If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Department of Emergency Medicine and Services, University of Helsinki and Helsinki University Hospital, Helsinki, FinlandDepartment of Pathophysiology and Transplantation, University of Milan, Milan, Italy
Perfusion pressure and chest compression quality are generally considered key determinants of brain oxygenation during cardiopulmonary resuscitation (CPR) and the impact of oxygen administration is less clear. We compared ventilation with 100% and 50% oxygen during ineffective manual chest compressions and hypothesized that 100% oxygen would improve brain oxygenation.
Ventricular fibrillation (VF) was induced electrically in anaesthetized pigs and left untreated for 5 minutes, followed by randomization to ineffective manual CPR with ventilation of 50% or 100% oxygen. The first defibrillation was performed 10 minutes after induction of VF, and CPR continued with mechanical chest compressions (LUCAS2™) and defibrillation every 2 minutes until 36 minutes or return of spontaneous circulation (ROSC). Brain oxygenation was measured with near-infrared spectroscopy (rSO2) and invasive brain tissue oxygen (PbtO2) with a probe (NEUROVENT-PTO, RAUMEDIC) inserted into frontal brain tissue. Cerebral oxygenation was compared between groups with Mann-Whitney U tests and linear mixed models.
Twenty-eight pigs were included in the study: 14 subjects in each group. During ineffective chest compressions relative PbtO2 was higher in the group ventilated with 100% compared to 50% oxygen (5.2 mmHg [1.4–20.5] vs 2.2 [0.8–6.8], p = 0.001), but there was no difference in rSO2 (22% [16–28] vs 18 [15–25], p = 0.090). The use of 50% or 100% oxygen showed no difference in relative PbtO2 (p = 1.00) and rSO2 (p = 0.206) during mechanical CPR.
The use of 100% compared to 50% oxygen during ineffective manual CPR improved brain oxygenation measured invasively in brain tissue, but there was no difference in rSO2.
There are no randomised data available on optimal partial pressure of arterial oxygen (paO2), but smaller studies suggest improved rates of return of spontaneous circulation (ROSC) and survival with higher paO2.
We may assume that higher brain tissue oxygen (PbtO2) during CPR would decrease the risk of hypoxic brain injury. Invasive measurement of PbtO2 during cardiac arrest (CA) is only possible in experimental settings and if the probe is in place before CA.
Therefore, a lower FiO2 combined with high quality compressions might be sufficient and could reduce immediate post ROSC hyperoxia and mitigate reperfusion injury. Unfortunately, quality of chest compressions are poor.
Cerebral oximetry using near-infrared spectroscopy (NIRS) measures regional cerebral oxygen saturation (rSO2) and enables continuous, non-invasive, real-time monitoring of brain oxygenation even during CA when blood flow is nonpulsatile.
In a study with in-hospital cardiac arrest (IHCA) patients, higher mean rSO2 values were measured in patients treated with mechanical chest compressions, which suggests that the quality of chest compressions can be improved by using a mechanical chest compression device.
In the current experimental animal study, we compared the effect of ventilation with 50% or 100% oxygen on cerebral oxygenation measured with PbtO2 and rSO2 during ineffective manual chest compressions mimicking a real-life situation in the early phase of cardiopulmonary resuscitation. We hypothesised that the use of 100% FiO2 during ineffective chest compressions would result in better brain oxygenation.
Material and methods
We performed an experimental animal study in healthy pigs between October 2019 and April 2020. The study was conducted at the Laboratory Animal Centre of the Large Animals Unit in Viikki of the University of Helsinki and approved by the Finnish National Animal experimental board (ESAVI/15067/2018, ESAVI/35183/2019). The study results are in adherence with the Arrive Guidelines.
Twenty-eight 14–18-week-old landrace pigs of both genders, weighing between 21 and 50 kg, were included. The animals were fasted overnight but had free access to water. The animals were pre-medicated 30–45 minutes before the experiment with intra-muscular injection of ketamine (10 mg/kg) and medetomide (0.2 mg/kg). After cannulating a peripheral ear vein, an infusion on Ringer’s acetate was started. Anaesthesia was induced with intravenous (IV) boluses of fentanyl (2–4 mg/kg) and propofol (1.3–3.3 mg/kg), and maintained with a continuous propofol infusion titrated based on the clinical status of the animal. Boluses of fentanyl were administered at least every 45 minutes and before all painful procedures. A cuffed endotracheal tube was inserted into the trachea and pigs were mechanically ventilated (Servo Ventilator 900C; Siemens-Elema, Solna, Sweden) with 21% oxygen during the preparation period (Fig. 1). Minute ventilation was adjusted to correspond with end-tidal carbon dioxide (etCO2) of approximately 5–5.5 kPa. A bolus of 200–300 mL of Ringer’s acetate was given if the animal was hypotensive or tachycardic. The pigs’ body temperature was maintained at their normal level around 38–39 °C.
