CHD affects 1% of births in the United States of America. Reference Hoffman and Kaplan1 Medical and surgical advances have resulted in increased survival and life expectancy. This has been accompanied by recognition of short- and long-term neurodevelopmental consequences, some of which originate in the perioperative period. Reference Marino, Lipkin and Newburger2
For example, neonates with CHD undergoing cardiopulmonary bypass are at risk of acute symptomatic seizures, with a reported incidence between 5 and 30%. Reference Clancy, McGaurn and Wernovsky3–Reference Desnous, Lenoir and Doussau10 Acute symptomatic seizures have been associated with an increased length of hospital stay, Reference Ghosh, Philip and Patel6 in-hospital mortality, and adverse neurodevelopmental outcomes, including attention deficit hyperactivity disorder, Reference Wang, Weng and Chang11,Reference Hansen, Poole and Nguyen12 motor impairment Reference Karl, Hall and Ford9 and other developmental delays, Reference Rappaport, Wypij and Bellinger13 and autism. Reference Sigmon, Kelleman, Susi, Nylund and Oster14 While the incidence of perioperative seizures after CHD surgery has been well-described, the incidence of later epilepsy is less well-defined.
Previous studies have estimated the rate of epilepsy in patients with repaired CHD to be between 1.6 and 5.3%; Reference Ghosh, Philip and Patel6,Reference Desnous, Lenoir and Doussau10,Reference Leisner, Madsen, Ostergaard, Woo, Marino and Olsen15,Reference Massin, Astadicko and Dessy16 however, the majority of these studies relied on either billing codes or definitions of epilepsy that do not align with the current International League Against Epilepsy criteria for the diagnosis of epilepsy. Reference Fisher, Acevedo and Arzimanoglou17 We aimed to determine the incidence of epilepsy and identify risk factors in children with CHD who underwent surgical repair using CPB.
Materials and methods
Patient population
We conducted a single-centre retrospective study of children <2 years of age undergoing CHD repair using CPB between January, 2012 and December, 2013. Patients were identified from a prospective cardiac surgery database. Only patients who were followed in the outpatient clinics at our institution for at least 2 years after their first surgery were included to ensure adequate follow-up data. For patients requiring multiple surgical repairs using CPB, data from each surgery were included. An overview of the study is shown in Figure 1.
Flowchart of the Study.
Clinical variables
Demographic and cardiac data were obtained from the cardiac surgery database. A single-ventricle defect was defined based on the Society of Thoracic Surgeons (STS) – Congenital Heart Surgery Database Committee consensus. Reference Jacobs and Mayer18 The type of CHD was also classified based on the estimated procedure-specific relative risks of mortality using the STS Congenital Heart Surgery Database and the European Association for Cardiothoracic Surgery Congenital Heart Surgery Database. Reference O'Brien, Clarke and Jacobs19 As per the cardiac surgery database, an operative or procedural complication was defined as a complication occurring within 30 days after surgery or other cardiac procedure (e.g., cardiac catheterisation) whether the patient was inpatient or outpatient at the time the complication arose, or after 30 days during the same hospitalisation subsequent to the operation or other procedure. Review of the electronic medical record was used to supplement the cardiac surgery data with neurologic data.
Electrographic seizures were defined as a paroxysmal EEG change that was different from the background lasting >10 seconds, with a plausible field and evolution in morphology, frequency, and spatial distribution. Reference Chong and Hirsch20,Reference Payne, Zhao and Frndova21 Any single seizure lasting >30 minutes or recurrent seizures comprising >30 minutes of a 1-hour epoch (50% seizure burden) were considered as electrographic status epilepticus. Reference Payne, Zhao and Frndova21–Reference Pisani, Cerminara, Fusco and Sisti24 All EEGs were interpreted clinically by board-certified clinical neurophysiologists/epileptologists.
