毕业生知识库

精神健康问题已成为现代医学和心理学领域所面临的全球性挑战。世界范围内的统计数据显示，在导致人类伤残的前十大原因中，有一半与精神健康问题相关。精神健康覆盖了从正常到异常的广泛心理状态，涵盖情感、心理和社交功能的各个层面。常见的精神健康问题，如焦虑、抑郁和躁狂等，可能会干扰个体的正常生活秩序，严重时甚至会成为精神疾病的先兆。精神疾病可以视为精神健康连续谱系中的极端偏差，不仅让患者承受巨大痛苦，还给家庭和社会造成了沉重的经济负担。然而，由于精神健康背后的神经生物学和遗传学基础尚不完全明了，这严重制约了对各类精神疾病的精准诊断和有效治疗。

大多数精神健康问题表现出多基因遗传的特点，并且在大脑多个区域存在程度不一的结构异常。生物内表型（如脑影像）是相对明确的生理或行为测量，它搭建了基因和疾病症状表型之间的桥梁。近年来，随着大规模影像基因组学数据的涌现以及多组学技术的不断发展，大脑与遗传学的潜在关联为解析复杂的精神健康表型提供了新的视角。本文基于来自多个中心的大规模影像基因组学数据，从精神疾病到连续维度的症状评估，从表型相关到遗传相关，从相关关系到因果关系，从多视角研究了基因、脑结构与精神健康之间的复杂关系，试图为精神健康的神经机制及其背后的遗传学基础提供新的见解。论文的主要工作和创新点归纳如下：

1. 融合脑影像与基因组数据的精神分裂症分类研究

本文首先关注精神健康的极端表型-精神疾病的客观诊断分类，已有研究主要是基于脑影像特征的模型，忽略了基因组等遗传特征的影响。本文提出融合脑影像与基因组数据，可有效提高疾病分类准确率。以重性精神疾病-精神分裂症为例，本文基于来自中国汉族人群的8个中心的大规模数据集，创新性地将脑结构磁共振影像和多基因风险分数相结合，采用支持向量机、逻辑回归和集成学习等机器学习方法，训练了多个精神分裂症分类模型。留一站点的交叉验证结果显示，融合脑结构和遗传特征的模型在精神分裂症分类上优于单一数据源模型，并且在多个独立站点上展现出良好的泛化能力。对分类贡献较大的脑区主要位于颞上回、颞中回、眶回、中央前回、基底神经节和丘脑等区域，这些脑区灰质体积的改变与之前报道过的研究结果一致，为精神分裂症的诊断提供了潜在的生物标志物。此外，与精神分裂症关联最紧密的遗传位点对分类模型的贡献最大，表明遗传特征可以提供关于疾病风险的重要信息。该研究揭示了结合脑影像与基因组信息在精神健康研究中的重要性和必要性，为精神疾病的分类研究提供了新的思路，帮助深入理解精神疾病的病理生理学机制。

2. 脑结构与精神健康的关联模式及其遗传机制研究

精神病理学的维度方法旨在研究从正常到异常的整个变化范围，以提高对典型与病理状态之间差异的理解。基于此，本文在上述精神疾病分类研究的基础上，进一步扩展到一般人群的精神健康评估，以深入探索遗传效应如何作用于神经生物学过程，进而影响精神健康的具体机制。利用大规模英国生物样本库的脑结构磁共振影像、精神健康评估、认知行为测试和基因组学等数据，本文在广泛的一般人群中发现了脑结构与精神健康之间两种稳定的多元关联模式。其中，模式一主要表征了小脑体积减小与精神创伤增加的关联，并且对生存率有显著的危害作用；而模式二则主要表征了额颞叶等皮层区域的形态变化与精神困扰减少及躁狂症状增加的关联。遗传学分析表明，与精神健康症状的全基因组关联研究（genome-wide association study，GWAS）相比，脑结构模式的GWAS能更有效地识别遗传位点，并且揭示了多个对基因表达具有重要调控作用的位点。进一步的功能注释、富集分析、基于基因的关联和遗传相关性等分析发现，脑结构模式一主要涉及基因表达的调控等生物过程，并且与精神疾病和认知存在显著的遗传重叠；而脑结构模式二则与儿茶酚胺等神经递质的代谢过程有关，并且与神经精神疾病中的认知异常存在显著的遗传重叠。通过多基因风险分数分析，本文在独立样本中验证了这些遗传重叠发现，为从遗传学角度预测个体的精神健康状况提供了可能。该研究首次揭示了脑结构与精神健康关联背后的遗传机制，并从新的视角建立了遗传变异、大脑结构与精神健康之间的关联，为全面理解精神健康问题的神经生物学机制和遗传学基础提供了新的科学依据，也进一步支持了精神疾病的连续性假说。

