第一部分 深髓静脉损害相关的脑结构特征谱

研究背景和目的：静脉胶原蛋白病早已被列入脑小血管病的范畴，但是脑小静脉病致病机制的研究非常有限。静脉引流障碍和淀粉样物质清除障碍两种机制合并存在，导致复杂脑结构损害模式。本课题拟在大样本社区人群列队中，探讨深髓静脉、脑白质和灰质结构损害、淀粉样物质清除障碍等各环节的时序特征和空间模式，进一步理解脑小静脉病的发生发展过程。

研究方法：本研究入组了顺义队列中完成了良好质量的基线3T核磁评估的受试者，包括T1、磁敏感加权成像（Susceptibility-weighted imaging，SWI）、弥散张量成像（diffusion weighted imaging， DTI）等序列，并进行5年头核磁随访。通过视觉评估SWI图像获得深髓静脉（Deep Medullary vein, DMV）的数量；使用SPM和freesurfer自动分割3D-T1取得白质高信号（White matter hyperintensities, WMH）体积和脑体积参数；通过FSL取得DTI参数。对基线和随访血浆样本行Aβ40、Aβ42及Aβ42/40浓度测定。数据分析分为全脑水平和体素/表面水平两个层次：（1）全脑水平：通多元线性回归分析深髓静脉数与脑结构参数的的相关性，使用中介模型分析深髓静脉数-脑结构关联的中介环节。（2）体素/表面水平：通过FSL软件中基于纤维束的空间统计（Tract-Based Spatial Statistics, TBSS）工具实现DMV数量与DTI参数的线性相关性分析；使用CAT12工具包和freesurfer实现DMV数量与灰质体积基于体素/表面的相关性分析。

研究结果：基线分析纳入977人（平均年龄56.8岁，34.7%男性），其中544人获得完整核磁随访（平均随访时间5.56年）。结果发现深髓静脉减少与广泛的白质微结构完整性破坏相关（深髓静脉数与分数各向异性β=0.000863，SE=0.000176，p<0.001），并且在深部白质具有更快的进展速度。更值得注意的是，深髓静脉数目减少与全脑萎缩有稳定和较强的相关性（深髓静脉数与脑实质分数β=0.03，SE=0.001，p<0.001），萎缩分布于双侧颞叶、海马等部位。进一步对深髓静脉损害导致脑萎缩的各致病环节进行了探索，发现白质微结构破坏在深髓静脉导致的脑萎缩中具有中介作用。在深髓静脉损害发展病程的不同阶段，脑结构改变的模式具有双向特征，提示不同的致病机制。

结论：深髓静脉损害导致广泛的白质和灰质结构损害，两者相互作用且在病程进展中有多种的机制参与。本研究揭示了脑小静脉病在衰老及神经系统变性病中的潜在机制，值得关注和进一步研究。

第二部分 深髓静脉损害与认知的关系探究

研究背景和目的：脑小静脉病理改变与神经系统变性病之间的关系受到关注，据此推测脑小静脉病在功能层面上可能导致认知下降为核心特征的临床表现，但是目前研究很少。本研究在长程随访的大样本社区人群列队中，探讨深髓静脉与全面认知功能指标、认知事件发生及进展的关系，并分析脑结构特征在这些关联中的作用，以进一步揭示脑小静脉病的临床症状特征及其相关机制。

研究方法：本研究入组了顺义队列中完成了全面的基线神经心理学评估的受试者，包括简易精神状态检查、蒙特利尔认知评估、FOM实物记忆测试、快速言语检索测试、积木测试和数字回忆测试，并进行4年后的认知随访。通多元线性回归分析深髓静脉数与脑结构参数的相关性，使用多元逻辑回归分析深髓静脉数对痴呆、轻度认知功能障碍（Mild cognitive impairment，MCI）等认知相关事件的预测价值，使用中介模型分析脑结构指标及血浆淀粉样蛋白浓度在深髓静脉数与认知关联的中介作用。

研究结果：基线分析纳入1040人（平均年龄55.98岁，36.28%男性），其中895人获得完整认知随访（平均随访时间4.33年），出现新发痴呆或MCI分别有31人和59人。在单因素模型中深髓静脉数减少与基线诊断痴呆的风险增加是显著相关的（OR=0.683，95%CI=0.500-0.934，P=0.017）。深髓静脉数减少与较差的词语检索能力相关（β=0.65，SE=0.17，P<0.001），且独立于年龄、性别、受教育年和血管危险因素存在，但是在随访期间未见显著进展。在中介分析中发现，白质微结构损害在深髓静脉导致执行功能下降中具有中介作用。

结论：本研究发现深髓静脉数减少与较差的执行功能相关，并且该关联由白质微结构损害介导，但是在纵向分析中其因果关系未能被验证。

Part I: Temporal and Spatial Patterns of Structural Abnormalities in Cerebral Small Vein Disease

Background and purpose: Cerebral small vessel disease (CSVD) has long been classified as a category of small vessel disease, while the mechanism of cerebral small venules’ disease is poorly understood. We speculate that a combination of venous drainage obstruction and amyloid clearance leads to complex brain structural damage patterns. This study aims to explore the temporal and spatial patterns of damage to the deep medullary vein (DMV), white matter, gray matter, and amyloid clearance in a large community population, to further understanding the occurrence and development of cerebral small venules’ disease.

