Cognitive function and physiological stress
in adults with 22q11 deletion syndrome
Lucas Lumeij student number: 10353062 Bachelor’s thesis Biomedical Sciences
University of Amsterdam
Supervisor: E.D.A. van Duin Msc. (Department of Psychiatry, Maastricht University; AMC) Co-assessor: Dr. H.J. Krugers (Swammerdam Institute for Life Sciences, UvA)
2
Index
Abstract………...3
Introduction………..4
22q11DS genotype and psychosis………..4
22q11DS genotype and stress………...5
22q11DS genotype, psychosis, stress and cognitive dysfunction……….7
Project aims………8
Methods and materials………..10
Results………14
Study 1………..………..14
Study 2………...19
Study 3……….23
Discussion………26
Acknowledgments……….30
References………..31
Appendices……….36
3
Abstract
The 22q11.2 deletion syndrome (22q11DS) is genetic syndrome that is caused by a hemizygous microdeletion on chromosome 22. It occurs in 1 of 2,000-4,000 live births. The phenotypic expression of 22q11DS is very diverse ranging from life-threatening conditions to only a few pathologies. However, 22q11DS is known as the largest molecular risk factor for psychotic disorders. Moreover, 22q11DS has been associated with multiple cognitive deficits. The study of 22q11DS may provide valuable insights into the genes that play a role in these cognitive deficits and the development of psychosis. Next to 22q11DS stress is also an important risk factor for psychosis. This shows the importance of studying effect of stress on the development of psychosis in 22q11DS. In this study cognitive performance in 22q11DS has been compared to healthy controls. Moreover, the difference in cognitive performance between 22q11DS with two variants (Val and Met) of the catechol-O-methyltransferase (COMT) gene, a gene that plays a large role in prefrontal dopamine metabolism, has been investigated. Also, the associations between psychotic symptoms and cognitive performance in 22q11DS have been investigated. Finally, associations between morning cortisol levels and psychotic symptoms in 22q11DS have been investigated to address the effects of physiological stress on psychosis in 22q11DS.
The 22q11DS group performed significantly worse than the healthy control group on multiple cognitive tasks. Also the Met variant of the COMT gene has been associated with worse cognitive performance in 22q11DS. The associations between psychotic symptoms and cognitive performance in 22q11DS showed mixed results. No correlations were found between morning cortisol levels and psychotic symptoms.
This study provides more insight in 22q11DS and the connections with cognitive dysfunction, stress and psychosis, but more research is needed to further elucidate these connections.
4
Introduction
The 22q11.2 deletion syndrome (22q11DS) is a genetic disorder that occurs in 1 of 2,000-4,000 live births and is caused by a hemizygous microdeletion in the long arm of chromosome 22, usually of a size of 1.5 or 3 megabases (Mb) (Schneider et al., 2014; Karayiorgou et al., 2010). The phenotypic expression of 22q11DS varies largely among patients ranging from life-threatening conditions to only a few features. Among the medical conditions are: cardiac anomalies, palatal anomalies, hypoparathyroidism/ hypocalcemia and subtle dysmorphic facial features (Bassett et al., 2011). Also, 22q11DS patients are reported to have greater rates of various psychiatric disorders than found in the general population. These disorders include autism spectrum disorders (ASD), attention deficit hyperactivity disorder (ADHD), anxiety disorders and psychotic disorders (Jonas et al., 2014; Michaelovsky et al., 2012). Moreover, 22q11DS patients are reported to have certain cognitive impairments. For example, the majority has a borderline intellectual level (IQ, 70-84) (Schneider et al., 2014; Evers et al., 2016; Fung et al., 2015; de Koning et al., 2015). However, also full scale IQ (FSIQ) differs greatly among 22q11DS patients. In some patients the intellectual and cognitive impairments are much worse than in others. Because of this strong variability in the clinical expression it is difficult to recognise the 22q11DS phenotype. Moreover, the clinical symptoms of a patient can change over time (Evers et al., 2009). This is why it is relevant to investigate the genetic and environmental factors that influence the different phenotypical expressions in 22q11DS.
22q11DS genotype and psychosis
Since 20% to 30% of the 22q11DS patients develop schizophrenia or affective psychosis (compared to 1% of the general population) and 1% to 2% of the schizophrenia cases are accounted for by 22q11.2 microdeletions, 22q11DS is considered one of the largest risk factors in developing psychotic disorders (Jonas et al., 2014; Beaton & Simon, 2011; Murphy 2002). Although there are many more risk factors for developing psychotic disorders, it is valuable to look into the genetic risk factors for psychosis that are specific for 22q11DS.
In the 3 Mb deletion approximately 60 known genes and in the 1.5 Mb deletion approximately 35 known genes are deleted (Karayiorgou et al., 2010). One largely investigated gene in this deletion is the catechol-O-methyltransferase (COMT) gene. Because of the deletion, in 22q11DS only one copy of the COMT gene present (haploinsufficiency). One important known function of COMT is the breakdown of dopamine (DA), particularly in the prefrontal cortex (PFC) (Tunbridge et al., 2006). Degradation of DA in other brain regions, for example in the striatum, is mainly done by monoamino-oxidase (MAO) and the dopamine transporter (DAT) (Chen et al., 2004). DAT, however, is relatively scarce in the PFC (Tunbridge et al., 2006). COMT is therefore more important for PFC DA degradation compared to DA degradation in other brain regions. Hence, haploinsufficiency of the COMT gene in 22q11DS causes lower levels of the enzyme that degrades DA and therefore higher DA levels in the PFC (Boot et al., 2011). Additionally, a functional polymorphism in the COMT gene involves a substitution of methionine (Met) for valine (Val) at codon 158 (Jonas et al., 2014). The COMT-Met allele is less stable than the COMT-Val allele at 37°C (normal physiological temperature) and therefore the COMT-Val type has a 40% higher enzymatic activity (Chen et al., 2004). This means that PFC DA levels are supposedly higher in 22q11DS patients because of their COMT haploinsufficiency. Moreover, 22q11DS patients carrying the COMT-Met allele would have even higher PFC DA levels.
This dopaminergic dysfunction in the PFC of 22q11DS patients might play a role in the increased risk for psychotic symptoms. However, there is not yet consensus on the exact effect of the COMT Val158Met polymorphism on the development of psychotic symptoms. Different effects of the
COMT genotype on psychosis risk in 22q11DS are reported. Some studies report that COMT-Met is associated with a higher chance of psychotic symptoms in 22q11DS (Gothelf et al., 2005; Gothelf et al., 2008; Bassett et al., 2007). Other studies report no difference between COMT-Val and COMT-Met with regard to psychosis risk in 22q11DS (Baker et al., 2005). There are even studies that found an
5 association between COMT-Val and higher psychosis risk in 22q11DS (Boot et al., 2011). This means that the effects of COMT Val158Met genotype on psychosis risk are still inconclusive.
The leading hypothesis in the appearance of psychotic symptoms is the presence of abnormally high subcortical DA activity and low levels of PFC DA (Hernaus et al., 2014; Šagud et al., 2010; da Silva Alves et al., 2008). Thus, an imbalance in these regions may play role in psychosis and this imbalance may also be partially affected by the COMT Val158Met polymorphism.
22q11DS genotype and stress
The 22q11.2 deletion is an important genetic risk factor for psychosis. However, there are many other factors that play a role in the development of psychosis. For example, a major environmental risk factor is psychosocial stress (Van Winkel et al., 2008; D Collip et al., 2011; Beaton & Simon, 2011; Reininghaus et al., 2016; Holtzman et al., 2013). This plays a large role in 22q11DS, because patients have additional physiological and psychosocial stressors beyond those of the general population (Beaton & Simon, 2011). Anxiety and stress symptoms are more common in 22q11DS patients, because of genetically derived temperament (Jabbi et al., 2007), early traumatic experiences (e.g. surgery because of medical conditions associated with 22q11DS), and also day-to-day challenges (e.g. medical and social difficulties) connected to the syndrome (Beaton & Simon, 2011). Their cognitive impairments also often make 22q11DS patients more anxious in social interactions during childhood (Swillen et al., 2001). This means that 22q11DS patients experience increased stress compared to the general population because of the syndrome.
