Head-to-Head Comparison of Peak Upright Bicycle and Post-Treadmill Echocardiography in Detecting Coronary Artery Disease: A Randomized, Single-Blind Crossover Study


Post-treadmill digital echocardiography (post-TME) is the most widely used form of exercise echocardiography, but ischemia can rapidly resolve in the postexercise period; peak upright bicycle digital echocardiography (UBE) has the advantage of providing images at peak exercise that reflect normal physiology. However, the comparative accuracy of the two methods in detecting ischemia in the same patients is unknown. To compare the relative diagnostic value of peak UBE and post-TME in detecting coronary artery disease, both tests were performed in 86 consecutive patients undergoing coronary angiography.


Eighty-six patients referred for evaluation of coronary disease underwent peak UBE (starting at 25 W, with 25-W increments every 3 min) and post-TME (Bruce protocol) in a random sequence. Digitized images of peak UBE and post-TME were interpreted in a random and blinded fashion.


More transient wall motion abnormalities were detected with peak UBE than post-TME (55 vs 42, P < .001), and such exercise-induced wall motion abnormalities were more extensive (5.5 ± 3.0 vs 3.4 ± 2.1 dyskinetic segments, P < .001) and more severe (regional wall motion score index, 2.7 ± 0.5 vs 2.5 ± 0.5; P = .003). By angiography, 59 patients had coronary artery disease (a coronary stenosis of ≥50% diameter narrowing); the sensitivity of peak UBE for detecting coronary artery disease was greater than that of post-TME in the population as a whole (88% vs 66%, P < .01) and in the single-vessel subgroup (72% vs 44%, P < .05), with no worsening in specificity (89% vs 89%, P = NS).


Peak UBE is more capable of detecting ischemia than post-TME, and this is achieved with no worsening of specificity. Thus, peak UBE should be preferred in patients able to perform bicycle exercise.

Exercise is the safest, most physiologic, and most effective type of stress induced to provoke myocardial ischemia and so is better accepted by patients. Exercise increases the heart rate, blood pressure, and cardiac work more than other types of stress. The cathechol drive associated with exercise also causes constriction of atherosclerotic coronary arteries and tachycardia, which shortens diastolic perfusion time, all factors that maximize ischemia. The combination of exercise with cross-sectional echocardiography has enhanced the ability to evaluate exercise-induced ischemia, but the best exercise echocardiographic modality is not yet known.

The most common approach to exercise echocardiography is post-treadmill digital echocardiography (post-TME), which involves the acquisition of stress images in the immediate postexercise period. Its widespread use is due mainly to an easier scanning procedure (resembling that of routine echocardiography) and because walking can induce greater maximal oxygen uptake and a faster heart rate than other types of exercise, thus potentially lengthening ischemia recovery time. However, it has a number of drawbacks in clinical terms: exercise-induced wall motion abnormalities (WMAs) can normalize very rapidly after the cessation of exercise, especially if only mild exercise-induced ischemia has been induced, as in less severe disease and/or in patients on β-blockers and in the young. The fact that no intermediate scans can be acquired has a negative impact on test interpretation and on patient safety, and the time constraint after stopping exercise can be stressful for the operator and hamper adequate scanning.

These shortcomings could be overcome by peak exercise echocardiography, which allows imaging during the development of ischemia and at its peak rather than when the ischemia is already resolving. However, there are few data comparing post-TME and peak exercise echocardiography, and those that do exist are related to supine bicycle exercise, which does not appropriately reflect normal physiology, or to technically challenging procedures (such as scanning during peak treadmill exercise) that worsen test specificity. For this reason, the choice of method is currently a matter of local preference and expertise.

Peak upright bicycle digital echocardiography (UBE) reflects normal physiology and thus resembles the condition in which spontaneous ischemia occurs, and it does not require special equipment. In addition, ultrasound scanning is feasible because the thorax does not move as much during pedaling, the upright position sometimes improves image quality surprisingly, and second-harmonic and digital technology helps improve image quality in difficult patients, as well as image acquisition and review. However, peak UBE is more technically demanding because of the limited windows and the difficulty of scanning with the patient in the upright position and in motion. Moreover, because the muscle mass engaged in bicycling may be less extensive, maximum oxygen uptake and the anaerobic threshold may be lower than with a treadmill.

Because no published study has yet compared peak UBE and post-TME in the same patients, we evaluated their relative feasibility and diagnostic potential in inducing ischemia and detecting coronary artery disease (CAD) in 86 patients scheduled for coronary angiography, who underwent both tests in a study with a randomized, single-blind crossover design.


