Quantitative metabolomics for prediction of lameness and elucidation of related mechanistic pathways in dairy cattle

人类和牲畜的复杂疾病很常见，管理起来也极具挑战性。许多因素导致个人罹患疾病的风险，包括遗传和非遗传因素。在试图管理疾病风险时，需要衡量和了解这些不同因素的相对重要性，这是很难实现的。归根结底，这些挑战往往导致缺乏关于如何预防或降低人群疾病风险的知识和理解。跛行(活动能力受损)是英国乳制品行业最优先考虑的疾病之一。这种令人痛苦和虚弱的疾病影响到牛的福利以及生产和健康。目前的估计表明，在任何时候，英国牛群中约30%的奶牛都患有跛行。其影响不仅限于奶牛，还包括奶牛养殖场的可持续性以及疾病降低效率对环境的影响。作为一种复杂的疾病，导致其发生的因素很多，跛行本质上是具有挑战性的，我们对导致跛行的疾病过程的理解存在重大差距。已知的是，进入第一次哺乳的奶牛(小母牛)在防止跛行方面是乳牛群中最重要的一组牛；小母牛代表着牛群的未来，一旦奶牛经历第一次跛行，脚部就会发生病变，使它们处于更高的复发风险。如果我们要从一开始就防止出现跛行，就应该把注意力集中在小母牛身上。目前，只有在疼痛导致奶牛以不同的步态行走时，才有可能在疾病的晚期检测到跛行。在此阶段之前及早发现跛行的能力将是对当前情况的巨大进步，从而改善健康、福利和生产力。代谢组学是一种允许测量影响疾病风险的遗传和非遗传因素的最终产品(代谢组)的技术。通过比较疾病和健康个体的代谢组，可以识别疾病的特征(生物标记物)。为了进行这些比较，可以使用复杂的统计方法；统计学和代谢组学的强大组合可以识别疾病生物标记物，并进一步了解导致疾病风险的过程。我们以前的工作表明，通过使用这种方法，从产犊前(跛行前1-10周)收集的尿样代谢组可以预测跛行的准确率为93%。该项目将使用代谢组学和人工智能的创新组合来识别奶牛跛行的生物标志物。将对160头小母牛分组进行长期监测(最长305天或每头奶牛一次哺乳)，定期收集尿液、血浆和牛奶样本以及与跛行、健康和生产有关的数据；为本项目和未来的项目提供宝贵的资源。配对样本将从跛行和非跛行的首次哺乳奶牛中挑选，在第一个跛行病例之前不久和产犊前不久。这些时间点被认为在导致跛行的病变的发展过程中是重要的。将开发计算模型来预测跛行的发生，方法是比较患病和健康个体的代谢组；识别代谢组中预测跛行的信号，并使用这些信号来了解发生的疾病过程。这些结果将为了解导致奶牛跛行的疾病途径和预测疾病风险提供至关重要的洞察。这反过来将为疾病管理提供信息，并提供一种早期预测跛行的工具，使奶牛跛行的检测和管理发生阶段性变化。这一方法适用于牲畜和人类的许多其他复杂疾病，并将在今后的工作中进一步使用。

Complex diseases of humans and livestock are common and extremely challenging to manage. Many factors contribute towards an individual's risk of experiencing disease, including genetic and non-genetic factors. When trying to manage disease risk, the relative importance of these different factors needs to be measured and understood, which is difficult to achieve. Ultimately, these challenges often result in a lack of knowledge and understanding of how to prevent or reduce disease risk in a population. Lameness (impaired mobility) is one of the highest priority diseases for the UK dairy industry. This painful and debilitating condition affects cattle welfare as well as production and health. Current estimates suggest that at any one time ~30% of dairy cows in UK herds are suffering from lameness. Implications extend beyond the cow to the sustainability of dairy farming and environmental impacts from disease reducing efficiencies. As a complex disease with many factors contributing to its occurrence, lameness is inherently challenging to tackle and major gaps exist in our understanding of disease processes contributing to lameness. What is known, is that cows entering their first lactation (heifers) are the most important group of cattle in a dairy herd in terms of preventing lameness; heifers represent the future of the herd and once cows experience a first episode of lameness pathological changes occur in the foot that place them at higher risk of recurrence. If we are to prevent lameness occurring in the first place, attention should be focussed on heifers. Detection of lameness is currently only possible at an advanced stage of disease when pain causes cows to walk with an altered gait. The ability to detect lameness early on, prior to this stage, would be a huge advancement on the current situation resulting in improved health, welfare and productivity.Metabolomics is a technique that allows the end products (metabolome) of genetic and non-genetic factors influencing disease risk to be measured. By comparing the metabolome of diseased and healthy individuals, signatures (biomarkers) of disease, can be identified. To make these comparisons, complex statistical methods can be used; a powerful combination of statistics and metabolomics can identify disease biomarkers and further our understanding of processes contributing to disease risk. Our previous work has shown that by using this approach lameness can be predicted with an accuracy of 93% from the metabolome of urine samples collected prior to calving (1 - 10 weeks prior to lameness). This project, will use an innovative combination of metabolomics and artificial intelligence to identify biomarkers for lameness in dairy cows. Groups of 160 dairy heifers will be monitored over a prolonged period of time (up to 305 days or one lactation per animal) with collection of regular urine, plasma and milk samples and data related to lameness, health and production; providing a valuable resource for this and future projects. Paired samples will be selected from lame and non-lame first lactation cows shortly before the first case of lameness and prior to calving. These time-points have been selected as being important in the development of lesions causing lameness. Computational models will be developed to predict the occurrence of lameness by comparing the metabolome of diseased and healthy individuals; identifying signals in the metabolome that predict lameness and using these to understand the disease processes occurring. Results will provide a vital insight into disease pathways contributing to lameness in dairy cows and the ability to predict disease risk. This in turn will inform disease management and provide a tool for early prediction of lameness, offering a step change in the detection and management of lameness in dairy cows. This approach is applicable to many other complex diseases in both livestock and humans and will be further utilised in future work.