2.2.1. Population cohort study: metformin interaction with LVH risk SNP rs9838915

Patients were derived from the Genetics of Diabetes Audit and Research Tayside Scotland (GoDARTS) study, the methods and cohort details have been described in full previously [63]. In brief, the GoDARTS study includes 10,149 patients with type 2 diabetes (DM) and 8,157 controls without DM. Clinical characteristics are collected at baseline and electronic health records are linked using a unique patient identifier for blood samples, prescribing and clinical outcomes. We also linked these patients through the same identifier to the Tayside echocardiography database which contains over 100,000 clinically requested scans. Patients were fully genotyped using the Affymetrix platform as previously described.

In this study we included all DM cases and excluded non-DM controls. Patients were stratified by genotype related to Kruppel-like factor 15 (KLF-15) (rs9838915) which we have previously shown is significantly associated with echocardiographic left ventricular hypertrophy (LVH) [53]. Genotyping and quality control have been described previously [64]. We defined left ventricular hypertrophy (LVH) according to the American Society of Echocardiography (ASE) criteria [65] as previously described [64]. Patients were classified as non-LVH controls if they either had echocardiography performed that confirmed no LVH (LV wall septal and posterior wall thickness <1.2 cm) or had no clinically requested echo and had never had a heart failure hospitalisation.

Logistic regression was used to determine the association of KLF-15 genotypes (GG vs. GA vs. AA) with presence of LVH with adjustment for age, gender, diabetes status, systolic blood pressure and BMI. Interaction testing was performed to determine whether the association of rs9838915 with LVH was different in those patients using metformin at the point of recruitment into the GoDARTS study. We used Angiotensin Converting Enzyme (ACE) inhibitors or Angiotensin II Receptor Blockers (ARB) use as a control group. A 2-sided p value of <0.05 was considered significant and analysis was performed using R version 3.4.3.