Twin studies have lacked statistical power to apply advanced genetic modelling techniques to the search for cognitive endophenotypes for bipolar disorder.
To quantify the shared genetic variability between bipolar disorder and cognitive measures.
Structural equation modelling was performed on cognitive data collected from 331 twins/siblings of varying genetic relatedness, disease status and concordance for bipolar disorder.
Using a parsimonious AE model, verbal episodic and spatial working memory showed statistically significant genetic correlations with bipolar disorder (rg = |0.23|–|0.27|), which lost statistical significance after covarying for affective symptoms. Using an ACE model, IQ and visual-spatial learning showed statistically significant genetic correlations with bipolar disorder (rg = |0.51|–|1.00|), which remained significant after covarying for affective symptoms.
Verbal episodic and spatial working memory capture a modest fraction of the bipolar diathesis. IQ and visual-spatial learning may tap into genetic substrates of non-affective symptomatology in bipolar disorder.