Author + information
- Received January 27, 1998
- Revision received May 27, 1998
- Accepted June 4, 1998
- Published online October 1, 1998.
- R.David Anderson, MD∗,* (, )
- E.Magnus Ohman, MD∗,
- David R Holmes Jr., MD†,
- Robert A Harrington, MD∗,
- Gregory W Barsness, MD†,
- Nancy M Wildermann, BS∗,
- Harry R Phillips, MD∗,
- Eric J Topol, MD‡ and
- Robert M Califf, MD∗
- ↵*Address for correspondence: Dr. R. David Anderson, P.O. Box 3850, Duke University Medical Center, Durham, North Carolina 27710
Objectives. We sought to determine the prognostic significance of a history of congestive heart failure above that provided by baseline ejection fraction in patients undergoing percutaneous coronary interventions.
Background. Left ventricular function is a known predictor of survival in patients with coronary artery disease, as is a history of congestive heart failure. The contribution of heart failure history independent of left ventricular function is unknown.
Methods. Data were pooled from four interventional trials and the Duke University database. The combined dataset included 5,260 patients undergoing percutaneous interventions, 334 with and 4,926 without a history of heart failure. Patients were defined by the treating physician as having a clinical history of heart failure at the time of enrollment.
Results. The 30-day and 6-month mortality were higher in patients with a clinical history of congestive heart failure than in those without such a history (2% vs. <1%, p = 0.002 at 30 days, 5% vs. 1%, p = 0.001 at 6 months). Heart failure history did not influence the incidence of myocardial infarction, use of angioplasty or the use of bypass surgery during follow-up. Multivariable analysis revealed that heart failure history added significantly to ejection fraction in predicting intermediate-term (6-month) mortality (p = 0.01). Stepwise logistic regression also revealed heart failure history to be an independent predictor of 6-month mortality (odds risk 1.9, 95% confidence interval 1.1 to 3.5).
Conclusions. A clinical history of congestive heart failure is associated with increased early and intermediate-term mortality in patients undergoing percutaneous revascularization. Congestive heart failure history appears to provide prognostic information independent of that available from a patient’s left ventricular function. These findings suggest that patients with a clinical history of congestive heart failure who undergo a percutaneous intervention should be closely monitored, especially those with the lowest ejection fractions.
The state of the left ventricle is a strong predictor of outcome in patients with coronary artery disease. Early studies of patients with coronary artery disease treated medically made clear that ejection fraction was most closely associated with survival, an even stronger predictor than the number of diseased vessels (1,2). A clinical history of congestive heart failure (CHF) has also been shown to be a significant prognostic factor in this patient population (3,4). Both the clinical and angiographic predictors, heart failure history and ejection fraction (EF), have also been shown to be prognostically important in patients after myocardial infarction (5–7). Data have suggested that the inverse relationship between EF and mortality holds even for the right ventricle in the postmyocardial infarction setting (8).
The prognostic significance of EF and heart failure history have also been shown in other settings. These characteristics are important to consider in patients undergoing surgical revascularization (9–12), and are predictive of perioperative and long-term mortality (13,14)as well as long-term symptom status after surgical bypass (15). Congestive heart failure history has been found to predict survival in uncorrected aortic stenosis (16)and was found to strongly correlate with 1-year survival in a study of postmyocardial infarction arrhythmia suppression (17). Many of the clinical and angiographic predictors of outcome in patients with coronary artery disease treated medically (and surgically) have also been found to be important in those who undergo a percutaneous revascularization procedure. Ejection fraction has been shown to be prognostically important after coronary angioplasty for unstable angina in both short- and long-term studies (18–20). A patient’s left ventricular function (EF) is also prognostically significant after angioplasty for acute myocardial infarction (21).
