The COVID-19 pandemic, caused by SARS-CoV-2, highlighted the wide variability in individual responses to infection, ranging from asymptomatic cases to severe disease and death. Identifying host factors contributing to this variability is crucial. Previous research indicated a potential link between psychiatric disorders and worse COVID-19 outcomes. Patients with schizophrenia (SCZ) have shown evidence of more severe COVID-19 compared to controls. This study aimed to explore the genetic overlap between psychiatric disorders and COVID-19 by constructing polygenic risk scores (PRSs) for SCZ, BD, and DEP and assessing their predictive value for COVID-19 susceptibility and severity. Given the reported immunological dysfunctions and shared pathophysiological mechanisms (such as chronic systemic inflammation and neuroinflammation) between psychiatric disorders and COVID-19, the researchers hypothesized that these PRSs would predict SARS-CoV-2 infection and COVID-19 severity, symptomatology, and the need for hospitalization.

The literature review highlights the existing evidence linking certain comorbidities (cardiovascular disease, respiratory diseases, metabolic disorders, etc.) to worse COVID-19 outcomes, many of which are frequently observed in psychiatric patients. Studies reported increased COVID-19 prevalence and severity among SCZ patients, especially in older individuals. The review discusses the immunological dysfunctions often reported in psychiatric disorders (SCZ, MDD, BD), such as persistent inflammation and impaired adaptive immunity, which may increase vulnerability to COVID-19. Furthermore, the literature points to shared pathophysiological mechanisms between these disorders and COVID-19, including chronic systemic inflammation, neuroinflammation, microbiome disruptions, and oxidative stress. While some medications used in treating psychiatric disorders have shown potential in preventing SARS-CoV-2 neurotropic adverse events, the extent to which shared genetic risk factors contribute to the observed clinical parallels remained unclear. The review mentions previous genetic studies implicating specific genes (ACE2, TMPRSS2, genes in the 3p21.31 region, ABO blood type locus, and MHC class I genes) in COVID-19 susceptibility and severity. Previous research also highlighted the association of B-lymphocyte activity and MHC loci with SCZ and BD and the involvement of genetic variants in inflammatory pathways across DEP, BD, and SCZ. Analyses of GWASs and WGS data revealed shared genes and biological pathways between COVID-19 and psychiatric disorders, suggesting a potential genetic overlap.

The study utilized a large COVID-19 cohort (11,977 cases) from 34 Spanish hospitals, along with a control group (5943 individuals with unknown COVID-19 status). Genotype data were obtained, quality control procedures were applied, and imputation was performed using the TOPMed version r2 reference panel. Polygenic risk scores (PRSs) were constructed for SCZ, BD, and DEP using summary statistics from the Psychiatric Genomics Consortium (PGC). The PRS-CS software was employed, using a Bayesian framework with continuous shrinkage priors to infer posterior effect sizes. Logistic regression models were used to analyze the PRSs' ability to predict SARS-CoV-2 infection (RISK), symptom presence (SYMP), hospitalization (HOSP), and critical status (CRIT), adjusting for sex, age, and principal components. The area under the receiver operating characteristic (ROC) curve (AUC) was calculated to evaluate predictive performance. An immune PRS was also constructed from variants in immune-related genes identified using MAGMA, focusing on SNPs from genes involved in KEGG pathways. Linkage disequilibrium score (LDSC) regression was performed to estimate the genetic correlation between the PRSs and COVID-19 variables. Multiple-comparison correction was applied to account for multiple tests.

The global SCZ PRS showed significant associations with COVID-19 risk (RISK), symptom presence (SYMP), and hospitalization (HOSP) in the total sample, which were maintained in the female subsample. In male subsample only SYMP analysis was significant. The odds ratios (ORs) increased across SCZ PRS deciles for all three variables. However, the AUC values did not significantly change when comparing models with and without the PRS, suggesting limited predictive accuracy beyond chance. No significant associations were found for the immune SCZ PRS or the global BD or DEP PRSs. LDSC regression analysis did not reveal significant genetic correlations between SCZ risk and COVID-19 case/control status or hospitalization status, using either the study's own GWAS summary statistics or those from the COVID-19 HGI GWAS meta-analysis. The findings suggest a potential association between SCZ polygenic risk and COVID-19 outcomes, particularly in women, although the predictive power was limited.

The significant associations found with the global SCZ PRS, but not the immune SCZ PRS, indicate that a broader range of functional categories beyond immune function may contribute to the genetic overlap between SCZ and COVID-19. This aligns with previous research showing shared transcriptomic markers and genes involved in various biological systems. The lack of significant findings in the LDSC regression analysis might be due to the use of posterior effect sizes in PRS estimation, differing from the original ORs used in genetic correlation analyses. Different results between PRS and genetic correlation analyses have been reported previously even for psychiatric disorders. The study's sex-stratified analyses reveal that the associations between SCZ PRS and COVID-19 outcomes were stronger in women, possibly reflecting sex-based differences in genetic architecture and pleiotropy. The lack of significant associations for BD and DEP PRSs may be attributed to differences in heritability and phenotype definition. The limited predictive accuracy of the SCZ PRS highlights the complexity of COVID-19 pathogenesis, which involves both genetic and non-genetic factors.

This study provides evidence for a potential SNP-based genetic overlap between schizophrenia and COVID-19, particularly concerning susceptibility, symptom presence, and hospitalization, especially among women. The limited predictive accuracy emphasizes the complex interplay of genetic and non-genetic factors in COVID-19 pathogenesis. Future research should incorporate sex loci, rare variants, and detailed clinical information, including medication use and psychiatric diagnoses to refine our understanding of this relationship.

The study's limitations include the lack of information on psychotropic medication use, which might influence COVID-19 response, and the absence of detailed psychiatric diagnoses in the COVID-19 cohort. While comorbid medical conditions were recorded, the impact of these on the results is not completely addressed. The study's focus on common variants may neglect the contribution of rare variations. The use of posterior effect sizes in PRS estimation could also explain the discrepancy between PRS and LDSC regression results.