The manuscript contains a post-hoc analysis of data originally obtained in related studies using our established pseudo-PEA model. 8,9 These studies were conducted in accordance with the guidelines of the National Research Council of the National Academies and with the approval of the Dartmouth College Institutional Animal Care and Use Committee. The model utilized was a variation on previously described porcine asphyxial P-EMD preparation. 5,6,10
Surgical Prep
Three cohorts (N = 38) of domestic farm raised Yorkshire swine weighing approximately 30 kg were fasted over-night with free access to water and then sedated with an intramuscular injection of ketamine (30 mg/kg). After endotracheal intubation, anesthesia was initiated and maintained with isoflurane (0.5 – 4 %) and oxygen (1 – 3 L/min). During preparation, ventilation was provided by a volume-controlled ventilator (GE Datex-Ohmeda Modulus SE, Madison, WI) with 100 % O2 (tidal volume of 15 – 20 cc/kg and ventilation rate of 8 – 15 breaths per minute) during initiation, reducing to 30% shortly thereafter. Ventilation rate and tidal volume were initially adjusted to maintain normocapnia (the end-expiratory partial pressure of CO2 between 35 – 45 mmHg) as measured continuously by a capnometer (CO2SMO, Novametrix, Wallingford, CT) placed in the airway. Arterial blood gases (I-Stat, Abbott Point of Care, Princeton, NJ) were analyzed to confirm adequate baseline ventilation. Throughout the experiment, the animals were monitored using ECG, end-tidal CO2, and arterial blood pressure. In addition, depth of anesthesia was continuously assessed.
The animals were secured in a supine position and were given normal saline at a rate of 10 ml/kg per hour through a vein to maintain a central venous pressure of ~ 5 mm Hg. Through either ultrasound guided percutaneous cannulations or surgical cut-down, micromanometer catheters with a lumen were placed into: 1) the right atrium via the femoral vein, and 2) the descending aorta through the femoral artery for pressure measurements. Cohort 1 utilized Tru-Wave Pressure Transducers (Edwards Lifesciences, Irvine, CA) and cohorts 2 & 3 utilized SPR-350 (Millar Instruments, Houston, TX). All catheters were positioned under fluoroscopic guidance, and unfractionated Heparin (100 units/kg) was given to prevent catheter clotting. Flow probes were placed around the carotid artery (3 PS probe, Transonic, Ithaca, NY) and jugular vein (2.5 PS probe, Transonic, Ithaca, NY) via a cut down procedure.
After instrumentation, baseline measurements were obtained for all variables including blood gas analyses. Analog outputs of the physiological parameters were digitized and stored in data files on a personal computer for further analysis using a 16-channel computerized data-acquisition system at a sampling rate of 1000 Hz (Powerlab 16SP, ADInstruments, Castle Hill, Australia). Raw data channels included ECG, aortic pressure, right atrial pressure, intra-cranial pressure, capnography, carotid blood flow, and jugular blood flow.
Pseudo-PEA induction and chest compression
Animals were converted to continuous intravenous anesthesia using ketamine (50 mcg/kg/min) and fentanyl (0.45mcg/kg/min), isoflurane was gradually discontinued. Depth of anesthesia was ensured via monitoring blood pressure, heart rate, and jaw tone. The IV anesthesia protocol was maintained 15 minutes to allow isoflurane washout and to establish a stable level of continuous IV anesthesia prior to initiation of the hypoxia protocol.
Arterial blood gases were measured at baseline and after each episode of pseudo-PEA (i-STAT, Abbott Point of Care, Abbott Park, IL). Measurements included: pH, pCO2, pO2, base excess, HCO3, TCO2, O2 percent saturation, Na (mmol/L), K (mmol/L), ionized calcium (iCa), glucose, hematocrit, and hemoglobin.
Once adequate anesthesia has been confirmed, the animals were paralyzed using Vecuronium (1.0mg/kg) to minimize gasping. 11 Baseline data were measured before any injury occurred. Round 1 data were measured after resuscitation from the first pseudo-PEA injury. Data were also measured after resuscitation from subsequent pseudo-PEA injuries. The number of injuries per animal are variable, but the maximum number of hypoxic episodes is 5. Animals were ventilated with a progressively hypoxic gas mixture of O2/N2. Gas concentrations were measured using an oxygen concentration analyzer (Oxygen Analyzer S-3A/II, Applied Electrochemistry, VMETEK) and the concentration of O2 was decreased until pseudo-PEA was achieved. Onset of pseudo-PEA was defined as sustained aortic systolic pressure ≤ 50mmHg recorded by the aortic catheter in the presence of an organized cardiac rhythm.
