Figure 1. Index exercise echocardiogram and outcomes. A indicates annualized mortality rate in percentages; ExE, exercise echocardiography; and RVS, repeated revascularization.
Figure 2. Survival by ischemia status at first postrevascularization stress echo among 2105 asymptomatic patients. A, All-cause death; B, cardiac-related death.
Figure 3. Adjusted survival based on multivariate Cox regression models at mean of covariates, according to the presence of ischemia and subsequent revascularization.
Harb SC, Cook T, Jaber WA, Marwick TH. Exercise Testing in Asymptomatic Patients After RevascularizationAre Outcomes Altered?. Arch Intern Med. 2012;172(11):854-861. doi:10.1001/archinternmed.2012.1355
Author Affiliations: Department of Cardiovascular Medicine, Cleveland Clinic Heart and Vascular Institute, Cleveland, Ohio.
Background Although exercise echocardiography (ExE) of asymptomatic patients early (<2 years after percutaneous coronary intervention [PCI] or <5 years after coronary bypass graft surgery [CABG]) after revascularization is considered inappropriate, the appropriateness of later testing is indeterminate. Treatment responses to positive test results in either setting have uncertain outcome implications. We sought to identify whether predictors of increased risk by ExE could lead to interventions that change outcome in asymptomatic patients with previous coronary revascularization.
Methods Exercise echocardiography was performed in 2105 asymptomatic patients (mean [SD] age, 64  years; 310 [15%] were women; 845 [40%] had a history of myocardial infarction; 1143 [54%] had undergone PCI and 962 [46%] had undergone CABG 4.1 [4.7] years prior to the ExE). Ischemia was identified as a new or worsening wall motion abnormality. Patients were followed for a mean (SD) period of 5.7 (3.0) years for cardiac mortality. The association of ischemia during ExE with survival was assessed using Cox proportional hazard models, and an interaction with revascularization was sought.
Results Of 262 patients with ischemia (13%), only 88 (34%) underwent subsequent revascularization. Mortality (97 patients [4.6%]) was associated with ischemia (hazard ratio, 2.10; 95% CI, 1.05-4.19; P = .04) in groups tested both early (P = .03) and late (≥2 years after PCI or ≥ 5 years after CABG) (P = .001). However, the main predictors of outcome were clinical and stress testing findings rather than echocardiographic features. Subgroup analysis showed that asymptomatic patients without diabetes mellitus, with normal ejection fraction (≥50%), and normal exercise capacity (>6 METs [metabolic equivalent for task]) were unlikely to have a positive test result or events. Even high-risk patients did not seem to benefit from repeated revascularization.
Conclusions Asymptomatic patients who undergo ExE after coronary revascularization may be identified as being at high risk but do not seem to have more favorable outcomes with repeated revascularization. From a health economic standpoint, appropriateness of such testing must be carefully reviewed.
Recurrent ischemia and cardiac events are common after coronary revascularization procedures—both percutaneous coronary intervention (PCI) and coronary bypass graft surgery (CABG).1,2 This reflects the progression of coronary artery disease as well as progressive failure of treatment, including in-stent restenosis and graft atherosclerosis. In symptomatic patients after revascularization, the American College of Cardiology/American Heart Association (ACC/AHA) guidelines recommend evaluation with stress imaging tests,3 including exercise stress echocardiography (ExE). The evaluation of asymptomatic patients is more controversial. The presence of ischemia on a stress echocardiogram is predictive of adverse outcome in post-PCI and post-CABG patients.4- 6 However, there is no evidence that repeated revascularization (RVS) based on positive testing changes the course of the disease or patient outcomes.
The inappropriate use of noninvasive testing is not only costly but also could lead to unnecessary downstream testing and interventions such as coronary angiography and revascularization. While stress echocardiography in postrevascularization patients with ischemic symptoms is appropriate,7 appropriateness in asymptomatic patients depends on the timing relative to intervention. Testing is considered inappropriate early after PCI (<2 years) and CABG (<5 years),7 but the justification for these cutoffs is ill defined. We sought to study the effectiveness of testing asymptomatic patients early and late postrevascularization by examining the frequency of a positive response, the association of test results with subsequent revascularization and mortality, and the presence of any interaction of revascularization with these event rates.