Over a 24-month period, a total of 125 consecutive patients with suspected or known CAD were recruited; only those patients without WMAs at rest (86 patients) were selected for the final analysis ( Table 1 ). No patient was excluded on the grounds of poor-quality imaging. Despite the higher efficiency of the crossover design, to guard against the exclusion of some patients, sample size (127 patients) was computed by Z test for comparison of proportions in parallel trials, hypothesizing a global diagnostic accuracy of 75% for post-TME and 90% for peak UBE, with a power of 0.80 and an α value of 0.05, on the basis of literature data.

Table 1

Characteristics of the 86 enrolled patients

Variable Value
Age (y) 57 ± 8
Men 66 (76%)
Diabetes 5 (6%)
Hypertension 34 (39%)
Previous MI/unstable angina 17 (20%)
Reason for referral
Typical angina 15 (17%)
Atypical angina 57 (66%)
Post-MI 8 (9%)
Unstable angina 6 (7%)
Antianginal drugs 42 (49%)
β-blockers 3
Other antianginal drugs 39

MI , Myocardial infarction.

Data are expressed as mean ± SD or as number (percentage).

Previous non–ST-segment elevation MI.

Asymptomatic for 72 hours before undergoing test.

At the time of exercise echocardiography.

Exclusion criteria were the persistence of at least one of the following conditions: clinical instability, severe valvular lesions, active unstable angina, New York Heart Association class III or IV symptoms, severe hypertension (systolic blood pressure > 180 mm Hg), previous episodes of ventricular tachycardia, or an inability to exercise.

Study Protocol

Each patient underwent peak UBE and post-TME on 2 consecutive days, at approximately the same time of day to take into account the circadian variation in threshold ischemia. A randomization list was generated to establish the test sequence for each subject. All patients were informed of the purpose and nature of the study and gave written informed consent to take part.

All echocardiographic examinations were performed by two experts (C.C. and D.C.) using commercially available equipment (Sonos 5500; Philips Medical Systems, Andover, MA) with a 2.5-MHz transducer; the same physician conducted both tests in individual patients.

Four resting images were acquired during both peak UBE and post-TME (in the upright position and the left lateral decubitus position, respectively) to explore the three myocardial coronary territories thoroughly. First, short-axis and long-axis parasternal views and four-chamber and two-chamber apical views were attempted; if the parasternal views were poor (most of the peak UBE cases), the three-chamber apical view and short-axis subcostal view at the midpapillary level were attempted ; in cases with very poor apical windows, the parasternal and subcostal windows were carefully explored (four-chamber and short-axis views).

Exercise Protocols

Bicycle exercise was performed using a cycle ergometer (Marquette Electronics/Hellige, Milwaukee, WI). After resting images were acquired, patients pedaled at a constant speed, starting with an initial workload of 25 W, which was increased by 25 W every 3 min; at peak stress, a sequence of images repeating those of the resting sequence was obtained. Scanning for the peak images was started when the patient had reached >85% of maximum age-predicted heart rate (if asymptomatic), presented initial signs of exhaustion, or else started to report chest pain. To optimize the examination, the patient was asked to keep the torso as motionless as possible during the recordings. Three patients who had never cycled before were successfully trained to do so, for 0.5 hours per day for 2 days before the tests.

Treadmill exercise was performed using a T2000 treadmill (Marquette Electronics/Hellige). The Bruce maximal exercise protocol was used for the treadmill test. At the end of the exercise period, the patient resumed the left lateral decubitus position, and images repeating those of the resting sequence were acquired: the elapsed time from exercise termination to the first and last image acquisition was 23 ± 8 sec (range, 7–47 sec) and 95 ± 31 sec (range, 43–126 sec), respectively.

During both tests, patients underwent 12-lead electrocardiographic (ECG) monitoring, and blood pressure was measured every 3 min during exercise, at peak stress, and 3 min later. The criteria for interrupting either exercise test were severe chest pain, an ST-segment shift of ±2 mm diagnostic for ischemia, extreme fatigue, an excessive increase in blood pressure (systolic blood pressure > 240 mm Hg, diastolic blood pressure > 120 mm Hg), limiting dyspnea, or reaching the maximum age-predicted heart rate.

For both stress echocardiographic modalities, continuous harmonic imaging acquisition was adopted. To optimize digital acquisition of the stress images and eliminate translation artifacts, the patients were taught to hold their breath for as long as possible (at least a few seconds) during acquisition. The digitized images were arranged in a quad-screen format for simultaneous comparison of the stress and resting images and stored on magnetic-optical disks for later review.

