Failing with women: How clinical trials fail our largest subgroup

More than 2.5 million women in the United States have heart failure, resulting in approximately 500,000 hospitalizations and nearly 130,000 deaths each year. Women have a greater risk of dying from heart failure than from breast, ovarian, or lung cancer. However, despite the significant morbidity and mortality associated with heart failure in women, there is a paucity of prospectively collected data guiding their treatment.

Etiology

Several cardiomyopathies show a higher prevalence in women. Hypertension is the most prevalent cause of heart failure in women, resulting in a greater percentage of women with diastolic heart failure compared with men.^1,2 Toxic cardiomyopathies, commonly resulting from the use of doxorubicin [Adriamycin, Rubex] and the HER-2 antagonists), occur predominantly in women because of the use of these drugs in the treatment of several gynecologic tumors.³ Stress-related or Takotsubo cardiomyopathy has been described as reversible left ventricular dysfunction in response to stress, and occurs predominantly in middle-aged to elderly women.⁴ Peripartum cardiomyopathy, characterized by the development of heart failure within the last month of pregnancy or within 5 months of delivery, affects an estimated 1 in 15,000 deliveries in the United States.^5,6 Although many other causes of cardiomyopathy may be gender-specific, susceptibilities or responses to inciting stimuli (such as thyroid abnormalities, catecholamines, alcohol, or infection) have not been systematically studied outside of animal models.^7-9

Clinical presentation

Individuals with heart failure, regardless of their sex, may present with varying degrees of dyspnea on exertion, fatigue, orthopnea, paroxysmal nocturnal dyspnea, and shortness of breath at rest. Nevertheless, a multitude of cardiovascular (ie, ischemia and arrhythmia) and noncardiovascular (ie, gastrointestinal, pulmonary, and hematologic) presentations may either obscure or coexist with a diagnosis of heart failure. At least one study has noted a higher incidence of nausea and chest pain in women, symptoms that could potentially detract from an accurate diagnosis of heart failure.¹⁰ At least 1 study has noted a higher incidence of nausea and chest pain in women with heart failure, symptoms that could potentially detract from an accurate diagnosis.¹⁰ Consistent with this is the finding of Lenzen and colleagues, who saw a statistically significant greater admission of men to cardiology versus general medicine wards.¹¹ Yet, few data are available elucidating these issues. Compared with men with similar left ventricular dysfunction, several studies have shown that women have more difficulty exercising and performing normal daily activities.^12-22 Women presenting to the emergency department with acutely decompensated heart failure may have a higher incidence of congestive signs of heart failure, such as jugular venous distention, rales, and lower-extremity edema, than men.¹⁰ In addition, women with heart failure are also more likely to be depressed than men.^23-25

Data describing the sensitivity, specificity, and accuracy of the physical examination in detecting and monitoring heart failure (ie, jugular venous distention, rales, S3, hepatojugular reflux, abdominal ascites, and peripheral edema) in women are not available, despite differences in general body habitus, neck contours, the risk of coexistent primary pulmonary disease, and differences in lymphatic drainage.

Diagnostic interventions

Diagnostic modalities, such as transthoracic echocardiography and the measurement of serum B-type natriuretic peptide (BNP or N-terminal pro B-type natriuretic peptide [NT-pro BNP]) levels, have been used to aid in diagnosing the presence and severity of heart failure. Although no obvious limitations in the performance or interpretation of echocardiograms in women have been noted, the higher incidence of hypertension-induced heart failure in women results in higher reported mean left ventricular ejection fractions and an increased relative incidence of diastolic dysfunction compared with men.^10,26-28

Studies have reported that BNP measurements are useful in 4 clinical contexts: screening for asymptomatic left ventricular dysfunction,²⁹ diagnosing the presence of new-onset heart failure, triaging patients with acute shortness of breath in the emergency department,³⁰ and the continued care management of heart failure patients.^31-34 Population-based analyses have reported that BNP and NT-pro BNP levels are higher and more variable in women than in men.^35-39 Although sex-specific receiver operating characteristic analyses have described a similar sensitivity and specificity of BNP in screening for structural heart disease in men and women,⁴⁰ data from the Framingham database have shown severe limitations in using the BNP level as a screen for heart failure in healthy young women.⁴¹ In the acute setting, the Breathing Not Properly study of 883 men and 703 women showed similar abilities of BNP values to predict the presence of heart failure in both men and women.⁴² However, across the sample cut points, BNP levels in women were consistently less sensitive and less specific than in men.⁴² In the Rapid Emergency Department Heart Failure Outpatient Trial (REDHOT) study, which included 464 patients admitted to the hospital for heart failure with BNP levels > 100 pg/mL, patients with gray-zone BNP levels (ie, 100-500 pg/mL) were most likely to be women (58.4% vs 41.6%), whereas patients with BNP levels > 500 pg/mL were more likely to be men (60% vs 40%).⁴³ Prognostically, it has been suggested that BNP levels are stronger predictors of mortality in women than in men.⁴⁴ Whether this will be confirmed with the larger Acute Decompensated Heart Failure National Registry (ADHERE) database remains to be determined.⁴⁵