Beside the environmental factors there are also genetic factors in 22q11DS that lead to increased stress. For example, beside its role in psychosis, the COMT gene also plays a role in stress reactivity (Armbruster et al., 2012; Hernaus et al., 2013; Stein et al., 2006). In this context the COMT gene is sometimes called the warrior versus worrier gene (Stein et al., 2006). The COMT-Val (warrior) allele is often associated with an advantage in processing aversive stimuli and the COMT-Met (worrier) allele is often associated with an advantage in memory and attention tasks, but also with higher stress reactivity.
In previous research has been found that COMT, beside its role in stress reactivity, plays a role in dopaminergic PFC function (Tunbridge et al., 2006; Boot et al., 2008; Chen et al., 2004; Jonas et al., 2014). This could mean that there is a connection between COMT genotype, PFC function and stress reactivity. This connection can be found in the working of the physiological stress system, so at first this stress system will be described. A physiological stress response involves activation of the hypothalamic-pituitary-adrenal (HPA) axis (Oswald et al., 2004). When activated neurons in the paraventricular nucleus (PVN) of the hypothalamus release corticotropin releasing hormone (CRH). This release stimulates adrenocorticotropic hormone (ACTH) release from the anterior pituitary. Subsequently, ACTH stimulates cortisol release from the adrenal cortex. Cortisol (a glucocorticoid) has profound metabolic effects that are part of the stress response. Also, cortisol causes negative feedback towards both the hypothalamus and the anterior pituitary. These two feedback loops are direct. However, there are also indirect feedback loops: for example, the amygdala exerts mostly excitatory feedback on the PVN of the hypothalamus, while the hippocampus, the anterior cingulate cortex (ACC) and the PFC mainly inhibit the PVN (Zschucke et al., 2015; Dedovic et al., 2009).
The involvement of the PFC in feedback towards the HPA via the hypothalamus may explain this connection between COMT genotype, PFC function and stress reactivity. For example, Zschucke and colleagues found a negative correlation between PFC activity and cortisol levels (Zschucke et al., 2015). Also, decreased activity in the orbitofrontal cortex is found in response to stress (Dedovic et al., 2009). Another important connection between the PFC and the HPA axis is the importance of glucocorticoid feedback towards the PFC. In rat studies the suppression of endogenous glucocorticoids is negatively correlated with dopaminergic transmission in the PFC, leading to impaired working memory (Mizoguchi et al., 2004). Another study associated the COMT-Met form with a higher HPA axis activation in response to naloxone (an opioid receptor antagonist that provokes HPA axis activity) (Oswald et al., 2004). An explanation that is proposed is that the low activity COMT-Met allele is associated with greater catecholaminergic (e.g. dopaminergic) activity. Catecholamine is a major
6 stimulator of CRH release from the PVN and therefore more cortisol is released with greater catecholaminergic activity. In summary, disrupted PFC function may cause disrupted feedback from the PFC towards the HPA axis and vice versa, disrupted HPA axis function may cause dopaminergic dysfunction in the PFC. These findings together display the large interconnectivity between (DA in) the PFC and the HPA axis. In figure 1the HPA feedback loops and the possible connection with the COMT genotype are depicted schematically.
From studies that investigated the connection between psychosis (risk) and cortisol levels multiple conclusions are drawn. Firstly, schizophrenia patients seem to have higher diurnal cortisol levels than healthy controls (Mondelli et al., 2010; Ryan et al., 2004). Also, schizophrenia patients appear to have a lower cortisol peak short after awakening (cortisol awakening response; CAR) (Mondelli et al., 2010). This CAR appears to have an arousal effect on the central nervous system to prepare an individual for the events of the day (Hodyl et al., 2015). In addition, schizophrenia patients show a blunted cortisol response to a psychosocial stressor compared to healthy control (Jansen et al., 1998). The difficulty in these studies is that many psychotic patients receive antipsychotic medication which affect cortisol levels (D Collip et al., 2011). That is why it is useful to investigate subjects that are at risk for developing psychosis, but who do not receive medication. In siblings of psychotic patients higher cortisol levels after a stressful event and higher diurnal cortisol levels are measured compared to healthy controls (D Collip et al., 2011). Moreover, research in non-psychotic subjects shows that exposure to traumatic experiences during childhood correlates with HPA axis activity during adulthood (Heim et al., 2000). Additionally, psychotic patients that were exposed to childhood trauma showed higher emotional and psychotic reactivity to stressful events during adulthood (Lardinois et al., 2011). This shows that early stressful life events can influence stress reactivity later in life. On a physiological level, greater stress reactivity and state anxiety has been associated with a blunted CAR (Walker et al., 2011). This is similar to what has been observed in psychosis.
Figure 1: Schematic representation of the feedback loops of the hypothalamic-pituitary-adrenal (HPA) axis. This is a suggestion of how COMT genotype is connected with the physiological stress system. On the left side the influence of 22q11DS and the COMT genotype on dopamine in the prefrontal cortex is depicted. Disrupted dopaminergic prefrontal cortex function can affect its inhibitory function on the HPA axis. COMT: catechol-O-methyltransferase; ACC: Anterior Cingulate Cortex; CRH: corticotropin releasing hormone; ACTH: adrenocorticotropic hormone
7 In conclusion, because stress plays such a large role in the development of psychosis it is useful to investigate (biological) stress reactivity in the 22q11DS group, because they are genetically at risk for psychosis. The non-psychotic 22q11DS patients are interesting in this respect, because unlike psychotic patients they do not receive medication. However, more important is the clear 22q11.2 deletion that provides a very interesting model to investigate the underlying mechanism of psychosis and the influence stress can have on developing psychosis.
22q11DS genotype, psychosis, stress and cognitive dysfunction
As found in previous studies, the 22q11DS patients have on average a lower FSIQ than the general population and they are reported to have a range of cognitive deficits (Schneider et al., 2014; Evers et al., 2016; Fung et al., 2015; de Koning et al., 2015). While relatively few research is done in cognitive function in 22q11DS there are certain cognitive deficits found in 22q11DS compared to the general population. These deficits are mostly observed in these domains: executive function, visuospatial function, attention, episodic memory and social cognition (Woodin et al., 2001; Antshel et al., 2010; Bearden et al., 2004; Yi et al., 2016; Baker et al., 2005). In this study the explanation for this cognitive dysfunction in 22q11DS will be approached from three angles: the 22q11DS genotype, the similarities between 22q11DS and psychosis and the increased stress that is associated with 22q11DS. Firstly, the role of the 22q11DS genotype on cognitive dysfunction will be addressed. While many different genes may play a role in this dysfunction the focus of this study will be on the COMT gene. As described in previous studies, the COMT haploinsufficiency in 22q11DS plays a role in cognitive dysfunction because of its role in disrupted dopaminergic PFC function (Gothelf et al., 2005; van Amelsvoort et al., 2008). Multiple studies suggested the crucial role of PFC DA in cognitive function (Cropley et al., 2006; Bäckman et al., 2006). A model that could explain the effect of PFC DA levels on cognitive function is the inverted-U-shaped model (Goldman-Rakic et al., 2000). According to the inverted-U-shaped model, PFC DA levels have to be in balance for a good cognitive function. This means that both too low and too high DA levels may cause cognitive dysfunction. Because of the COMT haploinsufficiency too high DA levels in the PFC are more often expected in 22q11DS. Thus, in 22q11DS, imbalance in PFC DA might be a cause for cognitive dysfunction.
