SA11DO PRS FOR SQUIZOPHRENIA, BIPOLAR DISORDER AND MAJOR DEPRESSION DISTINGUISH BETWEEN AFFECTIVE-PSYCHOSIS DIAGNOSTIC CATEGORIES? THE EUGEI STUDY

Abstract

Background: It is well-known that Bipolar Disorder (BD) and Major Depression Disorder (MDD) aggregate in families, with heritability rates as high as 80 and 40% respectively. It is believed that there are probably many hundreds or thou- sands of common alleles that influence susceptibility to the most common psychiatric disorders, where each gene con- fers only a fractional risk. A newly developed method called polygenic risk score (PRS) allows us to summarize individual risks for an illness across those associated variants into a quantitative score. However, the variance explained by PRS for identifying case-control status is lower than expected. Apart from the heterogeneity, another major reason is at- tributed to the shared genetic load between disorders. Us- ing data from the EUGEI case-control study of first-episode psychosis (FEP), we test whether PRS for schizophrenia (SZ), BP and MDD enable to distinguish between categorical diag- nosis under the umbrella of affective psychosis beyond the genetic overlap. Methods: DNA from blood tests or saliva sample was ob- tained from most participants at baseline (73.6% of 1130 cases and 78.5% of 1499 controls) among 16 European cities as part of the EUGEI study. PRS for SZ, BD and MDD were built using data from the Psychiatric Genomic Consortium (PGC). DSM-IV diagnoses were extracted from interviews and mental health records utilizing the Operational Criteria Checklist at baseline and were grouped into diagnostic groups (Bipolar Disorder, Psychotic Depression, Affective psychosis -combination of two previous-, Non-affective psychosis -other psychosis-) for clinical status comparisons (affective psychosis vs non-affective psychosis, affective psychosis vs control, bipolar disorder vs psychotic depres- sion, bipolar disorder vs control and psychotic depression vs control). We built different logistic regression models using separately the different PRSs as predictors for each clinical comparison, adjusted by population stratification using the first 10 principal components (PC). Results: PRS SZ was the only able to distinguish between affective and non-affective psychosis, showing an associa- tion with non-affective psychosis (OR=0.36, 95% CI .19 - .67, p=0.001). Both PRS SZ (OR=3.08, 95% CI 1.76 – 5.37 p>0.001) and PRS BP (OR=1.92 95% CI 1.39 – 2.63 p<0.001) identified affective psychosis from controls, whereas PRS MDD only showed a trend. When comparing bipolar disor- der with psychotic depression PRS SZ was the only able to distinguish between categories, showing an association with bipolar disorder (OR=2.89 95% CI 1.01 – 8.24, p>0.001). Both PRS SZ (OR=5.59 95% CI 2.52 – 12.36, p<0.001) and PRS BD (OR=2.33 95% CI 1.49 – 3.65, p<0.001) were signifi- cantly associated with bipolar disorder when compared with controls; but interestingly, only PRS BD was positively asso- ciated with psychotic depression against controls (OR=1.63 95% CI 1.07 – 2.48, p=0.022). PRS MDD failed to identify both bipolar disorder and psychotic depression versus controls. Discussion: These results suggest that once the explained variance becomes more significant, PRS could be a useful marker to help us understanding the architecture of af- fective psychosis; for both positioning it on the spectrum of psychotic or mood disorders and delineating diagnostic boundaries. This potential use in High Risk or FEP settings may imply important differences in treatment and prognosis prediction.