2026-05-12

113_ Association of polygenic scores for substance use disorders and cognitive performance in patients with major psychiatric

Research Question and Aims

Cognition involves mental processes that transform the information received by perception into a response, whether externalized or internalized. Cognitive domains among others include intelligence, executive functioning (EF), and learning and memory (PMID: 25266297). Genetic factors influence cognitive performance and while twin-based heritability of general cognition is estimated at >50%, genome-wide association studies (GWASs) reported associated loci, including those intersecting with neurodegenerative and neurodevelopmental disorders, physical and psychiatric illnesses, and brain structure (PMID: 29844566). This could partially explain the close relationship between cognitive decline and psychopathology, given that various psychiatric disorders are accompanied by cognitive impairment of different extent. Understanding which cognitive domains are affected by genetic risk factors in these patients, can be applicable for both diagnostics and therapeutic purposes.
Substance use disorders (SUDs), including drug dependence (DD), alcohol dependence (AD), and tobacco use disorder, produce a high burden of disease worldwide (PMID: 30392731). In Germany, the most common substance addiction relates to tobacco (8.3%), alcohol (4.2%), analgetic (2.8%), and cannabis (1%), indicating a relative high risk for SUDs in our study population (PMID: 40956679). With the US Centers for Disease Control and Prevention WONDER report a rapid increase in drug-related deaths, especially for opioids, understanding the consequences of this risk are of relevance. Genetic factors contribute to a higher risk of developing a SUD. Twin studies suggest quite high heritability estimates for SUDs range between 30% and 80% (PMID: 38828723). Using large GWAS studies, polygenic scores had been calculated for substance use disorders for: Problematic Tobacco Use (PMID: 38632388), Problematic Alcohol Use (PMID: 32451486), Cannabis Use Disorder (PMID: 37985822), Opioid Use Disorder (PMID: 35879402). Recent multi-phenotype GWASs (MP-GWAS) of SUDs have described a general addiction-risk-factor (PMID: 34750568). Genetic liability for SUDs have been related to cognitive functioning. For example, general cognitive ability, EF, and learning/memory in children unexposed to substances were associated with PGS for lifetime cannabis use (PMC9331817), while PGS for SUDs were associated with educational attainment (PMC12980298). The genetic overlap might be even stronger for some traits, as most of the variants related to the risk of Opioid use disorder (OUD) were associated with general cognitive ability in people of European ancestry (PMID: 42016913).
Nevertheless, a detailed description of the cognitive domains related to genetic risk of SUD remains underexplored. In this study we aim to describe the association between PRS for SUD and cognitive performance in three domains (crystallized intelligence, verbal learning and memory, and the common executive function factor) in patients with psychiatric disorders, as well as healthy controls.

Analytic Plan

1. Differences between the five PGS-SUD will be tested between the three diagnostic groups (Controls, Psychotic and Affective) using Anova and the TukeyHSD test.
2. Z-scores for three cognitive domains will be calculated: MWTB - crystallized intelligence, VLMT - verbal learning and memory, and the common Executive Function factor (cEF) - executive functions.
3. PGS for Problematic Alcohol Use (PAU), Problematic Tobacco Use (PTU), Cannabis Use Disorders (CUD), Opioid Use Disorder (OUD), and the Addition Risk Factor (ARF) will be calculated from the latest GWAS using PRS-CS (PMID: 38632388, PMID: 32451486, PMID: 37985822, PMID: 35879402, PMID: 34750568).
4. Cross-sectional analysis will be conducted using multivariate regressions of the first visit (cEF, MWTB), and second visit (VLMT), adjusting by age, sex, the first 4 genetic-ancestry PCs, and diagnosis and each PGS separately.
5. Longitudinal analysis for the scores of cEF and VLMT will be conducted using linear mixed models (LMM), using fixed effects for age, sex, the first four genetic-ancestries PCs, PGS, visit, and the interaction of PGSxVisit. Random intercepts will be included for center nested within patient, and diagnostic group.
6. P-values will be corrected using the FDR method, and standardized Betas per SD will be calculated as effect sizes.
7. Sensitivity analysis including the lifetime prevalence of substance use disorders, medication use, educational attainment, and potentially other confounders.

Resources needed

v1_id
v1_stat
v1_center
v1_sex
v1_age_m_birth
v1_age_f_birth
v1_ed_status
v1_ever_smkd
v1_age_smk
v1_alc_pst12_mths
v1_lftm_alc_dep
v1_evr_ill_drg
v1_sti_cat_evr
v1_can_cat_evr
v1_opi_cat_evr
v1_kok_cat_evr
v1_hal_cat_evr
v1_inh_cat_evr
v1_tra_cat_evr
v1_var_cat_evr
v1_nrpsy_mwtb
Antidepressants
Antipsychotics
nrpsy_tmt_A_rt
nrpsy_tmt_B_rt
nrpsy_vlmt_corr
nrpsy_vlmt_rec