Multimodal MRI Study On The Dynamic Patterns Of Cerebral Structural And Functional Networks In Type 2 Diabetes Mellitus Patients

Type 2 diabetes mellitus(T2DM)seriously threatens human health and the quality of life.Cognitive impairment persists from the onset of T2 DM.It is reported that the prevalence of mild cognitive impairment(MCI)in patients with T2 DM was 45.0%.Besides,T2 DM promoted the conversion from MCI to dementia at a high rate of 8.7% per person per year.MCI can seriously affect the self-management of T2 DM patients,and induce treatment-related complications,which in turn will further promote the occurrence and development of MCI,forming a vicious circle.However,the mechanism of MCI in T2 DM is still not well clear.In particular,T2 DM will not develop MCI over a long period of time.Some of T2 DM patients recover from MCI to normal cognitive levels.Therefore,further study on the critical pathway and neural mechanism of MCI in T2 DM patients may provide new strategies and targets to prevent MCI and to block the deterioration of the vicious circle between T2 DM and MCI.The brain network is the structural and functional basis for organizing and integrating high-level cognitive information.Recently,the study of functional magnetic resonance image(f MRI)on cognitive impairment in T2 DM has evolved from the study of neural activity in isolated brain regions to the exploration of network-level information interaction and efficiency.Dynamic functional connectivity(d FC)has been shown to be associated with specific mental states,cognitive states,and neurodegeneration,which are more closely similar to the temporal resolution of neural activity.The characteristics of d FC have been verified to help differentiate between subjective cognitive decline and MCI,and are closely associated with the progression of MCI and the development of AD.Effective connectivity(EC),on the other hand,can reflect the influence of one brain area on another,thus reflecting the direction of information communication.EC abnormalities between the hippocampus and the default mode network affect memory in T2 DM patients.Thus,it implies that the abnormality of cerebral dynamic organization and information flow are not only related to insulin resistance,memory and learning ability,but also the advanced central characteristics of cognitive impairment in T2 DM.On the other hand,T2 DM is associated with changes in the process of food reward.Besides,the atrophy of nucleus accumbens(NAC)in T2 DM patients can be alleviated by insulin therapy.Substantial preclinical and clinical studies demonstrated that changes in neural activity,volume,and neuroplasticity of the nucleus accumbens are the interface between cognition,emotion,and motivational behavior abnormalities.However,the pattern of dynamic EC(d EC)in T2 DM patients with cognitive impairment remains unclear.Though most of the pathogenesis overlaps,whether the neuroimaging mechanisms of T2 DM with MCI are consistent with those of pure MCI is unknown.Moreover,it is still unclear whether abnormalities in the nucleus accumbens and the reward circuit in T2 DM contribute to cognitive impairment.Therefore,to identify these problems,four consecutive experiments were performed as the following:Part 1Study on the characteristics of static and dynamic network abnormalities in single-centered type 2 diabetic patients without mild cognitive impairment Aims: To explore the structural abnormalities of gray matter(GM)and white matter(WM),to screen the altered EC network,and to further explore the d EC characteristics of the corresponding network in single-centered T2 DM patients without MCI.Materials and Methods: A total of 28 single-centered(s)T2DM patients(s T2DM-CN)and30 gender-and age-matched healthy subjects(s HC-CN)underwent 3T f MRI scans.The Gray volume matter(GMV),the cortical thickness and the cortical complexity indexes from high resolution structural images were estimated.Effective connectivity(EC)network matrixes were constructed.The T2 DM cognitive impairment associated subnetwork was screened using network-based statistic(NBS)analysis.Then,The d EC matrixes of the subnetwork was constructed and analyzed by clustering analysis and connection analysis.The white matter(WM)connectivity probability(WMcp)network was constructed using diffusion tensor imaging(DTI),and the NBS was used to calculate the group different WMcp subnetworks.Besides,WM network topology indexes were estimated.Age,sex and body mass index(BMI)were used as covariates to analyze the partial correlation between image characteristics and scale scores.Results:(1)Compared with the s HC-CN group,the s T2DM-CN group had no significant changes in the clinical information and the cognitive assessment scores,except for some subitems of executive ability and memory.There was no significant difference in GMV,cortical thickness,cortical complexity features,WMcp and the topology features of WM network between the two groups(P > 0.05).