Authors: Laurie Seltzer (aDepartment of Neurology, University of Rochester Medical Center, Rochester, NY.), Michael F. Swartz (bDepartment of Surgery, University of Rochester Medical Center, Rochester, NY.), Jennifer Kwon (aDepartment of Neurology, University of Rochester Medical Center, Rochester, NY.), James Burchfiel (aDepartment of Neurology, University of Rochester Medical Center, Rochester, NY.), Jill M. Cholette (cDepartment of Pediatrics, University of Rochester Medical Center, Rochester, NY.), Hongyue Wang (dDepartment of Statistics, University of Rochester Medical Center, Rochester, NY.), Dawn Sweeney (bDepartment of Surgery, University of Rochester Medical Center, Rochester, NY.; eDepartment of Anesthesia, University of Rochester Medical Center, Rochester, NY.), Heather R. Adams (aDepartment of Neurology, University of Rochester Medical Center, Rochester, NY.), Cecilia Meagher (cDepartment of Pediatrics, University of Rochester Medical Center, Rochester, NY.), Ron Angona (bDepartment of Surgery, University of Rochester Medical Center, Rochester, NY.), Ronnie Guillet (fDepartment of Neonatology, University of Rochester Medical Center, Rochester, NY.), George M. Alfieris (bDepartment of Surgery, University of Rochester Medical Center, Rochester, NY.)
Categories: Article, neurodevelopmental outcomes, deep hypothermic circulatory arrest, EEG
Source: The Journal of thoracic and cardiovascular surgery
Authors: Laurie Seltzer, Michael F. Swartz, Jennifer Kwon, James Burchfiel, Jill M. Cholette, Hongyue Wang, Dawn Sweeney, Heather R. Adams, Cecilia Meagher, Ron Angona, Ronnie Guillet, George M. Alfieris
Neonates with congenital heart disease are at risk for impaired neurodevelopment after cardiac surgery. We hypothesized that intraoperative EEG activity may provide insight into future neurodevelopmental outcomes.
Neonates requiring surgery had continuous intraoperative EEG and hemodynamic monitoring. The level of EEG suppression was classified as slow and continuous; moderate burst suppression; severe burst suppression; or isoelectric (no brain activity for >3 minutes). Follow-up neurodevelopmental outcomes were assessed using the Vineland Adaptive Behavior Scale II (Vineland-II).
Twenty-one neonates requiring cardiac surgery developed a slow and continuous EEG pattern after general anesthesia. Ten neonates (48%) maintained continuous brain electrical activity with moderate burst suppression as the maximum level of EEG suppression. Eleven neonates (52%) developed severe burst suppression that progressed into an isoelectric state during the deep hypothermic period required for circulatory arrest. However, the duration of this state was significantly longer than circulatory arrest times (111.1 ± 50 vs 22.3 ± 17 minutes; P <.001). At a mean follow-up at 5.6 ± 1.0 years, compared with neonates with continuous brain electrical activity, neonates who developed an isoelectric state had lower Vineland-II scores in communication. There was an inverse relationship between composite Vineland-II scores and duration of isoelectric activity (R = −0.75, P = .01). Of neonates who experienced an isoelectric state, durations of >90 minutes were associated with the lowest Vineland-II scores (125.0 ± 2.6 vs 81.1 ± 12.7; P <.01).
The duration of cortical isoelectric states seems related to neurodevelopmental outcomes. Strategies using continuous EEG monitoring to minimize isoelectric states may be useful during complex congenital heart surgery.
Neurodevelopmental delay is commonly diagnosed at 1 to 10 years after neonatal cardiac surgery,^1–4^ and likely results from a combination of pre- and intraoperative factors.^1–8^ However, despite increased awareness, the exact mechanism(s) responsible for neurodevelopmental delay remain elusive. In addition, only minimal advances have been made to improve neurodevelopmental outcomes in neonates during complex cardiac procedures.^7,8^
Although initial results of using selective antegrade regional perfusion were promising,^9^ data are conflicting as to whether this approach improves outcomes, compared with deep hypothermic circulatory arrest (DHCA).^10^ Furthermore, monitoring devices designed to identify and predict poor neurodevelopmental outcomes have not led to clear delineation of those at highest risk. Near-infrared spectroscopy, cerebral oximetry, and electroencephalograms (EEGs) have, in selected case reports, demonstrated benefit; however, that improvement has not been observed within a larger population.^11–14^
In recent intraoperative EEG monitoring, during the repair procedure for neonates who had congenital heart disease, we observed distinct EEG patterns that were primarily dependent on body temperature.^15^ We found that an isoelectric state in the brain, defined as absence of brain activity, began before, and extended beyond, the DHCA time. Therefore, the aim of the current study was to determine the impact of an intraoperative cortical brain isoelectric state on intermediate-term neurodevelopmental outcomes.
