Author + information
- Received July 31, 2001
- Revision received December 19, 2001
- Accepted January 18, 2002
- Published online April 17, 2002.
- Robert B McCully, MB, ChB, FACC*,* (, )
- Veronique L Roger, MD, FACC*,
- Douglas W Mahoney, MS†,
- Kelli N Burger, BS†,
- Roger L Click, MD, PhD, FACC*,
- James B Seward, MD, FACC* and
- Patricia A Pellikka, MD, FACC*
- ↵*Reprint requests and correspondence:
Dr. Robert B. McCully, Division of Cardiovascular Diseases and Internal Medicine, Mayo Clinic, 200 First Street SW, Rochester, Minnesota 55905 USA.
Objectives We sought to define the prognostic implications of the extent and severity of exercise echocardiographic abnormalities in patients with good exercise capacity.
Background The exercise capacity of patients with known or suspected coronary artery disease (CAD) is of prognostic importance, as is the extent of exercise-related left ventricular (LV) hypoperfusion or dysfunction.
Methods We examined the outcomes of 1,874 patients with known or suspected CAD (mean age 64 ± 10 years, 64% men) who had good exercise capacity (≥5 metabolic equivalents [METs] for women, ≥7 METs for men) but abnormal exercise echocardiograms and analyzed the potential association between clinical, exercise and echocardiographic variables and subsequent cardiac events.
Results Multivariate predictors of time to cardiac death or nonfatal myocardial infarction (MI) were diabetes mellitus (risk ratio [RR] 1.88; 95% confidence interval [CI] 1.2 to 3.0), history of MI (RR 2.44; 95% CI 1.6 to 3.6) and an increase or no change in LV end-systolic size in response to exercise (RR 1.61; 95% CI 1.1 to 2.5). Using echocardiographic variables that were of incremental prognostic value, we were able to stratify the cardiac risk of the study population; cardiac death or nonfatal MI rate per person-year of follow-up was 1.6% for patients who had a decrease in LV end-systolic size in response to exercise (n = 1,330) and 1.2% for patients who did not have any severely abnormal LV segments immediately after exercise (n = 868).
Conclusions In patients with good exercise capacity, echocardiographic descriptors of the extent and severity of exercise-related LV dysfunction were of independent and incremental prognostic value. Stratification of patients into low- and higher risk subgroups was possible using these exercise echocardiographic characteristics.
For patients with known or suspected coronary artery disease (CAD), the prognostic importance of exercise capacity is well established (1–5). The extent of exercise-related left ventricular (LV) hypoperfusion or dysfunction, as measured by nuclear scintigraphy, has also been shown to be of prognostic value (6,7). Published information is limited regarding the prognostic value of the degree of exercise-related LV dysfunction, as measured by two-dimensional echocardiography (8).
The prognosis of patients after normal exercise echocardiography is favorable. Recently, we demonstrated that women who can exercise into stage 2 of the Bruce protocol (≥5 metabolic equivalents [METs]) and men who can exercise into stage 3 (≥7 METs) have an excellent prognosis if the exercise echocardiogram is normal (9).
In this study, we examined the outcomes of patients with good exercise capacity but abnormal exercise echocardiograms. We sought to determine if there was an association between any clinical, exercise or echocardiographic variables and subsequent cardiac death or nonfatal myocardial infarction (MI). Late coronary revascularization was included as a cardiac event in a separate analysis. Echocardiographic variables analyzed included semiquantitative indices of the extent and severity of LV systolic dysfunction before and immediately after exercise.
Data for patients who had exercise echocardiography from January 1990 to December 1995 at the Mayo Clinic were prospectively entered into a computer database from which the study population was selected. Patients who had good exercise capacity (≥5 METs for women, ≥7 METs for men) and abnormal exercise echocardiograms were candidates for the study. These exercise capacity criteria were selected because the expected treadmill time for men is, on average, 2 to 3 min longer than that for women (10). Patients were excluded if they had been diagnosed as having hypertrophic, restrictive or idiopathic dilated cardiomyopathy or if they had significant valvular heart disease, congenital heart disease or left bundle branch block. A total of 1,963 patients fulfilled criteria for inclusion. The Institutional Review Board approved the study protocol.
