基于MRI的独立成分分析和双回归评价肌萎缩侧索硬化症患者大尺度运动及认知网络退变的研究

Evaluation on large-scale motor and cognitive network degeneration in patients with amyotrophic lateral sclerosis by independent component analysis and dual regression based on MRI

目的基于MRI应用独立成分分析和双回归结合的方法,探索肌萎缩侧索硬化症(ALS)患者大尺度运动及认知相关网络功能变化,及其与运动、认知功能的关系。方法前瞻性收集2014年1月至2016年6月于西安交通大学第一附属医院就诊的46例ALS患者(ALS组),另招募性别、年龄匹配的正常对照40名(对照组),均完成运动及多维度认知功能评价[包括简明精神状态评价(MMSE)、蒙特利尔认知评价量表(MoCa)、语义流畅性(SVF)、音韵流畅性(PVF)、向前数字广度(DS_F)、向后数字广度(DS_B)、额叶功能评估(FAB)评分,威斯康星卡片分类测试(WCST)的分类准确数、分类错误数、持续性反应分类数目、持续性分类错误反应数目、非持续性错误分类数目评分及汉密尔顿抑郁量表(HAMD)、汉密尔顿焦虑量表(HAMA)评分]。采集所有被试的静息态MRI数据,以多变量解释线性优化独立成分分解进行独立成分分析,挑选出运动网络(SMN)、默认网络(DMN)和额顶叶控制网络(FPCN);应用双回归分析进行基于体素的组间网络差异对比。ALS组与对照组不同认知功能评价指标的差异采用多变量协方差分析;2组间脑网络差异的对比采用非参数置换检验,以族错误率(FWE)校正,以P<0.008为差异有统计学意义;采用偏相关和多元线性回归评价组间差异脑区功能连接变化与认知功能的关系。结果ALS组患者MMSE、MoCa、SVF、PVF、DS_B、分类准确数评分均较对照组低(P<0.05),而分类错误数、非持续性错误分类数目、HAMD和HAMA评分均较对照组高(P<0.05)。校正性别、年龄的前提下,ALS组与对照组的SMN脑区比较差异无统计学意义(FWE校正,P>0.008);与对照组比较,ALS组DMN的左侧腹内侧前额叶皮层(vmPFC)功能连接增强,FPCN的右侧前扣带回(ACC)、右侧后扣带回、左侧顶下小叶以及左侧颞下回局部脑区功能连接下降(FWE校正,P<0.008)。ALS患者vmPFC功能连接与MoCa评分(r=-0.565,P<0.001)、FAB评分(r=-0.373,P=0.015)及WCST的分类准确数(r=-0.478,P=0.002)、SVF评分(r=-0.458,P=0.002)呈负相关;与WCST分类错误数和HAMA评分呈正相关(r=0.416,P=0.007;r=0.388,P=0.011)。FPCN多个脑区功能连接下降,其中ACC的功能连接与DS_F(r=0.341,P=0.027)和MMSE评分呈正相关(r=0.351,P=0.023)。vmPFC功能连接增加对MoCa评分的影响占49.6%;而对FAB评分和HAMA评分的影响分别占35.2%和34.2%;对WCST的分类准确数、分类错误数以及SVF的影响不足30%;ACC功能连接下降对DS_F评分的影响占37.7%。结论ALS患者存在以额颞叶核心脑区为主的大尺度脑网络改变,DMN和FPCN网络变化与ALS患者临床认知改变密切相关。

Objective To evaluate changes of large-scale motor and cognition related networks′function in patients with amyotrophic lateral sclerosis(ALS)and their relationship with corresponding clinical symptoms using independent component analysis combined with dual regression.Methods Forty-six ALS patients(ALS group)who were treated in the First Affiliated Hospital of Xi′an Jiaotong University from January 2014 to June 2016 were prospectively collected,and 40 gender-and age-matched normal controls(control group)were recruited.All the participants completed the motor and multi-dimensional cognitive function evaluation[including Mini-mental State Examination(MMSE),Montreal Cognitive Assessment(MoCa),Semantic Fluency(SVF),Phonological Fluency(PVF),Digital Span Forward(DS_F),Digital Span backward(DS_B),frontal assessment battery(FAB),Wisconsin Card Sorting Test(WCST)for classification accuracy,classification error,persistent response classification,persistent error response classification,non-persistent error classification and Hamilton Depression Scale(HAMD),Hamilton Anxiety Scale(HAMA)].The resting-state MRI data of all subjects were collected,and independent component analysis was carried out with multivariate interpretation linear optimization independent component decomposition.Dual regression analysis was performed to compare network differences between groups based on voxel level in sensorimotor network(SMN),default mode network(DMN)and frontal-parietal control network(FPCN).Multivariate covariance analysis was used to evaluate the differences of different cognitive function indexes between ALS group and normal control group,the comparison of brain network differences between the two groups was performed by nonparametric permutation test,corrected by family-wise error(FWE),P<0.008 as the statistical threshold;partial correlation and multiple linear regression were used to evaluate the relationship between changes in functional connectivity of different brain regions and cognitive functions.Results The scores of MMSE,MoCa,SVF,PVF,DS_B,and classification accuracy were lower,while the number of error classifications,the non-persistent error classifications,HAMD and HAMA scores were higher in patients with ALS group than those in control group(P<0.05).After adjusting for gender and age,there was no significant difference in the SMN between ALS group and control group(FWE correction,P>0.008).Compared with control group,patients with ALS showed increased functional connectivity in the left ventromedial prefrontal cortex(vmPFC)of the DMN,and decreased functional connectivity in the right anterior cingulate gyrus(ACC),the right posterior cingulate gyrus,the left inferior parietal lobule and the left inferior temporal gyrus of the FPCN(FWE correction,P<0.008).Increased functional connectivity of the vmPFC in ALS patients was negatively correlated with MoCa score(r=-0.565,P<0.001),FAB score(r=-0.373,P=0.015)and the classification accuracy of WCST(r=-0.478,P=0.002),SVF(r=-0.458,P=0.002)scores,and was positively correlated with the number of error classifications and HAMA scores(r=0.416,P=0.007;r=0.388,P=0.011).Decreased functional connectivity were detected in multiple brain regions of FPCN,and the functional connectivity of the ACC was positively correlated with the DS_F(r=0.341,P=0.027)and MMSE(r=0.351,P=0.023).The effect of increased vmPFC functional connectivity accounted for 49.6%changes on MoCa score;35.2%and 34.2%for FAB and HAMA respectively.While the impact of increased functional connectivity in the vmPFC was less than 30%on classification accuracy,classification error of WCST and SVF.The reduced functional connectivity in the ACC accounted for 37.7%impact on the DS_F score.Conclusions Large-scale brain network changes are dominated by the frontotemporal core brain regions in ALS patients.DMN and FPCN network changes are closely related to the clinical cognitive performance of ALS patients.