Heart failure (HF) is a leading cause of morbidity and mortality in Western society and a growing public health problem. Almost 30 million people all over the world have HF, with 5 million in the United States, and an incidence of 550,000 new cases each year.1 It is likely that the number of cases will rise in the coming years due to better survival of ischemic heart disease and expansion of the elderly population.2 Despite advances in management and treatment, survival with HF continues to be poor and the mortality rate is comparable to that of many forms of cancer.3–5 The detection of patients at high risk for death is a major challenge in the evaluation and management process. Several clinical, functional, and laboratory variables have been recognized to be predictive of mortality.6–17 We compared the prognostic value of 23 classical clinical Doppler echocardiography and cardiopulmonary exercise (CPX) indexes in a stable, moderately symptomatic, systolic HF outpatient population receiving optimal medical therapy.
©2012 Wiley Periodicals, Inc.
Heart failure (HF) is a leading cause of morbidity and mortality. The detection of patients at high risk for death is a major challenge in HF management. The authors compared the prognostic value of 23 clinical Doppler echocardiography and cardiopulmonary exercise indexes in a stable, moderately symptomatic, systolic HF outpatient population receiving optimal medical therapy. The end point was the incidence of overall mortality. Between January 2002 and December 2008, a total of 146 patients with left ventricular (LV) ejection fraction 0.31±0.8 and New York Heart Association functional class II or III were enrolled. The prognostic power of single variables was assessed using chi-square test for categoric variables and t test for continuous variables. Variables associated with the prespecified end point were included as predictors in a binary logistic regression multivariate model. At multivariate analysis, “restrictive” LV filling pattern (P=.004), ischemic etiology (P=.022), pulmonary artery systolic pressure (PASP) ≥50 mm Hg (P=.027), and peak oxygen uptake (VO2) <15.9 mL/kg/min (P=.046) resulted independent predictors of the outcome. A simple risk score was then obtained using these significant independent variables, excluding peak VO2 because of only borderline significance. Patients with ischemic etiology, restrictive LV filling pattern, and PASP ≥50 mm Hg have a very high risk of death (odds ratio, 33.77; 95% confidence interval, 5.74–198.8; P<.001, compared with patients with no risk factors). In this high-risk group, evaluation of peak VO2 could be superfluous. A very simple clinical echocardiographic model based on etiology-LV filling and pulmonary pressure is a powerful tool for risk stratification of systolic HF in ambulatory patients.
Consecutive HF outpatients who underwent their first ambulatory visit between January 2002 and December 2008 were recruited. Inclusion criteria were chronic symptomatic HF and New York Heart Association functional class II or III, left ventricular (LV) ejection fraction ≤0.40, and clinical stability while taking optimized medical therapy for at least 3 months. Patients with angina or silent ischemia, myocardial infarction or heart surgery during the past 6 months, HF due to primitive valvular disease, anemia (hemoglobin <11 g/dL), neoplasia, and contraindication to exercise testing for cardiac or noncardiac reasons were excluded. Enrolled patients performed clinical examination, electrocardiography, Doppler echocardiographic study, and CPX. The etiology of HF was defined as ischemic in patients with previous myocardial infarction and/or surgical or percutaneous revascularization and/or previous coronary angiography showing at least one critical coronary stenosis (>50% luminal narrowing).
The Hewlett-Packard Sonos 5500 echograph (Philips, Amsterdam, The Netherlands) was used. Ejection fraction was calculated by means of the modified Simpson’s rule.18 LV mass was calculated using the American Society of Echocardiography19 formula. LV filling was obtained by pulsed wave Doppler mitral velocity profile from the 4-chamber apical view by positioning a sample volume between the tips of the mitral valve leaflets in diastole. LV filling pattern was classified as “restrictive” (E/A ratio >2 or from 1 to 2 with E-wave deceleration time ≤140 ms) or “nonrestrictive” in the other cases. In case of atrial fibrillation, only E-wave deceleration time was used to discriminate the restrictive pattern. Five beats were measured and the mean value was utilized. Mitral regurgitation was evaluated from the apical view and graded as mild, moderate, or severe on the basis of the area covered by the regurgitant signals visualized with color Doppler flow.20 Pulmonary artery systolic pressure (PASP) was estimated from the 4-chamber apical or parasternal short-axis view, using continuous-wave Doppler between the tricuspid valve leaflets, applying the modified Bernoulli equation and adding right atrial pressure to systolic gradient across the tricuspid valve.21 Right atrial pressure was assumed to vary from 5 mm Hg to 20 mm Hg and was clinically rated from the height of jugular vein with the patient lying at a 45° angle.22 Doppler-derived PASP was categorized as <50 mm Hg or ≥50 mm Hg.20
Cardiopulmonary Exercise Testing
All tests were performed in the morning immediately after echocardiographic examination using a bicycle ergometer with the physician unaware of the result of the previous echocardiogram. One to 7 days before CPX, all patients performed a preliminary test in order to become familiar with it. After a 1-minute warm-up period at 0 W, a 10-W/minute incremental protocol was started. Ventilation (VE), oxygen uptake (VO2), and carbon dioxide production (VCO2) were monitored online (OXYCONDELTA; Jaeger, Wurzburg, Germany). Calibration with reference gases was performed immediately before each test. A standard 12-lead electrocardiogram was continuously recorded. Ventilatory anaerobic threshold was determined by the ventilatory equivalent for VO2 or by the V slope method.23 Peak VO2 was defined as the highest O2 consumption obtained during the test and peak VE/VCO2 was defined as the value of the ventilatory equivalent for VCO2 at the time of the peak VO2. Resting heart rate was defined as the lowest recorded in the upright position before exercising and peak heart rate as the highest obtained during exercise. The heart rate response to exercise was assessed by the chronotropic index calculated by the formula suggested by Robbins and colleagues11: peak heart rate-rest heart rate/220-age-rest heart rate ×100. Patients were encouraged to exercise until limiting dyspnea or fatigue.