(2)The static EC of s T2DM-CN group was significantly higher than that of s HC-CN group,and the left V5/MT area of occipital gyrus(LOc C＿L＿4＿2)was the core of the altered network.The network nodes mainly involved the visual network,the somatomotor network,the default mode network,the limbic network.The dynamic network fluctuated between state I of stronger EC and State II of weaker EC.The fractional windows(F)and the mean dwell time(MDT)of state I in s T2DM-CN were significantly lower than those in s HC-CN group(P < 0.001).The F and MDT of State II are the opposite(P < 0.001).There were no significant differences in number of state transfers(NT)of d EC between the two groups(P= 0.305).(3)In the s T2DM-CN group,the sum of d EC(sumd EC)was positively correlated with the number of short-delay free recalls.Sum of WMcp(sum WMcp)was negatively correlated with the number of animal names and the percentage of correct responses in the Montreal cognitive assessment scale.Conclusion: In the early stage of cognitive impairment,T2 DM did not show significant disruption in gray matter structure,WMcp network and whole brain topology.The visualmotor network centered on V5/MT may be the specific network of early cognitive impairment in T2 DM.Besides,its d EC clustering and connection strength were significantly correlated with memory.Part 2Study on the characteristics of cognitive impairment related dynamic functional and white matter network abnormalities in multi-centered type 2 diabetic patients Aims: To explore the d EC characteristics of visual-motor network of T2 DM patients with/without MCI,and to explore the relationship between the abnormal structural and functional network characteristics.Materials and Methods: Multi-centered age,sex and education matched twenty-eight T2 DM patients with normal cognition(m T2DM-CN),31 T2 DM patients with MCI(m T2DM-MCI)and 28 healthy subjects(m HC-CN)were recruited from Alzheimer’s disease imaging database(ADNI)3.We obtained high-resolution structural images,blood oxygen level dependent(BOLD)images,DTI and the cognitive scales reflecting the whole cognitive state,memory,executive ability and language fluency.The method of image analysis processing was the same as that in Part 1.Bonferroni method was used for multiple comparisons of the F,MDT,d EC and WMcp.Significant levels were set at P< 0.05.Results:(1)Compared with m HC-CN group,m T2DM-CN group had no significant change in all cognitive performance,while m T2DM-MCI group had significant difference in overall cognitive performance and subitems of multiple cognitive dimensions(P < 0.05).(2)The visual-motor network fluctuated between stronger EC State I and weaker EC State II.The F and the MDT of state I in m T2DM-MCI were significantly higher than those in m HC-CN and m T2DM-CN groups(P < 0.001).The F and MDT of State II are the opposite(P < 0.001).No significant difference in F value and MDT value was found between m HC-CN and m T2DM-CN groups(P > 0.05).(3)Compared with m HC-CN group,GMV in m T2DM-MCI group decreased significantly in right middle temporal gyrus of temporal pole,bilateral parahippocampal gyrus,left rectus gyrus,left caudate nucleus,left lingual gyrus and left middle frontal gyrus(P<0.05).Although there were no significant differences in the topology of WMcp networks among the three groups,compared with m HC-CN Group,WMcp of altered WM networks in m T2DM-CN and m T2DM-MCI groups were significantly higher involving all the eight functional networks with more altered WMcp in the latter group.Hubs in T2DM-related functional network were also found in the increased WMcp matrix,especially in m T2 DMMCI patients.(4)In the m T2DM-MCI group,sumd EC was negatively correlated with the use of cues in the logical memory test.sum WMcp was positively correlated with the number of animal classification in category fluency test.The sensitivity,specificity and accuracy of the linear model established based on sumd EC to distinguish m T2DM-MCI group and m HC-CN group were 0.903,1 and 0949,respectively Conclusion: In T2 DM patients with MCI,performance of multiply cognitive dimensions decreased significantly.Gray matter atrophy,the increased MDT in strong EC state and increased WMcp were observed.The enhancement of functional and structural connectivity may be the phenotype of compensation mechanism.The overlap of functional and structural abnormal nodes suggests that functional and structural networks also have synergistic effects on the mechanism of cognitive impairment.sumd EC was an effective diagnostic factor to differentiate the m T2DM-MCI group from m HC-CN group.Part 3A comparative study of the abnormal pattern of dynamic function and white matter network in mild cognitive impairment patients with or without type 2 diabetic Aims: Multi-modal f MRI was used to explore whether the d EC and WM network altered pattern in T2 DM