Levosimendan increases brain tissue oxygen levels after cardiopulmonary resuscitation independent of cardiac function and cerebral perfusion

Prompt reperfusion is important to rescue ischemic tissue; however, the process itself presents a key pathomechanism that contributes to a poor outcome following cardiac arrest. Experimental data have suggested the use of levosimendan to limit ischemia–reperfusion injury by improving cerebral microcirculation. However, recent studies have questioned this effect. The present study aimed to investigate the influence on hemodynamic parameters, cerebral perfusion and oxygenation following cardiac arrest by ventricular fibrillation in juvenile male pigs. Following the return of spontaneous circulation (ROSC), animals were randomly assigned to levosimendan (12 µg/kg, followed by 0.3 µg/kg/min) or vehicle treatment for 6 h. Levosimendan-treated animals showed significantly higher brain PbtO2 levels. This effect was not accompanied by changes in cardiac output, preload and afterload, arterial blood pressure, or cerebral microcirculation indicating a local effect. Cerebral oxygenation is key to minimizing damage, and thus, current concepts are aimed at improving impaired cardiac output or cerebral perfusion. In the present study, we showed that NIRS does not reliably detect low PbtO2 levels and that levosimendan increases brain oxygen content. Thus, levosimendan may present a promising therapeutic approach to rescue brain tissue at risk following cardiac arrest or ischemic events such as stroke or traumatic brain injury.

4 cardiac output or cerebral perfusion pressure. In the present study we provide evidence that NIRS fails to reliably detect low brain tissue oxygen levels and that levosimendan improves brain oxygen content. Levosimendan may therefore present a promising therapeutic approach to rescue brain tissue at risk in patients after cardiac arrest or other causes of cerebral ischemia or malperfusion such as stroke or traumatic brain injury.

Background
The number of successful resuscitations after cardiac arrest (CA) is increasing over the last decade, whereas survival and discharge rates, as well as long-term outcome have not improved substantially (AHA Report 2013, Heart Disease and Stroke Statistics) [1]. This discrepancy between successful resuscitation and patient outcome therefore requires new therapeutic strategies to achieve the ultimate goal -good neurocognitive function.
Levosimendan is an inodilatator and Ca 2+ -sensitizer, clinically established for the treatment of acute heart failure [2]. Based on its combination of inotropic and vasoactive characteristics, levosimendan has become a promising agent in the post CA-care. Potential benefits during cardiopulmonary resuscitation (CPR) were addressed in several studies, suggesting increased "return of spontaneous circulation" (ROSC) rates [3], reduced neuronal injury [4], and organ ischemia/reperfusion injury [5]. These effects are attributed to known mechanisms of action: Ca 2+ -sensitization and activation of ATP-sensitive K + -channels in the vascular bed, but also activation of mitochondrial ATP-sensitive K + -channels in cardiomyocytes [6]. These regulations result in positive inotropy as well as peripheral and coronary vasodilation.
On the other hand, in isolated hippocampal mouse brain slices subjected to mechanical trauma levosimendan reduced tissue injury [7]. After 40 minutes of aortic clamping and consecutive spinal cord ischemia, levosimendan-treated rabbits showed a better neurologic outcome [8]. The results indicate a neuroprotective effect, which cannot be attributed to cardiovascular changes and suggest a direct cellular effect in damaged tissues.
The causal link between myocardial stimulation and the observed protection after cardiac arrest has not been proven yet. This study was designed to determine the significance of the cardiovascular system for the levosimendan-mediated protection in a porcine model of CA, by targeting the effects of levosimendan on global cerebral perfusion, cerebral microcirculation and systemic hemodynamic parameters, and correlating these changes with brain tissue oxygen levels and cerebral oxygen saturation.

Subjects
After approval by the Federal Animal Care Committee (Landesuntersuchungsamt Rheinland-Pfalz, protocol number 23 177-07/G 13-1-0103), 19 male pigs (body weight: 28.4 ± 3.1 kg; age: two months) were subjected to CA and resuscitation and were randomized thereafter to levosimendan or vehicle treatment. Three animals did not achieve ROSC and were excluded prior to randomization.

