043_ Developing a polygenic score for lithium treatment response
Research Question and Aims
Using dataset from the International Consortium on Lithium Genetics (ConLi+Gen), we recently developed a lithium response polygenic score (Li+PGS) [for both dichotomous and continuous scales] in patients with bipolar disorder (BD) and demonstrated for the first time that Li+PGS is positively associated with response to lithium (adjusted OR=1.39 [95%CI: 1.26, 1.54]. Compared to bipolar patients in 1st decile of the risk distribution, individuals in the highest 10% of the distribution have ~3.47[95%CI: 2.22, 5.47] fold favorable response to lithium treatment (see the figure below). Li+PGS can help to stratify bipolar patients by the treatment pattern, and we proposed that it is time to apply polygenic scores in clinical care as a pharmacogenomic testing tool. We are requesting dataset from the PsyCourse cohort study to test whether our findings can be replicated in an independent cohort.
We aim to generate Li+PGS and test its association with lithium response in patients with BD.
Participants: Data from all samples in PsyCourse who have genotype data available will be included in this study.
Polygenic risk scores: GWAS summary statistic from the ConLiGen sample will be used as discovery sample for computing Li+PGS using “a polygenic risk score by continuous shrinkage (PRS-CS) software” that utilizes a Bayesian regression framework (Ge, Chen et al. 2019). PRS-CS infer posterior SNP effect sizes under continuous shrinkage (CS) using GWAS summary statistics and an external linkage disequilibrium (LD) reference panel. For the current analysis, the precomputed LD pattern of the 1000 Genomes European reference panel and (Southam, Gilly et al. 2017) and the discovery GWAS summary statistics will be used to calculate PGS scores.
Statistical analysis: To assess the association of Li+PGS with lithium treatment response, a binary logistic regression model will be applied for the binary outcome (lithium response versus non-response) and a tobit analysis model (censored regression) will be used for the continuous outcome (Alda total). Further, we will divide the study sample into deciles, ranging from the lowest polygenic load (1st decile, reference group) to the highest polygenic load (10th decile). Then, we will compare BD patients in the higher Li+PGS load deciles (2nd –10th deciles) with patients in the lowest Li+PGS load decile (1st decile). The proportion of phenotypic variance explained by PGSs will be computed as the difference in R2 of the models fit with and without the PGS scores. Each modelling analysis will be adjusted for covariates: age, sex, and principal components (PCs). Statistical significance will be determined at p?0.05.
Current psychiatric treatment
Lithium treatment information from Visit 1 to Visit 4
Alda scale in Visit 4
Genetic data (both genotype & imputed data).