection on brain and heart in rabbits submitted to equal or less than 10 min of cardiac arrest.^10,15  In this study, we propose to use more severe experimental conditions inducing kidney dysfunction after 15 min of cardiac arrest in rabbits. We hypothesized that ultrafast cooling with TLV can increase kidney resistance to the postcardiac arrest syndrome. Our endpoints were kidney function biomarkers, morphological appearance, and transcriptomic responses.

The experiments were conducted in accordance with French official regulations, after approval by the institutional Animal Care Committee (ComEth “Anses/ENVA/UPEC” n°16, Maisons-Alfort; protocol 13/12/11–5). It conformed with the guidelines laid out in the Guide for the Care and Use of Laboratory Animals from the National Academy of Science.

New Zealand rabbits (3.0–3.5 kg) were anesthetized with zolazepam, tiletamine, and pentobarbital (all 20–30 mg/kg i.v.). They were intubated and mechanically ventilated (Fio₂ = 100%). After administration of pancuronium bromide (200 µg/kg i.v.), two electrodes were implanted upon the chest and inserted into the esophagus for subsequent induction of ventricular fibrillation. Rectal, esophageal, and tympanic temperatures were continuously monitored by using thermal probes (Harvard Apparatus, Paris, France). Throughout the protocol, external electrocardiogram was recorded, as well as arterial blood pressure from a catheter implanted into the ear artery. Data were digitalized and analyzed by using the data acquisition software HEM v3.5 (Notocord, Croissy-sur-Seine, France).

As illustrated in figure 1, the animals were randomly assigned after a period of stabilization to the Sham group or two groups submitted to cardiac arrest (control and TLV groups). In these two groups, cardiac arrest was induced by ventricular fibrillation by passing an alternating current (10 V, 4 mA) between the implanted electrodes. After 15 min of untreated fibrillation, cardiopulmonary resuscitation was started by using cardiac massage (approximately 200 compressions per minute), electric attempts of defibrillation (5–10 J/kg), and intravenous administration of epinephrine (15 µg/kg i.v.). After resumption of spontaneous circulation, administration of epinephrine was still permitted with an infusion pump to maintain mean arterial pressure at approximately 80 mmHg. In the TLV group, the animals were cooled to 32°C by using TLV after resumption of spontaneous circulation. The lungs were filled with 10 ml/kg of perfluorocarbon (Fluorinert, 3M, Cergy, France), and the endotracheal tube was connected to our prototype of liquid ventilator (tidal volume = 7–10 ml/kg; respiratory rate = 6 breaths/min). The temperature of the perfluorocarbon was adjusted to maintain esophageal temperature at a target temperature of approximately 32°C. After 20 min of TLV and achievement of the hypothermic target temperature, the perfluorocarbon was evacuated from the lungs, and the endotracheal tube was again connected to a conventional mechanical ventilator. Hypothermia was maintained externally at 32°C during the subsequent entire follow-up. In all groups, the animals were followed during a total duration of 6 h after cardiac arrest. Blood samples were withdrawn at baseline, 15, 60, 180, and 360 min for the assessment of blood creatinine levels and blood gases partial pressure. Urine production as well as urinary creatinine concentration for calculation of the corresponding clearance was also measured. We also assessed blood and/or urinary osmolarity (freezing point depression osmometer; Roebling Osmometer, Burladingen, Germany) and levels of sodium, glucose, creatine phosphokinase (Hitachi/Roche Cobas; Roche Diagnostic, Meylan, France), N-acetyl-β-(d)-glucosaminidase (proximal tubule lysosomal enzyme; Roche Diagnostic, Mannheim, Germany), γ-glutamyltransferase (marker of acute renal injury; Roche Diagnostic), and β2-microglobulin (marker of proximal tubule dysfunction; Roche Diagnostic). Fractional sodium excretion was calculated from blood and urinary sodium levels. At the end of the follow-up, animals were euthanized and kidneys were sampled for electron microscopy, conventional histology, and molecular biology.

Kidneys samples were processed by transmission electron microscopy, as previously described.¹⁶  In brief, tissue sections of 1 mm³ were fixed in glutaraldehyde (3%; 2 h at 4°C), washed and postfixed in osmium tetroxide (1%; 1 h at 4°C). They were dehydrated in graded series of acetone and embedded in araldite. Ultrathin sections were cut and stained with uranyl acetate and lead citrate and were examined under an electron microscope (JEOL 1010, Tokyo, Japan). Mitochondria integrity (membrane damage and crest reduction), cellular edema, loss of brush border, cellular vacuolization, and lyses of intracellular organelles were evaluated. The degree of histological lesions was determined in the cortex and medulla by using a semi-quantitative graded scale from 0 to 10 according to lesion extension among the kidney samples (0, no alterations; 1, mild lesions <10% of the kidney; 2, lesions affecting 11–20%; 3, 21–30%; 4, 31–40%; 5, 41–50%; 6, 51–60%; 7, 61–70%; 8, 71–80%; 9, 81–90%; and 10, >91%). Scores were blindly attributed by two independent observers after examination of at least 10 different sections.

