毕业生知识库

轻度认知障碍（mild cognitive impairment，MCI）是介于正常（cognitively normal，CN）衰老和阿尔兹海默症（Alzheimer’s disease，AD）之间的过渡状态，是发展为AD的高风险因素。目前AD的药物治疗收效甚微，在MCI阶段进行干预治疗，有可能延缓AD的发生和发展。因此，探索MCI的发病机制，识别MCI早期诊断的生物标志物具有重要的临床意义。β淀粉样蛋白（amyloid-β，Aβ）和tau蛋白沉积是AD的两个特征性病理，也是MCI的生物标志物，然而二者的病理改变具有临床异质性。通过正电子发射断层成像（positron emission tomography，PET）技术对Aβ和tau研究发现，沉积过程的异质性与临床异质性有关。但已有研究仅揭示了Aβ在沉积过程中的时间异质性或空间异质性，缺少对Aβ时空异质性的表征。Tau作为另一个特征性病理，尽管发现了与记忆减退相关的tau时空亚型及沉积模式，但尚不清楚该亚型是否有助于MCI的早期诊断。此外，亚型背后的遗传基础也有待进一步探索。本论文基于Aβ和tau的PET影像数据，围绕二者在脑内沉积的时空异质性问题，首先揭示了Aβ沉积的时空亚型及其与临床特征的关系；其次研究了Aβ和tau的时空亚型对MCI早期诊断的影响；最后，分析了tau及其典型亚型背后的遗传基础。论文的主要工作和创新点归纳如下:

1.Aβ亚型及其与临床特征的关系

基于MCI及其早期阶段的Aβ影像数据，采用数据驱动的方法分解Aβ在脑中的沉积过程以揭示其时空异质性。在两个独立的数据集中一致地发现了两种时空亚型：皮层优先亚型和皮下优先亚型。在皮层优先亚型中，Aβ最先在扣带回沉积，随后扩散至皮层和脑岛，最后蔓延至皮下核团。在皮下优先亚型中，Aβ沉积始于皮下核团，随后蔓延至扣带回和脑岛，最后发展至皮层。两种亚型不仅沉积模式完全不同，其临床特征也存在着显著的差异，包括认知功能、记忆力和执行功能。与皮下优先亚型相比，皮层优先亚型中AD风险基因携带者更多，与AD相关的其他生物标志物也更严重。此外，皮层优先的亚型更容易转化为AD，而皮下优先的亚型疾病发展过程相对缓慢。进一步分析发现这两种亚型的分子途径也存在差异。该研究揭示了Aβ沉积的时空异质性与疾病发展轨迹的关系，为理解发病机制、定义疾病阶段提供了重要线索。

2.结合Aβ和tau亚型的MCI早期诊断

预测认知正常老年人是否会发展为轻度认知障碍具有重要的临床意义，而多模态的神经影像学数据为预测诊断提供了可能，尤其是一些生物标志物亚型的发现提供了更多有价值的信息。本章基于结构磁共振成像（structural magnetic resonance imaging，sMRI），结合Aβ和tau蛋白沉积的PET数据，对认知正常老年人两年后是否会转化为轻度认知障碍进行了预测。由于一些被试缺少tau蛋白影像数据，首先基于sMRI数据并结合深度学习方法对缺失的数据进行了填充。其次结合了Aβ和tau的沉积亚型，提取影像特征进行分类。实验结果表明，生成的tau蛋白影像数据是有效的，而结合沉积亚型提取病理特征能进一步提高预测的准确性。

3.Tau及其亚型的遗传基础

脑内tau的沉积与相关酶的活性失衡有关，也受tau与微管结合能力的影响。然而，该通路相关基因是否为tau沉积和其亚型的遗传风险尚不清楚。本章计算了包含该生物通路基因的多基因风险分数（polygenic risk score，PRS），评估了该PRS与tau蛋白沉积的生物标志物及记忆功能损伤的关系。研究发现在CN和MCI人群中，PRS对tau沉积及其纵向变化和亚型都有显著的影响。Tau蛋白沉积还是PRS与记忆损伤相关的中介变量，即PRS通过影响tau蛋白的沉积进而导致记忆损伤。此外，PRS与tau病理的关系还存在性别差异，女性更容易受遗传风险的影响。本研究发现特定生物通路的基因是tau病理的遗传基础，为理解tau病理潜在的神经机制提供了科学依据。

