|Home | About | Journals | Submit | Contact Us | Français|
Cerebrovascular autoregulation after resuscitation has not been well studied in an experimental model of pediatric cardiac arrest. Furthermore, developing noninvasive methods of monitoring autoregulation using near-infrared spectroscopy (NIRS) would be clinically useful in guiding neuroprotective hemodynamic management after pediatric cardiac arrest. We tested the hypotheses that the lower limit of autoregulation (LLA) would shift to a higher arterial blood pressure between 1 and 2 days of recovery after cardiac arrest and that the LLA would be detected by NIRS-derived indices of autoregulation in a swine model of pediatric cardiac arrest. We also tested the hypothesis that autoregulation with hypertension would be impaired after cardiac arrest.
Data on LLA were obtained from neonatal piglets that had undergone hypoxic-asphyxic cardiac arrest and recovery for 1 day (n=8) or 2 days (n=8), or that had undergone sham surgery with 2 days of recovery (n=8). Autoregulation with hypertension was examined in a separate cohort of piglets that underwent hypoxic-asphyxic cardiac arrest (n=5) or sham surgery (n=5) with 2 days of recovery. After the recovery period, piglets were reanesthetized, and autoregulation was monitored by standard laser-Doppler flowmetry and autoregulation indices derived from NIRS (the cerebral oximetry [COx] and hemoglobin volume [HVx] indices). The LLA was determined by decreasing blood pressure through inflation of a balloon catheter in the inferior vena cava. Autoregulation during hypertension was evaluated by inflation of an aortic balloon catheter.
The LLAs were similar between sham-operated piglets and piglets that recovered for 1 or 2 days after arrest. The NIRS-derived indices accurately detected the LLA determined by laser-Doppler flowmetry. The area under the curve of the receiver operator characteristic curve for cerebral oximetry index was 0.91 at 1 day and 0.92 at 2 days after arrest. The area under the curve for hemoglobin volume index was 0.92 and 0.89 at the respective time points. During induced hypertension, the static rate of autoregulation, defined as the percent change in cerebrovascular resistance divided by the percent change in cerebral perfusion pressure, was not different between postarrest and sham-operated piglets. At 2 days recovery from arrest, piglets exhibited neurobehavioral deficits and histologic neuronal injury.
In a swine model of pediatric hypoxic-asphyxic cardiac arrest with confirmed brain damage, the LLA did not differ 1 and 2 days after resuscitation. The NIRS-derived indices accurately detected the LLA compared to laser-Doppler flow measurements at those time points. Autoregulation remained functional during hypertension.
After global hypoxic brain injury from cardiac arrest, the recovering brain remains vulnerable to secondary injury. The brain is protected from ischemic injury during hypotension or hypertension by cerebrovascular autoregulation, the physiologic mechanism that maintains relatively constant cerebral blood flow (CBF) across changes in cerebral perfusion pressure (CPP). Traditional methods of continuous autoregulation monitoring were derived from invasive intracranial pressure (ICP) monitors or transcranial Doppler (TCD), which are used after traumatic brain injuries (TBI).1-3 However, ICP monitoring is not routinely used in children recovering from cardiac arrest, and TCD requires technical expertise not widely available. Therefore, we developed methods that use near-infrared spectroscopy (NIRS) to monitor autoregulation continuously: the cerebral oximetry (COx) and hemoglobin volume (HVx) indices. COx is calculated by a moving, linear correlation coefficient that represents the relationship between cerebral oximetry and arterial blood pressure and is based on the assumption that changes in tissue oxygen saturation are proportional to changes in CBF over brief periods with stable cerebral metabolic rate.4-6 HVx is a moving, linear correlation coefficient between relative total tissue hemoglobin (rTHb, measured by NIRS) and blood pressure. HVx is based on the assumption that autoregulatory vasodilation and vasoconstriction produce changes in cerebral blood volume that are proportional to changes in rTHb.6 Compared to laser-Doppler flowmetry, both COx and HVx accurately identify the lower limit of autoregulation (LLA) in swine models of hypotension.4-6
The impact of pediatric cardiac arrest and resuscitation on autoregulation remains unclear, and the utility of COx and HVx monitoring after pediatric cardiac arrest has not yet been explored. Studies of adult patients7,8 and a rodent model of cardiac arrest9 indicate that cardiac arrest disrupts cerebral autoregulation and that the LLA shifts to a higher blood pressure.8 In contrast, LLA was not increased in cerebral cortex of neonatal swine at 6 hours after resuscitation from hypoxic-asphyxic cardiac arrest.10 However, most cortical neurons have not yet died at this early recovery time in this model. Here, we extended the duration of recovery to 1 and 2 days when many of the cortical neurons in primary sensorimotor cortex are undergoing cell death. We hypothesized that: 1) the LLA would shift to a higher CPP as time of recovery progressed after cardiac arrest, 2) the NIRS-derived indices, COx and HVx, would accurately detect the LLA in comparison to laser-Doppler flow (LDF) measurements of cortical red blood cell flux, and 3) autoregulation would be impaired during hypertension in a swine model of pediatric, hypoxic-asphyxic cardiac arrest.
