Long-Term Prognostic Value of Exercise Echocardiography Compared With Exercise 201Tl, ECG, and Clinical Variables in Patients Evaluated for Coronary Artery Disease
Background—The accuracy of exercise echocardiography and 201Tl single photon emission computed tomography (SPECT) is similar in the diagnosis of coronary artery disease (CAD). However, comparative data on long-term prognosis are lacking.
Methods and Results—Clinical variables and exercise, echocardiographic, and 201Tl tomographic parameters were studied in 248 patients (age, 56±12 years [mean±SD]; 189 men) who underwent simultaneous treadmill exercise 201Tl SPECT and echocardiography. Follow-up was obtained in 225 patients (91%) at a mean of 3.7±2.0 years. A total of 64 cardiac events occurred. With the use of stepwise logistic regression, 4 models simulating clinical stress testing scenarios were evaluated in the prediction of all cardiac events, ischemic events, and/or cardiac death. The best clinical models were exercise echocardiography with exercise ECG and exercise 201Tl SPECT with exercise ECG. Both models were comparable in the prediction of cardiac events. For the exercise echocardiography model, exercise wall motion score index and induction of ischemia were the strongest predictors of events with ORs of 2.63 per unit increment (95% CI, 1.34 to 5.17; P=0.005) and 4.1 (95% CI, 1.32 to 12.79; P=0.015), respectively. For the model with exercise 201Tl SPECT, the strongest predictor was ischemic perfusion defect (OR, 4.93; 95% CI, 1.72 to 14.08; P=0.003). The absence of ST changes during exercise decreased the risk of events. For the prediction of ischemic events and/or cardiac death, echocardiographic and 201Tl parameters were the only predictive variables.
Conclusions—In patients evaluated for CAD, exercise echocardiography and 201Tl combined with ECG variables provide comparable prognostic information and can be used interchangeably for risk stratification.
Several studies demonstrated that in addition to conventional cardiac risk factors, resting left ventricular function and the presence or absence of stress-induced ischemia are among the most powerful predictors of cardiac events.1 2 3 Currently, exercise echocardiography and radionuclide perfusion imaging are the 2 best modalities for the detection of exercise-induced ischemia and are superior to conventional stress ECG.4 5 Although exercise echocardiography evaluates global and regional function and the radionuclide technique assesses relative myocardial perfusion, studies comparing these techniques have shown comparable results in the detection of coronary artery disease (CAD).4 6
Current literature suggests that exercise echocardiography and radionuclide perfusion imaging are both capable of detecting subgroups of patients at higher risk for future cardiac events.4 7 8 9 However, to date there has been no comparison of the 2 techniques for risk stratification in the same patients. The purpose of this study, therefore, was to compare the relative prognostic value of exercise echocardiography and exercise 201Tl single photon emission computed tomography (SPECT) in patients with suspected or known CAD. The incremental prognostic value of both techniques compared with clinical variables and exercise ECG was evaluated.
The population consisted of 248 patients (189 men; mean age, 56.3±12 years) evaluated between 1986 and 1993 during a single treadmill exercise session who underwent exercise echocardiography simultaneously with 201Tl SPECT. Most of these patients (85%) participated in a prospective study comparing the diagnostic value of exercise echocardiography with SPECT 201Tl.6 Patients were excluded if they had a recent myocardial infarction (<2 months), significant valvular heart disease, dilated or hypertrophic cardiomyopathy, or previous cardiac transplantation. This investigation was approved by the institutional review board of Baylor College of Medicine. Baseline characteristics are depicted in Table 1⇓. One hundred ninety-four patients (78%) were receiving cardiac medications consisting of calcium channel blockers in 47%, β-blockers in 37%, nitrates in 31%, and aspirin in 44%. Indications for the test were evaluation of chest pain in 35%, screening for CAD in 52%, evaluation of ischemia 2 to 3 months after acute myocardial infarction in 8%, palpitations in 3%, and heart failure in 2%. Historical data were retrieved from medical records. Pretest probability of CAD10 was low (≤25%) in 58% of patients, intermediate (26% to 69%) in 18%, and high (≥70%) in 24%. Cardiac catheterization was performed within 3 months in 84 patients, of whom 27 had single-vessel and 43 had multivessel disease.