Kidneys slices were also fixed in formaldehyde (4%) for conventional histology after hematoxylin–eosin–saffron staining. We used a 0 to 5 score system to blindly quantify the severity of the lesions in the cortex and the medulla (0, normal appearance; 5, extensive necrosis). For each animal, the sum of these two scores (cortex and medulla) led to an overall score from 0 to 10. Detection of macrophages was performed on paraffin tissue sections by using the RAM11 antibody against rabbit macrophages (Dako, Trappes, France), as previously described.¹⁷  The brush-border integrity was also evaluated by immunochemistry staining by using the CD10 antibody (Dako).

In all animals, kidney samples were fixed immediately after organ removal by using liquid nitrogen. For cortical tissue RNA extraction, we used a commercial kit (Macherey Nagel, Hoerdt, France). Genomic DNA was removed by using DNA-free kit (Applied Biosystems, Saint Aubin, France), and first-strand reverse transcription (Applied Biosystems) was performed. Real-time polymerase chain reaction assays were performed on a RotorGene Q (Qiagen, Courtaboeuf, France) following the manufacturer’s recommendations. Rabbit DNA primers were designed using OligoPerfect™ (Invitrogen, Carlsbad, NM), QuantPrim (Universität Potsdam, Max-Planck-Gesellschaft), and OligoAnalyzer (Integrated DNA Technologies, Coralville, IA), with the sequences detailed in Supplemental Digital Content 1, https://links.lww.com/ALN/B5. Finally, expression level of messenger RNA, relative to expression in healthy kidneys, was quantified with the Pfaffl method (expressed as relative fold change), using ribosomal L19, β-actine, and ribosomal protein large P0 as gene references.

Data were expressed as mean ± SEM. Statistical analyses were performed with the use of a statistical software (SigmaStat 3.5; Systat Software Inc., Chicago, IL). Hemodynamic and biochemical parameters were compared between groups by using a two-way ANOVA for repeated measures. Post hoc analyses were performed between groups at each time point using a Student t test with Bonferroni correction (two-tailed). Values were not compared between the different time points to avoid multiple comparisons. Histological scores and molecular biology markers were compared between groups by using a Kruskall–Wallis nonparametric test. Significant differences were determined at a P value of 0.05 or less.

Twenty-four animals were included in the different groups (n = 8 per group), with no missing data. In control and TLV groups, resumption of spontaneous circulation was obtained in 4.1 ± 0.7 and 3.7 ± 0.5 min after cardiac arrest, respectively.

As illustrated in figure 1B, a mild and reversible decrease in esophageal and rectal temperatures was observed in the control group after cardiac arrest. In comparison, temperatures decreased very rapidly in the TLV group and achieved 32°C within 5 min after the onset of TLV in the esophagus. As shown in table 1, this was associated with a strong decrease in heart rate in the TLV group as compared with control and Sham groups (e.g., −32% in TLV vs. control groups at t = 360 min after cardiac arrest). Only minor changes were observed regarding mean blood pressure as our goal was to support values of approximately 80 mmHg using epinephrine infusion. The total doses administered to achieve this goal were significantly higher in control versus TLV groups (990 ± 179 and 361 ± 23 µg/kg, respectively), showing favorable hemodynamic effects of hypothermia. Despite epinephrine administration, blood pressure was moreover significantly decreased in the control group at the end of the follow-up (t = 360 min after cardiac arrest) as compared with the TLV group. In these two groups, we also observed a dramatic increase in glucose and lactate blood levels, as well as acidosis and decrease in blood oxygen partial pressure (table 1). During the TLV episode (15 min after cardiac arrest), blood carbon dioxide partial pressure was also significantly higher in the TLV group as compared with control and Sham. Creatine phosphokinase blood levels were similarly increased in TLV and control versus Sham.

As shown in table 2, blood creatinine levels were significantly increased at the end of the follow-up (360 min) in both groups submitted to cardiac arrest as compared with the Sham group. They were significantly higher in the control group as compared with TLV. Creatinine clearance was also significantly reduced in both control and TLV groups when compared with Sham. This reduction tended to be more important in the control versus TLV group (+50%), but this did not achieve statistical significance. Total urine output also nonsignificantly decreased in control as compared with Sham and TLV groups. As shown in table 3, the beneficial effect of TLV on tubule function was evidenced by preserved fractional sodium excretion and urine concentration capacity. We also observed a limited glucose urinary excretion in TLV as compared with control animals, despite similar blood glucose levels (table 1). The urinary concentrations of selective markers of tubular damages, that is, N-acetyl-β-(d)-glucosaminidase, β2-microglobulin, and γ-glutamyl transferase, were also significantly decreased in TLV versus control despite not completely normalized as compared with Sham animals.

