Introduction
Risk factors or markers for mortality in heart failure (HF) include a long list of clinical parameters, including history, physical examination, laboratory values, hemodynamics, cardiac structure and function, and biomarkers ( Table 3.1 ). This chapter reviews predictors of survival in HF and compares acute versus chronic HF, HF with reduced ejection fraction (HFrEF), and preserved ejection fraction (HFpEF), as well as different subpopulations such as transplant candidates and elderly patients.
| Demographic parameters | Age, etiology, gender, race |
| Functional parameters | NYHA class, peak VO 2 , % predicted peak VO 2 , VE/VCO 2 ratio or slope, anaerobic threshold, circulatory power (peak VO 2 × systolic blood pressure), cardiac power (cardiac output × mean arterial pressure/451), oxygen kinetics (oxygen debt, recovery time), oxygen uptake efficiency slope, exercise oscillatory breathing pattern, chronotropic incompetence (i.e., failure to achieve 85% of age predicted maximum HR or low chronotropic index HR adjusted to maximal exercise test level), HR recovery, 6-minute walk test |
| Physical signs | ↑ HR, ↓ blood pressure, S 3 , BMI, mitral regurgitation |
| Ventricular structure and function | LVEF, ventricular volumes, mitral regurgitation severity, RVEF, tricuspid annular plane systolic excursion |
| Hemodynamic parameters | RAP, PVR, PCWP, cardiac output, cardiac index, LVSWI, RVSWI, dP/dt, load-corrected PCW |
| Laboratory values | Sodium, BUN, creatinine, hemoglobin, hypoalbuminemia, ESR, WBC, urine albumin excretion, insulin resistance, cholesterol, INR, bilirubin |
| Neurohormones | NE, BNP, N terminal-pro BNP, angiotensin II, aldosterone, endothelin, vasopressin, adrenomedullin, IGF-I, testosterone, copeptin, dehydroepiandrosterone, pro-ANP, MR-pro-ANP, galectin-3, tumor necrosis factor, C-reactive protein, interleukin-6, soluble CD 14 |
| Biomarkers | Homocysteine, carbohydrate antigen 125, troponin I, troponin T, cystatin C, neutrophil gelatinase–associated lipocalin, growth differentiation factor 15, ST-2, type 1 collagen telopeptide |
| Electrocardiogram parameters | ↑ QRS duration, ↑ Q-Tc interval, abnormal SAECG, T-wave alternans, ↓ HR variability, atrial fibrillation, ventricular tachycardia |
| Comorbidities | Diabetes, obesity, renal insufficiency, dementia, sleep apnea, mobility disorders |
| Medical therapy | Inotropes, inability to tolerate beta blockers or ACEIs |
| Recent heart failure hospitalization |
Although the list of univariable predictors in Table 3.1 is long, there is no perfect prognostic indicator—a variable that, in isolation, would reliably and easily identify a patient’s risk. However, some variables are more powerful than others, such as peak oxygen consumption (VO 2 ), which will be reviewed in detail. Additional key prognostic indicators include the New York Heart Association (NYHA) and Interagency Registry for Mechanically Assisted Circulatory Support (INTERMACS) classifications, severely reduced left ventricular ejection fraction (LVEF), hypotension (systolic blood pressure < 90 mm Hg), hyponatremia, elevated B-type natriuretic peptide (BNP or NTproBNP), troponin, and renal dysfunction. However, an ideal biomarker or test—one that has high specificity and sensitivity, is reproducible, has low cost, is generally available, is easily measured, and has been prospectively tested and validated across all demographic groups—does not exist. Hence, prediction of patient prognosis in HF ultimately relies on integration of a careful clinical and laboratory evaluation interpreted in the context of the specific patient. Analysis of the many clinical and laboratory parameters can be performed using statistical models as reviewed at the end of the chapter.
Acute versus chronic heart failure
Most data on risk prognostication are derived from populations with chronic HF. However, populations with chronic and acute HF are obviously intimately connected, and for most patients, the distinction between acute and chronic HF is simply temporal. In fact, approximately two-thirds of patients admitted with HF have an antecedent diagnosis of chronic HF, and likewise, most patients with chronic HF will, at some point in time, have suffered from acute HF.