All patients underwent a multistage treadmill exercise with the standard Bruce protocol. Criteria for interrupting the test were severe chest pain, diagnostic ST-segment shift ≥2 mm, extreme fatigue or dyspnea, systolic blood pressure >240 mm Hg, diastolic pressure >120 mm Hg, or the reaching of maximal age-predicted heart rate. ECG was diagnostic for ischemia when there was ≥1-mm flat or downsloping ST-segment depression 0.08 seconds after the J point. ST-segment depression in the presence of left bundle-branch block, ventricular hypertrophy, or digoxin was considered to be nondiagnostic.
Two-dimensional echocardiograms were performed at rest and immediately after exercise with digital technology as previously described.6 The left ventricle was divided into 16 segments,6 and wall motion was semiquantified as follows: 1=normal or hyperdynamic, 2=hypokinetic, 3=akinetic, and 4=dyskinetic. Images were interpreted by one of 2 experienced investigators without knowledge of ECG, additional diagnostic tests, or clinical outcome. Overall interpretation was as follows: normal=normal or hyperdynamic function with exercise, ischemia=new wall motion abnormality with exercise, fixed abnormality=resting wall motion abnormality without evidence of ischemia, and mixed abnormality=resting wall motion abnormality with worsening function or contiguous or distant ischemia during exercise. A wall motion score index (WMSI) was derived as the sum of the individual scores divided by the total number of segments, and ejection fraction was determined by subjective estimation or with the multiple diameter method.11
Exercise 201Tl SPECT
At peak exercise, 3 mCi 201Tl as thallous chloride was injected intravenously, and the patient was asked to exercise for an additional minute. After acquisition of the echocardiographic images, SPECT was performed by use of a large–field-of-view, single-crystal, rotating gamma camera (ADAC, ARC 3000–3300).5 Redistribution images were obtained 4 hours later. Reconstructed tomographic slices were oriented in the short, horizontal long, and vertical long axes and displayed sequentially to assess regional perfusion.5
Experienced observers unaware of the results of the echocardiography, additional tests, or clinical outcome performed the analyses. 201Tl uptake was scored as follows: 1=normal, 2=mildly reduced, 3=moderately reduced, and 4=severely reduced. Perfusion defects were analyzed for the presence of complete redistribution (ischemia), no redistribution (fixed defect), or partial redistribution (mixed defect). Computerized polar maps were generated and compared with a normal data bank.5 201Tl defect size (percent) was obtained by computer in most patients (n=228).5 In 20 patients studied in 1986 whose images could not be adequately processed, defect size was calculated as the number of exercise abnormal segments divided by the total number of segments (n=16) and expressed in percent. This semiquantitative analysis (percent abnormal segments) was also performed for all patients.
Follow-up data were obtained by review of patient records, personal communication with the patients’ physicians, and telephone interviews and were available in 225 patients (91%). The remaining 23 patients could not be contacted but had similar baseline characteristics as the 225 patients. The following outcome events were recorded: myocardial infarction, revascularization procedures (angioplasty or coronary artery bypass), congestive heart failure or unstable angina requiring hospitalization, and death (cardiac and noncardiac). Analysis was censored at the onset of the first event.
Comparison of the prognostic power of exercise echocardiography and exercise 201Tl was accomplished in 3 steps. Step 1 involved univariate testing to determine which variables distinguished patients with from those without cardiac events. These analyses used the χ2 test for association or Fisher’s exact test for categorical variables and Student’s t test or the Kruskal-Wallis test for continuous variables.12
Step 2 involved using stepwise logistic regression to develop models capable of predicting events.13 Three definitions for cardiac events were used (1) all cardiac events, (2) ischemic events (myocardial infarction, hospitalization owing to unstable angina) and cardiac death, and (3) cardiac death alone. The reason for this approach was to analyze the group of “hard events” separately from events such as revascularization procedures and congestive heart failure, which may be viewed as “soft events” but are also important in terms of morbidity and healthcare costs.
For each of the 3 definitions of events, 4 logistic models were developed. The first had as its starting point the clinical and exercise ECG variables important in univariate analyses. The second model included resting echocardiographic variables in addition to the first model. The third model included exercise echocardiographic variables in addition to clinical and exercise ECG variables. The fourth model comprised 201Tl SPECT variables in addition to clinical and exercise ECG variables. Variables that are highly correlated (eg, WMSI and ejection fraction) should not be included in the same model. In this situation, separate models with each of these variables were constructed, and the 1 with the best fit was reported. Pearson’s χ2 test was used to assess the fit of each of the models.13
Step 3 consisted of comparing the predictive ability of the models using area under the receiver-operating characteristics curve (AUC). The AUC was determined by use of the trapezoidal rule.14 AUCs were compared by the method of Hanley and McNeil.14 15 A secondary aim was to compare event-free survival of patients on the basis of the results of exercise ECG, echocardiography, or 201Tl SPECT. The Kaplan-Meier curves are shown and were compared by use of the log-rank test.12 Significance was set at P<0.05.
