【作者】 罗丹；

【导师】 李静；

【作者基本信息】 四川大学 ， 精神病与精神卫生学（专业学位）， 2023， 博士

【摘要】 目的:甲基苯丙胺（Methamphetamine,METH）滥用是全球范围内严重的公共卫生问题,METH的易成瘾性、高复吸率和难治性造成了极重的疾病负担。但截止目前,METH成瘾的作用机制尚不完全清楚。因此,阐明METH成瘾的作用机制,找到干预靶点,降低复吸率,提高治愈率是当前社会亟待解决的问题。本研究拟探讨甲基苯丙胺使用障碍（Methamphetamine use disorder,MUD）者冲动特质的特征及其在MUD发生和发展中的作用机制,通过核磁共振成像全面了解MUD者功能脑网络连接特征,利用第二代全外显子测序技术探索与MUD相关的罕见变异遗传生物学标记,最后融合功能脑影像和遗传变异探索MUD的脑功能改变潜在的遗传成分和作用机制。材料和方法:本研究是在四川大学华西医院心理卫生中心开展的病例对照研究。从四川省成都市强制隔离戒毒所和四川大学华西医院心理卫生中心门诊招募和筛选MUD者110例,同时在社区招募人口学信息匹配的健康对照（Health control,HC）55例。所有受试者需要进行临床特征评估、脑影像数据收集和外周血采集。临床特征评估包括一般情况量表、毒品使用模式、冲动量表和视觉模拟量表。脑影像数据采集在华西医院成办分院放射科磁共振中心完成。采用3.0T MR成像系统对所有受试者的头颅结构进行静息态功能磁共振（rs-fMRI）和高分辨三维T1加权图像（3D-T1WI）扫描。外周血采集由护士完成,使用含EDTA抗凝剂的5ml真空采血管收集血样本约3ml,在4小时内将血样离心后将血细胞和血浆分装保存于-80℃的冰箱。样本统一送至诺禾致源公司进行第二代高通量全外显子测序。完成数据采集后使用下述方法对数据进行了分析:1.在冲动特质研究中,使用独立样本t检验比较了HC组和MUD组之间冲动量表得分差异;偏相关分析探索了MUD者冲动特质与毒品使用情况之间的关系;利用中介分析调查了冲动性在毒品使用和MUD严重程度的作用。2.在功能脑网络研究方面,使用图论分析和基于网络的连边分析比较HC组和MUD组基于图谱的稳态脑网络拓扑属性和功能连接（Functional connectivity,FC）强度差异,并采用偏相关分析调查了拓扑属性与毒品使用和冲动特质的关系。使用基于数据驱动的独立成分分析构建大尺度脑网络,采用单因素协方差分析比较HC组和MUD组大尺度功能脑网络之间连接强度差异,并利用偏相关分析调查MUD组脑功能网络连接（Functional network connectivity,FNC）与毒品使用和冲动特质的关系。使用隐马尔可夫模型（Hidden Markov Model,HMM）构建动态脑网络,利用双样本t检验比较HC组和MUD组之间状态窗口占比（Fractional occupancy,FO）、状态平均居留时间（Mean dwell time,MDT）、状态切换率（Switching rate,SR）的差异,进一步采用偏相关分析调查动态衡量指标与毒品使用和冲动特质的关系。使用滑动时间窗法提取大尺度脑网络时间序列构建动态脑网络,采用双样本t检验比较HC组和MUD组各状态FNC连接强度及动态衡量指标（FO、MDT、SR）差异,使用偏相关分析调查各状态动态指标与毒品使用和冲动特质的关系。3.在遗传研究中,对测序数据进行质控、变异检测和注释后,纳入外显子和剪接区域的非同义突变,参考次等位基因频率（Minor allele frequency,MAF）过滤变异。在不同MAF下进行基于基因的多变量折叠（Multivariate and Collapsing,CMC）负荷分析（Burden test）,并对MAF<0.01潜在有害性类别分析中p<0.05的基因进行GO/KEGG（Gene Ontology/Kyoto Encyclopedia of Genes and Genomes）富集分析,以探索与MUD易感性有关的基因和通路。4.最后,使用平行独立成分分析（Parallel independent component analysis,p ICA）对功能脑影像数据和遗传变异数据进行融合分析,探索MUD脑功能改变的遗传基础,进一步以遗传变异作为自变量,脑功能改变作为中介变量,MUD诊断作为结局变量进行中介分析以探寻基因-脑-MUD疾病之间的关系。结果:本研究共纳入男性健康对照55例,男性甲基苯丙胺使用障碍者110例。在临床特征评估部分,HC组2例和MUD组5例未完成评估,因此此部分纳入健康对照53例,甲基苯丙胺使用障碍105例。在脑影像学数据分析部分,HC组2例和MUD组6例因无法坚持或磁共振数据不全而被排除,在数据预处理和质控后,HC组2例和MUD组2例头动不合格而被排除,因此此部分纳入健康对照51名,甲基苯丙胺使用障碍102例。在全外显子测序和分析部分,仅有145例完成了外周血采集,MUD组3例因血液标本不合格而被排除,因此此部分纳入健康对照51例,甲基苯丙胺使用障碍91例。具体研究结果如下:1.在冲动特质研究方面,比较HC组和MUD组冲动量表得分发现,MUD者在运动冲动、认知冲动、无计划冲动和冲动总分得分均更高（p<0.05）;控制人口学信息、吸烟和饮酒情况后进行偏相关分析发现,冲动特质评分与多个毒品使用相关变量及MUD严重程度显著相关（p<0.05）;中介分析发现,冲动总分在开始吸毒年龄与MUD严重程度之间以及在渴求和MUD严重程度之间分别存在完全中介作用（中介效应占比:47.99%;95%CI:-0.27,-0.07）和部分中介作用（中介效应占比:19.47%;95%CI:0.02,0.17）;无计划冲动在渴求和MUD严重程度之间存在部分中介作用（中介效应占比:12.81%;95%CI:0.01,0.12）。2.功能脑网络分析结果如下:2.1稳态功能连接分析研究结果显示MUD者和HC都具有小世界属性（σ>1）,但相较于HC,MUD者小世界属性降低(λ_MUD>λ_HC,p<0.05),其他拓扑属性也具有降低趋势,但无统计学差异（p>0.05）;节点拓扑属性的组间比较发现MUD组小脑蚓部的节点集聚系数（p<0.001,通过Bonferroni校正）和节点局部效率增加（p<0.001,通过Bonferroni校正）;在MUD组,控制协变量进行相关分析发现,小世界属性与甲基苯丙胺（Methamphetamine,METH）使用频率、心理渴求、冲动特质显著相关（p<0.05）,全局效率与运动冲动呈正相关（p<0.05）;左侧角回、颞上回、豆状苍白球的拓扑属性与METH使用剂量显著相关（p<0.001,通过Bonferroni校正）;双侧楔前叶、右侧额中回节点拓扑属性与运动冲动显著相关（p<0.001,通过Bonferroni校正）;右侧海马旁回节点拓扑属性与认知冲动显著相关（p<0.001,通过Bonferroni校正）。稳态FC的组间比较分析发现,MUD者两个子网络功能连接强度减弱,一个子网络功能连接强度增强（p<0.0005,NBS校正,置换5000次）。2.2在稳态功能网络连接（Functional network connectivity,FNC）研究中,采用独立成分分析得到36个独立成分,最后确定19个典型的独立成分,分为8个脑网络。通过比较HC和MUD组间脑网络功能连接差异发现,相较于HC,MUD者突显网络（Salience network,SAN）内功能连接增强,SAN与小脑网络（Cerebellar network,CEN）、背侧注意网络（Dorsal attention network,DAN）、视觉网络（Visual network,VN）间的功能连接减弱;默认模式网络（Default mode network,DMN）与VN、听觉网络（Auditory network,AN）间功能连接减弱;额顶网络（Frontoparietal network,FPN）与DAN之间功能连接减弱（所有p<0.05,通过FDR校正）。