Pharmacogenetic Regression Tree Analysis of the Vitamin D Intervention Effects in Patients with Type 1 Diabetes

Presentation Number: SAT 340
Date of Presentation: April 2nd, 2016

Klaus Badenhoop*1, Natalie Filmann2, Dimitra Bogdanou3, Marissa Penna-Martinez3, Eva Herrmann4 and Ulrike Koehl5
1Goethe-University Hospital, Frankfurt, Germany, 2University Hospital/Goethe-University Frankfurt, 3University Hospital, Goethe-University Frankfurt am Main, Germany, 4Goethe University Frankfurt am Main, Germany, 5Institute for Molecular and Therapeutics (MHH), Hannover, Germany


Background: Type 1 diabetes mellitus is an autoimmune disease mediated by T cells with regulatory T cell (Treg) defects. Vitamin D deficiency, which is highly prevalent in patients with type 1, may even aggravate immunodeficiency.

Aim of this study is to analyze whether and to what degree pharmacogenetic variation directs the response on vitamin D levels and Tregs in type 1 diabetes patients treated with vitamin D supplementation. This is of particular interest for future treatment approaches in terms of personalized vitamin D supplementation.

Methods: We utilized blood samples from 39 type 1 diabetes patients undergoing a double-blind randomized trial with a sequential cross-over design of 4000 IU vitamin D for three months followed by three months of placebo or vice-versa (EudraCT 2010-022677-34, Bogdanou et al., submitted). Treatment effects were evaluated on the basis of within subject changes between treatment and placebo periods. Regression trees were constructed using recursive partitioning, with an implementation of an unbiased tree algorithm for conditional inference trees (Hothorn et al., 2006).

Results: The vitamin D baseline level had significant impact on vitamin D increase (p<0.001) in interaction with body weight (p=0.07) and a genotype of the glucocorticoid receptor GR rs6198 (p=0.02). Pharmacogenetic variation of the vitamin D system had significant or trend influence on changes in Tregs, where we found a four-way-interaction between VDRA (p=0.008), VDRF (p=0.001), VDRB (p=0.05), and DBP rs7041 (p=0.016). An interaction with the glucocorticoid system was also observed in %monocytes: Here we detected a three way interaction between VDRF (p=0.031), vitamin D increase (p=0.036) and GR rs6198 (p=0.023), and accordingly DBP rs7041 (p=0.06).

All p-values were generated without correction for multiple comparisons.

Discussion: Although obtained from a study cohort with a small sample size, our results confirm and extend previous findings on functional effects of the VDR alleles, in particular VDRF, and on the interaction between genes of the glucocorticoid system and genes of the vitamin D cascade.  In conclusion, studies with more patients and longer treatment periods are needed to validate our results and to extend and deepen the findings to obtain rational means for individualized dosing schemes in vitamin D supplementation for patients with type 1 diabetes.


Nothing to Disclose: KB, NF, DB, MP, EH, UK