Exemplar scoring identifies genetically separable phenotypes of lithium responsive bipolar disorder

Area of Science:

• Psychiatry
• Genetics
• Pharmacogenomics

Background:

• Predicting lithium response in bipolar disorder (BD) is challenging due to patient heterogeneity.
• Genomic markers for lithium response are difficult to identify in diverse patient groups.

Purpose of the Study:

• To investigate if patients with "exemplary phenotypes" show greater genetic distinguishability for lithium response (LiR) and non-response (LiNR).
• To assess if genetic data can better classify lithium responders and non-responders when focusing on patients with highly predictive clinical features.

Main Methods:

• Calculated a "clinical exemplar score" based on phenotype-LiR/LiNR predictability in 1266 BD patients.
• Compared genetic classification accuracy between "best clinical exemplars" (top 25%) and "poor clinical exemplars" (bottom 25%) of LiR/LiNR in 321 genotyped patients.

Main Results:

• Best clinical exemplars of LiR often had later onset, episodic course, no rapid cycling/psychosis, and fewer comorbidities.
• Genetic data achieved higher classification accuracy (AUC 0.88) for best clinical exemplars versus poor clinical exemplars (AUC 0.66, p=0.0032).
• Alzheimer's amyloid-secretase, G-protein-coupled receptor, muscarinic acetylcholine, and histamine H1R pathways were informative.

Conclusions:

• Focusing on patients with "exemplary phenotypes" enhances genetic prediction of lithium response in bipolar disorder.
• This approach may aid in discovering more robust genomic markers for mood stabilizer response.