在MUD组中,与临床症状进行偏相关分析发现SAN与感觉运动网络（Sensorimotor network,SMN）间的功能连接与METH使用时长呈显著负相关（p=0.002,通过FDR校正）;CEN与DAN间的功能连接与METH使用时长呈显著正相关（p<0.001,通过FDR校正）。2.3在动态功能连接（Dynamic functional connectivity,dFC）部分,根据最小自由能,得到12个动态脑网络状态。比较HC和MUD组动态指标发现,状态3的FO和MDT存在组间差异（p<0.05,但未通过Bonferroni校正）。控制协变量进行偏相关分析发现,状态3的MDT与开始吸毒年龄呈负相关（p=0.004,通过Bonferroni校正）,状态9的MDT与METH使用时长显著相关（p=0.001,通过Bonferroni校正）,SR与METH使用剂量显著相关（p=0.025,通过Bonferroni校正）。2.4在动态功能网络连接（Dynamic functional network connectivity,dFNC）研究中,通过K-means聚类得到5个动态脑网络状态,比较动态功能脑网络连接差异发现,相较于HC,MUD者状态3中VN内、FPN内、FPN和DMN间功能连接增强（p<0.05,通过FDR校正）,状态4中DMN和FPN间、DMN和CEN间功能连接增强（p<0.05,通过FDR校正）;状态2的MDT增加（p<0.01,通过FDR校正）,状态2和状态5的状态出现频率减少（p<0.01,通过FDR校正）。3.通过CMC负荷分析发现SPATA21罕见变异位点可能与MUD遗传风险高负荷有关;GO/KEGG富集分析显示,与MUD相关的罕见变异位点基因参与了ABC型异种活性转运蛋白（ABCC6、ABCC1、ABCB11）、基转移酶复合物、组蛋白甲基转移酶（SETD1A、KDM5B、MGA、PHF20、WDR5B、RIOK1）和ABC转运蛋白通路（ABCC6、ABCC1、ABCB11、ABCA12）的表达。4.通过对功能脑影像与遗传变异的融合分析,我们发现一对显著相关的功能脑影像遗传变异成分对（fMRI25-VAR4,r=-0.525,p<0.001,通过Bonferroni校正）,fMRI25独立成分主要是左侧颞叶和枕叶等脑区的功能改变;对VAR4前5%权重的基因位点进行富集分析发现其参与轴丝动力蛋白（DNAH1、DNAH14）的表达,与突触可塑性和神经炎症调节过程相关;fMRI25的负荷系数与无计划冲动、METH使用频率呈正相关（p<0.05）;VAR4的负荷系数与认知冲动呈负相关（p<0.05）;中介分析发现独立成分fMRI25的负荷系数部分介导了VAR4的负荷系数对MUD疾病的影响。结论:1.MUD者具有更高的冲动性,其在物质滥用的发生、发展中发挥重要作用,高冲动性可作为预测成瘾预后的潜在行为学标记。2.MUD者大脑整合和分离信息的能力下降以及额顶叶之间功能连接强度的改变为METH对大脑认知控制能力造成损害提供神经生物学依据。小脑拓扑属性改变、SAN与CEN之间功能连接强度的改变,证明小脑参与成瘾的过程。ECN与DMN间功能连接强度改变可能是METH成瘾的动态脑网络特征,可能可以作为MUD戒毒疗效的预测指标。3.我们揭示了SPATA21、SETD1A、KDM6B及ABC基因家族位点变异可能与MUD的易感性有关,其机制可能与免疫调节紊乱、不良的神经发育、突触可塑性缺陷有关,从罕见变异角度为MUD的遗传病因学机制提供了新见解。4.通过功能脑影像与遗传变异融合分析,我们揭示了MUD中脑功能改变的潜在遗传成分,发现MUD者脑功能改变可能与不良的神经发育有关。另外,被识别的遗传成分可能通过脑功能活动改变作用于MUD,脑功能活动作为中间表型,在对MUD遗传病因学机制的解读中发挥重要作用。综上,本研究分别利用基于图谱的硬分割和基于数据驱动的软分割方式分别构建静态脑网络和动态脑网络,从多个维度对MUD者脑网络功能连接特征进行了全面的分析,发现了MUD特征性影像学生物标记。同时,本研究还利用第二代全外显子测序技术,发现了MUD致病候选基因及作用机制,进一步融合功能脑影像和遗传变异两个模态,考察了MUD患者脑功能改变潜在的遗传机制。这为探索MUD的神经脑影像和遗传病因学机制奠定了基础,进一步为MUD的早期识别、干预、防复吸提供理论依据。更多还原

【Abstract】 Objective:Methamphetamine（METH）abuse is a serious public health problem worldwide.Its high addictive potential,high relapse rates,and difficulty in treatment result in a heavy disease burden.However,to date,the mechanisms underlying METH addiction are not fully understood.Therefore,elucidating the mechanisms ofMETH addiction,identifying intervention targets,reducing relapse rates,and improving cure rates are urgent issues that need to be addressed in society.This study aims to explore the characteristics of impulsivity traits and its mechanism in the occurrence and development of drug use in individuals with methamphetamine use disorder（MUD）.Nuclear magnetic resonance imaging（MRI）technology and the second-generation whole exon sequencing（WES）technology were respectively used to explore the characteristics of functional brain network and the genetic biological markers of rare variations related with MUD participants.Finally,functional brain imaging and genetic variations were fused to investigate the potential genetic components and mechanism of action related with the change of brain function in participants with MUD.Materials and Methods:This study was a case-control study and conducted in Mental Health Center,West China Hospital of Sichuan University.110 participants were recruited and screened from Compulsory Detoxification Centers in Chengdu,Sichuan Province,and the outpatient department of the Mental Health Center,West China Hospital of Sichuan University.Meanwhile,55 healthy controls（HC）matched for demographic information were recruited from the