All procedures were approved by the Animal Care and Use Committee at Johns Hopkins University and complied with the United States Public Health Service Policy on Humane Care and Use of Laboratory Animals and the Guide for the Care and Use of Laboratory Animals. Animal care was in accord with National Institutes of Health Guidelines and ensured the animals' comfort. Neonatal male piglets (3–5 days old, 1–2.5 kg) underwent sham surgery or hypoxic-asphyxic cardiac arrest with resuscitation. To evaluate the LLA, we divided piglets into 3 treatment groups: 1 day of recovery after arrest, 2 days of recovery after arrest, and 2 days of recovery after a sham procedure. To evaluate the autoregulatory response to hypertension, we divided a separate cohort of piglets into 2 treatment groups: 2 days of recovery after arrest and 2 days of recovery after a sham procedure.
Anesthesia was induced with 5% isoflurane in a 50%/50% nitrous oxide/oxygen mixture. The animals were tracheally intubated and their lungs mechanically ventilated. Femoral veins and arteries were cannulated, and anesthesia was maintained with fentanyl (10 mcg/kg + 10–20 mcg/kg/hr, IV), pancuronium (1 mg/kg/hr), 50%/50% nitrous oxide/oxygen, and 1.5% isoflurane. Upon completion of surgery, isoflurane was decreased to 0.8%. Piglets received D5½ normal saline at 10 cc/hr and cefazolin 25 mg/kg. After piglets were resuscitated from arrest or sham surgery, their anesthesia was discontinued and they recovered overnight in a cage.
One or 2 days after resuscitation or 2 days after sham surgery, anesthesia was reinduced, the animals were tracheally intubated and their lungs mechanically ventilated, and anesthesia was maintained as described. For piglets in the hypotensive cohort, a 5-French balloon catheter was placed in the femoral vein and advanced into the inferior vena cava for later inflation to induce hypotension. For piglets in the hypertensive cohort, an aortic balloon catheter was placed in the descending aorta for later inflation to induce hypertension. Three craniotomy sites were made for: (1) a ventricular catheter for ICP monitoring, (2) a 1-mm diameter LDF probe (Moor Instruments, Devon, U.K.; model DRT4; 60 Hz), and (3) an epidural temperature probe. The LDF probe was positioned in the frontoparietal cortex 5 mm from the ICP monitor, secured with rubber washers, and cemented to the skull. A NIRS probe (Covidien, Boulder, CO; light-emitting diode to shallow and deep photodiode distances of 30 mm and 40 mm, respectively) was placed over the frontoparietal cortex contralateral to the craniotomy sites. The NIRS monitor measured cerebral oximetry and rTHb continuously.
The fractional inspired oxygen concentration (FiO2) was decreased to 10% by adding nitrogen for 45 minutes. Nitrous oxide was continued during this period. Five minutes of room air (without nitrous oxide) was then supplied to briefly reoxygenate the heart. This brief reoxygenation period is required for successful cardiac resuscitation. The endotracheal tube was then occluded for 7 minutes to produce asphyxia. The piglets were resuscitated with 100% oxygen by mechanical ventilation, manual chest compressions, and epinephrine (100 mcg/kg, IV). Chest compressions were administered at a goal rate of 100 beats per minute and to attain goal mean arterial blood pressure (MAP) ≥40 mmHg. Animals that did not regain spontaneous circulation after 3 minutes of chest compressions were considered “failed to resuscitate.” After resuscitation, the FiO2 was decreased to 30% to maintain oxyhemoglobin saturation > 93%, nitrous oxide and isoflurane were restarted, and sodium bicarbonate and calcium chloride were administered for metabolic acidosis and hypocalcemia. Sham-operated piglets received the same anesthesia, surgery, and duration of anesthesia, but without arrest. Their lungs were ventilated with FiO2 30% throughout the experiment.
On days 1 or 2 after arrest or sham surgery, the MAP, ICP, LDF, and NIRS-derived indices (see below) were monitored continuously. MAP, ICP, and LDF measurements were sampled from an analog-to-digital converter at 100 Hz with ICM+ software (Cambridge University, Cambridge, UK). CPP was calculated as MAP – ICP and recorded every 10 seconds. Piglets were monitored for 4 hours with spontaneous changes in CPP. During this time, to ensure that CPP remained above the LLA before hypotension was induced, a low-dose phenylephrine infusion was administered when needed to maintain CPP ≥45 mmHg.
