2026-01-23

110_ Pathway-Level Genetic Contributions to Executive Functioning in the PsyCourse Study

Research Question and Aims

Executive function (EF) impairments are commonly observed across a wide range of psychiatric disorders. Multiple studies suggest that individual differences in EF, and particularly in the latent Common Executive Function (cEF) factor, are heritable and share genetic influences with psychiatric phenotypes. The primary aim of this project is to investigate whether genetic variation in biologically defined pathways is associated with stable inter-individual differences in cEF factor measured across repeated assessments. Using longitudinal linear mixed models in combination with kernel-based variance component testing, we will assess whether genetic similarity within candidate pathways contributes to consistent differences in cEF across time, while accounting for practice effects and within-person variability. This approach builds on prior large-scale work (Hatoum et al., 2018) identifying genetic correlates of cEF in cross-sectional samples and extends it to a longitudinal framework that improves the precision and reliability of the cognitive phenotype.
As a complementary objective, we will conduct a multivariate analysis in which individual baseline EF test scores are modeled as outcomes without aggregation into a single cEF factor. This multivariate approach allows us to examine whether genetic variation in the same pathways explains shared covariance across multiple EF domains at baseline, providing an alternative and complementary perspective on executive functioning. Together, these analyses address the question of whether pathway-level genetic variation contributes to stable differences in EF in a longitudinal psychiatric cohort.
The pathway analyses will focus on those pathways implicated in executive functioning and cognitive control, including but not limited to those previously reported by Hatoum et al. (2023): synaptic signaling, neurotransmitter systems, neuronal development, and calcium-dependent signaling processes. Pathways will be selected from Reactome database, with size thresholds applied to ensure stable kernel estimation and interpretability.

Analytic Plan

This project will combine longitudinal mixed-effects modeling and multivariate baseline analyses to investigate the contribution of pathway-level genetic variation to EF. Analyses will focus on two complementary phenotypic representations of executive function: (i) a longitudinal Common Executive Function (cEF) factor derived from repeated EF task performance, and (ii) baseline performance across multiple EF tasks modeled jointly in a multivariate framework.
The primary longitudinal phenotype will be the Common Executive Function (cEF) factor, which captures variance shared across multiple executive function tasks and reflects stable individual differences in executive control. Repeated measurements of cEF across study visits will be used to model within-person change and between-person differences over time. The longitudinal factor analysis used in Navarro et al. (submitted) will be used for the analysis of the latent cEF. As a secondary analysis, executive function will be examined at baseline using two complementary approaches: a cross-sectional Common Executive Function (cEF) factor, and a multivariate outcome of individual EF task scores. This allows for a comparison between latent-construct and multivariate representations of EF, while allowing for the investigation of shared genetic influences across distinct EF domains. The feasibility of extending the multivariate approach to a longitudinal framework will be explored.
Longitudinal analyses will be conducted using linear mixed models (LMMs), with repeated cEF measurements within individuals. Time (Visit) will be included as a fixed effect to account for systematic changes over repeated testing, such as practice effects, and random intercepts and random slopes for time will be included to capture individual-specific deviations. Genetic effects will be modeled using a kernel-based variance component term h(Gi), representing genetic similarity across variants within predefined biological pathways. This approach allows for the joint modeling of potentially many variants with small effects, without assuming a specific direction or linearity of individual SNP effects.
The primary hypothesis of the longitudinal analysis is that genetic variation within specific pathways explains stable inter-individual differences in cEF across repeated assessments, above and beyond within-person variability over time. Evidence for a pathway effect is obtained by testing whether the variance component associated with h(Gi) is significantly greater than zero.
To complement the longitudinal cEF analysis, a multivariate model will be fitted in which individual EF task scores are treated as a vector-valued outcome for each participant. This analysis will be restricted to baseline measurements to avoid complications arising from repeated testing and learning effects. Genetic pathway effects will again be modeled using kernel-based variance component testing, assessing whether genetic similarity within pathways explains covariance across multiple EF domains. This multivariate approach provides insight into whether pathway-level genetic effects are shared across executive function components or are more domain-specific.
Genetic data have undergone standard quality control procedures. To account for population stratification, the first 10 ancestry principal components derived from genome-wide genotype data will be included as covariates in all analyses. Additional covariates, such as age and sex and testing center, will be included where appropriate to control for potential confounding effects. Corrections for multiple testing will be performed. Statistical power is expected to be modest, particularly for small pathway effects, and results will be interpreted with appropriate caution. Findings that survive multiple testing correction will be considered candidates for further investigation and replication in independent samples. For replication study data from the research consortium FOR2107 will be used.

Resources needed

v1_id
v1_stat
v1_center
v1_tstlt
v1_sex
v1_yob
v1_scid_dsm_dx
v1_scid_dsm_dx_cat
psyc_id
gsa_id
visit
nrpsy_dg_sym
nrpsy_dgt_sp_bck
nrpsy_dgt_sp_frw
nrpsy_tmt_A_err
nrpsy_tmt_A_rt
nrpsy_tmt_B_err
nrpsy_tmt_B_rt