community.Clinical characteristics,brain imaging data and peripheral blood were collected in all participants.Clinical characteristics were assessed by the general scale,drug use pattern,Barratt Impulsivity Scale Chinese version 11（BIS-11）and visual analogue scale（VAS）.Resting state functional magnetic resonance imaging（rs-MRI）and T1 weighted structural magnetic resonance imaging（s MRI）scans were performed with a 3.0T magnetic resonance scanner at the Magnetic Resonance Center of Radiology Department,Hospital of Chengdu Office of People’s Government of Tibetan Autonomous Region（Hospital.C.T.）.Approximately 3 m L of peripheral venous blood were collected by a sterile venipuncture using a sterile EDTA vacutainer,which was completed by nurse.After collection,the blood samples were centrifuged within 4 hours,and the blood cells and plasma were stored in a refrigerator at-80℃.The samples were sent to Novogene company for second-generation high-throughput whole exon sequencing.After data collection,the methods to analysis the data were as follows:1.In the study of trait impulsivity,independent samples t-tests were applied to compare the difference of BIS-11 scores between HC group and MUD group.Among participants with MUD,partial correlation analysis was applied between impulsivity traits and variables of drug use to investigate their relations,and mediation analysis was performed in impulsivity,drug use pattern and MUD severity.2.In the study of functional brain network,graph theoretic analysis and network-based statistics were used to explore the differences of topological attributes and functional connectivity strength of static brain network based on the brain atlas between HC and MUD groups,and partial correlation analysis was performed to investigate the relations of topological attributes with drug use model and impulsivity.The data-driven independent component analysis was used to construct large-scale brain networks.Controlling age,one-way analysis of covariance（ANCOVA）was used to analyze the differences of large-scale functional brain network connectivity strength between HC and MUD groups,and partial correlation analysis was used to test the relations of functional network connectivity（FNC）strength with drug use patterns and impulsivity in MUD patients.Dynamic brain networks were constructed using Hidden Markov Model（HMM）.The differences in fractional occupancy（FO）,mean dwell time（MDT）,switching rate（SR）between HC and MUD groups were tested with two-sample t-test,and the correlations of dynamic indicators with drug use related variables and impulsivity traits were analyzed with partial correlation analysis.Sliding time window method was performed to extract large-scale brain network time series and then to construct dynamic brain networks.Then,the differences of dynamic functional network connectivity strength and parameters of time series（FO,MDT,SR）in different states were identified using two-sample t-test,and the relations of dynamic indexes of different states with drug use and impulsivity were explored using partial correlation analysis.3.In the genetic study,after quality control,identifying variants and annotation of sequencing data,nonsynonymous mutations in the exon and splicing region were retained.According to minor allele frequency（MAF）,rare variants ware screened,and then Multivariate and Collapsing（CMC）burden test and GO/KEGG（Gene Ontology/Kyoto