The presence of CHF is associated with adverse events after percutaneous revascularization procedures (22,23). In an analysis from the National Heart, Lung and Blood Institute (NHLBI) registry of adverse events 8 years after angioplasty, a history of CHF at baseline was the strongest predictor of outcome among six clinical characteristics evaluated (24). To date, the independent contribution of CHF history to the prediction of morbidity and mortality after percutaneous revascularization has not been evaluated. We therefore studied, in a pooled analysis of interventional trials, whether a baseline history of CHF contributed significant information, above that contained in a patient’s EF, to the prognosis of patients undergoing percutaneous procedures.
Patients were pooled from four randomized trials (the Perfusion Balloon Catheter Study Group [PBC] Trial, Coronary Angioplasty Versus Excisional Atherectomy Trial [CAVEAT]-I and -II, Integrelin to Manage Platelet Aggregation to Prevent Coronary Thrombosis-II [IMPACT-II] and one interventional (Duke) database. These data were collected from April 1989 to August 1991 in the PBC trial, July 1991 to July 1992 in CAVEAT-I, March 1992 to September 1993 in CAVEAT-II, November 1993 to November 1994 in IMPACT-II and between April 1986 and June 1989 in the Duke dataset. The four trials and prospective cohort combined include data from 6,602 patients. Of these patients, 5,260 had both EF and CHF history available and provide the basis for this analysis.
The randomized trials in this analysis have been published (25–28). The randomized trial of gradual prolonged (perfusion) vs. standard balloon inflation in elective angioplasty (PBC trial) provided 547 patients. These patients had stable or unstable angina. Included were 1,012 patients from the CAVEAT-I trial of angioplasty vs. directional atherectomy in native coronary arteries. The data from 305 patients enrolled in the CAVEAT-II trial of angioplasty vs. directional atherectomy in saphenous vein bypass grafts were also included. The majority of patients in CAVEAT-I and -II had unstable angina. The IMPACT-II trial included 4,010 patients undergoing elective, urgent or emergency percutaneous coronary intervention. More than half of these patients were considered low risk (58%), while the remainder were enrolled with either unstable angina (38%) or myocardial infarction (3%).
The interventional database included all patients undergoing elective first-time angioplasty of native coronary artery lesions screened at Duke University Medical Center. This process identified 2,176 patients, of which a random sample of 1,056 were included in the database. After exclusion of 328 patients with acute myocardial infarction, 728 remained for this analysis.
All patients in this database received preprocedural aspirin (325 mg/day) and intraprocedural heparin. Patients in the PBC trial also received preprocedural dipyridamole, 75 mg three times daily. All patients except those in the PBC and IMPACT-II trials received a calcium-channel blocker for approximately 1 month. All angioplasty and atherectomy procedures were performed using standard techniques (29,30). The vascular access sheaths were removed from all patients between 4 and 24 h after the procedure (4 to 6 h in IMPACT-II) at the discretion of the treating physician.
The data from the four randomized trials were all collected prospectively on case report forms during the individual trials and forwarded to the coordinating center at Duke University. All four trial datasets were verified using consistency checks and double data entry; sites were queried about any missing data. Demographic and procedural data from all patients in the Duke database were collected prospectively and entered into the Duke Databank for Cardiovascular Disease. The complete dataset was verified by retrospective review of patient charts. Complete copies of the datasets from the four randomized trials were combined with the Duke data into a single database for the purposes of this analysis.
Patients in CAVEAT-I, CAVEAT-II and the PBC and IMPACT-II trials all had 6-month follow-up for the clinical events of death, myocardial infarction, repeat angioplasty or bypass surgery. In CAVEAT-I and the PBC trial follow-up continued for up to 1 year. The follow-up of in-hospital events was performed by retrospective chart review for patients from the Duke database. Patients were contacted by telephone or mail to assess their status at 1 year (the occurrence of death, myocardial infarction, repeat angioplasty or surgical bypass). The average duration of follow-up on patients in this study was 8 months. All patients in the randomized trials underwent protocol recatheterization at 6 months, except in IMPACT-II. All patients with a successful procedure in the Duke database were asked to return within 6 months for repeat angiography, regardless of symptom status.