Interpretation of Exercise Echocardiograms

All exercise echocardiograms were interpreted by a single experienced observer (C.C.) who was unaware of the type of exercise, the echocardiographic methodology (all the identifiers on the digitizer’s quad screen were switched off), symptoms, and the ECG response to exercise, as well as the cardiac catheterization data. Regional ventricular function was interpreted on the basis of American Society of Echocardiography recommendations using a 16-segment left ventricular model, and wall motion was scored as follows: 1 = normal or hyperdynamic, 2 = hypokinesia, 3 = akinesia, and 4 = dyskinesia. A normal response was defined as normal or hyperdynamic function during exercise and ischemia as the development of a new WMA. The spatial extent of regional ischemia induced by either test was assessed as the total number of ischemic segments and its severity as the regional wall motion score index (i.e., the wall motion score in the ischemic segments divided by the number of ischemic segments).

Intraobserver and Interobserver Variability

For interobserver variability, a second reader (D.S.), blinded to the initial results, reread the stress echocardiographic studies (both peak UBE and post-TME) from a group of 24 patients randomly selected from the entire group of 86; the scores ascribed were compared with the original scores given by the first reader (C.C.). For intraobserver variability, the first reader who had reviewed the examinations (C.C.) reread all the stress echocardiographic studies made in the same 24-patient group a second time, ≥3 months after the first reading.

Image Quality

The technical quality of the images during each stress session was assessed using a four-point scoring system: 1 = good visualization of the endocardial edge for all segments in all four echocardiographic views; 2 = the same segment(s) not visualized in all views, provided that all 16 segments could be analyzed in at least one view; 3 = some segments not visualized at all but all coronary territories at least partially visualized; and 4 = failure to visualize one or more coronary territories.

For the comparison with coronary angiography, we used a modified version of the scheme proposed by Segar et al. in which the apical lateral, apical inferior, and basal posterior segments were considered as overlap areas. The overlap area was considered as a part of that territory containing additional WMAs.

Coronary Arteriography

Coronary angiography was performed using the standard Judkins method and the femoral approach. Any coronary stenosis was visually assessed by means of multiple projections. CAD was considered present when one or more epicardial vessels showed a ≥50% reduction in luminal diameter.

Statistical Analysis

Data are expressed as mean ± SD. Student’s t test was used to compare continuous variables, and two-way repeated-measures analysis of variance was used to compare the peak UBE and post-TME data; nonparametric tests were used when appropriate. Categorical variables were compared using χ 2 , Fisher’s exact, or McNemar’s tests. Effect-size statistics in the crosstabs procedure were assessed using Cramer’s V coefficient. Matched comparison between positive and negative predictive values of the two diagnostic tests was made according to the approach described by Bennett. Kappa statistics were used to evaluate concordance in ECG results and echocardiographic reading variability: κ > 0.5 represents moderate agreement, κ > 0.7 indicates good agreement, and κ > 0.8 indicates very good agreement according to Peat. Sensitivity, specificity, and 95% confidence intervals (CIs) were calculated in the usual manner. P values < .05 were considered significant. The data were analysed using SPSS version 19.0 for Windows (SPSS, Inc., Chicago, IL).


Coronary Angiography

All 86 patients underwent coronary angiography; 59 patients had CAD, 25 with single-vessel disease and 34 with multivessel disease. The remaining 27 patients had no significant atherosclerotic lesions. Forty-five patients had significant left anterior descending coronary artery (LAD) stenosis, 29 had left circumflex coronary artery (LCx) stenosis, and 33 had right coronary artery (RCA) stenosis. Fifteen patients with CAD (25%) were found to have intermediate-severity CAD; most (11 patients) belonged to the single-vessel subgroup. Thirteen patients with CAD had angiographically visualized intercoronary collateral vessels.

Hemodynamic Changes and Clinical and ECG Variables

There was no significant difference between treadmill and bicycle exercise in terms of the percentage of patients reaching ≥85% of their age-predicted maximal heart rate or in terms of maximum heart rate, ECG ischemia, or typical angina at maximum stress ( Table 2 ). However, maximum blood pressure and, consequently, the peak rate-pressure product, were slightly but significantly higher (mean difference, 1.027 ± 3.343; 95% CI, 301–1.352) during peak UBE ( Table 2 ). The ECG results, dichotomized as ischemic and nonischemic, show good agreement between the two exercises: 73 of 84 patients (86%) (κ = 0.72, P < .001).

