Changes in Brain Natriuretic Peptide and Norepinephrine Over Time and Mortality and Morbidity in the Valsartan Heart Failure Trial (Val-HeFT)

Changes in Brain Natriuretic Peptide and Norepinephrine Over Time and Mortality and Morbidity in the Valsartan Heart Failure Trial (Val-HeFT) Inder S. Anand, MD, DPhil; Lloyd D. Fisher, PhD; Yann-Tong Chiang, PhD; Roberto Latini, MD; Serge Masson; Aldo P. Maggioni, MD; Robert D. Glazer, MD; Gianni Tognoni, MD; Jay N. Cohn, MD; for the Val-HeFT Investigators Background Neurohormones are considered markers of heart failure progression. We examined whether changes in brain natriuretic peptide (BNP) and norepinephrine (NE) over time are associated with corresponding changes in mortality and morbidity in the Valsartan Heart Failure Trial. Methods and Results Plasma BNP and NE were measured before randomization and during follow-up in 4300 patients in the Valsartan Heart Failure Trial. The relation between baseline BNP and NE and all-cause mortality and first morbid event (M&M) was analyzed in subgroups, with values above and below the median, and by quartiles. The change and percent change from baseline to 4 and 12 months in BNP and NE were also analyzed by quartiles for subsequent M&M. Risk ratios for M&M were calculated using a Cox proportional hazard model. Risk ratio of M&M for patients with baseline BNP or NE above the median was significantly higher than that for patients with values below the median. Baseline BNP and NE in quartiles also showed a quartile-dependent increase in M&M. BNP had a stronger association with M&M than NE. Patients with the greatest percent decrease in BNP and NE from baseline to 4 and 12 months had the lowest whereas patients with greatest percent increase in BNP and NE had the highest M&M. Conclusions Not only are plasma BNP and NE important predictors of heart failure M&M, but changes in these neurohormones over time are associated with corresponding changes in M&M. These data further reinforce their role as significant surrogate markers in HF and underscore the importance of including their measurement in HF trials. (Circulation. 2003;107:1278-1283.) Key Words: heart failure trials, clinical natriuretic peptides norepinephrine The syndrome of heart failure (HF) is characterized by hemodynamic abnormalities, 1 impaired exercise capacity, 2 neurohormonal activation, 3,4 and a relentless progression with high mortality. 5 Although the mechanisms are not entirely clear, HF progresses through a process of structural remodeling of the heart, to which neurohormonal activation may make an important contribution. 6 Several lines of evidence support the role of neurohormones in the progression of HF. Norepinephrine (NE) 7 and angiotensin II 8 may be directly toxic to cardiac myocytes; the degree of neurohormonal activation in HF is proportional to disease severity, increases with the progression of HF, and is related to prognosis. 9 Furthermore, treatment with drugs such as angiotensin-converting enzyme inhibitors (ACE-I) that decrease HF morbidity and mortality also cause a reduction or attenuation in neurohormones. 10,11 However, there is no evidence to show that changes in neurohormonal activation over time, occurring either spontaneously or in response to pharmacological therapy, are associated with proportional changes in subsequent mortality and morbidity. Such a relationship would further support the hypothesis that changes in neurohormone levels could serve as a useful surrogate for outcomes. 12 See p 1231 In the Valsartan Heart Failure Trial (Val-HeFT), which evaluated the efficacy of valsartan in patients with moderate to severe HF, blood was sampled for brain natriuretic peptide (BNP) and NE assay in all patients at baseline and at 4, 12, and 24 months during follow-up. The neurohormonal database in Val-HeFT is the largest ever collected in a HF trial. 