Moreover, the COMT Val158Met polymorphism plays a role in cognitive function. In multiple
studies the high activity COMT-Val has been associated with worse cognitive function than COMT-Met (Tunbridge et al., 2006; Egan et al., 2001). The explanation has been attributed to too low PFC DA levels. These studies, however, have been conducted in subjects that were not hemizygous for the COMT gene. In studies in 22q11DS the COMT Val158Met polymorphism shows different effects. In
certain 22q11DS studies COMT-Met has been associated with worse overall cognitive function (Gothelf et al., 2005; Baker et al., 2005), while in others COMT-Val has been associated with worse overall cognitive function (Bearden et al., 2004). Moreover, in 22q11DS research, tendencies have been found of lower FSIQ in COMT-Met carriers than in COMT-Val carriers (Baker et al., 2005; Bearden et al., 2004; Gothelf et al., 2005). The worse cognitive performance in COMT-Met compared to COMT-Val could also be explained by the inverted-U-shaped model. In 22q11DS the PFC DA levels are already expected to be high, because of the COMT hemizygosity. In carriers of the less active COMT-Met allele these PFC DA levels would be even higher and further away from the optimum where there is a good balance in PFC DA.
Secondly, because 22q11DS is such a large risk factor for psychosis, there might be similarities in cognitive function between 22q11DS and psychosis. In non-22q11DS schizophrenia patients cognitive impairment is considered an important feature too (Šagud et al., 2010). In schizophrenia dopaminergic dysfunction plays a role in worsened cognition as well. Therefore, in schizophrenia patients the effect of COMT genotype on cognitive function has been investigated. In these studies the COMT-Met allele was positively correlated with cognitive performance in certain domains (e.g. working memory, executive functioning and processing speed), but the COMT-Met was also negatively correlated with other domains (cognitive flexibility and switching) (Twamley et al., 2014). Important to consider is that the subjects in these studies were not hemizygous for COMT, like in 22q11DS. Also, similarities are found in cognitive dysfunction between 22q11DS and non-22q11DS subjects who are
8 at high risk for schizophrenia (Baker et al., 2005). This means that a similar neurodevelopmental disruption could be present in 22q11DS patients and persons who are at (familiar) higher risk for schizophrenia.
Thirdly, Beside the 22q11DS genotype and psychosis, stress is also thought to be associated with cognitive dysfunction (Shansky & Lipps, 2013; Aas et al., 2011). Glucocorticoids, that are released during the stress response, are reported to have effects on cognitive function via interactions with the catecholamines, DA and noradrenaline (NA), in non-human primates and rodents (Shansky & Lipps, 2013). Glucocorticoids affect DA and NA levels by blocking catecholamine transporters. This leads to increased extracellular DA and NA levels in the PFC. These high levels of DA and NA can overstimulate the dopamine and noradrenergic receptors and therefore cause PFC dysfunction.
Furthermore, an association between abnormal HPA axis activity (a blunted CAR) and cognitive dysfunction in psychotic patients is found (Aas et al., 2011). However, the direction of this effect is unknown. Namely, it could be the case that abnormal hippocampus function in these patients accounts for both the HPA axis (via feedback towards the hypothalamus) and cognitive dysfunctions. Moreover, a meta-analysis study found a negative effect of acute increases in cortisol on working memory, but a positive effect on inhibition and no effect on set-shifting (Shields et al., 2015). Finally, also an effect of early life stress on cognitive function was found in healthy adults. Exposure to childhood trauma seems have an negative effect on working and long-term memory in adulthood in subjects without significant anxiety or depression symptoms (Majer et al., 2010).
In conclusion, 22q11DS genotype, psychosis and stress all have effects on cognitive function. There is overlap in the way these three things affect cognitive function. For example, in 22q11DS, psychosis and stress disruptions in dopaminergic PFC function are found. This overlap is not surprising, since both 22q11DS and stress are risk factors for developing psychosis, so similar neuronal dysfunctions are expected.
Project aims
As shown in previous research cognitive dysfunction is associated with both 22q11DS and psychosis. However, fairly few research is done in cognitive function in 22q11DS and the effects of the COMT Val158Met polymorphism on cognitive function are still unclear. Moreover, stress has been
shown to increase psychosis risk and 22q11DS has also been associated with increased stress reactivity. That is why it is important to investigate physiological stress in 22q11DS and investigate whether there is an association with psychosis risk. This study has been divided in three sub studies to assess these questions.
Study 1: 22q11DS genotype and cognitive task performance
In the first study cognitive task performance was compared between 22q11DS subjects and healthy controls. Based on previous research it was hypothesised that 22q11DS subjects perform worse on these tasks than the healthy controls. Moreover, because 22q11DS is associated with increased stress reactivity it was expected that the biggest difference will be found in the more stressful cognitive tasks. Within the 22q11DS group an exploratory analysis has been conducted to investigate the difference in cognitive task performance between COMT-Val and COMT-Met carriers. Because of worsened PFC DA degradation in the COMT-Met it was hypothesised that the COMT-Met group would perform worse on the cognitive tasks than the COMT-Val group.
Study 2: (prodromal) psychotic symptoms and cognitive task performance
In the second study associations between (prodromal) psychotic symptoms and cognitive task performance were investigated within the 22q11DS group. Because cognitive dysfunction has been associated with psychosis in previous studies it was hypothesised that (prodromal) psychotic symptoms would be associated with worse cognitive task performance.
9 In the third study associations between morning cortisol levels and (prodromal) psychotic symptoms were investigated. Because in previous studies psychosis has been associated with blunted morning cortisol levels it was hypothesised that lower morning cortisol levels would positively correlate with (prodromal) psychotic symptoms. Moreover, the associations between morning cortisol levels and early life trauma and state anxiety were investigated. Earlier research shows that early life trauma can increase stress reactivity and state anxiety during adulthood. Additionally, greater state anxiety and stress reactivity were associated with blunted morning cortisol levels. That is why it was hypothesised that early life trauma and state anxiety would be negatively correlated with morning cortisol levels.
10
Methods and materials
Subjects
Study 1: 22q11DS genotype and cognitive task performance
For a comparison in cognitive task performance between the 22q11DS group and the healthy control 26 adult 22q11DS subjects were included and 26 sex and age matched healthy controls.
19 adult 22q11DS subjects were included for a post hoc study of cognitive task performance in the COMT genotypes (COMT-Val and COMT-Met). In these subjects the COMT Val158Met
polymorphism was determined.
Study 2: (prodromal) psychotic symptoms and cognitive task performance
44 adults with 22q11DS were included to study the association between (prodromal) psychotic symptoms and cognitive task performance.
Study 3: morning cortisol and (prodromal) psychotic symptoms and childhood trauma
28 22q11DS subjects were included in a cortisol sampling study. In these subjects morning cortisol samples were collected. Also, questionnaires were assessed to determine (prodromal) psychotic, state anxiety and childhood trauma scores.
To participate in the study all 22q11DS subjects had to: I) have a deletion 22q11.2 deletion confirmed by Fluorescence In Situ Hybridisation (FISH), Multiplex Ligation-dependent Probe Amplification (MLPA) or micro-array analysis; II) have an age between 18 and 65 years; III) be able to give informed consent; IV) give informed consent.
The participants were recruited via the 22q11 inpatient clinic of the University Hospital Maastricht, the University Medical Centre Utrecht and the University Hospital Leuven. Additionally, participants were recruited via the Dutch 22q11 foundation (Stichting Steun 22q11) and by advertising in the Dutch 22q11 newsletter. Each participant gave written informed consent after being informed on the full study procedure. The protocol was approved by the Ethics Committee of Maastricht University.
Clinical assessment
In both 22q11DS subjects and healthy control subjects FSIQ was estimated using a shortened version of the Wechsler Adult Intelligence Scale – III (WAIS-III) (Canavan et al., 1986). Furthermore, schizophrenia symptoms were assessed using the Positive and Negative Symptom Scale (PANSS) (Kay et al., 1987). Also, the 16-item version of the Prodromal Questionnaire (PQ-16) was used to assess psychosis risk via prodromal symptoms (Loewy et al., 2005; Ising et al., 2012). Beside the prodromal symptoms the distress associated with these symptoms was also assessed (PQ distress). The PANSS and the PQ-16 were only assessed in the 22q11DS group.