Acute decompensated HF (ADHF) is a leading cause of hospitalization in the developed world. Although the rate of HF hospitalizations in the United States has decreased from > 500 to approximately 400 per 100,000 individuals per year since the early 2000s, more than 900,000 patients are admitted for HF every year. Hospitalization as an event identifies a highly vulnerable period in the HF patient’s journey: data from large US registries reveal that mortality is 3%–4% in-hospital, 10%–12% at 30 days, and 30%–40% after 1 year. Rehospitalization rates are high and are generally reported to be 45%–65% within the first year.
ADHF encompasses a heterogeneous group of patients whose clinical presentation may identify patients at very high risk. Patients admitted with cardiogenic shock after myocardial infarction have a 30-day mortality that exceeds 40%. The great majority of patients admitted with ADHF are not in cardiogenic shock. In those who are, admission profiles associated with greatest risk include patients with acute myocardial infarction, ischemia as evidenced by electrocardiographic changes or elevated troponin T or I, hypotension (systolic blood pressure < 115 mm Hg), renal insufficiency (blood urea nitrogen [BUN] > 43 mg/dL and creatinine > 2.74 mg/dL), tachycardia, hyponatremia, reduced ejection fraction, increased BNP or NTproBNP, older age, and presence of multiple comorbidities. Patients admitted who require vasoactive drugs have a poor prognosis with increased risk of death. The need for inotropic support in the Acute Decompensated Heart Failure National Registry (ADHERE) study corresponded to an in-hospital mortality rate of 12% to 13%, and even higher rates have been reported outside the United States (26% in the Acute Heart Failure Global Survey of Standard Treatment [ALARM-HF] registry).
In contrast, hypertensive patients (systolic blood pressure > 160 mm Hg) may appear most acutely ill, but they generally have the best prognosis, with a low 60-day mortality. Similarly, patients presenting with ADHF secondary to medical noncompliance, poorly controlled hypertension, and normal cardiac troponin I levels also have a good prognosis ( Table 3.2 ).
| Admission Profile Associated With Good Outcome (Low In-Hospital Mortality) | Admission Profile Associated With Poor Outcome |
|---|---|
| Hypertensive | Elderly |
| Normal renal function | Blood pressure < 100 mm Hg |
| Serum sodium > 136 mEq/L | Creatinine > 2.5 mg/dL |
| Normal troponin | Elevated troponin or ischemic ECG changes |
| Absence of comorbidity | Pneumonia or significant comorbidity (i.e., dementia, cancer, CVA) |
| Noncompliance | Good compliance with medications and diet |
| HFpEF | Markedly elevated BNP |
The INTERMACS classification was recently introduced to capture clinical characteristics of patients with advanced HF, both hospitalized and ambulatory. By defining seven clinically recognizable groups, patients can be categorized with specific focus on potential timing for advanced therapies, especially mechanical circulatory support ( Table 3.3 ). The INTERMACS classification has been shown to correlate with the outcome of patients, even if not managed by mechanical circulatory support. In a study of ambulatory, the 12-month overall survival in non–inotrope-dependent HF patients was 60%, 74%, and 84% in patients in INTERMACS 4, 5, and 6/7, respectively ( Fig. 3.1 ).
| INTERMACS Class | Clinical Description |
|---|---|
| 1 | Critical cardiogenic shock describes a patient who is “crashing and burning,” in which a patient has life-threatening hypotension and rapidly escalating inotropic pressor support, with critical organ hypoperfusion often confirmed by worsening acidosis and lactate levels. |
| 2 | Progressive decline describes a patient who has been demonstrated “dependent” on inotropic support but nonetheless shows signs of continuing deterioration in nutrition, renal function, fluid retention, or other major status indicator. |
| 3 | Stable but inotrope dependent describes a patient who is clinically stable on mild–moderate doses of intravenous inotropes (or has a temporary circulatory support device) after repeated documentation of failure to wean without symptomatic hypotension, worsening symptoms, or progressive organ dysfunction (usually renal). |
| 4 | Resting symptoms describes a patient who is at home on oral therapy but frequently has symptoms of congestion at rest or with activities of daily living (ADL). He or she may have orthopnea, shortness of breath during ADL such as dressing or bathing, gastrointestinal symptoms (abdominal discomfort, nausea, and poor appetite), disabling ascites, or severe lower extremity edema. |
| 5 | Exertion intolerant describes a patient who is comfortable at rest but unable to engage in any activity, living predominantly within the house or housebound. This patient has no congestive symptoms but may have chronically elevated volume status, frequently with renal dysfunction, and may be characterized as exercise intolerant. |
| 6 | Exertion limited also describes a patient who is comfortable at rest without evidence of fluid overload but who is able to do some mild activity. ADLs are comfortable and minor activities outside the home such as visiting friends or going to a restaurant can be performed, but fatigue results within a few minutes of any meaningful physical exertion. |
| 7 | Advanced NYHA Class 3 describes a patient who is clinically stable with a reasonable level of comfortable activity, despite history of previous decompensation that is not recent. |
Hospitalization as a Prognostic Marker
Hospitalization for HF is an event indicating disease progression, increased mortality, and increased risk of early rehospitalization. Analysis of the health care usage database during the period 2000–2004 in British Columbia, Canada, demonstrated that successive heart failure hospitalizations were associated with incremental mortality. ( Fig. 3.2 ). In ambulatory patients with chronic HF, hospitalization for HF is associated with a marked increase in mortality.
