Exercise electrocardiography (XECG) has been a widely used test in diagnosing and prognosticating individuals with suspected coronary artery disease (CAD) ((1),(2),3). Current guidelines recommend XECG as the first diagnostic step of suspected CAD in patients who are able to exercise (1). However, its usefulness is limited by a modest sensitivity and specificity of 68% and 77%, respectively, across a wide range of patient subsets (2).

Recently, coronary computed tomography angiography (CTA) was introduced as a novel, noninvasive approach for the evaluation of CAD. Because coronary CTA demonstrates high specificity and negative predictive value in the exclusion of CAD ((3),(4),5), it has been suggested as a potential noninvasive method to rule in or rule out obstructive CAD. Although previous studies revealed the superior diagnostic accuracy of coronary CTA compared with XECG ((6),7), to date, the prognostic value of coronary CTA has not been compared adequately with that of stress tests ((8),9). In addition, the usefulness of coronary CTA as an alternative or an adjunct to stress tests (including XECG) in the diagnostic work-up as well as risk stratification of patients with chest pain remains to be studied. Because the value of any noninvasive diagnostic strategy is determined in large part by its prognostic benefit, we thus sought to assess the prognostic value of coronary CTA in relation to XECG in patients with suspected CAD.

The initial study sample included 3,944 consecutive patients who had undergone both coronary CTA and XECG within 90 days for evaluation of suspected CAD at Severance Cardiovascular Hospital from May 2003 through April 2009 without any other cardiovascular testing. Patients were excluded who 1) were younger than 30 years (n = 63); 2) had a history of prior myocardial infarction (MI), coronary revascularization, or cardiac transplantation (n = 21); 3) had inadequate XECG (137 patients); 4) had insufficient medical records or uninterpretable coronary CTA results (n = 26); and 5) were without at least 1 of following symptoms or signs: angina, angina equivalent symptoms, or abnormal resting electrocardiography (ECG) results (n = 720). After exclusion according to the study criteria, a total of 2,977 patients remained for final analysis. The median number of days between coronary CTA and XECG was 8 days (interquartile range: 2 to 14 days). Clinical indications of coronary CTA and XECG are listed in (5). Pretest likelihood of CAD was determined based on American College of Cardiology/American Heart Association guidelines, which were modified from the literature review of Diamond and Forrester ((10),11) (5).

Clinical data were collected at the time of the index visit. Hypertension was defined by current use of antihypertensive medications or a blood pressure of 140/90 mm Hg or more. Diabetes mellitus was defined as receiving antidiabetic treatments or a fasting plasma glucose of 126 mg/dl or more. Current cigarette smoking was defined as any cigarette smoking in the past month. Dyslipidemia was defined as use of cholesterol-lowering medications or having a total serum cholesterol of 200 mg/dl or more. Institutional review committee approval and informed consent were obtained.

Data acquisition and image analysis were carried out as described previously (12). Briefly, patients without a contraindication to beta-adrenergic blocking agents (bronchial asthma, overt heart failure, and atrioventricular conduction abnormalities) and with initial heart rates higher than 65 beats/min received a single oral dose of 40 mg propranolol hydrochloride (Pranol; Dae Woong, Seoul, Korea) 1 h before coronary CTA. The patients' mean heart rate during the CT examination was 58 ± 7 beats/min (range: 34 to 110 beats/min). Two types of CT system configurations were used: 1) a 64-slice CT scanner (Sensation 64, Siemens Medical Solutions, Forchheim, Germany) using retrospective ECG gating with tube current modulation from 2003 through 2009 with the following parameters: rotation time: 330 ms, tube voltage: 100 to 120 KeV, tube current: 400 to 800 mA, and pitch factor: 0.2; and 2) a 64-row CT scanner (LightSpeed VCT XT, GE Healthcare, Milwaukee, Wisconsin) using prospectively an ECG-gated axial technique from 2008 through 2009 with the following parameters: rotation time: 350 ms, tube voltage: 100 to 120 KeV, and tube current: 300 to 900 mA. A real-time bolus-tracking technique was applied to trigger the initiation of the scan. Contrast enhancement was achieved with 75 ml iopamidol (370 mg iodine per milliliter, Iopamiro, Bracco, Milan, Italy) injected at 5 ml/s, followed by an injection of 50 ml of saline at 5 ml/s by using a power injector (Envision CT, Medrad, Indianola, Pennsylvania) via an antecubital vein.

