Multivariate associations between structural brain changes and cognitive impairment in partially or fully remitted persons with bipolar disorder

Abstract

Background

Cognitive impairment across cognitive domains and brain structure alterations are well documented in persons with bipolar disorder (BD) but the association between them is still unclear. Previous studies have generally applied univariate models to investigate brain-cognition correlations, which limits the discovery of complex association patterns. The aim of this study was to apply canonical correlation analysis (CCA) to identify multivariate associations between brain structure and cognitive impairment in BD.

Methods

Cognitively impaired persons with BD (n = 169) in full or partial remission were included from four prior pro-cognitive intervention studies. We included healthy controls (HC, n = 40) for the calculation of covariate-adjusted brain and cognition z-scores. All participants underwent structural magnetic resonance imaging and an extensive cognitive test battery. We conducted principal component analysis on the brain data within the BD group to reduce the number of variables in the dataset. We then applied CCA to investigate multivariate associations between brain structure and cognition in the BD cohort.

Results

Poorer performance across working memory, psychomotor speed, executive functions, and verbal learning and memory correlated with lower grey matter volume in frontotemporal regions, the right hippocampus, and the left caudate nucleus, and with larger frontotemporal and right posterior cingulate gyral thickness.

Conclusions

The association between cognition, reduced grey matter volume and larger thickness in frontotemporal and posterior cingulate regions suggests that cognitive impairment originates from dysregulated, rather than simply reduced, neuroplasticity processes. Aberrant volume and thickness measures in these regions are potential treatment targets to promote cognition in BD.