Cardiac arrest
Anesthesia was induced with intravenous injections of fentanyl (4 µg/kg) and propofol (3 mg/kg) and was maintained by continuous infusion of fentanyl (8 µg/kg / h) and propofol (8 mg/kg/h). A single dose of atracurium (1.5 mg/kg) was administered prior to endotracheal intubation. Volume-controlled ventilation (AVEA Care-Fusion, San Diego, CA) was conducted (tidal volume 8 mL/kg; positive endexpiratory pressure 5 cmH 2 O, FiO 2 = 0.3; inspiration to expiration ratio 1:2; and variable respiration rate to achieve an end-tidal pCO 2 < 6 kPa). Temperature was monitored continuously and maintained constant with a heating blanket.
Via ultrasound guidance five femoral vascular catheters were placed: central venous line, PiCCO cardiac output system (Pulsion Medical Systems, Feldkirchen, Germany) and three introducer sheaths for a) a pacing catheter, b) a catheter for microspheres injection placed in the left heart ventricle and c) an arterial pressure catheter placed in the thoracic descending aorta.
After baseline-measurements, CA was induced with a pacing catheter in the right heart ventricle. After ventricular fibrillation (VF), ventilation and general anesthesia were discontinued. Immediately following 7 min of VF, ventilation was started at a rate of 10 min − 1 , external chest compressions were initiated with a thumping device (LUCAS 2®, Physio-Control Inc., Lund, Sweden) at a rate of 100 min − 1 , and 0.4 U/kg vasopressin was injected. After 2 min of persisting VF a first defibrillation with 200 J was performed and additional 0.4 U/kg vasopressin was administered.
Afterwards, chest compressions were continued for 2 min, followed by a second defibrillation with 360 J. All animals included in this study achieved ROSC after the second defibrillation. ROSC was defined as presence of heart contractions and mean arterial blood pressure (MAP) above 30 mmHg. Immediately after ROSC 30 mL/kg normal saline were given, anesthesia was resumed, and the animals received either 7 0.3 µg/kg/min or the equivalent fluid volume of normal saline (VEH). Randomization and preparation of the infusion was done by a third party not involved in the experimental setting.
After ROSC, norepinephrine infusion was started at a rate of 0.3 µg/kg/min. If mean arterial blood pressure (MAP) was below 60 mmHg at 5 minutes after ROSC, an additional normal saline bolus (15 mL/kg) was administered. Target MAP was defined as 75 ± 10 mmHg. Further norepinephrine infusion was adjusted accordingly. Normal saline infusion was adjusted to 5 mL/kg/h and additional boluses were injected if global end-diastolic volume index (GEDI) dropped below 20% of baseline values (before CPR).
The investigators were blinded to the experimental groups. Apart from the continuous infusion of levosimendan or vehicle solution, both groups were treated identically (see Fig. 1).

Hemodynamic variables
An arterial thermo-dilution system (PiCCO-System) was used to determine and record blood pressure, cardiac index (CI), global end-diastolic water index (GEDI), intrathoracic blood volume index (ITBI), and systemic vascular resistance index (SVRI).

Brain hemoglobin oxygen saturation
We quantified cerebral oxygen hemoglobin saturation (rSO 2 ) with a near infrared Cerebral tissue oxygen Brain tissue oxygen (PbtO 2 ) content was determined with an ultrafast fiber-optic, aluminum-jacketed fluorescence quenching pO 2 probe presenting an uncoated ruthenium complex at the tip (Foxy-AL300, Ocean Optics, Dunedin, FL) [9].
A craniotomy (1 × 1 cm) was performed on the left hemisphere, 5 mm apart from the midline and 5 mm behind the coronal suture, allowing the insertion of the probe (diameter 0.5 mm) through and 14 mm below the dura.

Laser-doppler flowmetry (O2C)
Cerebral microcirculation was measured using the O2C system (LEA Medizintechnik, Gießen, Germany). Based on the doppler principle and light spectroscopy, the system is able to measure regional blood flow 8 mm below the surface. The probe was positioned 1 cm lateral of the Foxy-AL300 probe on the intact dura.
Cerebral perfusion by microspheres Fluorescent-labeled microspheres (15 µm diameter) were applied for measurement of regional organ blood flow [10]. They were injected via catheter into the left ventricle to be distributed throughout the body and become trapped in capillaries.
Four different colors were used in a randomized sequence. 1 mL (equaling 10 6 spheres) was injected at baseline (BL, before CPR) and 30 min, 3 h, and 6 h after ROSC. After scarification, brain and kidneys were dissected, and the microspheres were recovered. Fluorescence was determined by high-performance liquid chromatography (HPLC), allowing for a highly sensitive quantification for each tissue as described before [11].