13 In this study, we examined the relationship between baseline BNP and NE and their changes over time, with subsequent mortality and morbidity. Received September 10, 2002; revision received December 5, 2002; accepted December 5, 2002. From the VA Medical Center and University of Minnesota (I.S.A.), Minneapolis, Minn; Department of Biostatistics (L.D.F.), University of Washington, Seattle, Wash; Novartis Pharmaceuticals Corporation (Y.-T.C., R.D.G.), East Hanover, NJ; Istituto Mario Negri (R.L., S.M., G.T.), Milan, Italy; ANMCO Research Center (A.P.M.), Florence, Italy; and University of Minnesota (J.N.C.), Minneapolis, Minn. Drs Latini, Maggioni, and Tognoni have received honoraria for conferences. Correspondence to Inder S. Anand, MD, Director of Heart Failure Program, Veterans Administration Medical Center, One Veterans Drive 111-C, Minneapolis, MN 55417. E-mail anand001@tc.umn.edu 2003 American Heart Association, Inc. Circulation is available at http://www.circulationaha.org DOI: 10.1161/01.CIR.0000054164.99881.00 1278

Anand et al Changes in BNP and NE Over Time and Events 1279 Methods Patients Val-HeFT was a randomized, placebo-controlled, double-blind, parallel-arm multicenter trial. Five thousand ten patients with stable, symptomatic HF who were undergoing prescribed HF therapy and had left ventricular (LV) ejection fraction (EF) 40% and LV internal diameter in diastole adjusted for body surface area (LVIDd/ BSA) of 2.9 cm/m 2 were enrolled in the study. Results of the main trial have been presented in detail previously. 13 Blood Sampling and Hormone Assays Blood samples were collected for measurement of BNP and NE before randomization and 4, 12, and 24 months thereafter. Plasma was separated, and samples were stored at 70 C until assay. Plasma NE was measured with high-performance liquid chromatography (HPLC) with electrochemical detection and BNP with radioimmunoassay by Shionogi (CIS, Tronzano). 14 Figure 1. RR for all-cause mortality and first morbid event in patients with baseline BNP or NE above and below median. BNP exhibited a stronger association for mortality and morbidity than NE (P 0.0001). Data Analysis Baseline BNP and NE as a Prognostic Marker Two analyses were made based on baseline subgroups. For BNP and NE, we first analyzed the relation between baseline values above and below the median (97 pg/ml for BNP and 394 pg/ml for NE) with respect to all-cause mortality and first morbid event (defined as death, sudden death with resuscitation, hospitalization for heart failure, or intravenous inotropic or vasodilator therapy for at least 4 hours). An analysis was also done in subgroups of patients defined by baseline BNP and NE quartiles. All analyses were carried out irrespective of the assigned treatment group. For both all-cause mortality and first morbid event, the hazard risk ratio (RR) and its 95% confidence interval (CI) were calculated relative to the subgroup below the median in the first analysis and relative to the subgroup in the first quartile for the second analysis. In addition, a global test across all quartiles was performed. For the baseline median subgroup analyses, a Cox proportional hazard model was used with NYHA class, LVEF, baseline ACE-I and -blocker category, etiology, and age as covariates. For the baseline quartile subgroup analyses, a Cox model without covariates was used. In an additional analysis, the relative predictive value of baseline BNP was compared with that of baseline NE for mortality and morbidity by comparing the estimated relative risks based on standardized BNP (ie, BNP/SD BNP ) and standardized NE (ie, NE/SD NE ) in a Cox model including both standardized covariates. Changes in BNP and NE Over Time and Subsequent Events For this analysis, change from baseline and the percent change in BNP and NE from baseline to 4 months and 12 months were analyzed by quartiles for subsequent mortality and first morbid