After the 4-hour period, we determined the LLA by decreasing MAP from baseline to near zero over a 2–3 hour span by inflating a balloon catheter in the inferior vena cava.4-6 If the piglet required phenylephrine, this infusion was stopped before inflating the balloon catheter. The autoregulation response to hypotension was examined by plotting LDF as a function of CPP. Using SigmaPlot 11.0 (Systat Software, Inc.), we dichotomized the plot to give two datasets with regression lines having the lowest combined error squared. The intersection of these two lines was defined as the LLA.4-6
The NIRS-derived autoregulation indices, COx and HVx, were simultaneously calculated. Cerebral oximetry and rTHb were synchronously sampled from the digital output of the NIRS monitor at a refresh rate of 30/min as previously described.4,6 COx was calculated from a continuous moving 5-minute window as the Pearson correlation between cerebral oximetry and CPP, and HVx was similarly calculated from the correlation of rTHb and MAP.4-6 COx and HVx are continuous measures of autoregulation that range from −1 to +1. When autoregulation is functional, the indices are near zero or negative because CPP and cerebral oximetry (for COx) and MAP and rTHB (for HVx) are either not correlated or are negatively correlated. When autoregulation is not functional, the indices become positive and approach +1 because blood pressure and cerebral oximetry or rTHb directly correlate.
On day 2 after arrest or sham surgery, the MAP, ICP, and LDF were monitored continuously for 4 hours while allowing CPP to change spontaneously, as previously described. Hypertension was then induced over 30 minutes by inflating the aortic balloon catheter and infusing phenylephrine and dopamine. The autoregulatory response to hypertension was examined by calculating the static rate of autoregulation (sROR). The sROR is equal to the percentage change in cerebrovascular resistance (CVR) divided by the percentage change in CPP (%ΔCVR / %ΔCPP; CVR = CPP/CBF). An sROR of 0 represents a passive relationship between LDF and CPP, whereas 1.0 represents perfect autoregulation.11,12 CVR was calculated by dividing CPP (as a percentage of baseline) by LDF (as a percentage of baseline; CVR = CPP/LDF). The sROR was then determined by plotting CPP against CVR and calculating the slope of the linear regression line.
To further evaluate the neurologic injury, subsequent experiments were performed in which 6 piglets underwent cardiac arrest and 2 days of recovery without cranial instrumentation or blood pressure manipulation. Five sham piglets were used for comparison. Examiners who were blinded to treatment group were available to perform neurobehavioral tests in 4 of the postarrest piglets and in 5 sham piglets. The neurobehavioral scoring system assigns points for deficits in consciousness, motor and sensory function, brainstem reflexes, and common behaviors (0 = best outcome, 154 = worst outcome). Seven primary components are evaluated by this system: level of consciousness (0–15), brainstem function (0–22), sensory responses (0–20), motor function (0–46), behavioral activities (0–16; includes psychomotor activity, appetite, vocalization, and social interactivity), spatial orientation (0–20), and excitability (0–15).13
Piglets were perfused transcardially with phosphate-buffered saline for exsanguination and then with 4% paraformaldehyde. The brains were removed, and forebrain slabs were cut into 10 μm sections, embedded in paraffin, and used for hematoxylin and eosin staining. Coronal sections were matched for anterior-posterior level as determined by anatomical regions and by using white matter and neocortical layer 6 as landmarks. Ischemic and viable neurons were then counted in cortical layer 5 of parasagittal cortex (representing primary sensorimotor cortex) in 10 non-overlapping microscopic fields at 1,000× magnification in 2 sections spaced 200 μm apart. Criteria for ischemic cytopathology included nuclear pyknosis, eosinophilic cytoplasm, and cytoplasmic vacuolation.14
Data are presented as means ± S.D. unless otherwise noted. Differences were considered significant at p ≤ 0.05. Statistical analyses were performed and graphs were generated with STATA (v11.1), SigmaPlot (v11.0), Microsoft Excel, and GraphPad Prism (v5.03, GraphPad Software, Inc.). The duration of chest compressions was compared among groups with one-way ANOVA and Newman-Keuls tests. In the hypotensive cohort, comparisons of LLA, physiologic variables, blood gas variables, hemoglobin levels, and electrolytes were compared among groups with one-way ANOVA. Comparisons of LLA between piglets that received phenylephrine and those that did not were performed with a Mann Whitney rank sum test for sham-operated piglets and a t-test for postarrest piglets. In the hypertensive cohort, data for physiologic variables, blood gas variables, hemoglobin levels, electrolytes, and sROR were compared by using t-tests. Mann-Whitney rank sum tests were used for comparisons of neurobehavioral scores and histologic neuron counts.