The 28-item version of the Childhood Trauma Questionnaire (CTQ) was used to screen for maltreatment histories during childhood (Bernstein et al., 2003). It measures five subscales: physical abuse and neglect, emotional abuse and neglect and sexual abuse. In addition, a short version of the Retrospective Bullying Questionnaire was used to measure bullying experience during childhood (Schäfer et al., 2004). Also, the State-Trait Anxiety Inventory (STAI) was used to measure state anxiety (Spielberger 1985). The CTQ, Bullying Questionnaire and the STAI were only assessed in the 28 22q11DS subjects that participated in the cortisol sampling study.
For all questionnaires and test methods Dutch versions were used. Study procedure
The assessment of the participants’ data was mostly done at their homes. The WAIS-III and the PANSS were assessed by the researcher, while the PQ-16, the CTQ, the bullying questionnaire and the STAI were filled in by the participants themselves. Furthermore, cognitive performance was assessed
11 using the Cambridge Neuropsychological Test Automated Battery (CANTAB) under supervision of the researcher. Moreover, from all 22q11DS subjects blood samples were taken to analyse the genotype. This was done at their homes or in a clinic when necessary. Finally, in participants that were considered capable a six-day cortisol sampling study was assessed. During these six days the proceedings were monitored by the researcher by telephone.
Genotyping
From all 22q11DS subjects blood samples were collected. DNA was isolated from the blood using the Nucleospin Blood L Kit (Macherey-Nagel, Düren, Germany). The kit consisted of buffer BQ1 (lysis leukocytes), protease K (lysis cell nuclei), buffer 50% BQ1 – 50% ethanol (bind DNA to column), buffer B5 (denaturation of free proteins), buffer BW (washing buffer) and buffer BE (elution buffer, at 65°C).
Subsequently, the DNA analysis was performed as previously described (de Koning et al., 2012). COMT Val158Met polymorphism (rs4680, CCAGCGGATGGTGGATTTCGCTGG C[A/G]T GAAGGACAAGGTGTGCATGCCTGA) was determined with 5’-nuclease TaqMan assay (Life Technologies, Foster City, CA, USA). For this particular single-nucleotide polymorphism (SNP) specific TaqMan probes were used (C. 25746809 – A/G). The probes were labelled with either fluorescent VIC or 6-FAM on the 5’-end and with a non-fluorescent quencher on the end. A minor groove binder (MGB) on the 3’-end stabilises the double-stranded structure that is formed between the target and the probe. The TaqMan genotyping assay is depicted in figure 2.
The genotyping was carried out using a LC-480 384-well Lightcycler (Roche Diagnostics, Mannheim, Germany). The conditions consisted of 1 cycle of 10 min at 95°C for denaturation, 40 cycles of 1 min at 60°C for amplification and 1 cycle of 1 min at 37°C for cooling. All DNA samples were genotyped in duplicate and H2O samples were added as controls. To analyse end point fluorescence
the Lightcycler LC-480 Software release 1.5.0 (Roche Diagnostics, Basel, Switzerland) was used.
Figure 2: Schematic depiction of TaqMan genotyping assay. Retrieved from: http://www.dnavision.com/taqman-genotyping-assays.php
12 Salivary cortisol
28 22q11DS and 24 healthy control subjects were given an electronic momentary assessment technology device (PsyMate) (figure 3) (Myin-Germeys et al., 2011). On each day over a period of 6 consecutive days the PsyMate emitted up to 10 ‘beep’ signals at random moments between 07:30 and 22:30. The participants were asked to fill in a small questionnaire to assess their emotional state (these data are not used for this paper). During these questionnaires the participants collected a saliva sample using a cotton swab (Salivette, Sarstedt, The Netherlands). These swabs were placed in a salivette tube and the exact collection times were recorded by the participant. The samples were stored in the participants’ home
freezers before being moved to the laboratory where they were kept at -20°C until analysis.
Cortisol was analysed from the saliva samples in duplicate using radio-immunoassays. A tracer solution 3CMO coupled with 2-[125 I]histamine and specific antibodies raised against cortisol-3CMO-BSA was used (D Collip et al., 2011). The salivary cortisol analyses were conducted at Dresden University of Technology in Germany.
In this study morning cortisol levels were measured. Samples that were collected within 1 hour after awakening were included. For each individual an average of the earliest morning cortisol measurements of each day was made. Before analysis the cortisol values were log transformed to reduce skewness of distribution (D Collip et al., 2011).
Cognitive task performance
To determine cognitive performance tests were assessed at the participants’ homes using the CANTAB (Cambridge Cognition, Bottisham, UK) (figure 4) (Fray et al., 1996). This is a computerised set of neuropsychological tasks. The CANTAB can be used to study a variety of cognitive functions and it has been used for a wide range of clinical populations (Roque et al., 2011). The main functions that can be examined by the CANTAB are executive functioning and working memory (Roque et al., 2011; Fray et al., 1996). In this study 8 different CANTAB tasks are assessed: Paired Associates Learning
(PAL), Verbal Recognition Memory (VRM), Rapid Visual Information Processing (RVP), Spatial Working Memory (SWM), One Touch Stockings of Cambridge (OTS), Reaction Time (RTI), Intra-Extra Dimensional Set Shift (IED) and Emotional Recognition Task (ERT). The different procedures of the tasks and the outcome measures that were used are described in box 1. For all tasks the Dutch versions were used.
Statistical analysis
For differences between groups in sex distribution the Fisher’s exact test was used. For differences in age, clinical data and cognitive performance scores between groups an analysis of variance (ANOVA) was conducted. When FSIQ was included as a covariate an analysis of covariance (ANCOVA) was conducted. For correlations between clinical scores and cognitive performance scores Pearson’s product-moment correlation (Pearson’s r) was conducted, providing a correlation coefficient (r). Where necessary in Pearson’s r FSIQ was also introduced as a covariate, providing a partial correlation coefficient. Also, for correlations between morning cortisol, early life trauma and clinical scores Pearson’s r was conducted. To test for normal distributed data the Shapiro-Wilk test was used. A probability value of 0.05 was chosen as the significance value.
All statistical analyses were conducted in R version 3.2.4 (revised) for Windows (R Foundation, Vienna, Austria) and RStudio version 0.99.893 for Windows (RStudio Inc., Boston, MA, USA).
Figure 3: PsyMate
13
Box 1: Procedures of the cognitive CANTAB tasks and the used outcome measures
h) g) f) e) d) c) b) a)
a) Paired Associates Learning (PAL)
Boxes are displayed on the screen and are opened in a randomised order. One or more boxes contains a pattern. One at a time the patterns are displayed in the middle of the screen and the participant must touch the box where the pattern was located. The patterns are re-presented if an error is made. The difficulty increases through the test.
Used outcome measure: mean errors of each stage before all patterns were located correctly (mean errors to success).
b) Verbal Recognition Memory (VRM)
The participant is shown a list of words and then asked to recall as many words as possible. After that they are asked to recognise the words from a new list containing the original words and distractors. Used outcome measure: total correctly recalled words in the ‘free recall’ phase (free recall – total correct).
c) Rapid Visual Information Processing (RVP)
Digits, ranging from 2 to 9, appear in a pseudo-random order at the rate of 100 digits per minute. Participants are asked to detect a target sequence of digits (e.g. 2-4-6, 3-5-7, 4-6-8) and press a pad when registered.
Used outcome measure: a signal detection measure of sensitivity to errors using the hits and false alarms (A).
d) Spatial Working Memory (SWM)
A number of coloured boxes are shown on the screen. The participant has to find a ‘token’ in each of the boxes. The token appears only once in every box. Through the stages the number of boxes increases gradually.