TABLE 1. Association Between Baseline BNP and NE in Quartiles and All-Cause Mortality and First Morbid Event (All Randomized Patients) Groups (1 vs 2) Group 1 n/n (%) Group 2 n/n (%) RR 95% CI. Cox P Baseline BNP Quartile 2 vs 1 152/1065 (14.3) 104/1069 (9.7) 1.474 1.149 to 1.892 0.0023 Quartile 3 vs 1 226/1090 (20.7) 104/1069 (9.7) 2.266 1.796 to 2.859 0.0001 Quartile 4 vs 1 350/1081 (32.4) 104/1069 (9.7) 3.952 3.175 to 4.920 0.0001 First morbid event Quartile 2 vs 1 244/1065 (22.9) 167/1069 (15.6) 1.495 1.228 to 1.821 0.0001 Quartile 3 vs 1 372/1090 (34.1) 167/1069 (15.6) 2.457 2.047 to 2.949 0.0001 Quartile 4 vs 1 526/1081 (48.7) 167/1069 (15.6) 4.086 3.432 to 4.864 0.0001 Baseline NE Quartile 2 vs 1 179/1083 (16.5) 147/1067 (13.8) 1.259 1.012 to 1.566 0.0384 Quartile 3 vs 1 247/1075 (23.0) 147/1067 (13.8) 1.890 1.541 to 2.319 0.0001 Quartile 4 vs 1 260/1076 (24.2) 147/1067 (13.8) 2.044 1.669 to 2.503 0.0001 First morbid event Quartile 2 vs 1 305/1083 (28.2) 243/1067 (22.8) 1.310 1.107 1.551 0.0017 Quartile 3 vs 1 352/1075 (32.7) 243/1067 (22.8) 1.629 1.383 1.919 0.0001 Quartile 4 vs 1 411/1076 (38.2) 243/1067 (22.8) 2.065 1.761 2.420 0.0001 Cox regression without covariate adjustments. Events occurring after permanent treatment discontinuation were not censored. Morbid events include death, sudden death with resuscitation, therapeutic IV HF therapy for 4 hours, or hospitalization for HF.

1280 Circulation March 11, 2003 Figure 2. Kaplan-Meier curves for all-cause mortality and first morbid event in subgroups by quartiles for BNP and NE. event. For mortality and first morbid event, the hazard RR and its 95% CI were calculated for quartiles 2 to 4 compared with quartile 1 using a Cox proportional hazard model with the baseline value as a covariate. Results The mean SD for plasma BNP at baseline (n 4305) was 181 230 pg/ml (median, 97 pg/ml) and for plasma NE at baseline (n 4301) was 464 323 pg/ml (median, 394 pg/ml). The mean SD plasma BNP was 142 167 pg/ml in NYHA class I/II patients compared with 244 296 pg/ml in class III/IV patients. The corresponding values for plasma NE were 425 250 and 522 409 pg/ml, respectively. Baseline values for both BNP and NE were available in 4284 patients. Baseline BNP and NE and Events The incidence of all-cause mortality and first morbid events was significantly higher for patients with baseline BNP or NE above the median than for those with values below the median. The RRs for these events (mortality and first morbid event) for values above versus below the median were higher for BNP [RR (95% CI), 2.1 (1.79 to 2.42) and 2.2 (1.98 to 2.52), respectively, for mortality and morbidity] than for NE [1.5 (1.28 to 1.71) and 1.4 (1.23 to 1.54), respectively] (Figure 1). BNP exhibited a relatively stronger association with mortality and morbidity than NE (P 0.0001). Baseline BNP and NE in quartiles also showed a significant quartile-dependent increase in mortality and first morbid event (Table 1, Figure 2). The baseline values for BNP in quartiles were 41, 41 to 97, 97 to 238, and 238 pg/ml and for NE were 274, 274 to 394, 394 to 572, and 572 pg/ml. Similar findings were observed for BNP and NE values at 4 months and 12 months and subsequent events (data not shown). Changes in BNP and NE Over Time and Subsequent Events The relationship between change in neurohormone level and outcome can be examined either as absolute change over time or as percent change. When absolute change was examined in quartiles, no direct relation to mortality was observed (Figure 3). Patients who had the greatest decrease ( 51 pg/ml for BNP and 100 pg/ml for NE at 4 months) or greatest increase ( 19 pg/ml for BNP and 141 pg/ml for NE at 4 months) had the highest mortality risk (Figure 3). In contrast, patients with either no change or a modest increase or decrease in BNP and NE over the 4 months had the lowest risk of mortality. Similar