Average measures of LDF, COx, and HVx were binned into 5-mmHg increments of CPP for analysis. The effects of CPP and treatment group (arrest with 1 day of recovery, arrest with 2 days of recovery, or sham surgery) on each measure of autoregulation (LDF, COx, and HVx) were analyzed with two-way ANOVA. Because the LLA differs among individual piglets, we also analyzed the data using CPP relative to each animal's LLA. Subsequent multiple comparison tests (Newman-Keuls) were performed as needed. Spline regression analyses were performed to evaluate COx and HVx responses to hypotension. Receiver operator characteristic (ROC) curves were calculated to test the sensitivity and specificity of COx and HVx in detecting CPP above or below the LDF-derived LLA. Both the spline regression and ROC analyses accounted for multiple intrasubject measurements. For the ROC curves, CPP was divided into 5-mmHg bins and coded as a binary variable above or below the bin containing the LLA, and COx and HVx were analyzed as continuous variables (−1 to +1).
To examine autoregulation during hypotension, we subjected 23 piglets to cardiac arrest and 13 piglets to sham surgery. In the arrest groups, 3 piglets died during hypoxia-asphyxia or could not be resuscitated, resulting in a resuscitation rate of 87%. After arrest, 1 piglet in the 1-day recovery group died from pulmonary edema a few hours after return of spontaneous circulation (ROSC), and 1 piglet in the 2-day recovery group died while recovering overnight (before the second surgery). In the 2-day recovery arrest group, 1 piglet was excluded from analysis because the laser-Doppler probe moved, and another was excluded because the NIRS monitor malfunctioned. In the sham group, 1 piglet died during the recovery period (before the second surgery), 2 piglets died during cranial surgeries, and 1 piglet died from cardiac arrhythmias as blood pressure was decreased for autoregulation testing. One sham piglet was excluded from analysis because the laser-Doppler probe moved. Data were analyzed on 24 piglets for a sample size of 8 in each of the 3 hypotensive treatment groups.
To examine autoregulation during hypertension in a separate cohort, we subjected 8 piglets to cardiac arrest and 5 piglets to sham surgery. One animal could not be resuscitated after hypoxia-asphyxia, resulting in a resuscitation rate of 88%. Two animals died on the second day after arrest (before the second surgery). Autoregulation data were analyzed on 5 postarrest and 5 sham piglets during hypertension.
For neurobehavioral testing and histology, a separate cohort of 6 piglets underwent arrest, and 5 piglets received sham surgeries without cranial instrumentation or blood pressure manipulation. All animals in the histology and neurobehavior testing groups survived.
For piglets in the hypotension groups, the average duration of chest compressions was 1.9 ± 0.8 minutes for piglets that recovered for 1 day and 1.5 ± 0.6 minutes for piglets that recovered for 2 days. In piglets that had neurobehavioral testing and histologic examinations, postarrest piglets received chest compressions for an average of 1.25 ± 0.5 minutes (p = 0.33 among piglets for induced hypotension and piglets for neurobehavioral scoring and histology). Piglets in the hypertensive cohort required a shorter average duration of chest compressions (0.6 ± 0.2 minutes) to attain return of circulation (p = 0.014 among groups for induced hypertension, induced hypotension, and neurobehavioral scoring and histology).
Among groups of piglets that underwent induced hypotension, heart rate, MAP, and temperature were similar at baseline, during hypoxia-asphyxia, and 1 hour after ROSC (Table 1). There were no differences in pH, PaCO2, PaO2, or hemoglobin at baseline, during hypoxia-asphyxia, or 1 hour after ROSC, with the exception of baseline PaO2, which was higher in piglets destined to recover 1 day (Table 2). Before hypotension was induced on days 1 and 2 after arrest or sham surgery, heart rate, CPP, ICP, temperature, pH, PaCO2, PaO2, hemoglobin, and electrolytes were similar among groups (Table 3). On days 1 and 2 after arrest or sham surgery, 2 postarrest piglets in the 1-day recovery group, 1 postarrest piglet in the 2-day recovery group, and 3 sham piglets received phenylephrine infusion to maintain CPP ≥45 mmHg during the 4-hour monitoring period. Piglets that recovered for 2 days after arrest had slightly lower rectal and brain temperatures than did those in other groups. Sham piglets had slightly lower sodium levels than did piglets in the other groups.