Exercise echocardiography protocol
Symptom-limited treadmill-exercise testing was performed in all patients. The Bruce protocol was used in 95% of patients. Two-dimensional echocardiographic images at rest and immediately after exercise were acquired, digitized, recorded and analyzed according to a previously published protocol (11). Both digitized and videotaped images were used to interpret the studies. By definition, all patients had rest or exercise-induced LV regional wall motion abnormalities. A 16-segment LV model was used, and all segments were scored semiquantitatively (12). The segments were scored at rest as follows: 1, normal; 2, mildly or moderately hypokinetic; 3, severely hypokinetic or akinetic; 4, dyskinetic; and 5, an aneurysmal segment. The segments of the images acquired immediately after exercise were scored as follows: 1, a normal segment at rest that had a normal hyperdynamic response to exercise; 2, a normal segment at rest that became mildly or moderately hypokinetic; 2, a mildly or moderately hypokinetic segment at rest that improved; 3, a severely hypokinetic or akinetic segment that improved or did not worsen; 3, a normal or mildly to moderately hypokinetic segment at rest that became severely hypokinetic or akinetic; 4, a segment that became dyskinetic; and 5, an aneurysmal segment. At rest and immediately after exercise, the percentage of LV segments that were abnormal (the number of segments with a score of 2, 3, 4 or 5 divided by 16, then multiplied by 100) and severely abnormal (that is, severely hypokinetic or worse [the number of segments with a score of 3, 4 or 5 divided by 16, then multiplied by 100]) was calculated. New or worsening wall motion abnormalities after exercise were interpreted as ischemic responses. The overall change in LV end-systolic size was assessed visually, comparing the side-by-side rest and postexercise digitized images in four standard views (parasternal long axis, parasternal short axis, apical four-chamber and apical two-chamber). The LV end-systolic size decreased, did not change appreciably or increased in response to exercise. Visual assessments were made of LV ejection fraction at rest and immediately after exercise (13).
Follow-up information was obtained from mailed questionnaires, scripted telephone interviews and reviewed medical records. Of the 1,963 patients who fulfilled the inclusion criteria, 24 patients declined to participate in the study. For another 65 patients (3%), there was no follow-up. Follow-up was complete for 1,874 patients (95% of eligible patients).
The primary end points considered in the analysis were cardiac death (sudden death or death preceded by an acute coronary syndrome or congestive heart failure) and nonfatal MI. In a separate analysis, late coronary revascularization was included as an event. All cardiac events were verified by contacting the patients’ primary physicians and obtaining medical records or death certificates. With regard to the timing of coronary revascularization after exercise echocardiography, we observed a clustering of procedures done within the first 30 days, after which there was a lower, steady rate of coronary revascularization. Therefore, in the analysis that included coronary revascularization as an event, we did not include patients who had coronary revascularization within 30 days in an attempt to reduce the post-test referral bias that might exaggerate the predictive value of the variables analyzed.
Baseline clinical, exercise and echocardiographic characteristics of the study group were summarized as the mean value ± SD for continuous variables and as a percentage of the group total for categorical variables. Cardiac event-free survival was estimated by the Kaplan-Meier method, using a time-to-first-event approach. Patients who died of noncardiac causes were censored at the time of death. Patients who had coronary revascularization were censored at the time of the procedure for the cardiac death or nonfatal MI-free survival analysis. For the analysis of event-free survival, which included late coronary revascularization as a cardiac event, patients who had early (≤30 days) coronary revascularization were censored. The association of clinical, exercise and echocardiographic characteristics with time to first event was investigated with the Cox proportional hazards regression model. Univariate and multivariate predictors were identified for cardiac events consisting of cardiac death or nonfatal MI and for all cardiac events, including late coronary revascularization. The pairwise interaction of variables within the final multivariate model was also investigated. Hierarchical or incremental prognostic modeling was also performed to examine the incremental prognostic value of information obtained from the exercise electrocardiogram (ECG) and the exercise echocardiogram.
Clinical and exercise echocardiographic characteristics
The mean age of the 1,874 patients was 64 ± 10 years. Clinical characteristics of the patients are summarized in Table 1. Of these patients, 897 (48%) had a history of MI or prior coronary revascularization. For the remaining patients, the pretest probability of CAD was high, intermediate and low in 203 (11%), 397 (21%) and 377 (20%) patients, respectively (9). The exercise echocardiographic characteristics of the patients are summarized in Table 2. The exercise ECG was positive for ischemia in 563 patients (30%). There were no regional wall motion abnormalities at rest in 513 patients (27%), and the exercise echocardiogram was positive for ischemia in 1,243 patients (66%).
