Using multi-trait polygenic scores to predict lithium responsiveness in patients with bipolar disorder

The predictive power of polygenic scores (PGSs) for lithium treatment response in bipolar disorder remains limited.

To enhance the prediction of lithium responsiveness by developing a multi-trait PGS (mt-PGS) combining genetic information from multiple phenotypes implicated in lithium response and/or bipolar disorder aetiology.

We analysed data collected from bipolar disorder patients who had received lithium treatment for at least 6 months and participated in the International Consortium on Lithium Genetics study (N = 2367). The Alda scale was used to assess lithium responsiveness, and treatment outcome was defined as continuous total Alda score (0-10) and categorical outcome (favourable ≥7 versus unfavourable response). PGSs were calculated for 60 phenotypes grouped into 5 clinical-biological clusters: psychiatric and behavioural (#23 phenotypes), cardiometabolic (#17), autoimmune/inflammatory (#5), neurocognitive (#8) and renal function (#7). We applied cross-validated machine learning regression approaches in both outcomes within each cluster, with the selected features from each cluster subsequently combined to construct the final mt-PGS models. Model performance was assessed using explained variance (R²) for the continuous outcome, and both McFadden's pseudo-R² and standard classification model parameters for the categorical outcome.

mt-PGS explained between 5.70% (continuous outcome) and 9.11% (categorical outcome) of the interindividual variability in lithium responsiveness. Classification accuracy (area under the curve) for the categorical outcome was 68.24% (95% CI: 64.98-71.66), with a Brier score of 0.257. Of the 5 clusters, the PGSs for psychiatric and behavioural phenotypes were most strongly associated with lithium responsiveness, accounting for 3.67-6.52% of its variability.

By integrating PGSs for multiple relevant phenotypes, predictive accuracy for lithium response improved substantially compared with single-trait methods. Future research incorporating larger, more diverse populations and combining genetic scores with clinical data holds promise for further enhancing prediction and advancing clinical implementation.