分子与遗传医学

Biomarkers for Development of Glucocorticoid-Induced Diabetes Mellitus - A Metabolomics-Based Prediction Model

Abstract

Klarskov CK, Havelund JF, Zegers FD, Færgeman NJ, Schultz HH, Persson F, Debrabant B, Bjergaard UP and Kristensen PL

Background: Glucocorticoid-induced diabetes mellitus (GIDM) is a serious side effect of glucocorticoid (GC) treatment that is associated with both increased mortality and morbidity, but not all patients develop GIDM when treated with GC. The reason is not known, and clinical risk factors predictive of type 2 diabetes do not predict GIDM. Previous metabolomics studies have found specific metabolic disturbances prior to clinical type 2 diabetes. This could also be true for GIDM. The primary aim of this study was to investigate whether distinct metabolic patterns in patients treated with high dose GC can predict development of GIDM.
Material and Methods: Serum from 116 patients about to be treated with or in the first days of treatment with high-dose GC (>100 mg prednisolone equivalent) was analyzed with liquid chromatography-mass spectrometry (LC-MS) based nontargeted metabolomics. Clinical data were collected at baseline and through a 3-week follow-up period. 52 patients developed GIDM and 64 did not (control group). A logistic regression model and a predictive model was build and differences in the metabolome due to treatment with GC was tested in serum from patients without GC treatment (n=6) and patients with GC treatment (n=107).
Results and Discussion: At univariate analysis three metabolites were associated with the development of GIDM. These metabolites could not be annotated to specific metabolites. A multi-metabolite approach could not predict GIDM, and this is different from previous findings in T2DM. This supports the hypothesis that the etiology of T2DM and GIDM is different. The biological significance of our finding remains unknown, but with the rapid development in the field of metabolomics and databases with increasing numbers of characterized metabolites, these metabolites may be identified. Conclusion: Our data indicate that the typical metabolic shifts in T2DM are not the same in GIDM. This supports the hypothesis that GIDM may have a pathophysiology different from T2DM. Furthermore, our data suggest that there is potential for identifying patients at risk of GIDM before clinical manifestation.