Term infants born with cyanotic congenital heart disease (CHD), including transposition of the great arteries (TGA) have been reported to have significant disturbances in cerebral oxidative metabolism and mild to moderate white matter injuries primarily in the form of periventricular leukomalacia before and after neonatal open heart surgery.[2, 3] Several medical and patient-related risk factors including pre-, intra-, and postoperative variables may contribute to their neurological status and long-term neurocognitive outcomes. Despite overall normal general intelligence, long-term cognitive impairments as well as a high prevalence of attention-deficit-hyperactivity disorder, ‘externalizing’ problems (e.g. aggressive behaviour), and ‘internalizing’ difficulties (i.e. social withdrawal) have been reported. Recent studies have also demonstrated significant impairments in understanding other people's inner emotional states, suggesting deficits in social cognition. Compared with children with typical development, children with TGA have been reported to display impairments on mental state attribution (first- and second-order theory of mind [ToM] tasks) as they had significant difficulties on making correct inferences on the false-beliefs held by others and on understanding and predicting their behaviour.[9, 10] Adolescents with TGA also had significantly worse performances on a test of complex inner state attribution (reading the mind in the eyes test) and exhibit more autistic traits than expected in the general population. These findings opened a new window for the study of neurocognitive consequences associated with CHD; however, important questions remain concerning the extent and generalization of these difficulties to the broad domain of social cognition.
Social cognition is a multifaceted concept that includes elementary processes such as the recognition of facial expressions of emotion and more complex abilities referring to the comprehension of mental and affective states. In typical development, early improvements on basic facial emotion recognition are reported in preschool-aged children. This progress parallels the comprehension of some affective states, such as the understanding that people may hold different desires or that emotions may have external causes (e.g. if somebody gets a present for their birthday, how would they feel?). Cognitive mental state attribution also undergoes significant progress during this period as children can recognize first-order false-beliefs. Furthermore, in middle childhood, at around 7 to 8 years, children are able to understand the possibility of hiding emotions (e.g. showing a happy face but feeling sad inside), and they come to understand second-order false-beliefs with affective contents (e.g. Paul thinks that Mary feels) followed by epistemic contents (Paul thinks that Mary thinks).[17, 18] Finally, at around 9 to 11 years, they begin to master mixed emotions (e.g. feeling happy and sad at the same time) and some aspects of morality (e.g. feeling guilt after telling lies). From a neurofunctional point of view, emotion recognition and comprehension in typical development involves networks including the amygdala and the prefrontal cortex. Furthermore, concomitant deficits in facial expression recognition and emotion understanding as well as in affective and cognitive ToM have been found in 6- to 8-year-old children after traumatic brain injury and in children with attention-deficit–hyperactivity disorder. Despite the growing concern regarding systematic neurocognitive screening and remediation strategies for children with CHD, no data concerning the extent and nature of social cognition impairments including emotional processing are available.
The objective of this study was to assess core components of social cognition including elementary aspects such as basic facial expression recognition (perceptual identification and labelling), as well as main milestones of emotion comprehension organized hierarchically according to their developmental path of acquisition (e.g. from desires to moral feelings comprehension). In addition, as school-aged children with TGA have been reported to have significant delays on the litmus test of ToM (cognitive false-belief attribution), we sought to investigate if delays would also be observed for same-structured tests with an affective component (second-order false-belief tasks with a positive or a negative emotional outcome). In typical development, affective components in second-order false-belief understanding seem to facilitate children's performances. We sought to assess if children with TGA would also benefit from this affective effect. Finally, we investigated potential neonatal medical risk factors including pre-, intra-, and postoperative variables associated with emotional and cognitive outcomes.
- Top of page
- Supporting Information
In recent times there has been a growing concern about social cognition impairments after CHD,[23, 31] yet no research to date has investigated the extent of these deficits. The aim of the current study was to assess the recognition of facial expressions and the gradual comprehension of emotions, as well as the understanding of affective and cognitive false-belief. When compared with the comparison group, children with TGA had significantly lower scores on the test of facial expression labelling despite normal identification of the same emotions when the verbal label was explicitly given. When a series of potential covariates (age, language scores, SES, and sex) was controlled, however, this difference was no longer significant. In typical development, the visuo-perceptive recognition of emotion facial expressions precedes the capacity to produce and correctly retrieve verbal labels for these emotions. The current results suggest that facial expression recognition generally may be preserved after TGA. Moreover, children with TGA did not exhibit an atypical pattern of performance according to the type of emotion. As observed in typical early development, negative emotions are less accurately recognized than positive ones and this also applies to our cohort with TGA.