The internal jugular vein was cannulated using Seldinger’s technique with an introducer catheter (Arrow®, size 6Fr) for central IV access and pacemaker catheter insertion. A temporary balloon-tipped pacing wire (Edwards D97120F5, bipolar pacing catheter) was inserted into the right ventricle with correct placement confirmed by initiating pacing (Medtronic 5348 Single Chamber Temporary Pacemaker) and observing the presence of continuous ventricular pacing on the electrocardiogram (ECG). The femoral artery was surgically explored and cannulated (Avanti®+, size 6Fr), enabling invasive blood pressure measurement and arterial blood gas sampling.
Ventilation parameters were monitored using a spirometry flow (D-lite, Gas sampler and flow sensor, GE Healthcare, USA) sensor connected to a respiratory module (GE Datex-Ohmeda). Dynamic variables were measured and recorded using an AS/3 Monitor (Datex-Ohmeda AS/3, GE Healthcare, Helsinki) and stored on a computer using data collection software (iCentral and S/5 Collect®, GE Healthcare, Helsinki, Finland). Arterial blood gases were analysed with a point-of-care device (Epoc® Blood Analysis System, Siemens Healthineers, Erlangen, Germany).
Cerebral oximetry and brain tissue oxygen
Cerebral oximetry (rSO2) was monitored with NIRS (INVOS 5100C Cerebral Oximeter, Somanetics Inc., Troy, MI, USA) fixed on the left side of the forehead. Invasive brain tissue oxygen (PbtO2) was measured with a probe (NEUROVENT-PTO, RAUMEDIC, Helmbrechts, Germany). A line was drawn from the right orbit to the crossing of the frontal-parietal and midline sutures, and a burr hole was performed at the first third from the crossing. The probe was inserted through the burr hole into the brain tissue approximately 1 cm below the dura and secured with a bolt kit. A data logger (MPR2 logO Datalogger, RAUMEDIC, Helmbrechts, Germany) was used for recording and storing data.
Prior to induction of VF, baseline measurements were collected for 5 minutes, and a blood gas analysis was performed (Fig. 1). The propofol infusion was paused immediately before induction of VF with a 9 V direct electrical current through the pacemaker. Ventilation was discontinued immediately at the onset of VF, and CA was left untreated for 5 minutes. After inducing VF, the pigs were randomised with sealed opaque envelopes into two groups: ventilation an FiO2 of 50% or 100%.
After 5 minutes of VF, poor-quality manual chest compressions were initiated with a 30 to 2 compression-to-ventilation ratio (Fig. 1). Manual bag valve ventilation (LAERDAL Silicone Resuscitator, Norway) was performed with supplemental oxygen flow titrated to correspond to an (FiO2 of approximately 50% or 100% (2–15 L/min). Oxygen flow was adjusted using continuous information obtained with a sidestream flow sensor (D-lite, Gas sampler and flow sensor, GE Healthcare) between the endotracheal tube and ventilation bag, and connected to a respiratory module (GE Datex-Ohmeda). The compressions were manually performed at a rate of 80–90/min and depth of 4 cm and monitored (ZOLL® X Series® monitor/defibrillator). The first defibrillation attempt was performed after 5 minutes of poor-quality chest compressions. CPR was continued immediately after defibrillation with continuous mechanical chest compressions (LUCAS2TM Chest Compression System, Lund, Sweden) for at least 2 minutes. If a pulsatile rhythm was not achieved, CPR was continued with defibrillations (if still in a shockable rhythm) every 2 minutes and adrenaline (epinephrine) boluses 1 mg IV every 4 minutes. During mechanical CPR, the pigs were ventilated with a frequency of 10/min. CPR was continued until ROSC or for a maximum of 36 minutes from the start of CA. ROSC was defined as a sustained restoration of an organised cardiac rhythm with mean arterial pressure (MAP) of more than 50 mmHg. In case of recurrent CA after sustained ROSC, resuscitation was not resumed. After ROSC, FiO2 was maintained at 50% or 100% according to experimental group, and sedation was continued. The MAP target was >70 mmHg and maintained with an infusion of noradrenaline (norepinephrine, 0.04 mg/mL) and adrenaline boluses (0.25 mg) if needed. The pigs were euthanised with a lethal dose of potassium chloride (40 mmol) at the end of the experiment, at 60 minutes from start of CA. Arterial blood gas (ABG) samples were analysed at predefined timepoints during VF and CPR and after ROSC (Table 1).
Table 1Vital signs during baseline, cardiac arrest, cardiopulmonary resuscitation and after return spontaneous circulation. Data are presented as median (interquartile range) or n(%).