The mean follow-up period after exercise echocardiography was 3.1 ± 1.6 years (maximum 8 years). There were 59 noncardiac deaths. The cause of death for these patients was cancer in 34, stroke in 6, infection in 6, trauma in 4 and other causes in 9. There were 107 major cardiac events observed during the follow-up period; 33 patients had a cardiac death and 74 patients had a nonfatal MI. Additionally, 70 patients had early coronary revascularization and 159 patients had late coronary revascularization.
Multivariate predictors of cardiac death or nonfatal MI
The overall cardiac death or nonfatal MI rate was 2% per person-year of follow-up. There were 14 univariate and 3 multivariate predictors of time to cardiac death or nonfatal MI (Table 3). The multivariate predictors were diabetes mellitus, history of MI and an increase or no change in LV end-systolic size in response to exercise. The effect of an abnormal LV end-systolic size response to exercise on outcome differed for diabetic and nondiabetic patients (p = 0.04). The estimated risk ratio (RR) was 3.85 (95% confidence interval [CI] 1.6 to 9.1) (p = 0.003) for diabetic patients and 1.4 (95% CI 0.9 to 2.1) (p = 0.18) for nondiabetic patients.
Incremental prognostic value of exercise ECG and exercise echocardiography
To examine the incremental prognostic value of exercise ECG and echocardiographic variables over clinical information, we sequentially added the exercise ECG and then the echocardiographic variables to a clinical prognostic model. The global chi-square value for the clinical model was 31. In this cohort with good exercise capacity and abnormal exercise echocardiograms, exercise ECG variables did not add prognostic information to the clinical model (largest obtainable global chi-square = 33, p = 0.16). For exercise echocardiographic variables, the change in LV end-systolic size in response to exercise provided the greatest incremental prognostic value (global chi-square = 37, p = 0.046 vs. the clinical model plus exercise ECG). This was similar to the chi-square value obtained when the percentage of LV segments that were severely abnormal immediately after exercise was considered (global chi-square = 37, p = 0.046).
Stratification of cardiac risk
For the 1,330 patients (71%) who had a decrease in LV end-systolic size in response to exercise, the cardiac death or nonfatal MI rate per person-year of follow-up was 1.6% versus 2.9% for the 544 patients (29%) with an increase or no change in LV end-systolic size (Figs. 1A and 2). ⇓For patients with diabetes (n = 130) who had an increase or no change in LV end-systolic size, the event rate was 7.5% per person-year of follow-up.
Patients who did not have any severely abnormal (severely hypokinetic or worse) LV segments immediately after exercise (mildly abnormal, n = 868, 46% of patients) had a favorable outcome, with a cardiac death or nonfatal MI rate of only 1.2% per person-year of follow-up. Those who had severe abnormalities in 1% to 25% of LV segments immediately after exercise (moderately abnormal, n = 657, 35% of patients) had a cardiac death or nonfatal MI rate of 2.2%, and patients who had severe abnormalities in more than 25% of LV segments immediately after exercise (markedly abnormal, n = 349, 19% of patients) had a cardiac death or nonfatal MI rate of 3.8% (Figs. 1B and 3) ⇓.
Patients who had early coronary revascularization
The 70 patients (4%) who had coronary revascularization within 30 days after the exercise echocardiogram were compared with the 1,804 patients who did not have early coronary revascularization. These 70 patients were just as likely to have diabetes mellitus or a history of MI, but more often had an increase or no change in LV end-systolic size in response to exercise (63% vs. 28%, p = 0.001). Consistent with this descriptor of severe postexercise LV dysfunction, these patients had more extensive ischemia (>25% of LV segments became ischemic in 59% vs. 20%, p = 0.001) and more extensive severe regional wall motion abnormalities immediately after exercise (>25% of LV segments were severely abnormal in 41% vs. 18%, p = 0.001). Of the 70 patients who had early coronary revascularization, 2 had periprocedural MIs; 1 MI was fatal. The other 68 patients had good outcomes, with a subsequent cardiac death or nonfatal MI rate of 0.5% per person-year of follow-up (vs. 2% for those patients who did not have early revascularization).
