(Circulation. 1997;95:2597-2599.)
© 1997 American Heart Association, Inc.
Articles |
Predicting Outcomes in Severe Heart Failure
the University of California, San Francisco, Department of Medicine.
Key Words: Editorials • transplantation • prognosis • heart failure
 |
Introduction |
|---|
 Top Introduction Prediction: How Useful for...Prediction in Severe Heart...Where Now?References |
|---|
The crisis in cardiac transplantation will soon be upon us. With current candidate-selection practices, heart-distribution priorities, and inadequate numbers of donor hearts, the cardiac transplant system is projected to reach a dynamic equilibrium in 1998.1 At that time, 100% of available hearts will go to high-priority inpatients who cannot be weaned from intravenous inotropic agents or mechanical assist devices in the intensive care unit setting. Because high-priority candidates have a higher posttransplant mortality than low-priority outpatient candidates, the number of patients who die due to transplantation or are left destitute on the waiting list will double.1 2 3 Current trends confirm this impending crisis, as the percent of candidates on the waiting list for >1 year has increased from 12.5% in 1988 to 45.4% in 1994.4 A proposed solution to the crisis would be to list only candidates younger than 56 years of age whose 1-year survival without clinical deterioration to high-priority status is projected to be
20%.1 This practice would result in the distribution of half of donor hearts to high-priority candidates and half to low-priority candidates and would reduce the number of postoperative deaths and destitute candidates, thus maximizing donor-heart utilization. The key to this potential solution is better prediction of outcomes. |
Prediction: How Useful for Decision Making? |
|---|
|
TopIntroductionPrediction: How Useful for...Prediction in Severe Heart...Where Now?References |
|---|
Is it supposed to rain today? Should I take an umbrella? These simple questions, which many of us face every day, also describe symbolically the issues faced by clinicians who try to use predictions to improve clinical decisions. In weather forecasting, sophisticated mathematical models produce better predictions than can be made by flipping a coin, asking the average citizen, or even relying on the experience and judgment of a human meteorologist. Nevertheless, no one expects the National Weather Service or the individual meteorologist aided by the service's data to be perfect. In fact, probably none of us can resist the temptation to peek out the window or otherwise develop our own impressions. In clinical terms, we have our own prior probability that in bayesian terms should then be modified on the basis of the formal weather prediction.
Even when the combination of inputs from official prognosticators and our own common sense suggests that rain is a possibility or even a high probability, we may or may not change our plans for the day or decide to carry an umbrella or wear a raincoat. Maybe we should wait until the rain really starts. Maybe the rain will not be disruptive enough to interfere with our schedules or warrant the inconvenience of carrying the umbrella, which we always seem to lose anyway.
Although issues related to rain and umbrellas may seem trivial compared with issues related to severe congestive heart failure and cardiac transplantation, the principles of predictions and decision making are similar. In the clinical realm, better predictions allow us to help our patients plan for the future and identify subgroups of patients for whom different approaches may, on average, be preferred. However, we must be careful to integrate any such predictions with other quantitative and qualitative impressions and to not make critical decisions prematurely on the basis of an abstract computation.Predictive information tends to be most useful when the data can be used to influence an immediate decision rather than when it provides general prognostic information. To be useful beyond an immediate decision at hand, the information also must be updated (and "updatable"), or it rapidly loses its relevance.
|
Prediction in Severe Heart Failure |
|---|
|
TopIntroductionPrediction: How Useful for...Prediction in Severe Heart...Where Now?References |
|---|
Numerous factors have been suggested as predictive of survival in heart failure patients. Among these factors, peak
O2 is the only cardiophysiological prognostic factor routinely used as a selection criterion in heart failure patients being evaluated for cardiac transplantation.3 5 The 2-year event-free survival rate for patients with peak O2 14 mL·kg-1·min-1 was 48% compared with 84% for patients with peak O2>14 mL·kg-1·min-1.6 The discriminatory power of peak O2 is not surprising because it is dependent on two physiological processes that are impaired in heart failure: the ability to increase forward cardiac output and the ability to dilate peripheral skeletal muscle blood vessels in response to exercise.7
The use of peak O2 as the only prognostic factor used in transplant evaluations does not do the field justice, particularly since several studies have shown that combinations of factors are more predictive. For example, Campana et al8 evaluated 388 heart failure patients for cardiac transplantation between 1985 and 1989 and found six factors independently associated with prognosis: cause of heart failure, New York Heart Association functional class, S3 gallop, cardiac output, mean arterial pressure, and pulmonary artery diastolic pressure or pulmonary capillary wedge pressure. On the basis of a risk score generated from these factors, patients were stratified into low-, intermediate-, and high-risk groups with 1-year event-free survival rates of 95%, 75%, and 40%, respectively; however, this risk model was never prospectively validated in another cohort of patients.
