S. Goswami,1 S. Yee,1 J. Mosley,2 M. Hedderson,3 M. Kabu,4 S. Maeda,5 D. M. Roden,2 M. D. Simpson,6 K. M. Giacomini,1 R. M. Savic1; 1University of California, San Francisco, CA, 2Vanderbilt University, Nashville, TN, 3Kaiser Permanente Division of Research, Oakland, CA, 4RIKEN Yokohama Institute, Yokohama City, Japan, 5RIKEN Yokohama Institute, Yokohama City, CA, 6Marshfield Clinic Research Foundation, Marshfield, WI

BACKGROUND: Roughly 35% of patients on metformin fail to achieve therapeutic response. To date, pharmacogenetic (PGX) studies have not investigated the dynamics of biomarkers related to Type 2 Diabetes (T2D) (e.g. HbA1c). The goals of this research were to 1) Construct a population pharmacodynamic (POP-PD) model to quantify disease progression (DP) and metformin response and 2) To explore the role of SNPs on patient specific DP.
METHODS: Data from 1,056 T2D patients on metformin with HbA1c measurements (up to 10 years) and genome wide SNP data were used for a nonlinear mixed effects analysis. A multivariate model-based analysis was performed on model parameters to account for demographic covariates. Genetic analysis on DP was performed using machine-learning (ML) algorithms, interrogating genes linked to T2D and metformin response.
RESULTS: The chosen DP model included a disruption of HbA1c homeostasis resulting in a steady and highly variable increase in HbA1c levels over time (yearly median increase = 0.04%, 95% CI (0.01-0.28%)). The overall metformin effect resulted on average, a 13% decrease from baseline HbA1c levels (half-life of effect = 40 days). Body weight and surrogate exposure were significant covariates on the effect magnitude of metformin (P<0.01). Two major genes emerged from the ML analysis and were associated with DP: 1) a SNP in transcription factor TCF7L2 (rs12243326) and 2) a SNP in SLC30A8 (rs13266634), a protein known to colocalize with insulin in granules of INS-1 cells.
CONCLUSION: This PGX study implements a POP-PD linked with a machine learning approach to quantify the extent of DP for T2D patients on metformin. This study is the first to highlight the importance of two genes that are known to be associated with risk for T2D, TCF7L2 and SLC30A8, on DP for patients on metformin.