TABLE OF CONTENTS

Listing of Tables 6

Listing of Figures 8

1.0 Synopsis 10

2.0 Introduction 13

2.1 Public Health Impact of Heart Failure in Black Patients 13

2.2 Current Treatment Strategies in Black Patients with Heart Failure 13

2.3 Pathophysiology of Heart Failure in Black Patients 14

2.4 Effect of Isosorbide Dinitrate and Hydralazine on Vascular

Nitric Oxide 15

2.5 Regulatory Background of BiDil® (Isosorbide Dinitrate and

Hydralazine Hydrochloride) 16

3.0 Large-Scale Controlled Clinical Trials with

Isosorbide Dinitrate and Hydralazine 18

4.0 Trials with Isosorbide Dinitrate and

Hydralazine Administered as Individual Drugs 19

4.1 Vasodilator Heart Failure Trial I (V-HeFT I 19

4.1.1 Study Overview 19

4.1.2 Study Organization 19

4.1.3 Study Population 20

4.1.3.1 Inclusion Criteria 20

4.1.3.2 Exclusion Criteria 20

4.1.4 Study Plan 20

4.1.5 Study Assessments 21

4.1.5.1 Major Endpoints 21

4.1.5.2 Minor Endpoints 21

4.1.5.3 Safety Assessments 22

4.1.6 Statistical Plan and Analyses 22

4.1.6.1 Sample Size Determination and Interim Monitoring Plan 22

4.1.6.2 Statistical Analyses 22

4.1.7 Results 22

4.1.7.1 Baseline Characteristics 22

4.1.7.2 Patient Disposition and Exposure to Study Medication 24

4.1.7.3 Efficacy Results 24

4.1.7.3.1 Overall Mortality 24

4.1.7.3.2 Retrospective Subgroup Analysis of Mortality 26

4.1.7.3.3 Hospitalizations 29

4.1.7.3.4 Maximum Oxygen Consumption at Peak Exercise 33

4.1.7.3.5 Left Ventricular Ejection Fraction 34

4.1.7.3.6 Other Endpoints 35

4.1.7.4 Safety Results 35

TABLE OF CONTENTS (Continued)

4.1.7.4.1 Adverse events regardless of relationship to study

drug 35

4.1.4.4.2 Adverse events leading to permanent withdrawal

of study drug 36

4.1.7.4.3 Other safety topics 37

4.1.8 Summary and Conclusions for V-HeFT I 37

4.2 Vasodilator Heart Failure Trial II (V-HeFT II) 39

4.2.1 Study Overview 39

4.2.2 Study Organization 39

4.2.3 Study Population 40

4.2.3.1 Inclusion Criteria 40

4.2.3.2 Exclusion Criteria 40

4.2.4 Study Plan 41

4.2.5 Study Assessments 41

4.2.5.1 Major Endpoints 41

4.2.5.2. Minor Endpoints 42

4.2.5.3 Safety Assessments 42

4.2.6 Statistical Plan and Analyses 42

4.2.6.1 Sample Size Determination and Interim Monitoring Plan 42

4.2.6.2 Statistical Analyses 42

4.2.7. Results 43

4.2.7.1 Baseline Characteristics 43

4.2.7.2 Patient Disposition and Exposure to Study Medication 45

4.2.7.3 Efficacy Results 46

4.2.7.3.1 Overall Mortality 46

4.2.7.3.2 Retrospective Subgroup Analysis of Mortality 47

4.2.7.3.3 Hospitalizations 51

4.2.7.3.4 Maximum Oxygen Consumption at Peak

Exercise 54

4.2.7.3.5 Quality of Life 56

4.2.7.3.6 Left Ventricular Ejection Fraction 56

4.2.7.3.7 Other Endpoints 57

4.2.7.4 Safety Results 57

4.2.7.4.1 Adverse events regardless of relationship to

study drug 57

4.2.7.4.2 Adverse events leading to permanent withdrawal

of study drug 58

4.2.7.4.3 Other safety topics 59

4.2.8 Summary and Conclusions for V-HeFT II 60

TABLE OF CONTENTS (Continued)

5.0 Trials with Isosorbide Dinitrate and

Hydralazine Administered as a Combination Product 63

5.1 African American Heart Failure Trial (A-HeFT) 63

5.1.1 Study Overview 63

5.1.2 Study Organization 63

5.1.3 Study Population 64

5.1.3.1 Inclusion Criteria 65

5.1.3.2 Exclusion Criteria 65

5.1.4 Study Plan 66

5.1.5 Study Assessments 66

5.1.5.1 Primary Efficacy Endpoint 66

5.1.5.2 Secondary Efficacy Variables 67

5.1.5.3 Safety Assessments 68

5.1.6 Statistical Plan and Analyses 68

5.1.6.1 Sample Size Determination and Interim Monitoring Plan 68

5.1.6.2 Actions of the Data and Safety Monitoring Board 69

5.1.6.3 Statistical Analyses 71

5.1.6.3.1 Primary Endpoint 71

5.1.6.3.2 Secondary Endpoints 72

5.1.7 Results 72

5.1.7.1 Baseline Characteristics 72

5.1.7.2 Patient Disposition and Exposure to Study Medication 74

5.1.7.3 Efficacy Results 76

5.1.7.3.1 Primary Efficacy Analysis 76

5.1.7.3.2. Secondary Endpoints – Components of

Composite Score 79

5.1.7.3.3. Mortality 79

5.1.7.3.4 Hospitalizations for Heart Failure 82

5.1.7.3.5 Quality of Life 84

5.1.7.3.6 Other Secondary Endpoints 86

5.1.7.3.6.1 Total Number of Hospitalizations

and Hospital Days 86

5.1.7.3.6.2 Newly Recognized Need for Cardiac

Transplantation 87

5.1.7.3.6.3 Total Number of Emergency Room and

Office Visits 87

5.1.7.3.6.4 Echocardiographic Evaluation of Left

Ventricular Function 87

5.1.7.3.6.5 Brain Natriuretic Peptide 87

5.1.7.4 Safety Results 88

5.1.7.4.1 Adverse events regardless of relationship to

study drug 88

TABLE OF CONTENTS (Continued)

5.1.7.4.2 Serious adverse events regardless of relationship to study drug 90

5.1.7.4.3 Adverse events leading to permanent withdrawal

of study drug 92

5.1.7.4.4 Other safety topics 93

5.1.8 Summary and Conclusions for A-HeFT 94

6. 0 Summary of Hydralazine and Isosorbide

Dinitrate for Heart Failure 96

6.1 Rationale for Combining Hydralazine and Isosorbide Dinitrate 96

6.2 The V-HeFT I Trial 97

6.3 The V-HeFT II Trial 101

6.4 The A-HeFT Trial 106

6.5 Consistency of Findings in the V-HeFT and A-HeFT Trials 108

6.5.1 Consistency of Effect on Survival 108

6.5.2 Consistency of Effect on Hospitalizations for Heart Failure 109

6.5.3 Consistency of Effect on Quality of Life 110

7.0 Conclusions 111

8.0 References 112

LISTING OF TABLES

Table 1. Characteristics of Major Trials with Isosorbide Dinitrate and

Hydralazine in Heart Failure 18

Table 2. Baseline Demographic and Clinical Characteristics in V-HeFT I 23

Table 3. Effects on All-Cause Mortality; V-HeFT I 24

Table 4. Changes in Maximal Oxygen Consumption Relative to

Baseline; V-HeFT I 33

Table 5. Changes in Left Ventricular Ejection Fraction; V-HeFT I 34

Table 6. Patients with Adverse Events 36

Table 7. Baseline Demographic and Clinical Characteristics; V-HeFT II 44

Table 8. Effects on All-Cause Mortality; V-HeFT II 46

Table 9. Changes in Maximal Oxygen Consumption; V-HeFT II 55

Table 10. Quality of Life in Black Patients; V-HeFT II 56

Table 11. Left Ventricular Ejection Fraction; V-HeFT II 57

Table 12. Patients with Adverse Events; V-HeFT II 58

Table 13. Effect of Race on Change in Systolic Blood Pressure with

Enalapril and ISDN/HYD; V-HeFT II 59

Table 14. Baseline Demographic and Clinical Characteristics; A-HeFT 73

Table 15. Baseline Cardiovascular History and Treatment; A-HeFT 74

Table 16. Patients on Study Drug at Various Time Points [n (%)]; A-HeFT 75

Table 17. Mean Number of Study Drug Tablets Prescribed Per Day at

Various Times 75

Table 18. Primary Efficacy Endpoint; A-HeFT 76

Table 19. Component Scores for Primary Efficacy Endpoint; A-HeFT 76

Table 20. Effect of BiDil® on All-Cause Mortality; A-HeFT 79

Table 21. Mode of Death; A-HeFT 80

Table 22. Effect of BiDil® on Risk of Hospitalization for Heart Failure; A-HeFT 82

Table 23. All-Cause Mortality or Hospitalization for Heart Failure; A-HeFT 83

Table 24. Change in Overall, Emotional, and Physical Scores in Minnesota

Living with Heart Failure Questionnaire at Six Months; A-HeFT 85

Table 25. Change in Overall, Emotional and Physical Scores in Minnesota

Living with Heart Failure Questionnaire at Endpoint; A-HeFT 85

Table 26. Hospitalizations for Heart Failure; A-HeFT 86

Table 27. Hospitalizations for Any Reason in A-HeFT 87

Table 28. Overview of Patients with Adverse Events; A-HeFT 88

Table 29. Adverse Events Occurring in ≥ 2% of Patients in Either Group 89

Table 30. Serious Adverse Events Occurring in ≥ 1% of Patients

in Either Group 91

Table 31. Adverse Events Occurring in ≥ 0.4% of Patients in Either Group and

Leading to Permanent Discontinuation of Study Drug 92

Table 32. Mean Change in Heart Rate, Systolic Blood Pressure and Diastolic

Blood Pressure (BP); A-HeFT 93

LISTING OF TABLES (Continued)

Table 33. Characteristics of Major Trials with ISDN/HYD in Heart Failure 97

Table 34. Effect of ISDN/HYD on All-Cause Mortality in Black Patients

Patients with Heart Failure 109

Table 35. Change in Quality of Life Produced by ISDN/HYD in Black Patients

with Heart Failure at Six and Twelve Months;

V-HEFT II and A-HeFT 110

LISTING OF FIGURES

Figure 1. Kaplan-Meier Time-to-Event Curves for All-Cause

Mortality; V-HeFT I 25

Figure 2. Hazard Ratios and 95% Confidence Intervals for Effect of

ISDN/HYD on All-Cause Mortality in Subgroups; V-HeFT I 27

Figure 3. Kaplan-Meier Time-to-Event Curves for All-Cause

Mortality in Black Patients; V-HeFT I 28

Figure 4. Kaplan-Meier Time-to-Event Curves for All-Cause

Mortality in White Patients; V-HeFT I 29

Figure 5. Kaplan-Meier Time to First Heart Failure Hospitalization – All

Patients; V-HeFT I 30

Figure 6. Kaplan-Meier Time to First Heart Failure Hospitalization – Black

Patients; V-HeFT I 31

Figure 7. Kaplan-Meier Time to First Heart Failure Hospitalization – White

Patients; V-HeFT I 32

Figure 8. Kaplan-Meier Time-to-Event Curves for All-Cause

Mortality; V-HeFT II 47

Figure 9. Hazard Ratios and 95% Confidence Intervals for Effect of

ISDN/HYD on All-Cause Mortality in Subgroups; V-HeFT II 48

Figure 10. Kaplan-Meier Time-to-Event Curves for All-Cause

Mortality in Black Patients; V-HeFT II 49

Figure 11. Kaplan-Meier Time-to-Event Curves for All-Cause

Mortality in White Patients; V-HeFT II 50

Figure 12. Time to First Heart Failure Hospitalization – All

Patients; V-HeFT II 52

Figure 13. Time to First Heart Failure Hospitalization – Black

Patients; V-HeFT II 53

Figure 14. Time to First Heart Failure Hospitalization – White 54

Figure 15. Effect of BiDil® on Composite Score in Subgroups (Mean + 95%CI) 78

Figure 16. Kaplan-Meier Time-to-Event Curves for All-Cause

Mortality; A-HeFT 79

Figure 17. Hazard Ratios and 95% Confidence Intervals for Effect of BiDil®

on All-Cause Mortality in Subgroups; A-HeFT 81

Figure 18. Kaplan-Meier Time-to-Event Curves for Heart Failure

Hospitalization; A-HeFT 82

Figure 19. Kaplan-Meier Time-to-First Event Analysis of All-Cause Mortality or

Hospitalization for Heart Failure; A-HeFT 83

Figure 20. Mean Change in Minnesota Living with Heart Failure Questionnaire

Overall score at Each Visit and at Endpoint; A-HeFT 84

Figure 21. Kaplan-Meier Time-to-Event Curves for All-Cause

Mortality in Black Patients; V-HeFT I 99

Figure 22. Kaplan-Meier Time-to-Event Curves for All-Cause

Mortality in White Patients; V-HeFT I 100

LISTING OF FIGURES (Continued)

Figure 23. Kaplan-Meier Time-to-Event Curves for All-Cause

Mortality in Black Patients; V-HeFT II 103

Figure 24. Kaplan-Meier Time-to-Event Curves for All-Cause

Mortality in White Patients; V-HeFT II 104

Figure 25. Kaplan-Meier Time-to-Event Curves for All-Cause

Mortality; A-HeFT 108

1.0 Synopsis

This document summarizes the clinical data presented in New Drug Application # 20-727 for BiDil® (isosorbide dinitrate and hydralazine hydrochloride) Tablets, in which NitroMed requests approval for its use to treat black patients with heart failure. This NDA is based on three clinical trials; V-HeFT I, V-HeFT II and A-HeFT. The previous sponsor of this NDA, Medco Research, submitted the overall results for V-HeFT I and V-HeFT II in the original NDA and received a not approvable letter in 1997 requesting additional clinical research. NitroMed subsequently acquired the NDA and determined in consultation with heart failure experts and FDA that the clinical trial most likely to yield a positive outcome was a trial in black patients with heart failure. This briefing document describes the scientific rationale and regulatory history of BiDil® and presents the data from the three trials that form the basis of the NDA.