Assuming a similar difference during chest compression of poor quality with a statistical power of 0.80 and significance level of 0.05, a study with 13 animals in each group would have the sample size needed to show a similar difference. Given the risk of possible technical problems, we decided to include 28 animals. Continuous variables are reported as medians with interquartile range (IQR), and categorical values are reported as counts and percentages. Continuous variables were compared with Mann-Whitney U tests and categorical values with chi-square tests. A median baseline PbtO2 was calculated for every individual animal and the PbtO2 values during VF, CPR, and ROSC were compared to their respective baseline value in order to obtain the relative change in % for every individual. The rSO2 values and relative PbtO2 values were compared with Mann-Whitney U tests during baseline measurements, VF, and manual chest compressions. During mechanical CPR and if ROSC was achieved, a linear mixed-effects model with compound symmetry as the covariance matrix was used to compare PbtO2 and rSO2. The analysis included the effect of interventional group, the effect of time, and the interaction between time and intervention group. Median PbtO2 and rSO2 values over time were plotted in a graph, with medians and IQR. A p-value less than 0.05 was considered statistically significant. Statistical analysis was performed using IBM SPSS Statistics 188.8.131.52.
Twenty-eight pigs were included in the study: 14 cases in each group. The rSO2 values are missing for one pig in the 50% group and for one pig in the 100% group. In addition, the baseline values are missing for one pig in the 50% group. One pig in the 50% group had immediate ROSC after the first defibrillation and, in this case, CPR was only performed manually. There were no differences in pre-arrest vital signs (Table 1) or arterial blood gases (Table 3) between groups. Arterial blood gases were similar between groups after 5 minutes of VF. After initiation of VF, both rSO2 and PbtO2 decreased over time (Fig. 2, Fig. 3).
The quality of the manual compressions was comparable between the groups (Table 2). There were no differences in etCO2 or ventilation parameters between the groups during poor-quality manual CPR (Table 1). After 5 minutes of poor-quality CPR, the paO2 was significantly higher in the 100% group compared to the 50% group (p = 0.002). Extreme hyperoxia (paO2 > 40 kPa) was present in six cases in the 100% group compared to none in the 50% group (p = 0.007). There was no statistically significant difference in the paO2 values during mechanical chest compressions between groups over time (p = 0.658; Table 3). After 10 minutes of mechanical CPR, three pigs in both groups had extreme hyperoxia (p = 0.121). The etCO2 increased in both groups when more effective mechanical chest compressions were initiated (p < 0.001). The alterations in MAP and etCO2 during CPR are presented in the Supplementary data (Fig. 1, Fig. 2).
Table 2Factors at resuscitation. Data are presented as median with interquartile range.
ROSC was achieved in nine cases in the 50% group versus in eight cases in the 100% group (p = 0.699); correspondingly, the median delay to ROSC was 14 (12–19) versus 16 (12–20) minutes (p = 0.730). After ROSC, the median paO2 was higher in the 100% group, and there was a significant interaction between paO2 and time (p < 0.001). After ROSC, there was no difference in MAP, but etCO2 (4.9 kPa [4.7–5.1] vs 5.2 kPa [4.9–5.4], p < 0.001) was higher in the FiO2 100% group.
During the baseline period, rSO2 was higher in the group randomised to ventilation with 50% oxygen compared to 100% oxygen (39% [32–51] vs 36% [33–40], p = 0.013). The rSO2 values were similar between FiO2 groups during VF. There was no significant difference in rSO2 during ineffective chest compressions (18% [15–25] vs 22% [16–28], p = 0.090) and mechanical chest compressions (15% [15–28] vs 19% [15–46], p = 0.206). There was no difference in rSO2 after ROSC between groups. Alterations in rSO2 during the experiment are presented in Fig. 2.
Brain tissue oxygen
The absolute baseline PbtO2 was 21 mmHg (7–24) in the group randomised to ventilation with 50% oxygen and 12 mmHg (8–21) in the 100% group (p = 0.587). The relative PbtO2 was similar between FiO2 groups during VF. During ineffective chest compressions, the achieved relative PbtO2 was higher in the 100% group (p = 0.001). During mechanical chest compression, there was no difference in relative PbtO2 between groups. After ROSC, the relative PbtO2 increased to a markedly higher level with 100% oxygen, but without statistical significance. The changes in relative PbtO2 during the experiment are presented in Fig. 3 and absolute PbtO2 values in the Supplementary appendix Fig. 3.
