Identifier: TDX:2845
Authors: Barrilero Regadera, Rubén
Abstract:
1H-NMR serum profiling is especially suitable for high-throughput epidemiological studies and large-scale nutritional studies and drug monitoring. It requires minimal sample manipulation and its quantitative response allows inter-laboratory comparison. A comprehensive 1H-NMR serum profiling consists of three measurements encoding different molecular species: lipoproteins, low-molecular-weight metabolites and lipids. 1H-NMR serum profiling provides information of size, particle number and lipid content of lipoprotein subclasses, as well as abundance of amino acids, glycolysis-related metabolites, ketone bodies, fatty acids and phospholipids, among others. However, the spectral complexity promotes errors in manual data analysis and the multiple molecular interactions within the sample compromise reliable quantifications. Developing robust methods of metabolite serum profiling is therefore desirable to consolidate high-throughput 1H-NMR in the clinical practice. This thesis presents several methodological and computational strategies to that end. In the first study, we developed generalizable regression methods for lipids in routine clinical practice (known as “lipid panel”), to be applied in healthy population and in a wide spectrum of lipid and lipoprotein abnormalities. These standard lipids are still the main measurements of cardiovascular risk and therapy targets. In the second study, we characterised the quantitative errors introduced by protein binding in 1H-NMR profiling of clinically-relevant LMWM in native serum. Then, we proposed a competitive binding strategy to achieve quantifications closer to absolute concentrations, being fully compatible with high-throughput NMR. Finally, the third study presents LipSpin: an open source bioinformatics tool specifically designed for 1H-NMR profiling of serum lipids. Moreover, some methodological aspects to improve NMR-based serum lipid analysis are discussed.
Repositori URV