In this issue of Circulation, Aaronson et al9 examined 80 clinical characteristics at the initial evaluation of 268 ambulatory heart failure patients (80% men) sent for consideration of cardiac transplantation to the Hospital of the University of Pennsylvania. Two predictive models were generated: a noninvasive model with seven independent variables (cause of heart failure, resting heart rate, left ventricular ejection fraction, mean blood pressure, intraventricular conduction delay, peak O2, and serum sodium); and an invasive model, which added an eighth variable (mean pulmonary capillary wedge pressure). Low-, medium-, and high-risk groups were defined with 1-year event-free survival rates of 93%, 72%, and 43%, respectively, almost identical to the results of the study by Campana et al,8 which used different variables. Notably, the invasive model did not perform better than the noninvasive model in predicting event-free survival.
The study by Aaronson et al,9 however, went further to validate their model in a prospective cohort of patients. The invasive and noninvasive models were applied to 199 ambulatory heart failure patients (81% men) evaluated at Columbia-Presbyterian Medical Center (1993 through 1995) for consideration of cardiac transplantation. Both models performed well in the prospective cohort. When the noninvasive model was used, the validation patients could be stratified into low-, medium-, and high-risk groups with 1-year event-free survival rates of 88%, 60%, and 35%, respectively. In both the derivation and validation cohorts, patients with medium- and high-risk scores had 1-year event-free survival rates lower than expected for patients undergoing cardiac transplantation (82%).4 The authors recommend that patients in the medium- and high-risk groups be listed immediately for transplantation but that the low-risk patients be deferred. This recommendation is consistent with the current practices of selecting patients for cardiac transplantation.
The prognostic model of Aaronson et al9 is methodologically sound and meets the major criteria proposed for valid and potentially useful clinical prediction models.10 As an example of the science of clinical epidemiology, the authors are to be congratulated for their care in identifying and studying their patients and in analyzing their data in both the original set of patients as well as in an independent set of patients in whom the accuracy of their predictions was validated.
The methods and results of the study by Aaronson et al9 raise three important questions: Is their model applicable to all heart failure patients referred for cardiac transplantation? Does their model represent an advance in our ability to predict survival in heart failure patients? Does their proposal help solve the cardiac transplantation crisis?
Aaronson et al9 should be further congratulated for including and examining the largest number of clinical parameters to date. However, even they failed to include several potentially important parameters, such as signal-averaged ECGs, Holter monitor abnormalities, neurohormones, and cytokines.11 12 13 14 15 16 17 18 Paradoxically, including peak O2 limits the general applicability of the model to patients able to exercise. This problem is magnified by the fact that the simple ability of heart failure patients to perform exercise testing is in and of itself prognostic.19
The model by Aaronson et al9 also may not be generalizable to the increasing percentage of heart failure patients treated with ß-blockers (
10% of their patients). ß-Blockers modulate resting heart rate and may increase left ventricular ejection fraction,20 both of which were important prognostic variables in their model. Analogous limitations may apply to patients treated with amiodarone, which also decreases resting heart rate and may improve left ventricular ejection fraction.21 Furthermore, only 20% of patients in this investigation were women, for whom ACE inhibitors may be less beneficial.22 23 24
Does this complex prognostic model represent an advance? In some respects, the answer is "yes." Although it was not more discriminatory than the model by Campana and coworkers,8 this new model used only noninvasive parameters and obviated the need for right heart catheterization. Furthermore, it was also prospectively validated and hence may, at least for now, become the new benchmark for predictions in such patients.
