Neural correlates of socio-emotional perception in 22q11.2 deletion syndrome

The 22q11.2 deletion syndrome (22q11DS), also known as DiGeorge or velocardiofacial syndrome (VCFS), is a neurogenetic disorder affecting approximately 1 in 4000 live births [1] and occurring in up to 1:1000 pregnancies [2]. The syndrome is associated with elevated risk for schizophrenia [3] and is characterized by high rates of distinct positive and negative symptoms. It is well established that social dysfunction is also a common feature of the 22q11DS profile [4‐6]. Individuals with 22q11DS are described as shy, withdrawn, and presenting social interaction problems particularly with their peers [4‐6]. Up until now, only a few neuroimaging studies have investigated the neural correlates of socio-cognitive processes in the 22q11DS population to better understand the emergence of social deficits associated with the syndrome [7‐9]. The neural substrates of social-emotional impairments hence remain largely unclear. In the current study, we therefore focused on the neural correlates of socio-emotional processes to extend our understanding of impairments in the 22q11DS population related to such mechanisms.

Social perception encompasses the processing of various social cues that individuals encounter in everyday life (e.g. face and voice). Social cue perception has been mostly investigated through face perception. In healthy individuals, face perception is mainly associated with increased activity in the fusiform face area, visual extrastriate cortex, lateral occipital gyri, anterior temporal pole, and posterior superior temporal gyrus [10, 11]. Moreover, the processing of emotional information in faces involves a similar network as emotion perception (see below), including limbic regions, inferior frontal gyrus, medial prefrontal gyrus, and putamen [11‐13].

Until recently, most studies in patients with schizophrenia focused on higher order processes of social cognition (e.g. theory of mind) while lower aspects as social perception have received less attention. To our knowledge, only one study has examined brain activations in response to simple visual social cues [14]. This study reported abnormal neural activity during processing of social information. Hypoactivation in regions associated with visual processing (occipital and temporal regions) and increased cingulate activity during the processing of social vs. non-social images has been observed in patients with schizophrenia compared to controls. In patients with 22q11DS, as in schizophrenia, most studies have focused on higher-order socio-cognitive processes, highlighting impairments in theory of mind and emotion recognition [15, 16], and few functional magnetic resonance imaging (fMRI) studies have been conducted on this topic [7‐9]. Andersson et al. [8] investigated social perception and reported hypoactivation of the fusiform in response to faces versus houses. In light of the high prevalence of social impairments in 22q11DS, additional studies examining the neural substrates of social perception and their link with social functioning are clearly needed.

A second aspect of social cognition that will be investigated in this study is emotion processing. Commonly, these processes are investigated while individuals are viewing pleasant or unpleasant images compared to neutral images. In healthy participants, a set of regions has been related to emotion processing, mainly including the limbic system (amygdala, anterior hippocampus, anterior insula, and cingulate gyrus), as well as brain stem nuclei, thalamus, ventral striatum, medial prefrontal cortex, posterior cingulate cortex, precuneus, lateral temporal cortex, and temporal pole [17, 18]. However, up until now, only three studies have investigated the neural basis of emotion processing in 22q11DS [7‐9]. van Amelsvoort et al. [7] compared brain activity during the presentation of mixed emotional facial expressions between eight individuals with 22q11DS and nine controls. Two types of facial emotions (happy or angry) and neutral faces were presented in a block design. Patients with 22q11DS showed less activation in the right insula and frontal regions and more activation in occipital regions compared to controls. However, this study used a block design (with mixed emotion), which made it impossible to differentiate neural responses as a function of the type of valence. Two additional studies used only negatively valenced stimuli [8, 9] and confirmed the hypoactivation in regions related to emotion processing (superomedial prefrontal cortices) in patients with 22q11DS. However, and contrary to van Amelsvoort et al. [7], some regions involved in socio-cognitive processing (fusiform, anterior cingulate cortex) were also hypoactivated. Moreover, only one study investigated the relationship between brain activation and social functioning in individuals with 22q11DS [9]. The authors found that decreased brain activation during emotion perception was related to social difficulties in patients with 22q11DS. So far, studies investigating emotional processing in participants with 22q11DS reported results from very small samples (8 to 15 patients) and only one type of valence was explored. Consequently, the neural bases of emotion processing still remain unclear in this population, and the distinction between positive and negative emotion perception has never been properly investigated.

