Metabolomics from a Controlled Feeding Study and Large Cohorts to Identify Diet-Associated Biomarkers, Augmented by Genetics to Infer Causality for Cardiometabolic Outcomes

饮食与健康和疾病有着密切的关系。许多研究已经将饮食组成的变化与心脏代谢结果如肥胖、2型糖尿病（T2 D）或心血管疾病（CVD）相关联。然而，相关性并不能证明因果关系。一个关键的挑战是确定饮食的哪些影响（如果有的话）与长期健康结果有因果关系。传统上，随机对照试验（RCT）用于证明因果关系。然而，这样的试验基本上是不可行的，以评估机制和因果关系的影响，饮食对长期的健康结果。饮食的改变改变了许多变量，对生理产生了无数的影响。此外，依从性通常比结果增加更快地减弱，并且需要强有力的生物标志物来指示参与者已经遵守饮食并且饮食已经产生了预期的效果。幸运的是，全面的代谢物分析，包括在一个大型的，控制良好的饮食RCT和基于人群的队列中，通过人类遗传学增强，提供了一个潜在的途径，以确定强大的生物标志物，并评估饮食模式，生物标志物和长期心脏代谢结果之间的预测和因果关系。我们建议利用新的代谢组学和遗传学方法来确定关键饮食模式的生物标志物，碳水化合物与脂肪的比例，并使用这些生物标志物来检查这种饮食模式与心脏代谢结果（肥胖，T2 D和CVD）之间的关系。我们将实现三个具体目标。在SA#1中，我们将建立在令人兴奋的初步结果的基础上，并从一个大型的精心控制的饮食RCT中生成全面的非靶向代谢物分析数据，以确定与体重稳定受试者的饮食碳水化合物-脂肪比相关的代谢物（已知和未知）。在SA#2中，我们将利用具有饮食和代谢组学数据的大型现有基于人群的队列来验证这些代谢物作为稳健的生物标志物。在SA#3中，我们将使用纵向数据来测试这些生物标志物是否预测心脏代谢结果，并将采用一种称为孟德尔随机化（RCT的遗传等效物）的方法来评估饮食相关代谢物是否是肥胖，T2 D或CVD的原因。我们将部署多种创新方法，包括饮食RCT中的代谢物分析，非靶向代谢物分析数据的大型遗传研究，避免孟德尔随机化已知困难的方法，以及协调研究中非靶向分析中未识别代谢物的方法。这些目标的成功完成将开发关键饮食模式（碳水化合物与脂肪比）的稳健生物标志物，测试这些生物标志物是否是长期健康结果的有用预测因子，并测试对心脏代谢结果的这种饮食模式的生理反应的因果关系。这种方法将为改善营养干预研究提供一个范例，以更好地了解饮食与健康之间的关系，并最终加强饮食治疗，以预防和治疗慢性疾病。

Diet has a strong relationship with health and illness. Many studies have correlated variation in dietary composition with cardiometabolic outcomes such as obesity, type 2 diabetes (T2D) or cardiovascular disease (CVD. However, correlation does not prove causation. A key challenge is to determine which (if any) of the effects of diet are causally linked to long-term health outcomes. Classically, randomized controlled trials (RCTs) are used to prove causality. However, such trials are essentially infeasible for assessing mechanism and causal effects of diet on long-term health outcomes. Dietary modifications alter many variables, with myriad effects on physiology. Furthermore, compliance typically wanes much more quickly than outcomes accrue, and robust biomarkers are needed that indicate that a participant has complied with a diet and that the diet has had the desired effect. Fortunately, comprehensive metabolite profiling, including in the setting of a large, well-controlled dietary RCT and population-based cohorts, augmented by human genetics, offers a potential route forward to identify robust biomarkers and to assess the predictive and causal relationships between dietary patterns, biomarkers, and long term cardiometabolic outcomes. We propose to take advantage of new metabolomics and genetic methods to identify robust biomarkers of a key dietary pattern, carbohydrate-to-fat ratio, and to use these biomarkers to examine the relationship between this dietary pattern and cardiometabolic outcomes (obesity, T2D, and CVD). We will achieve three specific aims. In SA#1, we will build on exciting preliminary results and generate comprehensive untargeted metabolite profiling data from a large carefully controlled dietary RCT, to identify metabolites (known and unknown) that are associated with dietary carbohydrate-to-fat ratio in weight-stable subjects. In SA#2, we will leverage large existing population-based cohorts with dietary and metabolomic data to validate these metabolites as robust biomarkers. In SA#3, we will use longitudinal data to test whether these biomarkers predict cardiometabolic outcomes, and will employ an approach called Mendelian randomization (the genetic equivalent of a RCT) to assess whether the diet-associated metabolites are causal for obesity, T2D or CVD. We will deploy multiple innovative approaches, including metabolite profiling in a dietary RCT, large genetic studies of untargeted metabolite profiling data, methods to avoid known difficulties with Mendelian randomization, and also methods to harmonize unidentified metabolites from untargeted profiling across studies. Successful completion of these aims will develop robust biomarkers of a key dietary pattern (carbohydrate-to- fat ratio), test whether these biomarkers are useful predictors of long term health outcomes, and test causality of the physiological responses to this dietary pattern for cardiometabolic outcomes. This approach will provide a paradigm for improving nutritional interventional studies, for better understanding the relationship between diet and health, and ultimately enhance dietary treatments to prevent and treat chronic diseases.