Search

Paranoia is the unwarranted belief that others intend harm, either now or in the future. It is one of the most predictive factors for transition into psychosis, as well as being the most common content of delusional beliefs, known as persecutory delusions. A prominent theory of psychosis suggests that dopaminergic dysregulation in midbrain regions lead to the production of positive symptoms, such as persecutory delusions, and indeed observational evidence demonstrates that blocking post-synaptic dopamine through the administration of antipsychotics can reduce some of these symptoms. However, most mechanistic evidence that aim to test explanatory frameworks of persecutory delusions has focused on the role of dopamine in non-social learning, general cognition and perception. Focusing on non-social mechanisms strongly neglects the clearly social content of persecutory delusions, and therefore the role dopamine may play in the formation of paranoid beliefs. Further still, the well evidenced role of social attributions in paranoia and persecutory delusions has not been fully explained with reference to potential neurocognitive mechanisms that may underlie them. This thesis aims to fill this knowledge gap by developing an experimental test of social attributions in paranoia, developing a computational model that may explain their development and maintenance, and using a psychopharmacological double-blind placebo cross over study to test the role of dopaminergic mechanisms in their generation. I first demonstrate the psychometric properties of the Beliefs and Values Inventory - a novel measure to assess five themes of belief down three dimensions -in a large online population (n = 1763). Secondly, I utilise large online populations (n = 1754) to test how pre-existing paranoid beliefs may influence in-the-moment attributions of harmful intent during a social task using real human decisions, a modified serial Dictator game. The Dictator game suggested that paranoid beliefs lead to faster and larger harmful intent attributions, but not attributions of self-interest. By applying a computational model to the serial Dictator game data, I also found that pre-existing paranoid beliefs and those most likely to attribute high harmful intent are more uncertain about their partner's policy and are more sensitive to the social environment. Finally, I ran a double-blind, within subjects, randomised-control trial using LDOPA and haloperidol to moderate presynaptic and postsynaptic dopamine, respectively, in 30 healthy participants. Following dosing, participants were asked to fill in the Beliefs and Values Inventory and play the within-subjects dictator game. Results suggested that haloperidol selectively reduced harmful intent attributions and increased self-interest attributions across the board, however, did not affect any other themes of beliefs. We conclude that dopamine is crucial in processing the self-relevance of social intent from others. In sum, this thesis demonstrates that paranoid beliefs i) are along an exponential spectrum in the general populations, ii) influence the live attribution of harmful intent toward ambiguous social actions, iii) increase the probability that in-the-moment harmful intent attributions will be made, iv) increase the uncertainty about another's policy, and v) dopamine is crucial to in-the-moment paranoia. I suggest these results converge and integrate with evidence from adjacent fields of social learning to update current models of paranoia.

Disturbances in sleep and circadian rhythm are commonly reported in the period preceding deterioration in mental state in people with psychosis. Sleep disturbance may therefore serve as an early sign of impending relapse, for prompting timely, preventative interventions. Also, establishing a temporal association between sleep and symptomatic deterioration would lend support to the hypothesis that sleep plays a causal role in generating psychotic symptoms, and may in itself form an important therapeutic target. I aim to explore both of these questions through a digital phenotyping approach, which harnesses continuous, real-time data from mobile and wearable devices, in real-world settings. In the first chapter of this thesis incorporating publication, I summarise the evidence for sleep-circadian disruption during the relapse prodrome in schizophrenia, and incorporate literature from the bipolar disorder field. I explore the neurobiological mechanisms by which sleep reduction and circadian misalignment may drive psychopathology, and provide an overview of how digital technologies are being harnessed as clinical tools in psychiatry. The second chapter, a systematic review and meta-analysis of case-control actigraphy studies, seeks to understand sleep-circadian dysfunction across the psychosis spectrum by comparing actigraphy parameters in schizophrenia and bipolar disorder, in the non-acute phase of illness. I show that in both disorders, patients show significantly longer total sleep time and time in bed, with lower daytime activity. However, sleep is also more disturbed, with greater sleep latency and fragmentation, pointing to a range of difficulties pervading all aspects of the 24 hour cycle. Effect sizes were greater in schizophrenia; higher antipsychotic dose was associated with longer sleep time, and greater use of sedative medication was associated with reduced sleep latency, suggesting that treatment effects play a role in these findings. This sets the scene for the subsequent examination of dynamic changes, across phases of illness. To date, sleep-circadian and psychopathology variables have not been studied for long enough to allow episodes of deterioration and relapse to be captured. Increasingly ubiquitous mobile and wearable technologies may allow continuous sampling of subjective and objective sleep-circadian variables in the home environment, over extended periods. In chapter three, I describe the findings of a pilot study, undertaken to develop and test the feasibility and acceptability of Sleepsight, a digital platform designed specifically for this objective. Although there were high rates of engagement with the technology, greater negative symptom burden was associated with poorer completion of smart-phone diaries. Importantly, when the study was conducted in a transparent, collaborative manner, concerns over privacy and security were not significant barriers to engagement and adherence. This tool was then used in a substantive study of 36 people with psychosis, each participating for 12 months, in order to capture natural fluctuations in mental state, including relapse episodes. In chapter four, I examine the association between subjective sleep-circadian variables and psychopathology data using the Differential Time Varying Effects Model (DTVEM), a novel analytic approach which elucidates the lag structure between sleep variables and mental state. The results clearly suggest that poorer subjective sleep quality and reduced subjective sleep duration precede deterioration in symptoms by 8-12 days. Although an inverse direction of effect of psychopathology on sleep was also found, this was shorter, and of smaller magnitude. These findings support the hypothesis that sleep disturbance precedes deterioration in psychopathology. In chapter five, I describe how raw digital phenotype data are classified into sleep-circadian variables using a preprocessing pipeline, and employ a range of analytic approaches to explore these data. Visualisation of rest-activity patterns demonstrated a wide range of sleep-circadian phenotypes, some with striking patterns of circadian dysfunction including relative co-ordination and non-24 hour rhythms. A second analysis using the DTVEM method, with objective data, sustained the finding that sleep reduction preceded symptomatic deterioration. Group-level comparisons between relapsing and non-relapsing participants did not demonstrate significant differences in clinical or sleep-circadian variables. However, visualising these variables in the participants who relapsed, centred on relapse event, suggested that a reduction in sleep duration and decrease in activity accompany the period before, during and after relapse. Mathematical modelling was able to simulate major circadian disturbances in a subset of participants, and crucially showed that increasing daylight in the model restored normal circadian rhythmicity. Some of the technological limitations currently inherent in the approach are also discussed. Taken together, these findings suggest that sleep reduction, fragmentation and poorer sleep quality are features of schizophrenia in remission, and are also associated with deterioration and relapse. The evidence supports these changes playing a causal role in the development of psychopathology. Despite technical challenges relating to data quality, these findings also suggest that meaningful objective data can be collected from smartphone and wearable devices, which show promise for use as decision-support tools.

Introduction: Schizophrenia and other psychotic disorders are complex disorders with multifactorial aetiology encompassing genetic and environmental factors. The genetic architecture of these disorders involves many numerous loci implicated in conferring risk for developing the illness. Cognitive impairments are considered as important core features, serving as potential intermediate phenotypes for psychotic disorders. Cognitive biases including jumping to conclusions (JTC) and social cognition impairments such as difficulties in facial emotion recognition (FER) have been reported to be associated with psychosis, yet their link to shared genetic pathways with the disorder, and their contribution to psychotic outcome are still open questions. My thesis aimed to clarify the relation of these two biases with psychosis liability and with outcome. Methods: Data used in this Thesis come from two large-scale studies in which I participated: the multisite incidence and case-control EU-GEI study, and the Biological Phenotypes, Environment, Genes and Psychosis Outcome (GAP follow-up) study. To examine the presence of the JTC bias at first episode psychosis (FEP), taking into account the influence of general cognitive ability, I conducted a mediation analysis between case/control status, JTC, and IQ in a large sample of patients experiencing their FEP, and population controls. I then explored the associations of JTC with both psychosis and general cognition liability, estimating multiple linear regression models using calculated polygenic risk scores for schizophrenia (SZ PRS) and IQ (IQ PRS). Global and specific facial emotion recognition was analysed through fixed and random effects in repeated measures mixed models in the same large sample. Multiple linear regression models were used to investigate the association with FER and SZ PRS. In a London subsample in which I contributed to recruitment and assessment, I then tested both JTC and FER measured at FEP for multi-domain outcome prediction at, on average, 5-year-follow-up, based on estimated linear, ordinal, and multinomial logistic regression models. Finally, to test the stability of jumping to conclusions and facial emotion recognition over the years, intra-class correlations and repeated measures mixed models were performed. Results: The association between JTC and psychosis was shown to be fully mediated by IQ, and JTC was explained by the genetic underpinnings of IQ but not SZ PRS. Psychotic patients were more prone to display deficits in facial emotion recognition than controls, specifically in fear and anger recognition compared to happy or neutral faces. SZ PRS was associated with worse global facial emotion recognition and specifically with impairments in anger recognition. Neither JTC nor FER measured at FEP predicted any outcome domains at follow-up, although JTC showed a small effect on social outcome. Finally, an overall steady pattern of performance over time was detected for JTC and general facial emotion recognition in both patients and controls, with patients making hastier decisions and recognising fewer emotional faces than controls. Conclusions: My thesis contributed to clarifying the association of jumping to conclusions with psychotic disorders as mediated by general cognition; furthermore, JTC appeared not directly associated with common genetic risk variants for the disorder. However, this finding raised also the necessity of improved assessment for the JTC bias to reduce cognitive noise. On the other hand, findings on facial emotion recognition provided more evidence regarding impairment in global and emotion-specific recognition of emotional faces in the context of psychosis, corroborating previous findings from family studies; the results I obtained with polygenic risk score analysis suggested that facial emotion recognition - specifically negative emotions - could serve as intermediate phenotype for psychosis.

Background: People with psychosis often experience social difficulties associated with key psychological processes, such as social anxiety, paranoia, interpersonal sensitivity and fear of negative evaluation. Virtual reality presents a unique opportunity to bring social environments into the therapy room. This study presents a development case-series of virtual-reality assisted cognitive behavioural therapy (VR-CBT) for people with first episode psychosis (FEP) who are experiencing social difficulties. Methods: Participants were recruited from two South London Early Intervention mental health teams for people experiencing a FEP. Participants received a course of 10 sessions of individual formulation-driven VR-CBT with a focus on improving social functioning. A novel VR environment and therapy package were piloted for the use of the assessment, formulation and treatment of social difficulties. The VR environment was used to conduct exposures, behavioural experiments and to practice coping strategies, which were then translated into the individuals’ real-life environments (e.g. at work, on public transport, in the gym etc). A psychometric assessment was conducted with each participant pre- and post-therapy. A feasibility assessment was conducted, regarding the process of recruitment and retention, required resources and barriers to implementation. To examine acceptability participants also provided qualitative feedback during and after their experiences, and completed an ‘unwanted effects’ questionnaire. Results: 11 participants were recruited to the study, which was above the target of 6-10. There was a 64% conversion rate from those who were approached to those who consented to participation. There was an 82% retention rate; although several participants therapy was somewhat affected by the Covid-19 pandemic and adjustments had to be made to their participation in the study. An assessment of study resources was completed and identified potential challenges for future implementation. Comparison of the pre- and post-therapy psychometric assessment scores revealed a reliable change on some of the assessed domains for all participants. Participants reported few or no unwanted effects and provided positive feedback about the inclusion of VR in the therapy. Conclusions: This study presents promising findings from an innovative VR-CBT package for people with FEP who are experiencing social difficulties. It appears to be feasible and acceptable to participants, and potential benefits were derived. However, a number of challenges to implementation were identified that should be carefully considered before further piloting or trailing is attempted. Therapist reflections on using the VR environment in therapy are also presented. The clinical implications and possible future directions for research are considered.

Background: Although many people report benign, psychotic-like or unusual experiences (UEs) which do not cause distress, rates of reported suicidal ideation and suicide attempts are elevated across the psychosis continuum. Recent research investigating potential risk factors for self-injurious thoughts and behaviours (SITBs) has focussed on the individual contribution of UEs. Whilst there is substantial evidence linking UEs to SITBs, there are inconsistent findings regarding the direction of associations and the role of other factors such as co-occurring mental health difficulties in clinic-based populations. Method: Searches were conducted across three electronic databases: Embase, PsycINFO and PubMed. The reference lists of included studies were also screened for any additional papers. Best practice guidelines for conducting systematic reviews were adhered to and a quality assessment was performed. No studies were removed on the grounds of quality, but this was noted where study findings were integrated and described.

Background: Cognitive models of psychosis propose that coping responses to psychotic experiences (PEs) play an important role in the development of need-for-care. One limitation of the existing literature on coping with PEs is that studies focus primarily on clinical groups, and comparisons are usually made to healthy groups without PEs. Individuals with persistent PEs who do not experience distress or impairment are also highly pertinent to understanding factors that protect from need-for-care in the context of PEs. We therefore aimed to identify and summarise literature exploring the coping responses of healthy individuals with PEs. Methods: We systematically reviewed quantitative, qualitative and mixed methods studies exploring coping either solely in nonclinical individuals with persistent PEs or studies comparing these individuals with a clinical group. Results. Seventeen studies met the inclusion criteria. Most studies met the majority of quality criteria as assessed by the Mixed Methods Appraisal Tool (MMAT; Hong et al.,2018). Findings indicated that PEs are responded to with different types of engagement in the clinical and nonclinical groups. Safety seeking behaviours were more commonly reported by clinical individuals, with less consistent findings for distraction. More neutral/mindful types of responding were found to be more commonly associated with nonclinical group membership. For reappraisal,findings were inconclusive. The differing methods of measuring coping across studies were a key limitation of the existing literature, and this may account for some inconclusive findings. Conclusions: Unlike previous findings on symptomatic coping styles suggesting that engaging and getting involved in the content of PEs is maladaptive, this review suggests that it is possible to engage with PEs in a healthy way, from a relative position of control. Disengaging response styles are likely to be more helpful from a position of acceptance of experiences rather than resistance/avoidance. It is likely that group differences in perceived controllability, perceived threat, and phenomenology of PEs play a role in determining how an individual responds and whether engaging or disengaging from PEs is helpful. Future research should focus on developing a clear and comprehensive taxonomy of coping to capture the complex nature of responding across the psychosis continuum.

