© 2008 European Society of Cardiology
Can B-type natriuretic peptides replace heart failure risk models?
University of Washington, Division of Cardiology Box 356422, 1959 NE Pacific Street, Seattle, WA 98195, United States E-mail address: levywc{at}u.washington.edu. Tel.: +1 206 221 4507; fax: +1 206 221 6835.
Received December 10, 2007; The article by Dr. Pfister and colleagues published in this issue of the European Journal of Heart Failure, compares the prognostic value of the biomarker NT-proBNP and glomerular filtration rate (GFR) with the multivariate CHARM CV morbidity and mortality model and the Seattle Heart Failure Model (SHFM) [1]. The authors found that NT-proBNP was a very powerful risk marker with a hazard ratio of 2.1 for 1 SD increase in log NT-proBNP with a 1-year ROC of 0.80.
However, the group of heart failure patients evaluated was unusual; 65% were evaluated at hospital discharge and 68% of these were NYHA 1 or 2. In addition, the median dose of furosemide was extraordinarily low at 20 mg/day. The annual mortality was 
10% with a CV morbidity and mortality of 16%/year over 1.4 years of follow-up compared to a 1-year mortality of 35% for most heart failure hospitalisations [2].
NT-proBNP is a very powerful risk marker in high risk coronary artery disease [3], post-myocardial infarction [4], and heart failure [5]. NT-proBNP levels increase with age, left ventricular size, NYHA class, diuretic use, ACEI use, and creatinine. Levels decrease with higher ejection fraction, β blocker use, and BMI [5]. In Val-HeFT, for NT-proBNP levels measured in 3916 patients, the ROC was more modest at 0.69 compared with 0.80 in the current report [5]. Is NT-proBNP level accurate in predicting actual morbidity/mortality rather than just risk stratification? Does a level of 1000 ng/L in a NYHA 4 vs. NYHA 2 patient or in a patient aged 30 vs. 80 years have the same risk? There are five key reports of the prognostic value of NT-proBNP; Val-HeFT [5], COMET [6], CORONA-HF [7], and GUSTO [4], and the current report [1]. All had similar 1-year mortality (8-10%). However, the median NT-proBNP varied approximately two-fold (from 669 to 1470 ng/L). In COPERNICUS, carvedilol did not alter NT-proBNP [8], suggesting that heart failure medications may not alter NT-proBNP levels. Thus, although NT-proBNP may risk stratify a single cohort, it is uncertain if the absolute estimate of risk will be similar between cohorts or with different heart failure regimens.
In the current report, renal function (creatinine) was a modest predictor (p=0.039), while GFR was a strong predictor [1]. However, GFR is a multivariate risk marker that includes age, gender, race, and creatinine. Addition of age, gender, and race to creatinine, increased the hazard ratio for a 1 SD change from 1.2 to 1.9. Given the poor performance of creatinine, I suspect, most of the benefit was due to the addition of age in calculating GFR rather than renal function per se. In many databases, age and gender alone may provide a ROC as high as 0.75 [9], whereas in severe heart failure such as in the PRAISE trial, age and gender were not multivariate predictors. In a recent analysis of the DIG trial, GFR<60 mL/min/m2 was a much more powerful marker in patients with diastolic dysfunction (hazard ratio 2.25 vs. 1.14 in patients with an ejection fraction of <35%) [10]. In the CHARM laboratory substudy [11], age and gender were much more powerful predictors than creatinine (
2=7). Four of the five biomarkers in the Seattle Heart Failure Model were more significant than creatinine, including lymphocytes (2=21), uric acid (2=12), haemoglobin (2=8), and sodium (2=8) in the CHARM laboratory substudy.
