Mean and visit-to-visit variability of glycemia and left ventricular diastolic dysfunction: A longitudinal analysis of 3025 adults with serial echocardiography

Objective We aimed to determine the mean glucose thresholds to increase the risk of left ventricular diastolic dysfunction (LVDD) and whether visit-to-visit variability of fasting plasma glucose (FPG) and glycated hemoglobin (A1C) could independently increase the risk in a cohort with serial echocardiography. Methods This was a 3.5-year (range, 0.5–8.3) retrospective longitudinal cohort study of 3025 adults (age, 55.15 ± 7.6 years; without diabetes, n = 2755) with LV ejection fraction > 50% by serial echocardiography between 2006 and 2016. Mean, standard of deviation (SD) and coefficient of variation (CV) of FPG and A1C obtained from three consecutive measurements preceding the first echocardiography. The definition of LVDD in this study was primarily based on early peak mitral inflow velocity and early diastolic mitral annulus motion velocity. Results LVDD developed in 611/3025 subjects (20.2%). Cox proportional hazard models showed increased adjusted hazard ratios (HRs) for incident LVDD in the highest quartile of FPG-mean (HR 1.76, 95% confidence interval [CI]; 1.36–2.30), FPG-SD (HR 1.63, 95% CI; 1.27–2.09), FPG-CV (HR 1.47, 95% CI; 1.15–1.89), and A1C-mean (HR 1.83, 95% CI; 1.41–2.38) versus the lowest quartile, which was consistent even in subjects without diabetes. Mean glucose thresholds for the increased risk were below the lower limits for pre-diabetes. Conclusions In terms of mean glycemia, LVDD may be initiated in the earliest diabetic continuum, and such changes could be measurable within several years. Visit-to-visit variability of FPG, but not that of A1C, predicted accelerated development of LVDD.