Developmental psychopathology, family, and genetic studies showed shared liability to schizophrenia (SCZ) and bipolar disorder (BP), as well as a certain degree of specificity that represents the basis for distinct diagnoses. Previous GWAS evidenced a genetic correlation of 0.70 between SCZ and BP based on common single nucleotide polymorphisms (SNPs) (Cross-Disorder Group of PGC, 2019). An analysis of the most recent GWAS of 11 major psychiatric disorders (Grotzinger et al., 2022) identified four genomic factors summarizing their overlapping genetic architecture. One factor was the “psychotic factor” clustering common SNPs for BP and SCZ.

Although many SNPs contribute to both disorders, SNP subsets have a larger effect in SCZ than in BP and vice-versa (Ruderfer et al., 2018), which may contribute to clinical specificity. Associations were also described between SCZ-SNP-sets derived from the SCZ-GWAS-2014 (Ripke et al., 2014) and some phenotypic traits of BP such as age of onset (AO) (Ruderfer et al., 2018; Kalman et al., 2021), psychosis (Leonenko et al., 2018; Calafato et al., 2018; Stahl et al., 2019), although these associations were not always replicated (Kalman et al., 2019). Stahl et al. (2019) showed that the SCZ polygenic risk score (PRS) loading is higher in BP-type I (BP-I) than in BP-II and Allardyce et al. (2018) found that the SCZ-PRS (2014 set) loading was higher in BP-I with psychosis than in BP-I without psychosis. Treatment response in BP was also shown to be influenced by the SCZ-PRS (2014 set) (International Consortium on Lithium Genetics, 2018; Schubert et al., 2021).

A problem with large-scale GWAS samples is their phenotypic heterogeneity generated by the population heterogeneity, the different diagnostic criteria and recruitment settings, and inclusion of several BP-types, usually BP-I, BP-II, schizoaffective patients (e.g. Kalman et al., 2019; Stahl et al., 2019).

The objective of our study was to investigate the predictive value of PRS derived from the PGC Schizophrenia GWAS 2022 (Trubetskoy et al., 2022) (SCZ3) for phenotypic traits of BP-I in two phenotypically homogeneous and well characterized samples: a Romanian (RO) sample and an UK sample.