2022-04-26

055_ Contributions of Polygenic Risk Scores for Genomic Common EF- and P-Factors to Cognitive Performance

Research Question and Aims

Executive Functions (EFs) are meta-cognitive functions that control and coordinate mental processes (e.g., PMID: 34408280). Albeit lacking a concise definition, EFs include an array of meta-cognitive processes such as inhibition, set-shifting, and working memory. During the past decades, evidence has accumulated that these higher-level cognitive functions are best assessed on a latent level, since any single EF test also captures non-executive processes (PMID: 27251123). It is well known that latent EFs factors are related to but distinct from general intelligence, highly polygenic, and, in functional neuroimaging studies, associated with activation of the frontal lobes. In an earlier twin-study, a latent "common EF" (cEF) factor was found to be highly heritable (96%, PMID: 27251123), making it an attractive target for further molecular studies. A recent large-scale UK Biobank GWAS (currently available as a preprint) has recently examined the genomics of this common EF factor in over 427,000 individuals (Hatoum et al, 2020, doi.org/10.1101/674515), and has identified 112 distinct genomic loci associated with it. Furthermore, based on analyses using LDSC, the cEF factor is genetically correlated to many psychiatric disorders (e.g., about -0.3 to -0.4 with schizophrenia, bipolar disorder, and major depression). Paralleling this line of research, it has also been shown that different psychiatric disorders share a common phenotypic, neural, and genomic basis (e.g., PMID: 33526822). GWAS summary statistics of this latent "p-factor" were calculated with GenomicSEM (PMID: 30962613) and are readily available. Moreover, in a genomic structural equation model, the p- and cEF factors are related (Hatoum et al, 2020, doi.org/10.1101/674515, -0.50).
This proposal seeks to assess influences of polygenic risk scores (PRS) for both the cEF as well as the p-factor on all (phenotypic) cognitive variables collected in PsyCourse.

Analytic Plan

1. We hypothesize that PRS for cEF and the p-factor are significantly correlated. To systematically assess this, we will calculate PRS at different p-value thresholds.

2. We hypothesize that PRS for cEF and the p-factor explain variability in at least some neuropsychological test results (both PRS assessed separately).

3. We aim to calculate a phenotypic common EF factor score for the PsyCourse participants, by performing a confirmatory factor analysis, and assess the relationship of PRS for cEF and the p-factor with phenotypic factor scores for the cEF factor. The following neuropsychological tests from Visit 1 will be used to calculate the phenotypic factor score, and were selected to ensure maximum overlap with the test used by Hatoum et al. (2020): TMT-A [reaction time], TMT-B [reaction time], Digit-Span Forward, Digit-Span Backward, and the Digit-Symbol-Test.

PRS will be calculated with both PRS-CS, and the clumping-and-thresholding approach, using GSA data.

The analyses will include covariates, such as age, sex, study center, diagnosis, severity of current psychiatric symptoms, cumulative medication, and the first four principal components of an ancestry PCA (or MDS).

Secondary analyses will be performed separately for cases and controls, including broad diagnostic groups.

Resources needed

Phenotype data (PsyCourse 5.0):
v1_id
v1_stat
v1_center

v1_Antidepressants
v1_Antipsychotics
v1_Mood_stabilizers
v1_Tranquilizers
v1_Other_psychiatric

gsa_id
v1_scid_dsm_dx
v1_sex
v1_age
v1_ed_status
v1_curr_paid_empl
v1_panss_sum_pos
v1_panss_sum_neg
v1_panss_sum_tot
v1_idsc_sum
v1_cgi_s
v1_gaf
v1_nrpsy_mtv
v1_nrpsy_tmt_A_rt
v1_nrpsy_tmt_A_err
v1_nrpsy_tmt_B_rt
v1_nrpsy_tmt_B_err
v1_nrpsy_dgt_sp_frw
v1_nrpsy_dgt_sp_bck
v1_nrpsy_dg_sym
v1_nrypsy_mwtb
v2_nrpsy_vlmt_check
v2_nrpsy_vlmt_corr
v2_nrpsy_vlmt_lss_d
v2_nrpsy_vlmt_lss_t

Genotype data:
Imputed and best guess genotypes (already available from other proposals)