Recent Advances in Our Understanding of Neurodevelopmental Outcomes in Congenital Heart Disease

Abstract

Purpose of Review

Patients with congenital heart disease (CHD) suffer from a pattern of neurodevelopmental abnormalities including deficits in language and executive function. In this review, we summarize recent studies that examine these outcomes, their risk factors, possible biomarkers, and attempts to develop therapeutic interventions.

Recent Findings

The latest literature has highlighted the role of genetics in determining neurologic prognosis, as we have increased our understanding of potentially modifiable perioperative risk factors. The role of potentially neurotoxic medical therapies has become more salient. One recent focus has been how neurodevelopment affects quality of life and leads to a high prevalence of mental illness. Neuroimaging advances have provided new insights into the pathogenesis of deficits.

Summary

While many risk factors in CHD are not modifiable, there is promise for interventions to improve neurodevelopmental outcomes in patients with CHD. Biomarkers are needed to better understand the timing and prognosis of injury, and to direct therapy. Research into psychosocial interventions is urgently needed to benefit the many survivors with CHD.

Introduction

Over the past decade there is an increasing recognition of the importance of understanding the risks of poor neurodevelopmental outcomes in children with congenital heart disease (CHD). As surgical, cardiopulmonary bypass, and intensive care advances have improved survival, our focus, as a field, has progressively turned to increasing the percentage of survivors with good neurologic outcomes. The American Heart Association (AHA) published a scientific statement in 2012 [1] with recommendations for risk stratification, follow-up, and management for children with CHD. Clinical research focused on this topic has increased rapidly over the past few years. The goal of this review is to highlight contributions to our understanding of neurodevelopmental outcomes in CHD from the last eighteen months. This period has seen the publication of follow-up from some of the original groundbreaking studies of neurodevelopment in CHD as well as technological advances in the fields of neuroimaging and genetics.

Text of the Review

The cause of poor neurodevelopmental outcomes in patients with CHD is likely multifactorial, with contributions from genetics, chronic cyanosis (including in utero), medical and surgical therapies, neurovascular events, co-morbid conditions, and a lack of exposure to normal developmental stimuli in the intensive care unit (ICU) [1]. Children with CHD have lower overall intelligence quotients (IQ) than peers without CHD. In particular, deficits are seen in language [2*,3], executive function [4], and visual processing. Unfortunately, studies of modifiable perioperative strategies have failed to show significant change in the risk of poor outcomes [5,6]. The most notable studies include the single ventricle reconstruction (SVR) trial which failed to show an effect of shunt type on neurodevelopment in patients with hypoplastic left heart syndrome (HLHS) [7] and the Boston circulatory arrest study (BCAS) which showed no effect of cerebral perfusion strategy in patients with dextro-transposition of the great arteries (D-TGA) [8].

The Effect of Genetics on Neurodevelopmental Outcomes

Studies of possible genetic contributions to neurologic outcomes have found evidence that much of the neurodevelopmental impairment is an unmodifiable consequence of genetics [9]. Novel candidate genes have been found using whole exome sequencing in patients with CHD [10–12]. Notably, in the largest genetic evaluation of a single CHD cohort [13], pathogenic gene mutations had a large overlap with genes also implicated in the pathogenesis of autism. This result highlights a possible common underlying basis for both CHD and impaired neurodevelopment. When focused genetic screening is performed on patients with both CHD and neurodevelopmental disability, the yield of genetic testing is much higher than in CHD patients with normal neurodevelopment [14*], with particular enrichment of genes in the Ras pathway.

Despite these findings, the utility of whole exome sequencing in clinical practice is unknown as no gene-specific therapeutics are available. While genetic screening may allow a more comprehensive prognosis and access to earlier implementation of targeted developmental interventions, it requires more knowledge about the pathogenic role of particular genes.

