The general factor of psychopathology and brain network topology

Abstract

Recently a bifactor model has been introduced to characterize psychopathology. This model contains a general factor that captures shared variance across disorders as well as a specific internalizing and externalizing factor that are not correlated. This can be contrasted with the correlated factors model which contains a correlated internalizing and externalizing factor. The general factor may be linked to global properties of brain network organization. By applying graph theory analytics to neuroimaging data, it is possible to study network topology. We examined relations of latent factors of psychopathology (bifactor and correlated factors model) with both structural (diffusion weighted imaging data and morphometric properties) and functional (task related connectivity of a reward task) network topology. In these studies, we utilized a large young adult community twin sample that was oversampled on risk for psychopathology and ranged from 325 to 437 subjects depending on the imaging modality. Nominally significant correlates of the general factor were identified across all modalities with a measure of global integration emerging as the most robust measure. We further identified nominally significant correlates of the other second-order factors from the bifactor model and significant correlates of the correlated factors model. The specific externalizing and externalizing factors had the largest number of correlates. The correlated factors model yielded the most robust correlates with several relations that survived correction for multiple comparisons. Overall, these findings provide additional evidence for the utility of applying dimensional approaches to study neural correlates of psychopathology.