The CPP at the LLA was 37 ± 8 mmHg for piglets that recovered for 1 day after arrest, 40 ± 5 mmHg for piglets that recovered for 2 days after arrest, and 41 ± 11 mmHg for sham piglets (p = 0.65). For postarrest piglets, the LLA was 35 ± 2 mmHg in piglets that received phenylephrine (n = 3) and 40 ± 7 mmHg in piglets that did not require phenylephrine (n = 13; p = 0.24). For sham-operated piglets, the LLA was 35 ± 11 mmHg in piglets that received phenylephrine (n = 3) and 45 ± 10 mmHg in piglets that did not receive phenylephrine (n = 5; p = 0.14). LDF, COx, and HVx were not different among treatment groups at each level of CPP (Figure 1; p = 1.00 for LDF; p = 0.80 for COx; and p = 0.90 for HVx). It is possible that the variability in LLA among animals prevented possible differences among groups from being detected. Hence, data were re-analyzed as a function of CPP–LLA for each animal. Even after this analysis, no differences in LDF, COx, and HVx were present among treatment groups (Figure 2; p = 1.00 for LDF; p = 0.94 for COx; and p = 0.11 for HVx).
COx and HVx increased as CPP was decreased, and the slope of the relationship with CPP was less than zero (Table 4). For piglets that recovered from arrest for 1 day, the relationship for COx was biphasic, with the slope below the LLA significantly more negative than the slope above the LLA (p = 0.014; Table 4, Figure 3). Based on the area under the ROC curves (Table 5), both COx and HVx were good predictors of whether CPP was above or below the LDF-derived LLA.
Baseline heart rate, MAP, temperature, pH, PaCO2, PaO2, and hemoglobin levels were similar among sham and arrested piglets (Tables 6 and and7).7). On the second day of recovery and before inducing hypertension, heart rate, CPP, ICP, temperature, pH, PaCO2, PaO2, hemoglobin, and electrolyte levels were also similar between arrested and sham animals (Table 8). None of the animals required phenylephrine during the 4-hour monitoring period that preceded hypertension induction. During the induction of hypertension, all animals received phenylephrine and dopamine infusions in addition to aortic balloon inflation.
In the hypertensive cohort, CPP was increased to 143 ± 11 mmHg in postarrest and 145 ± 22 mmHg in sham-operated piglets (p = 0.82). Although some postarrest piglets showed a trend toward impaired autoregulation with severe hypertension, the LDF during aortic balloon inflation was similar between groups (Figure 4A). The sROR during induced hypertension was not significantly different (p = 0.85) between postarrest and sham piglets (Figure 4B).
The average neurobehavioral score for postarrest piglets was 37 ± 6 (n = 4) and 0 ± 0 for sham piglets (n = 5; p = 0.016). All postarrest piglets regained consciousness and perambulated after tracheal extubation. At 2 days of recovery, 4 of 4 postarrest piglets had impairments in measures of consciousness, vision, light reflex, hearing, and gait; 3 of 4 had impairments in measures that assess the animal's relationship to the environment; 2 of 4 had impaired measures of olfaction and appetite; and 1 piglet had clinical seizures.
Postarrest piglets sustained robust injury to neocortex (Figure 5B). The number of viable neurons per microscopic field in layer 5 of primary sensorimotor cortex was 12 ± 2 for postarrest piglets (n = 6) and 22 ± 4 for sham piglets (n = 5; p < 0.001). When normalized by the average density of viable neurons in the sham group, the percentage of viable neurons was 46.2 ± 34.3 for postarrest piglets and 99.7 ± 0.7 for sham piglets (p < 0.001; Figure 5C).
This study demonstrates several new findings. First, in a neonatal swine model of pediatric cardiac arrest, the LLA was similar among piglets that recovered for 2 days from sham surgery and anesthesia and piglets that recovered for 1 or 2 days after resuscitation from hypoxic-asphyxic cardiac arrest. Second, the NIRS-derived indices accurately detected the LLA determined by laser-Doppler flowmetry in piglets after recovery from hypoxic-asphyxic cardiac arrest. Third, autoregulatory function during hypertension, as measured by sROR, was similar between piglets that recovered for 2 days after sham surgery and piglets that recovered for 2 days after resuscitation from hypoxic-asphyxic cardiac arrest. Although histologic injury and neurobehavioral deficits were present 2 days after resuscitation, a more severe hypoxic-asphyxic injury may be necessary to affect autoregulation.
Pediatric cardiac arrests are often due to respiratory insufficiency,15 and our model is representative of the resultant global hypoxic brain injury. The pattern of regional brain injury in this piglet model14,16 corresponds to the selective vulnerability of primary sensorimotor cortex, striatum, and thalamus seen in newborns with hypoxic-ischemic encephalopathy.17 For the autoregulation measurements, we positioned the LDF probe over vulnerable sensorimotor cortex. We focused on the first 2 days of recovery because this is a period of intense monitoring in this patient population. Because autoregulatory function differs by gender,18-20only male animals were used.