The initial and follow-up angiograms of patients in randomized trials were read in a core laboratory by independent reviewers blinded to treatment. All cineangiograms were evaluated by quantitative coronary angiography. All initial as well as follow-up angiograms of patients in the Duke database were adjudicated by the consensus of two observers. In all patients a vessel was considered diseased if it contained at least one lesion that was more than 50% occluded. All EF measurements were obtained by ventriculography. In-laboratory and in-hospital (before hospital discharge) abrupt closure was defined as angiographic evidence of Thrombolysis In Myocardial Infarction grade 0 or grade 1 flow in the previously dilated artery. The definition of abrupt closure for patients in the Duke and CAVEAT-II datasets also required clinical or electrocardiographic evidence of ischemia. All patients had angiography at baseline and at 6 months, excluding those in IMPACT-II. IMPACT-II patients did not undergo follow-up angiography (28).
A clinical history of CHF was defined slightly differently in each study. The CAVEAT-I and -II and the IMPACT-II case report forms collected the variable “history of CHF” as a dichotomous variable (yes/no) with no further definition. The case report form for the PBC trial also collected whether the patient had a “history of CHF.” In the Duke dataset, the patient data form collected a variable, “recent history of CHF,” defined as symptoms of dyspnea or fatigue not related to an acute myocardial infarction within the past 6 weeks. A successful procedure was defined as one in which the residual stenosis in the target vessel was 50% or less. Restenosis occurred when the target lesion was found to have more than 50% narrowing on follow-up angiography.
All baseline characteristics are presented as frequencies and percentages for discrete variables and as medians with 25th and 75th percentiles for continuous variables. Univariable analysis at 30 days and 6 months was performed using the likelihood ratio chi-square test. We used logistic regression analysis to determine the contribution of baseline EF and heart failure history in predicting the outcomes of death, myocardial infarction, repeat angioplasty, bypass surgery or the composite of any of these. Baseline EF was treated as a continuous variable. Stepwise logistic regression was also performed to determine the significant predictors of 6-month mortality. The clinical characteristics used in this model were EF, history of CHF, number of diseased vessels, age, sex and diabetes history.
Of 5,260 patients, 334 had a history of CHF and 4,926 did not. The distribution of devices used in this cohort was 66% standard angioplasty, 23% directional atherectomy and 11% prolonged angioplasty with a perfusion balloon. More than a single interventional device was used in 12% of patients with and 11% of those without a history of heart failure.
Patients with a clinical history of CHF were older, more often female and more frequently had angina of a limiting degree (Table 1). These patients also were more likely to have other risk factors, including diabetes, hypertension, prior myocardial infarction and a previous revascularization procedure.
As expected, the patients with a clinical history of heart failure had more significant coronary disease and worse left ventricular function at baseline (Table 2). However, there were no differences between the two groups in angiographic outcomes either early or at the 6-month follow-up (Table 3).
Analyses were performed at both 30 days and 6 months to assess the impact of CHF history on early and intermediate clinical end points. Patients who had a clinical history of heart failure were less likely to survive at either time point (Table 4). There were no associations between CHF history and the incidence of myocardial infarction or the occurrence of a further revascularization procedure at these time points.
Logistic regression revealed that both EF and CHF history alone were prognostically important when predicting both 30-day and 6-month mortality (Table 5). When heart failure history was added to the 30-day model of mortality with EF incorporated, its additional prognostic potential did not reach statistical significance (p = 0.20). At 6 months, EF and CHF history each had a significant ability to predict mortality.
Congestive heart failure history added to the 6-month mortality model with EF already accounted for; its independent contribution was significant (chi-square 6.30, p = 0.01). When CHF history was first included in the model, EF added significantly both at 30 days and 6 months (p = 0.0001 for both). Neither EF nor CHF history was a significant predictor of either the composite or end points other than death at 6 months. Stepwise logistic regression was used to model 6-month mortality, and a history of heart failure was again shown to be predictive in addition to EF and patient age (Table 6). There was no interaction between gender and outcome; there was, however, a significant effect with inclusion of a trial variable.
