Date of Thesis

Summer 2018


Metabolomics, the study of a set of small molecules (a.k.a. metabolites) present within an organism, is a beneficial area of study in health related fields. Through metabolomics studies, it may be possible to identify important metabolites that correlate with the presence or advancement of a disease; these metabolites are termed biomarkers. Nuclear Magnetic Resonance (NMR) is employed as an analytical technique for the purpose of identifying and quantifying metabolites in human serum samples. NMR, while having low sensitivity, is an advantageous technique due to its ability to give spectra with peak patterns that are unambiguously characteristic of each metabolite and to perform intrinsically reproducible measurements.

This thesis reports first on methodological work that develops a better understanding of the sensitivity of NMR spectroscopy of extracted aqueous metabolites. Several techniques were applied to sample analysis to determine their affects on sensitivity. These techniques included: increased sample volume, alternative solvent extractions, use of specialized NMR tubes, and use of a high field NMR instrument fitted with a cryogenic probe. A surprising result was that the 850 MHz cryogenic probe provided only small gains in sensitivity with these serum samples because the saline present in extracted sera appeared to negate the sensitivity enhancement that otherwise would have been obtained. However, it was also determined that methanol extractions and restricted volume (a.k.a. ShigemiTM) tubes did, in fact, improve sensitivity.

Following methodological and validation work, a large clinical metabolomics study was undertaken. The central hypothesis of metabolomic work is that the composition of metabolites in individuals is sensitive to disease states. Metabolite biomarkers for Type 2 Diabetes (T2D) have been studied extensively, and T2D has shared risk factors with Nonalcoholic Fatty Liver Disease (NAFLD), suggesting that NAFLD may also be associated with systematic metabolite changes. This thesis reports a NMR metabolomics study on bariatric surgery patient serum samples to determine possible metabolite biomarkers for NAFLD. This research confirmed and extended recent work that reported decreased ketone bodies (3-hydroxybutyrate, acetone, acetoacetate) as being associated with NAFLD patients. In contrast to prior work by others, this research more strongly identifies alanine as a sensitive marker of disease progression. We also report the discovery of 2-hydroxyisovalerate as a useful marker of NAFLD progression. Whereas prior work used smaller and more uniform research cohorts with high throughput methods, the combination of larger cohorts with improved high precision NMR profiling has clearly established alanine as a metabolite of interest in NAFLD. An innovative aspect of this work is the use of standard-of-care, biobanked sera to design cohorts that are believed to be significantly less uniform than clinical research cohorts. Observing metabolite markers in this work is therefore highly significant for the potential of metabolomics in patient classification, for which a survey of results will be shown that achieve classification success rates on the order of 70-85%. Future work involving larger cohort studies to further validate known biomarkers and to discern weaker biomarkers is described.


NMR, Metabolomics, Nonalcoholic Fatty Liver Disease

Access Type

Masters Thesis (Bucknell Access Only)

Degree Type

Master of Science



First Advisor

David Rovnyak