In the present study, we postulated that the LLA would shift to a higher blood pressure as time progressed after resuscitation from cardiac arrest. However, this hypothesis was not supported by our data. Sundgreen et al.8 reported an increase in the LLA in adult patients after cardiac arrest. Several patients in that study had preexisting hypertension, and chronic hypertension is thought to induce vascular remodeling and increase the LLA.21 In our study, the fact that LLA was similar in sham piglets and piglets that recovered for 1 and 2 days after cardiac arrest may reflect the absence of chronic cerebrovascular disease and hypertension in immature brain. We previously reported that the LLA shifted to a higher blood pressure in piglets with intracranial hypertension.22 In our current study, ICP was not substantially increased 1 or 2 days after arrest. It is possible that our model of arrest was not severe enough to induce sustained cerebral edema with increased ICP and an associated shift in the LLA.
Nevertheless, the model was sufficient to produce selective neuronal necrosis in approximately half of the neurons in layer 5 of primary sensorimotor cortex where LDF was measured. However, neuronal injury is selective for this area of cortex, striatum, and parts of hippocampus and thalamus.14 Frontal lobe and areas of association cortex are not as metabolically active as primary sensorimotor cortex at this stage of development and are more resistant to ischemic injury. The lack of a global neuronal injury may help to explain the lack of robust intracranial hypertension. It may also explain the lack of severe neurobehavioral deficits. Much of the patterned behavior and standard neurological assessments do not require higher cortical function, and the remaining cortical neurons may provide sufficient redundancy for basic neonatal behavior. Nonetheless, in a similar swine model of hypoxia-asphyxia, this neurologic scoring system was sufficiently sensitive for detecting improvements with specific therapies 23,24. However, because autoregulation remained intact in the present study, we were unable to correlate loss of autoregulation with neurobehavioral deficits. Moreover, the piglets regained consciousness within a few hours after resuscitation and perambulated after tracheal extubation. Clinically, patients may remain comatose after arrest for several days, and a study in adult comatose patients after resuscitation suggested that autoregulation was impaired after arrest.7 Thus, it is possible that more severe injury with widespread cerebral edema and brainstem injury could result in impaired autoregulation. We did not use a longer period of hypoxia or asphyxia in our model because this markedly reduces the ability to resuscitate the heart.
Piglets maintained autoregulation during hypertension after 2 days of recovery from cardiac arrest. In unanesthetized piglets without injury, increasing MAP to 100–120 mmHg produced increases in CBF that were attributed to increased prostanoid production.25 In our study, which used anesthetized piglets, we did not observe a consistent increase in LDF in the CPP range of 100–135 mmHg. Although the lack of a significant effect may have been due to the small sample size, we also reported intact autoregulatory function during induced hypertension 6 hours after recovery from hypoxic-asphyxic arrest.10 Perhaps anesthesia reduces the release of vasodilator prostanoids during hypertension or extends the autoregulatory upper limit by affecting baseline myogenic tone. We were unable to consistently increase CPP above 140 mmHg because heart failure occurred. To fully examine the autoregulatory response to hypertension in this model, more aggressive cardiopulmonary support, such as bypass, or a different anesthetic regimen that does not affect cardiac contractility may be needed.
Cerebral vasoconstriction during arterial hypertension relies largely on a myogenic response that is intrinsic to vascular smooth muscle. This myogenic response can be overridden by tissue hypoxia, hypercapnia, and release of metabolic vasodilators. Our results suggest that the myogenic response recovers after the asphyxic insult and that vascular smooth muscle signaling responsible for the myogenic response is relatively resistant to this level of injury. Cerebral vasodilation during arterial hypotension relies on a decrease in myogenic tone and increased release of vasodilators from surrounding cells. Although many of the neurons are dead by 2 days of recovery, the remaining neurons, glia, and endothelium apparently remain capable of contributing to vasodilation. This vasodilation, along with myogenic mechanisms, maintains autoregulation to arterial hypotension.
The NIRS-derived autoregulation indices, COx and HVx, accurately identified the LLA in comparison to LDF measurements in this study. Our previous work showed that these indices also accurately detected the LLA in a swine model of cardiac arrest and therapeutic hypothermia.10 Recent studies have shown good correlation between COx, HVx, and autoregulation measures from TCD and ICP in adult TBI and during adult cardiopulmonary bypass.26,27 COx has also been used to measure autoregulation during pediatric cardiopulmonary bypass28 and in premature infants.29 However, the robustness of COx and HVx use could not be fully evaluated in the present model because the postarrest piglets did not have altered autoregulation.
