A Longitudinal HbA1c Model Elucidates Genes Linked to Disease Progression on Metformin Therapy
Srijib Goswami, Sook Wah Yee, Kathleen M. Giacomini, Radojka M. Savic
University of California, San Francisco
Objectives: Metformin is first-line therapy for Type 2 diabetes, and is one of the most commonly prescribed drugs worldwide [1-3]. Glycosylated hemoglobin (hbA1c) is the primary surrogate biomarker for long-term glycemic control and drug response [4]. There is high variability in both baseline HbA1c and long-term HbA1c dynamics over time [5,6]. Our research aim is to converge multiple genetic methodologies to identify genes linked to the long-term dynamics of metformin response.
Methods: First, we developed a model for disease progression (DP) and metformin response using non-linear mixed-effects modeling. In the second part of the analysis, a genetic model was built using HyperLasso (HL) and model-based methods. A total of 7822 HbA1c measurements from 1056 patients were used to develop the model. Available PK information was taken into consideration in the model structure. Inter-individual variability (IIV) was estimated for baseline HbA1c, the magnitude of metformin’s effect, and the DP parameter. Demographic and clinical covariate selection was performed using a stepwise analysis. For the genetic analysis, variants within 50 kilobases of 267 genes selected from literature were investigated. A penalized regression based approach was used (HL) to select variants statistically associated with individual parameters outputted from the model [7,8]. The top variants from HL were investigated using a model-based approach.
Results: A turnover model with a symptomatic metformin effect on the synthesis rate of HbA1c best characterized the data. In the model structure, KIN increase was influenced by the DP parameter in a nonlinear manner. Metformin surrogate drug exposure (steady-state serum creatinine level) was a significant predictor on the metformin effect parameter. From the HL step, 16 variants were linked to DP, of which 11 were intronic, 1 was missense, and 4 were located within 50 kilobases of the gene. Of the prioritized 16 variants from HL, a model-based approach subsequently selected 9 significant variants within the model structure that had strong effect sizes. The top 9 variants accounted for approximately 1/3 of the entire variability in the DP model parameter. Minor alleles of SNPs in CSMD1 and SLC22A2 were most influential on DP.
Conclusions: Overall, our study has successfully integrated model-based approaches with genetic analyses methods to uncover genes linked to the long-term progression of HbA1c on metformin therapy. Genetic variants in 2 genes: CSMD1, and SLC22A2, were identified as influencers of DP with a potential for genetic interaction.
References:
[1] Scarpello JH, Howlett HC. Metformin therapy and clinical uses. Diab Vasc Dis Res. 2008;5(3):157-167. doi:10.3132/dvdr.2008.027.
[2] Gong L, Goswami S, Giacomini KM, Altman RB, Klein T. Metformin pathways: pharmacokinetics and pharmacodynamics. Pharmacogenet Genomics. 2013;22:820-827. doi:10.1097/FPC.0b013e3283559b22.Metformin.
[3] Viollet B, Guigas B, Sanz Garcia N, Leclerc J, Foretz M, Andreelli F. Cellular and molecular mechanisms of metformin: an overview. Clin Sci (Lond). 2012;122(6):253-270. doi:10.1042/CS20110386.
[4] Hamrén B, Björk E, Sunzel M, Karlsson M. Models for plasma glucose, HbA1c, and hemoglobin interrelationships in patients with type 2 diabetes following tesaglitazar treatment. Clin Pharmacol Ther. 2008;84(2):228-235. doi:10.1038/clpt.2008.2.
[5] Cook MN, Girman CJ, Stein PP, Alexander CM. Initial monotherapy with either metformin or sulphonylureas often fails to achieve or maintain current glycaemic goals in patients with Type 2 diabetes in UK primary care. Diabet Med. 2007;24(4):350-358. doi:10.1111/j.1464-5491.2007.02078.x.
[6] Goswami S, Yee SW, Stocker S, et al. Genetic Variants in Transcription Factors Are Associated With the Pharmacokinetics and Pharmacodynamics of Metformin. Clin Pharmacol Ther. 2014;(February):1-10. doi:10.1038/clpt.2014.109.
[7] Hoggart CJ, Whittaker JC, De Iorio M, Balding DJ. Simultaneous analysis of all SNPs in genome-wide and re-sequencing association studies. PLoS Genet. 2008;4(7):1-8. doi:10.1371/journal.pgen.1000130.
[8] Bertrand J, Balding DJ. Multiple single nucleotide polymorphism analysis using penalized regression in nonlinear mixed-effect pharmacokinetic models. :167-174. doi:10.1097/FPC.0b013e32835dd22c.