Heart failure, a serious, progressive and debilitating condition, presents a particular burden in the black community. Black patients suffer disproportionately from heart failure, are diagnosed at younger ages and appear to respond less well to currently approved therapies. In spite of the introduction of generally effective heart failure treatments, e.g., angiotensin-converting enzyme (ACE) inhibitors and beta blockers, a need for additional, more effective treatments for black patients remains for the following reasons:

· The current number of black patients diagnosed with heart failure in the United States of America is 725,000 and it is expected to grow to 900,000 by the end of the decade. Mortality rates for heart failure remain high and patients living with heart failure suffer from incapacitating symptoms and repeated hospitalizations.

· Current treatments for black patients with heart failure are insufficient. Published data suggest that ACE inhibitors may be less effective in black patients, and may cause higher rates of angioedema in this population. Beta blockers have generally not been tested in a meaningful number of black patients with heart failure; in addition, when used in the treatment of hypertension, beta blockers have a known attenuation of effectiveness in black patients.

Given the need for additional therapies for heart failure, three studies have evaluated the safety and efficacy of the combination of isosorbide dinitrate and hydralazine hydrochloride (ISDN/HYD; proposed trade name BiDil®) in this condition. The first study, the first Vasodilator Heart Failure Trial (V-HeFT I) evaluated the safety and efficacy of ISDN/HYD against placebo, added to standard therapy, in the treatment of heart failure. At the time this study was conducted, the standard of therapy for the treatment of heart failure was digitalis glycosides and diuretics. The second Vasodilator Heart Failure Trial (V-HeFT II) tested the ISDN/HYD combination against enalapril, added to standard therapy; standard therapy was again generally digitalis glycosides and diuretics. Initial results from these two trials were suggestive of a benefit in heart failure patients of all races but the dataset was not adequate to gain FDA approval.

Retrospective reanalyses of the results from V-HeFT I and II generated the hypothesis that the observed benefit of the ISDN/HYD combination occurred primarily in black patients. This finding was consistent with a growing body of scientific literature suggesting that ethnic groups might respond differently to certain treatments. Particularly in the cardiovascular area, differential responses to treatments for hypertension and heart failure have been observed and documented.

The principal investigator of V-HeFT I and V-HeFT II, Dr. Jay Cohn, and the current sponsor of the BiDil® NDA, NitroMed, approached FDA with these reanalyses in 1999 and 2000 and together developed a clinical plan to test the hypothesis. The discussions with FDA resulted in a letter to NitroMed in 2001 stating that “[g]iven the subset finding and the overall trend toward a survival effect in V-HeFT I, we believe a single, clearly positive study in a black CHF population would be a basis for approval of BiDil for the treatment of heart failure in blacks.” That trial, the African American Heart Failure Trial (A-HeFT), provides the largest body of evidence on BiDil® in black patients with heart failure, and strongly confirms and extends the results of V-HeFT I and II in black patients with heart failure. Taken as a whole, the results from these three trials support approval of BiDil® to treat heart failure in black patients.

As described in detail in this document, the three trials supporting approval of BiDil® used a variety of endpoints, and were compared to a different standard of therapy in each case. However, V-HeFT I, V-HeFT II and A-HeFT show consistency in their overall key findings.

· Black patients with heart failure treated with ISDN/HYD experience a meaningful reduction in relative risk of mortality.

o In A-HeFT when BiDil® was compared to placebo, black patients demonstrated a 43% reduction in relative risk of mortality (p=0.012).

o In V-HeFT I retrospective analysis showed that black patients demonstrated a 47% reduction in risk of mortality relative to placebo (p=0.04).

o In V-HeFT II, ISDN/HYD did not demonstrate a significant reduction in risk in mortality in all patients when compared to the active drug enalapril. However, retrospective analysis demonstrated that the hazard ratio for the mortality results in black patients in the two treatment arms was approximately 1, suggesting that BiDil was as effective as enalapril in this patient population.

In all three studies both ISDN/HYD and the comparator were added to standard heart failure therapy. When V-HeFT I and V-HeFT II were performed, standard therapy was digitalis glycosides and diuretics; during A-HeFT, standard therapies generally included ACE inhibitors/angiotensin receptor blockers, beta blockers, and/or aldosterone antagonists along with digitalis glycosides and diuretics.

These mortality results therefore demonstrate that BiDil® has a positive effect on survival over the heart failure spectrum and across a wide range of symptoms and background medications.

· Black patients with heart failure treated with ISDN/HYD experience a meaningful reduction in hospitalizations for heart failure.

o In A-HeFT, patients in the BiDil® group demonstrated a 39% decrease in relative risk of first hospitalization for heart failure (relative to placebo). The mean number of hospitalizations for heart failure per patient and mean the number of days hospitalized for heart failure as a percentage of days on study were both statistically decreased in the BiDil® group relative to placebo.

o In V-HeFT I in black patients, at the end of the first year, the cumulative heart failure hospitalization rate was 17.2% in placebo patients but only 6.3% in ISDN/HYD patients as determined by retrospective analysis.

o In V-HeFT II in black patients, at the end of the first year, the cumulative heart failure hospitalization rate was 5.0% in ISDN/HYD patients versus 13.1% in enalapril patients as determined by retrospective analysis.

The concordance of these findings in patients over the heart failure spectrum supports the conclusion that BiDil® reduces the risk of hospitalization for heart failure in black patients across a wide range of symptoms and background medications.

· Black patients with heart failure treated with ISDN/HYD experience an improvement in quality of life.

o An improvement in quality of life was observed in both A-HeFT and V-HeFT II as measured by patient questionnaires. (A quality of life patient questionnaire was not used when V-HeFT I was performed.) The improvement in quality of life produced in A-HeFT by BiDil® in black male and female patients with moderate-to-severe heart failure generally treated with ACE inhibitors/ARBs, beta blockers and/or aldosterone antagonists as well as digitalis glycosides and diuretics was concordant with the improvement in quality of life seen in V-HeFT II with a combination of ISDN/HYD in black men with mild-to-severe heart failure generally receiving only digitalis glycosides and diuretics.

The concordance of these findings in patients over the heart failure spectrum supports the conclusion that BiDil® improves the symptoms of heart failure that impair quality of life in black patients across a wide range of symptoms and background medications.

· The results of these three studies demonstrate that the combination of isosorbide dinitrate and hydralazine is safe and generally well tolerated as treatment for heart failure. The most common adverse events observed among patients receiving the ISDN/HYD combination in all three studies were headache, dizziness and other vasodilator-type reactions.

Based on the results of the V-HeFT I, V-HeFT II and A-HeFT clinical studies, NitroMed proposes the following indication for the fixed-dose combination BiDil® (isosorbide dinitrate and hydralazine hydrochloride) Tablets:

BiDil® is indicated for the treatment of heart failure in black patients. BiDil® has been shown to reduce the risk of mortality from any cause, to reduce the risk of heart failure hospitalization and to improve quality of life.

2.0 Introduction

2.1 Public Health Impact of Heart Failure in Black Patients

Heart failure afflicts nearly 5 million patients in the United States and is responsible for considerable disability and loss of life. Heart failure is a particularly important public health concern in African Americans (also referred to as black persons). Approximately 3% of all black persons have heart failure, and it is estimated that 25-30% of all patients with heart failure in the United States are black.^1,2

Several studies suggest that the black population may be disproportionately affected by heart failure, as compared with the non-black population.³ At the time of diagnosis, black patients frequently are younger, have more advanced left ventricular impairment, and more advanced clinical symptoms than non-black patients.^4-7 Some reports suggest that once the diagnosis of heart failure is made, black patients frequently receive less than optimal care, and this may lead to a higher risk of hospitalization and a higher risk of death.^8-14

2.2 Current Treatment Strategies in Black Patients with Heart Failure

To make matters more complicated, the optimal treatment of heart failure in black patients with the disease has not been clearly defined. Black patients are frequently under-represented in large-scale clinical trials, particularly those carried out primarily or exclusively outside of the United States.¹⁵ Even in trials conducted in this country, the number of black patients in each trial has generally been so small that estimates of the magnitude of treatment responses in this subgroup have been very imprecise. This has led to considerable uncertainty about the benefit to risk relation in black patients of many widely-used treatment strategies. For example,

• In a large-scale heart failure trial with an angiotensin-converting enzyme (ACE) inhibitor that enrolled black patients (Studies of Left Ventricular Dysfunction; SOLVD trial), black patients responded less favorably to ACE inhibition than non-black patients.¹⁶ In this trial, enalapril reduced mortality similarly in black and non-black patients, but reduced the risk of hospitalization less effectively in black than non-black patients. This finding was consistent with the attenuation of responsiveness to ACE inhibitors in black patients when these drugs are used for the treatment of hypertension.¹⁷ Furthermore, not only may ACE inhibitors be less effective in black patients, but they are known to cause potentially life-threatening angioedema more frequently in black patients than in non-black patients.¹⁸

• In a large trial with a beta blocker carried out in the United States (BEST trial), black patients responded less favorably to beta-blockade than non-black patients.¹⁹ In this trial, bucindolol reduced the risk of death in non-black patients but increased the risk of death in black patients. This finding was consistent with the attenuation of effectiveness of beta blockers in black patients when these drugs are used for the treatment of hypertension.²⁰ Although a trial with carvedilol showed benefits with the drug in black patients with heart failure,²¹carvedilol may exert pharmacological effects beyond beta-blockade that may contribute to its efficacy in heart failure.²² The efficacy of other beta blockers (metoprolol and bisoprolol) in black persons with heart failure remains uncertain, since trials with these agents failed to enroll meaningful number of black patients.^23,24

• Aldosterone antagonists have recently emerged as a treatment for heart failure, but the trials carried out with eplerenone in early-stage post-infarction heart failure and with spironolactone in late-stage patients with severe heart failure failed to enroll meaningful numbers of black patients.^25,26 Although eplerenone has been reported to be equally effective in reducing blood pressure in black and non-black patients with hypertension,²⁷ black patients have been reported to be resistant to the potassium-sparing properties of aldosterone antagonists.²⁸ This is noteworthy because the potassium sparing actions of aldosterone antagonists may contribute importantly to their survival effects in heart failure.²⁹

Therefore, the available data from large-scale trials have created considerable uncertainty about the efficacy and safety in black patients of most of the therapeutic interventions that have been shown in non-black patients to modify the course of the disease. Individual reports have raised concerns about diminished efficacy and/or safety, and the current level of uncertainty is heightened by the fact that some of the clinical trials that have led to the approval of key drugs for heart failure were carried out in Europe and did not have an opportunity to enroll large numbers of black patients. As a result, physicians are uncertain about the treatment of heart failure in black persons.

2.3 Pathophysiology of Heart Failure in Black Patients

The current level of uncertainty about the management of heart failure in black patients is heightened by recent evidence that the pathophysiology of heart failure in black patients may differ from that in non-black patients.