Our study has several limitations, and our findings may have limited clinical applicability. The experimental design was unbalanced in that a 1-day survival sham group was not included. Because autoregulation was intact 2 days after sham surgery and was similar to that previously observed on day 0 in sham-operated piglets that were not awakened from the initial anesthesia,10 we did not expect to find altered autoregulation 1 day after sham surgery. Therefore, to minimize the number of animals killed, we did not include a 1-day recovery sham group. Similarly, because autoregulation during hypertension remained functional 2 days after arrest in the current study and we recently reported functional autoregulation during hypertension 6 hours after arrest,10 we did not expect to see a change in autoregulation with hypertension 1 day after arrest. Therefore, again, in an effort to decrease the number of animals used, we did not examine autoregulation with hypertension after 1 day of recovery.
Furthermore, the animals received an average duration of less than 2 minutes of chest compressions before attaining ROSC, regained consciousness after arrest, and had overall low ICP. In comparison to the hypotensive cohort, piglets that had induced hypertension required shorter durations of chest compression and did not need phenylephrine during the 4-hour monitoring period before blood pressure manipulation. Because the experiments were performed sequentially, seasonal variability in litters cannot be excluded. Another consideration is that sham piglets did not receive 100% O2. Because the arrested group was exposed to 100% O2 only during the period of chest compressions (an average of less than 2 minutes) and the exact duration could not be predicted, we did not expose the sham animals to 100% O2. Moreover, the sample sizes were small, and the piglets that were monitored for autoregulatory function were not also evaluated for neurobehavioral function. Although the examiners who performed the neurobehavioral testing were blinded to sham/arrest treatment groups, the examiner who performed the histologic neuron counting was not blinded to treatment. Blinded examiners were only available for neurobehavioral assessments in 4 of the 6 postarrest piglets.
The loss of 4 piglets that could not be resuscitated and 4 piglets that did not survive the full duration after resuscitation may have produced a bias for selecting those with intact autoregulation, but the number of animals lost represents a small fraction of those included. Moreover, inability to resuscitate primarily depends on loss of coronary perfusion pressure, which may not necessarily be strongly associated with subsequent neuronal injury.
Before hypotension was induced in the hypotensive cohort, 3 postarrest and 3 sham piglets required phenylephrine to maintain CPP ≥45 mmHg to prevent potential cerebral ischemia during anesthesia. Phenylephrine, a selective α-1 vasoconstrictor, theoretically should not directly affect the cerebral vasculature and has been used in CBF autoregulation studies.11,30 However, recent studies in piglet models of TBI suggest that phenylephrine may worsen autoregulatory function in males and protect autoregulation in females.19,20 In our study of male piglets, we did not observe a difference in autoregulation after resuscitation from cardiac arrest between piglets that did and did not receive phenylephrine. We chose phenylephrine because it is commonly used clinically at our institution. Future studies are needed to examine the effects of vasopressors, gender, and autoregulation after pediatric hypoxic brain injuries. Also, it is unclear if our results in neonatal piglets apply to older children, who have more mature cortical function, higher MAP, and presumably more vascular smooth muscle mass and myogenic tone.
LDF measurements permit relative changes in red blood cell flux to be tracked in cerebral cortex during changes in CPP. A limitation is that the technique does not provide absolute flow units, which could differ after arrest. However, maintaining CBF at a new operating point for absolute CBF presumably is physiologically important for minimizing secondary ischemia. LDF also remains useful for paired analysis within an individual subject and produces less statistical noise than do discrete measurements obtained with the commonly used absolute CBF techniques.
In conclusion, autoregulatory function measured by the LLA during hypotension and by sROR during hypertension was unaltered 2 days after resuscitation from hypoxic-asphyxic cardiac arrest in this pediatric swine model. The NIRS-derived indices COx and HVx accurately detected the LLA. Future studies are needed to determine whether a more severe insult that results in comatose survivors shifts the LLA or affects autoregulation during hypertension.
We are grateful to Lee Martin, PhD, from the Johns Hopkins Department of Neuropathology for his assistance in the analysis of the neuropathology. We would like to thank Carol Thompson, MS, MBA, from the Johns Hopkins Bloomberg School of Public Health, Department of Biostatistics, for her assistance with the spline regression analysis. We are also grateful to Claire Levine for her editorial assistance.
Funding: Dr. Lee was supported by grants from the Foundation for Anesthesia Education and Research (FAER), the American Society of Critical Care Anesthesiologists, and Hospira; the American Heart Association; and the International Anesthesia Research Society. Dr. Brady was supported by the FAER foundation. Dr. Koehler was supported by NIH NS060703. Dr. Czosnyka and Dr. Smielewski were supported by National Institute of Health Research, Biomedical Research Centre, Cambridge University Hospital Foundation Trust.
Name: Jennifer K. Lee, MD
Contribution: Dr. Lee participated in the study design, conduct of the study, data analysis, and manuscript preparation.