Seizures were classified as acute symptomatic or subacute symptomatic based on the time of occurrence after a possible inciting event: acute if the seizure onset was within 7 days or subacute if the onset was between 8 and 30 days after an inciting event. Reference Billinghurst, Beslow and Abend25 An inciting event was defined as any acute CNS insult (metabolic, toxic, structural, infectious, or inflammatory) preceding the onset of the seizure. Reference Beghi, Carpio and Forsgren26
Seizures that occurred >30 days after an inciting event met the criteria for epilepsy based on the International League Against Epilepsy classification, Reference Fisher, Acevedo and Arzimanoglou17 if the patient had at least two unprovoked seizures occurring greater than 24 hours apart, or one unprovoked seizure and a probability of further seizures >60%, or diagnosis of an epilepsy syndrome. A modified form of the Engel classification Reference Glass, Hong and Rogers27 was used to determine long-term outcome at the last follow-up visit; Class 0 = seizure-free off anti-seizure medications ≥6 months; Class 1 = seizure-free ≥6 months on medication or seizure-free off medication for <6 months, Class 2 = <1 seizure/month; Class 3 = 1–4 seizures/month; Class 4 = 5–30 seizures/month; Class 5 = >30 seizures/month. Engel classification could not be obtained in two patients with epilepsy due to insufficient information in the last follow-up visit note.
A neurodevelopmental diagnosis was obtained from chart review based on the last follow-up visit in the Neurology, Developmental Pediatrics, or Psychology clinic. We used diagnoses as assigned by the paediatric neurologist caring for the patient at the time of the diagnosis or, if the diagnosis was made by a developmental paediatrician or psychologist, a formal developmental scale, such as the Bayley Scale of Infant and Toddler Development, was used. Unfortunately, while the summative data provided in the medical record discusses the use of the Bayley to arrive at the diagnosis, the domains, cut-offs, and specific scores were not typically included in the electronic medical record.
Neuroimaging
Imaging was obtained at the discretion of the clinical teams caring for the patient. MRI scans were performed on either a 1.5T (Discovery MR450; GE Healthcare, Waukesha, Wisconsin or Siemens Avanto, Erlangen, Germany) or 3.0T (Discovery MR750; GE Healthcare, Waukesha, Wisconsin, USA) scanner. The MRI scans consisted of T1- and T2-weighted images, susceptibility-weighted images, diffusion-weighted images, and in some cases, MR spectroscopy. If a clinically indicated MRI could not be obtained, a CT scan was performed. All MRI and CT scans were interpreted clinically by board-certified paediatric neuroradiologists.
Data analysis and statistics
Study data were collected and managed using Research Electronic Data Capture tools hosted at the Children’s National Hospital. Reference Harris, Taylor, Thielke, Payne, Gonzalez and Conde28 Demographics, clinical characteristics, and assessment outcomes were summarised across each group and reported with median and interquartile range for continuous variables and frequency and proportion for categorical variables. Considering the relative paucity of patients with epilepsy, Mann-Whitney U-test and Fisher’s exact test were used to examine the group difference between with and without epilepsy depending on the type of variable. Due to our limited sample size in this pilot study, all the analyses are for exploratory purposes. Statistical analysis was performed using SAS for Windows version 9.4 (SAS Institute Inc., Cary, NC, USA). Two-sided test with a significance level of 0.05 was used throughout all the data analysis. This study was approved by the Institutional Review Board of Children’s National Hospital (Pro00011200). Consent was not required.
Results
Two-hundred and twenty-one patients with CHD and surgical repair using CPB were identified. Sixty-four were excluded due to loss to follow-up (n = 49), mortality (n = 14), or missing data (n = 1). A total of 157 patients were therefore included in our analysis, 5 (3.2%) of whom went on to develop epilepsy.
Clinical characteristics
Demographic and clinical characteristics are shown in Table 1. There was no difference in sex, race, or ethnicity between patients with and without epilepsy. Although equivalent numbers of patients were born at term and pre-term, on univariate analysis, patients with epilepsy had a lower birthweight and a lower gestational age than those without epilepsy. There was no difference in the number of patients who had non-cardiac congenital abnormalities, chromosomal abnormalities, or syndromes.
Clinical characteristics of patients with CHD undergoing repair using CPB with and without epilepsy
ADHD = attention-deficit hyperactivity disorder; CBP = cardiopulmonary bypass; CHD = congenital heart disease; ECMO = extracorporeal membrane oxygenation; GA = gestational age; GDD = global developmental delay; ID = intellectual disability; LVAD = left ventricular assist device; STSEACTS = Society of Thoracic Surgeons Congenital Heart Surgery Database and the European Association for Cardiothoracic Surgery Congenital Heart Surgery Database.