Additional evidence that much of the variance in neurodevelopmental outcomes is driven by genetics is derived from patients with Down syndrome. The Down Syndrome Cognition Project performed neurodevelopmental testing on 234 patients with and without structural cardiac disease [15**]. Interestingly, no differences in neurocognitive testing were found between groups, supporting earlier single-center studies [16,17] that also showed no difference in neurodevelopment between Down syndrome patients with and without CHD. Furthermore, a population-based study in Paris (419 patients) found that patients with CHD that did not require open heart surgery (which included those needing only catheter-based interventions) had similar neurodevelopmental outcomes to those who required surgery [2*]. While these studies support the assertion that underlying genetic factors may be more important than surgical risk factors, the results do need to be interpreted in context. Down syndrome is strongly linked to delayed development and patients usually have milder forms of CHD. In the Paris study, the surgical cohort included patients with TGA and tricuspid atresia, but not HLHS. Thus, these results may not extrapolate to patients with more severe forms of single-ventricle CHD. Furthermore, all studies of long-term neurodevelopment are limited by the fact that they only recruit those patients able to come to clinic for testing and thus exclude those who have died or suffered devastating brain injury. So, in all studies, especially those with a large percentage of more mild forms of CHD may understate the prevalence of neurodevelopmental impairment.

Other Risk Factors for Altered Neurodevelopment

Recent cohorts have found that markers of the intrauterine environment (e.g., small head circumference, low birth weight), low socioeconomic status, and length of ICU stay are markers of lower cognitive outcomes [2*,18,19]. The number of subsequent rehospitalizations is a significant predictor of school performance [19]. While to some extent, these may be modifiable risk factors, they are likely indicators of underlying disease severity.

The ICU, where many infants with CHD spend considerable time, lacks typical stimuli from the traditional home environment (e.g., voices and toys) and has stimuli which may be harmful such as bright lights and loud noises. The vast majority of neonates show significant levels of stress, disorganization, and dysregulation of arousal while being cared for in the ICU [20*,21,22]. More hospitals are promoting family-centered care and creating more developmentally-appropriate ICU settings [23], however, further controlled studies are needed. Identifying infants most affected by the ICU environment may be helpful, as early developmental assessments may identify those at risk of delayed motor function [24]

Palliative and corrective surgery for CHD and their accompanying alterations in cerebral physiology are also implicated in harming subsequent neurodevelopment. For example, in children with HLHS after hybrid palliation, lower developmental scores (Bayley-III) are associated with higher Glenn pressures and increased cerebrospinal fluid volume [25]. This association may be amenable to targeted medical or surgical clinical trials. In addition, optimizing perioperative care to avoid the need for perioperative extracorporeal membrane oxygenation (ECMO) should improve neurologic outcomes. Neurodevelopmental scores in children with CHD who were supported with ECMO were below scores for both healthy children and those with CHD who did not require ECMO [26]. Children on ECMO showed a high prevalence of parenchymal brain injury on magnetic resonance imaging (MRI) suggesting that ECMO complications may be the predominant factor contributing to the lower outcome scores [26].

One newly recognized modifiable risk factor is the potential neurotoxic effects of medications [27] and medical devices. For example, diuretics, which are used judiciously in patients with CHD, are toxic to the inner ear, and these patients have a hearing loss prevalence of 16–22% [28**,29]. However, only one-fourth of these children were suspected to have hearing problems prior to their evaluation by the research study [28]. Recent attention has focused on the possible effects of neonatal exposure to chemicals in medical devices (e.g., catheters) that are thought to be neurotoxic. A small study of 18 CHD patients during the neonatal hospitalization found urinary concentrations of common industrial plastics chemicals were well above the levels seen in their mothers or the general population [30*]. While further study is necessary to determine whether the levels of these chemicals present in patients with CHD have adverse neurodevelopmental implications, the importance of this potential exposure is generally unrecognized and poorly documented.

