Polygenic scores may aid patient stratification and facilitate subphenotyping: for instance, among bipolar disorder cases polygenic scores for schizophrenia have been shown to distinguish schizoaffective cases from others 48. A large Psychiatric Genomics Consortium study recently demonstrated that a PRS derived from a study of schizophrenia predicted psychotic symptoms in patients with bipolar disorder 49, a finding replicated in an independent study 50. A further study showed that schizophrenia PRS was inversely correlated with response to lithium treatment 41. Recent research has also shown that PRS may be useful in predicting future psychiatric diagnoses, such that individuals with first-episode psychosis who subsequently developed schizophrenia had significantly different PRS from those who did not become chronically ill, although the discriminative accuracy is not yet sufficient for clinical utility 51. Furthermore, higher PRS in patients with first-episode psychosis significantly predicted higher post-treatment symptom scores after 12-week follow-up, indicating that patients with low PRS were almost twice as likely to be treatment responders than patients with high PRS (that is, based on a median split into two categories) 52.