Late coronary revascularization as a cardiac event
When late coronary revascularization was included as a cardiac event, the number of cardiac events observed was 266, and this cardiac event rate per person-year of follow-up was 5%. There were 24 univariate (data not shown) and 7 multivariate predictors of cardiac death, nonfatal MI, or late coronary revascularization. The multivariate predictors were diabetes mellitus (RR 1.80 [95% CI 1.3 to 2.5], p = 0.0003), history of typical angina (RR 1.65 [95% CI 1.3 to 2.2], p = 0.0002), history of MI (RR 1.45 [95% CI 1.1 to 1.9], p = 0.007), hypertension (RR 1.36 [95% CI 1.1 to 1.9], p = 0.02), angina during the treadmill test (RR 1.48 [95% CI 1.1 to 2.1], p = 0.02), peak rate-pressure product (RR 0.44 [95% CI 0.3 to 0.7], p = 0.0003) and percentage of LV segments that were severely abnormal immediately after exercise (RR 1.28 [95% CI 1.1 to 1.5], p = 0.001).
For the incremental prognostic model, the following global chi-square values for the different models were determined: clinical model, chi-square = 44 (important variables: diabetes mellitus, history of MI and prior coronary revascularization); clinical model plus exercise ECG variables, chi-square = 75, p < 0.0001 versus clinical model (important variables: angina during the treadmill test, peak rate-pressure product and exercise ECG positive for ischemia); and clinical model plus exercise ECG plus exercise echocardiography, chi-square = 84, p = 0.003 versus clinical model plus exercise ECG (important variable: the percentage of LV segments that were severely abnormal immediately after exercise).
Patients with excellent exercise capacity
Of the patients, 658 (35%) were able to exercise ≥10 METs. In this subgroup of patients with excellent exercise capacity, 20% had an increase or no change in LV end-systolic size in response to exercise (compared with 34% of those patients who had good exercise capacity [≥5 METs for women, ≥7 METs for men, but <10 METs]). Patients with excellent exercise capacity had similar outcomes compared with patients who had good exercise capacity; the one-, three- and five-year cardiac death or nonfatal MI-free survival rates for patients who exercised ≥10 METs were 97.9 ± 0.6%, 95.3 ± 1.0%, and 90.2 ± 2.1%, respectively, and for patients who exercised <10 METs, 98.8 ± 0.3%, 95.3 ± 0.7%, and 89.6 ± 1.5%, respectively.
It is known that patients with CAD who have excellent exercise capacity have a favorable prognosis (1,2). In this study of patients with known or suspected CAD, we examined the outcomes of patients with good exercise capacity (as defined) and different degrees of exercise echocardiographic abnormalities. The following important clinical question was addressed in this study: What, if any, are the prognostic implications of an abnormal exercise echocardiogram for patients who are able to exercise adequately? It was not our intention to examine the outcomes of all patients who had exercise echocardiography. We deliberately excluded patients with normal exercise echocardiograms and those with reduced exercise capacity.
The current study
In this large group of patients with good exercise capacity, certain exercise echocardiographic variables were of incremental prognostic value when added to clinical and exercise ECG data and facilitated cardiac risk stratification of the study population. These variables, the change in LV end-systolic size in response to exercise and the percentage of LV segments that were severely abnormal (severely hypokinetic or worse) immediately after exercise, are descriptors of the extent and severity of exercise-related LV dysfunction. These two variables were more important than other variables, such as rest and exercise ejection fraction, the extent and severity of rest LV wall motion abnormalities and the presence and extent of exercise-induced myocardial ischemia. Both variables are a summation of LV function or dysfunction at rest plus any exercise-induced myocardial ischemia. When late coronary revascularization was included as an event, the percentage of LV segments that were severely abnormal immediately after exercise was the most important echocardiographic variable for prognosis prediction and was of incremental prognostic value.
Comparison with previous studies
Many studies have validated the notion that the noninvasive evaluation of patients with known or suspected CAD by exercise or pharmacologic stress radionuclide perfusion imaging techniques can provide incremental prognostic information and lead to appropriate risk stratification and cost-effective management strategies (6). Several studies have demonstrated that radionuclide myocardial perfusion imaging has incremental prognostic value over clinical and exercise data when major cardiac events or total events (major events and late coronary revascularization) are used as end points (6).
