Magnetic Resonance Imaging Based Multimodal Brain Connectomics Of State- And Trait-related Dysfunctions In Bipolar Disorder

Objective:Bipolar Disorder(BD)is an affective disorder that is prone to recurring attacks,affecting more than 1%of the world’s population and causing serious family and social burdens.Magnetic resonance imaging technology has always been widely used in the research of psychiatric diseases due to its advantages of non-invasiveness,nonradiation,and high resolution.However,the large heterogeneity and poor reproducibility of BD results in imaging studies have always plagued and hindered our understanding of the pathophysiological mechanism.An important reason for the inconsistent results is that there are conversions in mood state of BD.BD is mainly manifested in repeated mania,depression,and mixed episodes,interspersed with periods of emotional stability.Certain abnormalities change with repeated mood phase conversions,or may be susceptible to factors such as disease progression,termed state-related features.However,some abnormalities remain relatively stable during the occurrence of the disease,and even exist in the early stage of the disease and in BD genetic high-risk groups(Genetic High Risk,GHR-BD),which may be a feature related to the disease itself,known as trait-related features.Studying the state-and trait-related characteristics of BD can help us understand the pathophysiological mechanisms in the development of the disease.BD is a highly genetic psychiatric disorder,and its trait-related features may be influenced by genetic factors.frontal-limbic circuit variants in BD may be a BD-specific loop abnormality and may already exist in people at genetic high risk for BD(GHR-BD).The human brain is complex;it is a system in which separation and integration are the main functions.The normal brain is always in a homeostasis of separation and integration.Brain networks are often portrayed as connected networks with brain regions as nodes and brain connections as edges.Analytical methods using graph theory have been widely used to explore the topological properties of brain networks in patients with mental disorders.Graph theory can be used to study both global network changes and changes affecting only specific regions and is well suited to study complex networks.The onset of BD is the result of multiple brain regions and neural tracts acting together.Analyzing the topological properties of BD and GHR-BD in the whole brain as a whole,in the prefrontal-limbic circuit,and in the nodes can help us gain a more comprehensive understanding of the trait-and state-related features of BD.Most previous studies have mixed BDs of different mood phases;only a small portion of possible connections have been examined using imaging data from a single modality,ignoring the role of connections on larger-scale networks.Therefore,this study will analyze the brain network characteristics of BD in different mood phases at the global,prefrontal-limbic circuit and node levels based on multimodal magnetic resonance imaging,and then clarify the relationship between these features and clinical symptoms.We will explore the state-and trait-related characteristics of the disease,and then determine whether these characteristics have changed in the genetic high-risk population that has not yet developed the disease,and explore the he dynamic changes in resiliency.Method:Experiment 1:A total of 227 subjects aged 18-50 were collected in this study,including 152 patients with Bipolar Disorder(BD)and 75 healthy controls(HC).BD patients:After a structured interview,the diagnosis of BD is currently in line with the Diagnostic and Statistical Manual of Mental Disorders-Ⅳ Version(DSM-Ⅳ).BD patients were divided into three groups:Euthymic Bipolar Disorder(EBD):64 patients,who met the diagnostic criteria of DSM-IV currently in emotional remission with the total score of Hamilton Depression Scale-17 items(HAMD-17)and Young Manic Rating Scale(YMRS)are both<7 points;Manic Bipolar Disorder(MBD):36 patients,patients meet the current diagnostic criteria of DSM-IV for manic episodes,with HAMD total score<7 and YMRS total score>13;Depressed Bipolar Disorder(DBD):52 patients,patients currently in line with DSM-IV diagnostic criteria for depressive episodes,HAMD total score>20,YMRS total score<5.All subjects underwent demographic and clinical data collection and were assessed by HAMD-17,HAMA,and YMRS scales,followed by brain structure magnetic resonance imaging and resting-state functional magnetic resonance imaging.First,the structural and functional connection networks are established respectively through the Automated Anatomical Labeling(AAL-90)template.The structural connection network is based on the 90 partitions of the AAL-90 template as nodes,and the weight of the edge is defined as the number of white matter fibers between two nodes,forming a 90＊90 matrix;the functional connection network is based on every two nodes in AAL-90 templates.The temporal correlation between brain regions constitutes a 90*90 matrix,which we use as the adjacency matrix of the fully connected network,and the correlation between each two brain regions is used as the weight of the connection edge.In addition,we select 22 brain regions of the prefrontal-limbic loop in the AAL-90 template as nodes to form a 22*22 matrix to complete the construction of the prefrontallimbic loop structure and functional network.Calculate its global network properties and node network properties.The calculated network parameters