Mild Cognitive Impairment (MCI) is a transitional state between cognitively normal (CN) and Alzheimer's disease (AD), and is a high-risk factor for conversion to AD. At present, medical treatment produces very little effect in AD. Interventions applied in MCI might have a better chance of changing disease trajectory. Therefore, exploring the pathogenesis of MCI and identifying biomarkers for early diagnosis of MCI become clinically important. Amyloid-β (Aβ) and tau protein deposition are pathologies of AD and biomarkers of MCI, but their changes are clinically heterogeneous. The study of Aβ and tau on positron emission tomography (PET) has found that the heterogeneity of deposition progression is related to clinical heterogeneity. Until recently, most of the studies only focus on the temporal or spatial heterogeneity induced by the deposition of Aβ, and lack of the characterization of Aβ spatiotemporal heterogeneity. Tau is another characteristic pathology. Although a spatiotemporal subtype with deposition pattern was identified, which was associated with memory decline. It is still unclear whether such subtype could improve the early diagnosis of MCI. In addition, the genetic basis behind the subtype remains to be further investigated. This paper aims to explore the spatiotemporal heterogeneity of Aβ and tau deposition in the brain d on PET imaging data. The spatiotemporal subtypes of Aβ deposition were identified and then the relationship between these subtypes and clinical features were evaluated. Moreover, we investigated the influence of spatiotemporal subtypes of Aβ and tau on the early diagnosis of MCI. Finally, the genetic basis behind tau and its typical subtypes was explored. The main work and innovations of the paper are summarized as follows:

1.Aβ subtypes and their relationship with clinical features

Based on Aβ imaging data of mild cognitive impairment and its early stages, a data-driven approach was used to uncover the deposition of Aβ in the brain to reveal its spatiotemporal heterogeneity. Two spatiotemporal subtypes were identified in two independent datasets. They are cortex-priority subtype and subcortex-priority subtype. In the cortex-priority subtype, the earliest amyloid accumulation occurred in the cingulate region, followed by cortical areas and the insula, and finally the subcortical regions. In the subcortex-priority subtype, amyloid deposition began in the subcortical regions, successively affected the cingulate and insula areas, and then progressed to the cortical regions. The two subtypes not only presented distinct regional progression patterns, but also had distinct clinical features, including cognitive function, memory, and executive function. Compared to the subcortex-priority subtype, the cortex-priority subtype has more carriers of AD risk genes and more severe Alzheimer’s disease-related profiles. In addition, the cortex-priority subtype is more likely to convert to AD, while the subcortex-priority subtype has a relatively slow disease progression. The two subtypes also revealed differences in molecular pathways. This study reveals the relationship between the spatiotemporal heterogeneity of Aβ deposition and disease trajectories, which provides important clues to understand the pathogenesis and stage the disease.

2. Early diagnosis of MCI combining Aβ and tau subtypes

Predicting whether CN individuals will convert to MCI is clinically important, and multimodal neuroimaging data offer the possibility of predicting the diagnosis. Moreover, the identification of some biomarker subtypes provides more valuable information. This work predicts the conversion of CN to MCI after two years d on structural magnetic resonance imaging (sMRI), Aβ-PET and tau-PET. Since some individuals lacked tau-PET, the missing data were imputed by sMRI d on generative adversarial networks. Then, using the Aβ and tau subtypes to extract features on PET for classification. The results showed that the imputed tau-PET were valid. Combining the subtypes of the two pathologies could improve the accuracy of prediction.

3.Genetic basis of tau and its subtype

Tau deposition in the brain is associated with an imbalance of related tau kinases and the binding affinity of tau to microtubules. However, there was no evidence ing genes involved in this biological pathway to tau accumulation and tau subtype. In this work, the pathway-specific polygenic risk score (PRS) was calculated and assessed in relation to biomarkers of tau protein deposition, tau subtype and memory impairment. It was found that the PRS was associated with cross-sectional tau aggregation, longitudinal changes and tau subtype in CN and MCI. Furthermore, tau significantly mediated the PRS effect on memory impairment. That is, PRS contributed to memory impairment by affecting tau protein deposition. In addition, gender has an influence on the relationship between PRS and tau pathology, and women are more susceptible to the genetic risk. Our results indicated that the pathway-specific PRS is the genetic basis tau pathology, providing scientific evidence for understanding tau pathology.