When compared with non-black patients, black patients demonstrate a markedly reduced ability of peripheral blood vessels to dilate in response to endogenous stimuli of nitric oxide.^30-33 This defect appears to be related in part to an increased frequency in black persons of polymorphisms in the genes that regulate both the synthesis of nitric oxide and the production of oxygen free radicals capable of degrading nitric oxide.^34-36As a result, black patients show decreased responsiveness to drugs that stimulate endothelium-dependent vasodilation (e.g., methacholine) and enhanced responsiveness to drugs that increase the delivery of nitric oxide to peripheral blood vessels (e.g., arginine).^32,37,38

The reduced vasodilator responsiveness seen in black patients may explain the high prevalence of hypertension in this racial group. Hypertension afflicts black patients far more than white patients³⁹ and has unique characteristics, i.e., hypertension is typically characterized by salt sensitivity and reduced production of nitric oxide in black patients whereas it is characterized by the activation of neurohormonal systems (the renin-angiotensin system and the sympathetic nervous system) in white patients.^40,41 Both characteristics may help to explain why hypertension in black patients is commonly associated with end-organ consequences (e.g., left ventricular hypertrophy)⁴²and responds less readily to treatment with neurohormonal antagonists [ACE inhibitors, angiotensin receptor blockers (ARBs) and beta-adrenergic blockers].^17,22

The deficiency of nitric oxide seen in black patients may not only increase the predisposition of black persons to hypertension, but also to the development of heart failure.^43,44 Whereas the most common risk factor for heart failure in non-black patients is coronary artery disease, the most common risk factor for heart failure in black persons is hypertension — even in the absence of a defined ischemic event.^43-45Furthermore, independent of race, heart failure is characterized by both defects in nitric-oxide-mediated vasodilation and enhanced superoxide-mediated nitric oxide destruction,^46-50 and the presence of defects in nitric oxide-mediated vasodilation identifies patients who are most likely to experience worsening heart failure, cardiac transplantation or death.^51,52 The high prevalence of deficient nitric oxide-mediated vasodilation in black patients may explain in part why heart failure develops disproportionately in black patients and why, once developed, heart failure progresses more rapidly in black than in non-black patients.^4-14

These observations support the hypothesis that a vascular deficiency of nitric oxide may contribute meaningfully to the development and progression of heart failure and suggests that pharmacological amelioration of this deficiency (by the administration of a nitric oxide donor, an anti-oxidant that prevents the degradation of nitric oxide, or both^53-57) may exert clinical benefits, particularly in patients most likely to be deficient.³⁷

2.4 Effect of Isosorbide Dinitrate and Hydralazine on Vascular Nitric Oxide

When the concept of using isosorbide dinitrate and hydralazine (ISDN/HYD) together for the treatment of heart failure was first introduced in the late 1970s, the combination was believed to produce its clinical benefits by exerting complementary effects to relax both peripheral arteries and veins and thereby improve cardiac performance. This mechanism of action was supported by the following observations:

• The oral administration of isosorbide dinitrate to patients with heart failure produced short- and long-term decreases in right and left ventricular filling pressures at rest and during exercise with minimal change in blood pressure or heart rate. These effects were accompanied by little change in blood flow to the limbs and kidneys.^62,67-72Doses of 10 mg or less produced little hemodynamic effect, but sustained effects were seen with 20-40 mg, given TID or QID.^68,73Small placebo-controlled trials of isosorbide dinitrate alonefailed to demonstrate between-group improvement in symptoms or exercise tolerance in patients with heart failure, possibly because of the inadequate size of the studies or the limitation of monotherapy with isosorbide dinitrate.^71,
74-75

• The oral administration of hydralazine to patients with heart failure produced short- and long-term increases in cardiac output and stroke volume at rest and during exercise with minimal change in blood pressure or heart rate. These effects were accompanied by an improvement in blood flow to the limbs and kidneys.^58-63Doses of 50 mg or less produced little hemodynamic effect, but sustained effects were seen with 300 mg daily, given as 100 mg TID or 75 mg QID.^58,59,64Two placebo-controlled trials using low doses (150-200 mg daily) of hydralazine alonefailed to demonstrate improvement in symptoms or exercise tolerance in patients with heart failure,^65,66 possibly because of the inadequate doses used, the small size of the studies or the limitations of monotherapy with hydralazine.

• The oral administration of ISDN/HYD together in patients with heart failure produced short- and long-term increases in cardiac output and decreases in cardiac filling pressures at rest and during exercise with minimal change in blood pressure or heart rate. When used in doses of 160 mg daily of isosorbide dinitrate and 300 mg daily of hydralazine, the complementary effects of the two drugs produced a hemodynamic response comparable both qualitatively and quantitatively to that produced by intravenous nitroprusside.^76-79

In addition to these complementary hemodynamic effects, several studies have also suggested that isosorbide dinitrate and hydralazine may exert complementary biochemical effects that could underlie or contribute importantly to their hemodynamic actions. Isosorbide dinitrate exerts vasodilator effects by acting as a nitric oxide donor within blood vessels.⁸⁰ However, the hemodynamic actions of nitrate therapy are frequently lost during repeated administration of the drug (“nitrate tolerance”),^81-83 and this has limited the utility of the drug as monotherapy. Several studies have postulated that oxidative stress contributes importantly to the development of nitrate tolerance.^84-87 It is therefore noteworthy that — in addition to its vasodilator effects — hydralazine exerts anti-oxidant effects^88,89 and its co-administration with isosorbide dinitrate can prevent the development of nitrate tolerance.^90-93

2.5 Regulatory Background of BiDil® (Isosorbide Dinitrate and Hydralazine

Hydrochloride)

BiDil® is a fixed-dose combination of two active ingredients, isosorbide dinitrate (ISDN) and hydralazine hydrochloride (HYD). Isosorbide dinitrate was approved in 1961 to treat angina, and hydralazine was first approved in 1952 to treat hypertension. Neither of the approved labels describes the use of the agent for the treatment of heart failure, either alone or in combination and the combination of ISDN/HYD has not been approved for any indication.

In the 1980s the combination of ISDN/HYD to treat heart failure was evaluated in two trials sponsored by the Department of Veterans Affairs (V-HeFT I and V-HeFT II). These studies were licensed by Medco Research (currently King Pharmaceuticals) and an NDA was developed and filed in 1996 for the proposed use of BiDil® (a fixed-dose combination of ISDN/HYD) for the treatment of heart failure in patients who could not tolerate treatment with an angiotensin- converting enzyme inhibitor. BiDil® was assigned NDA number 20-727.

The Division of Cardiovascular and Renal Drugs convened a meeting of its Advisory Committee to consider the BiDil® NDA in February 1997. The Advisory Committee recommended against the approval of BiDil® for the proposed indication. A not approvable letter was issued in July 1997 stating that additional clinical data would be required. Medco Research opted not to pursue the development of BiDil®.

Based on a growing body of data suggesting that ethnic groups might differ in their mechanisms of disease and their response to treatment, Dr. Jay Cohn, the principal investigator for V-HeFT I and II, conducted additional retrospective analyses of the studies and observed that the black patients in V-HeFT I had remarkably better responses to ISDN/HYD than white patients.⁹⁴ In light of these new analyses, Dr. Cohn approached NitroMed, a company dedicated to research and development of nitric oxide enhancing technologies, to pursue the development of BiDil®. In 1999, NitroMed became the official sponsor of the BiDil® NDA.

NitroMed and Dr. Cohn met with FDA in 1999 and 2000 to explore further development of BiDil® for the treatment of heart failure in black patients. After these discussions and review the V-HeFT reanalyses and supporting documentation, FDA issued a letter in March 2001 stating that “[g]iven the subset finding and the overall trend toward a survival effect in VHeFT I, we believe a single, clearly positive study in a CHF population would be a basis for approval of BiDil® for the treatment of heart failure in blacks.” In May 2001, NitroMed launched the African American Heart Failure Trial (A-HeFT).

In July 2004, A-HeFT was stopped following a unanimous recommendation of its Data and Safety Monitoring Board (DSMB) and Steering Committee. This recommendation was based on the observation of a 43% reduction in relative risk of mortality in patients treated with BiDil® relative to placebo.

The complete response (or NDA amendment) to the original 1997 not approvable letter including the A-HeFT study report was submitted to the FDA in December 2004. The PDUFA date is June 23, 2005.

The indication being pursued by NitroMed is for the “treatment of heart failure in black patients. BiDil® has been shown to reduce the risk of mortality from any cause, to reduce the risk of heart failure hospitalization and to improve quality of life.”

A Data and Safety Monitoring Board composed of clinicians who did not participate in the trial, periodically reviewed study results and evaluated the treatments for excess events. The members of the DSMB were:

Richard Gorlin, M.D., Mount Sinai School of Medicine (chair)

Yick-Kwong Chan, Ph.D., West Haven VA Hospital

Leon Goldberg, M.D., Ph.D., University of Chicago

Genell Kantterud, Ph.D., Maryland Research Institute

William Parmley, M.D., University of California

David Shand, M.D., West Haven VA Hospital

4.1.3 Study Population

4.1.3.1 Inclusion Criteria

· Men, 18 to 75 years old.

· Heart failure as evidenced by reduced exercise tolerance for at least 3 months. Reduced exercise tolerance was defined as maximal oxygen consumption < 25 mL/kg/min during graded bicycle ergometry testing.

· Persistent symptoms despite treatment with digitalis glycosides and diuretics.

· Cardiothoracic ratio on chest x-ray > 0.55, or an echocardiographic left ventricular internal dimension > 2.7 cm/m², or a radionuclide or contrast left ventricular ejection fraction < 0.45.

4.1.3.2 Exclusion Criteria

· Myocardial infarction or cardiac surgery within 3 months.

· Hypertrophic cardiomyopathy or hemodynamically significant aortic or mitral valve or pericardial disease.

· Patients with hypertension requiring antihypertensive drugs other than diuretics.

· Angina pectoris severe enough to require long-acting nitrates or frequent administration of sublingual nitroglycerin (more than 4 tablets per week).

· Chronic treatment with a beta-blocking drug, calcium channel blockers, or vasodilators other than occasional sublingual nitroglycerin.

· History of systemic lupus erythematosus or history of intolerance to isosorbide dinitrate, hydralazine or prazosin.

· Chronic lung disease sufficient to limit exercise tolerance.

· Severe intrinsic renal disease or primary hepatic disease.

· Hematocrit < 30%.

· Disease that was expected to limit survival within 2 years.

4.1.4 Study Plan

After each patient was screened, he entered a baseline period of two weeks’ duration to establish optimal therapy with a digitalis glycoside and a diuretic and to allow any nonstudy drugs to be discontinued. Patients fulfilling all inclusion criteria and none of the exclusion criteria were randomized to one of three treatment groups: placebo, prazosin or the combination of ISND/HYD. Randomization was stratified by the presence or absence of clinically suspected coronary artery disease. Randomization was carried out within each stratification group at each center in blocks of 7, with 3 patients assigned to placebo and 2 patients assigned to each of the active drug regimens. This ratio was used because it allowed nearly optimal power for comparisons between the two active treatments and for a possible comparison between both active treatments combined and placebo.

Following randomization, patients were instructed to take 1 tablet QID and 1 capsule QID. The tablet contained ISDN 20 mg or placebo. The capsule contained HYD 37.5 mg, prazosin 2.5 mg or placebo. After 2 weeks, if tolerated, the patients were uptitrated to 2 tablets QID and 2 capsules QID. The target doses for the study were 160 mg/day of ISDN and 300 mg/day of HYD, or 20 mg/day of prazosin.

If the study medications were not tolerated, the patient could reduce the dose of one or both of the study drugs. The goal was to achieve the highest tolerated dose of the study medication, and the doses of other medications could be adjusted as clinically indicated.

Following randomization, each patient was to be seen as an outpatient every 2 weeks until two successive visits revealed stability, and then every 1-3 months for the duration of the trial. Chest x-ray, M-mode echocardiography, Holter monitoring, physician assessment of quality of life, radionuclide imaging for assessment of left ventricular ejection fraction, and maximum exercise testing were performed at baseline, at 2 and 6 months, and every 6 months thereafter.

4.1.5 Study Assessments

The study protocol described several major and several minor endpoints. However, the study was envisioned primarily as a mortality study, and mortality was the only variable that was used to determine the sample size of the trial.

4.1.5.1 Major Endpoints

• All-cause mortality during the entire study period

• All-cause mortality at 2 years

• Number and duration of cardiovascular hospitalizations

• Maximum oxygen consumption during peak exercise

• Maximum treadmill exercise time on a graded test

• Duration of exercise on submaximal test

4.1.5.2 Minor Endpoints

• Heart size by M-mode echocardiography

• Left ventricular function by M-mode echocardiography

• Heart size and pulmonary congestion by chest x-ray

• Ejection fraction by radionuclide ventriculography

• Arrhythmias assessed by Holter monitoring

• Patient and investigator global assessment of improvement

4.1.5.3 Safety Assessments

Safety assessments consisted of monitoring and recording all treatment-emergent adverse events and serious adverse events, the performance of physical examinations (which included the measurement of vital signs at every visit), and laboratory evaluations.

4.1.6 Statistical Plan and Analyses

4.1.6.1 Sample Size Determination and Interim Monitoring Plan

The study protocol projected a sample size of 720 patients (308 in the placebo group and 206 in each of the vasodilator regimens) in order to provide 84% power to detect a difference in survival curves if either vasodilator treatment reduced the annual mortality rate by 33% compared with the placebo group, assuming a dropout rate of 6% per year.

The Data and Safety Monitoring Board met at 6 month intervals throughout the study. The Committee used an O’Brien-Fleming boundary to guide decision-making during four interim analyses. The Committee made no decision to recommend modification of the course of the study.

4.1.6.2 Statistical Analyses

Mortality

Survival curves were compared among the three treatment groups using the log-rank test, and differences between the survival curve of each drug regimen vs that of placebo were analyzed using a protocol-specified one-sided α=0.025 (or two-sided α=0.05).

Non-Fatal Measures of Efficacy

Mean changes from baseline in exercise duration and capacity, left ventricular ejection fraction, heart size and quality of life assessments were calculated for each variable for those patients with data available at each study visit. [Such an approach does not account for the differences in survival between treatment groups.] The observed treatment difference was evaluated for significance using two-sample t-tests.

4.1.7 Results

4.1.7.1 Baseline Characteristics

A total of 642 patients were randomized to treatment with placebo (n=273), ISDN/HYD (n=186), and prazosin (n=183).