Background: Carer criticism, of all expressed emotion (EE) elements, has been linked most consistently with poorer clinical outcomes for patients transdiagnostically. Patient perception of carer criticism (perceived criticism) has been found to predict poorer outcomes across several disorders. The Perceived Criticism Measure (PCM) has featured predominantly in the literature. Despite the bidirectional effects of caregiving relationships on patient and caregiver outcomes,to date, there are no systematic reviews investigating the role of perceived criticism (PC) in both patient and carer outcomes in severe mental health conditions. Objectives: This review aimed to systematically synthesise the evidence available in relation to the role of PC in patient and carer outcomes in severe mental disorders as well as bridge the gap in current knowledge of existing PC measures. Method: A systematic review (Prospero ID: CRD42019122820) was conducted in accordance with PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines. Four databases (EMBASE, Medline and PsycINFO through OVIDSP and Web of Science) were searched, identifying literature focusing on the role of PC in patient and carer outcomes across severe mental disorders. For the purpose of this review, schizophrenia and related disorders, non-organic psychosis, bipolar disorder, anxiety disorders (including panic disorder, obsessive compulsive disorder, post-traumatic stress disorder, generalised anxiety disorder, social anxiety disorder or agoraphobia), eating disorders (including anorexia, bulimia or binge eating disorder) and depression were included under the severe mental illness (SMI) umbrella. The quality of the reviewed studies was evaluated using an adapted version of the “Quality Assessment Tool for Quantitative Studies” developed by the Effective Public Health Practice Project (EPHPP).

Background: Childhood onset mental health difficulties are known to be associated with later mental health disorders and worse prognoses in adulthood. Individuals who develop schizophrenia present from childhood onwards with cognitive deficits, psychotic-like experiences (PLEs), and internalising and externalising problems. People with a diagnosis of a schizophrenia spectrum disorder (SSD) are also more likely than people without this diagnosis to engage in aggressive behaviour towards others. This systematic review examines the evidence base investigating associations between childhood externalising problems (EP) and later psychotic symptoms. Methods: Searches were conducted on Ovid (Medline and Psychinfo), Pubmed and Scopus. PRISMA best practice guidelines for conducting systematic literature reviews were followed. Data was extracted from pre-defined items and assessed using a quality rating scale.

Background: Schizotypy has been described as a vulnerability factor to the development of schizophrenia-spectrum disorders with its three dimensions: positive, negative and disorganised, being associated to different patterns of symptoms and impairments. Digital technologies (DTs) have proved to be viable and accepted tools for assessing, treating and preventing a wide range of mental health disorders across both clinical and non-clinical populations. This review is focused on the use of DT in the assessment and research of any characteristics in high schizotypy in non-clinical individuals. Method: A systematic review was carried out across four online databases: PsyINFO, Embase, Ovid and Web of science. The search yielded 5,784 articles, after screening, seven studies met eligibility criteria, with the quality assessment conducted for each of them. Results: The seven studies included examined a total of 407 participants (156 males, 251 females). The number of participants in each study ranged from 24 to 117. The review illustrated the feasibility and safety of a wide range of DTs in assessing and researching a breadth of characteristics in healthy adult individuals including screening for schizotypy personality traits. The EPHPP tool was used to assess the studies’ qualities. All studies received a global rate of ‘weak’. Conclusions: Limited preliminary data suggest that DTs are viable assessment and research tools in schizotypy healthy populations. The review highlights the presence of insufficient information about the use of DTs to directly assess schizotypy. Suggestions for future research and areas that needs further exploration are discussed.

"Illness perceptions and outcomes in psychosis and bipolar: a systematic review". Introduction: Psychosis and bipolar are associated with significant negative impacts for individuals. Psychological and psychosocial interventions are recommended treatments, however engagement in mixed and there is limited understanding about mechanisms of effectiveness and predictors of outcomes. Illness perceptions have been extensively studied in a physical health context and found to be predictive of treatment engagement and outcomes. Evidence suggests that illness perceptions in psychosis and bipolar may show the same relationships, and thus form a potentially important treatment target. It is therefore important to better understand these complex relationships. Aim: To examine the relationships between illness perceptions and outcomes in psychosis and bipolar, including the direction of these relationships. Outcomes of interest included distress, wellbeing, quality of life, functioning, positive psychosis symptoms, engagement, and change over time in illness perceptions. Method: A comprehensive literature search of MEDLINE, EMBASE, and PsycINFO databases, returned 2573 references for screening. Inclusion and exclusion criteria were used to screen titles and abstracts, followed by full texts. References lists of included papers were also scanned resulting in a total of 33 included studies from 32 papers. Data extracted included details of participants, study design, illness perception and outcomes measures, and findings. A narrative synthesis was completed. Results: Included studies varied in design, measurement of illness perceptions and outcomes, methodological quality, and inconsistency in findings which limits the specificity of conclusions. However, overall the studies demonstrated that more positive or adaptive illness perceptions were associated with better outcomes, including psychological distress, wellbeing, quality of life, functioning, positive symptoms, and engagement in CBTp or psychoeducation interventions. However, as these findings were largely from cross-sectional analyses, causal inferences are limited to findings from only a few longitudinal studies demonstrating variation in stability of illness perceptions over time without specific interventions and that perceptions of control and loss could improve during CBT and psychoeducation interventions. Conclusion: Further intervention studies are needed to provide more evidence on the potential for changeability of illness perceptions and importantly to understand whether, and, if so, how, in the psychological treatment of psychosis and bipolar, change in illness perceptions is related to outcomes. "The role of illness perceptions on engagement and psychological outcomes in Cognitive Behavioural Therapy for Psychosis (CBTp)". Introduction: Illness perceptions are important mechanisms of engagement and change in physical health interventions and emerging evidence suggests they may play a similar role in psychological therapies, including the cognitive behavioural interventions recommended for psychosis and bipolar (CBTp). The limited number of CBTp studies to date that have investigated the role of illness perceptions have demonstrated that CBTp can improve illness perceptions and that this may relate to some outcomes. This relationship needs to be further understood to ascertain the potential to improve outcomes and effectiveness of treatment by specifically targeting illness perceptions in therapy. Aim: The present studied aimed to further understand the relationships between change in dimensions of illness perceptions and outcomes in CBTp, including the direction of these relationships. Method: A prospective cohort design was utilised with data collected as part of routine care across three time points: pre-therapy, mid-therapy and post-therapy. A modified version of the Brief Illness Perceptions Questionnaire (BIPQ) assessed ten dimensions of illness perceptions and the CORE-10 assessed levels of distress. Engagement was dichotomised as attending five or more therapy sessions (engaged) or ≤ 4 sessions (did not engage). Cross Lagged Panel Models (CLPM) were used to assess causal relationships between separate dimensions of illness perceptions and levels of distress during therapy. Results: Two dimensions of illness perceptions predicted engagement in therapy. Those who in engaged in therapy (n = 305) had perceptions of less personal control over their mental health problems but higher perceptions that talking therapy would be helpful than those who did not engage (n = 305). The subsample (n = 248) of participants with sufficient data to be included in CLPM analyses showed that, during therapy, illness perceptions predicted change in levels of distress, for perceptions of symptom identity and that changing the way you think or do things can improve mental health. Levels of distress predicted change in illness perceptions across six dimensions including: consequences, personal control, identity, concern, coherence, emotional response, changing the way you think or do things can improve mental health, and looking at things differently as helpful. No causal effects were found for either perceptions of timeline or treatment control with levels of distress in either direction. Conclusion: These findings suggest a strong causal role for distress on illness perceptions which has been neglected in previous research. Findings are potentially supportive of the possibility of incorporating explicit illness perception interventions into CBTp, but highlight the need for further controlled, longitudinal studies.

Evidence of neuroanatomical abnormalities in subjects with a recent first episode of psychosis (FEP) has been heterogeneous, possibly due to the increased risk of false positives and heterogeneous findings associated with small samples that dominate the literature. In addition, the clinical impact of such findings has been limited. Machine learning promises to overcome this limitation, however, initial attempts to identify FEP have yielded inconsistent results. Within this movement, deep learning has recently emerged as a promising approach in areas such as visual and speech recognition, as well as other areas of medicine. Its ability to capture highly abstract and complex interactions may be useful to capture the characteristic subtle and widespread neuroanatomical changes of FEP. The overarching aim of this doctoral thesis was to investigate neuroanatomical abnormalities in FEP at the individual level in a mega-analytic approach. Structural Magnetic Resonance Imaging (sMRI) data was collated from five independent studies, totalling 1074 participants. FEP and healthy controls (HC) were first compared using voxel-based morphometry in a large-scale megaanalysis. This was followed by a thorough review of the current evidence for deep learning in psychiatric and neurologic neuroimaging. A deep neural network, along with other wellestablished methods for comparison, were then used to classify FEP and HC at each site separately to test for the reproducibility of findings. Finally, a deep neural network was used to classify the two groups in a large-scale mega-analysis. Collectively, results revealed a pattern of fronto-temporal-insular changes identified both at group and individual level. Deep neural networks performed better than traditional machine learning approaches, albeit by a small margin. However, performances were lower than expected overall, ranging between 50 and 70%. Upon interpreting these results, I was able to show evidence for publication bias, suggesting that initial studies may have been over-optimistic. Consist with this, the large-scale deep learning analysis suggested that the reliable classification of FEP based on neuroanatomical data may be around 60%. In light of these results, future studies should continue the pursuit for larger samples combined with multimodal approaches to build more reliable and informative models.

Background: Existing reviews of trauma and psychosis have identified a relationship between childhood emotional abuse (CEA) and psychosis, but have not distinguished between different types of CEA. Thus, further investigation is needed. This systematic review aimed to explore the relationship between caregiver antipathy in childhood (i.e. criticism, hostility, coldness or rejection shown by parent figures towards the child) and psychosis diagnosis/symptoms. Method: Five databases were searched and studies were evaluated against inclusion/exclusion criteria. The relevant data was extracted and a narrative synthesis of key findings was completed. Estimates of effect size for antipathy-psychosis associations were calculated using standardised conventions. Assessment of study quality was undertaken using the Effective Public Health Practice Project (EPHPP) Quality Assessment Tool for Quantitative Studies. Results: Fourteen papers met all inclusion criteria. Twelve studies found significant associations between caregiver antipathy and psychosis and two did not. There was evidence that adults with schizophrenia-spectrum diagnosis reported more severe caregiver antipathy in childhood than non-clinical controls. In addition, caregiver antipathy was associated with more severe psychosis symptoms. Most studies received weak or moderate overall quality ratings. Conclusions: This is the first review to focus on investigating the relationship between caregiver antipathy and psychosis. The variability in study methods and quality makes it difficult to draw firm conclusions about the nature and strength of these associations. Further studies using more robust methodological and statistical procedures are required to consolidate the emerging evidence.

Aim: Our work builds on recent findings that (a) it is possible to devise algorithms for structured, focused and not timely treatment of schizophrenia (Agid et al, 2011) and (b) while prescribed later than it should be, Clozapine is the most effective antipsychotic treatment for adult treatment-resistant schizophrenia (Howes et al., 2012). However, there is no algorithm-based treatment for early onset schizophrenia, but rather generic, overly lengthy and non-specific guidelines (NICE 2012 and NICE 2013) and there is insufficient similar research in relation to, and therefore the use of, Clozapine treatment for treatment-resistant early onset schizophrenia. Hence our work aims to investigate the use of antipsychotics for early onset schizophrenia, and develop an hypothesis regarding a structured treatment to be tested prospectively like some studies did for adult schizophrenia (Agid et al, 2011) to reach clear, concise, algorithmic recommendations and become a solid basis for clinicians to make decisions for effective first-line treatment of early onset schizophrenia and effective treatment of treatment-resistant schizophrenia, hence having our study be a significant stepping stone towards the development of clinical decision making tools for the effective treatment of early onset schizophrenia. Hypotheses: Firstly, there are antipsychotics to prescribe for early onset schizophrenia, which are more effective than others and have manageable side effects. Secondly, Clozapine is the most effective antipsychotic for treatment-resistant early onset schizophrenia and is not sufficiently used due to unfounded concerns with regards to its side effects. Methods: Following a period of comprehensive review of available studies on the use of antipsychotics in children and adolescents, which we describe in chapters 1 and 2, we concluded we could not engage in a time limited (within the number of years we had set out to conduct our study) prospective study since doctors would not straight away engage in prospective studies to prescribe specific medications. This is because doctors follow guidelines (NICE 2012, NICE 2013) that are focused on the risks related to prescribing antipsychotics emphasizing what patient-dependent side effects to avoid when considering prescription and are therefore generic, unclear and ineffectively large citing many possible antipsychotic drugs. We instead opted to perform retrospective statistical analysis of two long-term treatments datasets documented with sufficient consistency over long periods to credibly be able to recommend future studies to engage in very specific prospective studies such as the algorithm-based approach to first-episode schizophrenia by Agid (Agid et al, 2011). Those long-term treatment datasets were: 1. The prescription collection records in a nation-wide unbiased population demographic and clinical dataset of youth with psychosis who were prescribed with Clozapine. The electronic records of 100 patients at the South London and Maudsley NHS Foundation Trust’s child and adolescent psychiatry services which we studied over 2 years. This was a population prescribed antipsychotics as part of their routine inpatient treatment, with data collected at baseline before antipsychotics switch / initiation and 3, 6 and 12 months thereafter. One-way ANOVA Power analysis was completed to ascertain appropriateness of the sample analyzed in the above-mentioned 2-year study. Results: 1. The whole body of literature on the use of antipsychotics for early onset schizophrenia highlights the need for further evidence in relation to efficacy and safety. 2. In the study we conducted with a 100-patient sample, which findings we ratified by performing a power analysis, we found that: A. Olanzapine was the most effective antipsychotic and patients prescribed with it were less likely to discontinue treatment, however it also resulted in the highest proportion of side effects; and B. Clozapine was the least prescribed antipsychotic (less than 0.5% of the sample) and only for one third of those with poor treatment response from Olanzapine. The antipsychotics examined were generally well tolerated. However, the overall rate of side effects was low suggesting that many side effects are either missed or not recorded in routine settings. 3. Clozapine is the most effective antipsychotic for patients with treatment-resistant early onset schizophrenia. Strict routine monitoring of its impact during treatment can provide an adequate level of comfort in its use. Conclusions: 1. All antipsychotics routinely prescribed to inpatients are associated with improvement in terms of global function. 2. Olanzapine showed evidence for efficacy superiority, suggesting that olanzapine should be considered as first line medication. This statistically significant fact was ratified by other studies completed in the year after the completion of our study (Zhu et al., 2017). 3. Clozapine is the most effective antipsychotic for patients with treatment-resistant early onset schizophrenia. Tools for its monitored prescription (such as blood measurement tools for youth) should serve as effective decision-making tools for clinicians wanting to use Clozapine more often and earlier in treatment.