Statistical analysis
All experiments were performed after randomization and experiments and analysis was performed by investigators blinded to group allocation. Prism 8 statistical software (GraphPad, La Jolla, CA) was used to perform the statistical analysis. Prior to analysis, we checked the test assumptions. Due to the limited power in small samples, we did not perform formal goodness-of-fit tests prior to the t test or analysis of variance (ANOVA), but instead relied on the graphical assessment of distribution characteristics. Normality was checked by inspecting the unimodality and symmetry of histograms, as well as by Q-Q plots. The equality of variances was checked by inspecting histograms and standard deviations. To evaluate group differences in repeated measurements from the same animals, repeated measures 2-way ANOVA was applied (factors: treatment and time), followed by Šidák multiple comparisons test. Comparisons between two independent groups were carried out by the Welch-t test. Values of p < 0.05 were considered significant. Correlation analysis between brain tissue oxygen level (PbtO 2 ) and cardiac index was performed using Spearman rank correlation coefficient. Data are presented as the mean and standard deviation (mean ± SD).

Experimental setting
Return of spontaneous circulation was achieved in 16 of 19 animals and all animals with ROSC survived the observation period. Before CPR, the experimental groups did not differ in respect to MAP, arterial and central venous oxygenation, lactate, total norepinephrine dose or total amount of fluid infusion. The total length of ischemia and time from VF-induction to ROSC was similar between both groups (LEVO 674 ± 24 sec, VEH 668 ± 18 sec). In all 16 animals, ROSC was achieved after 0.8 U/kg vasopressin injection and second defibrillation.
Blood pressure and heart rate MAP and heart rate were recorded throughout the experiments and were not different between both groups ( Fig. 2A, B). The vasopressor dose and normal saline volume required to maintain the target MAP were not significantly different between the groups (fluid balance: LEVO: 33 mL/kg, VEH: 30 mL/kg; norepinephrine dose: LEVO: 0.15 µg/kg/min, VEH: 0.22 µg/kg/min) (Fig. 2C, D).

Cardiac index
To determine the influence of levosimendan on post-CPR myocardial dysfunction, cardiac index (CI) was quantified by thermo-dilution technique and normalized to the body weight. As expected in animals without preexisting cardiovascular pathologies, CI was only temporarily impaired promptly after ROSC and recovered within 30 minutes to pre-arrest baseline values (LEVO: 138 mL/min/kg; VEH: 124 mL/min/kg; Fig. 2E) and without differences between both groups.

Influence on serum lactate
As a marker for global tissue ischemia, serum lactate (Fig. 2F)

Global parameters of cardiac preload and afterload
Global End-Diastolic water Index (GEDI, Fig. 3A) and the IntraThoracic Blood volume Index (ITBI, Fig. 3B) describe changes in the volume status at the end of the diastole and the estimated blood volume in the thorax. Both were used as parameters for the cardiac preload. Systemic Vascular Resistance Index (SVRI) was determined as parameter for the cardiac afterload (Fig. 3C) and to describe the influence of levosimendan on the vascular resistance. All three parameters were not influenced by the treatment.

Cerebral perfusion and cerebral oxygenation
To determine changes in cerebral perfusion, we used two independent techniques: before and 30 min, 3 h, and 6 h post ROSC total cerebral blood flow (CBF) was measured with fluorescent microspheres (Fig. 4A) and brain cortical blood flow was determined by O2C (Fig. 4B). To compare cerebral perfusion with the perfusion in other organ systems, at these time points the total blood flow in the kidney was quantified as index organ (Fig. 4C). To correct for interindividual variations, all parameters were normalized to the individual baseline levels obtained prior to CA.
Total cerebral perfusion (Fig. 4A) was stable during the experiments and not influenced by levosimendan. In contrast to the total cerebral perfusion data, regional cortical blood flow (Fig. 4B) decreased significantly to 72% of baseline (VEH) and 63% of baseline (LEVO) after ROSC and returned to baseline values at 3 h and 6 h post ROSC. In contrast to the cortical cerebral perfusion and similarly to the total brain perfusion data, kidney perfusion data did not show any differences between levosimendan and vehicle treatment (Fig. 4C).