Results demonstrated that children with TGA achieved significantly fewer components in the TEC than comparison children. However, when language comprehension scores, age, SES, and sex were controlled, this group difference disappeared. Indeed, it has been reported that language comprehension levels may explain a significant portion of variance of this test in normal development. When children's performances were compared according to each category specifically, however, the group with TGA performed significantly less well than the comparison group for the ‘mental emotions’ category even after controlling for all covariates mentioned including language scores. Children with TGA exhibit significantly lower scores on components such as dealing with desires, affective first-order false-belief or understanding concealed emotions, normally acquired at around 6 to 7 years of age. According to the gradual complexity of the TEC, our cohort displayed normal scores on the most elementary category taxing external emotions usually achieved at around 5 years of age. This developmental pattern indicates that basic processing of emotion is not impaired and difficulties apply to age-expected milestones. Thus, their performances were not significantly lower than those observed in the comparison group for the ‘reflective emotions’ category, which is expected to be acquired later in development. This suggests that children with TGA do not necessarily have an atypical early onset of complex emotion comprehension as their performances were similar to comparison children in the reflective category at age 7. Instead, it could be hypothesized that developmental lags may become apparent once these abilities have consolidated enough to age-expected levels, even if it is still unclear how these weakness may evolve in the long term. Bellinger et al. reported that 16-year-olds with TGA are significantly impaired in ‘reading the mind in the eyes’ test, suggesting that the processing of complex cognitive or affective mental states may have long lasting difficulties. Our data showed that success rates on first-order cognitive false-belief and second-order affective (positive and negative) false-belief were significantly lower for children with TGA and are in accordance with previous findings showing significant impairments in cognitive first- and second-order false-belief understanding in 7-year-old children with TGA. Taken together, these results add to the evidence for a potential generalization of impairments to emotional and affective components of social cognition. It is worth noting, however, that success rates on affective second-order false-belief understanding were significantly higher than those on a cognitive, yet similarly structured, second-order false-belief task. This could mean that the emotional outcome of the task facilitates inner states attribution for children with TGA, comparable to what is observed in typical development. The success rate of our comparison children at the cognitive second-order task (29%) was lower than that reported in a previous study on typical development (in Parker et al. 78% at age 7) and lower than our previous cohort of typical 7-year-olds. This could partially be explained by the inclusion of 6-year-olds in the current study, who systematically failed at this task. It is commonly admitted that typical children can pass second-order false-belief tests between the ages of 7 years and 9 years. It is also important to note that second-order ToM tasks, especially standard cognitive versions, have been reported to entail high language comprehension and information processing requirements. Failure at this task could reflect not only difficulties in ToM conceptual understanding per se, but also deficits in decoding and integrating this information as a result of limited linguistic and/or executive abilities. An important step in dissociating these deficits would be to test children with a linguistically facilitated version of second-order tasks that have been reported to elicit better performances in children with typical development.
Neural correlates of emotion processing including decoding facial emotion expressions and emotional ToM in typical development have been reported to involve cortical and sub-cortical structures including the amygdala, the orbitofrontal cortex, medial prefrontal cortex, and the right temporo-parietal junction. These studies have reported an increased neural specialization of these structures and, more importantly, a crucial role of neural connectivity as white matter volumes progressively increase with age during childhood. The functional relationships between regions underlying higher-order cognitive skills such as social cognition may require the integrity of association white matter tracts, as has been proposed in adult models.[37, 38] Term infants with cyanotic CHD requiring neonatal interventions have been reported to present white matter anomalies and delays in myelination, similar to what can be observed in preterm infants.[1, 2] No study to date has established an association between neonatal white matter disturbances and long-term neurocognitive outcomes in cardiac patients. It is critical to understand the impact that these early anomalies could have on the development of complex cognitive skills known to rely on intact brain connectivity throughout childhood.
Our data showed that among the pre-, intra-, and postoperative variables examined, preoperative factors only (presence of a VSD, younger age at the arterial switch operation, and a prenatal diagnosis of TGA) were significantly associated with better cognitive results on facial expression recognition and affective false-belief. An associated VSD plays a crucial role in cyanosis tolerance in the newborn infant as it allows for mixing of the arterial and pulmonary circulations and, therefore, improvement of oxygen saturation before cardiac surgery. Younger age at the arterial switch operation has been associated with a reduced risk for white matter injury in the form of periventricular leukomalacia in newborns with TGA and overall preoperative brain injury seems related to severity and duration of exposure to hypoxaemia in these patients. Finally, our results are in accordance with previous data showing a beneficial impact of prenatal diagnosis on neurocognitive outcomes in preschool-aged children with TGA. Indeed, prenatal diagnosis of cyanotic CHD is associated with reduced neonatal morbidity including better haemodynamic stability and neurological status as it reduces the odds of prolonged cyanosis before cardiac surgery.
Data from this study may suggest concomitant vulnerabilities in all core aspects of social cognition including age-expected emotion-processing abilities. Nevertheless, several limitations should be considered. As we sought to limit potential medical confounding variables including multiple cardiac operations and chronic cyanosis, we focused on patients who had TGA as they constitute a homogeneous group with cyanotic CHD and no associated genetic syndromes. Our results cannot be generalized to all children with CHD as variations in haemodynamic instability and operative management should be noted. Our study does not allow for investigation of the neuropathology related to cognitive dysfunction because no brain imaging was available for our patients. In addition, as ToM is also dependent on language development, future evaluations should systematically test this factor not only as a control variable but also including less demanding ToM tasks in terms of linguistic abilities. Moreover, observational studies (contrary to randomized controlled trials) may be subject to biases, notably participant selection and multiple confounding variables. We provided traditional statistical methods of adjustment for baseline covariates in all our models. Covariance analyses, however, should be conducted with caution and future studies in this population should consider the use of propensity score methods for optimal bias reduction.
Finally, future studies should determine the association of social cognition and emotion processing impairments assessed with neuropsychological testing to more ecological parental reports of children's functioning. This is particularly relevant as children with heart disease may display high levels of behavioural and emotional problems,[8, 42] indicating that daily life social functioning may be vulnerable in these patients. Recently, a scientific statement from the American Heart Association categorized children requiring open-heart surgery as at high risk for developmental disorders and recommended formal evaluations for these children. Social cognition screening should also be routinely conducted in all patients with CHD at high-risk, as deficits could affect social and emotional outcomes. Remediation or preventive interventions based on formal cognitive training such as ‘thought-bubble ToM training’ including pictures to depict mental states visually or interventions based on reinforcement of executive functions,[44, 45] may help reduce social cognition deficits after CHD.