3. 脑结构与精神健康之间的因果关系研究。

上述研究发现，脑结构与精神健康之间存在表型和遗传上的相关性，但它们之间是否具有因果关系尚不明确。本章在上述两种脑结构模式的GWAS结果基础上，结合了多项与精神健康相关的GWAS汇总统计数据，运用以遗传变异为工具变量的双向孟德尔随机化方法，揭示了脑结构对精神健康和认知相关性状有潜在的正向因果效应，而反向因果关系则不显著。通过水平多效性检验、异质性检验和留一法敏感性分析，本文发现脑结构模式一与精神疾病之间没有直接的因果效应，但脑结构模式一分数的增加会导致与精神疾病紧密相关的C反应蛋白和饮酒水平升高；同时，脑结构模式二分数的增加会导致自闭症谱系障碍和抑郁症状风险的增加，以及儿童期智力的下降。这些发现为脑结构与精神健康之间的因果关系提供了有力的遗传学证据，有助于在脑影像水平上更好地预测和干预精神疾病的潜在风险因素。

Mental health problems have become a global challenge in modern medicine and psychology. Worldwide statistics show that half of the top 10 causes of human disability are related to mental health problems. Mental health covers a wide range of mental states from normal to abnormal, covering all levels of emotional, psychological and social functioning. Common mental health problems, such as anxiety, depression and mania, may interfere with an individual's normal life order and even become a precursor of mental disorder. Mental disorder can be seen as an extreme deviation on the continuum of mental health, not only inflicting great suffering on the sufferer, but also imposing a heavy economic burden on families and society. However, because the neurobiological and genetic basis behind mental health is not fully understood, this seriously restricts the accurate diagnosis and effective treatment of various mental disorders.

Most mental health problems are characterized by polygenic inheritance and have varying degrees of structural abnormalities in multiple brain regions. Endophenotypes (such as brain imaging) are relatively explicit physiological or behavioral measures that bridge the gap between genes and disease symptom phenotypes. In recent years, with the emergence of large-scale imaging genomics data and the continuous development of multi-omics techniques, the potential connections between the brain and genetics have provided new perspectives for resolving complex mental health phenotypes. Based on large-scale imaging genomics data from multiple centers, this dissertation examines the complex relationship between genes, brain structure, and mental health from multiple perspectives, ranging from psychiatric disorders to symptom assessments on a continuum, from phenotypic correlations to genetic correlations, and from correlation to causation, in an attempt to provide new insights into the neural mechanisms of mental health and the genetic basis behind it. The main work and innovation of this dissertation are summarized as follows:

