In this last section, we will integrate the results from the two axes of this project. First, we will highlight the common pattern of brain alterations observed between the two axes, we will then discuss how the social motivation theory could be used to explain the social deficits observed in 22q11DS. Finally, we will propose a theoretical model of social motivation deficits in the 22q11DS.
3.3.1 The DMN as a common marker for reward and social perception impairments in 22q11DS
The results of studies 2, 5 and 6 point to impaired brain activity of the DMN during social reward processing and social perception, as well as an altered pattern of functional con-nectivity of this network during social perception. These results are in line with previous findings reported in patients with 22q11DS. Indeed, impaired DMN functional connectivity during resting-state has been observed in the 22q11DS population (Mattiaccio et al., 2016;
Padula et al., 2017; Schreiner et al., 2014). Although, previous studies conducted in the field of 22q11DS examined DMN impairments through a different modality (i.e. resting-state), the DMN has been shown to be activated during resting-state as during some specific cogni-tive tasks (Buckner et al., 2008). In particular, its role in socio-cognicogni-tive processes has been
well demonstrated (Andrews-Hanna, 2012; Spreng et al., 2009). Nevertheless, nowadays, the role of the DMN in reward processing appears unclear. More specifically, these three studies point out the role of the PCC in social reward processing and social perception deficits in 22q11DS. As mentioned in section 2.3, the PCC has been suggested to link motivational out-come to eye movements. Higher activation of the PCC with increase magnitude of salience stimuli has been reported in non-human primate studies (McCoy et al., 2003; McCoy and Platt, 2005). Thus, the PCC seems involved in the guidance of gaze behavior toward salient features. Of note, social information is highly salient by nature and both social reward pro-cessing and social perception are involved the visualization of social stimuli. It thus appears not surprising that the PCC sustains these two processes. Taken together these findings support the idea that the PCC might represent a marker for both reward processing and social perception impairments in 22q11DS. Nowadays, little is known about the mechanisms underlying DMN impairments in the 22q11DS. Nevertheless, impairment of parvalbumin in-terneurons (PV) has been proposed as a potential contributor. PV inin-terneurons play a role in activation/deactivation of pyramidal neurons that are necessary for synchronisation of local and distant brain regions (Ferguson and Gao, 2018). In 22q11DS mouse model (LgDel/+), a decrease of parvalbumin interneurons has been reported (Steullet et al., 2017)(Steullet et al., 2017). Therefore, reduction of parvalbumin interneurons in the 22q11DS population could impact the activity and connectivity of large-scale networks such as the DMN.
3.3.2 Deficits specific to social information in 22q11DS: relation to the social motivation hypothesis
Studies 2 and 5 argue in favour of a specific deficit in the processing of social information in patients with 22q11DS. Indeed in Study 2, we found brain alterations during social reward processing uniquely while no differences between 22q11DS individuals and healthy controls were found during monetary reward processing. In Study 5, we observed brain alterations during the perception of social versus non-social images, also suggesting a specific alteration during the processing of social information in people with 22q11DS. In ASD, specific al-terations in the processing of social information have been reported in line with the social motivation theory (Chevallier et al., 2012) (See section 2.1). As core features of ASD such as impairments in joint attention, gestural communication, initiating conversion and cir-cumscribed interests are also characteristics encountered in individuals with 22q11DS, prior studies have suggested a partial phenotypic overlap between ASD and 22q11DS (Kates et al., 2007). A recent meta-analysis also indicated that 22q11DS individuals have a 8-fold increased risk for ASD compared to the typical population (Richards et al., 2015), although the preva-lence reported appears highly variable across studies (ranging from 0-50%). Nevertheless, some authors have also suggested that ASD in 22q11DS more likely captures a generalized social impairment that differs from idiopathic autism (Eliez, 2007). In particular, the onset of social deficits appears very different between the two disorders. Individuals with 22q11DS appear often overly familiar during childhood and become progressively withdrawn during
adolescence (Eliez, 2007), while in ASD social withdrawal appears earlier during the first years of life. According to the social motivation theory that has been proposed in ASD, deficits in social interest at baseline will progressively reinforce social motivation impair-ments over time. However, contrary to ASD children it might be the case that children with 22q11DS have a preserved social motivation at baseline. Nevertheless, in our study 4 exam-ining the visual exploration of social information, we found that social perception is impaired from childhood to adulthood and additionally becomes more atypical with age in 22q11DS individuals. Thus, in the 22q11DS population, it seems that social deficits more likely derive from social perception impairments at baseline rather than social motivation issues. Indeed, social perception deficits at baseline might impact social cognition and therefore the under-standing of the social environment, this might lead to social motivation impairments. As individuals with 22q11DS might have difficulties to understand their social word and might progressively loose their social interest, this might accentuate social perception deficits over time as well as social difficulties. In sum, our studies point to a specific deficit of social in-formation processing in 22q11DS. In view of the elements described above and based on our clinical expertise, we thought that social deficits in 22q11DS although similar in appearance to those found in ASD result from a different causality. In the next section, we will propose a theoretical model of social motivation deficit in the 22q11DS and will explain in details how social perception deficits may lead to social motivation impairments in this syndrome.