Treatment resistance (TR), affecting approximately 20-30% of patients with psychosis, has a high burden both for patients and healthcare services. I provided a general introduction to TR in Chapter 6 and highlighted the need to identify TR earlier in the course of the illness, so that an effective treatment, such as clozapine, can be offered promptly. The purpose of this thesis was to identify clinical, demographic, and genetic predictors of TR, using information identified when a patient first presents to clinical services with psychosis. First, I reported the results of a systemic literature review, which synthesised predictors of TR identified in longitudinal, observational studies (Chapter 8). Younger age of onset was the most consistent predictor of TR, but this review also indicated that, to date, studies have not used statistical methods specifically designed to identify predictors that have a high chance of predicting TR in future patients. Existing literature has primarily used methods designed to capture the magnitude of association, between predictors and TR, within the study sample. Second, I reported the results of my own analysis using a dataset created by combining existing longitudinal, first episode psychosis cohorts. This dataset, known as STRATA-G (and described in Chapter 7), included patients who had a minimum of one year follow-up, provided a DNA sample, could be classified as either TR or non-treatment resistant (NTR). I found significant associations between TR and both younger age of onset and fewer years in education (Chapter 9). I created a predictive model of TR that selected nine first episode variables and could, after internal validation, correctly classify 59% of TR patients and 65% of NTR patients (Chapter 9). Third, I reported the results of a genome-wide association study (GWAS) of treatment resistance, which was used to create a polygenic risk score for TR (Chapter 10). This polygenic risk score was significantly associated with TR in the STRATA-G sample. Fourth, I investigated whether age of psychosis onset mediated the relationship between genetic burden for TR and subsequent TR, using STRATA-G (Chapter 11). Age of onset was not associated with genetic risk for TR and there was no evidence it mediated the relationship between the polygenic risk score for TR and subsequent TR. Finally, I expanded on the discussion of the findings in the preceding chapters (Chapter 12). Here, I placed my findings in the context of the wider literature and discussed the limitations of my data. I discussed the future of prediction modelling in TR; the potential uses of a prediction model, in both clinical practice and research studies, and the importance of future work to validate the findings reporting in this thesis.

Previous research has investigated socially based threats, such stigmatisation and shame, among people with psychosis. However, studies have typically considered these social experiences in terms of being the consequences, rather than causes, of psychosis. There have been some studies looking at stigma and shame as causal mechanisms in emotional symptoms (e.g. anxiety and depression) in the context of psychosis, and of post-psychotic trauma, but not of psychosis itself. One possible reason for why this may have been overlooked is that stigma and shame have not traditionally been regarded in ‘threat’ terms, with traumatic-like, perhaps dissociative, properties of their own. Social rank theory (Gilbert, 1992) offers an understanding of these socially-based experiences within their evolutionary context, with links to developing threat-monitoring, -processing, and -protective strategies. Social rank theory therefore provides a theoretical rationale within which to examine stigma and shame through the lens of threat and threat-protection mechanisms. These social rank threats can be investigated for links to traumatic, dissociative processes, and hence as potential contributors to causal pathways in psychosis. This thesis investigates the interaction of these social and dissociative processes in psychosis (Section A: Studies 1, 2, and 3). It also develops and tests the acceptability of a new psychological therapy that specifically targets social mechanisms, e.g. reducing shame and increasing social safeness, among people with psychosis (Section B: Study 4). The first study designed and tested a measure of psychotic-like experiences (PLEs) that could be used for the other studies. Study 1, a questionnaire design study, analysed data from online participants for a) item selection, and b) psychometric evaluation. Scores were provided by 532 participants, and a new scale, the Transpersonal Experiences Questionnaire (TEQ), was developed. The 19-item TEQ was found to be a unidimensional scale with satisfactory internal consistency (0.85), good utility for longitudinal hypothesis-testing, and particular reliability for people endorsing a high number of PLE items. Study 2 investigated the longitudinal relationship between dissociation and PLEs, and the moderation effects of social variables, shame and social safeness, on this relationship. Participants (n=314) provided online data at two timepoints (0 and 6 months) and the analysis demonstrated a significant interaction between dissociation and shame on PLEs, despite the absence of a direct association between dissociation and PLEs. Social safeness significantly moderated this interaction effect, but not in the expected direction: higher social safeness was associated with more pronounced dissociation-shame interaction effects. In Study 3, participants (n=30) were recruited from a ‘nonclinical- in-crisis’ population; i.e. those seeking support for PLEs, but from spiritual organisations rather than mainstream services. This sample had significant trauma histories and high levels of dissociation. Endorsing a higher number of PLE items was related to more distress; however, once total endorsement scores were controlled for, PLEs were rated as more positive (‘enriching’, ‘safe’ and ‘pleasant’) than negative (‘distressing’ and ‘disruptive’). The positive PLE dimensions were correlated with spirituality, post-experience growth, and social safeness. Negative PLE dimensions were correlated with depression. The second section of the thesis reports the development and initial piloting of a new clinical intervention called Compassion Focused Therapy for Psychosis (CFTp). CFTp specifically targets building social safeness experience and compassion motivation systems with the aim of regulating threat-based dissociative processes in psychosis. Study 4 was a case series study, in which NHS participants with distressing experiences were randomised to multiple baseline periods of 2, 4 and 6 weeks, before a 26-session individual therapy. Seven out of eight participants completed the full 26 sessions. At the single case level of analysis, over half the completers showed clinically reliable improvements in depression (5/7), stress (5/7), general wellbeing (5/7), anxiety (4/7), and voices (3/5). One showed a deterioration in anxiety (1/7) and dissociation (1/7). At the group level of analysis (n=7), there were significant improvements in group mean scores of depression, stress, wellbeing, voices and delusions, and at 6-8 week follow-up, the wellbeing, voices and delusions improvements remained significant. Further investigations are required of the theoretical relationships surfacing in this thesis, as well as the evaluation of CFTp. Specifically, the future modelling and testing of three-way interacting mechanisms (dissociation x social rank x attachment) in pathways to PLEs is recommended. For CFTp to be considered as a clinical treatment, it requires further testing in a pilot Randomised Controlled Trial, with different therapists, and with researchers who are unaware of the group allocation.

Background: The prevalence of psychotic experiences (PEs) in the general population is significantly greater than comparable rates of clinical disorder, consistent with a continuum view of psychosis. Cognitive models of psychosis propose that response style and reasoning biases play an important role in determining the clinical consequences of PEs. Non-clinical individuals, with persistent PEs but no ‘need for care’, have been found to be less likely to endorse maladaptive response styles compared to a clinical psychosis group. Research is typically based on retrospective interviews, with the potential confound that maladaptive responses simply reflect intrinsically more distressing experiences within the clinical group. A novel way of addressing this issue is by exploring response styles across the psychosis continuum, using experimentally controlled anomalous experiences. In a second strand of research, individuals with persecutory beliefs have been found to make decisions on the basis of limited evidence in probabilistic reasoning tasks; the so-called ‘Jumping to Conclusions’ (JTC) data-gathering bias. Lack of belief flexibility (defined as the meta-cognitive capacity of reflecting on one's own beliefs, changing them in the light of reflection and evidence, and generating and considering alternatives) is commonly reported in around 50-75 percent of people with clinical delusions. To date, no study has compared clinical and non-clinical individuals with persistent PEs on these reasoning biases. Furthermore, there has been recent interest in integrating the literature on reasoning biases in psychosis with broader, dual process models of reasoning which distinguish between fast, automatic, Type 1 (‘experiential’) processes and slow, analytic, Type 2 (‘rational’) processes. Research in this area has tended to focus on self-reported rational and experiential reasoning with the limitation that certain reasoning processes may occur outside of conscious awareness. Finally, despite the clinical and theoretical relevance, there has been a relative lack of research focusing on the alternative explanations held by people with clinical paranoia and no current method of categorising content-level differences in subtypes of alternative explanations. This thesis investigated the role played by response style and reasoning in determining the clinical consequences of PEs, addressing the following research questions: Research Question 1: Do individuals with persistent PEs, with and without a ‘need for care’, show differences in response styles following exposure to experimentally controlled anomalous experiences? Research Question 2: How might the literature on reasoning biases in psychosis be integrated with broader concepts of ‘fast’ and ‘slow’ thinking within dualprocess models of reasoning? Research Question 3: Do individuals with persistent PEs, with and without a ‘need for care’, show differences in reasoning biases and associated ‘fast’ and ‘slow’ thinking? Research Question 4: What types of alternative explanations (AEs) are reported by individuals with clinical paranoia to explain the evidence for their persecutory belief? Research Question 5: Do the reasoning biases under investigation combine to form latent ‘fast’ and ‘slow’ thinking factors, which predict paranoia severity within clinical samples? Research Question 6: What are the implications of this investigation of response style and reasoning for cognitive therapy for psychosis? Method: The research questions were pursued in four cross-sectional empirical studies and one narrative review: Empirical Study 1: Two groups with persistent PEs (clinical; n=84; non-clinical; n=92) and a control group without PEs (n=83) were compared on response styles following experimental analogues of thought interference (Cards Task, Telepath) and hearing voices (Virtual Acoustic Space Paradigm). Narrative Review: A narrative review was conducted on the recent literature on reasoning biases, delusional beliefs and psychosis. The state-of-the-art knowledge from systematic reviews and meta-analyses on the evidence for jumping to conclusions was summarised, before a fuller discussion of the empirical literature on belief flexibility as applied to delusions. Empirical Study 2: Two groups with persistent PEs (clinical; n=74; non-clinical; n=92) and a control group without PEs (n=83) were compared on ‘Jumping to Conclusions’ (JTC) and belief flexibility. A randomly selected subset (n=104) was investigated to examine differences in use of experiential and rational reasoning when explaining PEs. Empirical Study 3: Alternative Explanations (AE; a component of belief flexibility) were investigated in a large clinical sample with distressing persecutory beliefs (n=274). A system of categorising the AEs held by individuals to explain the evidence for their persecutory belief was developed. Reliability analysis was conducted, and categories presented descriptively with reference to conviction and associated distress. AEs (combining sub-types) and main belief were compared with respect to belief conviction and distress. Empirical Study 4: Drawing on the same sample as Empirical Study 3, associations between presence of AEs and other reasoning biases (Possibility of being mistaken (PM) and JTC) and paranoia severity were investigated. The hypothesis that paranoia involves reduced ‘slow thinking’ and over-reliance on ‘fast thinking’ was tested using formally assessed reasoning biases and a latent variable analytic approach. Results: Empirical Study 1: The non-clinical group with PEs were less likely to endorse maladaptive response styles, such as attempts to avoid, suppress, worry about or control mental experiences, compared to the clinical group on all tasks. The clinical group were more likely to endorse unhelpful response styles compared to controls on two out of three tasks. The non-clinical group performed similarly to controls on maladaptive responding across all tasks. There were no group differences for adaptive response styles, such as cognitive reappraisal or mindful acceptance of experiences. Narrative review: Dual process models of reasoning, popularised as ‘fast and slow thinking’, were considered to offer a useful theoretical framework for 7 integrating research on reasoning biases in psychosis with implications for research and clinical practice. Empirical Study 2: The JTC reasoning bias was more common in the clinical group than the other two groups. Unexpectedly, no group differences were observed between clinical and non-clinical groups with PEs on measures of belief flexibility. However, the clinical group was less likely to employ rational reasoning, while the non-clinical group was more likely to use experiential reasoning plus a combination of both types of reasoning processes, compared to the other two groups. Empirical Study 3: 129 individuals (47%) were able to provide an AE for the evidence they have for their main persecutory belief. AEs were categorised, with excellent reliability, into eleven categories: Thirty percent of the AEs were categorised under a broad ‘Illness’ category with unelaborated mental health explanations the single most common subcategory across all AEs. As well as being the most common, this category was associated with the lowest level of conviction. Over fifty percent of the AEs fell within the broad psychological category of ‘Misinterpretation’. AEs which involved recognition that the experiences were not personal or specifically targeted (‘Misinterpretation: nonpersonalising’) were associated with the lowest distress of any sub-category, although conviction in these AEs was comparatively low. When combined across all categories, AEs were rated as less distressing but also held with lower conviction when compared to the main distressing belief. Empirical Study 4: The presence of an AE was associated with PM and mean beads drawn on the JTC tasks. Individuals with an AE showed lower conviction and paranoia severity compared to those with no AE, across a range of assessments. The components of belief flexibility (AE and PM) formed a latent ‘Slow thinking’ factor, with higher levels of slow thinking associated with lower conviction and paranoia severity across all measures. The JTC tasks also formed a latent ‘Fast thinking’ factor which, as expected, was negatively correlated with ‘Slow thinking’. However, contrary to expectation, ‘fast thinking’ was not positively associated with paranoia conviction or severity.