P-values marked with bold indicate statistically significant differences between the groups.
The most common CHD types in the cohort as a whole were ventricular septal defect (19%), Tetralogy of Fallot (17%), atrioventricular canal (9%), hypoplastic left heart syndrome (8%), and transposition of the great arteries (8%). Patients with and without epilepsy had similar rates of single-ventricle defects, with 25% of all patients having a functionally single ventricle.
Surgical parameters were comparable between the two groups, including age at the first surgery, number of cardiac catheterizations, and total CPB, cross-clamp, deep hypothermic circulatory arrest (if utilised), and cooling time. However, patients with epilepsy were more likely to have had a greater number of cardiac surgeries and to require extracorporeal membrane oxygenation (p = 0.03) or a left ventricular assist device (p = 0.03). Both groups experienced similar rates of cardiac and other systemic complications, with arrhythmia, delayed sternal closure, and chylothorax being most common. On univariate analysis, patients with epilepsy suffered a higher rate of neurologic complications, including arterial ischaemic stroke, although rates of acute or subacute symptomatic seizures, cerebral sinus venous thrombosis, intracerebral haemorrhage, and hypoxic-ischaemic encephalopathy were similar in both groups. Patients who went to develop epilepsy also spent a longer time in the hospital.
Epilepsy diagnosis
The median (IQR) age at epilepsy diagnosis was 5.4 (3–51.2) months. Four patients developed focal epilepsy, while one patient had generalised epilepsy with multiple seizure types, including myoclonic and tonic seizures. Two patients had an Engel classification of 3 and one had an Engel classification of 2 at the last follow-up visit at 5.8, 7.5, and 6.4 years of age, respectively. Supplementary Table 1 summarises the clinical characteristics of patients with epilepsy.
EEG and imaging features of patients with epilepsy
Only one patient (20%) who went on to develop epilepsy had acute symptomatic seizures; these occurred after a surgery preceded by a cardiac arrest requiring the rapid deployment of extracorporeal membrane oxygenation. This is comparable to the rate of acute symptomatic seizures in patients who did not go on to develop epilepsy (n = 11, 7.2%; p = 0.33). There was also no difference in the occurrence of subacute symptomatic seizures between groups. Three patients had electroclinical (n = 2) or electrographic-only (n = 1) status epilepticus, none of whom went on to develop epilepsy.
Ten patients underwent EEG monitoring after the onset of clinical acute symptomatic seizures, one of whom went on to develop epilepsy. One patient had a normal EEG, while seven had an abnormal background, either slow for age (n = 6) or discontinuous (n = 1). Other electrographic features in patients with acute symptomatic seizures are shown in (Supplementary Table 2). No EEG feature was associated with epilepsy.
Amongst the patients with epilepsy, four underwent neuroimaging in the perioperative period (Supplementary Table 1). Stroke, confirmed by CT, occurred in one patient who went on to develop epilepsy and was significantly associated with the diagnosis of epilepsy (20% versus 0%, p = 0.03). This occurred in the context of being maintained on extracorporeal membrane oxygenation and a left ventricular assist device for 23 and 21 days, respectively. Another patient, who had already been diagnosed with epilepsy, had a middle cerebral artery territory infarct diagnosed 4 years after his index surgery. The remaining patients with epilepsy had structural abnormalities but no evidence of acute injury on perioperative imaging.
Developmental outcomes
A developmental assessment was performed in 72 patients (45.8%) at a median age of 36.2 months (IQR 30–72.7). Developmental assessments were performed at comparable rates in patients with and without epilepsy (80% versus 44.7%, p = 0.12). While 32.3% of patients without epilepsy had a global developmental delay or intellectual disability, all patients with epilepsy who had a developmental assessment (n = 4) had a global developmental delay or intellectual disability. Table 1 summarises the neurodevelopmental outcomes of children with epilepsy.
Discussion
We found a low incidence of epilepsy in children with CHD who underwent surgical repair utilising CPB. Predictors of epilepsy included earlier gestational age, lower birth weight, a greater number of cardiac surgeries, perioperative use of mechanical circulatory support, arterial ischaemic stroke, and a prolonged hospital length of stay. Patients with epilepsy were more likely than those without epilepsy to have a global developmental delay or intellectual disability.