The patients enrolled in V-HeFT I were middle-aged men, of whom approximately 27-29% were black in each group (see Table 2). The most common cause of heart failure was coronary artery disease. The mean left ventricular ejection fraction was approximately 30%, and the mean oxygen consumption was approximately 15 mL/kg/min. The groups were well-matched for baseline characteristics.

Table 2. Baseline Demographic and Clinical Characteristics; V-HeFT I

	Placebo n=273	ISDN/HYD n=186	Prazosin n=183
Demographic features
Age (years; mean)	58.5	58.5	58.3
Race (n, %)
White	192 (70.3%)	132 (71.4%)	NA
Black	79 (28.9%)	49 (26.5%)	NA
Other	2 (0.7%)	4 (2.2%)	NA
Cardiovascular history (n,%)
Coronary artery disease	129 (47.3%)	86 (46.2%)	86 (46.2%)
Alcohol excess	104 (38.1%)	80 (43.0%)	80 (43.0%)
Hypertension	118 (43.2%)	74 (39.8%)	74 (39.8%)
Diabetes	67 (24.5%)	32 (17.2%)	32 (17.2%)
Drug therapy (prior 6 mos; n, %)
Vasodilators	99 (36.3%)	78 (41.9%)	78 (41.9%)
Antiarrhythmics	73 (26.7%)	53 (28.5%)	53 (28.5%)
Sublingual nitroglycerin	53 (19.4%)	39 (21.0%)	39 (21.0%)
Anticoagulants	48 (17.6%)	34 (18.3%)	34 (18.3%)
Clinical data (mean)
Symptom score^a	5.6	5.6	5.6
Left ventricular ejection fraction (%)	30.4	30.4	30.3
Maximal O₂consumption (mL/kg/min)	15.0	14.5	14.4
Cardiothoracic ratio (%)	53.0	52.8	52.8
Exercise duration (min)	9.8	9.7	9.7

^a Sum of scores for dyspnea, fatigue, orthopnea, and paroxysmal nocturnal dyspnea; each symptom was scored

as 1 = none, 2 = moderate, and 3 = severe. Maximum possible score was 12.

4.1.7.3.4 Maximum Oxygen Consumption at Peak Exercise

For the first 2 years, exercise capacity (as assessed by the mean change from baseline in maximum oxygen consumption at peak exercise) was greater in the ISDN/HYD group than in the placebo group (Table 4). The difference in the response to treatment between the two groups was approximately 0.5 mL/kg/min at all time points, but was not statistically significant at any time point.

Table 4. Changes in Maximal Oxygen Consumption Relative to

Baseline; V-HeFT I

	Placebo	ISDN/HYD	p-value*

Baseline Mean (SD)	n=259	n=176
Baseline Mean (SD)	14.9 (3.9)	14.7 (3.9)

Week 8 Mean (SD)	n=221	n=151
Week 8 Mean (SD)	15.4 (4.4)	15.5 (4.3)
Mean change (SD)	+0.2 (3.0)	+0.7 (2.8)	p=0.125

Week 28 Mean (SD)	n=193	n=136
Week 28 Mean (SD)	15.4 (4.0)	15.3 (4.8)
Mean change (SD)	+0.1 (2.9)	+0.4 (3.8)	p=0.472

Year 1 Mean (SD)	n=155	n=113
Year 1 Mean (SD)	15.0 (4.0)	15.4 (4.1)
Mean change (SD)	–0.2 (3.7)	+0.6 (3.0)	p=0.056

Year 1.5 Mean (SD)	n=111	n=95
Year 1.5 Mean (SD)	15.1 (4.2)	15.3 (4.4)
Mean change (SD)	–0.2 (3.3)	+0.2 (3.8)	p=0.341

Year 2 Mean (SD)	n=99	n=73
Year 2 Mean (SD)	15.3 (4.2)	15.3 (3.4)
Mean change (SD)	–0.4 (3.5)	+0.2 (3.1)	p=0.270

* p-values refer to between-group comparisons vs placebo

Of note, the placebo-corrected increase in maximum oxygen consumption was generally larger in black patients than non-black patients (e.g., +1.64 mL/kg/min in black patients vs +0.84 mL/kg/min in non-black patients at one year).

When compared with placebo, prazosin had no effect on maximum oxygen consumption at any time in the study.

4.1.7.3.5 Left Ventricular Ejection Fraction

At all time points during the first 2 years of the study, mean change from baseline in left ventricular ejection fraction was significantly greater in the ISDN/HYD group than in the placebo group (p<0.03; Table 5).

Table 5. Changes in Left Ventricular Ejection Fraction (% Units); V-HeFT I

	Placebo	ISDN/HYD	p-value*

Baseline Mean (SD)	n=252	n=176
Baseline Mean (SD)	30.4 (13.5)	30.3 (12.9)

Week 8 Mean (SD)	n=230	n=143
Week 8 Mean (SD)	30.7 (13.8)	32.9 (14.4)
Mean change (SD)	+0.4 (6.2)	+2.9 (7.3)	p=0.0004

Week 28 Mean (SD)	n=199	n=141
Week 28 Mean (SD)	30.6 (14.0)	34.2 (15.0)
Mean change (SD)	+0.1 (7.4)	+3.7 (9.2)	p=0.0001

Year 1 Mean (SD)	n=166	n=124
Year 1 Mean (SD)	31.6 (15.0)	35.5 (15.6)
Mean change (SD)	+0.3 (9.2)	+4.6 (10.0)	p=0.0002

Year 1.5 Mean (SD)	n=128	n=101
Year 1.5 Mean (SD)	30.5 (15.1)	33.2 (16.1)
Mean change (SD)	–1.3 (8.7)	+2.0 (10.1)	p=0.0093

Year 2 Mean (SD)	n=107	n=85
Year 2 Mean (SD)	31.9 (15.9)	34.6 (17.2)
Mean change (SD)	–1.2 (8.5)	+2.0 (10.3)	p=0.0261

* p-values refer to between-group comparisons vs placebo

Of note, the placebo-corrected increase in left ventricular ejection fraction with ISDN/HYD was larger in non-black patients than black patients (+0.07 vs +0.01 units) at 1 year, but the race-by-treatment interaction was not significant (p=0.23).

When compared with placebo, prazosin had no effect on left ventricular ejection fraction at any time during the study.

4.1.7.3.6 Other Endpoints

The duration of tolerable exercise was greater in the ISDN/HYD group than in the placebo and prazosin groups at each time point during the first 2.5 years of the study. Slight mean increases from baseline were observed in the ISDN/HYD group, compared with slight mean decreases in the placebo group. However, none of the between-group differences was significant at any time point.

The cardiothoracic ratio assessed by chest x-ray was smaller in the ISDN/HYD group than in the placebo or prazosin groups early in the study. Compared with slight mean increases from baseline in the placebo group, the ISDN/HYD group experienced slight mean decreases from baseline in cardiothoracic ratio after 8 weeks (+0.1% placebo vs. –0.7% ISDN/HYD, p=0.046) and after 28 weeks (+0.1% placebo vs. –0.7% ISDN/HYD, p=0.063).

4.1.7.4 Safety Results

4.1.7.4.1 Adverse events regardless of relationship to study drug

Table 6 lists the number of patients who reported various adverse events. At each visit, investigators questioned the patients using a preprinted list of adverse events known to be associated with use of isosorbide dinitrate, hydralazine or prazosin; adverse events not on the preprinted list were recorded under “other.” A listing of specific “other” events is not available. Safety data for prazosin are not included in this briefing document.

Adverse events related to systemic vasodilation (headache, dizziness, flushing) or reflecting gastrointestinal distress (nausea, vomiting, diarrhea and abdominal pain) were more frequent in ISDN/HYD-treated than placebo-treated patients.

Table 6. Patients with Adverse Events; V-HeFT I

Preferred Term	Placebo N=273		ISDN/HYD N=186
Preferred Term	n	(%)	n	(%)
Headache	139	(50.9)	139	(74.7)
Dizziness	163	(59.7)	131	(70.4)
Arthralgias	158	(57.9)	118	(63.4)
“Other”	135	(49.5)	114	(61.3)
Palpitation	120	(44.0)	104	(55.9)
Nausea or vomiting	123	(45.1)	97	(52.2)
Ischemic chest pain	113	(41.4)	91	(48.9)
Diarrhea	106	(38.8)	87	(46.8)
Abdominal pain	95	(34.8)	84	(45.2)
Flushing	83	(30.4)	81	(43.6)
Rash	104	(38.1)	80	(43.0)
Fever	72	(26.4)	62	(33.3)
Syncope	65	(23.8)	49	(26.3)

Of these adverse events, about 30% were rated severe as assessed by the investigator. The proportion of patients who experienced one or more severe adverse events was higher in the ISDN/HYD group than in the placebo group (41.4% vs 20.5%). The most frequent severe adverse events were headache (3.3% placebo vs. 27.4% ISDN/HYD), dizziness (7.3% placebo vs. 12.9% ISDN/HYD), “other” (5.9% placebo vs. 6.5% ISDN/HYD), arthralgias (3.3% placebo vs. 5.4% ISDN/HYD), and nausea or vomiting (2.6% placebo vs. 5.4% ISDN/HYD).

4.1.7.4.2 Adverse events leading to permanent withdrawal of study drug

A patient was considered to have discontinued ISDN/HYD prematurely if he permanently discontinued both study medications prior to study end. A higher proportion of ISDN/HYD patients discontinued the study drugs prematurely because of adverse events (5.9% vs 1.1% on placebo). Adverse events leading to discontinuation of study medication included dizziness/syncope (0.7% placebo vs. 3.8% ISDN/HYD), headache (0.0% placebo vs. 3.2% ISDN/HYD), “other” (0.7% placebo vs. 2.7% ISDN/HYD), disorientation (0.0% placebo vs. 1.1% ISDN/HYD), arthralgia (0.0% placebo vs. 0.5% ISDN/HYD), and nausea (0.0% placebo vs. 0.5% ISDN/HYD).

4.1.7.4.3 Other safety topics

Vital signs

Neither systolic nor diastolic blood pressures were lower in the ISDN/HYD group when compared with the placebo group. When compared with placebo, both systolic and diastolic blood pressures were significantly lower in the prazosin group at 8 weeks but not at one year. Heart rates were also similar across the three treatment groups for the duration of the study.

Lupus syndrome

Arthralgias were considered severe and possibly or probably related to the study medication in 7 patients (6 in the ISDN/HYD group and 1 in the placebo group). A total of 12 patients (10 ISDN/HYD and 2 placebo) were discontinued from the study due to arthralgia. In 5 patients (3 ISDN/HYD and 2 placebo), arthralgias were associated with a significant increase (³ 1:160) in ANA titer; this increase was sustained (³ 2 consecutive assessments excluding baseline) in 1 patient in the placebo group and 3 patients in the ISDN/HYD group. In addition, two ISDN/HYD patients were diagnosed with lupus-like syndrome based on symptoms and immunologic (ANA titer and LE prep) assessments.

Clinical laboratory evaluations

No clinically relevant mean changes in values for clinical laboratory tests were seen during the study.

4.1.8 Summary and Conclusions for V-HeFT I

The findings of the first Vasodilator Heart Failure Trial (V-HeFT I) support the following conclusions:

· The long-term administration of a combination of ISDN/HYD to middle-aged men with mild-to-severe heart failure treated with digitalis glycosides and diuretics was associated with a 22% reduction in the relative risk of death. The p-value equaled 0.093 (protocol-specified log-rank test).

· In a retrospective analysis a reduction in the risk of death similar to that seen in the overall trial was seen across nearly all of the subgroups examined. The most striking effect was seen in black patients who experienced a 47% reduction in relative risk, as compared with white patients who experienced only a 12% reduction in relative risk, interaction p=0.15. The survival effect in black patients with ISDN/HYD was statistically significant in its own right (p= 0.04), even though black patients were one of the smallest subgroups and comprised only 30% of the patients in the trial.

· Further retrospective examination of subgroup effects suggested other subgroups might also respond well (with respect to survival) to the ISDN/HYD combination. These subgroups included: ejection fraction < 40% [25% reduction in risk when compared with 17% reduction in risk in patients with preserved ejection fractions]; younger patients [age < 59 years; 33% reduction in risk when compared with 9% reduction in risk in older patients]; diabetic patients [25% reduction in risk when compared with 5% reduction in risk in non-diabetics]; lower systolic blood pressure [< 118 mm Hg; 26% reduction in risk when compared with 14% reduction in risk in patients with higher systolic blood pressures].

· For the first two years of the study (the duration for which a meaningful proportion of the randomized patients were followed), the risk of hospitalization for heart failure was lower in the ISDN/HYD group than in the placebo group. A treatment effect in black patients contributed importantly to the overall differences.

· Although maximal exercise capacity was not significantly increased by the combination of ISDN/HYD in the overall trial, the magnitude of the functional improvement in black patients (who had an increase of 1.64 mL/kg/min) was greater than that seen in white patients (who had an increase of 0.84 mL/kg/min). This observation suggested that future trials might appropriately seek to confirm the efficacy of ISDN/HYD, using an endpoint that measures the effects of the drug on both clinical status and survival.

· The long-term administration of a combination of ISDN/HYD was associated with a consistent and meaningful improvement in left ventricular ejection fraction (about 3-4 units). Demonstration of the persistence of this effect for 2 years suggests that hemodynamic tolerance did not develop to this combination of ISDN/HYD during the course of long-term treatment of patients with heart failure.

