A Longitudinal Follow-Up Study Of The Characteristics Of Resting-State Functional Connectivity And Brain Network In Patients With Major Depressive Disorder

Objective:Major depressive disorder（MDD）is a kind of common mental disorder with high prevalence,high recurrence rate and low remission rate.Functional magnetic resonance imaging（f MRI）is expected to help detect the characteristics and possible biomarkers of emotional circuits in patients with MDD at the level of functional connectivity（FC）and brain network topology.The purpose of this study is to explore the characteristics of resting-state FC and brain network topology related to diagnosis and treatment in patients with MDD at the whole-brain level using network-based statistical（NBS）analysis and graph-theory-based analytic methods,based on a large sample of drug-naive patients with MDD before and after 6 months’antidepressant treatment.Methods:Two hundred and thirty-five drug-naive patients with with DSM-5 defined major depressive disorder and 163 healthy controls（HCs）were recruited at baseline.Demographic and clinical information of participants was collected,and the depression and anxiety symptoms were evaluated.Then the participants received resting-state f MRI（rs-f MRI）scan.After baseline assessment,patients with MDD received a first-line antidepressant treatment under the instruction of clinicians.At the end of the 2^nd week,1^st month,2^nd month,3^rd month,4^th month,5^thmonth and 6^th month,patients were assessed for depression and anxiety severity using the 24-item Hamilton depression rating scale(HAMD₂₄)and Hamilton anxiety rating scale（HAMA）.After 6 months’antidepressant treatment,the patients received a second rs-f MRI scan,which is the same as that at baseline.“Clinical remission”was defined as HAMD≤7 for two consecutive months during follow-up and persists till the end of the 6^th month.First,NBS and graph-theory-based analysis were used to explore the abnormalities of FC and brain network topological properties in patients with MDD compared with HCs.Then,correlation between the clinical variables and neuroimaging abnormalities were evaluated.Next,Changes of these abnormal characteristics in patients with MDD before and after treatment were further analyzed.Finally,differences in FC and brain network topological properties before and after treatment between remitted and unremitted patients were investigated.Results:A total of 196 patients with MDD and 143 HCs were included in the baseline analyses,and 114 patients with MDD were included in the follow-up analyses.Among the 114 patients with MDD who participated completed the 6-month follow-up,83（72.8%）achieved clinical remission（remitted group）,and the remaining 31（27.2%）patients did not achieve clinical remission（unremitted group）.（1）NBS analyses:a.Compared with HCs,patients with MDD showed significant difference in a subnetwork composed of 197 decreased FC and a subnetwork composed of 131 increased FC.The decreased FCs were mainly derived from the left precuneus cortex（located in the subcortical network）,left cerebellar cortex（located in the visual network II）and the right superior temporal gyrus（located in the motor network）.The increased FCs were mainly derived from the left cerebellar cortex（located in the visual network II）,left precuneus cortex（salience network）and left cerebellar cortex（located in the frontal-parietal network）.Total illness duration was positively correlated with 4 increased FCs between different cerebellum regions;episodes were positively correlated with 4increased FCs between the middle temporal gyrus and the hypothalamus,and between the middle temporal gyrus and the cerebellum;HAMD was positively correlated with 4 increased FCs between the dorsolateral prefrontal cortex and cerebellum,and between the supplementary motor area and cerebellum.b.One hundred and fifty five of the 197 decreased FCs and 100 of the 131 increased FCs remained unchanged after 6months’treatment,among which 102 decreased FCs remained lower than HCs and 61 remained higher than HCs,showing a state-independent feature.Forty two of the 197 decreased FCs were increased significantly after 6 months’treatment,among which 40 FCs exhibited no significant difference between patients at follow-up and HCs,showing a state-dependent feature.Thirty-one of the increased 131 FCs were decreased significantly after 6 months’treatment,among which 29 FCs exhibited no significant difference between patients at follow-up and HCs,showing a state-dependent feature.c.No significant difference between the remitted group and the unremitted group in the FCs at baseline,at follow-up or changes between baseline and follow-up.（2）Graph-theory-based analyses:a.The clustering coefficient and local nodal efficiency of the left hippocampus（located in the subcortical network）and the shortest path length of nodes in the left visuomotor area（located in the default mode network）of patients with MDD were significantly higher than those of HCs.The global nodal efficiency and degree centrality of the left visuomotor area（located in the default mode network）were significantly lower than those of HCs.The above indicators were not significantly correlated with the total illness duration,the current illness duration,episodes,HAMD₂₄,and HAMA.b.After 6months of antidepressant treatment,the clustering coefficient and nodal local efficiency of the left hippocampus of MDD patients showed no significant change compared with that at baseline,and were still higher than those in the HCs;the shortest path length and degree centrality of the left visuomotor area also showed no significant change compared with those at baseline,and were still lower than those in the HCs;the above 4indicators showed a state-independent feature.After 6 months of antidepressant treatment,the nodal global efficiency of the left visuomotor area was significantly higher than that at baseline,and was no longer significantly different from that of HCs,showing a state-dependent feature.c.The pre-and post-treatment change in clustering coefficient and nodal local efficiency of the left second visual cortex,the nodal global efficiency and degree centrality of the right angular gyrus,the betweenness centrality of the left second visual cortex,the shortest path length of the right angular gyrus,the degree centrality of the left primary motor area,and the betweenness centrality of the right dorsal cingulate gyrus showed significant difference between remitters and non-remitters.d.The clustering coefficient and nodal local efficiency of the left fusiform gyrus,and the shortest path length of the left primary motor area in the remitted group at baseline were significantly higher than those in the unremitted group.The nodal global efficiency and degree centrality of the primary motor area,and the nodal global efficiency and degree centrality of the left anterior supplementary motor area of the remitted group at baseline were significantly lower than those in the unremitted group.Conclusion:1.Compared with HCs,patients with MDD have multiple abnormalities in resting-state FCs and brain network topological properties.These abnormalities are mainly distributed in regions of the subcortical network,default mode network,salience network,frontal-parietal network,visual network II and motor network.2.Most of the abnormal resting-state FCs at baseline,the clustering coefficient and node local efficiency of the left hippocampus,the shortest path length and degree centrality of the left visual motor area are state-independent;a few abnormal resting-state FCs at baseline and the nodal global efficiency of the left visuomotor area are state-dependent.3.The clustering coefficient and nodal local efficiency of the left second visual cortex,the nodal global efficiency and degree centrality of the right angular gyrus,the betweenness centrality of the left second visual cortex,the shortest path length of the right angular gyrus,the degree centrality of the left primary motor area,and the betweenness centrality of the right dorsal cingulate gyrus may be related to the mechanism of action of antidepressants.4.The clustering coefficient and nodal local efficiency of the left fusiform gyrus,the shortest path length,nodal global efficiency and degree centrality of the left primary motor area,the nodal global efficiency and degree centrality of the left anterior supplementary motor area may serve as potential predictors for antidepressant effect.