This was an observational retrospective cohort study using prospectively obtained data in consecutive asymptomatic patients with a history of PCI or CABG who were referred for ExE at the Cleveland Clinic, Cleveland, Ohio, from January 2000 through November 2010. Our institution has no formal protocol regarding the assessment of patients who have had CABG or PCI and are symptom free; referral for ExE is solely at the discretion of individual physician treating the patient, usually on the basis of concerns regarding risk factor status or incomplete revascularization.
Before each ExE, a structured interview and medical chart review were done to gather data on demographics (age, sex), previous cardiac revascularization procedure (type, date), symptoms (chest pain, dyspnea), medications (aspirin, angiotensin-converting enzyme [ACE] inhibitors, angiotensin II receptor blockers [ARBs], statins, β-blockers), coronary risk factors (hypertension, diabetes mellitus, obesity [body mass index, calculated as weight in kilograms divided by height in meters squared, > 30], smoking), previous cardiovascular events (myocardial infarction [MI], stroke), and several cardiac and noncardiac diagnoses (atrial fibrillation [AF], congestive heart failure [CHF], and chronic obstructive pulmonary disease [COPD]). This clinical information was entered prospectively into the stress-testing database. The symptom status of patients was assessed by 2 independent reviewers who were blinded to each other's assessments and who were both unaware of the hypothesis of this study: 1 reviewer was from the registry group while the other was the exercise physiologist supervising the study. Only patients who were considered to be symptom free by both reviewers were included. Based on Appropriate Use Criteria,7 “early” testing was defined by investigation as testing less than 2 years after PCI or less than 5 years after CABG.
Information regarding revascularization was obtained by matching to the Cardiovascular Information Registry, a comprehensive database describing clinical and surgical characteristics of revascularized patients,8 which includes prospective follow-up for death, recurrent symptoms, and RVS. The databases and linkage were approved by the institutional review board.
Patients were prepared for exercise echocardiography in the traditional way,9 including abstaining from β blockade for 24 hours before the test. Maximal treadmill protocols, selected according to the patient age and functional status, were performed, and standard test end points were used.10 Ischemic symptoms, changes in ST segment, exercise capacity, and hemodynamic response to exercise were recorded for each patient. Exercise time, maximum ST-segment deviation, and the presence of exercise angina were used to calculate the Duke Treadmill score.11 In addition, we combined clinical (age, sex, angina, diabetes mellitus, smoking, hypertension) and stress data (exercise capacity, ST depression, angina, heart-rate recovery, ectopy during recovery) to derive a 5-year mortality risk index12 as previously described.
Baseline and stress images (performed immediately after cessation of exercise) were acquired from standard echocardiographic windows, including apical, parasternal, and subcostal.13 Studies were interpreted by experienced physicians, aware of the test indication but not of other clinical details of the patient. Standard left ventricular (LV) segmentation was used for LV regional wall motion analysis.14 Normal resting function with no worsening after exercise identified a normal test. Scar was characterized by akinesis or dyskinesis at rest. Ischemia was defined by new or worsening wall motion abnormalities. To express the extent of ischemia, myocardial segments were combined into vascular territories. Study results were considered abnormal if more than 1 segment showed evidence of ischemia or MI, and a normal study was defined by a normal response in all segments.
The primary end points were RVS and cardiac as well as all-cause mortality during a mean (SD) follow-up period of 5.7 (3.0) years. Repeated revascularization was attributed to the ExE result if it was performed within 6 months. End points were identified from the electronic medical record, state death records, and a national social security death index. End points were verified by reviewers who were not aware of stress testing or echocardiography results.
Differences in mortality between patients with and without ischemia during exercise echocardiography were evaluated by log-rank χ2 tests and compared using Kaplan-Meier survival curves. Adjusted estimates and tests were obtained using Cox proportional hazard models with adjustment for age, sex, type of prior revascularization (PCI vs CABG), ejection fraction, exercise capacity, smoking status, diabetes mellitus, hypertension, cardiovascular medications (aspirin, β-blockers, ACE inhibitors, ARBs, and statins), and cardiac history of AF, CHF, and MI. Log-likelihood ratio tests were used to evaluate any interaction terms. A series of subgroup analyses were performed in asymptomatic patients with low ejection fraction (EF <50%), diabetes mellitus, and low exercise capacity (<6 METs [metabolic equivalent for task]). Nested Cox regression models were used to assess the contribution of clinical factors, stress testing findings, and echocardiographic features in the prediction of mortality.