Psychosocial Effects of Altered Neurodevelopment

Neurodevelopmental impairment can impact quality of life (QOL), as assessed by both parents and older children. In one study, parents of patients with HLHS (ages 2–18 years) reported that their children had lower QOL than both healthy peers and those with other chronic illnesses [31*]. Additionally, those with more neurologic complications (e.g., stroke and seizure) had even lower QOL scores. Other studies have shown normal self-reported QOL in older children and adolescents with Fontans [32], patients after repair of interrupted aortic arch [33], and in a large German study with a range of cardiac defects [34]. However, recent meta-analyses found worse QOL in CHD survivors [35] and their parents [36]. The most recent follow-up data from the Boston Circulatory Arrest Study [37*], a sample of 135 patients with D-TGA, found that patients at 16 years of age had lower psychosocial and physical health compared with population norms. Patients with the worst attention scores at age 8 years had the worst psychosocial and physical health at age 16 years, which may indicate that interventions for attention deficits may offer the possibility to improve outcomes. Similarly, executive functioning has been shown to be a key mediator of QOL and the ability to cope with a chronic medical condition [4,38].

While QOL is affected by CHD severity as well as consequences of reparative and palliative surgeries, QOL is also affected by factors related to health care management. The number of doctors’ visits per year, the number of medications taken, self-esteem, and anxiety can be stronger factors than underlying disease characteristics [39]. While partially confounded by disease severity, simplifying clinic and medication schedules is a potential modifiable factor. Adults with CHD have a high prevalence of depression, anxiety, and post-traumatic stress disorder which impacts QOL [40*]. A higher burden of mental disorders was present in patients with lower socioeconomic status and those with more medical comorbidities. Perhaps paradoxically, those with less complex CHD actually had a higher prevalence of depression and anxiety. Mental health issues in this population lead to higher health care utilization with more cardiology and emergency department visits, and more hospitalizations [40*]. Despite the high prevalence of psychological distress, there have been very few controlled trials to examine the efficacy of psychosocial interventions with mixed results with regards to efficacy [41].

With the high prevalence of neurodevelopmental disability, efforts to improve long-term outcomes through therapy after hospital discharge are important. Neurodevelopmental clinics for multidisciplinary assessment were recommended by AHA guidelines. Such services are only effective for patients who attend. However, utilization of early intervention (EI) services is low even among patients from the SVR trial who were closely followed at major centers, with 38% of patients never having received EI [42**]. Loccoh et al. [43*] found that only 17% of patients returned for follow-up at age 3 years. Lower risk surgery and a shorter hospital stay predicted non-attendance, which may reflect parents or practitioners having lower concern for these patients even though they are still at risk of poor outcomes. Quality improvement initiatives have been effective at improving compliance with recommended assessments [44].

Biomarkers of Brain Injury

In order to improve early diagnosis and prognosis of brain injury in patients with CHD, many groups have examined neuroimaging biomarkers. CHD survivors have specific injury patterns on brain MRI including stroke [45] and white matter abnormalities [46]. Global reductions in brain volume [47*,48,49,50*] are seen which parallel poor head growth beginning in fetal life [51,52]. Recent neuroimaging studies have extended our knowledge of neonatal brain growth [47*,48] by showing that the trajectory of abnormal growth begins prenatally. Claessens et al. [46] found that white matter maturation in neonates with critical CHD occurs in the expected pattern (central to peripheral and posterior-to-anterior), although delayed relative to infants without CHD. While no association between white matter maturity and likelihood of injury were seen in this study, Guo et al [53*] showed that those brain regions that were undergoing active maturing where particularly vulnerable to white matter injury, resulting in more peripheral injury in comparison to premature neonates.

Studies with larger cohorts and longer follow-up have reinforced the link between neonatal white matter injury and developmental outcomes [54**,55]. While most studies of MRI findings in patients with CHD focus on the neonatal period and infancy, Watson et al. [56] studied a large cohort of adolescents after Fontan palliation. White matter microstructure was examined with diffusion tensor imaging, which revealed widespread changes throughout the white matter. Higher degrees of injury were associated with reduced processing speeds and IQ.