But is the model by Aaronson et al better than peak O2 alone for defining transplant selection criteria? Although the model defines three prognostic strata, two strata are recommended for transplant listing and one is not. Unfortunately, the noninvasive multivariate model proved to be no more discriminating in this regard in the validation cohort than was peak O2 alone.9
The comparison of this model with peak O2 is further complicated by the static nature of the model: it uses only clinical parameters present at the time of transplant evaluation and does not account for the dynamics of optimizing medical therapy. For example, at the University of California, Los Angeles, 90% of heart failure patients referred for cardiac transplant evaluation could be discharged from the hospital with oral vasodilator therapy after aggressive optimization of medical therapy.25 Forty-six percent of New York Heart Association class III and IV patients referred with left ventricular ejection fraction <20% and elevated pulmonary capillary wedge pressure responded to aggressive medical therapy and were discharged with oral vasodilator therapy with pulmonary capillary wedge pressure 16 mm Hg.26 This group exhibited a 1-year survival rate of 83%, similar to that seen with cardiac transplantation.26 27 Furthermore, this group of sustained medical responders had peak O2, 6-minute-walk tests, and other quality-of-life measures similar to cardiac transplant patients despite enormous differences in left ventricular ejection fraction (22±9% in the medical group versus 62±7% in the cardiac transplant group).27 A retrospective analysis of the V-HeFT trials also found that patients with serial improvement in left ventricular ejection fraction had better survival than patients without serial improvement.28 Thus, functional and hemodynamic responses to aggressive medical therapy are important prognostic factors in heart failure patients who at initial evaluation appear to have a grim prognosis but really have outcomes identical to patients undergoing cardiac transplantation.
|
Where Now? |
|---|
|
TopIntroductionPrediction: How Useful for...Prediction in Severe Heart...Where Now?References |
|---|
Some people avidly rely on weather forecasts to plan their schedules. Others plan their schedules and occasionally modify them only as the seriousness and high likelihood of accuracy of the forecast warrant. Still others forge on relentlessly until the realities of the weather force a change in plans. We should not be surprised that a similar divergence of responses may exist both for physicians and patients when provided with prognostic information.
The final litmus test of any heart failure prediction model may be whether it helps solve the impending cardiac transplantation crisis. There has been a recent paradigm shift moving away from the identification of heart failure patients "too well" to transplant25 26 27 toward identification of heart failure patients likely to die or deteriorate soon and therefore most likely to benefit from transplantation (References 1, 3, and 5, and L.W. Miller, MD, and D.G. Renlund, MD, unpublished letter to cardiologists of the American Society of Transplant Physicians, September 30, 1996). Any prediction model should be designed to optimize prediction at whatever prognostic threshold would be most helpful in this discrimination, such as the threshold at which medical prognosis equals transplant prognosis. An ideal model would apply as many independent, noninvasive, prognostic clinical parameters as needed to maximize discrimination while permitting dynamic flexibility to allow incorporation of newly identified prognostic variables, new treatments, and changes in a patient's clinical status. Although the ideal model does not yet exist, the models by Campana et al8 and Aaronson et al9 are clearly steps in the right direction. However, it is not at all clear that to substitute them for the current practice of discrimination on the basis of optimized peak O2 will help improve clinical decisions.
|
Footnotes |
|---|
Reprint requests to Lee Goldman, MD, University of California, San Francisco, School of Medicine, 505 Parnassus Ave, San Francisco, CA 94143-0120.
The opinions expressed in this editorial are not necessarily those of the editors or of the American Heart Association.
|
References |
|---|
|
TopIntroductionPrediction: How Useful for...Prediction in Severe Heart...Where Now?References |
|---|
1. Stevenson LW, Warner SL, Steimle AE, Fonarow GC, Hamilton MA, Moriguchi JD, Kobashigawa JA, Tillisch JH, Drinkwater DC, Laks H. The impending crisis awaiting cardiac transplantation: modeling a solution based on selection. Circulation. 1994;89:450-457.