Table 2

Hemodynamics, clinical, and ECG findings during treadmill and upright bicycle exercise

Variable Upright cycle ergometer Treadmill P
PMHR (%) 88 ± 11 88 ± 10 NS
Maximal HR (beats/min) 143 ± 18 143 ± 17 NS
Maximal RPP 28,466 ± 5,575 27,463 ± 5,390 .006
Maximal SBP 199 ± 28 192 ± 27 <.01
Angina 7 (8%) 11 (13%) NS
ECG ischemia 40 (46%) 46 (53%) NS

HR , Heart rate; PMHR , percentage of maximal age-predicted heart rate; RPP , rate-pressure product; SBP , systolic blood pressure.

Data are expressed as mean ± SD or as number (percentage).

Feasibility of Stress Echocardiography

Image quality was similar between peak UBE and post-TME, but the quality of the stress images was significantly poorer than that of the resting images in both cases ( Table 3 ).

Table 3

Quality of echocardiographic imaging: peak UBE versus post-TME

Score Peak UBE Post-TME
Rest Stress Rest Stress
1 61 (71%) 58 (67%) 59 (69%) 57 (66%)
2 22 (26%) 13 (15%) 22 (25%) 13 (15%)
3 2 (2%) 14 (16%) 5 (6%) 16 (19%)
4 1 (1%) 1 (1%) 0 0
Total 86 86 86 86

P < .01 versus peak UBE during stress.

P < .01 versus post-TME during stress.

Exercise-Induced WMAs

WMAs were more frequently detected during peak UBE (in 55 patients with peak UBE and in 42 with post-TME, P < .001; Figure 1 ) and were more extensive and severe, as shown by the total number of ischemic segments (peak UBE vs post-TME, 5.5 ± 3.0 vs 3.4 ± 2.1; P < .001) and regional wall motion score index (peak UBE vs post-TME, 2.7 ± 0.5 vs 2.5 ± 0.5; P = .003) in the 41 paired ischemia-positive tests ( Figure 2 , Videos 1 and 2 view video clips online).

Figure 1

Bar graph showing sensitivity (SENS), specificity (SPEC), positive predictive value (+PV), and negative predictive values (−PV) of peak UBE and post-TME in detecting CAD in the whole group and in the single-vessel and multivessel (MV) subgroups. The error bars indicate 95% CIs, and the numbers at the bottom of the bars indicate the numerator and denominator to calculate the bar value. Comparison of sensitivity and specificity was performed using McNemar’s test and of predictive values according to Bennett’s test.

Figure 2

The two quad screens are an example in the same patient of peak UBE (A) and post-TME (B) . An apical four-chamber (4 CH) view is represented at baseline and during stress. End-diastolic and end-systolic frames are reported in each stage. The exercise-induced regional mechanical dysfunction is much larger and more severe during peak UBE (A) than during post-TME (B) . During peak UBE, in fact, at peak stress, there is dyskinesia of the apex, the septum, and the middle part of the lateral wall (as indicated by the arrowheads); the increased telesystolic volume is also evident, a marker of extensive and severe ischemia very likely caused by a multivessel CAD. In real time (see Video 1 ; available at www.onlinejase.com ), the ischemic segments are seen as akinetic. On the contrary, during post-TME, there is only involvement of the distal part of the septum, which appears only mildly hypokinetic (better seen in Video 1 ), with no increment of the end-systolic volume. Coronary angiography showed a very proximal occlusion of the LAD, 80% stenosis of the LCx, and an important collateral circle. The patient was young (45 years old). It is also evident at baseline that the left ventricular volumes were slightly reduced during peak UBE, with a certain reduction of the ejection fraction as a consequence of the blood pooling that takes place in the upright position. DIAST , Diastole; SYST , systole.

Reading Variability

Among the selected patients ( n = 24), 11 were ischemic with peak UBE and six with post-TME at the first reading. Intraobserver and interobserver agreement regarding the presence or absence of ischemia was very good for both tests: intraobserver and interobserver agreement for peak UBE echocardiography was 100% (κ = 1.00, P < .001) and 96% (κ = 0.92, P < .001), respectively, and 100% (κ = 1.00, P < .001) and 96% (κ = 0.89, P < .001) for post-TME.

