Shared and unique functional connectivity correlates of geriatric depression subscales

• Published in: The American journal of geriatric psychiatry 2023-03, Vol.31 (3), p.S103-S104
• Main Authors: Saxena, Parnika, Turnbull, Adam, Sommer, Barbara, Kim, Daniel, Lin, Feng Vankee
• Format: Article
• Language: English
• ISSN: 1064-7481; 1545-7214
• DOI: 10.1016/j.jagp.2022.12.150

Major Depressive Disorder is a serious and disabling illness, yet its presentation is variable, making studies difficult to compare and causing inconsistent treatment responses. Depression in older adults can be harder to diagnose and specify subtypes due to its unique presentation, leading to a succession of ‘trial and error’ treatments. Identifying biomarkers to help with accurate and well-defined diagnosis is important to ensure the best possible treatment regimen. Functional connectivity (FC) biomarkers (which are statistical relationships between cerebral signals over time and thus potentially help to make the functional interactions between two or more brain regions) may be able to provide a means of distinguishing depression subtypes by providing insight into their shared and unique functional neural bases. Baseline data were retrospectively analyzed from a double-blind, randomized controlled trial that evaluated the effect of brain stimulation on mood in 40 subjects with mild cognitive impairment (MCI) (mean age = 71.7, SD =7.0; 24 females). We analyzed the FC correlates of six validated subscales from the Geriatric Depression Scale 30 (GDS-30): withdrawal-apathy-vigor (apathy), anxiety, memory impairment, hopelessness, dysphoric mood, and agitation. We compared multiple linear regression models with each subscale as the independent variable, with a similar model predicting the GDS-30 total, and we were able to identify FC relationships for geriatric depression in general, and each subscale specifically. Seven well-validated large-scale FC networks from Yeo & Krienen (2011) were used to generate within- and between-network FC scores that were used as dependent variables (separate models for within- and between-network FC). Findings were FDR-corrected within each model to account for the number of dependent variables. Neurodegeneration score (from t1 anatomical scan) and motion during the resting state scan were included as covariates. GDS-30 total score was significantly related to within sensorimotor (within-SM: B=-19.1, SE=6.9, p(FDR-corrected)=.035), within default mode network (within-DMN): B=62.6, SE=19.2, p(FDR-corrected)=.021), visual to dorsal attention network (visual-DAN): B=-43.3, SE=12.7, p(FDR-corrected)=.046), and DAN to DMN (DAN-DMN): B=-105.1, SE=29.8, p(FDR-corrected)=.046) FC. For the GDS-30 subscales, cognitive impairment showed the same relationships to FC as the GDS-30 total score (within-SM: B=-1.23, SE=0.45, p(FDR-corrected)=