Used outcome measure: total errors made in the entire task (total errors). e) One Touch Stockings of Cambridge (OTS)
Two patterns of three coloured balls are shown. They are displayed in a way that it is clear that the balls are stacked. The participant has to copy the upper pattern by moving the balls in the lower pattern. After a few practice rounds the participant must work out in his head how many moves it takes to solve the problem. This number has to selected.
Used outcome measures: number of choices before the right number is selected (choices to correct) and the latency before the right number is selected (latency to correct).
f) Reaction Time (RTI)
The participant has to keep his finger on a press pad and tap yellow dot when it is displayed and then go back to the press pad. In the first stage a yellow dot appears in one location and in the second stage the dot can appear in five different locations.
Used outcome measures: the time between the presentation of the dot and the release of the press pad in the five choice stage (five choice reaction time in ms) and the time between the release of the press pad and the touch of the yellow dot in the five choice stage (five choice movement time in ms). g) Intra-Extra Dimensional Set Shift (IED)
Firstly, two different colour-filled shapes are presented. The participant has to select the correct stimulus. Feedback teaches which stimulus is correct. After six correct responses the rule changes and another stimulus becomes correct. After a while other stimuli are presented and the participant has to use other criteria to select the correct stimulus.
Used outcome measure: total errors adjusted for the fact that a participant has less opportunity to make errors when failing at any stage (total errors {adjusted}).
h) Emotional Recognition Task (ERT)
Multiple faces expressing different are displayed, each for 200 ms. The participant has to select the right emotion choosing from 6 options.
14
Results
Study 1: 22q11DS genotype and cognitive task performance
1.1 Demographic and clinical data
In table 1 the demographic data of the 22q11DS subjects and the healthy controls are displayed. FSIQ was significantly lower in the 22q11DS group (ANOVA, p<0.001). Age and sex did not differ significantly between 22q11DS and the healthy controls.
For 19 subjects the COMT Val158Met polymorphism was determined. In table 2 the
demographic and clinical data of the 22q11DS COMT-Val and the 22q11DS COMT-Val group are displayed. 4 of the 19 subjects had the Val genotype, while 15 of the 19 subjects had the COMT-Met genotype. Age, sex and the clinical data (PQ, PQ distress and PANSS total) did not differ significantly between the COMT-Val and the COMT-Met group. FSIQ was significantly lower in the COMT-Met group (p=0.05) (figure 5).
Table 1: Demographic data of adults with 22q11DS and healthy controlsa,b
22q11DS Healthy controls Total F df pc
N 26 (50%) 26 (50%) 52 (100%)
Age 28.2 (7.5) 25.2 (4.2) 26.7 (6.2) 3.19 1 0.08
Sex (M/F) 14/12 (54%/46%) 14/12 (54%/46%) 28/24 (54%/46%) - - 1.00
Full scale IQ 74.9 (11.3) 107.3 (16.0) 91.1 (21.3) 71.16 1 <0.001***
a 22q11DS: 22q11 deletion syndrome.
b Age and full scale IQ presented as means and standard deviations in parentheses (analysis of variance, ANOVA); sex presented as frequency data (Fisher’s exact test). c *: p-value ≤ 0.05; **: p-value < 0.01; ***: p-value < 0.001.
Table 2: Demographic and clinical data of adults with 22q11DS grouped according to catechol-O-methyltransferase (COMT) Val158Met genotypea,b
a 22q11DS: 22q11 deletion syndrome; PQ: Prodromal Questionnaire; PANSS: Positive and Negative Symptom Scale.
b Age, full scale IQ, PQ, PQ distress and PANSS total presented as means and standard deviations in parentheses (analysis of variance, ANOVA); Sex presented as frequency data (Fisher’s exact test). c *: p-value ≤ 0.05; **: p-value < 0.01; ***: p-value < 0.001.
COMT-Val COMT-Met Total F df pc
N 4 (21%) 15 (79%) 19 (100%) Age 32.8 (3.6) 35.3 (10.1) 34.7 (9.1) 0.23 1 0.64 Sex (M/F) 2/2 (50%/50%) 4/11 (27%/73%) 6/13 (32%/86%) - - 0.56 Full scale IQ 81.5 (10.0) 72.7 (6.6) 74.6 (8.0) 4.57 1 0.05* PQ 3.5 (3.1) 3.3 (2.5) 3.3 (2.5) 0.02 1 0.89 PQ distress 5.5 (3.4) 8.6 (7.7) 7.9 (7.0) 0.59 1 0.45 PANSS total 36.7 (2.3) 38.1 (7.4) 37.8 (6.6) 0.10 1 0.75
15
Figure 5: Full scale IQ (FSIQ) of 22q11DS COMT-Val group (n=4) and COMT-Met group (n=15). The FSIQ of the COMT-Val group is significantly higher than the FSIQ of the COMT-Met group (p=0.05).
1.2 Cognitive task performance in 22q11DS compared with healthy controls
In table 3 the scores on the cognitive tasks, measured with the CANTAB, of the 22q11DS and the healthy control group are displayed. There has been verified whether age or FSIQ correlated with the scores on each cognitive task. There has also been verified whether the scores on the cognitive tasks differed between the sexes. FSIQ correlated significantly with the scores in every cognitive task except for RTI five choice reaction time (RTI5-rt) (Appendix 1). Age correlated with none of the scores on the tasks and there were no significant differences found in scores between the sexes. Because it correlated with score on most of the tasks FSIQ was considered a covariate, so the statistical values displayed in table 3 are controlled for FSIQ (see Appendix 2 for a table in which there is not controlled for FSIQ).
After controlling for FSIQ significant differences between the 22q11DS and the healthy control group were found in the following outcome measures: VRM total correct (p=0.002), RVP A (p<0.001), OTS latency to correct (p=0.002) and ERT total number correct (p<0.001). Also, for RVP A an interaction effect was found between group (22q11DS or healthy control) and FSIQ (p=0.01). This means that beside the fact that FSIQ was significantly correlated with RVP A outcome, FSIQ also has a different correlation with RVP A depending on the group. Namely, FSIQ was significantly correlated with RVP A in the 22q11DS group (p=0.004), but not in the healthy control group (p=0.97).
The RVP A scores and the OTS latency to correct scores of both the 22q11DS group and the healthy control group are depicted in figures 6 and 7 (the figures of the scores on the other tasks can be found in Appendix 3). The healthy control group scores significantly higher on RVP A than the 22q11DS group (p<0.001) meaning the healthy control group performed better on detecting the target sequence. Furthermore, the healthy control group had a significantly lower OTS latency to correct score than the 22q11DS group (p=0.002), so it took the healthy control group less time to find the correct answer than the 22q11DS group.
16
Table 3: Cognitive task performance of adults with 22q11DS and healthy controlsa,b
a 22q11DS: 22q11 deletion syndrome; PAL: Paired Associates Learning; VRM: Verbal Recognition Memory; RVP: Rapid Visual Information Processing; SWM: Spatial Working Memory; OTS: One Touch Stockings of Cambridge; RTI: Reaction Time; IED: Intra-Extra Dimensional Set Shift; ERT: Emotion Recognition Task.
b Scores on cognitive tasks presented as means and standard deviations in parentheses (analysis of covariance, ANCOVA). c P-values corrected for full scale IQ. d *: p-value ≤ 0.05; **: p-value < 0.01; ***: p-value < 0.001.
Figure 6: Rapid Visual Information Processing (RVP) A of 22q11DS (n=26) and healthy control group (n=26). RVP A is a signal detection measure of sensitivity to errors representing how well a participant can detect a target sequence (ranging from 0.00 to 1.00). The healthy control group had a significantly higher RVP A score than the 22q11DS group (p<0.001).