2. Stevenson LW, Hamilton MA, Tillisch JH. Decreasing survival benefit from cardiac transplantation for outpatients as the waiting list lengthens. J Am Coll Cardiol. 1991;18:919-925.[Abstract]
3. Mudge GH, Goldstein S, Addonizio LJ, Caplan A, Mancini D, Levine TB, Ritsch ME Jr, Stevenson LW. Task Force 3: recipient guidelines/prioritization. J Am Coll Cardiol. 1993;22:21-31.[Medline] [Order article via Infotrieve]
4. 1995 Annual Report of the US Scientific Registry of Transplant Recipients and the Organ Procurement and Transplant Network: Transplant Data—1988-1994. Richmond, Va: United Network for Organ Sharing; and Rockville, Md: Division of Transplantation, Bureau of Health Resources Development, Health Resources and Services Administration, US Department of Health and Human Services; 1995.
5.
Costanzo MR, Augustine S, Bourge R, Bristow M, O'Connell JB, Driscoll D, Rose E. Selection and treatment of candidates for heart transplantation: a statement for health professionals from the Committee on Heart Failure and Cardiac Transplantation of the Council on Clinical Cardiology, American Heart Association. Circulation. 1995;92:3593-3612.
6.
Mancini DM, Eisen H, Kussmaul W, Mull R, Edmunds LH Jr, Wilson JR. Value of peak exercise oxygen consumption for optimal timing of cardiac transplantation in ambulatory patients with heart failure. Circulation. 1991;83:778-786.
7.
Packer M. Is tumor necrosis factor an important neurohormonal mechanism in chronic heart failure? Circulation. 1995;92:1379-1382.
8. Campana C, Gavazzi A, Berzuini C, Larizza C, Marioni R, D'Armini A, Pederzolli N, Martinelli L, Vigano M. Predictors of prognosis in patients awaiting heart transplantation. J Heart Lung Transplant. 1993;12:756-765.[Medline] [Order article via Infotrieve]
9.
Aaronson KD, Schwartz JS, Chen T-M, Wong K-L, Goin JE, Mancini DM. Development and prospective validation of a clinical index to predict survival in ambulatory patients referred for cardiac transplant evaluation. Circulation. 1997;95:2660-2667.
10. Wasson JH, Sox HC, Neff RK, Goldman L. Clinical prediction rules: applications and methodologic standards. N Engl J Med. 1985;313:793-799.[Abstract]
11. Cohn JN, Rector TS. Prognosis of congestive heart failure and predictors of mortality. Am J Cardiol. 1988;62:25A-30A.[Medline] [Order article via Infotrieve]
12. Edwards BS, Rodeheffer RJ. Prognostic features in patients with congestive heart failure and selection criteria for cardiac transplantation. Mayo Clin Proc. 1992;67:485-492.[Medline] [Order article via Infotrieve]
13.
Mancini DM, Wong KL, Simson MB. Prognostic value of an abnormal signal-averaged electrocardiogram in patients with nonischemic congestive cardiomyopathy. Circulation. 1993;87:1083-1092.
14. Levine B, Kalman J, Mayer L, Fillit HM, Packer M. Elevated circulating levels of tumor necrosis factor in congestive heart failure. N Engl J Med. 1990;323:236-241.[Abstract]
15.
Ferrari R, Bachetti T, Confortini R, Opasich C, Febo O, Corti A, Cassani G, Visioli O. Tumor necrosis factor soluble receptors in patients with various degrees of congestive heart failure. Circulation. 1995;92:1479-1486.
16. Wilson JR, Schwartz JS, Sutton MSJ, Ferraro N, Horowitz LN, Reichek N, Josephson ME. Prognosis in severe heart failure: relation to hemodynamic measurements and ventricular ectopic activity. J Am Coll Cardiol. 1983;2:403-410.[Medline] [Order article via Infotrieve]
17. Meinertz T, Hofman T, Kasper W, Treese N, Bechtold H, Stienen U, Pop T, Leitner E, Andresen D, Meyer J. Significance of ventricular arrhythmias in idiopathic dilated cardiomyopathy. Am J Cardiol. 1984;53:902-907.[Medline] [Order article via Infotrieve]
18. Gottlieb S, Kukin M, Ahern D, Packer M. Prognostic importance of atrial natriuretic peptide in patients with chronic heart failure. J Am Coll Cardiol. 1989;13:1534-1539.[Abstract]
19. Stevenson LW, Couper G, Natterson B, Fonarow G, Hamilton MA, Woo M, Creaser JW. Target heart failure populations for newer therapies. Circulation. 1995;92(suppl II):II-174-II-181.
