Background: Thiazide diuretics, a commonly used antihypertensive, may cause QT interval (QT) prolongation, a risk factor for highly fatal and difficult to predict ventricular arrhythmias. We therefore examined whether common genetic variants modified the association between thiazide diuretic use and QT or its component parts: QRS interval (QRS), JT interval (JT).
Methods: For each outcome (QT, QRS, JT), 15 participating CHARGE PWG studies conducted genome-wide, ancestry-stratified analyses of European (66%), African (15%), and Hispanic (19%) descent populations (N=78,199). QT, QRS, and JT were measured using standard 12-lead electrocardiograms and diuretic use was assessed with medication inventories. Models leveraged longitudinal data when available, adjusted for heart rate and clinical covariates, and incorporated corrected standard errors to account for underestimation of variance. Ancestry-specific and trans-ethnic interaction effects for approximately 2.5 million HapMap2 SNPs were combined using inverse-variance weighted meta-analysis in METAL. We also conducted combined-phenotype meta-analysis of the t-statistic, weighted by sample size, across all three traits simultaneously using CPASSOC. Finally, we evaluated evidence for gene and pathway enrichment in pre-defined and biologically motivated ion transport, drug metabolism, heart or renal function, and transcription/translation pathways using MAGENTA.
Results: The interaction effects for 31 loci were genome-wide significant (P<5x10-8) following combined-phenotype meta-analysis but not single-trait analysis, including SGK1 (P=3x10-11), which encodes a kinase that regulates potassium, sodium, and calcium transport and KCNN2 (P=2x10-10), which encodes a calcium-activated potassium channel. Of note, the interaction effect for the KCNN2 index signal demonstrated opposite directions for JT (-2.56, 95% confidence interval (CI): -5.11, -0.01) and QRS (1.55, 95% CI: 0.18, 2.92), components which represent opposite segments of QT. Furthermore, we detected evidence of enrichment for genes associated with transcription/translation (ID1, CHRAC1, DHX15) and ion homeostasis (KCTD4, PKP2, MFN2, ALPL) as well as enrichment for ion transport pathways, including voltage gated calcium channels, sodium ion transport, and ligand-gated ion channel activity.
Conclusions: Our results suggest that loci involved in the control of ion transport modify the association between thiazide diuretic use and QT, QRS, and JT. Additional work is needed to validate these findings and identify the underlying variants. Furthermore, the potentially limited power to detect interaction effects in single-trait analyses suggests that innovative statistical methods are warranted in future efforts evaluating gene-environment interaction.