Psychiatric disorders are among the leading health concerns worldwide, with one in four people being affected by a mental health problem at some point in their life. While the economic burden has been on the rise, progress in treatments have remained stagnant due to the complexity of such disorders. To gain a better understanding in the aetiology of psychiatric disorders, in this thesis, I investigated the relation of environmental factors, genetic variation and methylation with brain structure/function and reinforcement-related behaviour and psychopathology. This work aims at (i) characterising mechanisms by which environmental influences, including psychosocial stress exert their effects on behaviour, (ii) identifying epigenetics/genetic markers related to behaviour and development of psychopathology, and (iii) examine how brain structure/function could relate the underlying biological mechanisms associated with behavioural outcomes. In the first half of this thesis, I examined the effects of DNA methylation in the longitudinal IMAGEN imaging - genetics cohort of 2000 adolescents and related them to measures of psychopathology relevant for different psychiatric disorders, including substance use. In two separate epigenome-wide studies, novel biological mechanisms underlying substance use wereidentified: allele-specific differential methylation of SPDEF (Chapter 3) and longitudinal differential methylation of roughly 11,000 genes (Chapter 4) were found to be associated with adolescent alcohol consumption respectively. In both studies, changes to the brain were also found to be involved in the gene-behaviour relationship. The second half of this thesis, I investigated the translational aspects of biological mechanisms underlying behaviour and psychopathology. In collaboration with other research groups, I replicated novel findings of alcohol-associated DLGAP2 methylation in post-mortem brain tissues (Chapter 5) and alcohol-related effects of SMPD3 in mice (Chapter 6), in normal human populations drawn from both the IMAGEN cohort and UK Biobank respectively. With virtually no progress in development of novel treatments over the last couple of decades and a constant increase in burden of psychiatric diseases, it is pertinent that psychiatric research aims at improving psychiatric nosology and discerning psychiatric pathology. Overall, this thesis contributes to existing literature and highlights the importance of multimodal approaches into better understanding the complex biological underpinnings of psychiatric disorders such as substance use.

Background: Immune, metabolic, and cardiac alterations are well recognised in schizophrenia, however it remains unclear if these alterations are present once the influence of lifestyle factors are accounted for, if they occur in all patients or only a subgroup, and if these alterations occur to a similar magnitude as alterations observed within the brain. Furthermore, it is unclear if metabolic changes in psychosis are reflective of a pro-inflammatory state. Systemic inflammatory conditions such as rheumatoid arthritis are characterised by insulin resistance and paradoxical reductions in cholesterol levels. In this context, specific hypotheses tested in this thesis are as follows: 1) Compared with controls, antipsychotic naïve patients with first episode psychosis will present with insulin resistance and reductions in total and low density lipoprotein cholesterol levels, reflective of a pro-inflammatory state; 2) Compared with controls, antipsychotic naïve patients with first episode psychosis will present with elevated pro-inflammatory immune parameters even once lifestyle factors are accounted for, and these alterations will be typically observed in patients rather than only occurring in a subgroup; 3) Compared with controls, patients with schizophrenia will present with fibroinflammatory changes within the myocardium, even after accounting for traditional cardiovascular risk factors; 4) In psychosis, alterations in metabolic, immune, and hypothalamic-pituitary-adrenal axis markers will be of a similar magnitude to brain structural, functional, and chemical alterations. Methods and Results: Introductory chapter 1 summarises the physical health burden associated with schizophrenia, considers outstanding issues that the papers in this thesis aim to address, and describes the statistical and imaging techniques employed to do so. Chapter 2 sets out to clarify if there is evidence of disturbed glucose homeostasis in antipsychotic naïve patients with FEP once the confounds of lifestyle and medication have been removed. The Embase, Medline and PsycINFO databases were systematically searched for studies examining measures of glucose homeostasis in drug-naïve individuals with FEP compared with controls. Of 3660 citations retrieved, 16 case control studies comprising 15 samples met inclusion criteria. The overall sample included 731 patients and 614 controls. Standardised mean differences in fasting plasma glucose, plasma glucose post-OGTT, fasting plasma insulin, insulin resistance, and HbA1c were calculated. Fasting plasma glucose (g = 0.20 (95% CI 0.02 – 0.38, p = 0.03)), plasma glucose post-OGTT (g = 0.61 (95% CI 0.16 – 1.05, p = 0.007)), fasting plasma insulin (g = 0.41 (95% CI 0.09 – 0.72, p = 0.01)) and insulin resistance (HOMA-IR) (g = 0.34 (95% CI 0.14 – 0.54, p = 0.001)) were all significantly elevated in patients compared with controls. Chapter 3 sets out to clarify if there is evidence of lipid dysregulation in antipsychotic naïve patients with FEP, even once the confounds of lifestyle and medication have been removed. The Embase, Medline and PsycINFO databases were systematically searched for studies examining lipid and adipocytokine parameters in individuals with FEP compared with controls. Studies reported fasting total cholesterol, low density lipoprotein (LDL) cholesterol, high density lipoprotein (HDL) cholesterol, triglycerides and leptin levels. Of 2070 citations retrieved, 20 case-control studies met inclusion criteria including 1167 patients and 1184 controls. Total cholesterol (g = -0.19; 95% CI = -0.32 - -0.06; p = 0.005) and LDL cholesterol (g = -0.22; 95% CI -0.35 - -0.09; p = 0.001) levels were significantly decreased in patients versus controls. Triglyceride levels were significantly increased in patients versus controls (g = 0.14; 95% CI 0.00 – 0.28; p < 0.05). However, HDL cholesterol (g = -0.22; p = 0.07) and leptin (g = -0.05, p = 0.78) levels were not altered in patients versus controls. Chapter 4 sets out to clarify if there is evidence of immune dysregulation in antipsychotic-naïve first episode psychosis (FEP) after controlling for environmental/demographic confounders. It also examines for evidence of an immune subgroup of patients. The Embase, Medline and PsycINFO databases were systematically searched for studies examining blood cytokines, C-reactive protein, and white cell counts in individuals with FEP compared with controls. Means and variances were extracted and analysed using multivariate meta-analysis of mean and variability of differences. Outcomes were: 1) variability in patients relative to controls, indexed by variability ratio (VR) and coefficient of variation ratio (CVR); 2) mean differences indexed by Hedges g; 3) Modal-distribution of raw immune-parameter data using Hartigan’s unimodality dip test. 35 studies reporting on 1263 patients and 1470 controls were included. Variability of interleukin-6 (IL6) (VR:0.19), tumour necrosis factor−α (TNFα)(VR:0.36), interleukin-1β (VR:0.35), interleukin-4 (VR:0.55), and interleukin-8 (VR:0.28) was reduced in patients. Results persisted for IL6 and IL8 after mean-scaling. 94% and 100% of raw data were unimodally distributed in psychosis and controls respectively. Mean levels of IL6 (g=0.62), TNFα (g=0.56), interferon-γ (IFNγ) (g=0.32), transforming growth factor- β (g=0.53), and interleukin-17 (IL17) (g=0.48) were elevated. Sensitivity analyses indicated this is unlikely explained by confounders for IL6, IFN γ, and IL17. Chapter 5 describes a cardiac MRI study which set out, as its primary objective, to examine for evidence of inflammation/fibrosis within the hearts of patients with schizophrenia. 31 participants (14 patients and 17 controls) underwent cardiac MRI assessing myocardial markers of fibrosis/inflammation, indexed by native myocardial T1 time, and cardiac structure (left ventricular (LV) mass) and function (left/right ventricular end-diastolic and end-systolic volumes, stroke volumes, and ejection fractions). Participants were physically fit, and matched for age, gender, smoking, blood pressure, BMI, HbA1c, ethnicity, and physical activity. Compared with controls, native myocardial T1 was significantly longer in patients with schizophrenia (effect size, d=0.89; p=0.02). Patients had significantly lower LV mass, and lower left/right ventricular end-diastolic and stroke volumes (effect sizes, d=0.86-1.08; all p-values <0.05). There were no significant differences in left/right end-systolic volumes and ejection fractions between groups (p > 0.05). Chapter 6 considers whether psychosis should be considered a multisystem disorder. First, a systematic meta-review of Pubmed was conducted to summarize effect sizes for both CNS (focusing on brain structural, neurophysiological, and neurochemical parameters) and non-CNS dysfunction (focusing on immune, cardiometabolic, and hypothalamic-pituitary-adrenal (HPA) systems) in FEP to determine summary effect sizes for each organ system individually. Case control data was extracted for 165 studies making up these meta-analyses, comprising a total sample size of 13,440. Conclusions: Chapter 7 considers the outcomes of the preceding 5 chapters. Specific conclusions are as follows: 1) Glucose homeostasis is altered from psychosis onset and in the absence of antipsychotic treatment, indicating that patients are at increased risk of diabetes mellitus; 2) Total and LDL cholesterol levels are reduced in FEP, indicating that high rates of hypercholesterolemia in patients with chronic schizophrenia is secondary and potentially modifiable. In contrast, triglycerides are elevated in FEP. Hypertriglyceridaemia is a feature of type 2 diabetes mellitus, therefore this finding adds to an increasing evidence base of glucose dysregulation in this cohort; 3) Elevated blood cytokines are observed in psychosis after accounting for confounds, supporting the theory that the condition is pro-inflammatory, however the hypothesis of an immune-subgroup in psychosis is not supported by analyses of variability or modal distribution; 4) Cardiac imaging outcomes suggest a diffuse fibro-inflammatory myocardial process in patients that is independent of established CVD-risk factors and could contribute to the excess cardiovascular mortality associated with schizophrenia; 5) Alterations in non-CNS systems in FEP are broadly similar in magnitude to a range of CNS alterations, which could be interpreted as psychosis representing a multi-system disorder. Chapter 7 also considers if immune and cardiometabolic alterations seen in psychotic illness can be explained by a single pathoaetiological model. This draws on evidence from the field of rheumatology where systemic pro-inflammatory conditions are associated with insulin resistance, paradoxical reductions in cholesterol levels, and cardiac fibrosis. Future directions are also considered, including the need to define if cardiac alterations are the effect of antipsychotic treatment or are intrinsic to schizophrenia, and if early intervention for management of glucose dysregulation in FEP is warranted.

Background: Offending behaviour among military personnel has been a concern throughout history. While research has largely focused on violent offending during the life course and post service, there is a dearth of research on offending behaviour during military personnel’s time in service. This thesis examines the prevalence of and risk factors for all types of in-service offending behaviour using official offence records. It also begins to explore the relationship between the perpetrator and victim of interpersonal violent offences. Method: This study used a large representative sample of the UK Armed Forces (N=11,418). Data on sociodemographics and experiences during service (including pre- and post-deployment), along with post-deployment mental health factors were collected by self-report questionnaire (collected at three time points, 2003-2005, 2007-2009 and 2013-2016). This data was linked the Ministry of Defence’s official offence record database of offending behaviour during military service, REDCAP. Findings: The prevalence of in-service violent offending among this sample was 10.4%. The prevalence of in-service non-violent offending was 16.9%. These results suggest that older age at enlistment, lower levels of educational attainment, having a history of pre-enlistment antisocial behaviour, being a member of the Army and being of lower rank were associated with in-service offending. Compared to non-deployed personnel, being deployed, regardless of role during deployment, was strongly associated with all measures of in-service offending. This was the case whether personnel had a record of offending prior to deployment or not. Being deployed in a combat role compared to a non-combat role was only associated with non-violent in-service offending (adjusted Hazard Ratio (AHR)=1.79, 95% Confidence Interval (CI): 1.21-2.65, p=0.003). While reported symptoms of PTSD were not associated in-service offending, alcohol misuse was independently associated with all measures of in-service offending. The presence of symptoms of common mental disorders was associated with any in-service offending (AHR=1.54, 95% CI: 1.08-2.21, p=0.019), but not in-service violent or in-service non-violent offending separately. Self-reported post-deployment anger was associated with any in-service offending (AHR=1.78, 95% CI: 1.06-3.01, p=0.031) and in-service violent offending (AHR=4.45, 95% CI: 1.69-11.73, p=0.003). Self-reported post-deployment violence towards family members was associated with having left service. Deployment in a combat role (compared to non-combat role) and post-deployment mental health variables were also associated with post-deployment family violence. Alcohol misuse was not associated with post-deployment family violence. Conclusion: The majority of offences perpetrated among this sample were non-violent offences. Violent behaviour and in-service offending towards family members were prevalent among this military sample and warrant further investigation. More research is needed to determine the factors associated with in-service interpersonal offending towards family members, particularly spouses and partners to build an evidence base on intimate partner violence among UK military personnel.

