Metabolic Biomarkers for Type I Diabetes and Cardiovascular Disease Risk in Adolescents

Bioinformatics Internship Presentation

Alexander Wong (Mentor: Dr. Simina Boca, Innovation Center for Biomedical Informatics, Department of Oncology, Georgetown University)

May 13th, 2016, 10:00pm, Room 1300, Harris Building

Diabetes mellitus type 1 or type 1 diabetes (T1DM) is a disease in which the host’s immune cells destroy the beta cells of the pancreas. Beta cells of the pancreas are involved in the creation of insulin hormone, which helps move glucose into the cells of the body to be used as fuel for normal processes. Studies have shown that cardiovascular disease (CVD) risk is elevated in populations that have diabetes.  The goal of this study was to identify metabolites significantly associated with T1DM and surrogate markers of CVD in adolescents. The study examined blood samples from 64 different adolescents (39 diabetic and 25 Non-diabetic), for which a panel of 185 metabolites was analyzed using FIA and LC-MS. The LDL particle number (LDLp), Apo protein B level, and glycated hemoglobin (blood sugar) levels were also measured on these patients. LDLp and ApoB are surrogate variables for CVD risk, with increased values conferring a higher future risk of CVD, while blood sugar levels are a measure of how well diabetes is controlled.

In order to identify significant metabolites associated with T1DM and CVD risk, we considered four multiple linear regression models for each metabolite, with each model including a variable related to T1DM (either disease status or blood sugar level) and a variable related to CVD risk (either LDLp or ApoB) and all models adjusting for age and sex. Given the large number of statistical tests, we applied a multiple comparison adjustment method (Benjamini-Hochberg procedure) to all of our models, which controlled the false discovery rate (FDR). At a FDR level of 0.1, two metabolites were associated with T1DM when adjusting for LDLp, two metabolites were associated with T1DM when adjusting for ApoB, four metabolites were associated with blood glucose when adjusting for LDLp, and four metabolites were associated with blood glucose when adjusting for ApoB.  We also determined common metabolites significantly associated with the T1DM and CVD risk variables in all four models by comparing the top ten metabolites ranked according to p-value. Sphingomyelin C:16 was the only metabolite within the top 10 which was associated with all the T1DM and CVD variables in all four models. Eight common metabolites were associated with the T1DM variables in all models, and five common metabolites were associated with CVD variables in all models when comparing the top ten metabolites. Thus, our approach is relatively robust in describing both T1DM and CVD risk, by using one of the two available variables for each disease process.