Institutional review board approval and parental consent were obtained. Neonates aged <30 days who required cardiac surgery, and were previously enrolled in a prospective observational study,^15^ were assessed for neurodevelopmental outcomes. Exclusion criteria included preoperative seizure activity, known central nervous system injury, and multiple extracardiac congenital anomalies. A limited central nervous system assessment was performed preoperatively by the treatment team, who assessed for gross neurologic dysfunction; a preoperative EEG, without sedation of the patient, was used to assess for seizures. Routine care at our institution did not include brain imaging prior to surgery, unless the clinical status suggested the presence of central nervous system dysfunction.
Gold-plated EEG electrodes were affixed to the scalp, and lead placement was in accordance with the international 10–20 system with the exception of FP1 (left frontal lobe) and FP2 (right frontal lobe), as described elsewhere.^15^ A baseline 30-minute preoperative EEG was obtained in all neonates 24 hours prior to surgery, with the patients awake and without sedation. The EEG recordings resumed upon arrival at the operating room and were made continuously for the duration of the procedure.
Analysis of EEG recordings was performed postprocedure by 2 of the investigators, under blind study conditions, and fluctuations in EEG activity were linked to events recorded from within the intraoperative record. The EEG activity was classified as (1) appropriate for normal for patient’s age; (2) slow and EEG with slower frequencies but remaining continuous; (3) having moderate burst intervals of <30 seconds between bursts; (4) having severe burst intervals of 31 to 179 seconds between bursts; or (5) no brain activity for >180 seconds (Figure 1).^15^ Intraoperative events were extracted from the intraoperative record according to a predetermined protocol, by a single individual, under conditions of study blinding to patients’ EEG data.
All neonates required cardiopulmonary bypass (CPB) for either cardiac repair or palliation. Fentanyl and pancuronium were used for anesthetic induction, and anesthesia was maintained with fentanyl and isoflurane throughout the remainder of the procedure. Arterial cannulation was aortic in all cases, except for neonates who received regional perfusion.
Regional perfusion (n = 4) was used in patients with hypoplastic left heart syndrome (HLHS) and was accomplished after the cannulation of a 3.5-mm graft sewn to the innominate artery. Next, neonates were cooled to 18 to 20°C, with flows of 20 to 30 ml/kg. In the absence of DHCA, all patients were cooled to 32 to 34°C. Cooling was achieved by lowering the blood temperature by 1°C every 1 to 2 minutes. Rewarming was initiated, typically after removal of the aortic crossclamp, and achieved via increasing the arterial blood temperature by 1°C every 3 minutes. While patients were on CPB, blood glucose was maintained between 80 and 200 g/dL (using insulin and 50% dextrose solution [D50]); PaO2 (ratio of partial pressure arterial oxygen and fraction of inspired oxygen) was maintained at >100 mm Hg; PaCO2 (partial pressure of carbon dioxide in the arterial blood) levels at 35 to 45 mm Hg; and hemoglobin >7.0 g/dL, corrected to 37°C, utilizing pH-stat cooling and alpha-stat rewarming.
In 2013, the parents of the 32 neonates who previously had intraoperative EEG recordings^15^ were contacted and asked to participate in the assessment of their child’s neurodevelopmental outcome. The Vineland Adaptive Behavior Scale, second edition (Vineland-II) was used to evaluate attainment of neurodevelopmental milestones. The Vineland-II is a reliable, well-validated omnibus measure of adaptive function that evaluates adaptive skills in multiple domains, including communication, motor function, activities of daily living, social skills, and maladaptive behaviors.^16,17^ The Vineland-II, which is appropriate for use across the lifespan, was chosen because of the age of the participants, and because it allows for evaluation of multiple areas of development.^16,17^
Parents who agreed to participate in this aspect of the study were mailed the Vineland-II questionnaire and asked to complete and return it. A post-completion telephone interview was performed by 1 of the investigators, under blind conditions, who was trained in the administration of the Vineland-II and could thereby ensure the accuracy and completeness of parental responses. The questionnaires were scored using standard software^18^ that generated an age-adjusted score—reflecting global performance in relation to same-age peers—and an individual score for each functional domain (communication, activities of daily living, social skills, and motor skills). In addition, z-values for the composite Vineland-II score (summarizing global adaptive skills across all areas assessed), as well as from individual domains, were determined to demonstrate the degree of neurodevelopment compared with the mean for age (z-value mean = 0; SD = 1.0).