These functional alterations were supported by kidney lesions in animals submitted to cardiac arrest as compared with Sham animals (figs. 2 and 3). As shown in figure 2, electron microscopy revealed altered microvilli (brush border) and loss in cytosolic and mitochondrial crest density in the cortex and medulla in the control group (fig. 2, C and G). In comparison, appearance of cortex and medulla was better preserved in the TLV group (fig. 2, D and H). In Sham animals, the appearance was normal at electron microscopy (fig. 2, B and F). This led to a significant decrease in lesion score in TLV versus control group with a significant increase for both groups when compared with Sham (fig. 2A). The glomerular apparatus was preserved in all groups (fig. 2E). These differences were confirmed using conventional histology (fig. 3A). The injuries were indeed particularly marked in the cortex of the control group as illustrated by a tubular necrosis in figure 3C. In the TLV group, lesions were attenuated with “only” dilation of the proximal tubes (fig. 3D). Virtually no macrophage infiltrations were detected by using the RAM11 antibody (fig. 3, E–H), even in territories with extensive necrosis in the control group (fig. 3F). Figure 3G illustrates one of the rare macrophages in a territory with a normal appearance in this same group. Immunochemistry marking of the brush-border membrane showed extensive degradation in control animals (fig. 3J) as compared with Sham animals (3I). These alterations were not prevented in all animals in the TLV group. Some kidneys indeed showed normal appearance (fig. 3K) whereas others had clear brush-border membrane alteration (fig. 3L).

As shown in figure 4A, we observed a dramatic increase in the expression of hypoxia markers in both control and TLV groups as compared with Sham (heme oxygenase-1, erythropoïetin, hypoxia-inducible factor-1α, and vascular endothelial growth factor). The expression of the apoptotic marker Fas and the mobility marker RhoA was also similarly expressed in TLV and control groups (fig. 4B). As illustrated in figure 4, C and D, the expression of endothelial activation and innate immunity markers was also not different between these two groups, including E-selectin, vascular cell adhesion molecule-1, interleukin-10, interleukin-1β, interferon-γ, tumor necrosis factor-α, interleukin-18, monocyte chemoattractant protein-1, and toll-like receptors 2 and 4.

In the current study, we demonstrate that ultrafast cooling induced by TLV protects kidneys in a severe model of cardiac arrest in rabbits. This was strongly supported by improved renal function and preserved morphology by using electron microscopy and conventional histology. Transcriptomic profiles were not much affected by TLV regarding numerous genes involved in innate immunity and hypoxic responses.

We previously showed that ultrafast hypothermic TLV can strongly prevent both cardiovascular and neurological dysfunctions after cardiac arrest.¹⁰  These investigations were conducted in rabbits after shorter duration of cardiac arrest comprised between 5 and 10 min.¹⁰  In these conditions, we observed dramatic cardiac and neurological dysfunctions, whereas renal dysfunction was very mild. In this study, we used a prolonged episode of 15 min of no-flow to obtain acute renal dysfunction and likely very poor neurological prognosis. TLV preserved urine output and significantly limited the increase in blood creatinine levels. The latter benefit was importantly not related to a decreased release of creatinine with hypothermia because creatine phosphokinase blood levels were similarly enhanced in TLV and control groups. Surprisingly, we did not observe any significant improvement in creatinine urinary clearance in TLV versus control. It is rather difficult to analyze this result according to the complex effect of hypothermia on urine production. After out-of-hospital cardiac arrest, Zeiner et al.¹⁸  also showed a delayed recovery in creatinine clearance with therapeutic hypothermia (32°–34°C). One could also argue that hypothermia directly decreased urine production or conversely increased the production of diluted urine as previously showed in hypothermic patients treated for acute ischemic stroke¹⁹  or with profound hypothermia.^20,21  This was not the case in our experimental conditions as hypothermia preserved urine production and osmolarity in the TLV group as compared with control. Importantly, the functional benefit afforded by TLV was also supported by a significant improvement of all markers of tubular damage, including N-acetyl-β-(d)-glucosaminidase, β2-microglobulin, and γ-glutamyl transferase. It would also be relevant to investigate original and specific biomarkers such as neutrophil gelatinase–associated lipocalin and kidney injury molecule-1.^22,23 