Follow-up was obtained through clinic visit or phone call after 1 year and then every 6 months. The end point was mortality for any cause. In all cases, the events were confirmed by review of clinical record and/or death certificate. Minimum follow-up lasted 23 days and the longest was 3740 days.
Numeric values were expressed as mean±standard deviations. Prognostic power of single variables was assessed using the chi-square test for categoric variables and the t test for continuous variables. All variables that were significantly associated with the prespecified end point were included in a binary logistic regression multivariate model. Cumulative lifetime survival rates were generated with the Kaplan-Meier method and compared using the log-rank test. Hazard ratios (HRs) with 95% confidence intervals (CI) were estimated. A P value <.05 was considered significant. All calculations were performed with SAS software release 8.2 for Microsoft Windows (SAS Institute Inc, Cary, NC).
Twenty patients were unable to perform CPX and were therefore excluded. Thereafter, of the initially screened 166 patients, 146 were evaluated. The characteristics of the patient population are listed in Table I and pharmacologic therapy in Table II. All continuous variables had an approximate normal distribution. No patient was lost during follow-up. After 4.80±2.62 years, there were 43 deaths (29.5%). On univariate analysis, variables significantly associated with mortality were peak VO2, LV filling pattern, pulmonary artery hypertension, heart rate, and blood pressure response to exercise (Table III). Heart rate response to exercise, as expressed by chronotropic index, was significantly lower in patients taking β-blockers (0.54 vs 0.67, P<.05); however, the risk did not differ according to β-blocking status. Of the clinical and therapeutic variables, ischemic etiology, treatment with amiodarone or digoxin and exclusion from therapy with renin-angiotensin system (RAS) inhibitors were significantly related to mortality. Patients with a contraindication to RAS inhibitors (8 of 147 patients) were significantly older (age 74±4 years vs 65±5 years; P<.01) and had more impaired functional capacity (peak VO2 11.6±4 mL/kg/min vs 16.2±3 mL/kg/min; P<.05). On multivariate analysis, only restrictive LV filling pattern (B: −1.30, P=.004; Exp(B): 0.27; 95% CI, 0.11–0.66), ischemic etiology (B: −1.01, P=.022; Exp(B): 0.36; 95% CI, 0.15–0.86), and PASP ≥50 mm Hg (B: −1.11, P=.027; Exp(B): 0.32; 95% CI, 0.12–0.88) maintained an independent predictive value. Peak VO2 <15.9 mL/kg/min showed a weaker association with mortality (B: 0.124, P=.046; Exp(B): 0.88; 95% CI, 0.78–0.99). Using the three most potent independent risk factors (etiology, LV filling pattern, and PASP), a simple risk score was obtained. Peak VO2 was excluded because of only borderline significance. Patients with no risk factors (score 0) were assumed as the reference. Different levels of risk were identified by the presence of 1, 2. or 3 variables (Table IV). Patients with ischemic etiology, restrictive LV filling pattern, and PASP ≥50 mm Hg (score 3) showed an incidence of death of 100% at 72 months (Figure).
