Altered Cortical Morphology And Topological Properties Of Structural Covariance Networks In Parkinson's Disease

BackgroundParkinson's disease(PD)is the second-most common neurodegenerative disease after Alzheimer's disease.The main pathological features of PD are the progressive loss of dopaminergic neurons in the substantia nigra and the appearance of intracytoplasmic inclusions composed of ?-synuclein aggregates known as Lewy bodies.As the disease progresses,Lewy body pathology gradually spreads from lower brainstem nuclei to cortical gray areas leading to structural abnormalitiesSo far,most structural MRI researches in PD mainly focused on cortical volume or surface cortical thickness(CT)analysis,investigations ofcortical surface morphology are scarce and need further clarification.The surface of the brain is composed of numerous closely arranged gyri and sulci.With the influence of aging or disease,the gyri decrease and shrink,and the sulci widen and deepen,resulting in changes of cortical surface morphology.Local fractal dimension(LFD),Local gyrification index(LGI)and Sulcal depth(SD)can help us to explore the altered characteristics of cortical surface morphology in PD patients from three perspectives of cortical complexity,gyrification pattern and sulcal morphology.Hence,we can have a more comprehensive understanding of the neuroanatomical changes of PD patients.In addition,studies have shown that there is a significant positive correlation between LFD and LGI.LFD is more sensitive than LGI to monitor the changes of brain morphology caused by aging,but it is not known which metric is more sensitive to detect the changes of cortical morphology in PD patients.SD is mainly affected by CT and the volume of white matter under gyrus,but which factor plays a leading role?The distribution characteristics of SD and CT changes in the same group of PD patients are also unclearAffected by common factors such as maturation,training or disease,brain regions with the same function will undergo common changes in structure,namely,structural covariance networks(SCN).Based on this hypothesis,the graph theory analysis method is widely used to study the changes of the SCN.Many studies have indicated that PD degeneration involves a battery of brain networks.Therefore,the study of SCN in PD patients can increase our understanding of PD pathology at the large-scale network level.Objective:In this study,several cortical morphological features were used to explore the change of cortical morphology and structural covariance networks based on cortical morphological features in PD patients.We will comprehensively understand the neuroanatomical changes of PD patients from regional brain regions and large-scale networks,which can improve the accuracy of clinical diagnosis of PD and actively carry out targeted early rehabilitation.Methods1:A total of 60 PD patients and 56 age-gender-and education matched healthy controls(HC)participated in this study.Clinical information and 3DT1WI structural MRI were collected.The LFD and LGI of the two groups were computed by Computational Anatomy Toolbox 12(CAT12),and vertex-wise two-sample t-tests were conducted to compare the differences of LFD and LGI between PD and HC.In addition,we also explore the correlation between LFD and LGI and clinical indicators.2:Using the same subjects and software,SD and CT were calculated and compared by the same methods as above.In addition,we also explore the correlation between SD and CT and clinical indicators.3:Using the GAT toolbox,SD-based SCN was constructed for PD patients and HC patients respectively,and network topology properties were compared between the two groups.The differences between global network parameters(clustering coefficient,path length,global efficiency and small-world index)and regional network parameters(clustering coefficient,node degree and node betweenness)are explored within a certain range of sparsity.Then,the area under the curve(AUC)within the sparsity range of all network properties is compared.Results1:Compared to HC,PD patients showed widespread LFD reductions mainly in the left pre-and postcentral cortex,the left superior frontal cortex,the left caudal middle frontal cortex,the bilaterally superior parietal cortex and the right superior temporal cortex(TFCE,FWE-corrected at P<0.05).These regions were mainly located in the left hemisphere.No regions showed increased LFD in PD patients compared to HC.For LGI,there was no significant difference between PD and HC.In PD patients,significant negative correlation was found between LFD of the left postcentral cortex and duration of illness(DOI)(Bonferroni correction,P<0.05).2:Compared to HC,PD patients showed widespread SD reductions mainly in the bilaterally insula,the superior temporal gyrus,the supramarginal,the left pars opercularis,the left precentral cortex,the left pars triangularis,the left superior parietal lobule(TFCE,FWE-corrected at P<0.05).No regions showed increased SD in PD patients compared to HC.For CT,PD patients showed Widespread SD reductions mainly in the left frontal lobe,the left inferior parietal and the posterior cingulate.Using multiple linear regression analysis,significant negative correlation was found between SD of right temporoparietal junction and H-Y stage,and significant negative correlation was found between SD of right dorsolateral prefrontal lobe and MMSE scores(none=0.001,cluster FWE-corrected at P<0.05).3:The SD-based structural covariance networks of PD and HC groups showed obvious small-world properties(Sigma>1).The AUC analysis of all global network parameters found no significant differences between the two groups.However,compared with HC,the Gamma(0.22)and Sigma(0.37 and 0.39)of PD patients were significantly higher.On the trend,Cp increased,and Lp,Lamda and Eglob decreased in PD patients.The comparison of regional network parameters found that clustering coefficient of the right cingul-post-ventral increased and regional betweenness of the right precentral-sup-part increased in PD patients(P<0.05,FDR correction).Conclusions1:Our results of widespread LFD reductions,but not LGI,indicate that LFD may provide a more sensitive diagnostic biomarker and encode specific information of PD.The significant negative correlation between LFD of the left postcentral cortex and DOI suggests that LFD may be a biomarker to monitor disease progression in PD.2:PD patients showed widespread SD reductions,which is correlated with some clinical indicators,indicating that SD may be a sensitive imaging indicator to evaluate and monitor the neuroanatomical changes of PD pathology.In addition,by comparing the distribution patterns of decreased CT in the same subjects,we found that the SD reductions may be more dependent on the retraction of gyral white matter.3:SD-based structural covariant network of PD patients showed altered global and regional properties.This indicates that suboptimal topological recombination,increased separation and decreased information integration ability occurred in the structural network of PD patients,which may lead to the neural mechanism of a series of clinical symptoms of PD.