The aim of the current work was twofold. First, to develop methods that allow for the useful integration of distinct methods of examining the human brain, namely positron emission tomography and functional magnetic resonance imaging. Second, to apply these methods to better understand the neurobiological mechanisms underlying the pathways between environmental exposures, schizophrenia, and psychotic symptoms. The introductory chapter provides an overview of the schizophrenia concept, the neurobiological systems under examination, and the methodologies used. Chapter 2 is a review article that discusses recent developments in our understanding of striatal structure and function, and specifically how striatal dysfunction may underlie many of the symptoms observed in schizophrenia. A specific issue discussed in Chapter 2 is, 'where' precisely does striatal dopamine dysfunction occur in schizophrenia? Chapter 3 attempts to answer this using a quantitative approach. A meta-analysis of all studies that have used positron emission tomography (PET) to measure striatal dopamine function in schizophrenia, shows that dopamine dysfunction displays marked spatial variability, and does not occur uniformly across the striatum. I find that dopamine dysfunction occurs predominantly in the associative striatum, refuting the hypothesis that dysfunction within the limbic striatum specifically underlies psychotic symptoms. Chapter 4 builds on the results of the meta-analysis and uses an integrative PETfMRI approach to ask whether spatial variability in striatal dopamine function shapes psychopathology in psychosis. I show that dopamine function within striatal regions functionally linked to cortical sensorimotor networks is associated with baseline motor symptoms, while negative and affective symptoms, are linked to dopamine function in striatal regions linked to default mode and cinguloopercular networks respectively. Chapter 5 again looks at the question of psychopathology, but here I study individuals who have been exposed to risk factors for psychosis, but who have not yet developed a psychiatric disorder. I show that exposure to environmental risk factors is associated with reduced adaptive and increased aberrant salience measures. I also find that these differences in behaviour are related to differences in corticostriatal connectivity. Finally, Chapter 6 uses PET and fMRI to answer an ongoing question regarding neurobiological mechanisms underlying salience processing. Specifically - what is the relationship between mesolimbic dopamine function and the cortical salience network? This is a question of interest because both neural systems share an overlapping role and have been implicated in psychotic disorders. I show that while striatal dopamine 'release' capacity is associated with reduced salience network connectivity, striatal dopamine 'synthesis' capacity is associated with greater connectivity within the salience network, and that this is particularly the case for 'hub' nodes playing a central role in information processing. In summary, I show that dopamine dysfunction in schizophrenia does not exist uniformly across the striatum, but is greatest in the associative striatum. The clinical relevance of this is demonstrated by the finding that psychotic symptoms relate to variation in the spatial profile of striatal dopamine dysfunction. In addition, I find that striatal connectivity is altered in individuals exposed to environmental risk factors. Finally, I show how systems known to be crucial to salience processing are linked. Together these findings advance our understanding of schizophrenia by suggesting mechanisms via which striatal and cortical function are linked, how risk factors are associated with neurobiological changes, and how neurobiological abnormalities may shape psychotic symptoms.

The current systematic review aims to understand the relationship between childhood maltreatment, emotional dysregulation and subsequent depression in adulthood. Existing research was identified following a search conducted on relevant scientific databases: Ovid MEDLINE(R) In-Process & Other Non-Indexed Citations and Ovid MEDLINE(R) (1946 to May 2019), Psycinfo (1806 to May 2019), PsycArticles and EMBASE (1974 to May 2019). Further hand searches were also conducted within relevant journals and publications identified from retrieved articles. Publications for the current systematic review were selected in accordance with published PRISMA checklist criteria. Identified publications were then appraised utilising the Quality Assessment Tool for Quantitative Studies measure. Twelve of the thirteen identified studies reported that emotional dysregulation mediates the relationship between childhood maltreatment and subsequent depression in adulthood. Furthermore, there is some evidence to suggest that emotional abuse and neglect may be associated with the highest likelihood of emotional dysregulation, increasing the risk of depression in adulthood. Nevertheless, significant methodological limitations were observed and as such, recommendations for future research are presented.

Background: Women with bipolar disorder (BD) face an increased risk of suffering a severe episode of illness in the perinatal period, requiring them to make difficult choices regarding pregnancy. Decisions are made more difficult by a lack of research, particularly on the teratogenicity of psychotropic drugs. The studies in this thesis aimed to explore women's views on making these decisions, together with the views of health professionals who treat them. Methods: 1) a systematic review and meta-synthesis of the qualitative literature on motherhood for women with severe mental illness (SMI); 2) a qualitative study of women with BD's views on pregnancy via individual interviews and an online forum; and 3) qualitative studies of health professionals' views via a focus group and Individual interviews. Results: The review identified 23 studies on the views of women with SMI, and eight papers on professionals' views. Themes identified included guilt, stigma and problems with service provision. In the second study (21 women interviewed, 50 internet contributors), centrality of motherhood, contextual factors, stigma and fear were major themes. Fear and stigma were among themes echoed in the studies with perinatal specialists (11) and general psychiatrists (14). The idea of a decision aid was broadly welcomed by all three groups. Conclusions: This thesis highlights the complexity of the challenges facing women with BD when they have children and the ways in which issues such as stigma and fear mitigate against the establishment of a meaningful therapeutic relationship with health professionals. Improved access to information and specialist advice is needed and research on the possible usefulness of a decision aid would be helpful. It is further suggested that many health professionals would benefit from more training on this topic to better understand the particular importance of BD in the perinatal period and to reduce stigmatizing attitudes.

This thesis investigates the relationship between interpretation biases, social anxiety and paranoia. The overarching hypothesis is that differences in the manner of interpretation of emotionally ambiguous information may determine the type of symptoms which predominate, perhaps even which pathology emerges. For example, an interpretation which assumes an intent of harm (e.g. seeing two people whispering while looking at you means 'they are plotting against me') is likely to precipitate negative paranoid thoughts, perhaps ultimately persecutory delusions. In contrast, an interpretation of the same ambiguity, but which assumes the self is being negatively evaluated by others (e.g. seeing two people whispering while looking at you means 'they are talking about my faults') is likely to precipitate socially anxious thoughts and perhaps ultimately social anxiety disorder. Experiment 1 (n=84) examined the form and strength of the associations between the hypothesised different types of interpretation bias (socially anxious resolutions/paranoid resolutions) and the traits of social anxiety versus paranoia. The results showed that there was a significant association between different types of interpretation bias and its content specific emotion trait. The negatively evaluated (socially anxious resolutions) interpretation bias was more strongly associated with social anxiety than the persecutory (paranoid resolutions) interpretation bias. Conversely, the persecutory (paranoid resolutions) interpretation bias was more strongly associated with paranoia than the negatively evaluated (socially anxious resolutions) interpretation bias. Experiment 2 (n= 80) extended Experiment 1 by altering the task design to directly contrast paranoid and socially anxious interpretations (rather than contrasting paranoid versus non paranoid and, separately, socially anxious versus non-socially anxious). Under this forced-choice experimental design, the results revealed that there was a significant association between persecutory interpretation and its content specific emotion (paranoia). However, both types (negatively evaluated and persecutory) of interpretation bias were significantly associated with social anxiety. Experiment 3 (n = 71) was a longitudinal follow up study (for which Experiment 1 served as the baseline). This study tested whether content specific interpretation biases would predict corresponding traits six months later, as an indirect test of the causal role of interpretation biases in precipitating social anxiety and paranoia. The results showed that negatively evaluated interpretation bias predicted subsequent social anxiety traits, and likewise persecutory interpretations predicted subsequent paranoia traits. Moreover, there was an interaction effect between the two types of bias suggesting that the predictive power of paranoid interpretation bias was especially strong when coexisting with a strong socially negative interpretation bias, which lends support to some theories of clinical paranoia. Experiment 4 used a combination of existing and new data collected over a six-month period to test the hypothesis of reciprocal causality (do traits contribute to the exacerbation of congruent biases in addition to biases maintaining traits, such that a vicious cycle is established?). The data showed a reciprocal causality of the trait social anxiety in corresponding negatively evaluative interpretation bias, while there was an absent contribution of the trait paranoia to persecutory interpretation bias, which was instead due to the trait social anxiety. Experiment 5 recruited a large clinical sample plus healthy controls (n= 102) to examine similar questions to those tested in the previous subclinical samples. The results revealed a 3 pattern broadly consistent with the content specific interpretation biases in both social anxiety and paranoid psychosis patients. Findings of this thesis confirm the cognitive theories of psychopathology in social anxiety and early psychosis. The study demonstrates a reciprocal causality between social anxiety and both persecutory and negatively evaluated interpretation bias. It has confirmed the vicious circle proposed by cognitive theories of social anxiety, and suggested a mechanism for the maintenance of persecutory interpretation bias in paranoia. Although content-specific interpretation bias was weighted the most in the prediction of its corresponding trait characteristics, the effects from the content-unmatched interpretation bias could not be neglected. This finding extends previous observations using similar methods in subclinical samples, and suggests an aetiological pattern that goes beyond the notion of content specificity.

Traditionally, antipsychotic drugs have targeted the dopaminergic and glutamatergic neurotransmitter systems, with varying efficacy. While dopaminergic antipsychotics are effective at treating positive symptoms, they are less effective for negative symptoms and cognitive impairments, glutamatergic antipsychotics have had inconsistent results at the clinical trial level; and both have poor side effect profiles. Recently, interest has grown in the antipsychotic potential of cannabidiol (CBD), a non-intoxicating extract of cannabis. This has stemmed from cannabis research linking the endogenous cannabinoid (eCB) system to psychosis, with clinical studies of CBD showing relative success in patients with psychosis. Considered a multimodal drug, which may directly or indirectly affect multiple distinct modes of neural signalling (including the eCB system, and also the dopaminergic and glutamatergic systems); CBD displays none of the side-effects of traditional antipsychotics, has high tolerability, and may also display neuroprotective properties. However, what effect a challenge of the eCB system with acute CBD would have on the neurophysiological and neurochemical substrates implicated in psychosis is currently unknown. Therefore, the aims of the current thesis were to provide novel mechanistic insight into these neurophysiological and neurochemical processes potentially underlying the antipsychotic effects of CBD, in psychosis. To do this, I conducted four experiments within the same group of psychosis patients, employing functional magnetic resonance imaging (fMRI) and proton magnetic resonance spectroscopy (1H-MRS), with a CBD challenge. Patients were administered a single dose of oral CBD (600mg) compared to a matched placebo, on separate days, at least one week apart, using a double-blind, randomized, placebo-controlled, repeated-measures, within-subject cross-over design. Three hours after taking the study drug, participants were scanned using i) a block design fMRI paradigm, while performing a verbal paired associate learning task; ii) a resting-state 1H-MRS paradigm; and iii) a resting-state fMRI paradigm. 15 psychosis patients completed both study days. All patients were within 5 years of illness onset, and receiving standard antipsychotic treatment. 19 healthy controls were also scanned using the same task fMRI and 1H-MRS paradigms under identical conditions, and 14 separate healthy controls were scanned using the same resting-state fMRI paradigm under identical conditions. The healthy controls were not administered any drug. In the four experiments, I investigated the effects of a single dose of CBD in psychosis patients in relation to: i) brain function in regions implicated in psychosis during a verbal memory task; ii) the context-dependent functional connectivity between brain regions implicated in psychosis; iii) hippocampal glutamate, and the relationship between hippocampal glutamate and memory related functional activation; and iv) resting-state functional connectivity of the hippocampus, and of the three large-scale functional networks. In the first three experiments, CBD partially normalised abnormal memory related functional activation, functional connectivity, and the associations between hippocampal glutamate and memory related functional activation, in the psychosis patients. Acute CBD resulted in a non-significant increase in hippocampal glutamate in the psychosis patients, which was positively associated with task performance. Additionally, CBD resulted in trend-level reductions in symptoms, and in anandamide and palmitoylethanolamine (components of the eCB system), within the psychosis patients. These reductions did not correlate with any neurophysiological changes. No CBD related differences were observed in the resting-state fMRI experiment. Collectively, these findings provide novel insight into the neurophysiological and neurochemical mechanisms potentially underlying the antipsychotic effect of CBD in psychosis, and emphasise the importance of investigation of these effects with sustained treatment, and at different doses. Furthermore, these findings support a link between the potential antipsychotic effect of CBD and glutamatergic signalling pathways, which could have implications for psychosis patients who do not respond to traditional dopaminergic antipsychotic drugs.