|NYHA functional class II or III||105/41|
|Coronary artery disease||81 (55%)|
|Chronic atrial fibrillation||25 (17%)|
|Left bundle branch block||24 (16%)|
|CRT and/or AICD||41 (28%)|
|Left atrial, mm||45±6.6|
|Mitral regurgitation, mean||1.46±0.9|
|“Restrictive” LVFP||50 (34%)|
|PASP ≥50 mm Hg||31 (21%)|
|Peak VO2, mL/kg/min||15.9±4.2|
|Peak SBP ≤130 mm Hg||21 (14%)|
|Peak SBP fall ≥20 mm Hg||7 (5%)|
|Drugs||Patients, No.||Daily Dosage, mg|
|Variable||Source||P Value||HR||95% CI|
|Peak VO2 <15.9 mL/kg/min||Mean||<.001|
|“Restrictive” LVFP||Previous studies||<.001||5.03||2.35–10.80|
|Chronotropic index <0.59||Mean||<.001|
|PASP ≥50 mm Hg||Previous studies||<.001||4.16||1.81–9.55|
|Double product <21,200||Mean||<.001|
|Ischemic etiology||Previous studies||<.01||2.74||1.27–5.92|
|Peak SBP fall ≥20 mm Hg||<.05||6.64||1.24–35.71|
|Peak SBP <130 mm Hg||<.05||3.20||1.24–8.23|
|Therapy with RAS inhibitors||<.05||0.23||0.05–0.98|
|Therapy with amiodarone||<.05||2.61||1.05–6.49|
|Therapy with digoxin||<.05||2.14||1.04–4.41|
|Therapy with β-blockers||NS||0.58||0.28–1.20|
|NYHA functional class II or III||NS||2.10||0.84–5.24|
|Chronic atrial fibrillation||NS||2.19||0.90–5.31|
|Left bundle branch block||NS||0.91||0.42–1.97|
|Left atrial size||NS|
|LV mass index||NS|
|Inclusion in training programs||NS||1.34||0.58–3.09|
|Score||Patients||Odds Ratio||95% Confidence Interval||P Value||Incidence of Death, %|
|0 (no risk factors)||41||–||13|
|1 (1 risk factor)||59||3.94||1.05–14.75||.032||38|
|2 (2 risk factors)||35||13.41||3.48–51.71||<.001||72|
|3 (3 risk factors)||11||33.77||5.74–198.8||<.001||100|
The annual mortality rate for patients with HF can range from 10% to 50%.3–6 Many studies have addressed the question of prognostic stratification of HF and many risk factors for death or acute decompensation have been identified, although most did not have independent prognostic power.24 Moreover, often the real patient is different from those of the study population from which the marker was deduced. The timing of patient evaluation in relation to therapy optimization and physical deconditioning is not consistently defined, nor is the effect of contemporary use of different evidence-based treatments, including β-blockers and cardiac resynchronization therapy-automatic implantable cardioverter-defibrillator devices. Finally, prognostic stratification based on a single risk factor has limited predictive value.25 For these reasons, several models integrating different variables have been proposed.26–30 However, many of these prognostic models cannot be easily utilized in ambulatory patients due to their complexity and the need for additional noninvasive as well as invasive measurements. Our study compared the prognostic power of 23 easy-to-obtain variables for predicting death in stable systolic HF outpatients, moderately symptomatic and treated with the best medical therapy available today. Univariate analysis confirmed previous evidence that peak VO2, LV restrictive filling pattern, PASP ≥50 mm Hg, blood pressure, and heart rate response to exercise are most significantly associated with prognosis. Regarding therapy, patients taking amiodarone or digoxin or with contraindication to RAS inhibitors are identified with a worse prognosis. On multivariate analysis, only ischemic etiology, LV restrictive filling pattern, and PASP ≥50 retained an independent risk predictive power, whereas peak VO2 achieved only borderline statistical significance. We therefore suggest a simple method based on the integration of these three independent risk factors that are extremely simple to use and powerful for stratification of systolic HF prognosis. Patients with ischemic etiology, LV restrictive pattern, and PASP ≥50 mm Hg (score 3) are at extremely high risk for death and should be aggressively treated and closely monitored. In this very high-risk group the measurement of peak VO2 could be superfluous, whereas CPX can attain the most important adjunctive value in patients with a moderately adverse prognosis (risk score 1 and 2). Finally, because HF is a dynamic condition, it is important to stress that prognostic stratification must be performed when the patient is stable and on optimal medical therapy.
The main limitation of this study is a small sample size. Moreover, this etiology-filling pulmonary pressure risk score has not yet been validated in an independent validation group. The score is derived from a stable HF population eligible for exercise testing. Indeed, our results cannot be extended to patients unable to perform CPX for cardiac or noncardiac reasons. Many patients were treated with β-blockers and we could speculate that β-blocking therapy affects blood pressure, heart rate, and exertional ventilation and could change the prognostic value of some exercise variables. Finally, the small number of women involved and the absence of populations different from Caucasians indicates that the results cannot be extended to women and to non-Caucasian races.
Our findings suggest that a very simple clinical echocardiography risk score based on etiology, LV filling pattern, and pulmonary pressure is a powerful tool for risk stratification of systolic HF ambulatory patients. The score does not require additional invasive or noninvasive measurements and, after the completion of a validation study, can be proposed in everyday clinical practice.
Disclosures: No grants and other support received for this article.