Use of multiple polygenic risk scores for distinguishing schizophrenia-spectrum disorder and affective psychosis categories in a first-episode sample; the EU-GEI study

Victoria Rodriguez, Luis Alameda, Diego Quattrone, Giada Tripoli, Charlotte Gayer-Anderson, Edoardo Spinazzola, Giulia Trotta, Hannah E Jongsma, Simona Stilo, Caterina La Cascia, Laura Ferraro, Daniele La Barbera, Antonio Lasalvia, Sarah Tosato, Ilaria Tarricone, Elena Bonora, Stéphane Jamain, Jean-Paul Selten, Eva Velthorst, Lieuwe de HaanPierre-Michel Llorca, Manuel Arrojo, Julio Bobes, Miguel Bernardo, Celso Arango, James Kirkbride, Peter B Jones, Bart P Rutten, Alexander Richards, Pak C Sham, Michael O'Donovan, Jim Van Os, Craig Morgan, Marta Di Forti, Robin M Murray*, Evangelos Vassos*

*Corresponding author for this work

Research output: Contribution to journalArticleAcademicpeer-review


BACKGROUND: Schizophrenia (SZ), bipolar disorder (BD) and depression (D) run in families. This susceptibility is partly due to hundreds or thousands of common genetic variants, each conferring a fractional risk. The cumulative effects of the associated variants can be summarised as a polygenic risk score (PRS). Using data from the EUropean Network of national schizophrenia networks studying Gene-Environment Interactions (EU-GEI) first episode case-control study, we aimed to test whether PRSs for three major psychiatric disorders (SZ, BD, D) and for intelligent quotient (IQ) as a neurodevelopmental proxy, can discriminate affective psychosis (AP) from schizophrenia-spectrum disorder (SSD).

METHODS: Participants (842 cases, 1284 controls) from 16 European EU-GEI sites were successfully genotyped following standard quality control procedures. The sample was stratified based on genomic ancestry and analyses were done only on the subsample representing the European population (573 cases, 1005 controls). Using PRS for SZ, BD, D, and IQ built from the latest available summary statistics, we performed simple or multinomial logistic regression models adjusted for 10 principal components for the different clinical comparisons.

RESULTS: In case-control comparisons PRS-SZ, PRS-BD and PRS-D distributed differentially across psychotic subcategories. In case-case comparisons, both PRS-SZ [odds ratio (OR) = 0.7, 95% confidence interval (CI) 0.54-0.92] and PRS-D (OR = 1.31, 95% CI 1.06-1.61) differentiated AP from SSD; and within AP categories, only PRS-SZ differentiated BD from psychotic depression (OR = 2.14, 95% CI 1.23-3.74).

CONCLUSIONS: Combining PRS for severe psychiatric disorders in prediction models for psychosis phenotypes can increase discriminative ability and improve our understanding of these phenotypes. Our results point towards the potential usefulness of PRSs in specific populations such as high-risk or early psychosis phases.

Original languageEnglish
Pages (from-to)3396-3405
Number of pages10
JournalPsychological medicine
Issue number8
Early online date25 Jan 2022
Publication statusPublished - Jun 2023


  • Affective psychosis
  • bipolar disorder
  • diagnosis
  • genetics
  • polygenic score
  • psychosis
  • psychotic depression
  • schizophrenia-spectrum disorder


Dive into the research topics of 'Use of multiple polygenic risk scores for distinguishing schizophrenia-spectrum disorder and affective psychosis categories in a first-episode sample; the EU-GEI study'. Together they form a unique fingerprint.

Cite this