Heart failure with preserved ejection fraction
The clinical profiles of patients with HFpEF versus HFrEF are different, with patients with HFpEF being more frequently female, elderly, hypertensive, and obese, with atrial fibrillation and chronic obstructive pulmonary disease. There is more clinical heterogeneity in HFpEF than in HFrEF. The profile of patients with LVEF in the medium range (LVEF 40%–50%), the so-called HFmrEF, resembles that of HFrEF more than HFpEF, but the prevalence of most clinical characteristics and comorbidities falls in between that of HFpEF and HFrEF. The survival rates of patients with HFpEF are mostly reported as better than the survival rates of patients with HFrEF, especially when correcting for the considerable difference in age ( Fig. 3.3 ). The survival of patients with HFmrEF appears to be better than both for patients with HFpEF and HFrEF, underscoring that LVEF is an imperfect tool for risk stratification.
Information about predictors of mortality in HFpEF is less abundant than in HFrEF. As in patients with HFrEF, age, higher NYHA class, higher BNP or NT-proBNP, expired volume-to-carbon dioxide consumption (VE/VCO 2 ) ratio, pulmonary artery pressures, exercise oscillatory breathing, reduced peak VO 2 , renal insufficiency, diabetes, anemia, hyponatremia, dementia, and peripheral arterial disease are associated with reduced survival in HFpEF while male gender and coronary artery disease have not been identified as predictors of reduced survival in this condition.
Cardiac amyloidosis, originally thought to be rare, has proven to be a common cause of HFpEF, affecting more than 10% of those with this diagnosis. The later stages of cardiac amyloidosis are characterized by severe restrictive cardiomyopathy, and in some patients, progression to HFrEF is seen. Prognosis in cardiac amyloidosis is worse than in HFpEF associated with other etiologies. In a recent large randomized clinical trial testing tafamidis in patients with transthyretin cardiac amyloidosis (mainly NYHA II–III), 43% of patients in the placebo group had died after 30 months. In patients with the other major form of cardiac amyloid, light chain or primary amyloidosis, mortality is even higher, with reported 2-year mortality rates exceeding 50%. Troponin I or T and especially BNP or NT-proBNP have emerged as potent plasma markers of mortality in cardiac amyloidosis.
Sudden cardiac death versus progressive heart failure
Historically, up to 50% of patients with HF have died suddenly, the majority from malignant ventricular arrhythmias. The use of primary and secondary prophylactic defibrillators has changed this picture in populations where defibrillator use is widespread. However, even in populations with low uptake of implantable cardioverter defibrillator (ICD) therapy, the rate of sudden cardiac death (SCD) has declined. In a recent analysis of randomized drugs trials in HF over the last 20 years, the annual rate of SCD decreased from more than 6% to 3%. The proportion of SCD to total mortality, however, was constant at approximately 30%–40% in these trials.