Encyclopedia of Genes and Genomes）enrichment analyses for potential deleterious gene（p<0.05）ofMAF<0.01 was performed to explore the pathogenic genes and pathways related to MUD.4.In the final part,the parallel independent component analysis（p ICA）was used to fuse functional brain image modality and genetic variants modality to probe the genetic basis of the altered brain function in MUD patients.With genetic variation as independent variable,altered brain function as mediating variable and MUD diagnosis as outcome variable,the relationship in gene-brain-MUD disease was probed using mediation analysis.Results:A total of 55 male healthy controls and 110 male individuals with MUD were included in this study.In the part of clinical features assessment,7 cases（HC=2 and MUD=5）were excluded because they did not complete the questionnaire.In the study of functional brain network,8 cases were excluded because they were unable to finish the MRI scan or had incomplete MRI data.After preprocessing and quality control to MRI data,4 cases（HC=2 and MUD=2）were excluded due to unqualified head movement.Therefore,51 healthy controls and 102 participants with MUD were included in this part.In the study of genetics,only 145 participants completed peripheral blood collection,and 3 patients in the MUD group were excluded due to unqualified blood samples.Therefore,142 cases（HC=51 and MUD=91）were included in this part.The results are as follows:1.In the study of trait impulsivity,we found that scores of motor impulsivity,attentional impulsivity,nonplanning impulsivity and BIS-11 total score were significantly lower in the MUD group than HC group（all p<0.05）.After controlling for general demographic data,smoking and alcohol status,partial correlation showed that impulsivity traits were significantly associated with multiple drug use variables and MUD severity（p<0.05）.The mediating analysis indicated complete mediation effects of BIS-11score in the relationships between the initial age of drug use and MUD severity（mediation effect:47.99%;95%CI:-0.27,-0.07）,and partial mediation effects of BIS-11score（mediation effect:19.47%;95%CI:0.02,0.17）and nonplanning impulsivity（mediation effect:12.81%;95%CI:0.01,0.12）,respectively,in the relationships between craving and the severity ofMUD.2.The results of functional brain network analysis are as follows:2.1 In the study of functional connectivity,our results demonstrated the small-world property of the resting state network in the MUD and HC groups（bothσ>1）.However,when compared to HC,individuals with MUD had significantly lower small-world attribute(λ_MUD>λ_HC,p<0.05),and the other global topological attributes reduction was similar with small-world,but the trend tests failed to reach statistical significance（p>0.05）.in addition,individuals with MUD had a significantly increased node clustering coefficient and node local efficiency in the cerebellar vermis（both p<0.001,corrected by Bonferroni）.In the MUD group,after controlling for covariance,results of partial correlation analysis showed that small-world property was significantly associated with METH use frequency,psychological craving and impulsivity trait（all p<0.05）,and global efficiency was positively correlated with motor impulsivity（p<0.05）.The topological attributes of bilateral precuneus and right middle frontal gyrus were significantly correlated with motor impulsivity（p<0.001,corrected