Introduction: Neuronal surface autoantibodies (NSAbs) such as those targeting the NMDA receptor (NMDAR) were originally described in association with autoimmune encephalopathies which initially present with psychosis, usually alongside multiple neurological features. NSAbs might also contribute to the onset of illness in a subgroup of patients with psychotic disorders, and while they are evident in a proportion of patients at the first episode of psychosis, different antibody testing methods appear to give rise to different estimates of seroprevalence in psychiatric patient groups and in healthy controls. Furthermore, it is not known if NSAbs are present prior to illness onset, whether they occur as a sequel to infection and whether they associate with presenting symptomatology or with clinical outcomes. Methods: Sera from a) a cohort of 254 subjects at clinical high risk (CHR) for psychosis; b) a cohort of 398 first episode psychosis (FEP) patients and c) 116 healthy volunteers were tested for neuronal autoantibodies targeting multiple antigens, using a combination of fixed and live cell-based assays (CBAs), immunohistochemistry using fixed brain slices and immunocytochemistry using live cultured hippocampal neurons. Antibodies to common infectious pathogens were also measured in the CHR cohort. I compared the prevalence of these antibodies between groups, and examined their relationship to psychopathology, cognitive function, inflammatory markers and MRI measures. Antibody detection techniques were compared to establish if different methods detected different seroprevalence rates in cases and controls. Results: Seroprevalence of NSAbs did not differ between cases and controls, regardless of the assay used. Nevertheless, within the CHR sample, NMDAR antibody seropositive subjects had more severe affective and catatonic symptoms and poorer cognitive function than seronegative subjects. Interactions between the presence of NSAbs and the level of S100B, a marker of blood brain barrier (BBB) integrity, were found such that in patients with high BBB permeability, seropositivity for NSAbs was associated with more severe psychopathology across multiple domains and with larger volume of limbic structures. NSAbs did not predict transition to psychosis over two years in CHR subjects. In FEP subjects, NMDAR antibodies associated with shorter duration of psychosis and, unexpectedly, a better response to antipsychotic treatment. Live CBA consistently detected greater NSAb seropositivity rates than fixed CBA and appeared to detect antibodies which had more clinically relevant associations. GABAA receptor antibodies showed no useful clinical associations in these psychiatric cohorts. Toxoplasma exposure in CHR subjects was associated with more severe negative symptoms and poorer functional outcomes. Epstein-Barr Virus was associated with reduced risk of being at CHR for psychosis, reduced rates of transition to psychosis and better functional outcomes. NMDAR autoantibody seropositivity was associated with measures of exposure to Toxoplasma and influenza. Conclusion: Serum neuronal autoantibodies - and NMDAR antibodies in particular - are detectable at appreciable rates in early psychosis and appear to be associated with psychopathological, cognitive, and neuroanatomical abnormalities, particularly before the onset of frank psychosis. The pattern of findings in CHR subjects is qualitatively similar to that in autoimmune encephalitis, consistent with the notion than in a minority of patients with psychosis, the onset of illness may be related to autoantibodies to CNS antigens, and that they may have a useful biomarker role. Exposure to infectious pathogens is associated with clinical status and outcome in individuals at CHR for psychosis, and may in some cases be a trigger for autoimmunity in these individuals. These studies raise the exciting possibility of future targeted therapeutic or even preventative interventions focusing on either autoimmunity, infection or both at the earliest phases of psychotic illness.

Psychotic phenomena, such as hearing voices and being extremely paranoid, represent some of the most extraordinary and distressing mental states. Children and adolescents are significantly more likely to experience psychotic phenomena when they have been exposed to multiple forms of victimisation (poly-victimisation) although a substantial proportion of poly-victimised individuals will not develop these experiences. This thesis investigates why certain high-risk individuals do not develop psychotic phenomena in order to eventually inform early intervention efforts to prevent the emergence of psychotic phenomena among poly-victimised youth. Specifically, this thesis comprises four studies which investigate individual, family and community-level factors which are protective in relation to psychotic phenomena among poly-victimised children and adolescents. Analyses use data from the Environmental-Risk (E-Risk) Longitudinal Twin Study, a birth cohort of 2,232 twin children born in 1994 and 1995 in England and Wales. The first study investigates multi-level protective factors for age-12 psychotic phenomena among children exposed to polyvictimisation. The second study explores multi-level protective factors for age-18 psychotic experiences among poly-victimised adolescents. The third study considers gender differences in terms of the protective influences of social support on psychotic phenomena among poly-victimised adolescents. The fourth study utilises discordant twin methods to investigate whether the association between social support and the absence of adolescent psychotic experiences is environmentally mediated, after accounting for family-wide (including genetic) factors. Collectively, these studies identify multi-level protective factors for psychotic phenomena amongst poly-victimised youth, although these also appear to be associated with an absence of psychotic phenomena in this cohort regardless of polyvictimisation exposure. If replicated, these findings will have practical implications for interventions aiming to prevent the occurrence of early psychotic phenomena and the potential to prevent subsequent mental health problems.

People at Clinical High Risk for psychosis (CHR-P) present with a clinical syndrome that includes attenuated positive psychotic symptoms and impairments of social/emotional functioning. The CHR-P state is associated with a 20% risk of developing psychosis over a two-year period. However, treatment options remain limited-no licensed pharmacological therapies are currently available, and the comparative efficacy of the available treatments remains unknown. The purpose of this PhD was to address aspects of these outstanding issues in two distinct but complementary ways. Part 1: of this thesis provides an evidence synthesis of the CHR-P treatment literature. Following a qualitative introductory review, I used systematic review and network meta-analyses to compare and summarise the relative efficacy and acceptability of current treatments for (a) preventing transition to psychosis from a CHR-P state [Paper 1], and (b) reducing attenuated positive psychotic symptoms [Paper 2]. The results of Papers 1 and 2 indicated that, to date, there is a lack of evidence that any specific intervention is particularly effective over any others in preventing transition to psychosis from a CHR-P state or in reducing attenuated positive psychotic symptoms. These results also suggest a need to identify potential novel therapeutics that may better target the pathophysiological mechanisms underlying psychosis onset and that may thereby alter the course of the disorder. Part 2: of this thesis sought to examine the neurophysiological basis for the effects of a potential novel treatment strategy, the neuropeptide oxytocin, in an intranasal oxytocin vs placebo acute challenge study using magnetic resonance imaging. Part 2 starts with the rationale for selecting oxytocin from the numerous candidate compounds indicated for those at CHR-P, followed by a discussion of the oxytocinergic system and its links to psychosis and the CHR-P state. I then present two experiments that tested whether oxytocin modulated (a) resting cerebral perfusion using arterial spin labelling, with a particular focus on the hippocampus [Paper 3], and (b) neurochemical metabolite levels (particularly glutamate, and glutamate plus glutamine, Glx) in the hippocampus, thalamus, and anterior cingulate cortex, as measured using proton magnetic resonance spectroscopy [Paper 4]. The results of the two neuroimaging experiments suggested that oxytocin can modulate hippocampal perfusion in CHR-P individuals [Paper 3] - which is a key pathophysiological mechanism strongly implicated in psychosis onset- but does not appear to have effects on glutamate or Glx concentrations in the hippocampus, thalamus, or anterior cingulate cortex [Paper 4]. However, based on the hippocampal perfusion findings alone, oxytocin merits further investigation as a candidate novel treatment for this group. Part 3: of this thesis extends the discussion of the findings from Part 1 and Part 2. Here, I use the results of the evidence syntheses and wider literature to highlight a number of critical challenges, and proposed solutions, for future CHR-P treatment research. I then place the experimental oxytocin findings in the context of CHR-P pathophysiology, before bringing together the two distinct methodological approaches (evidence synthesis and experimental medicine) to show how both-combined-could facilitate a new era of improved interventional research in the CHR-P field.

The human cerebellum is one of the earliest brain regions to develop, but continues to grow until late adolescence and adulthood. This makes it especially vulnerable for disorder and disease, specifically affecting neurodevelopmental disorders such as obsessive compulsive disorder (OCD) and autism spectrum disorder (ASD). Cerebellocortical loops further allow the cerebellum to interact with the neocortex, enabling changes in the cerebellum to influence cortical structures. This thesis aimed to identify cerebellar and neurocognitive mechanisms modifying (1) typical and (2) abnormal cerebellar growth and brain development. The first aim has been addressed in a series of two studies, exploring plastic changes in cerebellar and cortical functional connectivity (Chapter 2.1), volume, shape and white matter microstructure (Chapter 2.2) in healthy adolescents after learning to drum for 8 weeks. In order to explore distinct and shared cerebellar contributions to abnormal brain development, the role of the cerebellum and brainstem to classify adult and paediatric ASD was studied (Chapter 3.1); the reshaping of cerebellar and brainstem vertex (Chapter 3.2) and neocortical thickness and surface area (Chapter 3.3) with repetitive behaviours and symptom severity was explored in paediatric ASD and OCD. As compulsive and repetitive behaviours are one of the main measurements assessing diagnostic features in psychiatric animal models, distinct brain-behaviour correlates for OCD-phenotypical compulsive and anxious behaviours were studied in context of cerebellar insulin regulation in a mouse model of diabetes mellitus (Chapter 4). The data provided underline the cerebellum’s central role in neurocognitive mechanisms and functions important for neurotypical development and functioning. While abnormal cerebellar development co-occurs with neocortical dysfunction and has been connected to a range of childhood disorders, its plastic changes to complex skill learning during adolescence also co-occur with neocortical functional and structural changes. This opens up an opportunity of the cerebellum as a target for possible early intervention for childhood disorders.

Background Stigma is an important and influential factor in the process of help seeking. Contrary to previous research on stigma and help seeking across populations (Clement, 2014), a recent systematic review of stigma as a barrier to help seeking within the Armed Forces (AF) (Sharp, 2015) found no association between stigma and help seeking. The contribution of qualitative literature in this area has largely been ignored, despite the value this research brings to our understanding of complex social constructs such as stigma. The aim of the current systematic review of qualitative studies was to identify appropriate literature, assess the quality and synthesis findings across studies regarding evidence of stigma related barriers and facilitators to help seeking within the AF. Method. A multi-database text word search incorporating searches of PsycINFO, MEDLINE, Social Policy and Practice, Social Work Abstracts, EMBASE, ERIC and EBM Review databases between 1980 to April 2015. In addition references of reviews and grey literature were also searched. Literature was quality assessed using the CASP tool. Thematic synthesis was conducted across the literature. Results The review identified 8 studies with 1012 participants meeting inclusion criteria. Five overarching themes were identified across the literature: 1) Non-disclosure 2) Individual beliefs about mental health, 3) Anticipated and personal experience of stigma 4) Career concerns and 5) Influencing stigma. Conclusions The findings from the current systematic review found that unlike inconsistent findings in quantitative literature, there was significant evidence regarding the relationship between stigma and help seeking for mental health difficulties within the AF. The study advocates for future refinement of measures to accurately capture the complexity of stigma and help seeking.

Psychosis is a complex condition that features in many psychiatric disorders, and significantly affects the quality of life for both patients and family members. As part of the Genetics and Psychosis (GAP) study, this thesis presents one of the largest blood gene expression datasets on first-episode psychosis patients to date. This work aimed to characterise the blood-based biological perturbations in psychosis and to investigate the predictive ability of gene expression data. Firstly changes in expression, between healthy controls and first-episode psychosis patients was explored, to identify genes associated with psychosis. I identified hundreds of differentially expressed genes and found associations to pathways involved in transcription, oxidative stress and viral replication. Secondly, network approaches were used to construct modules of genes based on co-expression. I identified modules correlated to the severity of positive symptoms, and enrich-ment for pathways associated with the stress response and multiple brain regions. Thirdly regularised generalised linear models with bootstrapping were used to generate predictions based on combinations of gene expression, genetic and demographic data. The highest performance was found for models incorporating gene expression data, with minimal improvement using additional data. Prediction accuracy for identifying psychosis samples was found to increase with severity of positive symptoms in schizophrenia samples, but not in other psychoses. Finally, machine learning methods were used on public schizophrenia gene expression data to build a variety of predictive models. These models were tested on the Genetics and Psychosis (GAP) gene expression data. The results show increased predictive performance on samples with a schizophrenia diagnosis, compared to other types of psychosis. Overall the thesis presents work analysing a novel gene expression dataset. The results suggest that blood gene expression signatures are more predictive for positive symptoms in schizophrenia than for other psychoses. This work also highlights expression differences in innate immune pathways and the stress response.

In recent decades, the rate of very preterm birth (< 32 weeks of gestation) has increased, and it is now estimated as occurring in approximately 1 in 9 live births. Multiple lines of research suggest that very preterm birth is associated with a range of psychiatric and neurodevelopmental problems throughout childhood and adolescence. However, little is known about adult outcomes. This PhD sought to delineate the cognitive and psychiatric profile of adults born very preterm. It included four related studies. Firstly, IQ trajectories were examined in order to understand whether IQ improves or remains stable from the age of 8 to the age 30 years. The results indicated that individuals who were born very preterm and especially those born at the lower end of the gestational age spectrum, continue to be at higher risk of cognitive impairment in adult life, affecting Performance IQ in particular. Secondly, cognitive outcomes were compared between very preterm born adults and age-matched controls, with an emphasis on executive function. The influence of cognitive outcomes on social functioning and achievement was also examined. Individuals born very preterm performed worse than controls on measures of IQ and executive function. They also demonstrated significantly lower achievement levels in terms of years spent in education, employment status, and on a measure of functioning in work and social domains. Persisting executive function impairments in very preterm survivors were associated with achievement in several real-life domains. Thirdly, considering that very preterm birth is associated with an increased rate of psychiatric disorders, a dimensional approach was utilized to examine psychiatric symptomatology in adults who were born very preterm and controls. Moreover, the specificity of this risk was examined in order to better understand their adult clinical profile. Very preterm individuals demonstrated elevated psychiatric symptomatology compared to full-term controls. Psychiatric risk was characterized by a non-specific clinical profile and was associated with lower IQ. Lastly, salience attribution, thought to underlie psychiatric symptomology, was examined. Adults born very preterm displayed reduced capacity to process adaptive salience, indicating they may have difficulties in distinguishing between high and low probability stimuli features. Salience processing was associated with negative and positive psychotic-like symptoms, lending support to the hypothesis that very preterm individuals may be a distinct subgroup of high-risk individuals, characterized by increased ‘cognitive’ psychotic-like experiences. This study aimed to understand the neuropsychiatric and cognitive profile of adults born very preterm. It demonstrated that in adulthood they experience similar difficulties to those that are evident earlier in life. These findings emphasize the need for cognitive remediation programmes to be delivered to vulnerable groups, which thus far have targeted specific executive function components (e.g. working memory, cognitive control), and may one day show generalizable benefits for a successful overall life adjustment.

