2026-07-23

115_ Exploring cognitive performance in major psychiatric disorders and its genetic liability (Amendment to 009)

Research Question and Aims

Human cognitive performance can be deeply impacted by major psychiatric disorders like schizophrenia (SZ), schizoaffective disorder (SZA), bipolar disorder (BD) and major depressive disorder (MDD). These impairments can affect individuals during both acute and remission phases of the disorder. (1) Such conditions have been demonstrated to have intersecting symptomatology and genetics, making the hypothesis of a spectrum between these disorders realistic (2). It has been shown that individuals with either SZ, SAZ, or BD often have increased cognitive impairment, with BD patients performing better than SZ patients (3)(4). For example, Lynham et al. (5) tested this hypothesis of increased cognitive impairment from BD (n=78) to SZA bipolar type (n=76), to SZ (n=558) and SZA depressive type (n=112) (controls: n=103). However, Lynham et al., considered neither MDD, a disorder on the affective end of the affective-to-psychotic spectrum, nor genetic aspects. The continuum between such disorders is further supported by genetics, showing strong correlations between MDD polygenic risk scores (PRS) and BD II, and between SZ PRS and BD I (6).
Despite several studies attempting to link the PRS of SZ and BD with cognitive performance (7-9), this possible continuum has been poorly investigated, particularly yielding heterogeneous results.
The main aim of this project is trying to establish a link between the diagnosis of major psychiatric disorders and cognitive impairment in the trans diagnostic PsyCourse (10), possibly reaffirming a continuum between such conditions. Moreover, to further deepen the knowledge of genetic liability and its effect on severe mental disorders, we aim to better characterize the possible relationship between PRS for major psychiatric disorders (SZ, BP I, BP II, MDD) and a latent (“g”-like) dimension of cognitive performance.
This project…
- …is a partial replication of Lynham's approach (at least 4 out of 7 cognitive domains)
- …differentiates, unlike Lynham's study, between BD I and BD II
- …includes MDD as further disorder
- …presents the cognitive profile of the PsyCourse cohort at visit 1 (data for Visit 2 will be used for the Verbal Learning and Memory Test)
- …considers a possible association of PRS for SZ, BD, and MDD and a latent (“g”-like) dimension of cognitive performance.

Analytic Plan

Hypotheses:
H1: Cognitive performance declines on a spectrum from neurotypical over affective to psychotic disorders (specifically healthy controls > MDD > BP-II > BP-I > SZA > SZ).
H2: Each PRS for major psychiatric disorders (SZ, BP- I, BP- II, MDD) is associated with the latent “g”-like dimension of cognitive performance.

Phenotypes:
TMT- A, TMT-/B, digit symbol test (V1), verbal digit span forward and backwards (V1), and the verbal learning and memory test (V2) and MWT-B (V1, crystallized intelligence).

Genomic data:
-SZ-, BD-, MDD-PRS and ancestry principal components

Analytic methods
All cognitive test results will be z-standardized to allow a comparison between them. For all tests, sensitivity analyses may be carried out for variables such as medication and center.

1. Comparing cognition between diagnostic groups
1.1 ANCOVA for each cognitive test followed by Tukey’s post-hoc test: independent variable: DSM-Dx, dependent variable: cognitive tests, covariates: age, sex, center, inpatient status.
1.2 Ordinal linear Model for each cognitive test: Diagnostic group (criterion, y~), cognitive performance and covariates (predictors Ð).
Approaches 1.1 and 1.2 are a direct replication of Lynham et al. If feasible, we will additionally use an ordinal linear mixed model (LMM) (1.3), to research if the modeling of hierarchical structures in the data (by inclusion of the random variables id and center) will improve results.

2. Examining cognition as latent a “g”-like dimension across diagnostic groups
To derive the latent dimension, a confirmatory factor analysis with one “g”-like factor will be carried out, and factor scores will be calculated for PsyCourse participants. These factor scores will then be used in ordinal linear models like the analyses described in 1.2 and 1.3.

3. Association of the latent “g”-like dimension with PRS for major psychiatric disorders
In these analyses, for each PRS, the general approach is to compare two regression models:

1. A linear model (or LMM, depending on whether this approach was successful in previous analyses) with the latent “g”-like dimension as the criterion and the predictors as outlined above, additionally including covariates of an ancestry PCA.

2. The same model as in 1., additionally including PRS for major psychiatric disorders.

If the model including the PRS fit improves the adjuster R-squared, this will be interpreted as evidence for the importance of PRS.

Resources needed

v1_id
v1_stat
v1_center
v1_interv_date
v1_sex
v1_age
v1_marital_stat
v1_partner
v1_twin_slf
v1_school
v1_prof_dgr
v1_ed_status
v1_curr_paid_empl
v1_disabl_pens
v1_cur_work_restr
v1_cur_psy_trm
v1_age_1st_out_trm
v1_age_1st_inpat_trm
v1_dur_illness
v1_1st_ep
v1_Antidepressants
v1_Antipsychotics
v1_Mood_stabilizers
v1_Tranquilizers
v1_Other_psychiatric
v1_fam_hist
v1_lftm_alc_dep
v1_scid_dsm_dx_cat
v1_scid_age_MDE
v1_scid_no_MDE
v1_scid_ever_psyc
v1_panss_sum_pos
v1_panss_sum_neg
v1_panss_sum_gen
v1_panss_sum_tot
v1_idsc_sum
v1_ymrs_sum
v1_gaf
v1_nrpsy_lng
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_nrpsy_mwtb
v2_clin_ill_ep_snc_lst
v2_cur_psy_trm
v2_curr_paid_empl
v2_disabl_pens
v2_cur_work_restr
v2_Antidepressants
v2_Antipsychotics
v2_Mood_stabilizers
v2_Tranquilizers
v2_Other_psychiatric
v2_panss_sum_pos
v2_panss_sum_neg
v2_panss_sum_gen
v2_panss_sum_tot
v2_idsc_sum
v2_ymrs_sum
v2_gaf
v2_nrpsy_lng
v2_nrpsy_mtv
v2_nrpsy_vlmt_check
v2_nrpsy_vlmt_corr
v2_nrpsy_vlmt_lss_d
v2_nrpsy_vlmt_lss_t
v2_nrpsy_vlmt_rec
v4_opcrit
gsa_id