Finally, it is important to take into consideration that emotions can influence social processing, as it appears that there is an overlap between social information and emotion processing. Indeed, it has been argued that social cues are inherently emotional and that social information processing will therefore also involve emotion-processing networks [19]. To test this hypothesis, several studies in healthy individuals have been conducted and confirmed an interactive processing of social content and valence [19‐21]. While a social content × valence interaction effect was found in the thalamus, superior temporal sulcus, middle orbito-temporal cortex, as well as in the anterior insula and lateral medial prefrontal cortex in several studies, Norris et al., Scharpf et al., and Vrtička et al. [19‐21] also observed such interaction effect in the amygdala, fusiform gyrus, anterior superior frontal gyrus, and middle occipital cortex. Conversely, in the 22q11DS, the influence of valence on social and non-social information processing is currently unknown.

This study examined neural correlates of socio-emotional processing patients with 22q11DS compared to healthy controls. First, we investigated neural correlates of social perception. In line with the literature in schizophrenia [14], we expected to observe altered activations in the social perception network in participants with 22q11DS. Secondly, we examined neural response to positively and negatively valenced stimuli. Based on previous studies in participants with 22q11DS, we expected to observe a significant decrease in regions involved in emotion (e.g. insula, frontal regions) and socio-cognitive processing (e.g. fusiform). Thirdly, in order to investigate the influence of emotions on social perception in 22q11DS, we also examined a social content × valence interaction. Finally, we hypothesized that impaired activation of socio-emotional networks would be related to socio-cognitive deficits in participants with 22q11DS.

The present study aimed at identifying the neural correlates of socio-emotional perception in individuals with 22q11DS. Results indicated neural hyporesponsiveness within the social perception network in participants with 22q11DS compared to controls. Second, comparable to healthy controls, individuals with 22q11DS showed activation in regions related with emotion processing during the presentation of positive and negative stimuli. No correlation between brain activation and the clinical measures was observed in the 22q11DS population. Finally, results showed no social content × valence interaction differences between patients and controls.

This is the first study investigating the neural correlates of complex social stimuli perception in patients with 22q11DS. In line with our hypothesis, participants with 22q11DS showed atypical patterns of activation during the processing of social information compared to healthy controls. Specifically, we observed reduced brain activation in regions belonging to the default mode network (DMN). Indeed, the DMN encompasses the PFC as the anterior cingulate cortex (ACC) in its anterior part; laterally, it comprises the bilateral IPL and the medial temporal lobes whereas posteriorly it includes the posterior cingulate cortex (PCC) and the precuneus [37, 38]. In the current study, we found hypoactivation of these regions (except for temporal lobes) during the perception of social information in individuals with 22q11DS compared to controls. The DMN is a resting-state network more activated in the absence of a cognitive task and has been well described as being implicated in various socio-cognitive processes such as self-referential processing or theory of mind [39, 40]. Nevertheless, recent literature established that there is an overlap between regions attributed to the DMN and regions activated during socio-cognitive tasks [41, 42]. In disorders characterized by social functioning deficits, such as schizophrenia or autism spectrum disorder (ASD), alterations of the DMN have been previously reported. The majority of studies conducted in schizophrenia points to a decreased functional connectivity of the DMN [43‐47], whereas in ASD, altered functional connectivity of DMN regions has been reported during both rest [48‐56] and during social tasks [57].

