Pathomechanisms and Signatures in the Longitudinal Course of Psychosis



017_ A Genome-Wide Association Study of the Longitudinal Course of Executive Functions

Research Question and Aims

The term Executive functions (EFs) denotes higher cognition that involves active control and coordination of mental processes. EFs are impaired in severe mental disorders, and the course of this impairment is correlated with changes in the frontal cortex (Heilbronner et al., 2016). In this project, which began in 2018, we research the molecular genetic basis of the longitudinal course of EFs. EFs can be broadly separated into the correlated subfunctions of Set-Shifting, Updating, and Inhibition (Miyake et al., 2000). Two of these core aspects of EFs are assessed in the PsyCourse Study and will be studied: Set-Shifting by the Trail-Making-Test, Part B (TMT-B) and Updating by the Verbal Digit Span Backwards (VDS).

Analytic Plan

We will continue to perform two separate GWAS, assessing the course of the two core executive functions mentioned above, using the latest imputed PsyCourse genotype dataset. To fully take advantage of the PsyCourse sample, we included all genotyped patients and healthy control individuals in which the phenotypes mentioned above were assessed at least once. To study whether a SNP is interacting with time course, we have performed linear mixed model (LMM) analyses with outcomes VDS or log(TMT-B; transformed to normal distribution). We modeled subject-specific time courses allowing for random intercepts and slopes. We also included SNP, time, SNP x time, age, sex, DSM-IV diagnoses and the top five ancestry principal components as fixed effects and recruiting center as additional random effect in each LMM. Our focus of interest is the SNP*time interaction term, i.e. the slope. This main model is analyzed with best guess, dosage data and genotype probabilities for imputed genotypes in order to check robustness of the results. We will also check whether for some SNPs quadratic time profiles need to be allowed for. An efficient analysis strategy using different regression models will be used to assess the model for all SNPs above significance as well as borderline significant SNPs close to interesting genes. If possible, we would like to replicate our results using a different sample (e.g. the FOR2017 cohort), requesting a very limited number of SNPs from collaborators. If possible, we might also want to validate genotyping of a positive significant finding. Based on the same data we would also like to carry out a nonparametric longitudinal analysis as mentioned in the original KFO WP. Due to time issues, this will be outside of the current approach/time frame for reaching publication, and possibly carried out by another statistician of the Institute of Genetic Epidemiology.

Resources needed

Phenotype Data:
We have used the following variables from the PsyCourse3.0 dataset:

Recruitment data:
Participant identity column
Clinical/Control Status
Data of interview
Recruitment center

Demographic information:
Age at first interview
Marital status
Relationship status
Living alone

Country of birth
Country of birth mother
Country of birth father

Psychiatric history:
Current psychiatric treatment
Times treated as day-or inpatient

Clinical participants
Control participants

Family history of psychiatric illness

Substance abuse:
Illicit drugs

DSM-IV Diagnosis/MINI-DIPS for healthy controls

Psychiatric rating scales:
Major depressive episode
Mania and hypomania
Psychotic and associated symptoms
PANSS Positive sum score
PANSS Negative sum score
PANSS Total score
IDS-C30 Total score

Neuropsychology (cognitive tests):
Verbal digit span
GWAS analysis ID

Genotype data:
We have access to the latest version of the imputed PsyCourse GWAS dataset.