39. POLYGENIC SCORES FOR PSYCHIATRIC DISORDERS PREDICT QUANTITATIVE SYMPTOM SCORES IN THE POPULATION LARGELY DUE TO THE PERVASIVE GENETIC INFLUENCE OF GENERAL PSYCHOPATHOLOGY (P)

• Published in: European neuropsychopharmacology 2024-10, Vol.87, p.69-70
• Main Authors: Keser, Engin, Liao, Wangjingyi, Allegrini, Andrea, Eley, Thalia, Rimfeld, Kaili, Malanchini, Margherita, Plomin, Robert
• Format: Article
• Language: English
• ISSN: 0924-977X
• DOI: 10.1016/j.euroneuro.2024.08.153

Although polygenic scores (PGS) can be used to predict genetic liability to psychiatric problems, they often show transdiagnostic associations with multiple psychopathology traits, suggesting that these scores index genetic liability to general psychopathology in addition to trait-specific liability. Therefore, an important direction for genomic research in psychiatric disorders is to move beyond these PGS to PGS that predict specific psychopathology in the sense that they are corrected for these pervasive transdiagnostic effects. Accounting for transdiagnostic effects provides a way to investigate specificity in psychiatric disorders and obtain a clearer picture of the associations between different psychopathology traits. We investigated the extent to which PGS that are uncorrected and corrected for the transdiagnostic effects across 11 major psychiatric disorders and a polygenic score for general psychopathology (p) predict diverse psychopathology symptoms and diagnoses of common mental health problems. We used Genomic Structural Equation Modelling to isolate transdiagnostic effects across 11 major psychiatric disorders from genetic effects specific to each psychiatric condition. We then used summary statistics from these GWA analyses of the residual variances to construct PGSs that are corrected for the transdiagnostic effects. In a representative UK sample of young adults from the Twins Early Development Study (N up to 6,000, average age 25), we regressed quantitative symptom scores relevant to eight of the psychiatric disorders (ADHD, Alcohol Use, ASD, GAD, MDD, bipolar disorder, PTSD and psychosis) on uncorrected and corrected PGS, and on a polygenic score for the general factor of psychopathology (p). We also analysed extremes of the symptom scores as well as diagnoses. Finally, we assessed the extent to which these PGS predict mental-health related traits that often co-occur with psychiatric disorders including emotional and behavioural difficulties, self-harm, low subjective well-being, callous-unemotional traits, and drug use. The average correlation between the PGS and their target symptom scores was 0.10 for uncorrected PGS and 0.02 for PGS corrected for p, indicating that most of the modest PGS prediction of symptom scores is due to p. Moreover, a single polygenic score for p predicted the target symptom scores significantly better than the target PGS for 7 of 12 comparisons. We found similar results for the high extremes of the symptom score