The prognostic value of exercise echocardiography in ambulatory, outpatient populations has been examined more recently (8,14–16). The first study to compare the prognostic power of exercise echocardiography and nuclear scintigraphy in the same patients showed that the variables of cardiac function and perfusion measured by these two imaging techniques were equally powerful in distinguishing patients at low risk for cardiac events from those at high risk (8). In that study, the exercise wall motion score index, another descriptor of the severity of LV dysfunction, was the most important echocardiographic predictor of cardiac events. The wall motion score index is a summary of LV segmental wall motion abnormalities and does not readily differentiate between abnormal and severely abnormal segments. In the present study, we chose to make the distinction between abnormal and severely abnormal segments and counted them. This enabled us to analyze the potential prognostic implications of abnormal versus severely abnormal segments. We were thus able to identify a large subgroup of patients without severe wall motion abnormalities who were at low risk of having a major cardiac event. Investigators using exercise nuclear scintigraphy have similarly been able to stratify patients into low- and higher-risk subgroups, on the basis of the extent and severity of exercise-related LV myocardial perfusion abnormalities (17–20).
The only high-risk subgroup of patients that we identified included patients with diabetes who had an abnormal response of the LV end-systolic size to exercise. The 70 patients who were censored in the event-free survival analyses because they had early coronary revascularization could also have been a higher risk subgroup, because most had an increase or no change in LV end-systolic size in response to exercise and extensive exercise-induced myocardial ischemia. Analysis of this patient subgroup suggests that early coronary revascularization can offer an outcome advantage to some patients with good exercise capacity but abnormal exercise echocardiography.
Our definition of good exercise capacity is approximate because exercise capacity is a function of both gender and age. By gender- and age-specific aerobic capacity criteria, some patients had only fair or low exercise capacity (21). We selected the cutpoints of 5 METs for women and 7 METs for men because we showed previously that the outcome of such patients is excellent if the exercise echocardiogram is normal (9). These cutpoints are also simple to use in clinical practice.
Patient follow-up was not 100% complete, and it averaged three years. The results of exercise echocardiography were known to the patients’ physicians and undoubtedly played a role in patient management decisions. We attempted to minimize the effects of post-test referral bias by censoring patients who had early coronary revascularization.
The exercise echocardiograms were interpreted by experienced level III echocardiologists who work in a large clinical echocardiography laboratory at a tertiary medical center. The visual assessment of rest and exercise regional and global LV function, on which this and other studies are based, is an interpretive skill that can be mastered, but one that requires extensive training and ongoing experience. One of the more straightforward aspects of exercise echocardiographic interpretation, however, is the side-by-side comparison of LV end-systolic size at rest and immediately after exercise that digitizing technology permits. This assessment of the change in LV end-systolic size in response to stress has been validated by comparing it with the more exacting biplane “method of discs” measurement of LV end-systolic volume (22,23). Because there is interinstitutional variability in interpretation of stress echocardiograms (24), the findings of this study ideally should be confirmed in other patient populations.
For patients with good exercise capacity but abnormal exercise echocardiograms, echocardiographic descriptors of the extent and severity of exercise-related LV dysfunction were of independent and incremental prognostic value. This information can be used to stratify patients into low- and higher-risk subgroups.
☆ Supported by the Mayo Foundation.
- coronary artery disease
- confidence interval
- left ventricular
- metabolic equivalents
- myocardial infarction
- risk ratio
- Received July 31, 2001.
- Revision received December 19, 2001.
- Accepted January 18, 2002.
- American College of Cardiology Foundation
- Weiner D.A,
- Ryan T.J,
- McCabe C.H,
- et al.
- Ellestad M.H,
- Wan M.K
- Beller G.A
- Lee K.L,
- Pryor D.B,
- Pieper K.S,
- et al.
- Olmos L.I,
- Dakik H,
- Gordon R,
- et al.
- McCully R.B,
- Roger V.L,
- Mahoney D.W,
- et al.
- Schiller N.B,
- Shah P.M,
- Crawford M,
- et al.
- Marwick T.H,
- Mehta R,
- Arheart K,
- Lauer M.S
- Ladenheim M.L,
- Pollock B.H,
- Rozanski A,
- et al.
- Machecourt J,
- Longere P,
- Fagret D,
- et al.
- Marie P.Y,
- Danchin N,
- Durand J.F,
- et al.
- Hachamovitch R,
- Berman D.S,
- Shaw L.J,
- et al.
- McArdle W.D,
- Katch F.I,
- Katch W.L
- Chuah S.C,
- Pellikka P.A,
- Roger V.L,
- McCully R.B,
- Seward J.B
- Hoffmann R,
- Lethen H,
- Marwick T,
- et al.