In patients with 22q11DS, alterations of resting state networks, including the DMN, have also been demonstrated using various methods. Whereas studies investigating functional connectivity of multiple resting state networks found hypoconnectivity of the DMN in individuals with 22q11DS [57‐59], a multimodal approach revealed both structural and functional connectivity disruption of the DMN [60]. A few studies also examined the behavioural correlates of these alterations. In particular, associations between the strength of the DMN long-range connectivity and social functioning [58] and between a functional decrease of the DMN and prodromal symptom severity have been reported [57].

Our results are in accordance with previous literature in the 22q11DS population and demonstrate a hypoactivation of the DMN during the perception of social information, a basic socio-cognitive process. Taken together, these findings support the involvement of the DMN in socio-cognitive processes and point to a central role of this network in the pathophysiology of 22q11DS. The present study also suggests that alterations of this network could account for social dysfunctions observed in the syndrome. However, it should be noted that post hoc analyses did not reveal any significant correlations between activations within regions of the DMN and the socio-cognitive measures in the group of patients with 22q11DS. Because our sample size is relatively small and we used a non-direct measure of social functioning (parent-reported questionnaire), this finding should be interpreted carefully. Moreover, previous studies showing associations between social functioning and the DMN investigated functional connectivity within the DMN whereas the present study examined BOLD signal. Consequently, further studies investigating the functional connectivity during social perception as well as the association between impaired social perception network and social functioning are required.

Surprisingly, we did not find any significant difference between groups during social perception in the fusiform gyrus, while previous findings found reduced fusiform response during face perception in individuals with 22q11DS [8, 9]. We also failed to see differences in the amygdala and superior temporal sulcus, whereas these two regions appear to be significantly activated during social vs. non-social perception in healthy controls [21]. The lack of result could be due to thresholding differences applied between studies. Indeed, while we applied very strict corrections, previous studies reported results with lower and less-sensitive thresholds [8, 9, 21]. Further studies investigating the activation of these regions during social perception are required.

Taken together, our results clearly indicate alterations in the social perception network in participants with 22q11DS. These alterations could lead to difficulties in social information processing, which could in turn influence various aspects of social cognition and thus contribute to social functioning impairments frequently observed in this syndrome [4‐6].

Several explanations could be made to explain differences in the social perception network in 22q11DS. As proposed by Azuma et al. [9], one possible explanation may be a variation in dopamine, which is known to be a neuromodulator of socio-cognitive processes [61]. Individuals with 22q11DS are hemizygous for some genes, including the catechol-O-methyl transferase (COMT) gene that plays a role in dopamine degradation. Consequently, dopamine levels in regions of the social brain could be modified in 22q11DS [62, 63] and lead to altered brain response during social processing. It has already been suggested that dopaminergic genetic variation impacts social perception and behaviour [64]. Indeed, some studies investigating the effects of dopamine variation on ventral striatum response to social reward have demonstrated that heightened dopamine signal is associated with increased response to social reward (impulsive or aggressive) [65, 66] and increased social approach behaviours. Future research investigating the impact of dopamine variation on social perception in 22q11DS is required. Alternatively, these results could also be driven by differences in brain anatomy. Indeed, a previous study conducted in participants with 22q11DS reported a positive relationship between fronto-striatal grey matter volume and social behavioural difficulties [67]. Moreover, structural changes in regions that are part of the social network, such as frontal regions, PCC, and ACC, have already been reported in the 22q11DS literature [68‐70]. Thus, larger studies are required to examine the relationship between functional and structural alterations within the social network in the 22q11DS population.

