Dynamic Whole-Brain Connectivity underlying Abnormal Brain States in Late-onset Depression
Mingze Xu1,2, Shiyang Chen2, Bing Ji2,3, Jiuquan Zhang4, Huaiqiu Zhu1, Yi Zhang5, Yonggui Yuan6, Jiahong Gao1, Yijun Liu1, and Xiaoping Hu2

1Biomedical Engineering, Peking University, Beijing, China, People's Republic of, 2Biomedical Engineering, Emory University & Georgia Institute of Technology, Atlanta, GA, United States, 3University of Shanghai for Science & Technology, Shanghai, China, People's Republic of, 4Department of Radiology, Southwest Hospital, Third Military Medical University, Chongqing, China, People's Republic of, 5School of Life Science and Technology, Xidian University, Shaanxi, China, People's Republic of, 6Department of Psychosomatics and Psychiatry, ZhongDa Hospital, School of Medicine, Southeast University, Nanjing, China, People's Republic of

Synopsis

We conducted dynamic whole-brain connectivity analysis in Late-onset depression (LOD) to investigate alterations in brain networks. All subjects’ ROI-to-ROI dynamic FC were explored using a data-driven method to obtain the most explanatory states. Each state indicate a particular ROI-to-ROI FC pattern. The property of each state were determined based on its scores across time. Besides decreased FC in normal state, we found LOD patients switch between brain states more frequently and tend to enter LOD-risk states, due to and its high states variance and dominating increased FC in LOD-risk states. These results suggest neural mechanisms of disorder from dynamic perspective.

PURPOSE

Late-onset depression (LOD) is associated with an increased risk for developing Alzheimer’s disease[1]. However, neurobiological features of LOD remain unclear. Dynamics in resting state functional connectivity could provide additional insights regarding brain networks [2]. Therefore, we performed whole-brain dynamic connectivity analysis in LOD patients to examine possible alterations in brain networks. We hypothesized that LOD patients frequently switch into the “LOD” states from other states.

METHODS

Subjects and Scans

This study was conducted with approval from the medical ethics committee for clinical research of ZhongDa Hospital affiliated with Southeast University of China. Thirty four LOD patients and forty four normal controls (NCs), matched for age, gender, cerebrovascular risk factor and education, were studied. All subjects were assessed with a neuropsychological test battery, including overall cognition, Hamilton depression rating scale (HAMD), and working memory, and underwent resting-state functional MRI scans using a Siemens Verio 3.0 Tesla scanner (Siemens Medical Solutions, Erlangen, Germany). All imaging data were manually checked by two experienced radiologists for quality controls before processing.

Dynamic Functional Connectivity

rfMRI data were first preprocessed using functional realignment, functional slice timing, structural segmentation & normalization, functional normalization and band pass filter, detrending and regression out WM, CSF and other noise. A total 132 ROIs were defined as the combination of 91 cortical ROIs and 15 subcortical ROIs based on the FSL Harvard-Oxford maximum likelihood atlas and 26 cerebellar ROIs from the AAL atlas (26 ROIs). Subsequently, ROI-to-ROI connectivity matrices were calculated using CONN [3]. A data-driven method (http://www.nitrc.org/projects/conn) based on PsychoPhysiological Interaction model was applied to obtain the most explanatory discrete functional connectivity states by exploring all subjects’ ROI-to-ROI connectivity matrices across time. Each state is associated with a particular ROI-to-ROI connectivity pattern. Also the property of each state were determined based on its scores across time. Finally, state properties including temporal variability and rate of state change in each state were determined and statistically analyzed for differences and correlations. In addition, whole brain functional connectivities in each state were statistically compared between LOD and NC.

Results

States

As shown in Fig. 1, 6 states were derived reflecting the largest temporal modulation effects observed in connectivity matrix cross all subjects. The time courses at the bottom depict the scores of six states in subject NC01 over time.

States Properties

Temporal variability and rate of state change were calculated for each state. Results are presented in Fig. 2. LOD patients showed significantly higher state variance (P<0.05) than NCs in states 2 and 4. Rates of state change were found significantly correlated with HAMD. Several other neuropsychological test scores also exhibited similar correlation.

ROI-to-ROI connectivity in each States

Each state is associated with an ROI-to-ROI connectivity pattern. Whole brain functional connectivities were compared between NC and LOD groups in each pattern. Results are shown in Fig. 3.The central ring illustrates the conventional functional connectivity results. The surrounding rings depict the dynamic functional connectivity results in each state (FDR corrected P<0.05, red indicated NC>LOD and blue indicated LOD>NC).

DISCUSSION

In this study, temporal variability and rate of state change were found significantly altered in LOD and their neuropsychological test scores, suggesting that LOD patients switch brain states more frequently. In addition, the functional connectivity in each state was also changed. Decreased functional connectivities in LOD were mostly seen in state 1 and in connections present in conventional functional connectivity map. Some increased connectivities were seen in states 2-6 indicating LOD might exhibit stronger connections in these states, especially in states 4-6, which might represent LOD-risk states. These results confirmed our hypothesis, as illustrated in Fig. 4, that unlike healthy person, LOD patients switch between brain states more frequently and tend to enter LOD-risk states. State 2 and 3 might indicate a middle state, from which LOD-risk state is entered.

CONCLUSION

This study found that LOD patients exhibit altered brain network dynamics, with probably more time spent on LOD-risk state. This result have revealed some neural mechanisms of LOD and provided new insight for prevention and treatment in LOD and other disorder disease.

Acknowledgements

We wish to thank all the participants in Yijun Liu's Lab and Xiaoping Hu's Lab and the financial support of the National Natural Science Foundation of China (81371488, Yonggui Yuan).

References

1. Bai, F., et al., Topologically convergent and divergent structural connectivity patterns between patients with remitted geriatric depression and amnestic mild cognitive impairment. The Journal of Neuroscience, 2012. 32(12): p. 4307-4318.

2. Li, X., et al., Dynamic functional connectomics signatures for characterization and differentiation of PTSD patients. Hum Brain Mapp, 2014. 35(4): p. 1761-78.

3. Whitfield-Gabrieli, S. and A. Nieto-Castanon, Conn: a functional connectivity toolbox for correlated and anticorrelated brain networks. Brain connectivity, 2012. 2(3): p. 125-141.

Figures

Fig.1 6 states were derived reflecting the largest temporal modulation effects observed in connectivity matrix cross all subjects. Each matrix size is 132*132 and suggests a particular ROI-to-ROI connectivity pattern. For example, state 1 indicates the positive FC among the cortical ROIs. The time courses at the bottom depict the scores of six states in subject NC01 over scanning time.

Fig.2 LOD patients showed significantly higher state variance (P<0.05) than NCs in states 2 and 4. Rates of state change were found significantly correlated with HAMD. Several other neuropsychological test scores also exhibited similar correlation.

Fig.3 The central ring illustrates the conventional functional connectivity results. The surrounding rings depict the dynamic functional connectivity results in each state (FDR corrected P<0.05, red indicated NC>LOD and blue indicated LOD>NC).

Fig.4 Due to dominated strengthened FC in state 4-6 for LOD, we hypothesis state 4-6 represent LOD-risk states. Healthy people showed high FC in normal state (state 1) and is less impossible to enter LOD-risk states. LOD patients switch between brain states more frequently and tend to enter LOD-risk states. State 2 and 3 might indicate a middle state, from which LOD-risk state is entered.



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
4331