22q11DS Healthy controls Total F df pc,d
N (%) 26 (50%) 26 (50%) 52 (100%)
PAL mean errors to success 5.2 (3.9) 2.5 (4.0) 3.9 (4.2) 7.3 1 0.47
VRM free recall - total correct 5.4 (2.2) 8.7 (2.0) 7.0 (2.7) 32.6 1 0.02*
RVP A; hits and false alarms 0.83 (0.04) 0.95 (0.03) 0.89 (0.07) 125.15 1 <0.001***
SWM total errors 16.2 (10.7) 7.7 (7.9) 11.9 (10.2) 11.26 1 0.38
OTS choices to correct 1.9 (0.4) 1.3 (0.3) 1.6 (0.5) 43.24 1 0.58
OTS latency to correct in ms 44714 (21577) 21013 (6646) 32864 (19826) 28.09 1 0.002**
RTI reaction time in ms 304.0 (45.1) 300.0 (28.9) 302.0 (37.6) 0.14 1 0.69
RTI movement time in ms 294.6 (78.1) 219.6 (40.6) 257.1 (72.3) 19.74 1 0.16 IED total errors {adjusted} 32.1 (26.8) 11.2 (5.3) 21.6 (21.8) 16.46 1 0.40 ERT total number correct 85.3 (22.7) 127.2 (10.8) 105.8 (27.6) 80.02 1 <0.001***
17
Figure 7: One Touch Stockings of Cambridge (OTS) latency to correct (in ms) of 22q11DS (n=26) and healthy control group (n=26). OTS latency to correct represents how much time it took on average before the subject selected the correct answer in each stage. The 22q11DS group had a significantly higher latency than the healthy control group (p=0.002).
1.3 Exploratory post hoc analysis: cognitive task performance within 22q11DS: COMT-Val compared with COMT-Met genotype
In table 4 the scores on the cognitive tasks, measured with the CANTAB, of the 22q11DS COMT-Val and the 22q11DS COMT-Met carriers are displayed. Again, there has been verified whether age or FSIQ correlated with cognitive task performance and there has been verified whether task performance differed between the sexes. In this group of 19 subjects FSIQ and age did not correlate with cognitive performance. Moreover, no differences in score on the tasks between the sexes were found. However, since FSIQ correlated with the scores in most cognitive tasks in a larger sample size, there was decided to introduce FSIQ as a covariate. The statistical values displayed in table 4 are therefore controlled for FSIQ (see Appendix 4 for a table in which there is not controlled for FSIQ).
After controlling for FSIQ the scores on the PAL task differed significantly between the COMT-Val and the COMT-Met group (p=0.004). The COMT-COMT-Val group made on average significantly less errors before a stage was completed (figure 8).
18
Table 4: Cognitive task performance of adults with 22q11DS grouped according to catechol-O-methyltransferase (COMT) Val158Met genotypea,b
a 22q11DS: 22q11 deletion syndrome; PAL: Paired Associates Learning; VRM: Verbal Recognition Memory; RVP: Rapid Visual Information Processing; SWM: Spatial Working Memory; OTS: One Touch Stockings of Cambridge; RTI: Reaction Time; IED: Intra-Extra Dimensional Set Shift; ERT: Emotion Recognition Task.
b Scores on cognitive tasks presented as means and standard deviations in parentheses (analysis of covariance, ANCOVA). c P-values corrected for full scale IQ. d *: p-value ≤ 0.05; **: p-value < 0.01; ***: p-value < 0.001.
Figure 8: Paired Associates Learning (PAL) mean errors to success of 22q11DS COMT-Val group (n=4) and COMT-Met group (n=15). The PAL mean errors to success represent how many errors on average were made before each stage was completed. The COMT-Val group made significantly less errors than the COMT-Met group (p=0.004).
COMT-Val COMT-Met Total F df Pc,d
N (%) 4 (21%) 15 (79%) 19 (100%) - - -
PAL mean errors to success 2.2 (0.9) 7.2 (2.8) 6.1 (3.3) 11.64 1 0.004**
VRM free recall - total correct 5.3 (1.7) 6.0 (1.4) 5.8 (1.4) 1.17 1 0.30
RVP A 0.85 (0.07) 0.82 (0.05) 0.83 (0.05) 0.008 1 0.93
SWM total errors 8.5 (10.1) 20.1 (9.2) 17.6 (10.3) 3.61 1 0.08
OTS choices to correct 1.7 (0.4) 2.0 (0.5) 1.9 (0.5) 0.73 1 0.41
OTS latency to correct in ms 52342 (24162) 41050 (20416) 43427 (21068) 0.11 1 0.75 RTI reaction time in ms 278.4 (34.9) 308.7 (53.9) 302.3 (51.3) 0.28 1 0.61 RTI movement time in ms 287.5 (93.7) 289.2 (78.74) 288.8 (79.3) 0.26 1 0.62 IED total errors {adjusted} 24.5 (27.1) 28.3 (15.2) 27.5 (17.5) 0.37 1 0.55
ERT total number correct 81.3 (18.6) 81.6 (31.0) 81.5 (28.4) 0.71 1 0.41
19
Study 2: (Prodromal) psychotic symptoms and cognitive task performance
2.1 Demographic and clinical data
In table 5 the demographic and clinical data are displayed of the 22q11DS subjects that were included in the study that investigated associations between (prodromal) psychotic symptoms and cognitive task performance. The study was divided in three sub analyses investigating the prodromal symptoms (PQ and PQ distress) and the positive and negative psychotic symptoms (PANSS). For the 44 included subjects 42 PQ and PQ distress and 37 PANSS scores were available. PQ and PQ distress were not significantly correlated with age, FSIQ or PANSS. Also, PQ and PQ distress scores did not differ between the sexes. PQ and PQ distress where, however, significantly positively correlated with each other (p<0.001). PANSS was not significantly correlated with age, PQ or PQ distress and the PANSS score did not differ significantly between the sexes. However, there was a significant negative correlation between PANSS score and FSIQ (p=0.05) (figure 9). This means that a higher PANSS was associated with a lower FSIQ.
Table 5: Demographic and clinical data of adults with 22q11DS and correlations with Prodromal questionnaire (PQ), PQ distress and Positive and Negative Symptom Scale (PANSS) scoresa,b
PQ analysis PQ distress analysis PANSS analysis
Mean (SD) pc Mean (SD) pc Mean (SD) pc
N 42 - 42 - 37 - Age 31.8 (11.0) 0.26 31.8 (11.0) 0.24 31.7 (10.7) 0.31 Sex (M/F) 13/29 (31%/69%) 0.28 13/29 (31%/69%) 0.35 13/24 (35%/65%) 0.68 Full scale IQ 74.7 (10.5) 0.69 74.7 (10.5) 0.71 74.6 (10.6) 0.05* PQ 2.5 (2.2) - 2.5 (2.2) <0.001*** 2.3 (2.0) 0.99 PQ distress 6.3 (6.3) <0.001*** 6.3 (6.3) - 5.1 (4.6) 0.27 PANSS total 39.4 (9.2) 0.99 39.4 (9.2) 0.27 39.5 (9.1) -
a 22q11DS: 22q11 deletion syndrome. b Age, full scale IQ, PQ, PQ distress and PANSS total presented as means and standard deviations in parentheses (Pearson’s product-moment correlation, Pearson’s r); sex presented as frequency data (analysis of variance, ANOVA). c *: p-value ≤ 0.05; **: p-value < 0.01; ***: p-value < 0.001.
20
Figure 9: Correlation between Positive and Negative Symptom Scale (PANSS) and full scale IQ (FSIQ) in 22q11DS (n=37). There was a significant negative correlation between PANSS and FSIQ (p=0.05).
2.2 Associations between (prodromal) psychotic symptoms and cognitive task performance
In table 6 the correlations between respectively, PQ, PQ distress and PANSS with the scores on the cognitive tasks, measured with the CANTAB, are displayed. Again, there has been verified whether FSIQ correlated with the scores on the cognitive tasks. FSIQ was significantly correlated with VRM total correct, RVP A, SWM total errors. OTS choices to correct RTI five choice movement time and IED total errors (adjusted) (Appendix 5). Therefore, in these analyses FSIQ has been introduced as covariate as well.