Management and therapy

In the majority of therapeutic studies of heart failure, women account for only 15% to 30% of the total researched population. There is a paucity of prospective outcomes studies on women, and the majority of data used to treat women were gathered from inherently limited post-hoc analyses.⁴⁶ Of all the sex-specific differences shaping the evidence-based management of heart failure, none weighs as heavily as the apparent lower risks of arrhythmia and sudden death in women compared with men. Despite accounting for 28% to 35% of all cardiac deaths in women in the Nurses' Health Study,⁴⁷ women with systolic dysfunction appear to suffer less syncope and arrhythmia-associated sudden death than do men,^48,49 and may be less susceptible to ventricular arrhythmias.⁵⁰ It is therefore reasonable to question whether women derive as much benefit from therapies that reduce sudden death as do men.

With respect to diuretics, mortality data are limited for both sexes. One multivariate analysis, consisting of 362 furosemide (Lasix)-treated heart failure patients (162 women and 200 men), demonstrated an increased 5-year mortality risk only among the female subjects taking at least 80 mg of furosemide daily (P = .02).⁵¹ While awaiting confirmation in other populations, these data are consistent with a sex-selective increase in diuretic-induced hypomagnesemia,⁵² glucose intolerance, and thiamine deficiency.⁵³

According to the Heart Failure Society of America 2006 guidelines, beta blockers and angiotensin-converting enzyme (ACE) inhibitors are recommended for all women with heart failure secondary to left ventricular systolic dysfunction, both symptomatic (class B) and asymptomatic (class B for ACE inhibitors, class C for beta blockers). Carvedilol (Coreg), metoprolol succinate (Toprol-XL), and bisoprolol (Zebeta) are the only 3beta blockers that have been shown to be effective in reducing both the morbidity and mortality of heart failure patients.^54-56 Women accounted for 17.9% of study participants (1766 women and 8114 men). Post-hoc analyses have suggested that although carvedilol may be associated with improved survival and reduced hospitalization rates for women,^54,55 the use of metoprolol succinate in women has only been shown to reduce hospital readmission.⁵⁷ Whether these differences would be present in a prospective trial is unknown. Moreover, the importance of the differential actions of these beta blockers (metoprolol succinate is a beta-1 selective antagonist, whereas carvedilol is a nonselective beta-adrenergic antagonist with alpha blocking properties) has not yet been determined.

Angiotensin-converting enzyme inhibitors have been shown to reduce the risk of dying for the combined end point of death or hospitalization in women, as evidenced by 2 sex-based meta-analyses comparing ACE inhibitors with placebo.^58,59 Because the studies evaluated by these meta-analyses were all performed before the era of widespread beta blocker use, and given the wide confidence intervals (CIs) that each contains, the validity of these findings is being debated. This is not trivial based on the finding of a greater risk of ACE-inhibitor—induced side effects in women compared with men.⁶⁰ With regard to angiotensin receptor blockers (ARBs), candesartan (Atacand) is the only ARB that has been shown to reduce mortality in women with systolic heart failure,⁶¹ but as a class, ARBs are inferior to ACE inhibitors in reducing the risk of dying.

Two of the 3 studies of oral vasodilators (hydralazine [Apresoline] in addition to nitrates) for patients with heart failure failed to include any women because of their Veterans Affairs populations.^62,63 The recent African American Heart Failure Trial (A-HeFT), which included a patient population of 40% women, compared the addition of hydralazine and nitrate to optimal medical management (consisting of ACE inhibitors and/or ARBs) with optimal therapy alone.⁶⁴ The study suggested similar clinical benefits in the combined end point of death and cardiovascular hospitalization (33% reduction for men vs 38% reduction for women). However, data concerning the sex-specific side effects of this combination, which can be limiting in a large number of patients, have not yet become available.