Figure 1shows the probability of mortality at 6 months as a function of baseline EF for patients with and without a clinical history of CHF. Patients with a heart failure history undergoing a percutaneous intervention, especially patients with an EF less than 45%, were at greater risk of death than those without such a history.
A history of CHF is a marker for reduced survival in many different clinical settings, as is EF. This study confirms the prognostic importance of both EF and a history of CHF in 5,260 patients undergoing percutaneous revascularization procedures, as noted in other studies (18,20,22–25). The most important finding in the present study, however, is not only that a history of CHF was predictive of mortality, but that it also added significant information to that already provided by a patient’s EF. This suggests that for a given EF (Fig. 1), a patient undergoing a percutaneous intervention with a history of CHF is at greater risk of intermediate-term mortality than one without such a history. This difference appears to become more pronounced at lower EFs.
Prognostic significance of CHF history
Few studies have assessed the prognostic significance of heart failure history in those undergoing angioplasty, and none has addressed its contribution after adjusting for EF. In an analysis of differences in outcome after angioplasty between men and women, the hazard ratio for long-term mortality with a history of CHF was 2.6 (p = 0.0001) with EF in the model, and 2.94 (p < 0.0001) without it (31). The contribution of EF when included in the model containing a history of heart failure suggests that it adds prognostically to the latter. The reverse analysis, however, was not reported. In another regression analysis, CHF history and left ventricular EF were both strong predictors of mortality (risk ratios 3.0 and 3.3, respectively) (23), but both variables appeared to be in the model together, and the contribution of each, when adjusted for the other factor, was not evaluated. In the NHLBI PTCA registry (24), a history of CHF was the strongest correlate of 4-year mortality (adjusted risk 4.33, 95% confidence interval 2.67, 7.02). When the 8-year follow-up of these patients was available, CHF history was again the risk factor most strongly associated with adverse events, regardless of the end point (death, death/infarction, death/infarction/bypass surgery) (24). The independent contributions of EF and heart failure history were not reported in these studies.
The clinical characteristics of patients in this cohort were similar to those series previously reported (20,31–34). Procedural success was 80% in both groups in this study. While at first this may seem low, it is consistent with published data. An 84% success rate was reported in 427 patients who underwent angioplasty in 1981 (32), and other estimates have ranged from 82% to 88% during this time period (20,33,34). The lower rate seen in our study reflects the large number of angioplasty patients in this cohort from CAVEAT-I and -II who had significantly lower success rates (25,26). In addition, all angiograms in the present analysis were read in a core laboratory with the use of quantitative coronary angiography techniques; stenoses assessed in this manner tend to be graded as more severe than those assessed at individual sites (35)and may partly account for these lower procedural success rates.
The 6-month outcomes of patients in the current study are also consistent with previous studies. Mortality rates compare with those from reports of stable and unstable angina patients and also from the NHLBI angioplasty registry (33,34). Myocardial infarction and revascularization rates were also similar in this study compared with previous publications (19,22,23,32).
Interestingly, even in patients with mild or no left ventricular dysfunction (EFs >40%), a history of CHF is associated with an increased probability of short-term mortality (Fig. 1). As many as 36% to 42% of patients with the syndrome of CHF have a normal or near-normal (≥45%) EF (36,37). Diastolic dysfunction may play a role in the syndrome of CHF experienced by patients in this analysis, but its influence on this patient population remains speculative.
Nearly 40% of patients in the history of CHF cohort had diabetes. Recently published results of the Bypass and Angioplasty Revascularization Investigation Trial suggested that diabetics had a higher 5-year survival with bypass surgery as compared with angioplasty as the primary mode of revascularization (38). The decision to proceed with surgery as the mode of revascularization based solely on a history of diabetes, however, remains controversial (39). This decision algorithm should likely include the consideration of other factors such as angiographic descriptors of disease and clinical factors such as the presence of CHF.