1. Study on the classification of schizophrenia by integrating brain imaging and genomic data.

In this dissertation, we first focus on the objective diagnostic classification of the extreme phenotype of mental health—psychiatric disorders. Existing research, primarily model-based on brain imaging features, has overlooked the influence of genomic and other genetic characteristics. This dissertation proposes that integrating brain imaging and genomic data can effectively improve the accuracy of disease classification. Taking schizophrenia, a severe mental disorder, as an example, based on large-scale data sets from 8 centers of Chinese Han population, this dissertation innovatively combined brain structure magnetic resonance imaging and polygenic risk scores, and used machine learning methods such as support vector machine, logistic regression and ensemble learning to train multiple schizophrenia classification models. The leave-one-site-out cross-validation results indicated that models integrating brain structure and genetic features was superior to models based on a single data source in classifying schizophrenia, and showed good generalization ability at multiple independent sites. The brain regions that contributed more to the classification were mainly located in the superior temporal gyrus, middle temporal gyrus, orbital gyrus, anterior central gyrus, basal ganglia and thalamus. The changes in gray matter volume in these brain regions were consistent with previously reported results, providing potential biomarkers for the diagnosis of schizophrenia. Furthermore, the genetic loci most closely associated with schizophrenia make the greatest contribution to the classification model, indicating that genetic features can provide important information about disease risk. This dissertation reveals the importance and necessity of combining brain imaging and genomic information in mental health research, provides a new idea for the classification of mental disorders, and helps to deeply understand the pathophysiological mechanism of mental disorders.

2. Study on the association patterns and genetic mechanism between brain structure and mental health.

The dimensional approach to psychopathology aims to investigate the entire range of changes from normal to abnormal in order to improve understanding of the differences between typical and pathological states. Based on this, this dissertation further extends to the mental health assessment of the general population on the basis of the above classification of mental disorders, in order to deeply explore the specific mechanisms of how genetic effects act on neurobiological processes and thus affect mental health. Using brain structure magnetic resonance imaging, mental health assessments, cognitive behavioral tests, and genomics data from a large UK Biobank, we found two stable multivariate association patterns between brain structure and mental health in a broad range of the general population. Among them, mode 1 mainly characterized the association between cerebellar volume reduction and increased mental trauma, and had a significant harmful effect on survival. Mode 2 mainly characterized the relationship between the morphological changes in cortex such as the frontal and temporal lobes and the reduction of mental distress and the increase of mania symptoms. Genetic analyses have shown that compared to the Genome-Wide Association studies (GWAS) of mental health symptoms, the GWAS of brain structure patterns can more effectively identify genetic loci, and have revealed multiple loci with significant regulatory effects on gene expression. Further analysis of functional annotation, enrichment analysis, gene-based association and genetic correlation found that brain structure mode 1 is mainly involved in biological processes such as the regulation of gene expression, and had significant genetic overlap with mental disorders and cognition; while brain structure mode 2 is related to metabolic processes of neurotransmitters such as catecholamines and has significant genetic overlap with cognitive abnormalities in neuropsychiatric disorders. Through polygenic risk score analysis, we validated these genetic overlap findings in independent samples, providing the possibility to predict an individual's mental health status from a genetic perspective. This dissertation reveals for the first time the genetic mechanism behind the association between brain structure and mental health, establishes the association between genetic variation, brain structure and mental health from a new perspective, and provides new scientific evidence for a comprehensive understanding of the neurobiological mechanisms and genetic basis of mental health problems, further supporting the continuity hypothesis of mental disorders.

3. Study on the causal relationship between brain structure and mental health.

The above studies have found that there are phenotypic and genetic correlations between brain structure and mental health, but it is still unclear whether there is a causal relationship between them. Based on the GWAS results of the above two brain structure modes and combined with multiple GWAS summary statistics related to mental health, we used two-sample Mendelian randomization method with genetic variants as the instrumental variables, revealing that potential direct causal effects of brain structure on mental health and cognitive related traits, while the reverse causal relationship was not significant. Through horizontal pleiotropy test, heterogeneity test and leave-one-out sensitivity analysis, we found that there was no direct causal effect between brain structure mode 1 and mental disorders, but the increase of brain structure mode 1 score would lead to the increase of C-reactive protein and alcohol consumption, which are closely related to mental disorders. An increase in brain structure mode 2 scores would lead to the increased risk of autism spectrum disorder and depressive symptoms, as well as decreased intelligence in childhood. These findings provide strong genetic evidence for a causal relationship between brain structure and mental health and will help to better predict and intervene in potential risk factors for mental disorders at the brain imaging level.