Long-term test-retest reliability of resting-state networks in healthy elderly subjects and with amnestic mild cognitive impairment patients.

The investigation of cerebral resting-state networks (RSNs) by functional magnetic resonance imaging (fMRI) is a promising tool for the early diagnosis and follow-up of neuropsychiatric and neurodegenerative disorders like Alzheimer's disease (AD). In this context, the determination of inter-session reliability of these networks is crucial. However, data on network reliability in healthy elderly subjects is rare and does not exist for patients with amnestic mild cognitive impairment (aMCI), a prodromal stage of AD. Therefore, the aim of this study was to investigate the long-term test-retest reliability of RSNs in both groups. Twelve healthy controls (HC) and 13 aMCI patients underwent resting-state fMRI and neuropsychological testing (CERAD test battery) twice, at baseline and after 13-16 months. Resting-state fMRI data was decomposed into independent components using independent component analysis. Inter-session test-retest reliability of the resulting RSNs was determined by calculating voxel-wise intra-class correlation coefficients. Overall test-retest reliability of corresponding RSNs was moderate to high in both groups, but significantly higher in the HC group compared to the aMCI group (p < 0.001), while the cognitive performance within the CERAD test battery remained stable over time in either group. Most reliable RSNs derived from the HC group and were associated with sensory and motor as well as higher order cognitive and the default-mode function. Particularly low reliability was found in basal frontal regions, which are known to be prone to susceptibility-induced noise. We conclude that stable RSNs may represent healthy aging, whereas decreased RSN reliability may indicate progressive neuro-functional alterations before the actual manifestation of clinical symptoms.