findings were observed for 12-month data and for the first morbid event (data not shown). As noted in Figure 3, the baseline values of neurohormones in these 4 quartiles were strikingly different, being highest in the quartiles with the highest mortality. Therefore, percent change from baseline should provide a more meaningful assessment of change over time. The quartiles based on percent change exhibited similar baseline values (Figure 4). That analysis (Figure 4) reveals a direct relationship between percent change in BNP and NE and adverse outcomes. Quartiles showing the greatest percent increase in BNP and NE from baseline to 4 months exhibited the highest mortality, whereas those with the greatest decline exhibited the lowest mortality. The RRs for mortality and first morbid event increased significantly between quartile 1 and quartile 4 (Table 2 and Figure 5). Similar findings were seen for NE, although the RRs for both events were lower than those for BNP (Table 2 and Figure 5). Effect of Valsartan on Changes in BNP and NE Over Time and Subsequent Events A report on effects of valsartan on BNP and NE in Val-HeFT has been published. 14 Both BNP and NE increased progressively over the duration of the study in the placebo group. Valsartan caused a sustained reduction in BNP and attenuated the increase in NE over the course of the study. The difference in the change from baseline in BNP and NE between the 2 groups was highly significant at each time

Anand et al Changes in BNP and NE Over Time and Events 1281 Figure 3. Change in BNP (pg/ml) and NE (pg/ml) from baseline to 4 months in quartiles and total mortality. The mean baseline BNP and NE, the mean change from baseline, and the mean percent change from baseline to 4 months for each quartile are also shown. point. At last observation, the mean SEM BNP increased by 23 5 pg/ml in the placebo group and decreased by 21 5 pg/ml in the valsartan group; NE increased by 41 6 pg/ml and 12 6 pg/ml in placebo and valsartan groups, respectively. This effect of valsartan on BNP and NE might have contributed to the observed 13.2% reduction in morbidity in the overall population of Val-HeFT. When the association between changes over time in BNP/NE and mortality/morbidity was assessed separately in the 2 randomized groups, a similar relationship was observed. It indicated that the association holds irrespective of treatment groups. Discussion Baseline NE and BNP are important and independent prognostic markers in patients with HF. 3,15 BNP has also been shown to predict mortality and morbidity in other cardiovascular conditions, such as acute coronary syndromes and acute myocardial infarction. 15 17 The present data represent the largest study to confirm the prognostic importance of both these neurohormones in patients with moderate to severe HF. Furthermore, the findings also demonstrate that BNP is a more sensitive predictor than NE of morbidity and mortality in HF. More importantly, the data provide evidence for the first time that changes in these neurohormones over time are associated with corresponding changes in subsequent mortality and morbidity. Because neurohormone levels increase with the progression of HF and are strikingly correlated with outcomes, these data suggest that BNP and NE levels could Figure 4. Percent change in BNP and NE from baseline to 4 months in quartiles and total mortality. The mean baseline BNP and NE, the mean change from baseline, and the mean percent change from baseline to 4 months for each quartile are also shown. serve as guide to the severity of heart failure and the efficacy of its treatment. Because neurohormones have not been measured sequentially in most earlier HF clinical trials, the association of changes in NH over time with prognosis has not been described previously. In the BEST study, NE was measured at baseline and at several time points during the trial. A