by Bonferroni）,and the topological attributes of the right parahippocampal gyrus were significantly associated with attention impulsivity（p<0.001,corrected by Bonferroni）.The results of network-based statistical analysis showed two decreased functional connectivity of subnetworks and one increased functional connectivity of subnetwork in participants with MUD（p<0.0005,NBS correction,permutation 5000times）.2.2 In the study of functional network connectivity,36 independent components were obtained by independent component analysis,and 19 typical independent components were determined and divided into 8 brain networks.It was found that within-network connectivity in salience network（SAN）was enhanced,between-network connectivity was decreased in the SAN and cerebellar network（ECN）,dorsal attention network（DAN）,and visual network（VN）,the default mode network and VN,and visual network（AN）,and the frontoparietal network（FPN）and DAN（all p<0.05,corrected by FDR）.In the MUD,the results of partial correlation analysis showed that between-network connectivity in the SAN and sensorimotor network（SMN）was negatively correlated with METH use duration（p=0.002,corrected by FDR）,while between-network connectivity in the CEN and DAN was positively associated with METH use duration（p<0.001,corrected by FDR）.2.3 In the study of dynamic functional connectivity,according to the minimum free energy,12 dynamic brain network states were obtained by constructing HMM.The results of two-sample t-test showed that there were statistical differences of FO and MDT of state between HC and MUD groups（p<0.05,uncorrected by Bonferroni）.After controlling covariances,partial correlation analysis displayed that MDT of state3 was negatively correlated with the initial age ofMETH use（p=0.004,corrected by Bonferroni）,MDT of state 9 was significantly correlated with METH duration（p=0.001,corrected by Bonferroni）,and SR was significantly correlated with METH use dosage（p=0.025,corrected by Bonferroni）.2.4 In the study of dynamic functional network connectivity,5 dynamic brain network states were obtained by K-means clustering analysis.When compared with HC,individuals with MUD had higher dynamic functional within-network connectivity strength in the VN,andFPN in the state 3（both p<0.05,corrected by FDR）,and between-network connectivity strength between FPN and DMN in the state3（p<0.05,corrected by FDR）,and between-network connectivity strength between DMN andFPN,and DMN and CEN in the state 4（both p<0.05,corrected by FDR）.In addition,we found that MDT of state 2 increased,andFO of states 2 and 5 decreased in MUD group（all p<0.01,corrected by FDR）.3.In the genetic study,the results of CMC burden analysis showed that rare variants of SPATA21 may be related to genetic high risk ofMUD.Pathways of ABC-type xenobiotic transporter activity（ABCC6,ABCC1,ABCB11）,methyltransferase complex and histone methyltransferase complex（SETD1A,KDM5B,MGA,PHF20,WDR5B,RIOK1）,and ABC transporters were enriched using rare variants association with MUD by GO/KEGG enrichment analysis（ABCC6,ABC）C1,ABCB11,ABCA12).4.Fusion analysis of functional brain images and genetic variants revealed that one pair of components displayed significant correlation（fMRI25-VAR4）.After adjusting for covariates,the correlation remained significant（r=-0.525,p<0.001,corrected by Bonferroni）.The fMRI 25 component was mainly located in left occipital lobe and