After controlling for FSIQ, PQ was significantly negatively correlated with OTS latency to correct (p=0.03) (figure 10). Additionally, PQ distress was significantly negatively correlated with OTS latency to correct (p=0.02) (figure 11). This means that a higher PQ and PQ distress were associated with better performance on the OTS task. Moreover, PQ distress was significantly positively correlated with RTI five choice reaction time (p=0.02) (figure 12). Meaning that a higher PQ distress score was associated with worse reaction time in the five choice RTI task. PANSS had no significant correlations with any of the cognitive task scores.
21 M ea n ( SD ) P ar ti al c o rr el at io n c o ef fi ci en t c, d M ea n ( SD ) P ar ti al c o rr el at io n c o ef fi ci en t c, d M ea n ( SD ) P ar ti al c o rr el at io n c o ef fi ci en t c, d PA L m ea n er ro rs to s uc cess 5. 6 (3 .6 ) -0 .2 2 (t=-1. 40 ) ( p=0 .1 7) 5. 6 (3 .6 ) -0 .0 8 (t=-0. 52 ) ( p=0 .6 1) 5. 7 (3 .8 ) 0. 20 (t=1 .1 8) (p =0 .2 5) V R M f ree r ec al l to ta l c o rr ec t 5. 5 (2 .1 ) -0 .1 7 (t=-1. 10 ) ( p=0 .2 8) 5. 5 (2 .1 ) -0 .2 0 (t=-1. 25 ) ( p=0 .2 2) 5. 6 (2 .0 ) -0 .1 6 (t=-0. 96 ) ( p=0 .3 5) R V P A 0. 84 (0 .0 5) -0 .2 3 (t=-1. 49 ) ( p=0 .1 4) 0. 84 (0 .0 5) -0 .1 4 (t=-0. 86 ) ( p=0 .3 9) 0. 84 (0 .0 5) -0 .1 7 (t=-0. 98 ) ( p=0 .3 3) SW M to ta l er ro rs 20 .0 (1 0. 6) 0. 03 (t=0 .1 7) (p =0 .8 6) 20 .0 (1 0. 6) 0. 08 (t=0 .5 3) (p =0 .6 0) 19 .2 (1 0. 7) 0. 04 (t=0 .2 5) (p =0 .8 0) O TS c ho ic es to c o rr ec t 1. 9 (0 .4 ) 0. 19 (t=1 .2 0) (p =0 .2 4) 1. 9 (0 .4 ) 0. 16 (t=1 .0 3) (p =0 .3 1) 1. 9 (0 .4 ) 0. 22 (t=1 .3 4) (p =0 .1 9) O TS la ten cy to c o rr ec t in m s 46 37 9 (2 22 75 ) -0 .3 4 (t =-2. 25 ) (p =0 .0 3) * 46 37 9 (2 22 75 ) -0 .3 6 (t =-2. 41 ) (p =0 .0 2) * 47 16 4 (2 17 49 ) -0 .0 4 (t=-0. 24 ) ( p=0 .8 1) R TI r ea cti o n ti m e in m s 30 4. 4 (4 1. 5) 0. 30 (t=1 .9 7) (p =0 .0 6) 30 4. 4 (4 1. 5) 0. 36 ( t=2 .3 5) ( p =0 .0 2) * 30 4. 0 (3 8. 4) 0. 07 (t=-0. 40 ) ( p=0 .6 9) R TI m o vem en t ti m e in m s 28 6. 1 (7 0. 4) 0. 22 (t=1 .3 8) (p =0 .1 7) 28 6. 1 (7 0. 4) 0. 14 (t=0 .8 7) (p =0 .3 9) 29 0. 3 (6 6. 8) 0. 18 (t=1 .0 6) (p =0 .3 0) IE D to ta l er ro rs { ad ju sted } 31 .7 (3 2. 3) -0 .1 5 (t=-0. 91 ) ( p=0 .3 7) 31 .7 (3 2. 3) -0 .1 4 (t=-0. 90 ) ( p=0 .3 8) 32 .8 (3 3. 5) 0. 10 (t=0 .5 9) (p =0 .5 6) ER T to ta l n um ber c o rr ec t 86 .6 (2 4. 6) -0 .2 0 (t=-1. 24 ) ( p=0 .2 2) 86 .6 (2 4. 6) -0 .1 6 (t=-0. 99 ) ( p=0 .3 3) 85 .6 (2 4. 2) -0 .2 2 (t=-1. 32 ) ( p=0 .1 9) P Q ( n =4 2) P Q d is tr es s (n =4 2) P A N SS ( n =3 7) Ta b le 6 : P a rti a l c o rr ela ti o n c o eff ici en ts b et w ee n re sp ecti ve ly PQ, PQ d istr es s a n d P ANSS a n d co g n it ive t a sk sco res in a d u lts w it h 2 2q 1 1 D S a ,b a 22q 11DS: 22q 11 d eleti o n sy n d ro m e; PQ : Pr o d ro m a l Qu esti o n n a ir e; PAN SS: Po si ti ve a n d N eg a ti ve Sy m p to m Sca le; PAL : Pa ir ed A ss o ci a tes Lea rn in g ; V R M: V er b a l R eco g n iti o n Mem o ry ; R V P: R a p id V isu a l In fo rm a ti o n P ro ces si n g ; SW M: S p a ti a l W o rki n g Me m o ry ; OT S: On e To u ch Sto ck in g s o f Ca m b ri d g e; R TI: R ea ct io n T im e; IE D: In tr a -E xtra Di m en si o n a l Se t Sh ift; ER T: Em o ti o n R eco g n iti o n T a sk . b Sco res o n co g n iti ve ta sk s p resen te d a s m ea n s a n d st a n d a rd d ev ia ti o n s in p a re n th eses (Pe a rso n ’s p ro d u ct -m o m en t co rr ela ti o n , Pe a rso n ’s r) . c Sta ti st ica l v a lu es co rr ected fo r fu ll sc a le I Q g iv in g a p a rti a l co rr ela ti o n co ef fic ien t. d *: p -valu e ≤ 0. 05; * *: p -val u e < 0.01; ** *: p -val u e < 0.001.
22
Figure 10: Correlation between Prodromal Questionnaire (PQ) and One Touch Stockings of Cambridge (OTS) latency to correct (in ms) in 22q11DS (n=42). OTS latency to correct represents how much time it took on average before the subject selected the correct answer in each stage. Full scale IQ was introduced as a covariate giving a partial correlation coefficient (r). There was significant negative correlation between PQ and OTS latency to correct (p=0.03).
Figure 11: Correlation between Prodromal Questionnaire (PQ) distress and One Touch Stockings of Cambridge (OTS) latency to correct (in ms) in 22q11DS (n=42). OTS latency to correct represents how much time it took on average before the subject selected the correct answer in each stage. Full scale IQ was introduced as a covariate giving a partial correlation coefficient (r). There was significant negative correlation between PQ distress and OTS latency to correct (p=0.02).
partial r = -0.34
23
Figure 12: Correlation between Prodromal Questionnaire (PQ) distress and five choice Reaction Time (RTI) (in ms) in 22q11DS (n=42). Five choice RTI represents how fast a subject lifted its finger when presented with the target stimulus. Full scale IQ was introduced as a covariate giving a partial correlation coefficient (r). There was a significant positive correlation between PQ distress score and five choice RTI (p=0.02).
Study 3: Morning cortisol, (prodromal) psychotic symptoms and early life
trauma
3.1 Demographic data
In table 7 the demographic data are displayed of the 22q11DS subjects that were included in the study that investigated the associations between morning cortisol levels and (prodromal) psychotic symptoms, state anxiety and early life trauma (cortisol analysis). Also, in table 7 the demographic data are displayed of the 22q11DS subjects that were included in two separate analyses that investigated the associations between early life trauma and (prodromal) psychotic symptoms and state anxiety (childhood trauma and bullying analyses).
