B-type natriuretic peptide (BNP), peak oxygen consumption (VO 2 ), blood urea nitrogen (BUN), systolic blood pressure (SBP), and pulmonary capillary wedge pressure are all established predictors of mortality or urgent transplantation in patients with advanced heart failure (HF). However, their comparative predictive ability in estimating prognosis has not been well studied. We analyzed 1,215 patients with advanced systolic HF referred to a university center from 1999 to 2009. BUN, BNP, VO 2 , SBP, and pulmonary capillary wedge pressure were measured as a part of the initial evaluation. The patients were divided into groups according to the best cutoffs for predicting both 1- and 2-year mortality from the analysis of the receiver operating characteristic curves (BNP ≥579 pg/ml, peak VO 2 <14 ml/kg/min, BUN ≥53 mg/dl, SBP <118 mm Hg, and pulmonary capillary wedge pressure ≥21 mm Hg). During a 2-year follow-up, 234 patients (19%) died, and 208 (17%) required urgent transplantation. BNP (odds ratio 4.3, 95% confidence interval 3.3 to 5.5) and peak VO 2 (odds ratio 4.5, 95% confidence interval 2.6 to 7.8) were the strongest predictors for death or urgent transplantation. On multivariate analyses, BNP and peak VO 2 were the strongest predictors for both death or urgent transplantation and all-cause mortality. The c-statistic was 0.756 for BNP, 0.701 for VO 2 , 0.659 for BUN, 0.638 for SBP, and 0.650 for pulmonary capillary wedge pressure. In conclusion, of the 5 established predictors of outcomes in advanced HF, BNP was the most robust discriminator of risk and thus could be useful, along with other more traditional prognostic variables, in patient counseling regarding prognosis and determining the timing for heart transplantation.
Numerous variables have been identified that yield prognostic information in advanced heart failure (HF) and, therefore, help guide the aggressiveness of therapy, counseling of patients and their families, and listing for heart transplantation. In 1990, Stevenson et al identified the pulmonary capillary wedge pressure as an independent predictor of survival during vasodilator therapy tailored for hemodynamic goals. Soon after, Mancini et al demonstrated the utility of analyzing measures from cardiopulmonary exercise testing in determining the listing for heart transplantation. Specifically, peak oxygen consumption (VO 2 ) is now routinely used in the determination for candidacy for transplantation. More recent studies have shown that systolic blood pressure (SBP) and blood urea nitrogen (BUN) provide prognostic information in decompensated HF and, although less well-established, in chronic HF. In the past decade, B-type natriuretic peptide (BNP) has emerged as a major predictor of death and cardiac events in all phases of the disease. However, it remains unclear which of these 5 variables has the strongest prognostic power. The present study sought to compare the predictive power of BNP, VO 2 , BUN, SBP, and pulmonary capillary wedge pressure in patients with advanced HF receiving optimal medical therapy.
Methods
The initial study population consisted of 3,239 patients referred to a HF center at a university hospital for HF management and/or transplant evaluation from 1983 to 2009. Of the 3,239 patients, 345 with a left ventricular ejection fraction ≥40% were excluded. Because BNP, as a laboratory assay, emerged in clinical practice around 1999 to 2000, the 1,679 subjects with referral dates before 1999 were excluded. The final study population consisted of 1,215 patients. All patients were followed up at a comprehensive HF management program, as previously described.
Detailed information on the patients’ baseline characteristics and medications were recorded at their initial visit and each follow-up visit. BNP and BUN were measured at the initial referral visit, and VO 2 , SBP, and pulmonary capillary wedge pressure were measured after optimization of hemodynamics using right heart catheterization. Repeated measurements were not used in the analyses. VO 2 was measured (MedGraphics CardiO 2 , Medical Graphics Corporation, St. Paul, Minnesota) during cycle ergometry, with a 6-minute warm-up period followed by a brief rest, 2 minutes of unloading cycling, and a 15-W ramp to limiting symptoms, as previously described. The left ventricular ejection fraction was determined by echocardiography at the referral visit.
The primary end point of the study was a composite of death and urgent transplantation. Deaths were confirmed from the Social Security Death Index. Urgent transplantation was included in the primary end point because these patients, who were by definition status 1A, were expected to live <1 week without transplantation and were dependent on intravenous medications, a ventricular assist device, or mechanical ventilation. All transplantations were coded as nonfatal end of follow-up at the transplantation. Ventricular assist device implantation was either destination therapy or a bridge to transplantation and was not censored as an end point. Patients lost to follow-up were censored at the time they were last known to be alive and well.
