Individualized prediction of three- and six-year outcomes of psychosis in a longitudinal multicenter study: a machine learning approach.

Schizophrenia and related disorders have heterogeneous outcomes. Individualized prediction of long-term outcomes may be helpful in improving treatment decisions. Utilizing extensive baseline data of 523 patients with a psychotic disorder and variable illness duration, we predicted symptomatic and global outcomes at 3-year and 6-year follow-ups. We classified outcomes as (1) symptomatic: in remission or not in remission, and (2) global outcome, using the Global Assessment of Functioning (GAF) scale, divided into good (GAF ≥ 65) and poor (GAF < 65). Aiming for a robust and interpretable prediction model, we employed a linear support vector machine and recursive feature elimination within a nested cross-validation design to obtain a lean set of predictors. Generalization to out-of-study samples was estimated using leave-one-site-out cross-validation. Prediction accuracies were above chance and ranged from 62.2% to 64.7% (symptomatic outcome), and 63.5-67.6% (global outcome). Leave-one-site-out cross-validation demonstrated the robustness of our models, with a minor drop in predictive accuracies of 2.3% on average. Important predictors included GAF scores, psychotic symptoms, quality of life, antipsychotics use, psychosocial needs, and depressive symptoms. These robust, albeit modestly accurate, long-term prognostic predictions based on lean predictor sets indicate the potential of machine learning models complementing clinical judgment and decision-making. Future model development may benefit from studies scoping patient's and clinicians' needs in prognostication.

The relationship of IQ and emotional processing with insula volume in schizophrenia.

OBJECTIVE: The insula is involved in general and social cognition, in particular emotion regulation. Aim of this study is to investigate whether insula volume is associated with Intelligence Quotient (IQ) and emotional processing in schizophrenia patients versus healthy controls (HC). METHODS: Magnetic resonance imaging (MRI) brain scans, IQ and emotional processing tests (Benton Facial Recognition Test [BFRT], Degraded Facial Affect Recognition Task [DFAR], Emotional Mentalizing Task [EMT]) were administered in 246 subjects (133 schizophrenia patients and 113 controls). First order linear regression analyses were performed with group as independent variable and IQ/emotional processing test scores as dependent variables. Second order stepwise linear regression analyses were performed with IQ/emotional processing test scores as independent variables (as well as intracranial volumes, age, gender and cannabis abuse) and right/left insula volumes as dependent ones. A final mediation analysis (Sobel test) was performed to verify if IQ or emotional processing test scores could explain the eventual differences in insula volumes between the two groups. RESULTS: Schizophrenia patients presented lower insula volumes (left: F = 9.72, p < 0.01; right: F = 10.93, p < 0.01) as compared with healthy controls. Smaller insula volumes in schizophrenia patients are mediated by lower IQ scores (Sobel tests: 3.07, p < 0.01 for right insula; 2.72, p < 0.01 for left insula), but not by impairments in emotion processing. CONCLUSIONS: IQ, but not emotional processing mediates smaller insula volumes in schizophrenia patients.

Associations between olfactory identification and (social) cognitive functioning: A cross-sectional study in schizophrenia patients and healthy controls.

Schizophrenia patients have difficulties identifying odors, possibly a marker of cognitive and social impairment. This study investigated olfactory identification (OI) differences between patients and controls, related to cognitive and social functioning in childhood and adolescence, to present state cognition and to present state social cognition. 132 schizophrenia patients and 128 healthy controls were assessed on OI performance with the Sniffin' Sticks task. Multiple regression analyses were conducted investigating OI in association with cognitive and social functioning measures in childhood/adolescence and in association with IQ, memory, processing speed, attention, executive functioning, face recognition, emotion recognition and theory of mind. Patients had reduced OI performance compared to controls. Also, patients scored worse on childhood/adolescence cognitive and social functioning, on present state cognitive functioning and present state social cognition compared to controls. OI in patients and controls was significantly related to cognitive and social functioning in childhood/adolescence, to present state cognition and to present state social cognition, with worse functioning being associated with worse OI. In this study, findings of worse OI in patients relative to controls were replicated. We also showed associations between OI and cognitive and social functioning which are not specific to schizophrenia.

Biological aspects and candidate biomarkers for psychotic bipolar disorder: A systematic review.

AIM: We carried out a systematic review of the available literature about potential biomarkers of psychotic bipolar disorder (BD-P), a specific subset presenting worse outcome and greater risk of relapse than non-psychotic bipolar disorder (BD-NP). METHODS: We searched the main psychiatric databases (PubMed, ISI Web of Knowledge, PsychInfo). Only original articles with the main topic of BD-P compared to schizophrenia/BD-NP/healthy controls (HC) written in English from 1994 to 2015 were included. RESULTS: BD-P patients presented higher kynurenic acid levels in the cerebrospinal fluid, elevated anti- S accharomyces cerevisiae antibodies levels, and lower serum levels of dehydroepiandrosterone sulfate and progesterone than BD-NP/HC. Event-related potentials abnormalities have been identified in BD-P with respect to BD-NP. BD-P patients also presented bigger ventricles but similar hippocampal volumes compared to BD-NP/HC. Although the results are contrasting, some cognitive deficits seemed to be related to the psychotic dimension of bipolar affective disorder, such as impairment in verbal/logical memory, working memory, verbal and semantic fluency and executive functioning. Finally, polymorphisms of genes, such as NRG1, 5HTTLPR (s), COMT, DAOA and some chromosome regions (16p12 and 13q), were positively associated with BD-P. CONCLUSION: Data about the identification of specific biomarkers for BD-P are promising, but most of them have not yet been replicated. They could lead the clinicians to an early diagnosis and proper treatment, thus ameliorating outcome of BD-P and reducing the biological changes associated with a long duration of illness. Further studies with bigger samples are needed to detect more specific biological markers of the psychotic dimension of bipolar affective disorder.

