The Interactions Of Cognitive-related Brain Network Subsystems In Alzheimer’s Disease:A Resting-state FMRI Study

Background and purposeAlzheimer’s disease(AD)is a progressive neurodegenerative disease characterized by early memory impairment,although disturbances in other cognitive functions also occur as the disease progresses.Studies have consistently suggested that the default mode network(DMN)is earliest and most prominently disrupted in the courses of AD.Recent evidence has revealed that DMN is not united entity,but consists of at least three distinct subsystems: medial temporal lobe(MTL)subsystem,dorsomedial prefrontal cortex(DMPFC)subsystem and Core subsystem.Now,little is known about the alterations of DMN subsystems.Additionally,recent studies have indicated that the DMN doesn’t work alone,but interacts with dorsal attention networks(DAN)and frontoparietal control networks(FPCN)to support complex cognitive functions.As the disease progresses,damage occurs in the more cognitive domains of patients with AD,therefore,the interactions among DMN,FPCN,and DAN may be changed.FPCN can be divided into two parts: FPCN-A and FPCN-B subsystems.However,little is known regarding the alterations in the pattern of interactions among the DMN subsystems,FPCN subsystems and DAN in AD patients and relationship between these alterations and cognitive impairment;The underlying causes of these changes are also unclear.Therefore,aiming at the above problems,this study aims to: 1.Explore the alterations of different DMN subsystems in AD patients.2.Explore the changes in information flow strength(IFS)of nodes belonging to DMN subsystems,FPCN subsystems,and DAN in patients with AD;Explore the correlations between changes of intra-network nodal IFS and changes of inter-network nodal IFS in AD patients;Explore the changes of inter-subsystem IFS in AD patients;Explore changes in correlations between feed-forward IFS and feedback IFS of inter-subsystems in AD patients;Probe vulnerable subsystems in AD patients;Explore the correlation between changes of cognitive-related brain network subsysytems and impaired cognitive function in AD patients;Verify the repeatability of between-groups differences of inter-subsystem IFS.3.Explore alterations in temporal co-evolution of inter-subsystem dynamic effective connectivity in AD patients.4.Explore the changes in the regulatory effects of the FPCN subsystem on the DMN subsystems,interactions between DMN subsystems and DAN.MethodIn this study,resting-state functional magnetic resonance Imaging(rs-f MRI)data were collected from 42 AD patients and 45 normal elderly people.1.Regions of interest(ROIs),within the DMN subsystems were selected,and FC between the brain regions were measured to construct FC networks.We calculated intra-subsystem and inter-subsystem functional connectivity strength(FCS)and functional connectivity number(FCN),and conducted group-difference analysis to explore the changes of the DMN subsystems;between-group differences analysis in inter-nodal FC was used to identify the pairwise connections that caused the subsystems’ alterations;Degree centrality analysis was performed to detect the nodes contributing most to the changes of the DMN subsystems.2.This study selected ROIs included in DMN subsystems,FPCN subsystems and DAN,and then used GCA to construct the effective connectivity network for each subject;Support vector machine(SVM)classification algorithm was used to screen consensus effective connections that can stably distinguish AD from the normal elderly,for each subject’s effective connectivity network,only the effective connections corresponding to consensus effective connections were retained for subsequent network analysis.The steps for analyzing the effective connectivity networks are as follows: nodal IFS were calculated for each node,and group-difference analysis was conducted to explore changes of nodal IFS;Correlation analyses of intra-network nodal IFS with group-difference and inter-network nodal IFS was performed to explore the relationship between changes within networks and changes of inter-network;Calculated the inter-subsystem IFS,and conducted group-difference analysis to investigate specific inter-subsystem interactions changes;Performed correlation analyses of inter-subsystem’s forward IFS and backward IFS,and conducted between-group comparison to explore the changes in the coordination of inter-subsystem interactions;This study applied a SVM linear classifier,with the nodal IFS or subsystem’s IFS that had significant group difference and higher absolute T statistic values as classification features,to identify the vulnerable subsystems;Nodal IFS or subsystem’s IFS with highest classification accuracy of the SVM classifier was used to make correlations with MMSE scores to explore the relationships between changes of subsystems and cognitive function;Based on the validation data,the between-group differences of inter-subsystem IFS between the AD group and the NCs group were analyzed,and the statistical results of the experimental data and the validation data were compared to verify