EFFECTIVE CONNECTIVITY WITHIN THE DEFAULT MODE NETWORK IN MILD COGNITIVE IMPAIRMENT AND ALZHEIMER'S DISEASE
Cole John Cook,Gyujoon Hwang,Veena A. Nair,Andrew L. Alexander,Piero G. Antuono,Sanjay Asthana,Rasmus Birn,Cynthia M. Carlsson,Guangyu Chen,Dorothy Farrar Edwards,Malgorzata Franczak,Joseph S. Goveas,Sterling C. Johnson,Steven Kecskemeti,Arman P. Kulkarni,Elizabeth M. Meyerand,Rosaleena Mohanty,Andrew S. Nencka,Ozioma C. Okonkwo,Mary-Elizabeth Pasquesi
DOI: https://doi.org/10.1016/j.jalz.2018.06.2095
2018-01-01
Alzheimer s & Dementia
Abstract:The default mode network (DMN) is well-recognized as being involved in memory and affected by dementia due to Alzheimer's disease (AD), with differences in this network previously noted between healthy controls and AD patients. Previous studies of DMN effective connectivity in AD, such as used models of hemodynamic effective connectivity along with t-tests to compare connections between groups. Dynamic causal modeling (DCM) was used to explore neuronal effective connectivity differences between healthy controls and a group of mild cognitive impairment (MCI) and AD patients. Resting state fMRI data were acquired on 3T GE 750 scanners at the University of Wisconsin-Madison and Medical College of Wisconsin using whole-brain simultaneous multi-slice imaging (8 bands, 72 slices, TR 802 ms, voxel size 2 mm isotropic). Participants included 11 controls (6 females, average age=69.45 years) and 11 patients (5 females, average age=69.82 years), 8 MCI and 3 AD, (Table 1) enrolled in the NIH-sponsored Alzheimer's Disease Connectome Project. Following the Human Connectome Project minimal preprocessing pipeline, six motion parameters, motion parameters’ derivatives, and white matter and CSF signals were regressed out. DMN seed regions were extracted using coordinates (Table 2) in Van Dijk et al. (2010) to construct fully connected spectral DCMs. DCMs were the priors in a parametric empirical Bayes (PEB) model. Parameters in the second level Bayesian regression were group, sex, and mean centered age. Non-evident connections were removed via automated search. The final PEB model included seven connections (Figure 1) which differed between patients and controls following thresholding of automated search results for parameters with a posterior probability of 95% of being non-zero. Connections for controls are depicted in Figure 2. Patients showed decreased self-inhibition, denoted by a positive value, in the posterior cingulate cortex (PCC), left and right hippocampal formation (LHF, RHF), and left and right lateral parietal cortex (LLPC, RLPC) compared to controls. Additionally, decreased connectivity was noted from LLPC to the medial prefrontal cortex and increased connectivity was noted from RLPC to PCC. Connection differences between mild cognitive impairment and Alzheimer's disease patients and healthy controls in the default mode network for which strong evidence (posterior probability=0.95) exists. Self-connections are indicated by circles around the region label with the connection strength affixed on the bottom of the circle. For connections between regions, the connection's originating region is represented by the line color with the estimated strength near the terminating region. Connections in the default mode network of healthy controls for which strong evidence (posterior probability=0.95) exists. Self-connections are indicated by circles around the region label with the connection strength affixed on the bottom of the circle. For connections between regions, the connection's originating region is represented by the line color with the estimated strength near the terminating region. Connections with differed by sex are labeled ‘a/b’, with ‘a’ being the connection for males. If ‘a’ is a that means this value was not in the male network. Effective connectivity differed between AD/MCI and control within the DMN. Decreased self-inhibition would increase time to return to baseline, potentially influencing behavioral outcomes.