| Objective:It's difficult to make accurate diagnosis of bipolar disorder(BD)in clinical practice because approximately half of BD present with a depressive episode as their first episode,misdiagnosing BD as major depressive disorder(MDD)may have great adverse impact,including inappropriate treatments,a high risk of conversion to mania,and poor clinical outcome.These make patients face a worse prognosis,more severe damage to social function and economic burden.Our study aims to use diffusion tensor imaging technology(DTI)and graph-based complex network methods to perform a 5 years clinical natural follow-up observation.In the acute and the remission phase,the differences in network properties in the prefrontal cortex among BD group,MDD group and healthy controls(HC)group were explored,looking for the stable and reliable markers to differentiate BD from UD patients.Meanwhile,relevant imaging markers and clinical risk factors are discussed in association with each other to facilitate early diagnosis and early warning.Method: Participants were enrolled from inpatient facilities at the Nanjing Brain Hospital,with a diagnosis of MDD or BD.Age-,gender-,and education-matched HC were recruited in the same period.All subjects underwent DTI scan at baseline,and the demographic data,clinical data of patients were also assessment.Patients were required to have at least 5 years of bi-annual follow-up care.Remission patients underwent the second scan during the 2-3 month.After the termination of the follow-up,regroup.The DTI data were processed with the functional magnetic resonance imaging brain software library(FSL),and Fiber assignment by continuous track in(FACT)was performed on white matter fibers based on anatomical auto labeling template(AAL),finally,all the tested white matter networks were constructed based on complex theoretical methods.The brain connectivity toolbox(BCT)were used to calculate the local and global efficiency of the brain structural networks.Discussion from the following methods:1.Acute phase data: After the termination of the follow-up,the subjects enrolled in the baseline were regrouped,depressive patients which converted into BD during the follow-up were classified as BD group,and patients still diagnosed with MDD during the 5-year follow-up.DTI data,demographic data,clinical risk factors and other data at the baseline were analyzed.From the perspective of global and local efficiency attributes of brain network,statistical analysis was conducted among the converted BD group,MDD and HC group.Meanwhile,node efficiency attribute values and clinical factors were used as predictive values to depict the receiver operating characteristic curve(ROC)and explore the specificity of early diagnosis of BD with imaging characteristics.2.Remission period data: After the end of follow-up,the subjects scanned in remission were regrouped,MDD patients converted into BD during the follow-up process and diagnosed with BD in the baseline were classified as the BD group.Patients not turned into BD during the process of follow-up classified as MDD group.The DTI data,demographic data in remission were analyzed.Based on the results of the first part positive difference,the frontal lobe was selected as the main region of interest.Global and local efficiency value were analyzed among the three groups,at the same time,node efficiency values and clinical factors were used as predictive values to depict the ROC and explore the specificity of diagnosis of BD in remission phase.Statistical analysis was carried out using Statistical Product and Service Solutions(SPSS25.0)version.Duration and severity among the groups were tested by independent two-sample T test.Gender among the three groups was tested by chisquare test.Age,years of education,global efficiency and local efficiency of nodes in the white matter network were analyzed by one-way Variance(ANOVA).The significance level of general population and clinical data was set as P<0.05,and the global and local brain efficiency values were corrected by multiple comparisons using the False discovery rate(FDR)method to obtain brain regions with statistical differences(P<0.05 after FDR correction).Results:1.Acute phase1.1 Demographic and clinical data: From September 2006 to July 2010,80 patients were diagnosed with depression,and 78 patients completed the 5-year,semiannual telephone or outpatient follow-up study.Among them,12 patients were diagnosed with BD and 64 patients were diagnosed with MDD.Finally,12 patients with BD,44 patients with MDD and 37 HCs were included in the statistical analysis.There was no significant difference in age,gender,years of education,HMAD17 among the groups(P >0.05),and there was significant difference in age of onset and frequency of recurrence between the two groups(P <0.05).1.2 Global and local efficiency among BD,MDD,and HC: No significant statistical difference was found in the global efficiency among the three groups,and the trend of difference was observed between the three groups.BD group < MDD group < HC group(F=6.814,P=0.003,without FDR corrected).Left inferior frontal Gyrus(LIFG)differs significantly in local efficiency(F=10.900,P=0.0004,FDR corrected).1.3 Local efficiency between the BD converters and non-converters and ROC curve: Compared with the non-converters group,the local efficiency of the nodes of the networks in BD converters decreased significantly in the left opercula part of the IFG(P=0.0002,survived critical FDR).The local attribute index was used to establish the ROC curve between groups.The recognition rate was 0.80,and the 95% confidence interval was(0.64,0.95).After integrating demographic characteristics and clinical risk factors,the discriminant rate was significantly increased to 0.96,and the 95% confidence interval was 95%.(0.91,1.00).2.Remission period2.1 Demographic and clinical data: From May 2012 to June 2014,120 patients with MDD and BD were followed up for 5 years with semi-annual telephone or outpatientobservation.60 healthy controls are matched at the same period.Considering the influence of the course on the structure of the brain,4 cases with a total duration of illness of 20 years and above,34 cases of 6 months or less were screened,40 patients with BD,40 patients with MDD,and 40 cases of healthy controls were enrolled in statistical analysis.There were no significant differences in age,gender,years of education and HAMD17 score among the groups(P >0.05).2.2 Global and local efficiency among BD,MDD,and HC: ANOVA analysis showed that there was no significant difference in the whole brain information transfer efficiency among the MDD,the BD and the control group,and the trend was relatively consistent.In terms of local efficiency properties,there is a significant difference in LIFG(F=11.991,P=0.000018,corrected by FDR).2.3 Local efficiency between the BD converters and non-converters and ROC curve: Between groups,there was no significant difference in the local efficiency of the frontal lobe between the BD and the UD groups,and the UD versus HC.Compared with the HC group,the local efficiency of the nodes in BD decreased significantly in the left IFG(P=0.00003,t=5.01,corrected by FDR).The local attribute index was used to establish the ROC curve between groups.The discriminant rate between the UD and BD groups is low,the area under the curve is only 0.59,and the 95% confidence interval is(0.47,0.72).After combined with the age of the onset,the discriminating rate is higher than before,and the area under the curve is 0.67.The 95% confidence interval is(0.55,0.79).Conclusion:1.Graph theory provides a series of quantitative indicators that can be used to describe the brain structure networks and functional networks.Our findings extend the previous cross-sectional studies that showed the potential impact of IFG in BD,highlighting the deficiencies in the IFG before converting to BD.To identify the reliability of these biomarkers,future study should address those limitations by increasing the sample sizeand measure from onset of BD during progression and in remission stage.2.There was no statistically significant difference in local efficiency of IFG between MDD and BD in the remission period,but the BD group still had significant difference compared with the healthy control group,suggesting that the IFG could serve as a potential biomarker in the identification of BD.The biomarkers were not significantly in the remission episode of depression,combined with clinical risk factors,the better recognition rate can be obtained. |
PDF Full Text Request