CAD Detection

Using angiography as the gold standard, the sensitivity of peak UBE in detecting CAD was significantly better than that of post-TME in the whole study population (88% vs 66%, P < .001) and in the single-vessel subgroup (72% vs 44%, P < .05) ( Figures 1 and 2 , Videos 1–3 view video clips online). The negative predictive value of peak UBE was also significantly better than that of post-TME in the whole study population (77% vs 54%, P < .05 ) and in the single-vessel subgroup (77% vs 63%, P < .05) ( Figure 1 ), yielding a disease prevalence of 69% (95% CI, 58%–79%). In the subgroup affected by more extensive atherosclerosis (the multivessel subgroup), the sensitivity of peak UBE tended to be greater, but the difference was not significant (100% vs 82%, P = NS; Figure 1 ).

Peak UBE was also better at correctly predicting the multivessel extension of CAD: among 34 patients with angiographically proven multivessel disease, peak UBE and post-TME detected multivessel CAD in 23 and seven patients (sensitivity, 68% [95% CI, 49%–82%] and 22% [95% CI, 9%–40%]; P < .01), respectively. In the group without multivessel disease, peak UBE and post-TME falsely detected multivessel CAD in four and two patients, respectively (specificity, 83% [95% CI, 63%–95%] and 92% [95% CI, 73%–99%]; P = NS) ( Figure 2 , Videos 1 and 2 ; available at www.onlinejase.com ).

Considering only the CAD subgroup with intermediate severity (40%–70% luminal diameter narrowing), both single vessel and multivessel, peak UBE showed a trend toward higher sensitivity than post-TME: 67% (10 of 15) and 27% (4 of 15), respectively ( P < .07).

Analyzing the diagnostic accuracy for individual vessel lesions, peak UBE was shown to be significantly better than post-TME at precisely identifying disease of both the LCx and RCA ( Figure 3 , Video 3 ; available at www.onlinejase.com ); the superiority of peak UBE was less evident, even if significant, in assessing LAD disease ( Figure 3 ).

Figure 3

Bar graph showing the sensitivity (SENS) and specificity (SPEC) of peak UBE and post-TME in detecting specific significant lesions of the epicardial coronary arteries (LAD, LCx, and RCA); the numbers at the bottom of the bars indicate the numerator and denominator of each specific sensitivity and specificity value. Comparisons of sensitivity and specificity were performed using McNemar’s test.

Analyzing the accuracy in detecting the exact region at risk (seven areas at risk by angiography: LAD, LCx, RCA, LAD + LCx, LAD + RCA, LCx + RCA, and three vessels), peak UBE rated significantly better than post-TME: peak UBE depicted the exact region at risk as assessed by angiography in 24 patients (15 multivessel, 9 single vessel) but post-TME in only nine patients (2 multivessel, 7 single vessel), yielding sensitivity of 41% (24 of 59) and 15% (9 of 59), respectively ( P < .02, McNemar’s test).

Both approaches were highly specific, with comparable positive predictive values (94% vs 93%, P = NS; Figure 1 ).

Subgroups with Concordant ECG Stress Test Results

To better understand the relative role of the type of imaging versus the type of exercise in explaining the results, a subgroup with perfectly concordant stress ECG results (73 patients), in which the impact of the type of exercise should be quite homogeneous, was analyzed. In this subgroup, peak UBE remained superior to post-TME in terms of sensitivity, with 86% sensitivity (43 of 50) compared with 70% (35 of 50) for post-TME ( P = .004) and 87% specificity (20 of 23) compared with 91% (21 of 23) ( P = NS).

Cross-Matching of Exercise and Angiographic Variables with Three CAD Subgroups Categorized by WMA Detection by Post-TME and Peak UBE

The patients with angiographic CAD who developed exercise-induced WMAs ( n = 52) were subdivided into three groups on the basis of the differences of WMA detection by post-TME with respect to peak UBE ( Table 4 ): group 1 ( n = 21), in which post-TME detected exercise-induced WMAs as equally large and severe as peak UBE; group 2 ( n = 18), in which both studies were positive for ischemia, but WMAs were less extensive with post-TME than peak UBE ( Videos 1 and 2 ); and group 3 ( n = 13), in which post-TME was negative for ischemia but peak UBE was positive ( Video 3 ).

Jun 1, 2018 | Posted by in CARDIOLOGY | Comments Off on Head-to-Head Comparison of Peak Upright Bicycle and Post-Treadmill Echocardiography in Detecting Coronary Artery Disease: A Randomized, Single-Blind Crossover Study

Full access? Get Clinical Tree

Get Clinical Tree app for offline access