We identified 2105 patients with previous revascularization—1143 with prior PCI (54%) and 962 with prior CABG (46%)—who were asymptomatic at time of referral for the first follow-up ExE from January 2000 through November 2010. Their baseline clinical characteristics are summarized in Table 1. Of note, 40% had a history of MI, and the revascularization procedure had been performed at a mean (SD) of 4.1 (4.7) years prior to the exercise echocardiogram. Late testing (>2 years post-PCI and >5 years post-CABG) was performed in 434 PCI patients (38%) and 435 CABG patients (45%). The stress results overall suggested low risk—the mean exercise capacity was 8.7 (2.4) METs and the mean resting ejection fraction was 53% (8%). Ischemia was noted in 262 patients (13%), but only 1158 (55%) were completely normal.
Ischemia was detected at the index test in 262 patients, of whom 88 (34%) underwent RVS. Among those with a nonischemic index test, 908 patients (49%) had a second or subsequent test (Figure 1). Of these patients, the rate of initial or subsequent revascularization ranged from about 33% of those with a positive test result to 17% of symptomatic patients without ischemia and 12% to 14% of asymptomatic patients without ischemia.
In total, RVS was performed in 354 patients (17%) after a mean (SD) interval of 3.7 (3.2) years from the index ExE—mostly as PCI (12%), although 4% had CABG and 1% had both PCI and CABG (Table 2). The relationship between ischemia on the initial ExE and RVS is summarized as follows: 262 patients (13%) had ischemia, and of those, 174 (66%) did not undergo RVS. A total of 1845 patients (87%) did not have ischemia, and of these, 1579 (86%) did not undergo RVS. Among those who underwent RVS, 266 (75%) did not have ischemia on the initial exercise echocardiogram. However, RVS was associated with the detection of ischemia on subsequent testing (63 patients [24%]) performed for the development of spontaneous symptoms. Patients who developed symptoms were about twice as likely undergo RVS. By comparison, of the 1579 who did not undergo RVS, 197 (12%) became symptomatic at follow-up. The decision to proceed with RVS was based more on the clinical status of the patient than on the sole result of the test. Table 3, showing patients who had ischemia at any test, compares clinical characteristics in those who did and did not undergo RVS. Symptom status was the only variable significantly associated with the decision to proceed with RVS among patients with ischemia (P = .008).
Ninety-seven deaths (4.6%) occurred over a mean (SD) period of 5.7 (3.0) years of follow-up, an annualized mortality rate of 0.8%. Patients with ischemia on any postrevascularization ExE had a higher mortality compared with those without ischemia (8.0% vs 4.1%; P = .03). Figure 2 summarizes the difference in cardiac and all-cause mortality between patients with and without ischemia. Most deaths (76 patients [78%]) occurred in patients without ischemia, among whom the annualized mortality rate was 0.7% with RVS and 0.8% in patients without RVS (P = .15). There were 21 deaths in patients with ischemia, among whom the annualized mortality rates were 1.4% with RVS and 2.1% in patients without RVS (P = .38). By comparison, there were 20 deaths in the scar group (annualized mortality rate, 1.0%) and 32 deaths in the nondiagnostic group (annualized mortality rate, 1.4%).
The independent associations of clinical, stress testing, and echocardiographic characteristics with mortality are summarized in Table 4. Ischemia on exercise echocardiography was strongly associated with death (hazard ratio [HR], 2.10; 95% CI, 1.05-4.19; P = .04). Prior CABG, an abnormal ExE result even without ischemia (ie, scar or nondiagnostic), an EF of less than 50%, an exercise capacity of less than 6 METs, and diabetes mellitus were also associated with death. However, to be useful clinically, ExE needs to be not only of independent value but also incremental value. In nested models (Table 5), clinical and stress testing findings, but not echocardiographic features, were associated with both all-cause and cardiac mortality. Accordingly, we applied the predictive index based on clinical and exercise but no echocardiographic variables. When comparing patients with a predicted probability of 5-year survival below the median (ie, >4.0% risk of mortality over 5 years) to patients with higher predicted survival, the first group had a higher incidence of ischemia (15.7% vs 9.3%; P < .001) and a higher mortality rate (13.3% vs 3.9%; P < .001).
