Background The topological architecture of the whole-brain functional networks in those with and without late-life depression (LLD) and amnestic mild cognitive impairment (aMCI) are unknown.

Aims To investigate the differences in the small-world measures and the modular community structure of the functional networks between patients with LLD and aMCI when occurring alone or in combination and cognitively healthy non-depressed controls.

Methods 79 elderly participants (LLD (n=23), aMCI (n=18), comorbid LLD and aMCI (n=13), and controls (n=25)) completed neuropsychiatric assessments. Graph theoretical methods were employed on resting-state functional connectivity MRI data.

Results LLD and aMCI comorbidity was associated with the greatest disruptions in functional integration measures (decreased global efficiency and increased path length); both LLD groups showed abnormal functional segregation (reduced local efficiency). The modular network organisation was most variable in the comorbid group, followed by patients with LLD-only. Decreased mean global, local and nodal efficiency metrics were associated with greater depressive symptom severity but not memory performance.

Conclusions Considering the whole brain as a complex network may provide unique insights on the neurobiological underpinnings of LLD with and without cognitive impairment.