Cerebral oxygenation
To determine changes in cerebral oxygenation, two independent techniques were used to quantify this parameter. We measured the cortical cerebral hemoglobin oxygen saturation (rSO 2 ) by near-infrared spectroscopy (NIRS, Fig. 5A) and the PbtO 2 in the frontal cortex by real-time O 2 fluorescence quenching (Fig. 5B). To avoid injury to the dura mater before CA and CPR and to prevent mechanical brain tissue injury during CPR procedure, we did not attempt to place the brain tissue probes before CPR and performed the placement of the Foxy-probe immediately after ROSC. To maintain optimal blood oxygenation, ventilation was adjusted as described above. Settings were confirmed by arterial blood gas analysis. Using these settings, German landrace pigs show typically a PbtO 2 of 39.8 mmHg [12]. rSO 2 was not altered by the treatment with levosimendan (Fig. 5A). In contrast to the NIRS readings, PbtO 2 was significantly lower at 30 minutes after ROSC. The showed pathologically low PbtO 2 (Fig. 5C).

Discussion
This is the first report that levosimendan directly improves cerebral oxygen levels after global cerebral ischemia-reperfusion injury without improving CI or improving brain tissue perfusion.
After resuscitation, a complex series of events begins during reperfusion, leading to secondary brain damage [13]. The beneficial effects of levosimendan were subject of different studies [14] showing increased ROSC-rates [3,15], and less postresuscitation myocardial dysfunction [16], brain injury [4], and kidney ischemia/reperfusion injury [17]. All these positive effects were attributed to enhanced cardiac output [14] or CBF [4] as underlying mechanism. In contrast to this assumption, levosimendan did not improve CI, hemodynamic parameters nor cortical perfusion. Our study confirms data from healthy pigs without signs of leftventricular dysfunction showing that levosimendan does not enhance cardiac function [18].
There is a growing body of evidence for a neuroprotective mechanism, which is independent of any cardio-circulatory effect. Following transient spinal ischemia, levosimendan has been shown to ameliorate neurological damage [8], as well as reperfusion injury in a rat middle cerebral artery occlusion model [19] without affecting systemic hemodynamic parameters. In an in vitro model of traumatic brain injury, levosimendan reduced secondary tissue injury [7].
We show that levosimendan improves PbtO 2 (2-fold) 30 minutes after CA without influencing perfusion on the micro-or macrocirculatory level. This uncoupling between CBF and tissue oxygenation is highly relevant, since cerebral oxygenation is considered to be directly dependent on cerebral flow. Interestingly, tissue oxygenation in the buccal mucosa was also improved by levosimendan compared to norepinephrine in a rat model of septic shock without major changes in microcirculation and general hemodynamics [20]. We are convinced that our data set is highly reliable since very exact methods were used to measure tissue oxygenation [9], and tissue perfusion, such as the gold standard fluorescence microspheres method, laser-doppler flowmetry and NIRS [21].
In summary, cardiac output, cerebral perfusion and tissue oxygenation data indicate that levosimendan most likely acts at the cellular level by e.g. improving mitochondrial function. The present results could be the missing link between in vitro and in vivo studies. A putative mechanism of action is the activation of mitoK ATP [ 22]. Levosimendan has also been shown to interact with hydrophobic targets of the respiratory chain complexes, lowering the function of the respiratory chain and possibly reducing ischemia/reperfusion injury in subsarcolemmal mitochondria [23].
The selected methods require tissue digestion to liberate the microbeads to accurately quantify perfusion. It was therefore not possible to also perform molecular and histological analysis in brain tissue. Further investigations are required to confirm the role of mitoK ATP and the respiratory chain complex for the observed influence of levosimendan on brain tissue oxygenation. In addition, healthy animals were used in this study, which do not show cardiovascular comorbidity typically found in patients with CA of cardiac origin. The present study therefore more closely resembles the clinical situation of CA due to e.g. hypoxia or hypovolemia.

Conclusion
Cerebral oxygenation is the key to minimizing neurological damage during and after cardiac arrest. We provide evidence that NIRS fails to reliably detect low brain tissue oxygen levels and that levosimendan improves parenchymal brain oxygen content. This effect was not accompanied with improved cardiac output or cerebral perfusion. Our data therefore suggest a direct levosimendan-mediated at the cellular level. Levosimendan may therefore present a promising therapeutic approach to rescue brain tissue in patients with acute critically low tissue oxygen levels after CA, stroke, or traumatic brain injury.