preliminary report from that study shows that compared with the group of patients with no change or a small change in NE over time, the groups with a large increase or large decrease in NE had a higher mortality and morbidity, 18 similar to the observations in our patient subgroups. The discrepancy between grouping by change and percent change in levels of BNP and NE is understandable. Circulating hormone levels would be expected to change as a percentage of baseline levels after a change in the hormone secretory rate or sympathetic nerve-firing rate. High initial levels reflecting activation of the systems would be expected to exhibit greater absolute declines than low initial levels, although the percent change in these two groups might be comparable. Indeed, quartiles with the greatest absolute change (rise or fall) exhibited higher baseline levels that render them at higher risk for morbidity and mortality regardless of the change in levels over time (Figure 3). When BNP and NE were related over time to outcome based on quartiles of percent change of neurohormones from baseline, a consistent and progressive increase in outcome risk was observed with increasing percent changes in BNP and NE. Patients with the greatest

1282 Circulation March 11, 2003 TABLE 2. Association Between Percent Change in BNP and NE From Baseline to 4 Months in Quartiles and Subsequent All-Cause Mortality and First Morbid Event, Adjusting for Baseline BNP or NE as a Covariate (All Patients Who Did Not Die or Did Not Have a First Morbid Event Before Month 4) Comparisons (1 vs 2) % Change in BNP from baseline to 4 months, Group 1, n/n (%) Group 2 n/n (%) RR 95% CI Cox P Quartile 2 vs 1 145/935 (15.5) 127/935 (13.6) 1.303 1.026 to 1.656 0.0302 Quartile 3 vs 1 141/935 (15.1) 127/935 (13.6) 1.359 1.065 to 1.735 0.0136 Quartile 4 vs 1 179/935 (19.1) 127/935 (13.6) 1.923 1.522 to 2.429 0.0001 First Morbid Event Quartile 2 vs 1 213/905 (23.5) 175/904 (19.4) 1.410 1.153 to 1.725 0.0008 Quartile 3 vs 1 231/904 (25.6) 175/904 (19.4) 1.666 1.364 to 2.036 0.0001 Quartile 4 vs 1 270/905 (29.8) 175/904 (19.4) 2.195 1.804 to 2.672 0.0001 % Change in NE from baseline to 4 months Quartile 2 vs 1 150/937 (16.0) 135/936 (14.4) 1.228 0.971 to 1.553 0.0861 Quartile 3 vs 1 147/936 (15.7) 135/936 (14.4) 1.280 1.009 to 1.622 0.0415 Quartile 4 vs 1 164/937 (17.5) 135/936 (14.4) 1.480 1.171 to 1.870 0.0011 First morbid event Quartile 2 vs 1 214/906 (23.6) 212/905 (23.4) 1.096 0.905 to 1.328 0.3472 Quartile 3 vs 1 230/906 (25.4) 212/905 (23.4) 1.272 1.052 to 1.538 0.0129 Quartile 4 vs 1 239/906 (26.4) 212/905 (23.4) 1.381 1.141 to 1.671 0.0009 Cox regression with baseline covariate adjustment. Events occurring after permanent treatment discontinuation were not censored. Morbid events include death, sudden death with resuscitation, therapeutic IV HF therapy for 4 hours, or hospitalization for HF. percent decrease in BNP or NE at 4 months had the lowest subsequent mortality and first morbid event rates, and patients with the highest percent increase had the highest subsequent increase in event rates. Thus, both baseline and percent change in BNP and NE are important determinants of subsequent mortality and morbidity. These observational data from Val-HeFT do not allow any conclusions about a possible pathophysiologic link between hormone level and outcome. Norepinephrine has been implicated as a contributor to structural remodeling of the left ventricle, and the inhibition of its cardiac effects by Figure 5. RR (versus first quartile) for month 4%-change-from-baseline BNP and NE quartiles, with baseline as a covariate. Probability values for individual RRs (versus first quartile) and global probability value for test of no quartile differences are shown. -blockade has been associated with reversal of ventricular remodeling. Thus, it would be appropriate to suggest the possibility that the