The patients were divided into groups according to the best cutoffs for predicting 2-year death or urgent transplantation from analysis of the receiver operating characteristic curves (BNP ≥579 pg/ml, VO 2 <14 ml/kg/min, BUN ≥53 mg/dl, SBP <118 mm Hg, and pulmonary capillary wedge pressure ≥21 mm Hg). The results are presented as the mean ± SD for continuous variables and as percentages of the total for categorical variables. The independent samples t test and chi-square test were used for comparison of the continuous and categorical variables, respectively. Survival curves were calculated using the Kaplan-Meier method. Univariate and multivariate Cox regression analyses were used to calculate the estimated 2-year odds ratio for death or urgent transplantation and all-cause mortality with the 95% confidence interval. The variables entered into the multivariate model included BNP, BUN, SBP, VO 2 , pulmonary capillary wedge pressure, age, gender, HF etiology, left ventricular ejection fraction, New York Heart Association class, creatinine, and hypertension. Binary logistic regression analysis was used to determine the c-statistic for all 5 variables. The Statistical Package for Social Sciences for Windows, version 16.0 (SPSS, Chicago, Illinois) was used for all analyses.
This research was supported by the Ahmanson Foundation (Los Angeles, California). The authors are solely responsible for the design and conduct of the present study, all study analyses, and the drafting and editing of the report.
Results
The baseline characteristics of the population are listed in Table 1 . The patient age range was 17 to 85 years. The HF etiologies included ischemic (43%), idiopathic (21%), valvular, toxin-induced, peripartum, and hypertrophic. The mean and median length of follow-up was 868 ± 894 days and 507 days (range 119 to 1484), respectively.
| Variable | All Patients (n = 1,215) | Event-Free Survival (n = 773) | Death or Urgent Transplantation (n = 442) | p Value |
|---|---|---|---|---|
| Age (years) | 53 ± 13 | 53 ± 13 | 54 ± 14 | 0.181 |
| Men | 75% | 74% | 75% | 0.527 |
| Race | ||||
| White | 59% | 60% | 56% | 0.006 |
| Black | 10% | 13% | 8% | 0.043 |
| New York Heart Association class | 0.0001 | |||
| III | 43% | 51% | 32% | |
| IV | 32% | 17% | 62% | |
| Ischemic etiology | 43% | 44% | 46% | 0.021 |
| Diabetes mellitus | 29% | 28% | 30% | 0.652 |
| Hypertension | 39% | 41% | 35% | 0.082 |
| Smoker | 52% | 54% | 47% | 0.063 |
| Left ventricular ejection fraction (%) | 23 ± 7 | 23 ± 7 | 22 ± 7 | 0.003 |
| Left ventricular end-diastolic volume (mm) | 67 ± 11 | 67 ± 10 | 67 ± 12 | 0.765 |
| Severe mitral regurgitation | 16% | 14% | 21% | 0.013 |
| Peak oxygen consumption (ml/kg/min) | 13.1 ± 4.8 | 13.6 ± 4.6 | 11.1 ± 5.0 | 0.0001 |
| Implantable cardioverter-defibrillator | 53% | 57% | 51% | 0.049 |
| Ventricular assist device | 5.8% | 1.4% | 15.3% | 0.0001 |
| Median B-type natriuretic peptide (pg/ml) | 575 (190–1,300) | 354 (111–826) | 1,140 (581–1,900) | 0.0001 |
| Sodium (mmol/L) | 136 ± 5 | 137 ± 4 | 134 ± 5 | 0.0001 |
| Blood urea nitrogen (mg/dl) | 29 ± 20 | 25 ± 16 | 35 ± 24 | 0.0001 |
| Creatinine (mg/dl) | 1.5 ± 1.4 | 1.5 ± 1.6 | 1.6 ± 0.8 | 0.166 |
| Hemoglobin (g/dl) | 13.1 ± 2.0 | 13.4 ± 1.8 | 12.5 ± 2.1 | 0.0001 |
| Total cholesterol (mg/dl) | 159 ± 54 | 172 ± 52 | 138 ± 50 | 0.0001 |
| Hemodynamics ⁎ | ||||
| Systolic blood pressure (mm Hg) | 108 ± 19 | 111 ± 19 | 102 ± 17 | 0.0001 |
| Pulmonary capillary wedge pressure (mm Hg) | 17.3 ± 7.3 | 15.7 ± 6.4 | 19.5 ± 7.9 | 0.0001 |
| Mean right atrial pressure (mm Hg) | 8.4 ± 5.0 | 7.4 ± 4.0 | 9.5 ± 5.7 | 0.0001 |
| Cardiac output (L/min) | 4.4 ± 1.2 | 4.6 ± 1.1 | 4.2 ± 1.2 | 0.0001 |
| Cardiac index (L/min/m 2 ) | 2.6 ± 1.2 | 2.7 ± 0.9 | 2.5 ± 1.5 | 0.134 |
⁎ Hemodynamics recorded after pulmonary artery catheter-guided therapy.