temporal lobe.The results of enrichment analysis of the top 5%of VAR4component showed that DNAH1 and DNAH14 were involved in the expression of axial filament dynamic proteins.In addition,we found that the load coefficient of fMRI25 was positively correlated with the frequency of nonplanning impulsivity and METH use frequency,and the load coefficient of VAR4 was negatively correlated with attentional impulsivity.Mediation analysis showed that the load coefficient of fMRI25 partially mediated the effect load coefficient of VAR4 on MUD.Conclusion:1.Individuals with MUD had higher cognitive impulsivity and behavioral impulsivity than HC,and impulsivity traits mediated the effect of the initial age ofMETH use and craving on the severity ofMUD,which indicates that impulsivity traits play an important role in the occurrence and development of drug abuse,and high impulsivity can be regarded as a potential behavioral marker to predict the prognosis of addiction.2.The decreased ability of information integration and segregation in the brain and changes in functional connectivity strength between the frontal and parietal lobes in MUD patients provide a neurobiological basis for METH-induced impairments in cognitive control.Changes in the topological properties of the cerebellum and functional connectivity strength between the SAN and CEN demonstrate the involvement of the cerebellum in the addiction process.The changes in functional connectivity strength between the CEN and DMN may be dynamic brain network characteristics ofMETH addiction and could serve as a predictive indicator of detoxification efficacy for MUD.3.It was found that genetic variants in the SPATA21,SETD1A,KDM6B and ABC gene families might be significantly associated with susceptibility to MUD,with the mechanisms possibly involving immune dysregulation,impaired neural development,and defective synaptic plasticity.Our findings provide a new insight into the genetic etiological mechanism ofMUD from the perspective of rare variants.4.We initially used fusion analysis between functional brain images and genetic variants in addiction research to reveal brain functional changes and its underlying genetic components,and it was found that both adverse neurol development might lead to changes of brain function in individuals with MUD.In addition,we also found that the genetic components may induce MUD by affecting brain function.As an intermediate phenotype,brain functional changes play an important role in the interpretation of the genetic etiological mechanism ofMUD.In summary,in this study,we comprehensively explored the connective characteristics of functional brain network through multiple dimensions respectively using atlas-based and data-driven segmentation to construct static and dynamic brain network,and we found the characteristic neuroimaging biomarkers in individuals with MUD.In addition,using second-generation whole-exome sequencing technology,the study discovered SUD candidate genes and their mechanisms of action,and further integrated functional brain imaging and genetic variation analysis to investigate the potential genetic mechanisms underlying brain functional changes in MUD patients.These findings lay the foundation for the exploration of mechanism of neuroimaging and genetic etiology ofMUD,and provide a theoretical basis for further early recognition,intervention,and relapse prevention ofMUD.更多还原