Our findings are in keeping with those of prior studies, which have similarly demonstrated a low rate of epilepsy in children with CHD. Reference Ghosh, Philip and Patel6,Reference Desnous, Lenoir and Doussau10,Reference Leisner, Madsen, Ostergaard, Woo, Marino and Olsen15,Reference Billett, Cowie, Gatzoulis, Vonder Muhll and Majeed29 In contrast to prior studies, we used ILAE criteria, which are considered as the standard definition of epilepsy, to define our incidence. Reference Fisher, Acevedo and Arzimanoglou17 Ghosh et al used criteria most similar to ours, defining epilepsy as recurring seizures 30 days after cardiac surgery, and found a similar rate of epilepsy. Reference Ghosh, Philip and Patel6 Desnous et al defined epilepsy as any seizure recurrence more than 21 days after surgery, Reference Desnous, Lenoir and Doussau10 while the other prior studies did not specify the criteria used to define epilepsy. Reference Leisner, Madsen, Ostergaard, Woo, Marino and Olsen15,Reference Billett, Cowie, Gatzoulis, Vonder Muhll and Majeed29
The predictors of epilepsy identified in our study are also similar to those seen in prior studies, including a need for ECMO, Reference Desnous, Lenoir and Doussau10 acute brain injury, Reference Ghosh, Philip and Patel6 and a prolonged length of hospital stay. Reference Ghosh, Philip and Patel6,Reference Desnous, Lenoir and Doussau10 Similar to our findings, Leisner et al Reference Leisner, Madsen, Ostergaard, Woo, Marino and Olsen15 found that the risk of epilepsy was highest amongst those who had multiple surgeries.
It is interesting that children with CHD have a low incidence of epilepsy given the high rate of brain injury and perioperative seizures in this population, Reference Clancy, McGaurn and Wernovsky3–Reference Helmers, Wypij and Constantinou8,Reference Newburger, Jonas and Wernovsky30,Reference Naim, Gaynor and Chen31 especially when taking into consideration the relatively high rate of epilepsy in other populations of children with brain injury. Reference Billinghurst, Beslow and Abend25,Reference Frey32,Reference Beslow, Abend and Gindville33 Several factors could explain the low incidence of epilepsy in children with CHD, with the potential role of chronic tissue hypoxia being especially intriguing. For example, Zhen et al found that in rats undergoing a hypoxic preconditioning protocol, the frequency of pilocarpine-induced seizures and neuronal apoptosis was lower when compared to a group of naïve rats. Reference Zhen, Wang and Zhou34 In addition, Xie et al showed similar effects of a high-altitude environment on the seizure threshold in a young-rat seizure model. Reference Xie, Qin and Zhang35 The specific mechanisms involved and clinical implications of chronic hypoxia in patients with CHD require further investigation.
Our understanding of neurodevelopmental disabilities related to CHD has evolved considerably since the inception of the Boston Circulatory Arrest Study. The earliest data from this study showed that children with TGA who underwent arterial switch operation using DHCA had poorer outcomes, manifest as an increase in acute seizures, Reference Newburger, Jonas and Wernovsky30 lower motor skills at 1 year of age, Reference Bellinger, Jonas and Rappaport36 and an increased rate of behavioural, speech, and language abnormalities by the age of 4, when compared to children who underwent arterial switch using low-flow CPB. Reference Bellinger, Jonas and Rappaport36 By the age of 16, patients who underwent DHCA had a higher risk of executive dysfunction, were prescribed more psychotropic medications, and had a greater need for behavioural therapies. Reference Xiong, Gonzalez and Mu37 Although several additional studies have shown that a prolonged DHCA time increases the risk of acute symptomatic seizures, Reference Gaynor, Nicolson and Jarvik5,Reference Desnous, Lenoir and Doussau10,Reference Naim, Gaynor and Chen31,Reference Clancy, McGaurn and Wernovsky38,Reference Kornilov, Sinelnikov and Soinov39 our findings suggest that the duration of DHCA and CPB is not related to the development of epilepsy. This is in keeping with the findings of Ghosh et al Reference Ghosh, Philip and Patel6 However, these results should be interpreted with caution given the small number of patients with epilepsy in our study.
