Improving Mass Spectrometry-Based Metabolite Identification and Quantification and Application to Cardiovascular Disease
High-throughput molecular profiling is being increasingly applied to identify novel biomarkers and mechanisms of health and disease. One such application is the use of mass spectrometry-based metabolomic profiling in cardiovascular disease (CVD), whose underlying pathophysiology and risk prediction models are incompletely understood. Two general approaches have been taken in these applications: targeted and non-targeted profiling. The targeted approach identifies and quantifies select known or potential biomarkers in CVD, often via isotope-labeled internal standards. The non-targeted approach attempts to profile the full spectrum of the metabolome, with identification of metabolites aided by existing spectral libraries. In contrast to many successful applications of targeted metabolomics to CVD, early applications of non-targeted profiling have resulted in several pitfalls due to lack of rigor in study design, immature technologic platforms, and challenges in metabolite identification and quantification both at the experimental and computational level. These pitfalls highlight the importance of experimental design and method development in non-targeted metabolomic profiling. The overall goal of this dissertation is to improve methods in non-targeted metabolomic profiling both at the experimental and computational level, and apply these improved methods to CVD human studies. Specifically, this dissertation aims to: 1) identify and modify factors that could influence metabolite identification and quantification in GC-MS based non-targeted profiling at the experimental level; (2) apply emerging methods for metabolite identification at the computational level to generate hypotheses for unknowns; and (3) apply metabolomic profiling to studies in human cardiovascular disease, using the refined methods from the first two aims.
For the first aim, we sought to identify and modify factors in GC-MS-based metabolomic profiling of human plasma that could influence metabolite identification and quantification at the experimental level. Our experimental design included two studies: 1) the limiting-dilution study, which investigated the effects of laboratory preparation and analysis on analyte identification and quantification, and 2) the concentration-specific study, which compared the optimal plasma extract volume established in the first study with the volume used in the current institutional protocol. We tested and confirmed our hypothesis that contaminants, concentration, intra- and inter-experiment variability are major factors influencing metabolite identification and quantification. In addition, we established methods for improved metabolite identification and quantification, which were summarized into recommendations for experimental design of GC-MS-based profiling of human plasma.
For the second aim, we applied emerging methods for metabolite identification level to generate hypotheses for unknowns at the computational level. Specifically, we tested and confirmed our hypothesis that integrating genomic, transcriptomic, and metabolomic data could generate hypotheses for unknowns. Combining the strengths of multiple omics platforms and metabolomic databases, we were able to generate hypotheses for three unknown metabolites implicated in CVD at the computational level.
For the third aim, we applied metabolomic profiling to two studies of CVD and tested the hypothesis that application of the refined methods developed in the first two aims of this dissertation are useful in CVD biomarker and mechanism discovery. In one study, we used heart failure with preserved ejection fraction (HFpEF) as a model to demonstrate the power of targeted metabolomic profiling in testing existing hypotheses of CVD biomarkers and mechanisms. In a second study, we used incident CVD events as a model to 1) apply the refined methods from the first two aims of this dissertation, and 2) demonstrate the power of non-targeted metabolomic profiling in generating novel hypotheses of CVD biomarkers and mechanisms.
This dissertation contributes to research in metabolomics and CVD in several ways. The most significant contribution is the set of recommendations for experimental design in non-targeted metabolomics, which has been incorporated into the workflow of non-targeted profiling at the Duke Molecular Physiology Institute for future studies. Additional contributions include the following: hypotheses for three unknowns implicated in incident CVD events, and novel biomarkers and mechanisms implicated in HFpEF and CVD. Future directions from this dissertation include the following: 1) application of the same principles to method development and validation of metabolomic profiling using other analytical technologies; 2) experimental validation of the hypotheses for unknowns generated by this dissertation; and 3) functional validation of the biomarkers and mechanisms implicated in CVD at the experimental level.
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 3.0 United States License.
Rights for Collection: Duke Dissertations
Works are deposited here by their authors, and represent their research and opinions, not that of Duke University. Some materials and descriptions may include offensive content. More info