Advanced MRI diffusion metrics for assessing the brain’s microstructure demonstrate differences between CHD patients and controls [57**,58,59]. However, whether these novel techniques will result in more accurate prediction of functional outcomes has not yet been explored. Kelly et al. [57*] used higher-resolution MRI to assess brain maturation and found results consistent with abnormal dendritic arborization, with reductions in cerebral oxygenation associated with greater degrees of impaired development. While neurodevelopmental outcomes were not assessed in this cohort, this study provides early evidence that improvements in cerebral oxygenation may be a way to modify neuronal development and long-term brain organization. Complementing these structural imaging studies, MRI-derived resting-state functional connectivity demonstrated that while the brain’s global network structure was preserved preoperatively in CHD patients, local disturbances, particularly in subcortical regions such as the putamen, caudate, globus pallidus, and thalamus, were found which might be a mediator between structural changes and observed behavioral outcomes [60*].

Despite these advances, MRI is not a bedside tool and can only provide snapshots of brain injury and maturation. New technologies are needed for real-time longitudinal monitoring of infants and children being cared for in the ICU during the critical periods when interventions may prevent or mitigate neuronal injury. Amplitude-integrated electroencephalography (aEEG) is feasible to be performed at the bedside; however, while abnormal EEG background and ictal discharges can be a sign of brain injury [61], sensitivity and specificity are low, and patterns are highly affected by sedatives [62]. Optical monitoring (e.g., with near-infrared spectroscopy) can measure regional cerebral oxygenation continuously at the bedside; however, commercial systems have difficulty quantifying cerebral oxygenation in the setting of chronic hypoxemia. Newer more advanced systems (particularly in combination with diffuse correlation spectroscopy) can overcome these limitations and provide insight into the temporal evolution of cerebral oxygenation preoperatively [63].

Serum biomarkers provide information regarding the health of brain cells including neurons and astrocytes. Higher levels of glial fibrillary acidic protein (GFAP) measured intraoperatively were associated with lower Bayley-III scores at 12 months, however, the GFAP level only accounted for a small portion of the variance in outcomes [64]. S100B which is elevated in the setting of impaired cerebral oxygenation in neonates with CHD [65] may have a better positive and negative predictive values for neurodevelopmental outcomes in CHD patients [66].

Due to the heterogeneity of the clinical population, future insights will be gained from preclinical animal models where more anatomical and treatment covariates can be controlled. A major step in this direction was recently taken with the developed a fetal lamb model [67*]. In this model, chronic hypoxia led to impaired myelination and decreased neuronal density, which is similar to that seen in humans with CHD.

Conclusion

Many children with CHD suffer from poor neurodevelopmental outcomes. While recent research has demonstrated that a large amount of the variance between patients may be due to underlying genetic factors, there are likely modifiable factors related to CHD treatment. Improving outcomes will require more knowledge of imaging, serum, and behavioral biomarkers so that targeted interventions can begin earlier. While EI services are recommended for CHD children, attendance remains lower than desired. It is essential that all members of the health care team reinforce the importance of neurodevelopmental monitoring throughout these patients’ lifetimes. Additionally, there is a need for research on effective psychosocial interventions to improve the QOL of children and adults with neurodevelopmental impairment.

Key Points.

• Patients with CHD suffer from neurodevelopmental disability, especially in language and executive function.

• Much of the neurodevelopmental impairment may be explained by genetic factors. However, possible modifiable risk factors include improving cerebral oxygenation, reducing the use of neurotoxic medications, and improving the availability of appropriate developmental stimuli in the intensive care unit.

• Neurodevelopmental disability leads to anxiety, depression, and lower quality of life. Few studies of strategies psychosocial interventions have been performed, leading to a lack of effective therapeutic options to improve the lives of older children and adults with congenital heart disease.

• Neuroimaging has increased our knowledge of how injuries lead to outcomes. However, the relationships are often complex, making prognosis in an individual patient difficult. Furthermore, bedside modalities will be necessary to allow timely intervention.