Metabolomic Analysis of Changes in the Serum of Diabetic Patients Caused by Ulcers Arsenty Melnikov (Presenter) Internatrional Tomography Center SB RAS Presenter Bio: Arsenty Melnikov is currently an undergraduate student at Physics Department of the Novosibirsk State University (Novosibirsk, Russia). He is working at the ITC SB RAS (Novosibirsk, Russia) in the fields of metabolomics, mass-spectrometry and data analysis since 2016. He is experienced in high-resolution LC-MS techniques and up-to-date statistical methods of data analysis. He is highly skilled in application of programming languages including Python and R for statistical analysis and has experience in univariate and multivariate statistical approaches for metabolomics, such as PCA, PLS-DA, OPLS-DA, SVM-RFE. Current research interest is mainly devoted to applications of metabolomics in analysis of human diseases, especially diabetes and its application.

Authors: Arsenty Melnikov (1,2), Vadim Yanshole (1,2)
(1) International Tomography Center SB RAS, (2) Novosibirsk State University

Patients with diabetes mellitus have approximately a 20% risk of developing an ulcer, 7% cases of which leads to an amputation. Despite widespread occurrence of diabetic ulcers, this complication is poorly studied, making this disease a topical object of study.
Current report provides the data on metabolomics changes in the serum of diabetic patients without complications as compared to the serum of diabetic patients with ulcers. More than 200 metabolites were detected with the use of two independent HPLC-MS method (HILIC and RPLC). The concentrations of several metabolites are found to differ significantly: phospholipids, bilirubin, 1,5–anhydroglucitol, oleamide.

Introduction

There are two main types of diabetes mellitus (DM): type 1 DM is characterized by absolute insulin deficiency whereas type 2 DM is a result of insulin resistance. The result of this disease is the inability of cells to utilize glucose, which leads to increased catabolism of lipids and proteins. This often causes the development of different complications, such as diabetic ulcer. Patients with DM have approximately a 20% risk of developing an ulcer, 7% cases of which leads to an amputation. Despite widespread occurrence of diabetic ulcers, this complication is poorly studied.

At the moment, metabolomic changes caused by the development of diabetes are rather well studied [1, 2], however there are limited number of metabolomic studies on changes in serum of diabetic patients without complications as compared to serum of diabetic patients with ulcers. Current research is devoted to analysis of metabolomic changes caused by the development of diabetic ulcers.

Methods

Two high-performance liquid chromatography methods (HILIC & RPLC) in conjunction with high-resolution ESI-q-TOF mass-spectrometric detection (LC-MS) were used to obtain metabolomic profiles of serum. Feature detection in raw spectra, retention time correction and peak filling were performed using XCMS in R. A self-developed clustering algorithm similar to RAMClust [3] was implemented in Python and used to group features which correspond to the same metabolite. Information on exact mass, fragment ions and isotope distribution were used to annotate chemical compounds. Differences in metabolite abundances were assessed using Mann-Whitney U-test and visualized on volcano plot. Moreover, areas under ROC curves (AUC) were calculated to assess the efficiency of the found metabolites as a biomarker.

Results

1034 features were found using XCMS in the data obtained with HILIC, yielding 180 highly possible metabolites among them by clustering. For the majority of metabolites, their levels in the serum of diabetic patients without complications compared to the serum of diabetic patients with ulcers are similar, with few exceptions including 1,5–anhydroglucitol (fold change = 3.2, p-value = 10-4, AUC = 0.96 ± 0.02), bilirubin (fold change = -3.1, p-value = 10-3, AUC = 0.89 ± 0.03) and a group of phospholipids with the mean fold change = -1.48 ± 0.02 (p-value < 0.05). Although there are some differences in phospholipids HILIC rather poorly retents and separates lipids (including phospholipids) and it could lead to significant deviations in the data. In order to get rid of these deviations RPLS was used (667 features were found yielding 162 metabolites among them). This allowed us to detect more phospholipids with an average fold change = 1.77 ± 0.05 (p-value < 0.05). Also, the concentration of several lipid-related metabolites, including oleamide (fold change = -1.9, p-value = 0.02), have a large difference between the two groups of samples.

Conclusions & Discussion

It is believed that increased catabolism of lipids and proteins in patients with DM leads to the development of a wide range of complications. This study confirms that the development of diabetic ulcers is accompanied by a decrease in the concentrations of phospholipids and some lipid-related metabolites. For example, bilirubin is considered as an antioxidant in relation to lipid peroxidation. 1,5–anhydroglucitol is a validated biomarker of short-term glycemic control [4], but currently it is not clear why concentration of 1,5–anhydroglucitol is higher in serum of patients with diabetic ulcer as compared to diabetic patients without complications. This topic is currently under investigation in our lab.

1. Thomas J Wang, Martin G Larson, et al. Metabolite profiles and the risk of developing diabetes. Nature, Medicine, April 2011, p. 448-454.

2. Anna Floegel, Norbert Stefan, et al. Identification of Serum Metabolites Associated with Risk of Type 2 Diabetes Using a Targeted Metabolomic Approach. Diabetes, Vol. 62, February 2013, p. 639-648.

3. Corey David Broeckling, Fayyaz ul Amir Afsar Minhas, Steffen Neumann, et al. RAMClust: a novel feature clustering method enables spectral-matching based annotation for metabolomics data. Anal. Chem., 2014 Jul 15 ;86(14): 6812-7.

4.T Yamanouchi, N Ogata, T Tagaya, T Kawasak, et al. Clinical usefulness of serum 1,5-anhydroglucitol in monitoring glycaemic control. The Lancet Volume 347, Issue 9014, 1 June 1996, Pages 1514-1518

This work is supported by Russian Foundation for Basic Research (grants №17-03-00656, №18-415-543006)