Structural MRI Differences between Patients with and without First Rank Symptoms: A Delusion?

OBJECTIVE: It has been suggested that specific psychotic symptom clusters may be explained by patterns of biological abnormalities. The presence of first rank symptoms (FRS) has been associated with cognitive abnormalities, e.g., deficits in self-monitoring or in the experience of agency, suggesting that a specific network of neural abnormalities might underlie FRS. Here, we investigate differences in cortical and subcortical brain volume between patients with and without FRS. METHODS: Three independent patient samples (referred to as A, B, and C) with different mean ages and in different illness stages were included, leading to a total of 348 patients within the schizophrenia-spectrum. All underwent magnetic resonance imaging of the brain. In addition, the presence of FRS was established using a diagnostic interview. Patients with (FRS+, A: n = 63, B: n = 129, and C: n = 96) and without FRS (FRS-, A: n = 35, B: n = 17, and C: n = 8) were compared on global and local cortical volumes as well as subcortical volumes, using a whole brain (cerebrum) approach. RESULTS: Nucleus accumbens volume was significantly smaller in FRS+ as compared with FRS- in sample A (p < 0.005). Furthermore, FRS+ showed a smaller volume of the pars-opercularis relative to FRS- in sample B (p < 0.001). No further significant differences were found in cortical and subcortical volumes between FRS+ and FRS- in either one of the three samples after correction for multiple comparison. CONCLUSION: Brain volume differences between patients with and without FRS are, when present, subtle, and not consistent between three independent samples. Brain abnormalities related to FRS may be too subtle to become visible through structural brain imaging.

Emotion processing in schizophrenia is state and trait dependent.

BACKGROUND: Substantial evidence exists about emotion processing (EP) impairments in schizophrenia patients. However, whether these deficits are present primarily during psychosis (i.e., state dependent) or an integral part of the disorder (i.e., trait dependent) remains unclear. METHODS: EP was assessed with the degraded facial affect recognition task in schizophrenia patients (N=521) and healthy controls (N=312) at baseline (T1) and after a three year follow-up (T2). In schizophrenia patients symptomatic remission was assessed with the Positive and Negative Syndrome Scale (PANSS) remission tool. Patients were divided into four groups: remission T1 and remission T2 (RR); remission T1 and non-remission T2 (RN); non-remission T1 and non-remission T2 (NN) and non-remission T1 and remission T2 (NR). Factorial repeated measures ANCOVA was used to compare EP performance over time between groups. Age, gender and general cognition were included as covariates. RESULTS: Schizophrenia patients performed worse than healthy controls on EP at T1 (p=0.001). The patients that were in symptomatic remission at both time points (the RR group) performed worse than the healthy controls at T2 (p<0.001). Significant group×time interactions were found between RR and RN (p=0.001), and between NR and RN (p=0.04), indicating a differential EP performance over time. No group×time interaction was found between NN and NR. CONCLUSION: The results show relatively poor EP performance in schizophrenia patients compared to healthy controls. EP performance in schizophrenia patients was associated with symptomatic remission. The results provide support for the hypothesis that EP deficits in schizophrenia are both state and trait dependent.

[The association between cognitive deficits and different outcomes of schizophrenia]. / Deficyty poznawcze a efekty schizofrenii.

AIM: Schizophrenia is a disorder with different outcomes. Besides the positive and negative symptoms, cognitive impairment is an important core feature of schizophrenia and often pre-dates the disorder. Cognition has consistently been related to outcome in schizophrenia. Given this finding and the fact that diagnosing and treating schizophrenia as early as possible has better outcome chances, the current study investigated the hypothesis that cognitive performance is associated with two seemingly opposite outcomes: clinical remission and forced hospitalization three years after first assessment. METHODS: Subjects in the current study were schizophrenia patients not in an active psychosis during cognitive testing (N = 321). The results of the cognitive tests were used as predictor variables for the status of remission or the occurrence of a forced hospitalization in the three years following the cognitive testing. The cognitive tests included were WAIS-III subtests (Digit symbol, Information, Arithmetic, Block Design), Benton Facial Recognition task, Hinting task and the Rey Auditory Verbal Learning task. Besides these cognitive predictors, several relevant covariates (gender, age, education, number of psychotic episodes, duration of illness and amphetamine, cannabis or cocaine intoxication) were analyzed. Two multinomial logistic regression and analyses were conducted with the cognitive tests as independent variables and remission and forced hospitalization as dependent variables in separate models. RESULTS: The results showed that better performance on the verbal tasks (WAIS-III arithmetic score (b=0.17) and the WAIS-III information score (b=0.22)) and less psychotic episodes (b=-0.64) was associated with remission status. Worse performance on the memory task (b=-0.20) and more psychotic episodes (b=0,85) was related to forced hospitalization. CONCLUSIONS: This three-year longitudinal study showed that higher verbal IQ is a protective factor and poor memory and higher number of psychotic episodes are risk factors of the outcome of schizophrenia. This suggests that future research on prediction tools for the outcome of schizophrenia should include assessment of verbal IQ and verbal memory.