A limitation of this analysis is the short duration of follow-up: 6 months for the trials included in this cohort. While the results of our study may not be generalized to the long-term outcome of patients undergoing percutaneous procedures, similarities with long-term studies are worth noting. In the regression analysis performed in this study, we found that EF and heart failure history were both predictors of short-term mortality, but did not correlate with later infarction or further revascularization during the follow-up period. In the NHLBI PTCA registry, EF was correlated with mortality at 4 years but was not significant in predicting the combined end points of death/infarction or death/infarction/bypass surgery (23). In this same population, heart failure history was the strongest predictor of long-term adverse events, including combined end points, but this appeared to reflect mostly the differences in mortality (24). Finally, Ruygrok et al. (20)also noted that EF predicted mortality at 10 years of follow-up but did not appear to influence rates of myocardial infarction or the use of angioplasty or bypass surgery.
Another limitation of this analysis is its retrospective design. While there are known difficulties with retrospective studies, the present study represents a large patient population treated at a broad range of hospitals. Further, the patients in this study were treated with different percutaneous procedures, and follow-up was systematic and rigorous. While the techniques of percutaneous intervention have been refined over the last several years, improvements in procedural outcomes are unlikely to affect the influence of EF and CHF history on clinical outcomes. Whether these results can be generalized to the population of patients receiving a stent remains to be studied.
In conclusion, this study confirms the importance of baseline EF on the short-term mortality of patients undergoing angioplasty and other percutaneous procedures. In this population of patients included in randomized trials and a registry, a clinical history of heart failure adds prognostic information to that already contained in a patient’s EF. This suggests that in selecting a mode of revascularization for patients with coronary artery disease, one should consider not only the state of the left ventricle but also the recent symptom history of the patient. In patients who may not be fully compensated, as manifested by recent symptoms of heart failure, the decision to proceed with percutaneous revascularization should be made cautiously.
Presented in part at the 45th Annual Scientific Session of the American College of Cardiology, March 26, 1996, Orlando, Florida.
- Coronary Angioplasty Versus Excisional Atherectomy Trial
- congestive heart failure
- ejection fraction
- National Heart, Lung and Blood Institute
- Perfusion Balloon Catheter Study Group
- Received January 27, 1998.
- Revision received May 27, 1998.
- Accepted June 4, 1998.
- American College of Cardiology
- Mock M.B,
- Ringqvist I,
- Fisher L.D,
- et al.
- Burggraf G.W,
- Parker J.O
- Weiner D.A,
- Ryan T.J,
- McCabe C.H,
- et al.
- Emanuelsson H,
- Karlson B.W,
- Herlitz J
- Pigott J.D,
- Kouchoukos N.T,
- Oberman A,
- Cutter G.R
- Bounous E.P,
- Mark D.B,
- Pollock B.G,
- et al.
- Passamani E,
- Davis K.B,
- Gillespie M.J,
- et al.
- Killip T,
- Passamani E,
- Davis K,
- et al.
- Alderman E.L,
- Fisher L.D,
- Litwin P,
- et al.
- Participants in the VA Cooperative Study on Valvular Heart Disease
- Hallstrom A,
- Pratt C.M,
- Greene H.L,
- et al.
- Ruygrok P.N,
- de Jaegere P.P.T,
- van Domburg R.T,
- et al.
- Holmes D.R,
- Detre K.M,
- Williams D.O,
- et al.
- Holmes D.R,
- Detre K,
- Weh W.L,
- et al.
- Holmes DR, Topol EJ, Califf RM, et al. for the CAVEAT-II Investigators
- Ohman EM, Marquis JF, Ricci DR, et al. for the Perfusion Balloon Catheter Study Group
- Ellis S.G
- Safian R.D,
- Gelbfish J.S,
- Erny R.E,
- et al.
- Weintraub W.S,
- Wenger N.K,
- Kosinski A.S,
- et al.
- Talley J.D,
- Hurst J.W,
- King S.B,
- et al.
- Detre K,
- Holubkov R,
- Kelsey S,
- et al.
- Lincoff AM, Keeler GP, Berdan LP, for the CAVEAT Investigators
- Barsness G.B,
- Peterson E.D,
- Ohman E.M,
- et al.