Altered resting-state activity in seasonal affective disorder

Abstract

At present, our knowledge about seasonal affective disorder (SAD) is based mainly up on clinical symptoms, epidemiology, behavioral characteristics and light therapy. Recently developed measures of resting-state functional brain activity might provide neurobiological markers of brain disorders. Studying functional brain activity in SAD could enhance our understanding of its nature and possible treatment strategies. Functional network connectivity (measured using ICA-dual regression), and amplitude of low-frequency fluctuations (ALFF) were measured in 45 antidepressant-free patients (39.78 ± 10.64, 30 ♀, 15 ♂) diagnosed with SAD and compared with age-, gender- and ethnicity-matched healthy controls (HCs) using resting-state functional magnetic resonance imaging. After correcting for Type 1 error at high model orders (inter-RSN correction), SAD patients showed significantly increased functional connectivity in 11 of the 47 identified RSNs. Increased functional connectivity involved RSNs such as visual, sensorimotor, and attentional networks. Moreover, our results revealed that SAD patients compared with HCs showed significant higher ALFF in the visual and right sensorimotor cortex. Abnormally altered functional activity detected in SAD supports previously reported attentional and psychomotor symptoms in patients suffering from SAD. Further studies, particularly under task conditions, are needed in order to specifically investigate cognitive deficits in SAD.

Keywords: ALFF; ICA; fMRI; functional connectivity; resting-state; seasonal affective disorder.

Figure 1

Top, Framework of the group independent component analysis (ICA) (concatenation approach, as implemented in MELODIC). Individual fMRI datasets are concatenated temporally (after performing group‐based PCA reduction on each individual fMRI dataset). Temporally concatenated group‐data reduced by principle component analysis (PCA) reduction. Then using ICA group ICA maps and group mixing matrix are estimated. Bottom, Dual‐regression approach involves (1) spatial regression of ICA maps into each subject's fMRI data to find subject‐specific time‐courses, (2) back regression of these time‐courses into the subject's fMRI data to detected subject‐specific ICA maps. Subject‐specific ICA maps enable voxel‐wise statistical testing across subjects, separately for each original group‐ICA map (to detect between‐group differences).

Figure 2

Inter‐RSN correction framework. Subject‐specific spatial ICA maps are spatially concatenated before permutation tests as a procedure to account for false positives when testing multiple ICA maps.

Figure 3

Top, Eleven resting‐state networks (RSNs) (red‐yellow) showing increased functional connectivity (green) in SAD patients. SAD‐related increased functional connectivity involves the sensorimotor (RSNs 1–3), attentional (RSN 4) and visual (RSNs 5–11) networks. The left hemisphere corresponds to the right side. The t‐score threshold is shown at the right. Numbers at the bottom left of the images refer to MNI coordinates (xyz). Bottom, ALFF between‐group difference map show significant higher ALFF in the visual and right sensorimotor cortex. Results are corrected for multiple comparisons (inter‐RSN correction, P < 0.05).