The CHARM CV morbidity and mortality model is a powerful multivariate risk model that was derived in 7599 diverse heart failure patients including those with an ejection fraction
40% [12]. This model is dependent on age (2=182) with an 50% increased risk for each decade. This is the first validation of the CHARM model in another dataset. The ROC of the CV morbidity and mortality model was 0.75 in the derivation dataset and 0.79 in this validation cohort. Complete validation of a model requires comparable ROC and accuracy. For a mean CHARM score of 19 the anticipated annual estimate of CV morbidity and mortality is 11%. The actual event rate was 16%/year in this cohort, 50% higher than predicted by the model.
The Seattle Heart Failure Model is a multivariate model that includes demographics, heart failure medications and devices, along with five biomarkers; haemoglobin, % lymphocytes, uric acid, total cholesterol, and serum sodium, which reflect activation of the sympathetic nervous system, renin-angiotensin-aldosterone system, and inflammation. The model has been validated in 14,000 patients [13,14] with a ROC of 0.73. It was designed to predict mortality not morbidity as reported in this article, but remained predictive with a ROC of 0.69. The output of the published algorithm is the Seattle Heart Failure Score, which varies from –1 to 3 in outpatients, and is the correct variable to use in a Cox model. This score is exponentially converted to an estimate of survival from 1 to 5 years. The authors evaluated the average life expectancy, which is an estimate based on the predicted 5-year mortality. The dose of furosemide is very low for symptomatic heart failure patients at 20 mg/day. In MERIT-HF, with a similar mortality, the average furosemide dose was 66 mg/day [15]. It is uncertain if the authors included diuretic agents that are commonly used in Europe like piretanide, indapamide, and other thiazide diuretics. Post hoc addition of creatinine to the SHFM resulted in minimal improvements in the ROC of 0.001-0.0 1[13,14].
In the current report NT-proBNP added significance to both the CHARM and SHFM models. It is uncertain if it significantly altered the ROC. Addition of BNP in 4500 patients improved the ROC of the SHFM by +0.03 [13,14]. Addition of NT-proBNP in the HOPE trial improved the ROC by 0.04 [3]. In Framingham, addition of 10 biomarkers including NT-proANP and BNP improved the ROC by 0.02 [9]. Thus, it appears that natriuretic peptides can add to cardiovascular clinical models with ROC changes of 0.02 to 0.04.
In conclusion, B-type natriuretic peptides are simple, commonly used biomarkers that have significant prognostic information in cardiovascular disease. Whether they should be utilized as a single risk marker or as part of a multivariate model is uncertain, but warrants further investigation.
The Seattle Heart Failure Model copyright is owned by the University of Washington Technology Transfer. Thoratec, Astellas, Scios and Vasogen have supported research with the model. Medtronic has licensed the Palm version of the model.
 |
References |
|---|
 Top References |
|---|
- Pfister R., Diedrichs H., Schiedermair A., Rosenkranz S., Hellmich M., Erdmann E., et al. Prognostic impact of NT-proBNP and renal function in comparison to contemporary multi-marker risk scores in heart failure patients. Eur J Heart Fail (2008) 10(3):315–320. 10.1016/j.ejheart.2008.01.009.
[Abstract/Free Full Text] - Heidenreich P.A., Fonarow G.C. Are registry hospitals different? A comparison of patients admitted to hospitals of a commercial heart failure registry with those from national and community cohorts. Am Heart J (2006) 152(5):935–939.[CrossRef][Web of Science][Medline]
- Blankenberg S., McQueen M.J., Smieja M., Pogue J., Balion C., Lonn E., et al. Comparative impact of multiple biomarkers and N-terminal pro-brain natriuretic peptide in the context of conventional risk factors for the prediction of recurrent cardiovascular events in the Heart Outcomes Prevention Evaluation (HOPE) study. Circulation (2006) 114(3):201–208.