relationship between poor outcome and NE levels reflects an adverse effect of sympathetic stimulation rather than merely serving as a marker for more severe disease. It is interesting, however, that the relationship between BNP and outcome is even more powerful and BNP has not been implicated as a contributor to the progression of heart failure. Because BNP is produced in the heart, changes in circulating BNP may be considered a biological marker for changes in LV remodeling. However, there are few data correlating structural changes in the heart with changes in BNP. It is likely that the association of changes in BNP over time with mortality and morbidity noted in our study will correspond with structural and functional changes in the heart. Correlation of the echocardiographic and BNP data with mortality and morbidity in Val-HeFT may help to clarify this relationship. Such an analysis is in progress. The findings of this study raise important questions regarding the use of BNP and NE as guides to efficacy of HF therapy. In the case of BNP, preliminary data suggest that the use of a BNP-guided approach, targeted to reduce BNP to a predefined level, is associated with a significant reduction in cardiovascular events compared with a clinically guided approach. 19,20 These preliminary data would be supported by the findings in Val-HeFT. The evidence for the use of NE-guided approach is less clear, and the data from the PRIME II 21 and MOXCON 22 trials are troublesome in this regard. Both of these trials were terminated prematurely because of the adverse effects of ibopamine and moxonidine

Anand et al Changes in BNP and NE Over Time and Events 1283 on mortality despite significant reductions in plasma NE. Hence, although a physiological change in hormone levels may be a marker for outcome, strategies to reduce neurohormonal levels for mortality and morbidity benefit are at present premature and not justified. Conclusions Percent changes in BNP and NE over time are associated with corresponding changes in subsequent mortality and morbidity, thus supporting their role as important surrogate markers in HF. 12 These data, therefore, underscore the importance of including the measurement of neurohormones, especially of BNP in all patients, in future HF clinical trials. Because the annual mortality in patients with moderate to severe HF has now been reduced to 7% to 9% with appropriate use of ACE-I and -blockers, 13,23 it is likely that future HF trials will of necessity rely on markers for the progression of the disease to provide adequate power to demonstrate reduced morbidity. BNP and other surrogate measures such as LV structure and function could be considered sufficiently robust supplement morbidity end points to demonstrate the beneficial effects of a newer therapy in heart failure. Acknowledgments Supported by a grant from Novartis Pharma, Basel, Switzerland. We thank Susan Ziesche for her role as United States national study coordinator, Diane Judd, Marjorie Carlson, Yan Gao, Noeleen De Angelis, BiolD, and Mauro Torri, BiolD, for running the assays in the neurohormone core laboratories, Nora Aknay for facilitating international collaboration on manuscript preparation, and Allen Hester, PhD, and Joseph Yen, PhD, for statistical review of the manuscript. References 1. Wilson JR, Schwartz JS, St John Sutton M, et al. Prognosis in severe heart failure and relation to hemodynamic measurements and ventricular ectopic activity. J Am Coll Cardiol. 1983;2:403 409. 2. Szlachcic J, Massie BM, Kramer BL, et al. Correlates and prognostic implication of exercise capacity in chronic congestive heart failure. Am J Cardiol. 1985;55:1037 1042. 3. Cohn JN, Levine TB, Olivari MT, et al. Plasma norepinephrine as a guide to prognosis in patients with chronic congestive heart failure. N Engl J Med. 1984;311:819 823. 