Descriptive statistics were used to describe the study measures, such as mean, SD, median with 95th percentiles, frequency, and percentage. Continuous variables were compared between study groups, using a Mann-Whitney U test; categoric variables were compared using Pearson χ^2^ analysis or the Fisher exact test, where appropriate. The Pearson correlation coefficient was calculated to evaluate the associations of the duration of isoelectric activity, and of the DHCA, with the Vineland-II composite score. Multiple linear regression analysis was performed to test the effects of hypothermia, diagnosis, and duration of an isoelectric state on composite Vineland-II scores.
The F-test and standard diagnostic measures, such as residual plots, were used to check model adequacy and goodness-of-fit. Multicollinearity between predictor variables was assessed using multiple correlations. All statistics were completed using SPSS, version 21 (SPSS, Inc, Chicago, Ill) or GraphPad Prism (GraphPad Software, Inc). The Bonferroni correction was applied, when appropriate, to ensure that the type I error rate remained ≤5%. A power calculation was not performed, because this study was a follow-up of previously enrolled patients.
Of 32 neonates, 6 died during the follow-up period, and 2 were otherwise lost to follow-up. For 3 subjects, the parents did not complete the survey correctly, and appropriate values could not be determined from a phone interview. Therefore, these 3 children were excluded. The EEG data and subsequent neurodevelopmental outcomes were analyzed from the 21 remaining patients (Table 1). Baseline demographics demonstrated various congenital heart diseases, in which only 1 neonate with tetralogy of Fallot was cyanotic.
Baseline EEG measurements were normal for gestational age in 19 neonates, and suggested relative dysmaturity in 2 neonates (consistent with a gestational age of 2 weeks younger). With the induction of general anesthesia, the EEG changed, from normal for gestational age to slow and continuous. Moderate burst suppression was typically observed at the initiation of CPB for 15 of 21 (71.4%) neonates, and was present in all patients while they were on CPB. Ten neonates (48%) maintained continuous brain electrical activity (CBEA), and the maximum level of EEG burst suppression throughout the procedure remained moderate. All neonates who cooled to <32°C developed severe burst suppression, and a subsequent isoelectric state (Figure 1).
The mean arterial pressure, flow during CPB, and temperature at the initiation of each EEG class, are shown in Figure 2. Although at each stage, fluctuations were seen in mean arterial pressure and flow, these variables remained fairly constant despite changes in EEG suppression. In contrast, a decrease in temperature was the only variable directly associated with EEG suppression. Similarly, an increase in temperature at the conclusion of the procedure was associated with decreased EEG suppression. The onset of an isoelectric state occurred prior to DHCA in most of the neonates (10 of 11), at 23.2 ± 3.8°C.
Further, the return of EEG activity from an isoelectric state to severe burst suppression occurred at 34.0 ± 2.4°C, nearly 4°C warmer than at severe burst suppression initiation. As a consequence, isoelectric times were significantly longer than DHCA times (111.1 ± 50 vs 22 ± 17 minutes, respectively; P <.01). No differences were found in mean arterial pressure or flow on CPB, for neonates who maintained CBEA, versus those who developed an isoelectric state.
Follow-up was obtained at age 5.6 ± 1.0 years. Six children had interim events, all of which were reoperation and/or catheter-based 5 as subsequent staging for HLHS; 1 for augmentation of a transannular patch. As demonstrated in Table 2, children who were cooled, and subsequently developed an isoelectric state, had significantly lower Vineland-II scores in communication, which evaluates receptive comprehension, oral expressive language, and written language skills. The 3 children with the highest composite Vineland-II and communication scores had the shortest durations of an isoelectric state; neonates with times >90 minutes had the lowest Vineland-II scores (81.1 ± 12.7 vs 125.0 ± 2.6, respectively; P <.01).