the reproducibility of the relevant results of this study.3.Based on the inter-subsystem IFS with between-group differences obtained above,we further analyze correlations between variations in the IFS from the FPCN-A subsystem to the DMN subsystems across time and variations in IFS from the DMN subsystems to DAN(temporal co-evolution of inter-subsystem dynamic effective connectivity),and the between-group comparison was carried out to identify which subsystems play pivotal roles in the alterations of the temporal co-evolution of inter-subsystem dynamic effective connectivity.4.This study used partial correlation analysis in combination with pearson correlation analysis to quantify regulatory effect of the FPCN-A subsystem on DMN subsystems,and the interactions between the DMN subsystems and DAN,and between-group comparison was performed to explore the changes of the FPCN-A regulating effect in patients with AD.Results1.Based on DMN subsystems constructed by functional connections analysis,(1)Intra-subsystem FC analysis revealed decreased functional connectivity strength(FCS)and functional connectivity number(FCN)within the MTL subsystem in AD patients in comparison with the normal elderly,however,the FCS/FCN within the DMPFC and Core subsystems had no group differences,inter-subsystem FC analysis found diminished FCS and FCN between the MTL subsystem and the Core subsystem,decreased FCN between the Core subsystem and DMPFC subsystem in AD patients in comparison with the normal elderly;(2)Inter-nodal FCS analysis revealed decreased FCS between the posterior inferior parietal lobule(p IPL)and the hippocampal formation(HF),and parahippocampal cortex(PHC),and ventral MPFC(v MPFC)within MTL subsystem,HF and/or PHC of MTL subsystem showed the decreased FCS with the dorsomedial prefrontal cortex(d MPFC),temporal parietal junction(TPJ)of the DMPFC subsystem and the posterior cingulate cortex(PCC)of the Core subsystem,decreased FCS also was found between the PCC of the Core subsystem and the lateral temporal cortex(LTC)of DMPFC subsystem in AD patients in comparison with the normal elderly;(3)Degree centrality analysis revealed that compared with the normal elderly,the PCC,HF and PHC showed decreased degree centrality in AD patients.2.Based on the effective connectivity analysis across DMN Subsystems,FPCN Subsystems,and DAN,this study revealed that compared with the normal elderly,(1)most of nodes of DMN subsystems showed decreased intra-network nodal IFS and diminished inter-networks’ nodal IFS,and changes of intra-DMN nodal IFS was significantly related to the changes of inter-network’s nodal IFS in AD patients;(2)The decreased forward IFS and backward IFS were found among the DMN subsystems,the forward IFS and backward IFS between the DMN subsystems and FPCN-A subsystem also decreased,the IFS initiated from the MTL and FPCN-A subsystems to the FPCN-B subsystem decreased,IFS from the Core and DMPFC subsystems to the DAN decreased in AD patients,and the correlation values of the forward IFS and backward IFS between the DMN subsystems and FPCN-A subsystem diminished in AD patients;(3)With nodal IFS of the left PCC,right PCC of the Core subsystem and left ventral PCC(v PCC)of the MTL subsystem(including intra-in+out IFS of the PCC.L,intra+inter-out IFS of the PCC.R,and intra-in IFS of the v PCC.L)as classification features,the SVM classifier can distinguish AD Patients from NCs more efficiently,and these nodal IFS were positively correlated with MMSE scores in AD patients;(4)The results of between-group differences in inter-subsystem IFS based on validation data were consistent with the results of this study.3.Variations in the IFS from the FPCN-A subsystem to the Core subsystem/DMPFC subsystem across time were significantly correlated with variations in IFS from the Core subsystem /DMPFC subsystem to DAN in NCs;Temporal co-evolution between the IFS from the FPCN-A subsystem to the Core/DMPFC subsystem and IFS from the Core/DMPFC subsystem to DAN were decreased in AD patients in comparison with NCs.4.The FPCN-A subsystem had regulating effect on Core subsystem,MTL subsystem,DMPFC subsystem and the interactions between DMN subsystems and DAN in NC group;Compared with the NCs group,the modulation effect of the FPCN-A subsystem on the Core subsystem increased in patients with AD.Conclusion1.The DMN subsystems of AD patients were impaired to varying degrees,among which the memory-related MTL subsystem was significantly impaired.2.The damage of DMN subsystems in AD patients led to the damage and discoordination of its interactions with FPCN subsystems and DAN;The damage within the DMN and the damage between the DMN subsystems and the FPCN subsystems and DAN were associated with cognitive decline.The results of between-group differences in the brain network inter-subsystems IFS in AD patients and NCs groups were reproducible.3.DMN subsystems play a key role in the disturbance of temporal co-evolution of brain network inter-subsystems dynamic effective connectivity.4.The damage of DMN subsystem resulted in the increase of regulatory load of FPCN-A subsystem.