During follow-up (up to 8 years; mean, 3.7±2 years), 64 cardiac events occurred: 7 cardiovascular deaths, 8 nonfatal myocardial infarctions, 9 hospitalizations for unstable angina, 11 hospitalizations for heart failure, 28 revascularizations (16 coronary bypass surgery and 12 angioplasty), and 1 cardiac transplantation because of ischemic heart disease and severe heart failure. One patient suffered a noncardiovascular death but was not considered an event for the purposes of this study. Most events (59%) occurred after 90 days. Most of the “early” events (<90 days) were revascularizations with either angioplasty (n=7) or bypass surgery (n=10). In contrast, ischemic events and cardiac death accounted for most of the late events (71%). Of the 17 patients with “early” revascularization, 15 had abnormal studies by both stress echocardiography and 201Tl, and 2 were normal by both. Of the 11 patients hospitalized with heart failure, 5 (45%) had normal ejection fractions (>50%) at the time of study, and only 2 had ejection fractions <35%. Of the 11 patients, 9 had abnormal exercise echocardiography, and 8 had abnormal exercise 201Tl studies.
Clinical Parameters and Treadmill Exercise ECG
Among the clinical variables tested (Table 1⇑), smoking was the only variable associated with all cardiac events (P=0.02), with a trend observed for chest pain (P=0.06) and hypertension (P=0.08). No clinical variables were associated with ischemic events and cardiac death. Diabetes mellitus was associated with cardiac death (P=0.05), whereas old age was of borderline significance (P=0.09).
Exercise duration in the 225 patients was 8.04±2.78 minutes. Peak exercise heart rate was 146±23 bpm with maximal systolic and diastolic blood pressures of 156±29 and 77±11 mm Hg, respectively. Sixty-eight percent of patients reached 85% of their maximal age-predicted heart rate. A normal stress ECG and a higher maximal exercise heart rate characterized a group of patients with a lower risk for all cardiac events (Table 2⇓). On the other hand, an ischemic response on exercise ECG was a weak predictor of risk. There was no association between exercise ECG variables and ischemic events or cardiac death.
Exercise 201Tl SPECT
The presence of a normal or an abnormal 201Tl study discriminated patients without events from those with any cardiac event, ischemic events and cardiac death combined, and cardiac death alone (Table 2⇑). An abnormal exercise 201Tl was a predictor of all cardiac events whether the study was classified as ischemia, fixed defect, or mixed defect. Quantitative analysis of the perfusion defect size or semiquantitative analysis with percent of abnormal segments also differentiated patients with and without events in each of the 3 cardiac events groups (Table 2⇑).
Concordance between exercise echocardiography and 201Tl for normal and abnormal studies was 86% and for exact category of interpretation was 72%. The incidence of abnormal exercise echocardiography was higher in patients with any cardiac event (75%) compared with those without events (37%; P<0.0001; Table 2⇑). This incidence was also higher in patients with combined ischemic events and cardiac death (63%; P=0.018) and in those with cardiac death alone (71%; P=0.069). An abnormal exercise echocardiogram was a predictor of all cardiac events whether the study was classified as ischemia or fixed wall motion abnormality (Table 2⇑). Parameters of resting function, WMSI, and ejection fraction were predictors of all cardiac events. On the other hand, exercise WMSI and ejection fraction were significantly different in all subgroups of cardiac events, including death alone.
Patients with positive exercise ECGs for ischemia and nondiagnostic ECGs had worse survival than those with normal ECGs. These results reached statistical significance for the end points of all cardiac events and for cardiac death (P=0.02 to 0.001). For patients with normal stress ECGs, survival free from any event at 5.5 years was 75%. Figures 1⇓ and 2⇓ depict the survival analysis of the qualitative results of exercise echocardiography and 201Tl SPECT for the outcomes of all cardiac events and ischemic events or cardiac death, respectively. A significant difference was seen between patients with normal and abnormal studies for all end points, including death alone (P=0.04), for both modalities. The rates for various cardiac events in patients with normal 201Tl SPECT or normal exercise echocardiography are shown in Table 3⇓ and were comparable.