· The long-term administration of prazosin, another drug with arterial and venous vasodilating effects, did not demonstrate favorable effects on survival, ejection fraction or exercise capacity. This finding suggests that the mechanisms by which drugs exert their vasodilator effects appear to be relevant in determining their efficacy in the treatment of heart failure.

· The long-term administration of a combination of ISDN/HYD was associated with headache, dizziness and other vasodilator-type reactions.

· A meaningful proportion of patients failed to achieve target doses of both ISDN and HYD. Clinical benefits were seen despite the use of lower-than-target doses, suggesting that future trials might appropriately target lower doses of ISDN/HYD.

In conclusion, the findings of V-HeFT I suggested that the combination of isosorbide dinitrate and hydralazine was likely to have favorable effects on survival and functional status when used in the treatment of heart failure and that the subgroup of black patients might be particularly sensitive to these benefits.

4.2 Vasodilator Heart Failure Trial II (V-HeFT II)

4.2.1 Study Overview

The second Vasodilator Heart Failure Trial (V-HeFT II) was a multicenter, randomized, double-blind, parallel group, active controlled trial conducted at 13 sites in the United States under the auspices of the Department of Veterans Affairs, which compared the vasodilator combination of ISDN/HYD and the angiotensin-converting enzyme inhibitor enalapril.

The intent in V-HeFT II was to compare two different drug treatments that had been shown in separate trials to reduce the risk of death in patients with chronic heart failure: (1) ISDN/HYD, which had favorable effects on survival in V-HeFT I; and (2) enalapril, which had favorable effects on survival in the Cooperative North Scandinavian Enalapril Survival Study (CONSENSUS). Both drug regimens were known to produce hemodynamic benefits by exerting dilating effects on systemic blood vessels, but it was unclear whether the distinctly different mechanisms by which the treatments exerted vasodilator effects might result in distinctly different patterns of clinical benefit.

As was the case of V-HeFT I, V-HeFT II enrolled men who generally had class II-IV symptoms, had heart failure associated with both impaired and preserved ejection fraction, and were generally taking only digitalis glycosides and diuretics. The entry criteria, design and endpoints of V-HeFT II closely paralleled those of V-HeFT I.

4.2.2 Study Organization

The Executive Committee was the management and decision-making body for the operational aspects of the conduct of the study. The members of the Committee were:

Jay Cohn, M.D., Minneapolis VA Hospital (chair)

Donald Archibald, M.Phil., West Haven VA Hospital (biostatistician)

Frederick Cobb, M.D., Durham VA Hospital

Ross Fletcher, M.D., Washington DC VA Hospital

Gary Francis, M.D., Minneapolis VA Hospital

Clair Haakenson, R.Ph., Albuquerque VA Hospital (research pharmacist)

Gary Johnson, MS., West Haven VA Hospital (biostatistician)

Pravin Shah, M.D., West Los Angeles VA Hospital

Maylene Wong, M.D., West Los Angeles VA Hospital

Susan Ziesche, R.N., Minneapolis VA Hospital

A Data and Safety Monitoring Board; composed of clinicians who did not participate in the trial periodically reviewed study results and evaluated the treatments for excess events. The members of the DSMB were:

Richard Gorlin, M.D., Mount Sinai School of Medicine (chair)

Dorothea Collins, M.S., West Haven VA Hospital

Leon Goldberg, M.D., Ph.D., University of Chicago

Genell Kantterud, PhD, Maryland Research Institute, Baltimore

John Oates, M.D., Vanderbilt University

William Parmley, M.D., University of California San Francisco

4.2.3 Study Population

4.2.3.1 Inclusion Criteria

· Men, 18 to 75 years old.

· Persistent symptoms despite digitalis glycosides and diuretics.

· Cardiothoracic ratio on chest x-ray > 0.55, or an echocardiographic left ventricular internal dimension > 2.7 cm/m², or a radionuclide or contrast left ventricular ejection fraction < 0.45.

4.2.3.2 Exclusion Criteria

· Myocardial infarction or cardiac surgery within 3 months.

· Hypertrophic cardiomyopathy or hemodynamically significant aortic or mitral valve or pericardial disease.

· Patients with hypertension requiring antihypertensive drugs other than diuretics.

· Angina pectoris severe enough to require long-acting nitrates or frequent administration of sublingual nitroglycerin (more than 4 tablets per week).

· Chronic treatment with a beta-blocking drug, calcium channel blockers, or vasodilators other than occasional sublingual nitroglycerin.

· History of systemic lupus erythematosus or history of intolerance to isosorbide dinitrate, hydralazine or enalapril.

· Chronic lung disease sufficient to limit exercise tolerance.

· Severe intrinsic renal disease or primary hepatic disease.

· Hematocrit < 30%.

· Disease that was expected to limit survival within 2 years.

It should be noted that 129 patients who had completed V-HeFT I in either the placebo or prazosin treatment group and who met the eligibility criteria for V-HeFT II, were randomized into V-HeFT II.

4.2.5.2 Minor Endpoints

• Heart size and pulmonary congestion by chest x-ray

• Ejection fraction by radionuclide ventriculography

• Arrhythmias assessed by Holter monitoring

• Plasma norepinephrine

4.2.5.3 Safety Assessments

4.2.6 Statistical Plan and Analyses

4.2.6.1 Sample Size Determination and Interim Monitoring Plan

The study protocol projected a sample size of 952 patients in order to provide 87% power to detect a 30% difference in survival, assuming a mortality rate with ISDN/HYD similar to that observed with ISDN/HYD in V-HeFT I (α=0.05).

The Data and Safety Monitoring Board met at 6 month intervals throughout the study and used an O’Brien-Fleming boundary to guide decision-making during four interim analyses. The Committee made no decision to recommend modification of the course of the study.

4.2.6.2 Statistical Analyses

Mortality

Differences in survival were compared between the two treatment groups using a log-rank test (two-sided α=0.05). The final test for significance for mortality was set at α=0.042 after adjustment for four interim analyses using the O’Brien-Fleming group sequential boundary. Mortality risks within subgroup were assessed and compared using a Cox proportional hazard model. Exposure was censored at the time of heart transplantation in eight patients (six enalapril and two ISDN/HYD). For the comparison between treatment groups of survival rates at 2 years, 95% confidence intervals were used to generate the Greenwood standard errors using an asymptotically normal test statistic.

Non-Fatal Measures of Efficacy

Mean changes from baseline in exercise capacity, left ventricular ejection fraction, heart size and quality of life assessments were calculated for each variable for those patients with data available at each study visit. [Such an approach does not account for the differences in survival between treatment groups.] The observed treatment difference was evaluated for significance using two-sample t-tests. The chi-square statistic was used to test for significance of differences between groups in the number of hospitalizations and other major clinical events.

4.2.7 Results

4.2.7.1 Baseline Characteristics

A total of 804 patients were randomized to treatment with enalapril (n=403) or ISDN/HYD (n=401).

The patients enrolled in V-HeFT II were middle-aged men, of whom approximately 26-27% were black in each group (see Table 7). The most common cause of heart failure was coronary artery disease; slightly less than one-half of the patients had heart failure due to hypertension. The mean left ventricular ejection fraction was approximately 29%, and the mean oxygen consumption was approximately 13-14 mL/kg/min. The two treatment groups were well-matched for baseline characteristics, except that the mean duration of heart failure was longer in the ISDN/HYD group than in the enalapril group (p=0.0044), and the mean left ventricular internal dimensions were greater in the enalapril group than in the ISDN/HYD group (p=0.0192).

Table 7. Baseline Demographic and Clinical Characteristics; V-HeFT II

	Enalapril n=403*	ISDN/HYD n=401*
Demographic features
Age (years; mean, SD)	60.6 (8.3)	60.6 (8.5)
Race (n, %)
White	292 (72.5%)	282 (70.3%)
Black	106 (26.3%)	109 (27.2%)
Other	5 (1.2%)	10 (2.3%)
Duration of heart failure (mos; mean, SD)	31.2 (37.8)	40.2 (48.6)
NYHA class (n, %)
Class I	24 (6.0%)	22 (5.5%)
Class II	200 (49.6%)	210 (52.4%)
Class III	178 (44.2%)	167 (41.7%)
Class IV	1 (0.3%)	2 (0.5%)
Cardiovascular history (n, %)
Coronary artery disease	220 (54.6%)	213 (53.3%)
Alcohol excess	135 (33.5%)	147 (36.7%)
Hypertension	199 (49.6%)	182 (45.4%)
Diabetes	84 (20.8%)	80 (20.0%)
Drug therapy (prior 6 mos; n, %)
Vasodilators	250 (62.0%)	247 (61.6%)
Antiarrhythmics	100 (24.8%)	106 (26.4%)
Sublingual nitroglycerin	64 (15.9%)	67 (16.7%)
Anticoagulants	84 (20.8%)	88 (22.0%)
Clinical data (mean, SD)
Ejection fraction (%)	28.6 (10.9) (n=388)	29.4 (11.5) (n=384)
Maximal O₂consumption (mL/kg/min)	13.8 (3.5) (n=398)	13.5 (3.5) (n=400)
Systolic/diastolic BP (mm/Hg)	125/78	127/78
Heart rate (beats/min)	78.4 (12.1)	77.3 (11.9)
Cardiothoracic ratio (%)	53.7 (6.0) (n=392)	53.0 (6.2) (n=392)
Left ventricular internal dimension (cm/m²)	3.6 (1.4) (n=170)	3.2 (1.2) (n=159)
Plasma norepinephrine (pg/mL)	593 (388) (n=372)	544 (297) (n=371)
Plasma renin activity (mg/mL/hr)	19.9 (52.6) (n=371)	15.7 (28.1) (n=366)

* n= 403 or 401, unless otherwise specified

4.2.7.2 Patient Disposition and Exposure to Study Medication

The first of 804 patients was enrolled in March 1986, the last patient was enrolled in September 1990, and the study was completed in February 1991. The mean follow-up period was 2.5 years (range: 6 months to 4.9 years).

In the ISDN/HYD treatment group, the majority (67.3%) of patients achieved the target dose of both HYD and ISDN tablets by 6 months; at that time, a higher proportion had achieved the target dose of HYD than of ISDN (81.3% HYD vs. 72.1% ISDN). Approximately three fourths (74.8%) of ISDN/HYD patients achieved the target dose of both study medications at any time during the study; a higher proportion of patients achieved the target dose of HYD than of ISDN (84.5% vs. 78.6%). In comparison, the percentage reaching target dose was consistently higher in the enalapril group. The proportion of enalapril-treated patients who achieved the target dose at 6 months and at any time during the study was 92.8% and 94.8%, respectively.

In the ISDN/HYD group, the average number of HYD tablets taken per day was 5.4 (target dose = 8 x 37.5 mg tablet), and the average number of ISDN tablets taken per day was 2.5 (target dose = 4 x 40 mg tablet). Therefore, the average daily dose of HYD was 199 mg/day (67% of target dose) and the average daily dose of ISDN was 100 mg/day (63% of target dose). In the enalapril group, the average number of tablets taken per day was 1.5 (target dose = 2 x 10 mg tablet); therefore, the average dose was 15 mg/day (75% of target dose).

In the ISDN/HYD group, most (³ 74.3%) patients received treatment with HYD or ISDN for at least 6 months, and the majority (³ 62.3%) received treatment for at least one year. The proportion of patients exposed to drug was consistently higher in the enalapril group than in the ISDN/HYD group; the proportion of patients exposed to HYD was similar to that exposed to ISDN.

By the time of the final clinic visit, 22% of the patients assigned to enalapril had discontinued the drug, and an additional 8% had reduced the dose. In the ISDN/HYD group, 29% of the patients had discontinued HYD and 10% had reduced the dose, whereas 31% had discontinued ISDN, and an additional 10% had reduced the dose. Compliance with the prescribed regimen averaged 86%.

Twenty-five patients in the ISDN/HYD arm received ACE inhibitors, whereas in the enalapril arm, 5 patients were treated with HYD and 15 with ISDN.

4.2.7.3 Efficacy Results

4.2.7.3.1 Overall Mortality

By intention to treat, there were 132 deaths from all causes in the enalapril group (32.8%) as compared with 153 deaths in the ISDN/HYD group (38.2%; Table 8 and Figure 8). The log-rank p-value for the comparison of ISDN/HYD vs enalapril was 0.083. At the protocol-specified endpoint of 2 years, the cumulative mortality rate was 18% in the enalapril group and 25% in the ISDN/HYD group, a 28% difference in risk; p=0.016.

Table 8. Effects on All-Cause Mortality; V-HeFT II

# of Deaths (%)

Enalapril
(n = 403)

# of Deaths (%)

ISDN/HYD
(n = 401)

ISDN/HYD: enalapril

hazard ratio

(95% CI)

Log-rank

p-value

132 (32.8%)

153 (38.2%)

1.23

(0.97, 1.55)

0.083

Figure 14. Time to First Heart Failure Hospitalization – White Patients; V-HeFT II

4.2.7.3.4 Maximum Oxygen Consumption at Peak Exercise

Exercise capacity (as assessed by maximum oxygen consumption at peak exercise) was consistently greater in the ISDN/HYD group than in the enalapril group throughout the study (Table 9). The differences between the groups were significant or nearly so at most time points.