Background: People with schizophrenia (SCZ) suffer from impaired cognitive abilities and these are associated with poor functional outcomes. Cognitive Remediation Therapy (CRT) has been shown effective in improving the cognitive deficits of SCZ. Because there is evidence for CRT treatment heterogeneity of outcomes, there is a need to identify CRT predictors of differential response using moderation analysis of high dimensional psychiatric data, which typically contain relatively large percentages of missingness. This will contribute to precision medicine treatment, understanding mechanism responsible of differential therapy responses, and better prognosis. Aims: The primary aim of this PhD consisted of developing a CRT precision medicine model, using computer intensive statistical learning methods able to deal with high dimensional psychiatric data containing large percentages of missingness in the predictors and smaller per-centages in the outcome. Secondary aims were overcoming the following problems: variable selection or measurement of variable importance in the model, multicollinearity and overfitting, and summarising commensurate outcomes in one latent outcome. Methods: A simulation study comparing four statistical learning methods (Lasso, Elastic-net, Random Forests and Conditional Inference Random Forests) combined with two missing data imputation techniques (Multivariate Imputation using Chained Equations and MissForest) was run. The combined methods were assessed according to their optimism-corrected (via bootstrap internal validation) prediction accuracy and variable selection performance in differ-ent scenarios. The best method was chosen to develop a CRT precision medicine model using individual participant data from seven randomised controlled trials with approximately 400 pa-tients. Factor scores from a latent summary measure of cognitive commensurate outcomes, obtained via Factor Analysis, was used as the model dependent variable, to accommodate the above univariate statistical learning methods. Results: In the simulations, the method combining MissForest imputation with Lasso was the best compromise between prediction accuracy and clinical interpretability. MissForest-Lasso was then used to develop an internally validated precision medicine model, which se-lected only a weak moderator of treatment response. The model was therefore mainly prog-nostic. Conclusion: In future research, more modalities of data, such as genetics, OMICS and neuroimaging data, are recommended to successfully identify moderators of CRT success.

Approximately a third of patients with schizophrenia do not respond to antipsychotic treatment targeting the dopamine system, suggesting that a separable neural dysfunction may drive psychosis in these patients. This thesis aims to probe the mechanisms underlying treatment response by investigating two cognitive processes which have been implicated in schizophrenia – cognitive control and reinforcement learning – as well as brain myelination. The key hypotheses are that 1) treatment resistant schizophrenia emerges due to a failure to exert cognitive control, characterised by prefrontal hypoactivation and functional dysconnectivity, 2) treatment responsive schizophrenia is selectively associated with a subcortical dopaminergic dysfunction, evident in an abnormal neural signature of reward prediction error (RPE) during reinforcement learning, and 3) treatment resistant schizophrenia is characterised by exacerbated structural dysconnectivity as indexed by myelin content. To dissect these mechanisms, performance and neural activation during a cognitive control task and a reinforcement learning task, as well as myelin water fraction (MWF) were compared between 22 treatment resistant patients, 21 treatment responsive patients, and 24 healthy controls. Treatment resistant and responsive patients showed similarly impaired performance on both tasks compared to controls. During the cognitive control task, resistant patients showed an inverse correlation between frontal activation and psychotic symptoms as well as reduced functional fronto-thalamic connectivity compared to controls. During the reinforcement learning task, responsive patients showed reduced cortical and subcortical RPE related activation compared to controls and treatment resistant patients. MWF was reduced in patients compared to controls in several white matter regions but did not differ between the two patient groups. The findings support distinct neural mechanisms underlying treatment resistant and responsive schizophrenia despite similar behaviour. Functional dysconnectivity within the cognitive control network and a deterioration of frontal activation as a function of symptom severity may perpetuate psychosis despite dopaminergic treatment in treatment resistant schizophrenia, although this is not reflected in an exacerbated myelin dysfunction. The results highlight the importance of stratifying patient samples by treatment response status in future research.

The biological mechanisms through which childhood physical and sexual abuse increase the risk of psychosis are unclear. This thesis investigated the presence of abnormalities in brain structure and in the hypothalamic-pituitary-adrenal (HPA) axis activity in people at their first episode of psychosis (FEP) and controls, with and without a history of childhood abuse. Structural MRI scans were acquired in 86 FEP patients (49 abuse positive) 64 controls (30 abuse positive). The HPA axis was evaluated in 169 FEP (110 abuse positive) and 133 controls (67 abuse positive). In this population I explored differences in grey matter volume and cortical thickness and cortisol concentrations during the day and at awakening in association with abuse exposure. Finally I correlated cortisol levels with cortical thickness in regions sensitive to abuse. Childhood abuse was associated with smaller grey matter volume and reduced cortical thickness in frontal and occipital brain regions. I found an interaction between psychosis and abuse in frontal and parietal regions, which had greater grey matter volume and cortical thickness in controls and smaller in FEP exposed to childhood abuse. There was an interaction between psychosis and abuse in the cortisol awakening response (CARg) as FEP exposed to severe childhood abuse showed a blunted CARg while controls exposed to severe abuse had higher CARg. Finally there was a negative correlation between cortisol and cortical thickness: larger thickness in brain frontal and parietal regions sensitive to abuse exposure were correlated with lower cortisol levels during the day in controls but not in FEP. Childhood abuse has a differential impact on the brain structure as well as the HPA axis activity in FEP and controls. The relationship between cortisol and cortical thickness in regions sensitive to abuse found only in controls suggests an integration of brain structure and HPA axis activity and possibly an adaptive mechanism to environmental stress.

The early stages of the genome-wide association study (GWAS) era were dominated by studies focusing on single phenotypes, while in recent years there has been growing interest in multi-trait GWAS. A wide variety of multi-trait GWAS methods have been developed, but publications introducing new methods are highly inconsistent in their evaluation of method performance, obscuring their relative merit. Facilitated by burgeoning national biobank resources, multi-trait analyses are set to become more routinely applied, making understanding their relative performance increasingly important. We develop a simulation framework to model the complex networks underlying multivariate genetic epidemiology. We exploit our simulation framework to perform a comprehensive comparison study of the leading multi-trait GWAS methods, providing a web application and open-source software program implementing our simulation framework for further benchmarking of multi-trait GWAS methods. Motivated by our comparison results, we develop novel methodology and present a series of multi-trait analyses. We perform multi-trait genome-wide analyses on publicly available GWAS summary statistics on 19 traits – metabolic, anthropometric and psychiatric. We develop and apply two summary statistic methods: one that has increased power to detect pleiotropic effects on multiple traits, and one that is more powerful for detecting heterogeneous genetic effects. Polygenic risk scores (PRS) are now a commonly used tool for performing phenotype prediction from genetics, assessing the genetic aetiology underlying diseases, and testing for shared genetic aetiology among traits. Using UK Biobank data, we explore the predictive ability of PRS computed across multiple traits for Major Depressive Disorder (MDD). The MDD PRS itself has so far offered modest prediction of MDD case/control status; we explore the use of PRS built on traits correlated with MDD to improve predictive ability. We build main effect and interaction models, using both AIC and BIC stepwise variable selection, and cross-validation, to establish the most predictive models.

This! systematic literature review critically appraised and compared methods for measuring individual differences in inner speech in developmentally neurotypical adults. It also investigated whether individual differences in inner speech which were found through these methods of measurement, were related to psychopathology. Literature searches were performed in three databases. Twenty nine studies were included in this review. These studies measured inner speech through a variety of methods including questionnaire and interview, introspective, experimental, neuroimaging and lesion methodologies. Methods varied with regard to the characteristics of inner speech they measured, as well as the!qualities of the inner speech which the measurement method relied on. Only two of the included studies used two different types of inner speech measurement. This combining of different methods enabled reciprocal validation of each method, and indicated that some findings may be artefacts of the particular method used. Individual differences in inner speech found by these methods of measurement, were related to variation in other aspects of human experience. This included variation in psychological disorders and symptoms such as auditory verbal hallucinations, anxiety and self-esteem. These preliminary findings suggest that a more thorough assessment of the relationship between inner speech processes and psychological disorders and symptoms would be useful for the advancement of clinical psychology theory and treatment.

The Ultra High Risk (UHR) state is a clinical syndrome that is associated with a high risk for imminent psychotic disorder. Examination of the symptoms in the UHR population using a dimensional approach could improve our understanding of the clinical and neurobiological heterogeneity within UHR samples, as it has done in schizophrenia. Two previous studies in UHR subjects have found that the symptoms measured by the CAARMS cluster on to underlying psychopathological dimensions. Moreover, both found a link between severity of scores on negative and disorganised factors and the risk of later transition to psychosis. However, one study described five symptom dimensions whereas the other described a three dimensional model, and neither structure has been independently replicated. In the present study, CAARMS data from a total of 461 UHR subjects were used to perform principle axis factoring (PAF) and confirmatory factor analysis (CFA). Scores for each factor of the structure found to best fit the data were then correlated with resting regional cerebral blood flow (rCBF) at presentation and with clinical outcomes after two years of follow up. PAF revealed a five factor structure, which was confirmed using CFA. In the subsample of subjects who were studied with neuroimaging, scores on the disorganisation factor were associated with reduced rCBF in the thalamus, scores on the anxiety factor were associated with reduced rCBF in the insula, and Total CAARMS scores were associated with reduced rCBF in the left hippocampus. Higher scores on the anxiety factor and a higher total CAARMS score were associated with an increased risk of later transition to psychosis, while higher scores on the disorganisation factor was associated with worse functional outcome. These findings suggest that symptom dimensions in the UHR state have distinct neural substrates and clinical correlates. Variation in these dimensions may contribute to the marked heterogeneity of clinical presentations and outcomes within samples of UHR subjects.

Introduction: Psychotic disorders are known for their wide variability in clinical and social outcomes beginning from illness onset and throughout their course. Our current lack of understanding of the origins for this heterogeneity is further compounded by dearth in knowledge on how patients come to the attention of mental health services and methodological incongruity across different studies. Employing samples of first episode psychosis (FEP) patients, the aims of this thesis were to: 1) look back on the pathways to care patients used to enter mental health services and the use of prodromal services in South London; and 2) examine trajectories of the psychotic disorders and potential predictors of their longitudinal outcomes. Methods: Two large samples of patients with FEP (i.e., GAP and EU-GEI) were utilised in this thesis. For the study 1, information on pathways to care undertaken prior to coming to generic services for FEP was extracted from the Biomedical Research Centre (BRC) Case Register Interactive Search (CRIS) system. For studies 2, 3 and 4, using electronic clinical records, extensive information in three domains-clinical, social and service uses-was collated over 4-5 years after contact with mental health services. Results: Only a small fraction of individuals (4.1%) who present with FEP to the main secondary mental health provider have previously been in contact with prodromal services and made a subsequent transition to psychotic disorder; 77% this sub-group of patients entered their pathway to care via referral from General Practice or other health professional. In contrast, 45% of FEP group without prior contact with the prodromal services made first contact with mental health services via emergency services and 18% of this group were referred by the criminal justice system. Further, combining the baseline schizophrenia diagnosis with five symptom dimensions (i.e., positive, negative, excited, disorganised/concrete and depressed dimensions) generated the best model fit for predicting time to first remission. During the 5-year follow up after first contact with mental health services, a higher proportion of Black African and Black Caribbean ethnicity had compulsory re-admissions and instances of police involvement during an admission to a psychiatric unit compared with White British ethnic group. Patients of Black African and Black Caribbean ethnicity did not differ from White British ethnic group in overall functional disability and illness severity, or frequency of remission or recovery during the follow up period. However, patients of Black ethnicity become increasing socially excluded as their illness progress. In a sample of first-episode schizophrenia spectrum patients, 35% of the sample met the criteria for treatment resistance (TR) at the end of the first 5 years of follow up. Of these TR patients, 70% of these were treatment resistant from illness onset. Those who subsequently developed TR were more likely to have an early illness onset (<20 years) compared to those with non-TR. The relationship between an early age of onset (<20 years) and TR was specific to patients of Black ethnicity and patients of male gender. Conclusions: Very few of those who come to FEP services come after being seen for an at-risk-mental state by prodromal services suggesting that the scope for reducing or delaying onset of psychosis by this means may still be limited. My results indicate that supplementing the baseline categorical schizophrenia diagnosis with ratings on five symptom dimensions improves the prediction of delayed treatment response as measured by time to first remission. Further, the longitudinal trajectory of psychosis in patients of Black ethnicity did not show greater clinical or functional deterioration than white patients. However, their course remains characterised by more compulsion, and longer periods of admission. Finally, I showed that for the majority of the TR group, lack of response to antipsychotic treatment is present from illness onset, necessitating a consideration for an earlier use of clozapine.

Although cannabis use following a first episode of psychosis (FEP) has been linked to poor outcome such as relapse, current understanding is limited regarding the nature of this association. To shed light on the nature of this relationship, this thesis employed multiple methodological approaches: First, I pooled together data from previous studies in order to estimate the cross-sectional effects of continued and discontinued cannabis use following the onset of psychosis and risk of relapse, using a meta-analytic design (Paper 1). Second, I prospectively collected follow up data from a cohort of patients presenting with a first episode psychosis to psychiatric services in south London. Paper 2 examined the magnitude of effect of continued cannabis use on risk of relapse and related outcomes by grouping patients into classes of different patterns of cannabis use following onset of psychosis. In Paper 3, longitudinal modelling was employed to further examine whether the link between continued cannabis use and risk of relapse persists when non-causal explanations for the association are taken into account (reverse causation, premorbid genetic confounding). My findings indicate that that former regular users who stopped using cannabis after the onset had the most favourable illness course with regard to relapse. The most unfavourable course was present in those who continued to use high-potency cannabis in a high frequency manner. The results also suggest that this association reflects a dose-dependent association that is unlikely to be a result of reverse causation (e.g. selfmedication) or genetic and environmental (e.g. other illicit drug use, medication nonadherence) confounding. These findings point to reductions in cannabis use as a crucial interventional target to reduce risk of relapse in patients with a first episode psychosis.