Figures 3 through 5⇓⇓⇓ depict the results with quantitative parameters. Event-free survival in patients with perfusion defect size >15% (n=62) was significantly lower than in those with smaller defects (n=163). Similarly, the 67 patients with exercise WMSI ≥1.4, corresponding to 3 akinetic segments during stress out of the 16 segments or 18%, had higher cardiac event rates compared with the 158 patients with less extensive abnormality, regardless of the type of cardiac event analyzed (Figures 3 through 5⇓⇓⇓).
All Cardiac Events
The multivariate predictors for all the models are shown in Table 4⇑. In the model of exercise 201Tl SPECT, the most significant predictor was ischemia by 201Tl SPECT; normal exercise ECG was still a negative predictor. The addition of perfusion defect size did not improve the power of the model for all events (OR, 1.00; P=0.36). For the exercise echocardiography model, exercise WMSI was the strongest predictor, followed by ischemia by echocardiography and exercise ECG.
The incremental value for the 4 models tested, expressed as AUC and global X2, is shown in Table 5⇑ and Figure 6⇓. The addition of resting echocardiography to clinical and exercise ECG improved the predictive power of the model but did not reach significance (P=0.2). However, the AUCs for the models including imaging during exercise were significantly greater (P=0.05 to 0.017), confirming a true incremental power for prediction of cardiac events for exercise echocardiography and 201Tl SPECT (Figure 6⇓). The power of the exercise echocardiography and exercise 201Tl models was comparable (P=0.4).
Ischemic Events and Cardiac Death
The first 2 models tested were not significant predictors of ischemic events and cardiac death. For the model including 201Tl SPECT, the best multivariate predictor was an abnormal 201Tl SPECT (Table 6⇑). For the model comprising stress echocardiography, an abnormal exercise echocardiography was the only predictor (Table 6⇑). The power of these 2 models was comparable (Table 5⇑).
Models including cardiac imaging during stress were the only predictors of cardiac death (Table 6⇑). In the model of 201Tl SPECT, perfusion defect size was the strongest predictor, whereas for exercise echocardiography, exercise WMSI was the only significant predictor. The power of these 2 models was also similar (Table 5⇑).
The present study is the first to evaluate the comparative long-term prognostic power of exercise echocardiography and 201Tl tomography in patients undergoing both imaging modalities during the same exercise test. Parameters of cardiac function and perfusion by the respective modalities were equally powerful in separating patients with low or high risk for cardiac events, including hard cardiac events. Although exercise ECG discriminated between patients with and without events, the addition of cardiac imaging with either echocardiography or 201Tl during exercise improved the prognostic power and was similar for both imaging modalities.
Exercise Echocardiography Versus Exercise 201Tl SPECT
Improvements in ultrasound technology and digital image capture have propelled the use of stress echocardiography in the evaluation of patients with CAD. Although exercise echocardiography evaluates ventricular function and 201Tl scintigraphy reveals relative myocardial perfusion, studies comparing both techniques showed a similar accuracy in the diagnosis of CAD.4 6 A paucity of data exists, however, on the prognostic power of exercise echocardiography. The prognostic impact of exercise echocardiography in patients with chronic CAD followed for 1 year was initially reported by Krivokapich et al.8 More recently, longer-term follow-up was reported and demonstrated that a positive stress echocardiography is a strong independent predictor of outcome.9 The present investigation extends these observations and demonstrates that the long-term prognostic power of exercise echocardiography is similar to that of exercise 201Tl SPECT. The addition of imaging by either technique provided significant incremental value over that available solely from clinical and exercise parameters for all events and particularly for spontaneous cardiac events.
The impact of systolic ventricular function on prognosis is well established. Exercise WMSI, a parameter that incorporates resting function and extent of myocardium at risk during stress, was the main predictor of events and the sole echocardiographic predictor of cardiac death by multivariate analysis. Patients with an exercise ejection fraction of <60% had a worse prognosis. It is worth noting, however, that the impact of exercise ejection fraction on prognosis was not better than that of exercise WMSI, the latter being easier to determine during stress. The present study extends earlier observations with exercise radionuclide angiography3 to exercise echocardiography, further emphasizing the importance of evaluating the extent of ventricular dysfunction during stress, in addition to detection of ischemia.