Table 9. Changes in Maximal Oxygen Consumption; V-HeFT II

	Enalapril	ISDN/HYD	p-value*
Baseline Mean (SD)	(n=398) 13.8 (3.5)	(n=400) 13.5 (3.5)

Month 3 Mean (SD)	n=333	n=322
Month 3 Mean (SD)	13.9 (3.7)	14.1 (3.8)
Mean change (SD)	–0.05 (2.4)	+0.41 (2.4)	p=0.0152

Month 6 Mean (SD)	n=302	n=289
Month 6 Mean (SD)	14.4 (3.6)	14.4 (4.2)
Mean change (SD)	+0.25 (2.5)	+0.60 (2.7)	p=0.1099

Month 12 Mean (SD)	n=272	n=247
Month 12 Mean (SD)	13.8 (3.7)	14.2 (3.9)
Mean change (SD)	–0.32 (2.7)	+0.24 (3.0)	p=0.0185

Month 18 Mean (SD)	n=222	n=187
Month 18 Mean (SD)	14.0 (3.4)	14.3 (4.0)
Mean change (SD)	–0.24 (2.8)	+0.19 (3.2)	p=0.1462

Month 24 Mean (SD)	n=187	n=160
Month 24 Mean (SD)	13.9 (3.7)	14.4 (3.7)
Mean change (SD)	–0.67 (2.7)	+0.16 (2.5)	p=0.0035

^* p-values refer to between-group comparisons

4.2.7.3.5 Quality of Life

For all patients changes in quality of life were similar between the ISDN/HYD and the enalapril groups.

Of note, the improvement in quality of life with ISDN/HYD was primarily seen in black patients (Table 10). Specifically, in black patients, quality of life scores at twelve months in black patients improved by 0.67 units in the ISDN/HYD group but deteriorated by 1.04 units in the enalapril group (between group p=0.04). In white patients, quality of life at twelve months improved by 0.24 units in the ISDN/HYD group and by 0.26 units in the enalapril group (between group p=0.97). The race by treatment interaction p-value for changes in quality of life at twelve months was 0.09.

Table 10. Quality of Life* in Black Patients; V-HeFT II

Time on Study

ISDN/HYD

Enalapril

p-Value

# of Patients

QOL Change From Baseline

# of Patients

QOL Change From Baseline

3 Months

-0.29

0.23

0.43

6 Months

-0.29

0.8

0.18

12 Months

-0.67

1.04

0.043

* A Decrease in the “Quality of Life” score is favorable.

4.2.7.3.6 Left Ventricular Ejection Fraction

At the time of the first post-randomization measurement (3 months), the mean change from baseline in left ventricular ejection fraction was significantly greater in the ISDN/HYD group than in the enalapril group (p=0.026; Table 11). Although this difference in favor of ISDN/HYD persisted throughout the study, it was no longer significant after 3 months.

Table 11. Left Ventricular Ejection Fraction; V-HeFT II

	Enalapril	ISDN/HYD	p-value^*
Baseline Mean (SD)	n=388	n=384
Baseline Mean (SD)	28.6 (10.9)	29.4 (11.5)

Month 3 Mean (SD)	n=359	n=335
Month 3 Mean (SD)	31.0 (11.4)	32.3 (12.7)
Mean change (SD)	+2.1 (6.7)	+3.3 (7.1)	p=0.026

Month 12 Mean (SD)	n=308	n=275
Month 12 Mean (SD)	31.3 (12.4)	32.6 (13.7)
Mean change (SD)	+2.5 (8.4)	+3.6 (8.7)	p=0.12

Month 24 Mean (SD)	n=229	n=209
Month 24 Mean (SD)	31.6 (12.4)	33.1 (13.1)
Mean change (SD)	+2.5 (8.5)	+3.1 (9.9)	p=0.53

Month 36 Mean (SD)	n=141	n=137
Month 36 Mean (SD)	33.5 (13.7)	32.9 (12.8)
Mean change (SD)	+3.3 (10.3)	+3.7 (10.9)	p=0.68

^*p-values refer to between-group comparisons

4.2.7.3.7 Other Endpoints

The duration of tolerable exercise was somewhat greater in the ISDN/HYD group than in the enalapril group, but the difference between the groups achieved significance only at the first post-randomization measurement (month 3) and was not significant thereafter.

The cardiothoracic ratio assessed on a chest x-ray was similar in the two groups throughout the duration of the study.

4.2.7.4 Safety Results

4.2.7.4.1 Adverse events regardless of relationship to study drug

Table 12 lists the number of patients who reported adverse events. At each visit, investigators questioned the patients using a preprinted list of adverse events known to be associated with use of ISDN or HYD or enalapril; adverse events not on the preprinted list were recorded under “other.” A listing of specific “other” events is not available.

Adverse events related to systemic vasodilation (e.g., headache) were more frequent in patients receiving ISDN/HYD than enalapril.

4.2.7.4.3 Other safety topics

Vital signs

Following randomization, both systolic and diastolic blood pressures were consistently lower in the enalapril group than in the ISDN/HYD group; the difference between the two groups averaged about 5 mm Hg systolic and diastolic and was statistically significant at all time points.

Heart rate increased in the ISDN/HYD group and decreased in the enalapril group; the difference between the treatment groups in the mean change from baseline was statistically significant (p < 0.05) at each analysis at time point through month twelve and marginally significant from month fifteen through month twenty-one.

Further analysis demonstrated a significant interaction between changes in blood pressure and race (Table 13). During the first 6 months, systolic blood pressure decreased markedly in white patients treated with enalapril, decreased to an intermediate degree in black patients regardless of treatment, and increased slightly in white patients treated with ISDN/HYD. Hence, black patients showed a greater blood pressure reduction with ISDN/HYD than white patients, whereas enalapril and ISDN/HYD produced similar hypotensive effects in white patients. Therefore, the pattern of blood pressure effect closely paralleled the pattern of survival effects with the two treatments in the two racial groups.

Table 13. Effect of Race on Change in Systolic Blood Pressure With

Enalapril and ISDN/HYD; V-HeFT II

Time in Study	Non-black patients (mm/Hg)		Black patients (mm/Hg)		Race by treatment interaction p-value
Time in Study	Enalapril	ISDN/HYD	Enalapril	ISDN/HYD	Race by treatment interaction p-value
4 weeks	–7.4	–0.3	–4.0	–3.3	0.0067
3 months	–6.2	+0.9	–2.6	–2.8	0.0033
6 months	–4.6	+0.8	–1.7	–1.3	0.0694

Lupus syndrome

No patient in either treatment group was permanently discontinued from the study because of arthralgia. The number of patients who had arthralgia that led to dose reduction was somewhat higher in the ISDN/HYD group (n=44, 11.0%) than in the enalapril group (n=26, 6.4%). The arthralgias were considered severe and possibly or probably related to the study medication in 31 patients, 15 (3.7%) in the enalapril group and 16 (4.0%) in the ISDN/HYD group. In 27 patients, 10 in the enalapril group and 17 in the ISDN/HYD group, arthralgias were associated with a significant increase (³ 1:160) in ANA titer that was not preexisting and represented a worsening from baseline; this increase was sustained (³ 2 consecutive assessments excluding baseline) in 8 patients in the enalapril group and 12 patients in the ISDN/HYD group. In addition, a total of 15 patients (8 in the enalapril group and 7 in the ISDN/HYD group) interrupted or discontinued at least one study drug due to suspected lupus-like syndrome.

Clinical laboratory evaluations

No clinically relevant mean changes in values for clinical laboratory tests were seen during the study, except for small increases in blood urea nitrogen and serum creatinine in the enalapril patients as compared to the ISDN/HYD patients.

4.2.8 Summary and Conclusions for V-HeFT II

The findings of the second Vasodilator Heart Failure Trial (V-HeFT II) support the following conclusions:

· When compared with enalapril, the long-term administration of a combination of ISDN/HYD to middle-aged men with mild-to-severe heart failure treated with digitalis glycosides and diuretics was associated with a 23% greater risk of death. The p-value for this treatment difference 0.08 for the overall trial and 0.016 for the prespecified comparison of mortality rates at 2 years.

· The difference in survival between enalapril and ISDN/HYD seen in V-HeFT II was comparable in magnitude (23%) to the difference in survival seen in trials that have compared enalapril to placebo in mild-to-moderate heart failure (23% mortality reduction in a meta-analysis of all placebo-controlled trials of ACE inhibitors). Therefore, in the absence of V-HeFT I, the juxtaposition of the point estimates of V-HeFT II and the ACE inhibitors trials might suggest that the combination of ISDN/HYD had little effect on mortality in the majority of patients enrolled in V-HEFT II.

· It is therefore noteworthy that — although a superior survival effect of enalapril relative to ISDN/HYD was seen across nearly all of the subgroups examined — one notable exception was black patients. The hazard ratio for ISDN/HYD : enalapril was 1.32 in white patients but 1.01 for black patients, indicating that the superiority of enalapril over ISDN/HYD in the overall trial was driven primarily by the treatment difference seen in white patients. This could have occurred

— if enalapril was particularly ineffective in black patients as stated in Section 2.2 (a re-analysis of the SOLVD trial database has supported an attenuated effect of enalapril in black patients), or

— if the combination of ISDN/HYD was particularly effective in black patients (the subgroup analysis of V-HeFT I cited in Section 4.1.7.3.2 would support a particularly pronounced effect of the drug combination in black patients), or

— if both possibilities were correct.

· Further examination of a retrospective analysis of other subgroup effects in V-HeFT II did not confirm most of the subgroup hypotheses generated by V-HeFT I. Specifically, younger patients, diabetics, and patients with lower systolic blood pressures, responded better than placebo to ISDN/HYD in V-HeFT I, but responded worse to ISDN/HYD than enalapril in V-HeFT II (ISND/HYD : enalapril hazard ratios of 1.36 in younger patients, 1.35 in diabetics and 1.37 in patients with lower systolic blood pressures). Except for race, only one other subgroup effect seen in V-HeFT I was confirmed in V-HeFT II. Specifically, as in V-HeFT I, patients with an ejection fraction < 40% responded better to ISDN/HYD than patients with a higher ejection fraction (hazard ratio of 2.02 in patients with preserved ejection fractions as compared with 1.21 in patients with impaired ejection fractions).

· Time-to-event analyses of hospitalization for heart failure showed that during the first two years of the study, black patients treated with ISDN/HYD had a lower risk of hospitalization for heart failure than black patients treated with enalapril, whereas in white patients, the risk of hospitalization for heart failure was similar in the two treatment groups.

· Both enalapril and the combination of ISDN/HYD were associated with comparable improvements in left ventricular ejection fraction (about 2-3 units). The magnitude of this increase is similar to that which has been historically reported with angiotensin-converting enzyme inhibitors, which increase left ventricular ejection fraction by 2-3 units when compared with placebo. These data are consistent with the finding in V-HeFT I that the combination of ISDN/HYD increases left ventricular ejection fraction in heart failure.

· The combination of ISDN/HYD produced improvements in maximal exercise capacity in V-HeFT II that were generally superior to those produced by enalapril. This finding is noteworthy since several trials have reported that ACE inhibitors improve maximal exercise capacity.

· For all patients changes in quality of life were similar between the ISDN/HYD and the enalapril groups. At twelve months, a treatment difference in favor of ISDN/HYD was primarily seen in black patients with little difference seen in white patients (interaction p=0.09). This finding reinforces the impressions gained from V-HeFT I that future trials might appropriately seek to confirm the efficacy of ISDN/HYD, using an endpoint that measured both symptomatic and prognostic effects of drugs.

· Enalapril lowered both systolic and diastolic blood pressure more than the combination of ISDN/HYD. However, the greater hypotensive effects of enalapril were seen primarily in white patients. ISDN/HYD lowered systolic blood pressure more in black patients than white patients.

· The differences observed between ISDN/HYD and enalapril on survival, maximal exercise capacity and left ventricular ejection fraction reinforce the finding of V-HeFT I that the mechanisms by which drugs exert their vasodilator effects are relevant in determining their efficacy in the treatment of heart failure.

· The long-term administration of a combination of ISDN/HYD was associated with headache, dizziness and other vasodilator-type adverse reactions.

· A meaningful proportion of patients failed to attain target doses of ISDN/HYD. Clinical benefits were seen despite the use of lower-than-target doses, suggesting that future trials might appropriately target lower doses of ISDN/HYD.

The findings of V-HeFT II reinforced many of the key findings of and hypotheses derived from V-HeFT I. When taken together, the two trials suggest that the combination of ISDN/HYD may produce symptomatic and prognostic benefits that are particularly apparent in black patients; that such an effect might be most readily detected by an endpoint that simultaneously measures both effects; and that this benefit might be achieved at doses lower than the target doses used in V-HeFT I and V-HeFT II.

5.0 Trials with Isosorbide Dinitrate and Hydralazine Administered as a

Combination Product

5.1 African American Heart Failure Trial (A-Heft)

5.1.1 Study Overview

The African American Heart Failure Trial (A-HeFT) was a multicenter, randomized, double-blind, parallel group, placebo-controlled study conducted at 180 sites in the United States under the sponsorship of NitroMed. A-HeFT can be distinguished from V-HeFT I and V-HeFT II in the following ways:

• A-HeFT enrolled only African American patients.

