1H-NMR based metabolomics of urine samples from patients with acute myeloid leukemia and healthy control subjects

Details

Supervisors

Bindels, Laure B. ; Debier, Cathy

Faculty

Faculté des bioingénieurs

Degree label

Master [120] : bioingénieur en sciences agronomiques, à finalité spécialisée

Abstract

Acute Myeloid Leukemia (AML), one of the most frequent leukemias in adults, is characterized by the accumulation and proliferation of immature hematopoietic cells in the bone marrow and blood. Given that the gut microbiota is an accepted component regulating host metabolism and immunity, its role in the treatment of AML is attracting attention. In this context, the MicroAML study was launched, a multi-centric, prospective and observational clinical study, which main objective was to assess the composition and activity of the gut microbiota in patients diagnosed with AML. This master thesis takes part in this project. The main objective of this master thesis was to learn and participate to the implementation of metabolomics of human urine using 1H-NMR spectroscopy and interpret the obtained results. 1H-NMR spectroscopy was used to analyze urine samples from 30 patients with AML and 30 healthy control subjects collected during the MicroAML study. A typical metabolomics pipeline was followed. A special focus was made on the protocol for urine sample preparation, which is important to limit pH and ionic strength variations; on the normalization of spectra, which is crucial to make the spectra comparable by correcting dilution effects; on the alignment of spectra, which enables to correct peak shifts. No standard protocol for urine sample preparation was found in the literature, thus different parameters were tested to optimize it. Literature screening was performed to determine the most adapted normalization method, probabilistic quotient normalization. Different alignment tools were compared but none of them provided fully satisfactory results thus it was decided to not align spectra. An alternative solution to misaligned spectra was found in optimized bucketing. Optimized buckets were analyzed statistically to determine important buckets which, combined with literature screening, led to the establishment of a list of metabolites to screen. Eventually, 69 metabolites were identified, verified, and quantified. Subsequent data cleaning and imputation as well as descriptive, univariate and multivariate statistical analyses were performed. Taurine, carnitine, formate, hypoxanthine and 2-aminobutyrate were found significantly increased between AML patients and healthy controls as well as discriminating between both groups. Those results were then interpreted in the context of the MicroAML study. Simultaneously, a comparison between the results found in this master thesis and the results found in 16 selected articles was carried out. Variations in the general status of cancer patients were found and reflected in urine. AML patients seemed to be subject to greater oxidative stress than healthy control subjects. Metabolic specificities of AML cells were also reflected in urine. Further studies would need to be conducted in a larger cohort to confirm the obtained results.