In the overall population, RVS was not significantly associated with survival (P = .67); Figure 3 compares the adjusted survival of patients according to the presence of ischemia and subsequent revascularization. This lack of association persisted even after considering interaction with an abnormal ExE (P = .70 for interaction with ischemia; P = .41 for interaction with other abnormal findings, ie, scar or nondiagnostic). Subanalysis based on the magnitude of ischemia comparing patients with mild ischemia with those with moderate or marked ischemia showed that there was no difference regarding benefit from revascularization. Using stratification by groups defined by a single cardiac risk factor, including EF of less than 50%, exercise capacity of less than 6 METs, incomplete revascularization, prior CHF or MI, or presence of diabetes mellitus or hypertension, we were unable to isolate subgroups of asymptomatic patients who might benefit from testing and revascularization. Because the study was underpowered to detect differential risks among small subgroups, we also examined a composite measure of cardiac risk. Patients at low risk (age <65 years, no diabetes mellitus, nonsmoker) accounted for 775 patients (37%), and 91% had no ischemia (in contrast with 85% among the 1330 [63%] high-risk patients; P < .001). Although the RVS rate was similar (15% vs 18%; P = .14), the mortality of the low-risk group was less (4.3% vs 11.2%; P < .001). Table 6 presents an alternative stratification based on the 5-year mortality index.
The appropriateness criteria distinguish between testing early and late postrevascularization. Early testing was performed in 709 PCI patients (62%), and 527 CABG patients (55%). RVS was less associated with early than late ischemia (28% vs 38%; P < .001). Ischemia on ExE was a predictor of mortality in patients tested early (P = .03) and late after revascularization (P = .001). However, RVS did not significantly improve survival in patients tested either early or late. The timing of the ExE (early vs late) did not predict mortality in multivariate Cox regression (HR, 0.89; 95% CI, 0.56-1.40; P = .61) (Table 4).
There are 5 major findings from this study of asymptomatic postrevascularization patients. First, only 262 (13%) had evidence of ischemia. Second, abnormal test results were associated with increased risks for subsequent overall and cardiac death in both early- and late-tested groups. Third, the main predictor of outcome was exercise capacity, suggesting that risk evaluation could be obtained from a standard exercise test rather than exercise echocardiography. Fourth, ischemic results were often not acted on; among patients with evidence of ischemia, 66% did not undergo RVS, and of those who underwent RVS, 75% did not have ischemia on their ExE. The decision to proceed with RVS was based more on the change in the clinical status of the patient with development of ischemic symptoms than on the sole result of the test. Finally, RVS was not associated with more favorable outcomes.
Coronary revascularization is highly effective for symptom control, and changes the outcome of selected patient groups with extensive coronary disease.15 Unfortunately, no revascularization procedures are curative, and recurrent coronary problems are common, owing to stent restenosis, graft closure or new native coronary lesions.1,2 Screening tests are often considered in the hope of identifying and resolving these problems electively and reducing the likelihood of acute presentations.
Based on previous prognostic studies, the appropriate use criteria7 identify ExE as being of indeterminate value late after PCI (≥2 years) and CABG (≥5 years), and inappropriate before this time frame. These judgments were made on the basis of expert opinion and, in relation to prognostic value, do not seem to be supported by our data. Previous studies have shown that stress echocardiography is an accurate diagnostic tool for identifying significant coronary stenoses after either PCI16,17 or CABG.18,19 Studies of moderate size have shown stress echocardiography to have prognostic value in asymptomatic patients postrevascularization.4- 6 Other functional tests (eg, exercise nuclear perfusion imaging)20,21 have identified the detection of ischemia after 5 years as being associated with adverse outcome. In contrast with stress echocardiography, the nuclear appropriateness criteria conclude that stress testing is appropriate in asymptomatic patients 5 or more years after CABG, and indeterminate at less than 5 years after CABG.22 However, the appropriate use criteria indicate that the value of risk assessment with single-photon emission computed tomography in asymptomatic post-PCI patients is uncertain, with no evidence of survival benefit.22,23 Moreover, the exercise testing and nuclear guidelines are discordant about the value of testing in selected asymptomatic, high-risk patients.24,25
Unfortunately, it has not been well established as to whether the information provided by testing asymptomatic patients alters treatment, and, if so, whether such treatment changes alter outcomes. This is a critical issue because the detection of clinically silent coronary disease progression may expose the patient to the risks and expense of further revascularization without a survival benefit. There are also barriers to RVS, including comorbidities, lack of suitable target vessels, potential revascularization of nonischemic territory based on suitability of anatomy rather than physiology, and difficulties in balancing the risk of another procedure with the risk of adverse outcome from ischemia. Probably as a consequence, ischemic results were often not acted on. In addition, it seems that RVS did not provide prognostic benefit even among the asymptomatic patients at highest risk.