• A-HeFT enrolled men and women.

• A-HeFT enrolled patients with New York Heart Association (NYHA) class III-IV symptoms.

• A-HeFT enrolled patients with heart failure due to left ventricular systolic dysfunction.

• A-HeFT enrolled patients generally taking ACE inhibitors/ARBs, beta blockers and/or aldosterone antagonists, in addition to diuretics and digitalis glycosides.

• ISDN/HYD were formulated and administered as a fixed-dose combination tablet.

• Target doses in A-HeFT were ISDN 40 mg TID and HYD 75 mg TID.

In contrast, V-HeFT I and V-HeFT II enrolled all races but only men who had class II-VI symptoms, had heart failure associated with both impaired and preserved ejection fraction, were generally taking only digitalis glycosides and diuretics and were titrated to target doses of ISDN 40 mg QID and HYD 75 mg QID; the drugs were administered as individual agents.

Hence, A-HeFT focused on the subgroups that were concordantly identified in both V-HeFT I and V-HeFT II as showing the most favorable survival effects of ISDN/HYD: African Americans and patients with systolic dysfunction.

5.1.2 Study Organization

The Steering Committee provided leadership for the overall trial. Its primary responsibilities were to periodically meet to review study status, make recommendations and approve all protocol amendments, and to oversee the conduct of the study.

The members of the Committee were:

Anne L. Taylor, M.D., University of Minnesota (chair)

Kirkwood F. Adams, Jr., M.D., University of North Carolina at Chapel Hill

Peter Carson, M.D., Veterans Affairs Medical Center

Jay N. Cohn, M.D., University of Minnesota

Keith Ferdinand, M.D., Xavier University College of Pharmacy

Elizabeth Ofili, M.D., Morehouse School of Medicine

Adeoye Olukotun, M.D., Clinical and Regulatory Strategies

Malcolm Taylor, M.D., University of Mississippi School of Medicine

Clyde W. Yancy, Jr., M.D., University of Texas Southwestern Medical Center

Susan Ziesche, R.N., Minneapolis VA Hospital

The Independent Central Adjudication Committee was responsible for the review of subject information in order to determine if clinical events that occurred during the course of the trial met pre-defined criteria for efficacy endpoints. The Committee adjudicated the following clinical events: deaths, all hospitalizations, unscheduled emergency room visits and unscheduled office/clinic visits for the treatment of HF, and new heart transplant listings. The decisions of the Committee were used for the final efficacy analyses.

The members of the Committee were:

Peter Carson, M.D., Veterans Affairs Medical Center, Washington, DC (chair)

Inderjit S. Anand, M.D., Veterans Affairs Medical Center, Minneapolis, MN

Jalal Ghali, M.D., Louisiana State University, Health Sciences Center, Shreveport, LA

Joann Lindenfeld, M.D., University of Colorado, Health Sciences Center, Denver, CO

Allan B. Miller, M.D., University of Florida Health Science Center, Jacksonville, FL

Christopher M. O’Connor, M.D., Duke University Medical Center, Durham, NC

Felix E. Tristani, M.D., Cold Springs, MN

A Data and Safety Monitoring Board composed of clinicians and a statistician who did not participate in the trial, periodically reviewed study results, evaluated the treatments for excess events, determined whether the basic trial assumptions remained valid, and made recommendations to the A-HeFT Steering Committee and NitroMed. In addition to periodic reviews of the safety data from the trial, the DSMB reviewed the results of two interim analyses for the reassessment of sample size, which were performed to ensure that the assumptions regarding the composite score primary efficacy endpoint remained valid during the study.

The members of the Committee were:

David DeMets, Ph.D., University of Wisconsin (chair)
Richard Grimm, M.D., Hennepin County Medical Center
Pamela Ouyang, M.D., Johns Hopkins University
Jackson Wright, Jr., M.D., Case Western Research University

An independent Statistical Data Analysis Center (Ralph D’Agostino, Jr., Ph.D., Wake Forest University) received interim data from the data management center at Medifacts International (the Contract Research Organization for the trial) and performed the statistical analyses for the DSMB.

5.1.3 Study Population

A-HeFT enrolled African American or black patients. A person was defined as “African American” or black if he/she designated himself or herself as such. In addition, patients were required to fulfill all of the following inclusion criteria and none of the following exclusion criteria.

5.1.3.1 Inclusion Criteria

· Men or women, at least 18 years old.

· Chronic heart failure of at least 3 months’ duration.

· NYHA class III-IV symptoms.

· Receiving appropriate therapy for heart failure, which was expected (but not required) to include a diuretic, an angiotensin-converting enzyme inhibitor or an angiotensin receptor antagonist and a beta blocker and could have also included digitalis, spironolactone or other medications. Patients receiving beta blockers were to have been taking them for at least three months.

· Symptomatically stable while receiving a stable treatment regimen for heart failure. Stability was defined as no change in signs or symptoms of heart failure, no weight change of > 2.5%, and no permanent changes in heart failure medication in the two weeks prior to randomization.

· Resting left ventricular ejection fraction £ 35%, or a resting left ventricular internal dimension > 2.9 cm/m² BSA (or > 6.5 cm) combined with a left ventricular ejection fraction < 45%, within the prior 6 months.

· Outpatient or inpatient (if patient was ready for hospital discharge).

· Ability to comprehend and complete the Minnesota Living with Heart Failure questionnaire.

5.1.3.2 Exclusion Criteria

· Female who was pregnant, nursing, or of childbearing potential while not practicing effective contraceptive methods.

· Significant valvular heart disease, obstructive hypertrophic cardiomyopathy, active myocarditis, or uncontrolled hypertension.

· Unstable angina, myocardial infarction or cardiac surgery including percutaneous transluminal coronary angioplasty within three months or likely to require coronary artery bypass grafting or percutaneous transluminal coronary angioplasty during the ensuing year.

· Cardiac arrest or a sustained ventricular tachycardia considered life threatening and requiring intervention within three months, unless treated with an implantable cardiac defibrillator.

· Stroke within three months.

· Parenteral inotropic therapy within one month.

· Rapidly deteriorating or uncompensated heart failure such that cardiac transplantation would be likely over the ensuing one year.

· Symptomatic hypotension.

· Significant hepatic, renal, or other disease that might limit survival over the ensuing one year.

· Any condition which, in the opinion of the investigator or medical monitor, would jeopardize the evaluation of efficacy or safety.

· Any contraindications to the use of isosorbide dinitrate or hydralazine.

· Receipt of another investigational drug or device within 3 months.

· Requirement for hydralazine, long-acting nitrates or phosphodiesterase type 5 inhibitors like sildenafil (Viagra®), vardenafil (Levitra®) or tadalafil (Cialis®) at study entry.

5.1.4 Study Plan

Patients fulfilling all inclusion criteria and none of the exclusion criteria were randomized to receive either BiDil® or matching placebo (in a 1:1 ratio) for the remainder of the study. Randomization was stratified according to the use of beta blockers at baseline.

BiDil® was supplied in the form of tablets containing a fixed-dose combination of isosorbide dinitrate 20 mg plus hydralazine 37.5 mg. Patients were initially instructed to take 1 tablet three times daily, which was to be increased to 2 tablets three times daily 3-5 days later if the medication was well tolerated. The goal was to achieve the target dose of 120 mg/day of isosorbide dinitrate and 225 mg/day of hydralazine (2 tablets TID). These doses were approximately 25% lower than the target doses used in V-HeFT I and V-HeFT II. The study medication was added to pre-existing medications used for the treatment of heart failure.

If the medication was not tolerated at the target dose, the patient was prescribed the highest tolerated dose of the study medication and the doses of other medications could be adjusted as clinically indicated. Patients continued to receive the study medication even if they sustained a clinical endpoint unless they experienced intolerable adverse events, life threatening laboratory abnormalities, cardiac transplantation or pregnancy, or whenever the investigator considered it in the patient’s best interest. Patients who stopped taking the study medication for adverse events remained in the trial and complied with all scheduled visits and assessments.

Following randomization, each patient was to be seen as an outpatient every three months until either reaching a maximum of 18 months of treatment or until the last patient randomized had completed 6 months. [Because of the early termination of the study, patients recruited in the latter months of the trial were not followed for a minimum of 6 months.] At each visit, patients were assessed for the occurrence of major clinical events and adverse events. Quality of life was assessed by the Minnesota Living with Heart Failure Questionnaire at baseline and every 3 months. In addition, 2-dimensional echocardiograms and measurements of brain natriuretic peptide were performed at baseline and at 6 months.

5.1.5 Study Assessments

5.1.5.1 Primary Efficacy Endpoint

The primary efficacy parameter was a composite score of clinical outcomes, calculated as the sum of the patient’s vital status during the first 18 months of the study; heart failure hospitalization status during the first 18 months; and change in quality of life at 6 months. Specifically,

Vital status during study

· If patient died, the score for this component would be -3

· If patient was alive at the end of the trial, the score for this component would be 0

Heart failure hospitalization status during the study

· If patient was hospitalized for heart failure, the score for this component would be -1

· If patient was never hospitalized for heart failure, the score for this component would

be 0

A hospitalization for heart failure was defined as a hospital admission, whose primary reason was worsening symptoms or signs of heart failure, and during which the patient required intravenous medications specifically for the treatment of heart failure, and which lasted more than one calendar day.

Change in quality of life (Minnesota Living with Heart Failure) at 6 months relative to baseline

· If quality of life improved ≥ 10 units, then the score for this component would be +2

· If quality of life improved ≥ 5 and < 10 units, then the score for this component

would be +1

· If quality of life changed < 5 units, then the score for this component would be 0

· If quality of life worsened ≥ 5 and < 10 units, then the score for this component

would be –1

· If quality of life worsened ≥ 10 units, then the score for this component would be –2

Each patient’s composite score was obtained by summing the three components. As a result, the worst possible score was –6, i.e., the patient showed marked worsening of quality of life (-2) at 6 months, was hospitalized for heart failure (-1), and died (-3). The best possible score was +2, i.e., the patient was alive (0), was never hospitalized for heart failure (0), and showed marked improvement in quality of life at 6 months (+2). By design, each patient had a score of 0 upon entry into the trial, since the composite reflects a change from an individual's baseline status.

The composite score used in A-HeFT was developed specifically for this study and had not been used in other heart failure trials; it was developed in an attempt to incorporate changes in quality of life while also including the occurrence of major clinical events (death or hospitalization for heart failure).

5.1.5.2 Secondary Efficacy Variables

· Individual components of the composite score primary endpoint

· Death

- Time to death using time-to-event methods

- Adjudicated causes of death

· Hospitalizations

- Time to first hospitalization

- Total number of hospitalizations for heart failure

- Total number of hospitalizations for any reason

- Total days in hospital

- Number of adjudicated unscheduled emergency room and office/clinic visits

An unscheduled emergency room visit or unscheduled office/clinic visit was classified as due to heart failure, other cardiac causes or non-cardiac causes. Emergency room visits and office/clinic visits were attributed to worsening heart failure if the patient had worsening signs and symptoms of heart failure and received intravenous medication specifically for the treatment of heart failure.

· Quality of life

- Change in overall score, and physical and emotional component scores relative to baseline

during the trial

· Newly recognized need for cardiac transplantation

Listing of a patient for cardiac transplantation following persistent decline in functional capacity, repeated hospitalization for heart failure and need for intravenous treatment with positive inotropic or vasodilator drugs.

· Echocardiographic measures

- Change from baseline at six months in left ventricular ejection fraction, left ventricular internal diastolic dimension and left ventricular wall thickness

· Serum levels of brain natriuretic peptide

All deaths, hospitalizations, unscheduled emergency room visits and unscheduled office visits were adjudicated by the Independent Central Adjudication Committee.

5.1.5.3 Safety Assessments

Safety assessments consisted of monitoring and recording all treatment-emergent adverse events and serious adverse events and the performance of physical examinations (which included the measurement of vital signs at every visit). Given the well-characterized safety profile of the components of BiDil® (ISDN/HYD), there was no routine monitoring of hematology, blood chemistry and urine values.

5.1.6 Statistical Plan and Analyses

5.1.6.1 Sample Size Determination and Interim Monitoring Plan

The primary endpoint of the trial was the clinical composite score, which combined information regarding the occurrence of death and hospitalization for heart failure during the treatment period together with change in quality of life at 6 months relative to baseline to generate a single score, whose value could range from +2 to –6. The original sample size for the study was 600 patients, i.e., 300 patients per treatment group, which was expected to provide 80% power to detect a 0.5 unit difference in the primary endpoint (α=0.05), assuming that enrollment would be completed in 6 months and the mean duration of treatment would be 8-9 months. These estimates were based on the data collected in V-HeFT II.

Because the primary efficacy variable had not been used previously in a heart failure trial, NitroMed was uncertain about the validity of the study assumptions regarding the magnitude and variability of the treatment effect and wished to utilize interim results to re-estimate the sample size of the study. Normally, increasing sample size based on an interim estimate of a treatment difference would be expected to substantially inflate the probability of a type I error. However, in 1999, statisticians at the FDA had developed a new group sequential test procedure that — by modifying the weights used in the traditional repeated significance two-sample mean test — was able to preserve the probability of a type I error at originally targeted level while providing a substantial gain in power if the sample size were increased⁹⁵.

