Pathomechanisms and Signatures in the Longitudinal Course of Psychosis



071_ The impact of polygenic risk score and early trauma on functioning and illness severity in bipolar disorder

Research Question and Aims

Both genetic factors[1] and early life stress (ELS)[2,3] can negatively affect illness outcomes in individuals with bipolar disorder (BD), including cognitive, functional, and clinical outcomes (i.e., suicide). We aim to examine if ELS mediates the interaction between PRS for major psychiatric conditions (BD, schizophrenia, major depression) and BD outcomes. Furthermore, we aim to integrate ELS, PRSs, and clinical variables into a machine-learning algorithm able to dissect BD phenotypes (BD with poor cognitive and global functioning and more illness severity vs BD with good cognitive and global functioning and more illness severity) and predict BD.

1. PMID: 27743041
2. PMID: 33315268
3. PMID: 27500795

Analytic Plan

General Hypotheses:
Individual genetic liability for specific psychiatric traits and early life trauma, influence BD course and outcomes. The integration between genetic and clinical factors would be able to discriminate different phenotypes of BD and predict BD diagnosis.

Concrete Hypotheses:
1. Subjects with bipolar disorder (BD) will have higher polygenic risk scores (PRS) for major psychiatric conditions (BD, schizophrenia, major depressive disorder), higher early life stress (ELS), and worse cognitive and functioning outcomes and illness severity compared to healthy controls (HC)
2. ELS mediates the relationship between BD-PRS, SCZ-PRS, MDD-PRS and cognitive outcomes, functioning, illness severity and severe outcomes (i.e., suicide) in patients with BD.
3. Integrating genetic (PRS) with socio-demographical and clinical factors will significantly increase the accuracy of prediction of BD outcomes (in the total population of BD+HC) and BD risk (among individuals with BD).

For this purpose, we would use cross-sectional data derived from two different cohorts:
1. The Bipogent Cohort: composed of outpatients of the Bipolar and Depressive Disorders Unit of the Hospital Clinic of Barcelona, diagnosed with BDI or BDII, recruited from February 2017 to July 2019. In the cohort, we have a total of 150 individuals phenotyped and genotyped (78 patients and 72 matched healthy controls).
2. The PsyCourse Cohort: 560 participants of the PsyCourse study with bipolar-spectrum diagnoses (we will use only data of genotyped patients with BD or healthy controls)

Phenotype definition:
Baseline evaluations from the Bipogent and PsyCouse cohorts will be considered:
- Socio-demographics
- Psychiatric and medical history
- Medication data and treatment response
- Family history of psychiatric disease
- Clinical evaluations: mood assessment (HDRS, IDS, YMRS), functioning (FAST, GAF), early life stress (CTQ, CTS), illness severity (CGI), suicide (attempt, behavior)
- Neurocognitive assessment: Processing Speed (TMT-A total time); attention and working memory (WAIS-III digit span); executive functions (TMT-B); verbal memory (CVLT/VLMT), IQ (WAIS-III, MWT-B).

Polygenic Risk Scores:
Polygenic risk scores (PRS) for major psychiatric disorders (BD, schizophrenia, major depressive disorder) will be constructed by using the latest genome-wide association studies (GWAS) on the topics. PRS will be constructed using PLINK 1.9 independently in the two cohorts, that data will be meta-analyzed. The polygenic scores will be calculated based on summary statistics from the discovery dataset excluding rare SNPs (MAF < 0.5%), low-quality imputed variants (info score <90%), indels, and ambiguous markers (A/T and C/G). Data will be clumped in windows of 1000 kbp, discarding variants in LD (R2>.2), index variant p value threshold of 0.5 with another more significant marker. Scores will be calculated based on p-value thresholds ranging from p < 5 x 10-8 to p < 1.

Statistical Analyses:
First, several multivariate linear regressions will be used to examine the relationship between BD-PRS, MD-PRS, SCZ-PRS, ELS as main determinants and BD cognitive outcomes, functioning, illness severity and suicide attempts, as outcomes. For mediation analyses, the variance/covariance matrix will be obtained, calculating the correlation among each variable. Then parallel multiple mediation analysis will be performed using maximum likelihood estimation (MLE) path analysis to assess the effect of PRSs on clinical outcomes directly and indirectly through the postulated mediator (ELS). Statistical analyses will be performed with R, using the packages “glmnet” for the LASSO, “lavaan", and “tidySEM" for the mediation/moderation analysis. Random Forest Models will be trained integrating genetic and clinical variables to identify the presence of BD as the predicted condition in the entire sample. Next among people with BD, a second and a third Random Forest Models will be used to predict the presence of cognitive deficits and functional impairment. RF is a classification algorithm that combines multiple decision trees made by randomly selected bootstrap samples, mainly affected by unbalanced data.

Resources needed