patients with MCI are consistent with those in pure MCI patients in order to further analyze the mechanism of cognitive impairment in T2 DM.Materials and Methods: Multi-centered age,sex and education matched 31 pure MCI(mp MCI)patients,31 m T2DM-MCI and 28 m HC-CN were recruited from ADNI3.We obtained high-resolution structural images,BOLD images,DTI and the cognitive scales reflecting the whole cognitive state,memory,executive ability and language fluency.The method of clinical information and image analysis processing were the same as that in Part2.Results:(1)Compared with m HC-CN group,mp MCI group had significantly worse overall cognitive and memory performance.Compared with m T2DM-MCI,the performance of executive function and language fluency were significantly better in mp MCI group.Cortical thickness in mp MCI group was significantly decreased mainly in parietal lobe and temporal lobe(P<0.05).(2)The visual-motor network fluctuated between stronger EC State I and weaker EC State II.The F and the MDT of state I in mp MCI were significantly lower than those in m T2DM-MCI group(P<0.001).The F and MDT of State II are the opposite(P<0.001).No significant difference in F value and MDT value was found between m HC-CN and mp MCI groups(P>0.05).(3)Although there were no significant differences in the topology of WMcp networks among the three groups,we found increased WMcp involving the somatomotor,salient ventral attention,limbic,control,default and subcortical network and decreased WMcp involving visual,dorsal attention,salient,default and subcortical network in mp MCI patients compared to m HC-CN.WMcp of altered WM networks in mp MCI was significantly lower involving all the eight functional networks.Hubs in T2DM-related functional network were also found in the decreased WMcp matrix.(4)In mp MCI group,both the F value of the state I in backward network was significantly positively correlated with the GDS score.WMcp was negatively correlated with total insertion number in RAVLT.Conclusion: The dynamic network pattern of visual-motor pathway and WMcp network connection in m T2DM-MCI group were significantly different from those in mp MCI group.The increased connectivity of functional and structural networks may be a compensatory mechanism specific for the cognitive impairment process in T2 DM.Part 4Study on the pattern of nucleus accumbens-associated dynamic functional network abnormalities in type 2 diabetes mellitus Aims: To investigate the characteristics and role of the dynamic NAC-whole brain interactive network and classical reward circuit in the cognitive impairment process of T2 DM.Materials and Methods: The recruited subjects were the same as Part 1 and Part 2.The ztransferred low-frequency amplitudes(z ALFF),regional homogeneity(z Re Ho)and voxelmirrored homotopic connectivity(z VMHC)of NAC from AAL3,BNA and HOA templates were extracted and compared.Ventral tegmental area,NAC,amygdala,prefrontal regions and hippocampus were used as regions of interests(ROI)to construct reward networks.Besides,NAC was used as the ROI to construct the mutual network between NAC and other brain regions.The d EC clustering features of the two networks and their correlation with the cognitive performance were then explored.The remaining analysis and statistical methods for single-centerd and multi-centered data were the same as Part 1 and 2,respectively.Results:(1)In single-centered subjects,z ALFF of NAC increased significantly,while z Re Ho and z VMHC decreased significantly in s T2DM-CN group compared with s HC-CN group(P<0.05).The d EC of the two networks also fluctuated between stronger EC State I and weaker EC State II.The F and the MDT of state I in s T2DM-CN were significantly lower than those in s HC-CN group(P<0.05).The F and MDT of State II are the opposite(P<0.05).(2)In multi-centered subjects,z ALFF of NAC decreased significantly,while z Re Ho and z VMHC increased significantly in m T2DM-MCI group compared with m T2DM-CN and m HC-CN groups(P<0.05).The d EC of the two networks also fluctuated between stronger EC state I and weaker EC state II(P<0.001).The F and the MDT of state I in m T2DM-MCI were significantly higher than those in m T2DM-CN and m HC-CN groups(P<0.001).The F and MDT of State II are the opposite(P<0.001).No significant difference in F value and MDT value was found between m HC-CN and m T2DM-CN groups(P>0.05).(3)In s T2DM-CN group,the F value of the state I in backward reward network was positively correlated with congruent correct responses in Stroop color word test.In m T2DM-MCI group,the F value of the state I in backward reward network was negatively correlated with completion time of TMT A.Conclusion: The neural activity of NAC was useful for differentiating T2 DM patients with cognitive impairment from HC.NAC may play an important role in the mechanism of cognitive impairment in T2 DM.The reward circuit may be the susceptible pathway.