Working with the FDA and using the method of Cui et al., NitroMed designed A-HeFT as a group sequential design with two interim analyses of the primary endpoint at 25% and 50% information time followed by a final analysis at the end of the study (total of 3 planned looks). For each of the two planned interim analyses, a statistician (independent of both NitroMed and the Data and Safety Monitoring Board) would provide a sample size estimate to the Data and Safety Monitoring Board. However, only the results of the second interim analysis would be communicated to NitroMed and used to modify the sample size if necessary. This second interim analysis was planned to occur at 50% information time — when approximately 300 patients had completed six months in the study. O’Brien-Fleming type boundaries were used; the two-sided p-values required for statistical significance were 0.00001 at the first interim look, 0.0052 for the second interim look and 0.0480 at the final look, reflecting a penalty of 0.002 as a result of the group sequential procedure.

In addition to these considerations, NitroMed and the FDA agreed — before the start of the study — to increase the sample size based on the data available at the interim analysis to theoretically attain α=0.02, in order to increase the strength of evidence provided by the study. To implement this agreement, NitroMed agreed to utilize a target α=0.02 (rather than 0.05) in the sample size calculations when the trial was resized at the time of the second interim analysis. As a result, based on the treatment difference observed during the second interim analysis, the sample size for the study was increased from 600 to 1100. Note however that the p-value considered to demonstrate statistical significance remained at p = 0.048.

No plan for early termination of the trial for a mortality benefit was devised.

5.1.6.2 Actions of the Data and Safety Monitoring Board

The Data and Safety Monitoring Board carried out the interim monitoring plan as follows:

March 19, 2002

The Data and Safety Monitoring Board met for the first time to reach agreement on the operations of the Board. There was no review of blinded or unblinded data. At the time of the meeting, 221 patients had been randomized, and 3 patients had died. The Board adopted procedures to evaluate data and for the re-estimation of sample size.

August 23, 2002

The Data and Safety Monitoring Board carried out its first interim analysis to look at data after approximately 150 patients had completed the 6-month visit. [Specifically, 137 patients had completed the 6-month visit.] At the time of the meeting, 310 patients had been randomized, and 8 had died. Data were presented to the Board as Group A and Group B. Treatment differences were observed (favoring Group B in all parameters) for deaths (3 vs 5), hospitalizations (7 vs 17) and quality of life (–8.6 vs –2.5). NitroMed was informed that data variability for the primary endpoint was similar to that anticipated by protocol; no recommendation for protocol modification was made at this time.

March 3, 2003

The Data and Safety Monitoring Board carried out its second interim analysis to look at the data after approximately 300 patients had completed the 6-month visit. [Specifically, 313 patients had completed the 6-month visit.] At the time of the meeting, 528 patients had been randomized, and 23 had died. Data were presented to the Board as Group A and Group B. Treatment differences were observed (favoring Group A) for deaths (10 vs 13) and favoring Group B for quality of life (–7.4 vs –1.1). The composite score also favored Group B (-0.38 on control and 0.01 on treatment). Although the magnitude of the difference in the composite score was somewhat less than that specified in the protocol, the variability was similar. The DSMB provided several sample size options to NitroMed which selected an increase in sample size to 1100 (based on the FDA recommendation to size the study so that an α of 0.02 could theoretically be attained) and agreed to meet again one year later.

March 13, 2004

The Data and Safety Monitoring Board carried out its third data review one year after the second interim analysis. At the time of the meeting, 798 patients had been randomized and 59 had died. Data were presented initially to the Board as Group A and Group B. Treatment differences were observed (favoring Group B) for deaths (21 vs 38). The difference in deaths was associated with a log-rank Z value of 2.37 (p=0.018). The DSMB decided to unblind itself and Group B was identified as the group receiving BiDil®. Recognizing that no boundaries to terminate the trial for mortality had been formulated at the start of the study, the DSMB established an O’Brien-Fleming type group sequential alpha spending function as described by Lan and DeMets to guide further decision making. Of note, the treatment difference in mortality seen at this meeting fell just below the boundary value specified by the newly formulated boundaries. The DSMB recommended one additional safety review to take place in approximately 3-5 months.

July 7, 2004

The Data and Safety Monitoring Board carried out its final data review. At the time of the meeting, 1014 patients had been enrolled in the trial, and 75 had died. Data were presented as BiDil® and placebo. Treatment differences were observed for deaths (27 vs 48); this difference was associated with long rank Z=2.47 with a p=0.0132. [The nominal p-value for the monitoring boundary at this time was 0.031.] The DSMB adjourned so that additional analyses could be carried out. On July 9^th these analyses were discussed by the DSMB and showed treatment differences (BiDil® vs placebo) for deaths (27 vs 48, p=0.012), first hospitalizations for heart failure (64 vs 103, p=0.001), quality of life (–7.5 vs –2.5, p=0.002), worsening heart failure as an adverse event (66 vs 94, p=0.023) and the composite score primary endpoint (+0.13 vs –0.39, p=0.0001). In view of crossing of the monitoring boundary for the mortality difference and considering the consistency of the treatment benefit across all major endpoints, the Data and Safety Monitoring Board unanimously agreed to recommend early termination of the trial and to notify the Steering Committee and NitroMed of their recommendation.

Following consultation with the FDA, NitroMed stopped A-HeFT on July 19, 2004. At this point, 1050 patients had been enrolled in A-HeFT, and of these, 951 patients had reached a minimum of 3 months on study and thus, would have had the opportunity to participate in the first scheduled post-baseline quality of life measurement.

5.1.6.3 Statistical Analyses

5.1.6.3.1 Primary Endpoint

The primary endpoint of the trial was the clinical composite score, which combined information regarding the occurrence of death and first hospitalization for heart failure during the entire treatment period together with changes in quality of life at 6 months relative to baseline into a single score, whose values could range from +2 to –6. Differences between the two treatment groups were tested for significance by the two-sample t-test, as modified based on the method of Cui et al⁹⁵. The primary analysis specified no adjustment for covariates. The primary analysis population was based on the intention-to-treat principle and consisted of all randomized patients, whether or not they received at least one dose of study medication. This was the primary efficacy population for the primary endpoint and for all-cause mortality.

It was anticipated that various components of the composite score primary endpoint would be missing at the end of the trial. To accommodate this possibility, the protocol specified that the worst possible score would be assigned to each component with missing data (–3 for patients without data on vital status; –1 for patients with missing heart failure hospitalization data; and –2 for patients without quality of life data). Given the early termination of the trial, a large number of patients recruited during the 6 months before study termination could not undergo a 6-month quality of life assessment. As a result, if a 6-month quality of life assessment was not available as defined by the protocol, the last available on-study assessment before 6 months was used. Even so, 99 patients had been recruited so close to the early termination date of the study that some did not have the opportunity to undergo even a 3-month quality-of-life assessment. Some returned for a final study close out visit, but most did not. Therefore, if a post-baseline quality of life assessment was not available, the patient was assigned a worst score of –2.

Of note, at the conclusion of the trial, no patients were lost to follow-up for the assessment of vital status; 24 patients (2.3%) were lost to follow-up for the assessment of heart failure hospitalization and were assigned the worst score for this component (-1); and 81 patients (7.7%) had no quality of life measurement performed after starting study medication and were assigned the worst score (-2) for this component.

The following subgroups were analyzed for the composite score primary endpoint and for mortality: age (less than 65 years or at least 65 years); gender (male or female); etiology of heart failure (ischemic or non-ischemic); baseline systolic blood pressure (more or less than 125 mm Hg); patients with or without history of hypertension, diabetes mellitus or chronic renal insufficiency at baseline; and patients taking or not taking the following medications at baseline: angiotensin-converting enzyme inhibitors, angiotensin receptor blockers, beta blockers, calcium channel blockers, aldosterone antagonists, non-aldosterone antagonist diuretics, and digoxin.

5.1.6.3.2 Secondary Endpoints

Kaplan-Meier survival curves were used to display time-to-event analyses for death; for first hospitalization for heart failure; and for death or first hospitalization for heart failure. Patients who had cardiac transplantation were censored at the time of transplantation. Survival curves were compared using the log-rank test. Hazard ratios and 95% confidence intervals were generated using the Cox proportional hazards model.

For all variables other than the primary efficacy variable (the composite score) and the time-to-event survival analyses, comparisons were carried out on patients with paired data, with no imputation for patients with missing data. The change from baseline in Minnesota Living with Heart Failure quality of life overall score and its physical and emotional components was analyzed by two-sample t-tests. The number of hospitalizations for heart failure and for any cause and the total number of days in the hospital for heart failure and for any reason were analyzed by the Wilcoxon rank sum test.

The safety population consisted of all patients who were randomized, received at least one dose of study medication, and had at least one post-baseline measurement of safety.

5.1.7 Results

5.1.7.1 Baseline Characteristics

A total of 1050 patients were randomized into A-HeFT, 532 to placebo and 518 to BiDil®.

The patients enrolled in A-HeFT were middle-aged men and women (Table 14). The most common cause of heart failure was hypertensive heart disease; less than one-fourth of the patients had heart failure due to ischemic heart disease. More than 90% of the patients had NYHA class III symptoms. In general, the two treatment groups were well-matched for baseline characteristics; more men were randomized to the placebo group (p=0.01) and baseline diastolic blood pressure was higher in the BiDil® group (p=0.002).

	V-HeFT I	V-HeFT II	A-HeFT
Sponsor	Veterans Affairs	Veterans Affairs	NitroMed
Number of Patients	642	804	1050
Gender	Men	Men	Men & women
Race	All races	All races	African Americans
Drugs Studied	Placebo ISDN/HYD Prazosin	Enalapril ISDN/HYD	Placebo ISDN/HYD
Target Doses of ISDN/HYD	ISDN 40 mg QID HYD 75 mg QID	ISDN 40 mg QID HYD 75 mg QID	ISDN 40 mg TID HYD 75 mg TID
ISDN/HYD	As individual products	As individual products	As fixed-dose combination tablet (BiDil®)
Severity of Heart Failure	Mild-to-severe	Mild-to-severe	Moderate-to-severe
Background Therapy for Heart Failure	Digoxin Diuretics	Digoxin Diuretics	Digoxin Diuretics ACE inhibitors/ARBs Beta blockers Aldosterone antagonists

Preferred Term	Enalapril (n=403)		ISDN/HYD (n=401)
Preferred Term	n	(%)	n	(%)
Lassitude/fatigue	330	81.9	326	81.3
Headache	242	60.0	307	76.6
Arthralgias	288	71.5	276	68.8
Nasal congestion	272	67.5	271	67.6
Dizziness	269	66.8	268	66.8
“Other”	262	65.0	246	61.4
Palpitation	217	53.8	227	56.6
Nausea and vomiting	237	58.8	213	53.1
Chest pain	187	46.4	178	44.4
Constipation	176	43.7	169	42.1

	BiDil® (N = 518)	Placebo (N = 532)
Age (years)	56.8 (12.7)	56.9 (13.3)
Sex, men/women (n)	290/228†	340/192
Etiology of heart failure, n (%)
Ischemic	121 (23.4)	121 (22.7)
Idiopathic	127 (24.5)	147 (27.6)
Hypertensive	207 (40.0)	199 (37.4)
Valvular	13 (2.5)	17 (3.2)
Other	50 (9.7)	48 (9.0)
Ejection fraction, %, mean (SD)	23.9 (7.3) n = 517	24.2 (7.5) n = 532
Left ventricular internal diastolic dimension (cm), Mean SD	6.5 (0.9) n = 330	6.5 (1.0) n = 332
Baseline NYHA class, n (%)
I	1 (0.2)	1 (0.2)
II	9 (1.7)	2 (0.4)
III	493 (95.2)	503 (94.7)
IV	15 (2.9)	25 (4.7)
Missing	0 (0.0)	1 (0.2)
Systolic blood pressure, mm Hg mean (SD)	127.2 (17.5)	125.3 (18.1)
Diastolic blood pressure, mm Hg mean (SD)	77.6 (10.3)†	75.6 (10.6)
Heart rate, beats/min Mean (SD)	74.2 (12.3)	73.1 (11.0)

	BiDil® (N = 518)	Placebo (N = 532)
Cardiovascular history (n, %)
History of hypertension	472 (91.1)	468 (88.0)
Arrhythmias	169 (32.6)	184 (34.6)
Diabetes mellitus	232 (44.8)	197 (37.0)
Hyperlipidemia	289 (55.8)	263 (49.4)
Cerebrovascular disease	79 (15.3)	74 (13.9)
Peripheral vascular disease	58 (11.2)	71 (13.3)
Chronic obstructive lung disease	91 (17.6)	110 (20.7)
Chronic renal insufficiency	84 (16.2)	97 (18.2)
Valvular disease	186 (35.9)	194 (36.5)
Previous revascularization	111 (21.4)	96 (18.0)
Pacemaker or implantable defibrillator	86 (16.6)	92 (17.3)
Previous myocardial infarction	152 (29.3)	152 (28.6)
Current angina	75 (14.5)	78 (14.7)
Current smoking	143 (27.6)	140 (26.3)
Previous smoking	306 (59.1)	336 (63.2)
Background medications (n, %)