O4.8. CAN YOU SPOT EMOTIONS? FACIAL EMOTION RECOGNITION AND GENETIC RISK FOR PSYCHOSIS

Abstract

Background Facial emotion recognition (FER) is a key component of social cognition which has been found consistently impaired in schizophrenia. Deficits in global facial affect recognition have been also found in First Episode Psychosis (FEP) with the same severity as at further stages, especially for anger recognition. Literature to date has shown intermediate emotion recognition ability in either people with family history for psychotic disorders and unaffected relatives of psychotic patients, in a continuum between patients and healthy controls. Furthermore, Polygenic Risk Score (PRS) for schizophrenia has been found associated with social cognition, especially with facial emotion identification efficiency, suggesting a potential role of emotion recognition in the genetic risk for schizophrenia. This study aims to test whether PRS for schizophrenia (SZ) can predict facial affect recognition performance in a sample of FEP and healthy population controls. Methods Data on FER (Degraded Facial Recognition task – DFAR) have been analysed in a sample of 412 FEP and 805 population controls for which genetic information was available, recruited as part as the EU-GEI study across UK, Netherlands, France, Spain, Italy, and Brazil. A score <41 on the Benton Facial Recognition test was used to exclude cases and controls from the analysis on DFAR. SZ PRS was built using the results from large mega-analyses from Working Groups of the Psychiatric Genomics Consortium. In PRSice, individuals’ number of risk alleles in the target sample was weighted by the log odds ratio from the discovery samples and summed into the PRS at a 0.05 SNPs Pt-threshold (apriori selected). We excluded people of homogeneous African ancestry. Linear regression models were applied to predict DFAR scores using 1) case/control status, and 2) SZ PRS as predictive variables controlling for age, gender and 20 principal components (PCs) for population stratification. Results Linear regression analyses revealed SZ PRS to be negatively associated with DFAR scores, especially the total score (B= -3.9; 95% CI: -6.78 to -1.03; p= 0.008), DFAR FRIGHTENED (B=-3.68; 95% CI: -7.23 to -0.12; p=0.042), and DFAR ANGRY (B=-6.82; 95% CI: -10.9 to -2.75; p=0.001). DFAR performance on frightened faces was also negatively associated with case status (B= -3.31; 95% CI: -7.23 to -0.12; p=0.018). We ran the analyses on the total score and angry faces separately in cases and controls, finding that SZ PRS is a significant predictor of worse scores in controls only (total: B=-4.51; 95% CI: -8.15 to -0.88; p=0.015. Angry: B=-8.16; 95% CI: -13.29 to -3.03; p=0.002). Discussion This study supports the evidence that 1) the higher SZ PRS the worse the global ability to recognise facial emotions; 2) having a psychotic disorder is associated with worse emotion recognition, especially fear facial expressions; 3) higher genetic risk for schizophrenia predicts worse anger recognition in healthy controls.