include global network efficiency(E-loc),local network efficiency(E-loc),network clustering coefficient(Cp),network characteristic shortest path length(Characteristic Shortest Path)Length,Lp),standard network clustering coefficient(γ),standard network shortest path(λ),and smallworldness(σ).In this study,the statistical methods of analysis of variance and chi-square test were used to analyze the differences in general demographic data and clinical information among EBD,DBD,MBD and HC.The global network parameters,the prefrontal-limbic loop network parameters,and the node network parameters were analyzed by variance analysis and post-hoc test among the four groups.The global network parameters,prefrontal-limbic loop network parameters,and node network parameters with significant differences between groups were analyzed for partial correlation with the clinical symptoms and disease duration of the patients,controlling for age,gender,and mood state of groups.Experiment 2:A total of 166 subjects aged 18-50 were collected in this study,including 64 subjects with EBD,27 subjects with GHR-BD and 66 subjects with age-matched HC.GHR-BD are first-degree relatives of BD patients.Referring to the network construction steps of Experiment 1,the structure and function of the global network and the prefrontal-limbic loop network were respectively constructed.And we calculate the global network parameters,prefrontal-limbic loop network parameters and node topological properties.The calculating method is the same as experiment one.In this study,the statistical methods of analysis of variance and chi-square test were used to analyze the differences in general demographic data and clinical information among the three groups of EBD,GHR-BD and HC.For each network parameter,the analysis of variance among the three groups and the post-hoc test were carried out.Result:Result 1:(1)The global parameters of the structural network:the global network attributes E-Glo,E-Loc,and the global small-world attribute γ were statistically different in BD groups with different emotional states(P<0.05);E-glo was significantly lower than EBD and HC,DBD’s E-glo was significantly lower than MBD;DBD’s E-loc and γwere significantly lower than other groups.λ、σ were not statistically different.(2)Prefrontal-limbic loop network parameters and node attributes of the structural network:The Lp of the prefrontal-limbic loop structural network in each disease group was significantly lower than that of HC(P<0.05),node-level network analysis,each disease group There were group differences in the following brain regions of the prefrontallimbic circuit,bilateral orbital middle frontal gyrus,left anterior and paracingulate gyrus,left medial and paracingulate gyrus,left amygdala,right The Lp of the lateral orbital superior frontal gyrus was lower than that of the HC group(3)Correlation analysis of structural network attributes and clinical symptoms:structural network global smallworld attribute γ and depression(r=-0.269,p<0.05),anxiety level(r=-0.237,P<0.05)was significantly negatively correlated,and significantly positively correlated with the severity of mania(r=0.236,P<0.05);global network E-glo,E-loc,prefrontal circuit Lp and clinical scales There was no significant correlation between scores(P>0.05).(4)Global network parameters of functional network:the whole brain’s γ and σ of DBD were significantly lower than those of HC and EBD groups(P<0.05),and other network parameters had no significant difference between groups(P>0.05).(5)The network parameters and node attributes of the prefrontal-limbic loop of the functional network:there was no difference between the groups(P>0.05).(6)Correlation analysis of functional network attributes and clinical symptoms:There was no significant difference between functional network parameters and clinical scale scores with differences between groups(P>0.05).Results 2:(1)Global properties of structural network:There was no significant difference in global properties among EBD,GHR and HC groups.(2)Structural network prefrontal-limbic loop network parameters and node attributes:Compared with the HC group,the Lp of the prefrontal-limbic loop structural network in the GHR and EBD groups decreased significantly;The Lp of the lateral orbital middle frontal gyrus,right olfactory cortex,right medial and paracingulate gyrus were significantly lower than those of the HC group,and the Lp of GHR in the left hippocampus was lower than that of EBD and HC.(3)Functional network attributes:There were no significant differences in the functional network attributes of the three groups of brain global and prefrontal-limbic circuits.Conclusions:Conclusion 1:The global network of BD structure and function is unstable.The changes of the global network topological properties may be a state-related feature related to the BD mood phase,and the topological properties of the prefrontal-limbic loop exist in patients with each mood phase,which may be a qualitative trait-related feature related to the BD disease.Conclusion 2:The global network of structure and function of GHR is relatively stable.Altered structural network topology properties of prefrontal-limbic circuit are already present in genetically at-risk individuals without the disease and are reliable trait-related features of the disease.Bilateral orbital middle frontal gyrus,right olfactory cortex,right medial cingulate gyrus changes are risk factors for disease.The structural changes in the left hippocampus are protective factors for disease.