Background: Echocardiographic indexes of ventricular function have become indispensable in clinical cardiology but have not been tested as prognostic markers or therapeutic monitors in clinical trials. In two Veterans Administration trials on heart failure (Vasodilator-Heart Failure Trials I and II, V-HeFT I and II), echocardiographic variables were analyzed as predictors and monitors and were compared with other indicators of cardiac performance.
Methods and results: Echocardiograms were recorded before randomization and at follow-up intervals. Baseline measurements of left ventricular internal diameters (LVIDd, LVIDs), wall thickness (THd, THs), radius to thickness ratios (Rd/THd, Rs/THs), and mitral E-point septal separation (EPSS) were evaluated as predictors of mortality individually, in multivariate regression models with each other, and with nonechocardiographic predictors. Within-subject changes were compared between treatment groups. Cumulative survival curves were compared between strata formed by cut-points of EPSS and Rs/THs data. In Cox regression analyses, EPSS, LVIDs, and Rs/THs were significant predictors of mortality. In V-HeFT I, Rd/THd was a predictor in the presence of ejection fraction and peak oxygen uptake. In patients with EPSS > or = 21, there was an 83% increase in mortality in the subgroup of patients with Rs/THs > or = 2.5 compared with Rs/THs < 2.5 (p = 0.003), whereas there was no statistical difference for EPSS < 21. EPSS showed improvement in patients treated with hydralazine-isosorbide dinitrate compared with placebo at 2 and 18 months and a trend toward deterioration between 36 and 66 months. In V-HeFT II, there were no differences between enalapril and hydralazine-isosorbide dinitrate groups at follow-up.
Conclusions: Echocardiographic variables, EPSS, LVIDs, and Rs/THs were shown to be predictors of mortality and monitors of treatment for heart failure in clinical trials.