The incidence of SCD is higher in patients with mild HF than in patients with advanced disease, likely because of a higher risk of death from pump failure in the latter population. Data from MERIT-HF (Metoprolol CR/XL Randomised Intervention Trial in Congestive Heart Failure) showed that the distribution of the mode of death changes as HF advances. In patients with NYHA class II chronic HF, 64% of the deaths were classified as sudden death, and 12% were due to progressive HF, whereas in patients with NYHA class IV chronic HF, 33% died suddenly, and 56% died of progressive HF. In particular, beta-blockers and spironolactone have contributed to a reduction in the incidence of both sudden deaths and progressive HF deaths. Parameters that have been identified as prognostic indicators for SCD include the presence of severely reduced LVEF; ischemic etiology; prolonged QRS duration; presence of nonsustained ventricular tachycardia; prior episode of SCD; history of syncope; elevated BNPs; and in patients with an ischemic etiology, abnormal T-wave alternans. Elevated ST2, an interleukin-1 receptor family member, which has effects on myocardial fibrosis, has been reported to be predictive of SCD complementary to BNP. In addition, cardiac imaging using gadolinium late enhancement magnetic resonance imaging may identify patterns of myocardial fibrosis associated with a high risk of SCD both in ischemic and nonischemic cardiomyopathy.
Elderly versus transplant referral populations
The incidence of HF rises exponentially with age, and the average age of patients admitted with HF is > 75 years in most contemporary epidemiological studies. However, most of the literature describing the prognosis of outpatients has been derived from younger and middle-aged white male patients being evaluated for advanced HF therapies, including clinical trials or cardiac transplantation. The demographic profile of HF in elderly patients is distinct, with a greater percentage of female patients, a higher proportion of HFpEF, and more comorbidities. Recent studies have to some extent addressed this, as seen in studies like Get With the Guidelines (GWTG) study now enrolling > 300,000 hospitalized HF patients with a median age of 75 years. However, most studies that include a high proportion of elderly patients fail to provide information about prognostic indicators beyond simple clinical and laboratory values. The use of modalities such as advanced imaging and invasive hemodynamics in elderly individuals and the predictive power of these are still poorly described.
Renal dysfunction
Renal dysfunction is extremely common in HF. Among patients hospitalized for decompensated HF, more than 60% have moderate or severely reduced estimated GFR (eGFR) (< 60 mL/min) irrespective of cardiac systolic function. Serum creatinine and BUN have been identified in numerous studies as powerful prognostic factors in patients with acute and chronic HF. Hillege and coworkers showed that reduced eGFR was a continuous risk factor and that patients with an eGFR < 44 mL/min had twice the long-term mortality of those with normal renal function.
Similarly, development of worsening renal function during hospitalization for HF is associated with poor clinical outcomes. However, recent studies have clearly shown that eGFR cannot be viewed in isolation in HF, since changes induced by angiotensin-converting enzyme inhibitors or mineralocorticoid receptor antagonists (MRAs) are not associated with poor outcomes and since a decrease in eGFR during an admission for HF is detrimental only if accompanied by lack of decongestion. Hence, when using eGFR, creatinine, and BUN to evaluate prognosis in HF, they should be interpreted in a clinical context, incorporating medical therapy as well as the congestive state of the patient.
Biomarkers
As discussed earlier, no ideal biomarker has been described yet, but of all the prognostic biomarkers currently identified, BNP comes closest to that ideal parameter. This measurement entails an easily obtainable, rapidly processed, inexpensive blood test that has excellent sensitivity and good specificity for HF. BNP has very significant negative predictive accuracy for the diagnosis of HF. Another valuable biomarker in acute and chronic HF is serum troponin, both in patients with acute coronary syndrome and in patients with nonischemic ADHF.
Natriuretic Peptides
Natriuretic peptides have become a cornerstone in the diagnosis and management of HF. The vast majority of studies have used BNP or NT-proBNP, but atrial peptides, such as mid-regional pro atrial natriuretic peptide (MR pro-ANP), have also been demonstrated as a reliable biomarker in HF. It is currently a class I recommendation to use natriuretic peptides both for diagnosis and prognostication in HF. A low level of natriuretic peptide effectively rules out HF, whereas elevated levels do not necessarily indicate HF, as peptide specificity is modest, at best, in outpatients or patients presenting with acute dyspnea. Elevated levels of natriuretic peptides are associated with worse outcomes both in patients with mild HF and patients with advanced HF. However, while not necessarily linear, the risk function is continuous. As such, BNP values must be interpreted in a clinical context with the other markers mentioned here. Furthermore, it should be recognized that female gender and low body weight are associated with higher levels of natriuretic peptides irrespective of cardiac function. Also, cardiac (such as atrial fibrillation) and noncardiac conditions (such as renal dysfunction) increase BNP/NT-proBNP and changes should be evaluated in this context. Finally, it should be emphasized that neprilysin inhibitors, such as sacubitril/valsartan, specifically increase BNP, but not proBNP. Consequently, to use natriuretic peptides for monitoring of patients treated with neprilysin inhibitors, NT-proBNP must be applied.