To test the association between neurodevelopmental outcome and DHCA or an isoelectric state, linear regression was performed, and demonstrated an inverse correlation between duration of isoelectric activity and the composite Vineland-II score (Figure 3). However, within the limits of this dataset, no correlation was observed between DHCA times and the composite Vineland-II scores (R = 0.2; P = .6). Multiple regression analysis was performed to further evaluate the impact of the duration of the isoelectric state with hypothermia, and of cardiac diagnosis on the Vineland-II composite score. Multiple regression analysis demonstrated that only the duration of the isoelectric state was associated with lower Vineland-II composite scores (P = .007; 95% confidence 0.55 to 0.18); whereas temperature (P = .22; 95% confidence −3.3 to 0.8), and preoperative diagnosis (P = .6-.4; 95% confidence −24.6 to 57.4), were not significantly associated with lower Vineland-II composite scores.
In this series, neonates who required DHCA, and developed isoelectric EEGs during surgery, had lower Vineland-II scores in communication at 5.6 ± 1.0 years of follow-up. Neonates with isoelectric-state times >90 minutes had the poorest neurodevelopmental outcomes. Further, the duration of the isoelectric period on EEG, and not the duration of DHCA, inversely and independently correlated with a lower composite Vineland-II score.
The observed intraoperative EEG patterns have been reported for >30 years, but without demonstrating any clinical correlation.^14^ Previously, investigators have assessed the impact of EEG monitoring on predicting neurodevelopmental outcomes after congenital heart surgery.^11^ However, examination of the neurologic function was limited to the postoperative period and was not at an intermediate-term follow-up, the point at which others have previously observed the greatest neurodevelopmental impact.^4,19^
An isoelectric state is thought to be neuroprotective by decreasing the cerebral metabolism, thereby creating conditions that may be optimal during periods of hypoperfusion.^20^ However, the extent of the isoelectric state within these neonates was deep, profound, and continuous; in some cases, it approached 3 hours in duration. Further, to our knowledge, no benefit comes from lowering the body temperature below the point of isoelectricity, which in most cases developed at approximately 25°C. However, the question is unresolved as to whether minimization of hypothermia, and the subsequent duration of an isoelectric state, will result in improved neurodevelopmental outcomes, or exacerbate brain hypoxia at the time of the operation, potentially resulting in worse neurologic function.
Monitoring of EEG is only 1 type of intraoperative neurologic monitoring that can be performed during the repair/palliation of congenital heart disease. Unlike transcranial Doppler and cerebral oxygenation, EEG monitoring provides information about the overall cortical function of the patient.^11,14^ Transcranial Doppler and cerebral oximetry provide information only about overall blood flow to various regions of the brain and do not reflect the hypothermia-induced injury, which is likely substantial.
Previous reports have demonstrated that variables including younger gestational age, lower birth weight, and longer hospital length of stay are associated with worse neurodevelopmental outcomes.^1–3^ Unfortunately, due to the limited number of subjects, the current study is underpowered for identifying any association of an isoelectric state, these variables, and overall neurodevelopmental outcomes. The question remains of whether the duration of DHCA is an independent risk factor for poor neurodevelopmental outcomes.^21–23^
Early outcomes demonstrated no difference from the single-ventricle reconstruction trial,^4^ whereas intermediate-term data suggested that the duration of DHCA was associated with poor neurodevelopmental outcomes.^22,23^ The Boston Circulatory Arrest Study found that DCHA times >40 minutes were associated with poor neurodevelopmental outcomes at 8 years, but not at 16 years.^23^ Most of our subjects had DHCA times of <40 minutes, which may explain why DHCA was not inversely correlated with the composite Vineland-II score.
This work has several limitations. Subjects did not undergo preoperative imaging; therefore, the existence of preoperative central nervous system pathology cannot be excluded and could have confounded subsequent results. Monitoring of EEG was not coupled with other technology, such as near-infrared spectroscopy or cerebral oximetry, which would have provided additional intraoperative data. The small sample size and cohort represent a diverse group of patients who have varying cardiac anatomy and diagnosis, type of surgical intervention and subsequent operative details, and reintervention and neurodevelopmental outcomes.
These variables confound this study, and the possibility of differentiating among these differences is limited. Therefore, the correlations observed between the duration of the isoelectric state and the neurodevelopmental outcomes may be the result of these differences. Last, neurodevelopmental outcomes were assessed by survey only.
Results of this study demonstrate for the first time an inverse correlation between the duration of an isoelectric state in the brain, initiated primarily by profound hypothermia, and the development of poor neurodevelopmental outcomes. Neonates requiring surgery that resulted in an isoelectric-state duration of >90 minutes had the poorest outcomes. Future studies will investigate the use of moderate hypothermia and real-time EEG monitoring as a way to potentially moderate EEG suppression, which may lead to improved neurodevelopmental outcomes.