Deriving Mendelian Randomization-Based Causal Networks of Brain Imaging Phenotypes and Bipolar Disorder.

Area of Science:

• Neuroscience
• Psychiatry
• Genetics

Background:

• Previous studies observed neuroanatomical variations in bipolar disorder (BD).
• The causal links between these variations and BD remain unclear.

Purpose of the Study:

• To investigate the causal relationship between neuroimaging phenotypes and bipolar disorder (BD).
• To explore whether neuroanatomical variations cause or are caused by BD.

Main Methods:

• Mendelian randomization (MR) was applied to UK Biobank neuroimaging data and BD GWAS summary statistics.
• A novel inverse sparse regression model estimated direct causal effects (DCE).
• Causal scores were derived using DCE weights and applied to clinical cohorts.

Main Results:

• 28 significant causal relationships were identified, with 27 showing neuroimaging phenotypes affecting BD.
• White matter tract phenotypes demonstrated stronger causal effects on BD than vice versa.
• A neuroimaging-derived causal score predicted BD in an adolescent cohort.

Conclusions:

• Mendelian randomization suggests neuroanatomical variation, especially in white matter tracts, is a likely cause of BD.
• Findings highlight the potential causal role of brain structure in BD etiology.
• Further research using replication and triangulation is recommended.