Physiology of cognitive changes in ageing and dementia

随着年龄的增长，大脑会发生变化，但哪些变化我们应该认为是正常的，哪些是未来可能出现记忆障碍等认知问题的警告信号？我们如何知道一个人是否偏离了健康老龄化的轨道，最终会导致痴呆症，最常见的诊断是阿尔茨海默病（AD）？这些都是紧迫的问题，尤其是因为英国人口老龄化迅速，而且在未来10到20年里，全球阿尔茨海默病病例将翻一番。我的研究是关于识别大脑生理上最早的变化，这些变化会在以后的生活中引起认知障碍。动物模型的研究已经在细胞水平上确定了许多大脑衰老的候选机制，而大规模的流行病学研究已经确定了人们随着年龄增长通常会经历的认知变化。为了将这些发现联系起来，我们需要对大脑中与年龄相关的变化如何导致认知能力下降有一个机制上的理解。为了解决这个问题，我将解决一个关键的技术挑战。衰老和阿尔茨海默病可以同时影响神经元、大脑的血液供应（脉管系统）和它们之间的界面（神经血管耦合）。我们如何解开这些影响，并确定哪些与认知功能障碍有关？没有单一的医学成像方法可以做到这一点。一种测量大脑功能的标准方法，BOLD fMRI（血氧水平依赖的功能性磁共振成像），可以以高空间精度定位大脑中存在衰老影响的位置，然而它测量的信号反映了神经和血管的混合贡献。整合其他类型的测量可以帮助解决这种模糊性。动脉自旋标记（ASL） MRI提供了血液灌注的直接测量（血液输送到毛细血管床），但它比BOLD fMRI慢，因此灵敏度较低。大脑的直接电磁记录（脑电图，EEG或脑磁图，MEG）提供了对神经活动的精确观察，但它们对脉管系统不敏感。这些方法——BOLD fMRI、ASL MRI和EEG/MEG——提供了关于大脑的互补视角，提出了如何将它们整合起来以构建一个有凝聚力的衰老图景的问题。我的解决方案，也是我在提出这一建议之前提出的，是采取“世界上最好的”方法。我分析了合作者进行的大规模临床研究的数据，在这些研究中，志愿者接受了多种神经成像。例如，他们可能在执行一项简单任务时，首先用MEG测量他们的神经活动，然后在执行同样的任务时，用MRI测量他们的血氧和血流量。我的方法的新颖之处在于这些不同类型的数据被整合的方式。每个志愿者都有一个详细的生物学数学模型，用来描述他们的数据是如何产生的。该模型包含未知的量，如神经连接的强度，这是根据综合神经成像数据估计的。因此，这个模型就像一个“数学显微镜”，用来推断志愿者大脑中产生数据的生物过程。我将运用这种方法来确定与年龄相关的脑功能障碍相关的医疗和健康因素的特殊混合物。使用类似于天气预报的方法，我将调查这些因素中哪些影响认知能力下降的长期轨迹。未来，这可能使有针对性的早期干预成为可能，以防止或减缓衰老带来的认知能力下降。

The brain changes as we age, but which changes should we consider normal, and which are warning signs that cognitive problems such as memory impairment may develop in future? How do we know whether someone is deviating from the course of healthy ageing and is on a path that will eventually lead to dementia, the most common diagnosis being Alzheimer's disease (AD)? These are pressing issues, not least because the UK has a rapidly ageing population and cases of AD are set to double in the next 10-20 years worldwide.My research is about identifying the earliest changes in brain physiology that give rise to cognitive impairment in later life. Research in animal models has identified many candidate mechanisms of brain ageing at the cellular level, while large-scale epidemiological studies have identified the cognitive changes that people typically experience as they age. To connect these findings, we need a mechanistic understanding of how age-related changes in the brain give rise to cognitive decline. To address this, I will tackle a key technical challenge. Ageing and AD can simultaneously affect neurons, the brain's blood supply (vasculature) and the interface between them (neurovascular coupling). How can we disentangle these effects, and identify which of them are associated with cognitive dysfunction? No single medical imaging method can achieve this. A standard approach for measuring the function of the brain, BOLD fMRI (Blood-Oxygen-Level-Dependent functional Magnetic Resonance Imaging), can localise where in the brain there are effects of ageing with high spatial precision, however the signals it measures reflect a mixture of neural and vascular contributions. Integrating other kinds of measurements could help to resolve this ambiguity. Arterial Spin Labelling (ASL) MRI provides a direct measure of blood perfusion (delivery of blood to the capillary bed), but it is slower than BOLD fMRI and therefore affords lower sensitivity. Direct electromagnetic recordings of the brain (electroencephalography, EEG or magnetoencephalography, MEG) provide insights into neural activity with exquisite temporal precision, but they are not sensitive to vasculature. These methods - BOLD fMRI, ASL MRI and EEG/MEG - provide complementary perspectives on the brain, raising the question of how to integrate them to build a cohesive picture of ageing.My solution, which I have developed in the lead up to this proposal, is to take a 'best of all worlds' approach. I analyse data from large-scale clinical studies conducted by my collaborators, where volunteers have undergone multiple kinds of neuroimaging. For example, they may start by having their neural activity measured with MEG while performing a simple task, and then they will perform the same task while undergoing MRI, to measure their blood oxygen and blood flow. The novelty of my approach lies in the way these different kinds of data are integrated. For each volunteer, a biologically detailed mathematical model is specified that describes how their data were generated. This model includes unknown quantities, such as the strength of neural connections, which are estimated from the combined neuroimaging data. The model thereby acts like a "mathematical microscope", for inferring the biological processes in the volunteer's brain that gave rise to their data. I will apply this approach to identify the particular mixture of medical and health factors that are associated with age-related brain dysfunction. Using methods similar to weather forecasting, I will investigate which of these factors affect the long-term trajectory of cognitive decline. This could, in future, enable targeted early interventions to prevent or slow cognitive decline in ageing.