Animals were treated with mechanical chest compressions delivered at a depth of 5 cm and at a variable rate. Compressions were delivered for a total of 6 minutes. After 6 minutes of chest compressions the FiO2 was set to 100%, and chest compressions were continued until ROSC was achieved. ROSC was defined as a systolic pressure > 60 mm Hg without chest compressions. If ROSC was detected at any point during the chest compressions with ongoing hypoxia, compressions were terminated and FiO2 was set to 100%.
Post resuscitation treatment
Arterial blood gases were measured 10 minutes after ROSC in all animals and for all episodes of pseudo-PEA. Three related experiments that utilized the same method of hypoxic injury and resuscitation were performed on three separate cohorts of swine (Table 1). Animals in cohorts 1 and 2 did not receive ionized calcium. The observation that ionic calcium concentration decreases following pseudo-PEA led to the addition of Ca Gluconate to the treatment regimen post-resuscitation for the 14 animals in cohort 3. Animals in cohort 3 were administered ionized calcium in an ad hoc fashion during the recovery phase when MAP was not sufficient or stable. A bolus of Ca Gluconate would be delivered intravenously if the aortic blood pressure was decreasing after resuscitation. This meant that calcium might be given to treat blood pressures before the blood chemistries were drawn, resulting in some missing data.
In addition, epinephrine and sodium bicarbonate were also delivered to support perfusion pressures and correct post-resuscitation blood chemistry as needed. As a result, different animals received different numbers and dosages of calcium boluses. Individual animals received as many as five periods of hypoxia induced pseudo-PEA followed by resuscitation. Time of delivery of epinephrine, bicarbonate, and calcium were annotated in the physiological data file from each experiment.
Data Selection:
Of the 14 animals in cohort 3, 9 received calcium injections, for a total of 37 boluses of calcium. For each recorded calcium bolus, a data subset was exported that included time intervals before and after the bolus. Because the calcium was delivered in an un-protocolized manner, the amount of time before (or after) the bolus varied for each calcium injection. In the event that a calcium injection was preceded or followed by a different injection of calcium or injections of bicarbonate or epinephrine, the data subset started (or ended) half-way between the two injections. This was done so as to isolate the effect of the injection of calcium from the effects of other injections. One critical limitation of this method is that there was significant variation in the amount of time before and after a calcium injection. The minimum amounts of time before and after each calcium injection was 40 seconds and our analysis focused on these intervals.
Data Analysis:
Individual heartbeats were identified using a python script (Anaconda 1.9.6, Spyder version 3.1.2, Pandas version 0.22.0, Numpy version 1.14.0). Once the location of each heartbeat was identified, the mean arterial pressure (MAP), the systolic aortic pressure (AoS), and the diastolic aortic pressure (AoD) were all calculated using normal conventions. The effect of calcium on hemodynamics was clearly observable in most cases and the physiological effect could be approximated using a sigmoid shape. However, the heterogeneity of the mean blood pressures and their slope during recovery from pseudo-PEA would have obscured any analysis of the raw data. To understand the size of the pressor effect of calcium, blood pressure data were normalized such that the midpoint of observed effect occurred at a pressure of 0 mmHg and a time of 0 s. This normalization was performed by a second python script which used the zero-crossing point in the second derivative of the systolic aortic pressure, the second derivative of the coronary perfusion pressure and the second derivative of the cerebral perfusion pressure to approximate the midpoint of the sigmoid response to a calcium bolus. Once the midpoint was found, pressures were normalized by subtracting the pressure value at the midpoint and time was normalized by subtraction of the time when the midpoint occurred. Once the MAP, AoS, and AoD pressures were normalized, the mean values and standard deviations were calculated for the times -40 s < t < 40 s.
Statistical Methods:
The normality of the mean aortic pressure, systolic aortic pressure, and diastolic aortic pressure signals was verified using the Shapiro-Wilk test. The hemodynamics were compared from the time period -40 < t < -36 (before) and the time period 36 < t < 40 (after), where t represents the time normalized to the effect of IV administration of calcium. Measurements were averaged across these two 4 second periods and compared via a paired t-test (STATA v 15.1). The serum baseline, post-round 1, and post 2+ round ionic calcium concentrations were aggregated between cohorts 1 and 2. The normality of the aggregate initial iCa, post round 1 iCa, and average remaining post 2+ rounds iCa was verified using the Shapiro-Wilk test. These measurements were compared via one-way ANOVA (Excel v. 16.38).