Modeling based on genetic factors and clinical variables will enable better prediction of response to lithium treatment among patients with bipolar disorder.
Genetic and clinical data were collected from the International Consortium on Lithium Genetics project. Polygenic risk scores were determined for 692 patients with bipolar disorder.
The optimal linear model, which was based on polygenic risk scores and clinical variables, accounted for 5.1% of the variance in lithium response, while the optimal linear model, which was based on clinical variables only, accounted for 8.1% of the variance in lithium response. The non-linear model prediction improved to 13.7% with genomic stratification.