Blood Metabolomic Measures Associate With Present and Future Glycemic Control in Type 2 Diabetes

Leen M. 't Hart, Nicole Vogelzangs, Dennis O. Mook-Kanamori, Adela Brahimaj, Jana Nano, Amber A. W. A. van der Heijden, Ko Willems van Dijk, Roderick C. Slieker, Ewout W. Steyerberg, M. Arfan Ikram, Marian Beekman, Dorret I. Boomsma, Cornelia M. van Duijn, P. Eline Slagboom, Coen D. A. Stehouwer, Casper G. Schalkwijk, Ilja C. W. Arts, Jacqueline M. Dekker, Abbas Dehghan, Taulant MukaCarla J. H. van der Kallen, Giel Nijpels, Marleen M. J. van Greevenbroek

Research output: Contribution to journalArticleAcademicpeer-review


Objective: We studied whether blood metabolomic measures in people with type 2 diabetes (T2D) are associated with insufficient glycemic control and whether this association is influenced differentially by various diabetes drugs. We then tested whether the same metabolomic profiles were associated with the initiation of insulin therapy. Methods: A total of 162 metabolomic measures were analyzed using a nuclear magnetic resonance-based method in people with T2D from four cohort studies (n = 2641) and one replication cohort (n = 395). Linear and logistic regression analyses with adjustment for potential confounders, followed by meta-analyses, were performed to analyze associations with hemoglobin A1c levels, six glucose-lowering drug categories, and insulin initiation during a 7-year follow-up period (n = 698). Results: After Bonferroni correction, 26 measures were associated with insufficient glycemic control (HbA1c >53 mmol/mol). The strongest association was with glutamine (OR, 0.66; 95% CI, 0.61 to 0.73; P = 7.6 × 10-19). In addition, compared with treatment-naive patients, 31 metabolomic measures were associated with glucose-lowering drug use (representing various metabolite categories; P ≤ 3.1 × 10-4 for all). In drug-stratified analyses, associations with insufficient glycemic control were only mildly affected by different glucose-lowering drugs. Five of the 26 metabolomic measures (apolipoprotein A1 and medium high-density lipoprotein subclasses) were also associated with insulin initiation during follow-up in both discovery and replication. The strongest association was observed for medium high-density lipoprotein cholesteryl ester (OR, 0.54; 95% CI, 0.42 to 0.71; P = 4.5 × 10-6). Conclusion: Blood metabolomic measures were associated with present and future glycemic control and might thus provide relevant cues to identify those at increased risk of treatment failure.
Original languageEnglish
Pages (from-to)4569-4579
JournalThe Journal of clinical endocrinology and metabolism
Issue number12
Publication statusPublished - 2018

Cite this