Using network analysis to explore cognitive domains in patients with unipolar versus bipolar depression: a prospective naturalistic study

Abstract

Objective.Despite growing evidence in the field of cognitive function in mood disorders, the neurocognitive profiles of patients with unipolar and bipolar depression still need further characterization. In this study, we applied network analysis, hypothesizing this approach could highlight differences between major depressive disorder (MDD) and bipolar disorder (BD) from a cognitive perspective.Methods.The cognitive performance of 109 patients (72 unipolar and 37 bipolar depressed outpatients) was assessed through the Montreal Cognitive Assessment (MoCA), and a series of clinical variables were collected. Differences in cognitive performance between MDD and BD patients were tested using non-parametric tests. Moreover, a network graph representing MoCA domains as nodes and Spearman’s rho correlation coefficients between the domains as edges was constructed for each group.Results.The presence of mild cognitive impairment was observed in both MDD and BD patients during depression. No statistical significant difference was found between the two groups in terms of overall cognitive performance and across single domains. Nonetheless, network analytic metrics demonstrated different roles of memory and executive dysfunction in MDD versus BD patients: in particular, MDD network was more densely interconnected than BD network, and memory was the node with the highest betweenness and closeness centrality in MDD, while executive function was more central in BD.Conclusions.From a network analytic perspective, memory impairment displays a central role in the cognitive impairment of patients with unipolar depression, whereas executive dysfunction appears to be more central in bipolar depression. Further research is warranted to confirm our results.