Methods: We included participants in the Shunyi cohort study with complete baseline MRI, which included T1, SWI, and diffusion tensor imaging (DTI) sequences, was completed with good quality. A 5-yearsFollow-up MRI was conducted. The number of DMVs was obtained by visual assessment of SWI images; the white matter hyperintensity (WMH) volume and brain volume parameters were automatically segmented from 3D-T1 using SPM and Freesurfer; DTI parameters were obtained using FSL. The Aβ40, Aβ42, and Aβ42/40 concentrations in baseline and follow-up plasma samples were measured. We analyzed the relationship between DMV number and brain structure parameters was analyzed using mediation models. Using the tract-based spatial statistics (TBSS) tool, we analyzed the linear correlation between DMV number and DTI parameters. Using CAT12 and freesurfer, we performed and voxel/surface-based analysis on the correlation between DMV number and brain atrophy.

Results: The baseline analysis included 977 participants (average age 56.8 years, 34.7% male), of which 544 received complete MRI follow-up (mean follow-up time 5.56 years). The results showed that the reduction in DMVs was associated with widespread white matter microstructure damage (DMV number and fractional anisotropy β=0.000863, SE=0.000176, p<0.001) and progressed faster in deep white matter. More importantly, the reduction in DMV number was stably and strongly correlated with global brain atrophy (DMV number and brain parenchyma fraction β=0.03, SE=0.001, p<0.001), which was distributed in bilateral temporal lobes, hippocampus, and other regions. White matter microstructure damage played an intermediate role in deep medullary vein-induced brain atrophy. The pattern of brain structure changes showed bidirectional features at different stages of the development of deep medullary vein damage, indicating different pathogenic mechanisms.

Conclusion: Deep medullary vein damage leads to widespread damage to white matter and gray matter structures, with multiple mechanisms involved in disease progression. 

Part II: Investigation of the Relationship between Deep Medullary Veins Damage and Cognition

Background and Purpose: The relationship between cerebral small vessel pathology and neurodegenerative diseases has been recognized. We hypothesize that cerebral small venules' disease may lead to cognitive decline as a core clinical feature at the functional level, but the relevant evidence is lacking. This study aimed to explore the relationship between the deep medullary vein (DMV) damage and cognitive function as well as dementia in a large sample community population with long-term follow-up, and to analyze the role of brain structural parameters in these associations, in order to further reveal the clinical symptom characteristics and pathogenic mechanisms of cerebral small venules' disease.

Methods: Participants in ShunYi study that completed comprehensive baseline neuropsychological evaluations in the Shunyi cohort were included in this study. These evaluations included the Mini-Mental State Examination, Montreal Cognitive Assessment, Fuld Object Memory Evaluation, rapid verbal retrieval (RVR), Block Design Test, and Digit Span Test, and cognitive follow-up was conducted four years later. Multiple linear regression was used to analyze the correlation between the number of s (DMVs) and brain structural parameters, and multiple logistic regression was used to analyze whether the number of DMVs predicts dementia and MCI. Mediation analysis was used to analyze the effect of brain structural indicators and plasma amyloid protein concentration in the association between the number of deep medullary veins and cognition.

Results: We included 1040 participants the baseline analysis (mean age 56.8 years, 34.7% male), of whom 895 received complete cognition follow-up (mean follow-up time 4.33 years), and there were 31 and 59 new-onset dementia and MCI, respectively. In the univariate model, a decrease in the number of deep medullary veins was significantly associated with an increased risk of dementia (OR=0.683, 95%CI=0.500-0.934, P=0.017). The reduced number of deep medullary veins was also associated with poorer performance in RVR test (β=0.65, SE=0.17, P<0.001), independently of age, sex, years of education, and vascular risk factors, but no significant progression was observed during follow-up. In the mediation analysis, white matter microstructural damage mediated the association between DMVs number and RVR score.

Conclusion: Reduced number of DMV is associated with poorer executive function, is mediated by white matter microstructural damage. However, the causal relationship could not be validated in longitudinal analysis.