Since the initial report by Brown et al2 in 1983, several studies have demonstrated the prognostic significance of exercise 201Tl scintigraphy.7 16 17 18 In recent studies in which 201Tl SPECT was performed, the size of the perfusion defect was shown to be a powerful predictor of ischemic events and cardiac death.16 17 18 Iskandrian et al16 have shown that a perfusion defect size of >15% identified patients with a high likelihood of cardiac events. In the present investigation, similar findings were observed. Ischemia was the main multivariate predictor of all cardiac events. However, perfusion defect size successfully separated the study population into low and high risk and was the sole multivariate predictor of cardiac death. These findings further demonstrate that exercise 201Tl SPECT provides excellent risk stratification of patients with suspected CAD.
The association between exercise echocardiography and cardiac events was remarkably similar to that of 201Tl SPECT. Exercise WMSI (the counterpart of perfusion defect by 201Tl SPECT) separated the study population into low and high risk and was the best multivariate echocardiographic predictor of cardiac death. An important goal of any stress modality is the ability to identify patients with a low cardiac event rate in addition to those at high risk. The cardiac event rate in patients with normal exercise 201Tl SPECT was similar to that for exercise echocardiography, regardless of the type of outcome analyzed (Table 3⇑), and further support that the 2 modalities are comparable in risk stratification.
Most patients did not have angiography to evaluate the prognostic power of coronary anatomy. This stems from the design of the present investigation, which was to compare stress echocardiography and 201Tl SPECT in a clinical setting. However, the addition of coronary angiography to stress perfusion imaging was shown not to improve risk stratification.7 Availability of the stress test information to the clinician undoubtedly influences patient management. This may have triggered a revascularization procedure or alteration in medical management. For this reason, results were analyzed with all cardiac events and spontaneous events included. Furthermore, because this investigation is a comparative study, the influence on outcome affects the risk stratification of both modalities.
In patients evaluated for CAD, both exercise echocardiography and 201Tl SPECT significantly improve the prognostic power of exercise ECG in assessing subsequent cardiac events and provide comparable prognostic information. The choice of imaging modality in a particular institution, however, depends on several factors, including availability, feasibility, expertise, and cost considerations.
We thank Eula Landry for her assistance in preparing the manuscript.
Guest editor for this article was Richard L. Popp, MD, Stanford University Medical Center, Stanford, Calif.
- Received May 29, 1998.
- Revision received August 19, 1998.
- Accepted September 2, 1998.
- Copyright © 1998 by American Heart Association
McNeer JF, Margolis JR, Lee KL, Kisslo JA, Peter RH, Kong Y, Behar VS, Wallace AG, McCants CB, Rosati RA. The role of the exercise test in the evaluation of patients for ischemic heart disease. Circulation. 1978;57:64–70.
Jones RH, Johnson SH, Bigelow C, Pieper KS, Coleman E, Cobb FR, Pryor DB, Lee KL. Exercise radionuclide angiocardiography predicts cardiac deaths in patients with coronary artery disease. Circulation. 1991;84(suppl I):I-52–I-58.
Quiñones MA, Verani MS, Haichin RN, Mahmarian JJ, Suarez J, Zoghbi WA. Exercise echocardiography versus thallium-201 single photon emission computed tomography in the evaluation of coronary artery disease: analysis of 292 patients. Circulation. 1992;85:1026–1031.
Iskandrian AS, Verani MS. Risk assessment. In: Nuclear Cardiac Imaging: Principles and Applications. Philadelphia, Pa: FA Davis Publishers; 1996:242–278.
Quiñones MA, Waggoner AD, Reduto LA, Nelson JG, Young JB, Winters WL Jr, Ribeiro LG, Miller RR. A new, simplified and accurate method for determining ejection fraction with two-dimensional echocardiography. Circulation. 1981;64:744–753.
Rosner B. Fundamentals of Biostatistics. Belmont, Calif: Duxbury Press; 1995.
Hosmer DW, Lemeshow S. Applied Logistic Regression. New York, NY: John Wiley & Sons; 1989.
Machecourt J, Longere P, Fagret D, Vanzetto G, Wolf JE, Polidori C, Comet M, Denis B. Prognostic value of thallium-201 single-photon emission computed tomographic myocardial perfusion imaging according to extent of myocardial defect: study in 1,926 patients with follow-up at 33 months. J Am Coll Cardiol. 1994;23:1096–1106.
Marie P, Danchin N, Durand JF, Feldmann L, Grentzinger A, Olivier P, Karcher G, Juilliere Y, Virion JM, Beurrier D, Cherrier F, Bertrand A. Long-term prediction of major ischemic events by exercise thallium-201 single-photon emission computed tomography: incremental prognostic value compared with clinical, exercise testing, catheterization and radionuclide angiographic data. J Am Coll Cardiol. 1995;26:879–886.