During 2 years of follow-up, 234 patients (19%) died and 208 (17%) required urgent (status IA) transplantation. A total of 137 patients (11%) underwent nonurgent (status IB or II) transplantation. Of the 345 transplantations, 208 (60%) were status 1A. Progressive HF accounted for 114 deaths, and 40 deaths were considered sudden death, 6 were secondary to myocardial infarction, and 74 deaths were other or unknown. Two patients were lost to follow-up.
The mean VO 2 and SBP were significantly lower in the patients who died or had undergone urgent transplantation. In contrast, the levels of mean pulmonary capillary wedge pressure and BUN were significantly greater ( Table 1 ). The median and interquartile range for BNP was greater in the death or urgent transplantation group ( Table 1 ).
On univariate analyses, higher BNP and lower VO 2 were the strongest predictors of death or urgent transplantation and all-cause mortality ( Table 2 ). On multivariate analyses, BNP and VO 2 remained the strongest predictors of death or urgent transplantation and all-cause mortality ( Table 2 ). Neither higher BUN nor higher pulmonary capillary wedge pressure was a significant predictor of death or urgent transplantation on multivariate analyses. However, a lower SBP was a significant predictor of death or urgent transplantation, but not all-cause mortality.
| Variable | All-Cause Mortality (Events/Total Patients) | Multivariate Odds Ratio for All-Cause Mortality (95% CI) | Death or Urgent Heart Transplantation (Events/Total Patients) | Multivariate Odds Ratio for Death or Urgent Heart Transplantation (95% CI) |
|---|---|---|---|---|
| B-type natriuretic peptide (pg/ml) | ||||
| <579 | 7% (32/464) | — | 18% (84/464) | — |
| ≥579 | 27% (123/454) | 4.4 (1.5–13.0) | 56% (255/454) | 4.4 (2.3–8.5) |
| p Value | 0.0001 | 0.007 | 0.0001 | 0.0001 |
| Peak oxygen consumption (ml/kg/min) | ||||
| ≥14 | 3% (6/220) | — | 7% (15/220) | — |
| ≤14 | 13% (52/405) | 9.2 (1.1–76.9) | 26% (104/405) | 3.9 (1.6–10.0) |
| p Value | 0.0001 | 0.038 | 0.0001 | 0.004 |
| Blood urea nitrogen (mg/dl) | ||||
| <53 | 17% (169/997) | — | 35% (349/997) | — |
| ≥53 | 40% (48/121) | 1.7 (0.4–7.2) | 60% (72/121) | 2.9 (0.7–12.3) |
| p Value | 0.0001 | 0.451 | 0.0001 | 0.148 |
| Systolic blood pressure (mm Hg) | ||||
| ≥118 | 13% (37/277) | — | 24% (66/277) | — |
| <118 | 22% (175/810) | 3.3 (0.7–16.4) | 43% (349/810) | 2.8 (1.1–7.1) |
| p Value | 0.003 | 0.136 | 0.0001 | 0.026 |
| Pulmonary capillary wedge pressure (mm Hg) | ||||
| <21 | 19% (112/605) | — | 37% (223/605) | — |
| ≥21 | 23% (48/210) | 2.6 (1.1–3.0) | 60% (126/210) | 1.4 (0.7–2.8) |
| p Value | 0.172 | 0.0001 | 0.0001 | 0.289 |
The c-statistic for both BNP and VO 2 was >0.7 (BNP, c-statistic 0.756, VO 2 , c-statistic 0.701). SBP, BUN, and pulmonary capillary wedge pressure all provided less discrimination, with a c-statistic <0.7 ( Figure 1 ). Kaplan-Meier analyses revealed that the 2-year survival rate for those with BNP <579 pg/ml was 82% compared to 44% for those with BNP of ≥579 pg/ml. The 2-year survival rate for those with peak VO 2 of ≥14 ml/kg/min was 93% compared to 74% for those with VO 2 <14 ( Figure 2 ). When all 5 variables were combined into a single binary logistic regression model, the c-statistic was 0.748 for 1-year all-cause mortality, 0.771 for 1-year death or urgent transplantation, 0.734 for 2-year all-cause mortality, and 0.767 for 2-year death or urgent transplantation. The sensitivity, specificity, accuracy, positive predictive value, and negative predictive value were calculated for all variables at their receiver operating characteristic cutoffs ( Table 3 ).