[Abstract/Free Full Text] - James S.K., Lindahl B., Timmer J.R., Ottervanger J.P., Siegbahn A., Stridsberg M., et al. Usefulness of biomarkers for predicting long-term mortality in patients with diabetes mellitus and non-ST-elevation acute coronary syndromes (a GUSTO IV substudy). Am J Cardiol (2006) 97(2):167–172.[CrossRef][Web of Science][Medline]
- Masson S., Latini R., Anand I.S., Vago T., Angelici L., Barlera S., et al. Direct comparison of B-type natriuretic peptide (BNP) and amino-terminal proBNP in a large population of patients with chronic and symptomatic heart failure: the Valsartan Heart Failure (Val-HeFT) data. Clin Chem (2006) 52(8):1528–1538.
[Abstract/Free Full Text] - Olsson L.G., Swedberg K., Cleland J.G., Spark P.A., Komajda M., Metra M., et al. Prognostic importance of plasma NT-pro BNP in chronic heart failure in patients treated with a beta-blocker: results from the Carvedilol Or Metoprolol European Trial (COMET) trial. Eur J Heart Fail (2007) 9(8):795–801.
[Abstract/Free Full Text] - Kjekshus J., Apetrei E., Barrios V., Bohm M., Cleland J.G., Cornel J.H., et al. Rosuvastatin in older patients with systolic heart failure. N Engl J Med (2007) 357(22):2248–2261.
[Abstract/Free Full Text] - Hartmann F., Packer M., Coats A.J., Fowler M.B., Krum H., Mohacsi P., et al. Prognostic impact of plasma N-terminal pro-brain natriuretic peptide in severe chronic congestive heart failure: a substudy of the Carvedilol Prospective Randomized Cumulative Survival (COPERNICUS) trial. Circulation (2004) 110(13):1780–1786.
[Abstract/Free Full Text] - Wang T.J., Gona P., Larson M.G., Tofler G.H., Levy D., Newton-Cheh C., et al. Multiple biomarkers for the prediction of first major cardiovascular events and death. N Engl J Med (2006) 355(25):2631–2639.
[Abstract/Free Full Text] - Ahmed A., Rich M.W., Sanders P.W., Perry G.J., Bakris G.L., Zile M.R., et al. Chronic kidney disease associated mortality in diastolic versus systolic heart failure: a propensity matched study. Am J Cardiol (2007) 99(3):393–398.[CrossRef][Web of Science][Medline]
- Felker G.M., Allen L.A., Pocock S.J., Shaw L.K., McMurray J.J., Pfeffer M.A., et al. Red cell distribution width as a novel prognostic marker in heart failure: data from the CHARM Program and the Duke Databank. J Am Coll Cardiol (2007) 50(1):40–47.
[Abstract/Free Full Text] - Pocock S.J., Wang D., Pfeffer M.A., Yusuf S., McMurray J.J., Swedberg K.B., et al. Predictors of mortality and morbidity in patients with chronic heart failure. Eur Heart J (2006) 27(1):65–75.
[Abstract/Free Full Text] - Levy W.C., Mozaffarian D., Linker D.T., Sutradhar S.C., Anker S.D., Cropp A.B., et al. The Seattle Heart Failure Model: prediction of survival in heart failure. Circulation (2006) 113(11):1424–1433.
[Abstract/Free Full Text] - May H.T., Horne B.D., Levy W.C., Kfoury A.G., Rasmusson K.D., Linker D.T., et al. Validation of the Seattle Heart Failure Model in a community-based heart failure population and enhancement by adding B-type natriuretic peptide. Am J Cardiol (2007) 100(4):697–700.[CrossRef][Web of Science][Medline]
- Hjalmarson A., Goldstein S., Fagerberg B., Wedel H., Waagstein F., Kjekshus J., et alMERIT-HF Study Group. Effects of controlled-release metoprolol on total mortality, hospitalizations, and well-being in patients with heart failure: the Metoprolol CR/XL Randomized Intervention Trial in congestive heart failure (MERIT-HF). JAMA (2000) 283(10):1295–1302.
[Abstract/Free Full Text]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||