The results of our study also somewhat contradict the current guidelines about the selection of standard exercise electrocardiography (ECG) and stress imaging tests after revascularization. The ACC/AHA stress testing guidelines state that among patients who can attain an adequate level of exercise, symptom-limited treadmill or bicycle exercise is the preferred form of stress because it provides the most information concerning patient symptoms, exercise capacity, cardiovascular function, and the hemodynamic response during usual forms of activity.24 However, for diagnostic purposes in the postrevascularization population, stress imaging tests are proposed in preference to the standard exercise ECG, because a number of ECG changes are nonspecific in this group.26 Our results suggest that from a prognostic standpoint, a combination of clinical and exercise data is effective in identifying patients at the highest risk, even though they are unlikely to benefit from RVS. To our knowledge, no previous study of exercise echocardiography in asymptomatic patients postrevascularization addressed the incremental value of imaging to clinical and stress testing findings—all previous studies included either mixed stress modalities (exercise and pharmacologic) and/or had a combination of asymptomatic and symptomatic patients. In the nuclear literature, a single-exercise, thallium-perfusion testing study20 had a comparable population in asymptomatic patients after CABG. In that work, Lauer et al20 found that impaired exercise capacity was more predictive of death (adjusted relative risk [RR], 4.16, 95% CI, 2.38-7.29) and major events (adjusted RR, 3.61; 95% CI, 2.22-5.87) than thallium-perfusion defects (adjusted RR, 2.78; 95% CI, 1.44-5.39 for death and adjusted RR, 2.63; 95% CI, 1.49-4.66 for major events) in asymptomatic patients after CABG.
This is an observational study, which shows associations but cannot ascribe causality. However, while it is conceivable that some patients may have a more favorable outcome with RVS, this seems unlikely on the basis that risk stratification failed to identify asymptomatic patients who will benefit from RVS. Likewise, although these patients were treated as a specialized referral center, the levels of risk and risk factors seen in this study are compatible with the usual postrevascularization patient. Finally, all of our patients were tested with exercise echocardiography, raising the question whether the results of this study are generalizable to all forms of stress testing.
In conclusion, the results of this study suggest that asymptomatic patients who undergo treadmill ExE after coronary revascularization may be identified as being at high risk but do not seem to have more favorable outcomes with RVS. Given the very large population of post-PCI and post-CABG patients, careful consideration is warranted before the screening of asymptomatic patients is considered appropriate at any stage after revascularization.
Correspondence: Thomas H. Marwick, MD, PhD, MPH, Department of Cardiovascular Medicine, Cleveland Clinic Heart and Vascular Institute, J1-5, Cleveland Clinic, 9500 Euclid Ave, Cleveland, OH 44195 (firstname.lastname@example.org).
Accepted for Publication: March 10, 2012.
Published Online: May 14, 2012. doi:10.1001/archinternmed.2012.1355
Author Contributions:Study concept and design: Harb and Marwick. Acquisition of data: Harb, Jaber, and Marwick. Analysis and interpretation of data: Harb, Cook, and Marwick. Drafting of the manuscript: Harb, Jaber, and Marwick. Critical revision of the manuscript for important intellectual content: Cook. Statistical analysis: Harb, Cook, and Marwick. Administrative, technical, and material support: Harb, Jaber, and Marwick. Study supervision: Jaber.
Financial Disclosure: None reported.
Funding for Less Is More: Staff support for topics research funded by grants from the California Health Care Foundation and the Parsemus Foundation.