In this experimental animal study, we show that the use of a FiO2 of 100% during ineffective CPR resulted in higher oxygen measured in brain tissue compared to the use of 50%. When CPR was continued with effective mechanical chest compressions, the use of 100% oxygen did not result in better brain oxygenation measured in brain tissue oxygen or rSO2. Our findings suggest that using 100% oxygen may provide better brain oxygenation when chest compressions are of poor quality.
The PbtO2 reflects oxygen dissolved in plasma that diffuses into brain tissue according to the oxygen pressure differential between capillary and the tissue.
In a study on out-of-hospital cardiac arrest (OHCA) patients where CPR was performed by paramedics and nurse anaesthetists, chest compressions were given only half of the available time, and 60% of the chest compressions were too shallow (<38 mm).
Similar results indicating that CPR is deficient from guideline recommendations was observed in IHCA patients. The compression rates were below 90/min 30% of the time and too shallow (<38 mm) 37% of the time.
In our observational prospective study on OHCA patients treated by a physician-staffed helicopter emergency medical services unit, we studied associations between etCO2, paO2, and perfusion pressure with NIRS during clinical CPR.
In that study MAP was moderately correlated with rsO2 but paO2 did not. Contrary to our hypothesis in the present study, ventilation with 100% compared to 50% increased the paO2 but did not result in higher rSO2 during ineffective chest compressions. Both rSO2 and PbtO2 measure different aspects of focal brain oxygenation and do not necessarily represent the changes occurring in the whole brain during CPR. While PbtO2 represents oxygen levels in actual brain tissue, rSO2 gives an estimate of haemoglobin saturation of the venous blood (thus located in blood vessels) in monitored area of the brain. The fact that during poor quality CPR, ventilation with 100% oxygen improved brain tissue oxygen but not rSO2 may be related to the diffusion of the soluble oxygen which was notably higher in the 100% compared to the 50% group. On the other hand, with the high paO2 values in both groups the oxygen saturation is likely to have been close to 100% and this resulted in comparable rSO2 values.
There is little knowledge about factors that influence PbtO2 during CPR. Putzer et al compared the effect of different adrenaline boluses on PbtO2 during CPR and observed that adrenaline improved hemodynamics in the short term but the overall perfusion pressure was too low to show differences in PbtO2.
After ROSC, we noted a clear increase in PbtO2 compared to baseline values which is not unexpected given the higher FiO2.
Our study has several strengths, but the results should also be interpreted in the light of some limitations. This experimental study was performed in pigs, which are considered a good large animal for modelling CA.
The study was not blinded, but all procedures were standardised. The used animals were a homogenous group of young healthy pigs under general anaesthesia prior to the procedure, which may not completely reflect real clinical situations of CA. Using NIRS in pigs may be influenced by anatomical differences such as the thickness of the skull and overlying skin.
The NIRS signal might be also be influenced by vasoconstriction of blood vessels in extracerebral tissue resulting from adrenaline administration. Since both groups received adrenaline, it seems unlikely this would explain our findings. During the baseline period, rSO2 appeared higher in the group randomised to 50% oxygen and the difference between groups is most likely to be due to chance since the randomisation happened at a later time-point.
In this experimental CPR model, ventilation with 100% compared to 50% oxygen during poor-quality chest compressions resulted in improved brain oxygenation measured invasively in brain tissue but not when monitored with NIRS. In conclusion, our results support the use of 100% oxygen during CPR if high-quality chest compressions cannot be delivered.
Conflict of Interest Statement
Markus Skrifvars reports having received a research grant from GE Healthcare, travel reimbursements, and lecture fees from BARD Medical. Other authors report no conflict of interest.
CRediT authorship contribution statement
Annika Nelskylä: Methodology, Formal analysis, Investigation, Data curation, Visualization, Writing – original draft, Writing – review & editing. Jaana Humaloja: Formal analysis, Investigation, Visualization, Writing – review & editing. Erik Litonius: Conceptualization, Methodology, Investigation, Visualization. Pirkka Pekkarinen: Conceptualization, Methodology, Investigation, Writing – review & editing. Giovanni Babini: Investigation, Writing – review & editing. Tomi P. Mäki-Aho: Investigation, Resources. Juho A. Heinonen: Investigation, Writing – review & editing. Markus B. Skrifvars: Conceptualization, Methodology, Formal analysis, Investigation, Resources, Writing – review & editing, Visualization, Supervision, Project administration, Funding acquisition.
This study was financed with funding received from Sigrid Juselius stiftelse, Finska Läkaresällskapet, Medicinska Understödsföreningen Liv och Hälsa and Stiftelsen Dorothea Olivia, Karl Walter och Jarl Walter Perklens minne. The authors also wish to thank the staff of the Laboratory Animal Centre of the University of Helsinki and Olli Valtonen for their help in the care of laboratory animals.
Appendix A. Supplementary data
The following are the Supplementary data to this article: