Mapping development trajectories of default mode network dysfunction in major depressive disorder using the subtype and stage inference model.

BACKGROUND

Major depressive disorder (MDD) exhibits clinical neuroimaging heterogeneity, complicating efforts toward biologically informed stratification. In particular, heterogeneity in large-scale brain network function may reflect distinct patterns of dysfunction. This study aimed to identify data-driven functional heterogeneity within the default mode network (DMN) in MDD using resting-state functional MRI.

METHODS

We employed the data-driven Subtype and Stage Inference (SuStaIn) model to analyze cross-sectional magnetic resonance imaging (MRI) data from 815 patients with MDD and 816 healthy controls. SuStaIn was used to identify subtype-specific patterns and their relative ordering in DMN functional alterations based on cross-sectional data.

RESULTS

Two distinct fALFF-based patterns were identified. Subtype 1 showed early involvement of the right parahippocampal region, followed by ventrolateral prefrontal and temporal areas, consistent with a limbic-to-visuolinguistic functional gradient. Subtype 2 was characterized by predominant involvement of dorsolateral prefrontal, anterior cingulate, and insular regions, suggesting a dominant executive-control-related functional profile, as reflected by the model-inferred ordering. NeuroSynth-based cognitive decoding broadly supported these divergent functional associations.

CONCLUSIONS

This study identifies DMN-based functional subtypes of MDD with distinct model-inferred patterns of functional abnormality. By characterizing heterogeneity in large-scale brain network dysfunction, these findings provide a conceptual framework for future subtype-informed research and hypothesis generation in MDD.