The Randomized Aldactone Evaluation Study (RALES)⁶⁵ and Eplerenone Post—Acute Myocardial Infarction Heart Failure Efficacy and Survival Study (EPHESUS)⁶⁶ examined the use of mineralocorticoid antagonists in the management of heart failure in 2 populations. RALES showed a similar relative risk reduction for mortality in both sexes (30% in men and 29% in women) for stable New York Heart Association class III and class IV heart failure patients with systolic dysfunction, whereas EPHESUS showed a 25% reduction in men and a 17% reduction in women with heart failure in the peri-infarct period.^65,66 However, the significance of the reduction in women with peri-infarct left ventricular dysfunction remains unproven, as the CI passed through 1.0.

A higher mortality risk was observed in women with systolic heart failure treated with digoxin (Lanoxin) in a retrospective, post-hoc analysis of the Digitalis Investigation Group (DIG) trial.⁶⁷ Higher serum digoxin levels in women may possibly be the cause, resulting in a general recommendation to discontinue the drug or maintain a dose of 0.125 mg or less when use is necessary.⁶⁸

The Multisite Stimulation in Cardiomyopathy (MUSTIC),^69,70 Pacing Therapies for Congestive Heart Failure (PATH-HF),^71,72 Myocardial Ischemia Reduction With Aggressive Cholesterol Lowering (MIRACL), and MIRACL-Implantable Cardioverter-Defibrillator (ICD)^73,74 studies suggested that biventricular pacemakers could improve patient symptoms, exercise tolerance, and quality of life in carefully selected heart failure patients, although few women were available for analysis. The Comparison of Medical Therapy, Pacing, and Defibrillation in Heart Failure (COMPANION) study demonstrated that biventricular pacing could decrease the combined end point of risk of first hospitalization plus death from any cause at a mean of 17 months.⁷⁵ Although 33% of the population were women (493/1520) and similar efficacy was seen in both men and women, the CIs for both again passed through 1.0. The Cardiac Resynchronization-Heart Failure (CARE-HF) study demonstrated that biventricular pacemaker therapy reduced the primary end point of time to death from any cause or unplanned hospitalization for a major cardiovascular end point.⁷⁶ A statistically significant reduction in death from any cause and cardiovascular hospitalization (hazard ratio [HR] = 0.64; 95% CI, 0.42-0.97) was shown for the 27% of patients who were women (215 of 811 patients), which was similar to the results for men.

Finally, studies delineating the role of ICDs in patients with heart failure included few female patients.^77,78 Although the Multicenter Automatic Defibrillator Implantation Trial (MADIT)-II demonstrated the benefit of ICDs on overall mortality in patients with QRS prolongation and an ischemic cardiomyopathy (left ventricular ejection fraction ≤ 30%),⁷⁹ only approximately 16% of all patients studied were women, and the 95% CI passed through unity. The Defibrillators in Non-Ischemic Cardiomyopathy Treatment Evaluation (DEFINITE) ICD study demonstrated that ICD implantation in individuals already taking an ACE inhibitor and a beta blocker for nonischemic dilated cardiomyopathy, presenting with either ventricular premature contractions or nonsustained ventricular tachycardia, reduced the risk of dying by approximately 45% in men but did not show a survival advantage in women.⁸⁰ The Sudden Cardiac Death in Heart Failure Trial (SCDHeFT) evaluated both ischemic and nonischemic cardiomyopathy⁸¹ and showed a 23% improvement in survival in the overall population (HR = 0.96; 97.5% CI, 0.58-1.61); in this case, however, women did appear to have the same benefit as men (HR = 0.73; 97.5% CI, 0.57-0.93). The apparent overall diminished efficacy of ICDs in women may be the result of a lower prevalence of ventricular tachycardia and sudden death in women compared with men (52% of the men but only 34% of the women; P < .01).⁸¹ Although the mechanism for this lower prevalence is unknown, several authors point to the potential ameliorating effects of estrogen or the greater prevalence of hypertension-induced heart failure in women as possible explanations. Thus, heart failure therapies directed toward reducing the risk of sudden death require closer scrutiny in women.

Conclusions

The lack of gender-specific data severely limits the use of evidence-based medicine for managing heart failure in women. Despite clear gender-based differences in the pathophysiology and clinical presentation of the syndrome, adequately designed and prospective trials to assist with heart failure management in women are sorely lacking. For now, the data only indicate that we really don't know "how to treat a lady."