Comparison of the neural correlates of emotion processing within groups revealed that positive stimuli perception was associated with increased activation in the lingual gyrus in both patients and controls. Conversely, increased activation to positive images was found in the PCC and cuneus in controls only, while increased activation in the precuneus, postcentral gyrus, and superior parietal lobule was observed solely in patients. During negative stimuli perception, controls showed greater activation of the right lingual gyrus, bilateral cuneus, and left PCC, while no significant activation was found in patients. Between-group comparisons did not reveal significant difference. These results suggest that during negative and positive emotion processing, patients and healthy controls present a similar pattern of activation. Thus, compared to the social perception network that was clearly atypical in 22q11DS, the neural correlates of emotion perception appear to be preserved in this population. Our results are in contradiction with previous work in the 22q11DS that described hypoactivation of regions related to emotional processing. Indeed, hypoactivation in frontal regions during negative emotional viewing [8, 9] and reduced activation of the insula during mixed emotional faces [7] was found. However, it should be noted that van Amelsvoort et al. reported results from a very small sample (eight patients versus nine controls), so that differences observed between the two studies should be interpreted with caution.

Finally, we examined the influence of emotions on social perception by investigating interactive processing of social content and valence. Our results did not reveal any significant social content × valence interactions either within- or between-group comparisons. Those results are in contraction with previous studies conducted in healthy individuals where a social content × valence interaction emerged [19‐21]. Indeed, while Norris et al. [19] and Scharpf et al. [20] reported significant interactions in the thalamus, superior temporal sulcus, and middle orbito-temporal cortex, as well as in the anterior insula and lateral medial prefrontal cortex, Vrtička et al. [21] found such interaction within the amygdala, fusiform, anterior superior frontal gyrus, and middle occipital cortex. One explanation for this absence of social × valence content interaction is a lack of statistical power due the sample size. However, it should be noted that the previous studies reported whole brain or region of interest results with an uncorrected threshold, while we used strict corrections by combining a primary voxel-level statistical threshold at p < 0.001 (uncorrected, whole brain) and k ≥ 20, with a cluster-extent family-wise correction (FWEc < 0.05). As no differences in interactive processing of social content and valence between groups were found, this suggests that social processing is impaired regardless of the valence in the 22q11DS population.

The results of the present study should be interpreted in light of several limitations. First, we investigated the neural correlates of social perception whereas it is well documented that patients with 22q11DS present visual perception and processing deficits [70, 71]. Indeed, difficulties in object, face, and emotion recognition have been described in the 22q11DS population [71]. Moreover, Magnée et al. [70] reported abnormal transmission between higher and lower visual cortex areas during the presentation of visual stimuli in an event-related potential study. As visual processing deficits could influence social perception, their impact on the observed findings should be further examined. Secondly, although the number of participants included in this study is higher compared to previous work in this field of research [7‐9], our sample size remains relatively small. Consequently, results need to be interpreted carefully and studies including a larger sample are required. Thirdly, the neural correlates of social-emotional perception have only been investigated in a cross-sectional way. Thus, the time window during which functional alterations in the social perception network emerge remains unknown. Longitudinal studies examining the developmental trajectory of these alterations in 22q11DS are therefore required. An additional limitation is the potential influence of medication within the group of patients with 22q11DS, which was not investigated in the current study. However, we chose to exclude patients with schizophrenia or another psychotic disorder diagnosis, and only one patient was under antipsychotics at the time of testing. Finally, the neural correlates of additional socio-cognitive processes should be examined in further studies (e.g. theory of mind). Indeed, a better understanding of the neural correlates of social deficits in the 22q11DS population is required.

The present study provides evidence of specific alterations in the social perception network, irrespective of valence, in the 22q11DS population. Moreover, emotional processing of negative and positive stimuli seems preserved in individuals with 22q11DS. This is the first study investigating lower-order processes of social cognition in the 22q11DS population. The observed alterations are likely to influence higher order socio-cognitive processes. Deficits of social information processing could therefore be a key factor leading to socio-cognitive impairment in the 22q11DS population. Taken together, these findings highlight the need to better understand the emergence of social perception deficits in 22q11DS. Future studies investigating the developmental trajectory of social perception and its association with socio-cognitive processes are required.