Several decades of research attest to the importance of social relationships for mental and physical health. Social isolation and loneliness have often been considered adversities associated with aging. However, dissatisfaction with one’s quality or quantity of social relationships is particularly common among young people. This thesis comprises four studies investigating the experiences of social isolation and loneliness in children and young adults, using a longitudinal, genetically-sensitive study design. Data were drawn from the Environmental Risk (E-Risk) Longitudinal Twin Study, a birth cohort of 2,232 twins born in 1994 and 1995. The first empirical chapter investigates associations between social isolation and mental health in the early school years. Longitudinal data from childhood to preadolescence is used to test whether social isolation predicts increases in mental health problems over and above pre-existing difficulties, and vice-versa. The second empirical chapter examines the relationship between social isolation, loneliness and depression in young adults. Behavioural genetic modelling using twin data is used to test the relative contributions of genetic and environmental influences to these associations. The third empirical chapter investigates childhood antecedents of loneliness, and broadly documents the profile of lonely young adults in multiple domains of functioning, including mental health, coping strategies, career prospects and interpersonal perceptions. The fourth empirical chapter analyses the associations between loneliness and aspects of sleep quality in young adults, using the monozygotic twin-differences method to control for familial sources of confounding. A potential exacerbating role of past exposure to violence victimisation on the association between loneliness and sleep is tested. Together, the findings underscore the pervasive role of loneliness in the well-being of young adults, and reinforce the importance of early intervention to prevent its persistence.

The majority of schizophrenia patients exhibit significant, diffuse cognitive impairment. Understanding the course and nature of this impairment is essential to elucidating etiology and treatment. Previous studies have been unable to trace the course of cognition from infancy to adulthood and assess true developmental change. Cognitive deficits in individuals with subclinical psychotic experiences may provide important clues about clinical psychotic disorders. Yet, developmental change and the effects of age, sociodemographic and familial factors have not been examined. Moreover, while there is evidence for connection abnormalities in the brains of schizophrenia patients, little is known about the structure of cognitive functioning across the psychosis spectrum i.e. in subclinical psychotic experiences and clinical psychotic disorders. The first study examined the association between subclinical psychotic experiences and cognitive functioning in a general population sample. Adults with psychotic experiences showed significant verbal and memory, but not processing speed deficits. Only older adults with psychotic experiences showed medium to large verbal and memory deficits when adjusting for sociodemographic factors, psychiatric morbidity and cannabis use. First-degree relatives showed a significant verbal, but not memory impairment. The second and third studies used data from a longitudinal cohort to examine 1) the course of cognitive functioning from infancy to adulthood and 2) cognitive network structure in childhood and adulthood. Individuals with depression, psychotic experiences, affective psychotic disorder and non-affective psychotic disorder were compared to controls. Affective psychotic disorder, psychotic experiences and depression groups showed a degree of cognitive impairment across infancy and adulthood, but only those with non-affective psychosis showed large, progressive deficits across multiple domains. Individuals with non-affective, affective and subclinical psychosis also showed qualitative and quantitative abnormalities in cognitive network structure, with the two clinical groups showing larger, widespread anomalies. Controls showed increasing reliance on working memory between childhood and adulthood, while all other groups remained reliant on low-level cognitive processes. Altogether, these findings suggest that the course of cognitive impairment differs across the psychosis spectrum. Despite distinct profiles of impairment, abnormalities in cognitive network structure were seen across the spectrum, highlighting the importance of looking beyond cognitive deficits to how performance is achieved.

Background: Investigating long-term antipsychotic polypharmacy is key to unpacking the associations between serious mental illnesses (SMI) and detrimental outcomes, such as premature death and frequent hospital readmissions, observed in this population. However, existing research is sparse and hampered by methodological problems such as examining small and homogeneous samples and residual confounding. Objectives: 1) To identify cases on long-term antipsychotic polypharmacy (≥ 6 months) prescribing in South London and Maudsley electronic health records (EHR); 2) To identify factors that predict long-term antipsychotic polypharmacy prescribing for SMI patients in secondary mental health care; 3) To investigate whether outcomes such as hospital readmission and mortality are associated with long-term antipsychotic polypharmacy prescribing in secondary mental health care. Methods: Antipsychotic medication information was derived from the Clinical Record Interactive Search (CRIS), a de-identified electronic patient records system, for the period between 2007 and 2014. Data on mortality were extracted using existing linkages between CRIS and death certification (Office of National Statistics). Information about antipsychotic co-prescribing was extracted using a bespoke algorithm. Multivariable logistic models were built to investigate predictors of antipsychotic polypharmacy. To investigate the impact of antipsychotic polypharmacy on hospital readmission and all-cause mortality, I constructed multivariable Cox proportion hazard models. To test the association between long-term antipsychotic polypharmacy and cause-specific mortality I used competing risk regression. Implications: On a clinical level, this thesis provides an insight into factors that can predict clinical decision-making regarding antipsychotic polypharmacy prescribing in real-life clinical settings. On a patient level, the findings highlight patient burden associated with this antipsychotic regimen. In the wider treatment, service and policy context, the lack of patient benefit from antipsychotic polypharmacy highlights the need for programmes that target prescribers, to reduce antipsychotic polypharmacy.

The work presented in this doctoral thesis examines magnetic resonance imaging (MRI) data on a sample of young adults who were assessed for the presence of psychotic experiences (PEs) such as delusions, hallucinations, and thought interference. The manifestation of PEs is considered a risk factor for psychosis that becomes greater with persistence of symptoms. The aim of this body of work was to examine the state of the brain using MRI in individuals with PEs, taking into account previous reports on psychosis as well as typical neurodevelopment. We have carried out a range of imaging techniques and analyses on a large multi-modal imaging dataset of 111 healthy controls, 67 with transient PEs, and 69 with persistent PEs. I initially set out to examine brain structure and function in relation to working memory through voxel-based morphometry of grey matter and investigating neurofunctional signalling in relation to a working memory task. In addition, I sought to characterise the temporal dynamics of regional activation in frontal and parietal regions involved in working memory using dynamic causal modelling. Next, we carried out an in-depth analysis of the cortical surface morphometry and sought to characterise the underlying white matter connectivity using atlas-based tractography. Finally, we carried out virtual dissections of the corpus callosum, the inferior fronto-occipital fasciculus, and the superior longitudinal fasciculus. We found qualitative evidence of protracted development of working memory network configuration in relation to duration of PEs but neuroanatomical evidence did not show consistent support. We found evidence of disturbances in cortical folding in relation to persistence of PEs but little evidence of accompanying alterations in white matter. Analyses of dissected tracts did highlight differences along the tract in relation to PEs. We discuss these findings in relation to their implication for PEs as a risk factor for a psychotic disorder and as a manifestation of an atypical neurodevelopment trajectory.

ZNF804A has recently been identified as a disease candidate gene for several severe neuropsychiatric disorders including bipolar disorder, schizophrenia and autism. Schizophrenia is known to share several clinical, genetic, pathophysiological, neuroanatomical and cognitive features with bipolar disorder, and thus ZNF804A may represent a useful target for investigating the common mechanisms underpinning these diseases. The latest advances in stem cell research have provided a wide variety of cellular models for studying these mechanisms in cultures of human neurons. One of these lines, a conditionally immortalised, cortically derived human neural progenitor cell (NPC) line (CTX0E16), offers a valuable tool for studying ZNF804A in human cortical neurons. The primary aims of the current investigation were to characterise neurons derived from the CTX0E16 NPC line and to subsequently employ this line in conjunction with other neuronal models in the study of ZNF804A expression, subcellular localisation and biological function, specifically focusing on its potential impact on neuronal morphology. To this end, neuronal cultures differentiated from the CTX0E16 cell line were characterised using a broad set of techniques. These cells were shown to predominately adopt a glutamatergic fate, and demonstrated the morphology, functional responses, synaptic expression and electrophysiological characteristics consistent with immature cortical pyramidal neurons. When examined in human neurons, ZNF804A was shown to be distributed throughout the neuron and demonstrated a novel distribution at putative synapses. ZNF804A was also located within the nucleus of both NPCs and neurons. Interestingly, this pattern changed as NPCs terminally differentiated into excitatory neurons. Moreover, in mature neurons, the presence of ZNF804A within the nucleus was modulated by activity-dependent stimulation. siRNA-mediated knockdown of ZNF804A revealed a role for this protein in early neurite outgrowth, an effect potentially mediated by neuroligin 4x. Knockdown of the rodent homologue, Zfp804A, further revealed a role in the maintenance and plasticity of dendritic spines in mature neurons. Collectively, these data show that the CTX0E16 cell line provides a useful model of human cortical neuron development and disease, and that ZNF804A plays a role in the development, maintenance and plasticity of key aspects of pyramidal neuron morphology. As alterations in these factors are thought to be central to the pathophysiology of schizophrenia and bipolar disorder, it is thus possible that ZNF804A represents a common risk factor that contributes to the cellular disturbances in these diseases.

Background: Self harm is an increasingly common behaviour, associated with poor mental health, and an increased risk of death by suicide and other causes. It is one of the principle reasons for admission to inpatient psychiatric services, however very little is known about self harm on wards. Aims: This thesis set out to address a number of gaps in the literature identified following a systematic review of studies of inpatient self harm. Main aims were to describe the characteristics of self harming behaviour within a national sample of services, and to investigate perceptions of self harm and views of harm minimisation practices amongst inpatient nursing staff. Methods: Aims were addressed in two studies using a mixed methods approach. Study 1 investigated the characteristics of self harming behaviour within inpatient mental health services across the UK, through a cross-sectional documentary analysis of incident reports. Study 2 was a sequential explanatory study of nursing staff attitudes towards self harm, composed of two phases; Phase I measured staff attitudes and their relationship to staff characteristics, using the Self Harm Antipathy Scale, and Phase II was a qualitative interview study of staff understandings of self harm. Results: Inpatient self harm was more frequent within acute vs forensic services, largely took place in the private areas of the ward, during the evening hours, and constituted a wide range of behaviours of which cutting was the most common. Inpatient nursing staff generally demonstrated positive attitudes towards self harm, however being a healthcare assistant, or from a non-white ethnic group were associated with more negative attitudes, as were lower SF-36 scores. Staff differentiated between acts of ‘self harm’ and ‘attempted suicide’ using a wide range of criteria which differed between individual participants. Views of harm minimisation practices were mixed. Conclusions: Specialist training in mental health would be beneficial to all practitioners working with people who self harm, and should particularly focus on the interpersonal reasons for self harm. Amongst culturally diverse teams of staff there are likely to be multiple understandings of self harm, and those from high religiosity minority ethnic backgrounds may be less accepting of the behaviour. Use of the term ‘attempted suicide’ is problematic and should be avoided. A harm minimisation approach, whilst potentially beneficial to service users, will present significant challenges to some nursing staff.

Depressive disorders are common and debilitating conditions; however, current methods of diagnosis and treatment are suboptimal, largely due to a lack of understanding of the biological basis of these disorders. Neuroimaging has provided substantial insights in this area, but one particularly understudied area is the relationship between unipolar and bipolar depression. These disorders have similar symptom profiles but require different treatment strategies, making their diagnosis and management challenging for clinicians. The overarching aim of this thesis is to understand differences and similarities in the structure and neurochemistry of neurobiological systems underlying unipolar and bipolar depression. This question is addressed in three ways: Firstly meta-analyses structural neuroimaging studies looking at alterations in grey and white matter were performed to identify patterns of changes that were common or specific to either disorder. Secondly, an original investigation was carried out to identify patterns of neurochemical alteration that differ between unmedicated patients with unipolar and bipolar depression. Lastly, the appropriateness of a dimensional approach to bipolarity in depression was evaluated by looking for structural neural correlates of bipolar symptoms within patients with unip9olar and bipolar depression. The results of these studies show that although many neurobiological alterations are common to unipolar and bipolar depression, there are changes in grey matter volume that are specific to unipolar depression, and changes in white matter volume that are specific to bipolar depression. However, alterations in grey matter volume do not correlate with bipolarity when treated as a dimensional characteristic. These results contribute to our understanding of structural and neurochemical alterations in depressive disorders, and provide targets for future research into improved diagnosis of these conditions.

Early psychosis is a high-risk period for suicide. Of concern, suicide rates in psychosis have increased over recent years. Several risk factors have been established, including being male, white, recent loss, previous suicide attempts, depression, illegal drug use and agitation/restlessness. Although insight has been consistently associated with positive outcomes in psychosis, concerns have been voiced regarding the potential link between insight and increased suicidality. Thus, this thesis commences with a literature review on this topic, which reported mixed results and methodological limitations in previous studies, which were addressed in this investigation. Real-world data on suicides by patients with schizophrenia from the South London and Maudsley NHS Foundation Trust are also presented, which supports further the necessity for this work. Data from three large cohorts of first-episode psychosis (FEP) patients from the UK (n=112-181) and Spain (n=397) were analysed to ascertain the role of multiple insight dimensions (illness recognition, symptoms relabelling, awareness of the social consequences and awareness of the need for treatment) in risk for suicidal behaviour. Although bivariate analyses showed significant relationships between insight levels and risk of suicidal behaviour, only previous suicide attempts and depression, both of which were linked with insight, survived the multivariate analyses, hence emerging as the main predictors of suicidal behaviour. In other words, suicidal history and depression appear to explain the apparent association of insight with suicide risk in psychosis. Hence, no evidence was found supporting a direct relationship between insight and suicidal behaviour in early psychosis despite common assertions to the contrary, which has implications on clinical practice and future research. Insight, which is linked with better outcomes, does not increase suicidality. Guidelines should therefore prioritise improving insight interventions (e.g. talking therapies or antipsychotics) from first presentation with psychosis and future studies may examine whether this reduces suicide rates.