Chronic kidney disease (CKD) is prevalent in the U.S. population and is a major cause of end-stage renal disease (ESRD), cardiovascular disease morbidity, and mortality. Current gaps exist in characterizing the heterogeneity of CKD population. Obesity further increases the risks of adverse outcomes among people with impaired kidney function. However, the mechanism between obesity and CKD is not fully understood, and numerous studies have observed ?obesity paradox? or survival advantage being associated with higher body fat among advanced CKD patients. In this context, the Chronic Renal Insufficiency Cohort (CRIC) Study provides an ideal opportunity to study the mechanism of CKD with its rich clinical information and population characteristics, as well as its availability of non-targeted high-dimensional metabolomics data. To better understand the ?adiposity-obesity-related? (AOR) mechanisms of CKD and to discover novel biomarkers, we propose to identify distinct subgroups based on AOR attributes and to perform metabolomics analysis on the identified subgroups and CKD outcomes. The overall goal of this research and the training plan is to execute three studies as the subjects of a doctoral dissertation in metabolomics analysis and CKD epidemiology. First, in Aim 1, we will identify latent AOR subgroups in the CRIC CKD population using the data-driven approach of consensus clustering. We will also evaluate the utility and the independent risk discrimination performance of the AOR subgroups with key CKD outcomes (CKD progression, CVD, and death), using Cox regression model with adjustment for known CKD risk factors. Next, in Aim 2, we will perform metabolomics analysis and identify metabolites and metabolomic patterns that are associated with the different AOR subgroups. In response to the statistical challenges in analyzing the high-dimensional metabolomics data, we will implement both the machine learning random-forest algorithm and the conventional univariate and multivariate regression analysis. Finally, in Aim 3, we will investigate the relationship between CKD outcomes and the metabolites and metabolomic patterns identified in Aim 2, using Cox regression model, extended by pathway analysis and mediation analysis. Mining high-dimensional metabolomics data with AOR phenotypes in the CRIC Study holds promise to expand our knowledge of pathophysiology and molecular characterization of CKD. Future studies can be built upon these findings to improve more effective and personalized CKD management in the setting of the obesity epidemic. A rigorous curriculum including didactic and experiential learning in biomedical data mining, statistics, and advanced epidemiology will round out the applicant's training, preparing her to be an independent and collaborative investigator and a kidney disease epidemiologist at the completion of his PhD.
Chronic kidney disease (CKD) and its risk factors of obesity, diabetes, and hypertension are prevalent in the United States. To better understand the mechanism between adiposity-obesity-related CKD mechanism, the proposed research will implement machine learning algorithms, and identify distinct patient subgroups and discover novel CKD biomarkers, using the clinical and high-dimensional metabolomics data from a diverse cohort of CKD patients. The findings of our study hold promise to expand our knowledge of pathophysiology and molecular characterization of CKD, and facilitate personalized disease management.