In the cortisol analysis of the 28 subjects 3 were excluded, because no morning cortisol samples were collected. 1 subject was excluded, because it was an outlier. No correlations with age or FSIQ were found and no differences between the sexes.
24
Table 7: Demographic data of adults with 22q11DS and correlations with morning cortisol levels, childhood trauma score and bullying scoresa,b
a 22q11DS: 22q11 deletion syndrome. b Age and full scale IQ presented as means and standard deviations in parentheses (Pearson’s product-moment correlation, Pearson’s r); sex presented as frequency data (analysis of variance, ANOVA).
3.2 Associations between morning cortisol levels and (prodromal) psychotic symptoms, state anxiety and early life trauma
In table 8 the correlations between morning cortisol levels and (prodromal) psychotic symptoms, state anxiety and early life trauma scores are displayed. No significant correlations were found.
Table 8: Correlation coefficients between morning cortisol levels and PQ, PQ distress, PANSS total, STAI, childhood trauma and bullying in adults with 22q11DSa,b
a 22q11DS: 22q11 deletion syndrome; PQ: Prodromal Questionnaire; PANSS: Positive and Negative Symptom Scale; STAI: State-Trait Anxiety Inventory. b Data presented as means and standard deviations in parentheses (Pearson’s product-moment correlation, Pearson’s r).
3.3 Associations between early life trauma and (prodromal) psychotic symptoms and state anxiety In table 9 the correlations between early life trauma scores (childhood trauma and bullying) and (prodromal) psychotic symptoms and state anxiety are displayed. No significant correlations were found.
Morning cortisol Childhood trauma Bullying
Mean (SD) p Mean (SD) p Mean (SD) p
N 24 - 28 - 28 -
Age 33.3 (9.8) 0.32 34.0 (10.0) 0.95 34.0 (10.0) 0.88
Sex (M/F) 8/16 (33%/67%) 0.27 9/19 (32%/68%) 0.62 9/19 (32%/68%) 0.27 Full scale IQ 78.7 (10.8) 0.97 78.5 (10.3) 0.96 78.5 (10.3) 0.36
Morning cortisol
Mean (SD) Correlation coefficient
N 24 -
Cortisol level {nmol/L} 8.0 (3.3) -
PQ 2.6 (1.9) -0.25 (t=-1.19) (p=0.25) PQ distress 7.5 (5.0) -0.23 (t=-1.07) (p=0.30) PANSS total 35.2 (4.2) -0.34 (t=-1.65) (p=0.11) STAI 35.5 (10.1) -0.35 (t=-1.71) (p=0.10) Childhood trauma 33.4 (8.4) -0.18 (t=-0.85) (p=0.41) Bullying 4.8 (4.4) -0.28 (t=-1.39) (p=0.18)
25
Table 9: Correlation coefficients between respectively childhood trauma and bullying and PQ, PQ distress, PANSS total and STAI in adults with 22q11DSa,b
a 22q11DS: 22q11 deletion syndrome; PQ: Prodromal Questionnaire; PANSS: Positive and Negative Symptom Scale; STAI: State-Trait Anxiety Inventory. b Data presented as means and standard deviations in parentheses (Pearson’s product-moment correlation, Pearson’s r).
Childhood trauma Bullying
Mean (SD) Correlation coefficient Correlation coefficient
N 28 - - Childhood trauma 33.0 (8.1) - 0.29 (t=1.43) (p=0.17) Bullying 5.1 (4.3) 0.29 (t=1.43) (p=0.17) - PQ 3.1 (2.5) -0.13 (t=-0.61) (p=0.55) -0.02 (t=-0.09) (p=0.93) PQ distress 10.3 (8.9) -0.12 (t=-0.55) (p=0.59) 0.09 (t=0.43) (p=0.67) PANSS total 37.0 (7.4) -0.10 (t=-0.45) (p=0.66) 0.23 (t=1.10) (p=0.28) STAI 35.5 (9.7) 0.13 (t=0.61) (p=0.55) 0.30 (t=1.43) (p=0.17)
26
Discussion
This study is one of the few to investigate cognitive function in 22q11DS. In this study the differences in cognitive performance between 22q11DS and healthy controls were investigated, but also within 22q11DS. Furthermore, the associations between morning cortisol levels and (prodromal) psychotic symptoms in 22q11DS were investigated. Multiple interesting findings were done.
Study 1: 22q11DS genotype and cognitive task performance
In the comparison in cognitive task performance between 22q11DS and healthy controls in multiple tasks significant differences in performance are found. In line with previous studies and with the hypothesis the 22q11DS subjects performed worse on each of the tasks where differences were found (Woodin et al., 2001; Baker et al., 2005; Yi et al., 2016). The 22q11DS group performed worse than the healthy control group on the following tasks: VRM, RVP, OTS and ERT. The VRM task measures visual and short-term memory (Smith et al., 2013), the RVP task measures sustained attention and working memory (Coull et al., 1996), the OTS task measures executive function (Revsbech et al., 2015) and the ERT task measures facial emotion recognition (Russo et al., 2015). Since previous studies observed deficits in executive function, attention, episodic memory and social cognition in 22q11DS, it is not in conflict with the hypothesis that the 22q11DS group would perform worse on these particular tasks (Woodin et al., 2001; Antshel et al., 2010; Bearden et al., 2004; Yi et al., 2016; Baker et al., 2005). Additionally, an explanation for the worse performance in the 22q11DS group on the ERT task might be given by a previous fMRI study. This study found decreased activity in 22q11DS compared to healthy controls in brain regions that are involved in social cognition and emotion processing across emotion types (the fusiform-extrastriate cortex, ACC and superomedial PFC) (Azuma et al., 2015).
Important to consider is the fact that the average FSIQ in the 22q11DS group was much lower than in the healthy control group. Since FSIQ correlated significantly with performance on most of the cognitive tasks it was decided to introduce FSIQ as a covariate. Namely, it could have been the case that the proceedings of the tasks were more difficult to understand for the 22q11DS group. However, one could argue that FSIQ should not be left out when comparing the 22q11DS group to the healthy control group, since borderline intellectual level is such an important phenotypic manifestation in 22q11DS. That is why in Appendix 2 the results of the comparison of cognitive performance between 22q11DS and healthy controls can be found without FSIQ as a covariate.
In the exploratory analysis that compared cognitive task performance between 22q11DS COMT-Val and 22q11DS COMT-Met a significant difference in performance in the PAL task is found. The COMT-Met group performed worse on the PAL task than the COMT-Met group. The PAL task measures visuospatial cognition (Junkkila et al., 2012). Previous studies observed deficits in visuospatial cognition in 22q11DS, but not specifically in 22q11DS COMT-Met compared to COMT-Val (Bearden et al., 2004). Also a significant lower FSIQ in the Met group compared to the COMT-Val group was found. This is in line with previous 22q11DS research that found tendencies of lower FSIQ in COMT-Met compared to COMT-Val (Baker et al., 2005; Bearden et al., 2004; Gothelf et al., 2005).
In both the 22q11DS versus healthy control analysis and the COMT-Val versus COMT-Met analysis abnormally high PFC DA levels could explain the worse cognitive performance in respectively the 22q11DS group and the COMT-Met group. Firstly, because of the COMT haploinsiffuciency the PFC DA levels in 22q11DS are expected to be higher than in healthy controls (Gothelf et al., 2005; van Amelsvoort et al., 2008). In the 22q11DS COMT-Met these PFC DA levels are expected to be even higher, because of the lower activity of COMT-Met compared to COMT-Val (Chen et al., 2004). This DA imbalance in the PFC could be another explanation for worse cognitive function (Goldman-Rakic et al., 2000).