Conflict of Interest: None
Name: Zeng-Jin Yang, MD, PhD
Contribution: Dr. Yang participated in the conduct of the study and manuscript preparation
Conflict of Interest: None
Name: Bing Wang, MD, PhD
Contribution: Dr. Wang participated in the conduct of the study and manuscript preparation
Conflict of Interest: None
Name: Abby C. Larson, BS
Contribution: Ms. Larson participated in the conduct of the study and manuscript preparation
Conflict of Interest: None
Name: Jessica L. Jamrogowicz, BS
Contribution: Ms. Jamrogowicz participated in the conduct of the study and manuscript preparation
Conflict of Interest: None
Name: Ewa Kulikowicz, MS
Contribution: Ms. Kulikowicz participated in the conduct of the study and manuscript preparation
Conflict of Interest: None
Name: Kathleen K. Kibler, BS
Contribution: Ms. Kibler participated in the conduct of the study and manuscript preparation
Conflict of Interest: None
Name: Jennifer O. Mytar, BS
Contribution: Ms. Mytar participated in the conduct of the study and manuscript preparation
Conflict of Interest: None
Name: Erin L. Carter, RN
Contribution: Ms. Carter participated in the conduct of the study and manuscript preparation
Conflict of Interest: None
Name: Hillary T. Burman, AA
Contribution: Mr. Burman participated in the conduct of the study and manuscript preparation
Conflict of Interest: None
Name: Ken M. Brady, MD
Contribution: Dr. Brady participated in the study design and manuscript preparation
Conflict of Interest: Under a licensing agreement with Somanetics, Dr. Brady is entitled to a share of fees and royalty received by The Johns Hopkins University on the monitoring technology described in this article. The terms of this arrangement are being managed by The Johns Hopkins University in accordance with its conflict of interest policies.
Name: Peter Smielewski, PhD
Contribution: Dr. Smielewski participated in the study design and manuscript preparation
Conflict of Interest: ICM+ software is licensed by the University of Cambridge, Cambridge Enterprise Ltd. Dr. Smielewski has a financial interest in part of the licensing fee.
Name: Marek Czosnyka, PhD
Contribution: Dr. Czosnyka participated in the study design and manuscript preparation
Conflict of Interest: ICM+ software is licensed by the University of Cambridge, Cambridge Enterprise Ltd. Dr. Czosnyka has a financial interest in part of the licensing fee.
Name: Raymond C. Koehler, PhD
Contribution: Dr. Koehler participated in the study design, data analysis, and manuscript preparation
Conflict of Interest: None
Name: Donald H. Shaffner, MD
Contribution: Dr. Shaffner participated in the study design, data analysis, and manuscript preparation
Conflict of Interest: None
Reprints will not be available from the authors.
Jennifer K. Lee, Johns Hopkins University, Department of Pediatric Anesthesia and Critical Care Medicine, Baltimore, Maryland.
Zeng-Jin Yang, Johns Hopkins University, Department of Pediatric Anesthesia and Critical Care Medicine, Baltimore, Maryland.
Bing Wang, Johns Hopkins University, Department of Pediatric Anesthesia and Critical Care Medicine, Baltimore, Maryland; and Chinese Materia Medica, China Academy of Chinese Medical Sciences, Beijing, China.
Abby C. Larson, Johns Hopkins University, Department of Pediatric Anesthesia and Critical Care Medicine, Baltimore, Maryland.
Jessica L. Jamrogowicz, Johns Hopkins University, Department of Pediatric Anesthesia and Critical Care Medicine, Baltimore, Maryland.
Ewa Kulikowicz, Johns Hopkins University, Department of Pediatric Anesthesia and Critical Care Medicine, Baltimore, Maryland.
Kathleen K. Kibler, Texas Children's Hospital, Department of Anesthesiology and Pediatrics, Houston, Texas.
Jennifer O. Mytar, Johns Hopkins University, Department of Pediatric Anesthesia and Critical Care Medicine, Baltimore, Maryland.
Erin L. Carter, Johns Hopkins University, Department of Pediatric Anesthesia and Critical Care Medicine, Baltimore, Maryland.
Hillary T. Burman, Johns Hopkins University, Department of Pediatric Anesthesia and Critical Care Medicine, Baltimore, Maryland.
Ken M. Brady, Texas Children's Hospital, Department of Anesthesiology and Pediatrics, Houston, Texas.
Peter Smielewski, Addenbrooke's Hospital, Academic Neurosurgery, Cambridge, United Kingdom.
Marek Czosnyka, Addenbrooke's Hospital, Academic Neurosurgery, Cambridge, United Kingdom.
Raymond C. Koehler, Johns Hopkins University, Department of Pediatric Anesthesia and Critical Care Medicine, Baltimore, Maryland.
Donald H. Shaffner, Johns Hopkins University, Department of Pediatric Anesthesia and Critical Care Medicine, Baltimore, Maryland.