20.
Packer M, Bristow MR, Cohn JN, Colucci WS, Fowler MB, Gilbert EM, Shusterman NH, for the US Carvedilol Heart Failure Study Group. Effect of carvedilol on morbidity and mortality in chronic heart failure: US Carvedilol Heart Study Group. N Engl J Med. 1996;334:1349-1355.
21. Doval HC, Nul DR, Grancelli MO, Perrone SV, Bortman GR, Curiel R, for the Grupo de Estudio de la Sobrevida en la Insuficiencia Cardiaca en Argentina (GESICA). Randomized trial of low dose amiodarone in severe congestive heart failure. Lancet. 1994;344:493-498.[Medline] [Order article via Infotrieve]
22. Limacher MC, Yusuf S, for the SOLVD Investigators. Gender differences in presentation, morbidity, and mortality in the Studies Of Left Ventricular Dysfunction (SOLVD): a preliminary report. In: Wenger NK, Sporoff L, Packard B, eds. Cardiovascular Health and Disease in Women. Greenwich, Conn: Le Jacq Communications Inc; 1993:345-348.
23. Pfeffer MA, Braunwald E, Moye LA, Basta L, Brown EJ Jr, Cuddy TE, Davis BR, Geltman EM, Goldman S, Flaker GC, and the SAVE Investigators. Effect of captopril on mortality and morbidity in patients with left ventricular dysfunction after myocardial infarction: results of Survival And Ventricular Enlargement trial—the SAVE Investigators. N Engl J Med. 1992;327:669-677.[Abstract]
24. Swedberg K, Held P, Kjekshus J, Rasmussen K, Ryden L, Wedel H. Effects of the early administration of enalapril on mortality in patients with acute myocardial infarction: results of the COoperative New Scandinavian ENalapril SUrvival Study II (CONSENSUS II). N Engl J Med. 1992;327:678-684.[Abstract]
25. Stevenson LW, Dracup KA, Tillisch JH. Efficacy of medical therapy tailored for severe congestive heart failure in patients transferred for urgent cardiac transplantation. Am J Cardiol. 1989;63:461-464.[Medline] [Order article via Infotrieve]
26. Stevenson LW, Tillisch JH, Hamilton M, Luu, M, Chelimsky-Fallick C, Moriguchi J, Kobashigawa J, Walden J. Importance of hemodynamic response to therapy in predicting survival with ejection fraction <20% secondary to ischemic or nonischemic dilated cardiomyopathy. Am J Cardiol. 1990;66:1348-1354.[Medline] [Order article via Infotrieve]
27. Stevenson LW. Tailored therapy before transplantation for treatment of advanced heart failure: effective use of vasodilators and diuretics. J Heart Lung Transplant. 1991;10:468-476.[Medline] [Order article via Infotrieve]
28. Cintron G, Johnson G, Francis G, Cobb F, Cohn JN, for the V-HeFT VA Cooperative Studies Group. Prognostic significance of serial changes in left ventricular ejection fraction in patients with congestive heart failure. Circulation. 1993;87(suppl VI):VI-17-VI-23.
This article has been cited by other articles:
 |
|
 |
|
  M. Demir, M. Kanadasi, O. Akpinar, Y. Donmez, M. Avkarogullari, C. Alhan, T. Inal, M. San, A. Usal, and M. Demirtas Cardiac Troponin T as a Prognostic Marker in Patients With Heart Failure : A 3-Year Outcome Study Angiology, November 1, 2007; 58(5): 603 - 609. [Abstract] [PDF]
|
|
|
|
 |
|
 G. A Sagnella Practical implications of current natriuretic peptide research Journal of Renin-Angiotensin-Aldosterone System, December 1, 2000; 1(4): 304 - 315. [PDF]
|
|
|
|
 |
|
 E. Loh, T. R. Rebbeck, P. D. Mahoney, D. DeNofrio, J. L. Swain, and E. W. Holmes Common Variant in AMPD1 Gene Predicts Improved Clinical Outcome in Patients With Heart Failure Circulation, March 23, 1999; 99(11): 1422 - 1425. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||