Polygenic risk for schizophrenia modifies the real-world effectiveness of antipsychotic treatment in psychotic disorders

Introduction

Pharmacological treatment of psychotic disorders shows substantial interindividual variability in effectiveness and tolerability, contributing to reactive treatment decisions and suboptimal outcomes. Despite advances in the genetic understanding of major psychiatric disorders such as schizophrenia, bipolar disorder, and major depressive disorder (MDD) [1], their potential in understanding treatment response heterogeneity in real-world clinical settings remains unknown.

Aims

To evaluate whether polygenic risk score (PRS) for schizophrenia, bipolar disorder, or MDD moderate the effectiveness and failure of pharmacotherapy in individuals with psychotic disorders.

Methods

We used data from the SUPER-Finland study, part of the Stanley Global Neuropsychiatric Genetics Initiative, comprising 10,403 actively consented individuals with a psychotic illness (ICD-10 codes: F20–F29, F30, F31, F32.3, and F33.3). The follow-up period from linked registers was 1996–2018. The primary investigated treatment exposures were antipsychotics, and secondary exposures were mood stabilizers and antidepressants. The primary outcome was psychiatric hospitalization (ICD-10 codes: any F* hospitalization), which was used as a proxy for severe relapse, and the secondary outcome was all-cause discontinuation (i.e., discontinuation or switching of the studied medication). To eliminate selection bias, we employed within-individual analyses, in which each individual was used as their own control. Within-individual Cox regression models were adjusted to account for the temporal order of treatments, time since cohort entry, and the use of concomitant psychopharmacotherapies. Antipsychotic doses were classified using a defined daily dose (DDD). PRSs for schizophrenia, bipolar disorder, and MDD were calculated using established genome-wide association study weights.

Results

The analyzed sample comprised 9,408 individuals (4,741 males) with psychotic disorders, including schizophrenia (N=5,494, 58.4%), schizoaffective disorders (N=894, 9.5%), bipolar disorder (N=1,524, 16.2%), psychotic depression (N=513, 5.5%), and other psychoses (N=983, 10.5%), all with available PRS data and register linkage. The mean (SD) age at psychosis onset was 29.11 (11.24) years, and the mean (SD) follow-up duration was 15.08 (7.13) years. PRS for schizophrenia, but not for bipolar disorder or MDD, moderated the association between antipsychotic use and the risk of psychiatric hospitalization (Z = −4.90, P < 0.0001). Specifically, the adjusted hazard ratio (aHR) decreased across PRS deciles, from 0.95 (95% CI 0.87–1.05) in the lowest decile to 0.74 (95% CI 0.67–0.81) in the highest decile. No corresponding interactions were observed for antidepressants or mood stabilizers with any PRS. Among individual antipsychotics, significant interactions with schizophrenia PRS (indicating lower aHRs at higher PRS levels) were observed for first-generation antipsychotics (P = 0.0007), clozapine (P = 0.016), olanzapine (P < 0.0001), aripiprazole (P = 0.008), and antipsychotic polypharmacy (P < 0.0001), whereas no interaction was detected for risperidone, quetiapine, or long-acting injectable (LAI) antipsychotics. Across schizophrenia PRS strata, the lowest risk of rehospitalization was observed at standard dosing (DDD: 0.6-1.6), consistent with a U-shaped relationship, regardless of the genetic risk. Finally, compared with oral olanzapine, higher schizophrenia PRS was associated with increased discontinuation of LAI antipsychotics, quetiapine and risperidone.

Conclusions

These findings suggest that common genetic variation underlying schizophrenia contributes to heterogeneity in antipsychotic treatment response in real-world clinical settings.

References

Murray GK, Lin T, Austin J, McGrath JJ, Hickie IB, Wray NR. Could Polygenic Risk Scores Be Useful in Psychiatry? A Review. JAMA Psychiatry. 2021;78(2):210–219. doi:10.1001/jamapsychiatry.2020.3042