Changes in natriuretic peptides carry even greater prognostic information, and hence, these markers are useful for longitudinal follow up. This finding has led to studies testing the hypothesis that BNP-driven pharmacological strategy would be superior to one guided by clinical judgment. However, so far, outcomes have not been improved convincingly with BNP-guided treatment strategies in HF.
Metabolic and Inflammatory Markers
Metabolic markers such as serum cholesterol, uric acid, and hemoglobin have been variably reported to be predictive of outcome in HF patients. An inverse relationship between serum cholesterol and survival has been described in patients with HF, likely reflecting malnutrition and cachexia as poor prognostic factors. Insulin resistance is commonly observed in patients with HF and is associated with worse prognosis. Hyperuricemia is another metabolic abnormality associated with an adverse prognosis.
Elevations in bilirubin and international normalized ratio (INR) are believed to reflect right HF and are important markers of outcome, especially in patients treated with advanced therapies such as left ventricular assist devices (LVADs).
Subclinical inflammation is detectable in many patients with HF. Several biomarkers that have been identified as indicators of poor prognosis may simply reflect the degree of systemic or local inflammation (i.e., C-reactive protein, tumor necrosis factor, and interleukin-6, ST-2).
Physical capacity and mortality risk in heart failure
Six-Minute Walk Test
Inability to exercise is the cardinal symptom of chronic HF. Functional capacity is assessed in chronic HF by NYHA criteria, but even if studies show that, indeed, patients with NYHA class II HF have higher exercise capacity than do patients with NYHA class III–IV HF, the assessment of NYHA class remains highly subjective. The 6-minute walk test (i.e., the distance walked over a period of 6 minutes) is less subjective than NYHA functional class but it can be heavily influenced by the motivation of the patient, tester, or both. Additionally, it is important to recognize that the 6-minute walk test is close to a maximal exercise test in some patients with advanced HF, whereas in patients with NYHA class II HF, it is a test of submaximal exercise tolerance. However, the 6-minute walk test has several advantages: it is reproducible, easy to perform, and inexpensive.
Importantly, the 6-minute walk test provides prognostic information as demonstrated by the SOLVD (Study of Left Ventricular Dysfunction) investigators, who showed in a substudy of 898 HF patients in their registry that mortality risk was 3.7 times higher in patients with a 6-minute walk distance less than 350 m compared with patients who walked more than 450 m. Similarly, the risk of HF hospitalization was also higher in patients with reduced walk distance. Subsequent investigators have shown prognostic value of the 6-minute walk test in some cohorts, while others have not.
Maximal Oxygen Uptake and Prognosis in Heart Failure
Determination of peak oxygen uptake during a maximal symptom limited treadmill or bicycle exercise test is the most objective method to assess maximal functional capacity in patients with chronic HF and has been shown to be an excellent indicator of prognosis.
Peak VO 2 can be calculated from the Fick principle: peak VO 2 is the product of peak cardiac output and maximal arteriovenous oxygen difference. Because the maximal arteriovenous difference is similar among most sedentary individuals, peak VO 2 provides an indirect assessment of cardiac output reserve, and this largely underlies the effectiveness of peak VO 2 in risk stratification in HF. Several peripheral factors may also affect peak VO 2 , such as the metabolic activity of skeletal muscle mass, endothelial function, and demographics such as age and gender. As HF severity increases, skeletal muscle mass and its metabolic activity both decrease, and endothelial function is progressively impaired, and this may add to the prognostic utility of peak VO 2 .
To contribute with valid prognostic information, measurements of peak VO 2 must be accurate and it must be ensured that the value measured is representative of the patient’s maximal exercise effort. Patients with comorbidity, such as arthritis, may terminate the test before achieving their peak VO 2 . Several methods exist to ensure maximal effort such as a plateau in the VO 2 to power (watt) relationship, greatly elevated plasma lactate (> 18 mM), or, most commonly used, an elevated respiratory exchange ratio (RER; the ratio between expired CO 2 and consumed O 2 ) ( Fig. 3.4 ). An RER >1.05 is generally accepted as a measure of sufficient exercise to validate the peak VO 2 results on a cardiopulmonary exercise test.