Morphological data definitely confirm the renal-protective effect of TLV in our study by using either conventional histology or electron microscopy. Such a beneficial effect of moderate hypothermia has also been shown in animal models of regional renal ischemia–reperfusion in rats.²⁴  In the latter study, body temperature during ischemia dramatically affected the severity of injury, again showing the importance of per-ischemic and early cooling. Intraischemic hypothermia is however difficult to achieve during cardiac arrest, especially regarding abdominal and renal temperatures. In a recent meta-analysis including 19 studies and 2,218 patients, Susantitaphong et al.⁵  showed that therapeutic hypothermia prevented neither the development of acute kidney injury nor dialysis requirement but was associated with lower mortality. The current study suggests that the benefit of hypothermia could be still evidenced in very severe experimental conditions when it is induced very rapidly.

In the current study, the severity of the ischemic insult was also supported by kidney transcriptomic alterations. As an example, hypoxic stress led to a 30- to 50-fold increase in heme oxygenase-1 expression, as well as a 10- and 6-fold increases for erythropoïetin and hypoxia-inducible factor-1α, respectively. These expressions were not different between TLV and control groups, suggesting that the renal-protective effect of TLV was not directly mediated through these pathways. However, one could speculate that the potent up-regulation of erythropoïetin and heme oxygenase-1 could be associated with a facilitation of tissue repair, as previously suggested after tubulointerstitial injury and progressive nephritis.²⁵  We also did not observe any significant difference in transcriptomic profiles of cellular mobility, endothelial activation, and innate immunity markers between control and TLV groups. A nonsignificant decrease in toll-like receptor-2 and proinflammatory tumor necrosis factor-α and interferon-γ markers was however observed in the TLV group, as well as a mild and nonsignificant up-regulation of the regulatory interleukin-10 cytokine. Interestingly, the latter cytokine was shown to mediate delayed protection afforded by remote ischemic preconditioning against myocardial ischemia–reperfusion injury.²⁶  Conversely, expressions of toll-like receptor-2 and toll-like receptor-4 are well known to mediate kidney ischemia–reperfusion damages.^27,28  The lack of differences between groups could be a consequence of the early time point of organ removal (6 h after cardiac arrest). This also suggests that the protective effect of TLV is at least in part related to other mechanisms, likely an early antinecrotic effect through ultrafast hypothermia. Improvement in the hemodynamic status could also participate in the renal-protective effect.

The main limitation of the current study was therefore probably the short duration of follow-up before organ sampling and analyses after cardiac arrest. Animals were followed during only 6 h as it was difficult to maintain animals alive for much longer duration (e.g., 24 h) according to the severity of the cardiac arrest insult (15 min of no-flow). In future studies, it could be relevant to analyze kidney function at later time points, for example, after transplantation in recipient animals. Hypothermic TLV could indeed offers quite new and promising therapeutic perspectives for uncontrolled organ donation after cardiac arrest or for controlled organ donation after brain death in initially resuscitated patients.²⁹  It was not possible to conduct transplantation experiments in rabbits in the current experimental conditions, but we currently are working on a new technology of liquid ventilation that could be used in a relevant porcine model for a definitive proof-of-concept using organ transplantation.³⁰  There is indeed a high degree of proximity between human and pig kidneys with multilobular, multipapillary architecture, whereas mice, rats, dogs, and rabbits have unilobular, unipapillary kidneys.³¹  In dogs and rodents, segmental arteries are bypassed due to the lack of multiple medullary pyramids, whereas in humans and pigs an elaborated system of interlobar and segmental arteries is present to supply the numerous kidney lobes.³⁰ 

In conclusion, ultrafast cooling with TLV is renal protective after cardiac arrest and resuscitation. Further experiments with organ transplantation might be relevant to afford a proof-of-concept in this setting.

The authors gratefully thank Virginie Ameteau, B.Sc., and Maité Jacquard, B.Sc., for their excellent technical support (Inserm, U1082, Poitiers, France).

Supported by a grant (ABYSS-R12031JJ) from the “Agence Nationale pour la Recherche” (Paris, France). The study was also supported by Inserm, Université de Poitiers, Région Poitou-Charentes, Conseil général de la Vienne, Région Ile-de-France (CODDIM), University Paris Est Créteil and CHU de Poitiers. Dr. Tissier was a recipient of a “Contrat d’Interface Inserm-ENV.” Dr. Kohlhauer was supported by a doctoral fellowship from the “Region Ile-de France.” Patent application: Drs. Tissier and Berdeaux, Method and System for Treatment of a Body of a Mammal in Cardiac Arrest. U.S. Patent application 20120226337, September 6, 2012.

The authors declare no competing interests.