4. Swedberg K, Eneroth P, Kjekshus J, et al. Hormones regulating cardiovascular function in patients with severe congestive heart failure and their relation to mortality: CONSENSUS Trial Study Group. Circulation. 1990;82:1730 1736. 5. Massie BM, Conway M. Survival of patients with congestive heart failure: past, present, and future prospects. Circulation. 1987;7: IV11 IV19. 6. Cohn JN, Ferrari R, Sharpe N. Cardiac remodeling: concepts and clinical implications. A consensus paper from an international forum on cardiac remodeling. J Am Coll Cardiol. 2000;35:569 582. 7. Mann D, Kent R, Parsons B, et al. Adrenergic effects on the biology of the adult mammalian cardiocyte. Circulation. 1992;85:790 804. 8. Tan LB, Jalil JE, Pick R, et al. Cardiac myocyte necrosis induced by angiotensin II. Circ Res. 1991;69:1185 1195. 9. Anand IS, Chandrashekhar Y. Neurohormonal responses in congestive heart failure: effect of ACE inhibitors in randomized controlled clinical trials. In: Dhalla NS, Singhal PK, Beamish RE, eds. Heart Hypertrophy and Failure. Boston: Kluwer Academic Publishers; 1996:487 501. 10. Francis GS, Cohn JN, Johnson G, et al. Plasma norepinephrine, plasma renin activity, and congestive heart failure: relations to survival and the effects of therapy in V-HeFT II. The V-HeFT VA Cooperative Studies Group. Circulation. 1993;87(6 suppl):vi40 VI48. 11. Benedict CR, Francis GS, Shelton B, et al. Effect of long-term enalapril therapy on neurohormones in patients with left ventricular dysfunction: SOLVD Investigators. Am J Cardiol. 1995;75:1151 1157. 12. Anand IS, Florea VG, Fisher L. Surrogate end points in heart failure. J Am Coll Cardiol. 2002;39:1414 1421. 13. Cohn JN, Tognoni G. A randomized trial of the angiotensin-receptor blocker valsartan in chronic heart failure. N Engl J Med. 2001;345: 1667 1675. 14. Latini R, Masson S, Anand I, et al. Effects of valsartan on circulating brain natriuretic peptide and norepinephrine in symptomatic chronic heart failure: the Val-HeFT trial. Circulation. 2002;106:2454 2458. 15. Latini R, Masson S, De Angelis N, et al. Role of brain natriuretic peptide in the diagnosis and management of heart failure: current concepts. J Cardiac Failure. 2002;8:288 299. 16. Sabatine MS, Morrow DA, de Lemos JA, et al. Multimarker approach to risk stratification in non-st elevation acute coronary syndromes: simultaneous assessment of troponin I, C-reactive protein, and B-type natriuretic peptide. Circulation. 2002;105:1760 1763. 17. Richards AM, Nicholls MG, Yandle TG, et al. Plasma N-terminal pro-brain natriuretic peptide and adrenomedullin: new neurohormonal predictors of left ventricular function and prognosis after myocardial infarction. Circulation. 1998;97:1921 1929. 18. Bristow MR, Zelis R. Baseline and three month change in systemic venous norepinephrine as predictors of clinical outcomes in the BEST Trial investigators. J Am Coll Cardiol. 2001;37(suppl A):183A. 19. Murdoch DR, McDonagh TA, Byrne J, et al. Titration of vasodilator therapy in chronic heart failure according to plasma brain natriuretic peptide concentration: randomized comparison of the hemodynamic and neuroendocrine effects of tailored versus empirical therapy. Am Heart J. 1999;138(6 pt 1):1126 1132. 20. Troughton RW, Frampton CM, Yandle TG, et al. Treatment of heart failure guided by plasma aminoterminal brain natriuretic peptide (N-BNP) concentrations. Lancet. 2000;355:1126 1130. 21. Hampton JR, van Veldhuisen DJ, Kleber FX, et al. Randomised study of effect of ibopamine on survival in patients with advanced severe heart failure: Second Prospective Randomised Study of Ibopamine on Mortality and Efficacy (PRIME II) Investigators. Lancet. 1997;349:971 977. 22. Cohn J. The MOXCON study. Circulation. 1999;100:e128. 23. MERIT-HF Trial Investigators. Effect of metoprolol CR/XL in chronic heart failure: Metoprolol CR/XL Randomised Intervention Trial in Congestive Heart Failure (MERIT-HF). Lancet. 1999;353:2001 2007.