3.3.3 Toward a theoretical model of social motivation deficits in the 22q11DS
Based on the findings obtained in this project, we propose in this section a theoretical and explanatory model of social motivation deficits in the 22q11DS, which integrates social perception within the reward model (Figure 3.2).
We hypothesize that social perception, which are impaired since childhood in 22q11DS (study 4), might be at the origin of the various reward-related processes alterations observed in this syndrome. Indeed, we think that impaired social perception (1) will lead to poor integration of social information’s and consequently difficulties in reading social situations.
Indeed, a good understanding of social environment seems crucial to produce an adapted directed behavior in a given situation. According to Brown and Pluck (2000), goal-directed behaviors are guided by internal states but also by information’s derived from the environment (Brown and Pluck, 2000). Thus, it is likely that an individual who does not efficiently explore social stimuli will not be able to response adequately to a given situation.
Inadequate response to social situations (2) will interfere with the quality of social relation-ships and lead to social failures. Thus, given the poor quality of social relationrelation-ships, the pleasure felt during social situations (in-the-moment pleasure) will be altered (3). Individu-als will associate social situations to social failures (4) (associative learning), which will be encoded in long-term memory. When a new opportunity to engage in a social activity arises (reward representation) (5), for example receiving an invitation for a party, these previous
Figure 3.2: Theoretical model of social motivation deficits in the 22q11DS population
experiences will be reactivated and influence the way individuals with 22q11DS will antic-ipate future social activities. As previous social experiences have been associated to social failures, patients will therefore tend to anticipate future activities as less enjoyable (6). This will influence the value that patients attribute to social situations (7) as the effort that they will consider worth it to produce (8). In fine, the motivational resources - social motiva-tion (9) - will be altered, which could lead to social withdrawal, loss of initiative for social interactions (social avolition) and in addition patients will consider social situations as less enjoyable (social anhedonia). Thus, social perception deficits might lead to reward-related impairments and thereafter influence the emergence of motivational symptoms such as social withdrawal, anhedonia and avolition in the 22q11DS population. Based on this theoretical model, social perception might represent a valuable candidate for early therapeutic interven-tion. Indeed, targeting social perception may be promising for the prevention and treatment of motivational symptoms in 22q11DS. Moreover, recently a new technique aiming to improve social exploration in the field of psychosis emerged (Choi et al., 2017). This technique based on pupillometry feedback has been developed in order to improve visual scanning efficiency in CHR populations. In this method, a processing speed training is proposed to participants in order to improve their visual-motor and processing abilities. This program is incorporated within a game-like fantasy context aiming to enhance motivation for the cognitive task. The pupillometry technique allowed providing an immediate biofeedback to the training software
in order to adjust task parameters and levels and thus provide a personalized and efficient training program. After approximately 30 hours of training over the course of 2 months, pre-liminary results revealed that CHR individuals reported improvements in processing speed, which was related to improvements in social adjustment and social avoidance. In view of the promising results, this technique should thus be considered in the 22q11DS population.
