Review of the V-HEFT I Trial
Effect of vasodilator therapy on mortality in chronic congestive heart failure
N Engl J Med 1986; 314:1547-52
Background Into the mid-1980’s, digoxin and diuretics were the mainstay of chronic disease management for congestive heart failure. Vasodilator agents were also commonly used based on limited data of their favorable hemodynamic effects. No sufficiently powered trials in this space had been performed to assess whether administration of vasodilators or any other agents improved long-term morbidity or mortality for heart failure patients. The V-HEFT trial was undertaken to test the hypotheses that 2 widely used vasodilator regimens (prazosin or a combination of hydralazine and isosorbide dinitrate) were superior for reducing death versus placebo. The trial was sponsored by the Veterans Administration and only enrolled men.
Patients Men between the ages of 18 and 75 were recruited from 11 participating Veterans Administration hospitals and had to have chronic congestive heart failure based on either evidence of cardiac dilatation or left ventricular dysfunction (EF <45%) in association with reduced exercise tolerance. Patients were excluded if 1) exercise tolerance was limited by chest pain rather than breathlessness and fatigue, 2) they had a heart attack within the previous 3 months, 3) substantial obstructive valvular disease, 4) considerable obstructive myocardial disease, 5) chronic pulmonary disease, 6) any disease likely to limit 5-year survival, 7) required long-acting nitrates for angina, 8) required calcium channel blockers, beta-blockers or anti-hypertensive drugs other than diuretics, or 9) if the study drugs were contraindicated.
Baseline characteristics The average age of patients in the trial was 58 years. Over 40% of patients had coronary artery disease and/or previous MI and about 13% had previous bypass surgery. Excess alcohol use as well as hypertension were also present in about 40% of patients and diabetes was present in about 20%. There was a notable difference in the diabetes rates between groups that nearly reached statistical significance (placebo [24.5%], prazosin [18.7%], hydralazine-nitrate [17.2%]). Prior to randomization, about 40% of patients were taking vasodilators, 27% antiarrhythmics, 20% sublingual nitroglycerin and 20% anticoagulants.
The average symptom score of participants was 5.6 on a scale of 4-12. Symptoms included dyspnea, fatigue, orthopnea and PND each scored from 1 (none) to 3 (severe). The average BP was about 119/76 mmHg, HR 81 bpm, and EF 30%.
Procedures Prior to randomization, digoxin and diuretic therapy were adjusted to achieve a digoxin blood level >0.7 ng/mL and euvolemic volume status. Clinical evaluations and exercise-tolerance tests on 2 consecutive visits, two weeks apart, had to reveal clinical and exercise stability before randomization could occur. Following randomization, patients continued to receive the optimal dose of digoxin and diuretic along with 1 of 3 study regimens. The placebo group was given placebo tablets and placebo capsules and instructed to take them 4 times a day. The prazosin group took 2.5 mg prazosin capsules and placebo tablets 4 times a day. The hydralazine-isosorbide dinitrate group took 37.5 mg hydralazine capsules and 20 mg isosorbide dinitrate tablets 4 times a day.
In all groups, therapy began with 1 capsule and 1 tablet to be taken 4 times a day. In the absence of side effects, this was increased to 2 capsules and 2 tablets 4 times a day for a total of 20 mg of prazosin or 300-160 mg of hydralazine-isosorbide dinitrate. If drug-related side effects occurred, the dose could be reduced to half a tablet 4 times per day or to one capsule 2 times per day. If the dose was reduced, an attempt was made later to reinstitute the full dose.
In order to limit dropouts, rigorous criteria were established for “treatment failures.” Physicians were advised to hospitalize patients with worsening symptoms, and, if appropriate, to use temporary intravenous vasodilator or inotropic interventions for stabilization. Physicians were encouraged to resume study medications upon discharge. At least 2 such hospitalizations were required, along with objective evidence of deterioration, before the study medications were discontinued and replaced with known therapy.
Endpoints The primary endpoint was all-cause mortality.
Results 642 patients were enrolled (273 in placebo group, 183 in prazosin group and 186 in the hydralazine-isosorbide dinitrate group). Excluding discontinuations that took place within 1 month before death, 47 patients (17%) discontinued one or both types of placebos, 43 patients (23%) discontinued prazosin, and 60 patients (32%) discontinued either one or both drugs in the hydralazine-isosorbide group. Six months after randomization, the average prescribed doses were 18.6 mg per day of prazosin, 270 mg per day of hydralazine, and 136 mg per day of isosorbide dinitrate. More than 85% of the prescribed drugs were taken in each treatment group.
The mean follow-up was 2.3 years (range 6 months to 5.7 years). Only 4 patients were lost to follow up (2 in placebo group, 1 in prazosin group, and 1 in hydralazine-dinitrate group). There were 120 deaths in placebo group (44%; 19 per 100 patient years), 91 in the prazosin group (50%; 22 per 100 patient years), 72 in the hydralazine-dinitrate group (39%; 17 per 100 patient years). A reduction in mortality over the entire follow-up period was observed in the hydralazine-nitrate group compared with placebo (p = 0.093 on the log-rank test and p = 0.046 on the generalized Wilcoxon test, which gives more weight to treatment differences occurring in the earlier part of the mortality curves and less weight to the latter part, where the numbers are smaller). The absolute difference in mortality between these groups increased during three years and then began to diminish. The absolute difference in mortality between the placebo group and hydralazine-isosorbide groups at years 1 through 4 was 7%, 9%, 11% and 4%, respectively.
Prespecified subgroup analysis in CAD vs no CAD stratification showed no significant treatment effect heterogeneity for hydralazine-nitrate among those with CAD although the absolute difference in mortality between groups was numerically higher for patients with CAD.
At 8 weeks and 1 year, SBP (-4.1 and -4.6 mmHg) and DBP (-3.2 and -2.7 mmHg) decreased the most in the prazosin group compared to placebo. Hydralazine-nitrate was not associated with a statistically significant nor clinically significant difference in BP with exception of DBP at 8 weeks. The EF rose significantly at 8 weeks and 1 year in the hydralazine-nitrate group (+2.9 and +4.2) compared to placebo but not in the prazosin group.
Side effects were reported in 4.0% of placebo patients, 11% of prazosin patients and 19% of hydralazine-nitrate patients, respectively. The most common side effects were headache and dizziness. Headache was reported in 12% of hydralazine-nitrate patients.
Conclusions This study compared the combination of hydralazine-isosorbide dinitrate or prazosin to placebo in patients with chronic congestive heart failure who were optimized on digoxin and diuretic therapy. In what appears to be a young (58 years) and highly selected population of clinically stable, male veterans with dilated cardiomyopathies and low symptom burdens, the combination of hydralazine-isosorbide reduced death by 2 per 100 patient years, increased EF by 4% at 1 year and did not significantly alter BP compared to placebo. Side effects were reported in approximately 1 out of 5 patients with the most common being headache and approximately 1 out of 3 discontinued 1 or both study drugs. Prasozin did not reduce death or increase EF but did reduce BP compared to placebo. The internal validity of the study is high with only a few minor imbalances in baseline characteristics, which do not appear clinically relevant nor to consistently favor any one group. Less than 1% of patients were lost to follow up with no significant imbalances between groups. The external validity is limited by the fact that this is a population of male veterans and the etiologic distribution of cardiomyopathy and heart failure is likely different from a general heart failure population; etiologic causes of death are also likely to be different. Furthermore, the population is highly selected and its unclear how many patients from the general heart failure population would meet study criteria.