Our findings further support emerging evidence that the mechanisms underlying neurodevelopmental disabilities and epilepsy in patients with CHD are multiple and cumulative. Reference Donofrio and Massaro40–Reference Wernovsky and Licht42 Factors that may impact the risk of epilepsy include prematurity, Reference Hirvonen, Ojala and Korhonen43,Reference Li, Peng and Deng44 brain malformations, Reference Glauser, Rorke, Weinberg and Clancy45,Reference Jones46 acquired pre- and peri-operative brain injury, Reference Ghosh, Philip and Patel6,Reference Golomb, Garg, Carvalho, Johnson and Williams47,Reference Laugesaar, Vaher and Lõo48 and underlying genetic syndromes. Reference Pierpont, Brueckner and Chung49,Reference Towbin and Roberts50 The disease model of epilepsy in these patients is likely to be complex and at this time is poorly understood, but is critical for the development of targeted interventions.
Ghosh et al reported that the risk of seizures is highest in the first 12 months after surgery. Reference Ghosh, Philip and Patel6 In our cohort, three out of five patients who developed epilepsy were diagnosed within 6 months after surgery. In keeping with the findings of Ghosh et al, we found that EEG abnormalities in the acute perioperative period were not associated with epilepsy. Reference Ghosh, Philip and Patel6 However, this must be interpreted with caution given that only 10 patients underwent EEG monitoring in our cohort. While querying a cohort who underwent surgery between 2012 and 2013 allowed for sufficient duration of follow-up to assess for the development of epilepsy, it does limit our ability to draw conclusions about the relationship between acute EEG abnormalities and the later development of epilepsy, given that EEG monitoring was not as prevalent then as it is now. At the current time, all neonates and young infants who undergo CPB at our institution have ∼48 hours of continuous EEG monitoring, consistent with Naim et al Reference Naim, Gaynor and Chen31 and the American Clinical Neurophysiology Society. Reference Shellhaas, Chang and Tsuchida51
Notably, all patients who developed epilepsy and had a developmental assessment had a global developmental delay or intellectual disability. As noted above, it is known that patients with CHD are at higher risk of cognitive impairment. In our study, 39.7% of our overall cohort had an intellectual disability or global developmental delay at a median of 36 months of age, similar to previous reports. Reference Gaynor, Stopp and Wypij52 The increased risk of epilepsy co-occurring with higher rates of global developmental delay and intellectual disability further supports that the development of epilepsy in children with CHD is likely multifactorial, and in some patients, may be influenced more by co-occurring conditions than by the presence of CHD itself. This conclusion comes with the caveat that the diagnosis of GDD and ID was not made based on uniformly applied criteria. The diagnosis was either assigned by the paediatric neurologist caring for the patient at the time of the diagnosis or, if the diagnosis was made by a developmental paediatrician or psychologist, a formal developmental scale, such as the Bayley, was used. Unfortunately, while the summative data provided in the medical record notes use of the Bayley to arrive at the diagnosis, the domains, cut-offs, and specific scores were not typically included in the patient’s electronic chart.
Our study is further limited by the single-centre and retrospective nature of our data. Moreover, we may not have detected a subset of patients with epilepsy given that 22.2% of our cohort was lost to follow-up. Importantly, the small number of patients with epilepsy also limited our ability to perform a multivariate analysis to better identify associations between epilepsy and potential risk factors. In addition, not all data points were available for every patient in our cohort. Finally, all data come from a single institution and may not be generalisable to other paediatric centres.
Conclusion
Epilepsy in children with CHD is uncommon. The mechanism of epileptogenesis in these patients is the result of a complex interaction of patient-specific and treatment factors, some of which may be present before surgery. Further long-term follow-up studies are needed to better delineate this risk; however, our early data can be incorporated in current comprehensive neurodevelopmental programmes to screen, identify, and treat patients with potential risk factors for epilepsy. Future directions for research on epilepsy should focus on identifying clinical, electrographic, and radiographic risk factors and understanding the temporal course of epileptogenesis in these patients.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/S1047951121003279.
Acknowledgements
None.
Financial support
This research received no specific grant from any funding agency, commercial, or not-for-profit sectors.
Conflicts of interest
None.