3.4 Limitations
In this section, we will describe some of the limitations of this project, either related to the cohort of 22q11DS individuals either related to the techniques used in this project.
3.4.1 Limitations associated with the DTI sequence and analysis
The DTI sequence used in this project was implemented at the beginning of the Geneva cohort and was chosen regarding to its quality and short duration (5 minutes). The se-quence is characterized by the acquisition of 30 directions, which is not optimal for the reconstruction of crossing fibers. Indeed, it has been estimated that 90% of voxels contains fibers with multiple directions. Nevertheless, a minimum of 64 encoding directions, has been suggested as the minimum number of directions required for high quality DTI (Alexander et al., 2007). Thus, issues in reconstructing crossing fibers may influence the accuracy of ours results. In Study 1, we initially planned to reconstruct fibers within the entire reward system. Using the probabilistic tractography algorithm, we were however confronted to the following issues: for numerous participants, we were unable to reconstruct tracts mainly within cortical and subcortical regions. Thus, being limited by our DTI sequence and to ensure the veracity of the reported results, we decided to focus on the accumbofrontal tract for which a well-developed mask has been created (Karlsgodt et al., 2015) as a first step to identify brain correlates of hedonic alterations in 22q11DS. Henceforth, more advanced sequences are available to answer to cross-fibers issues. For instance, an approach based on the high angular resolution diffusion-weighted imaging (HARDI) consists to measure the diffusion-weighted signal using a larger number of distributed gradients (Tuch et al., 2002).
HARDI acquisitions allow identifying complex arrangements of fibers as fibers intersections in deep white matter and at subcortical level. Nevertheless, it should be noted that such sequences require longer acquisition and thus are more prone to motion artefacts.
3.4.2 Limitations associated with the fMRI paradigm and analysis
One issue when analyzing fMRI paradigms is to choose a tradeoff between the number of trials presented to participants and the duration of the task. Indeed, a certain amount of
trials are required to obtain a satisfying BOLD signal, but increased the number of trials will drastically increase the duration of the acquisition and thus the presence of motion artefacts.
In Study 5 and 6, for which we used a very simple paradigm in term of cognitive load, we fixed a low number of trials (N=20). Nevertheless, even with a simple task and short dura-tion length, the number of participants excluded based on the task (not enough hit trials) (N=8, 15%) or the motion criteria (N=6, 12%) was elevated. Consequently in Study 2, for which we used a more complex paradigm, we included a high number of trials in order to avoid the exclusion of too many participants (particularly in the group of individuals with 22q11DS). We used two tasks, each encompassing 96 trials. With this procedure, we suc-ceed to keep the majority of our participants on the task criteria (enough number of trials succeeded to analyze the BOLD signal) (N=3, 7%). Nevertheless, once again the number of participants excluded based on the motion criteria were elevated (N=9, 18%). Moreover, as patients with 22q11DS are more prone to motion, for sample size issue, we could not use a stringent threshold for the exclusion of subjects with excessive motion criteria. Indeed, we used a threshold of 3mm for translation and 3 degree for rotation instead of the threshold of 1mm commonly used. Recently, studies demonstrated association between head motion and brain activity (Friston et al., 1996) or connectivity (Ciric et al., 2018), which thus could have influence our results. Although, the 6 head motion parameters (3 for translation and 3 for rotation) were regressed out in the GLM model, we cannot exclude that residual motion artefacts may have influence our results.
3.4.3 Limitations associated with the negative symptoms evaluation
As briefly mentioned in the introduction, the PANSS and SIPS scales were used to measure negative symptoms in this project. Indeed, the importance of negative symptoms in this population being still unknown at the beginning of the Geneva cohort a global measure of the symptoms was chosen. Thereafter, the SIPS, which allow assessing symptoms more precisely in population at risk for psychosis, was introduced. However, as implemented latter on, this scale was not available for all patients of the cohort and therefore could not be used in some of the studies presented in this project. Henceforth, new assessment scales are available to measure negative symptoms according to its current definition. One of this scale, the BNSS (describe in the section 2.3) as thus been added to the Geneva protocol in 2017 and will allow a better assessment of negative symptoms for future studies.