Image reconstruction was performed on the scanner's workstation using commercially available software (Wizard, Siemens Medical Solutions, or GEAW, GE healthcare). Axial images were reconstructed retrospectively at 65% of the RR interval for each cardiac cycle. If artifacts appeared, additional data sets were obtained for various points of the cardiac cycle, and the data set with the minimum artifact was selected for further analysis. The reconstructed image data sets were transferred to an off-line workstation (Aquarius Workstation, TeraRecon, Inc., San Mateo, California) for postprocessing and analysis. Each lesion identified was examined using maximum intensity projection and multiplanar reconstruction techniques on a short axis and along multiple longitudinal axes. Lesions were classified by the maximal luminal diameter stenosis seen on any plane. Coronary CTA was evaluated by 2 experienced cardiac radiologists (Y.J.K. and B.W.C., with 6 and 9 years experience in coronary CTA, respectively), who were blinded to XECG results of each patient. In case of disagreement, a joint reading was performed to reach a consensus.

We used 3 different coronary CTA models to compare with prognostic values of XECG: 1) a binary obstructive CAD model; 2) an extent of CAD model; and 3) a severity of CAD model. Obstructive CAD was defined when coronary artery segments exhibited plaque with a luminal diameter stenosis of 50% or more. Extent of CAD was classified as the number of obstructive vessels (≥50%): no obstructive CAD (absence of obstructive CAD), 1-vessel disease (VD), 2-VD, and 3-VD. In addition, severity of CAD was classified into 4 categories according the degree of stenosis (13): normal (absence of CAD), mild (1% to 39% luminal narrowing in DS), moderate (40% to 69% luminal narrowing in DS); and severe (≥70% luminal narrowing in DS) stenosis.

A symptom-limited exercise treadmill test was performed according to the Bruce protocol (14). During exercise stress test, heart rhythm and blood pressure were recorded at rest, at the end of each stage of exercise, at peak stress, and during recovery. A 12-lead ECG was obtained every minute, and a 3-lead ECG for heart rhythm was monitored continuously. Indications for terminating the exercise test were as previously described (11). In the summary of the XECG data set (summary XECG), results of the stress test were classified as positive when the ST segment appeared horizontal or had a down-sloping depression of 1 mm or more for 60 to 80 ms after the end of the QRS complex (11). Inadequate stress tests of patients who had not reached the reference standards established for age, sex, and weight were excluded from the analysis. The detailed methods to record XECG parameters and to calculate Duke treadmill score are described in the (5).

Clinical follow-up data were obtained via review of electronic medical records and telephone contact by a dedicated physician, research nurse, or both, who were blinded to coronary CTA and XECG results. The primary endpoint was the occurrence of major adverse cardiac events (MACE), defined as cardiac death, nonfatal MI, unstable angina requiring hospitalization, and revascularization either by percutaneous coronary intervention or coronary artery bypass graft after 90 days of the index test. Coronary revascularizations occurring within 90 days after the index test were not included in the analyses to exclude any test-driven procedure from being considered as a MACE ((15),(16),17).

Discrete variables were presented as numbers (percentages), and continuous variables were expressed as mean ± SD or median with interquartile range, as appropriate. Differences between continuous variables were analyzed by analysis of variance tests, and those between categorical variables were analyzed by the chi-square test or Fisher exact test, as appropriate. Cumulative event rates as a function of time were calculated using Kaplan-Meier survival analysis for XECG results and coronary CTA-diagnosed CAD and were compared using the log-rank statistic. Five-year estimated incident MACE risk was calculated for each participant using Cox proportional hazards model. Univariate and multivariate models were calculated to identify XECG and coronary CTA predictors of outcome. From the Cox hazard models, hazard ratios (HRs) and 95% confidence intervals (CIs) were calculated.

To evaluate the discriminatory function of each model, C-statistics for the following models were calculated ((18),19): model 1, clinical risk factors (RF); model 2, RF + summary XECG; model 3, RF + summary XECG + Duke treadmill test score; model 4, RF + coronary CTA; model 5, RF + summary XECG + coronary CTA; model 6, RF + summary XECG + Duke treadmill test score + coronary CTA. p Values less than 0.05 were considered to be statistically significant. Statistical analysis was performed using SAS software version 9.2 (SAS Institute Inc., Cary, North Carolina) and R software version 2.13.1.

Overall, the study population consisted of 2,977 patients. The mean age of patients was 58 ± 10 years and 50% were male. The prevalences of clinical RF were as follows: diabetes, 16%; hypertension, 48%; dyslipidemia, 44%; and current smoking, 13%. In subset groups of the more severe CAD, patients were male, older, diabetic, and hypertensive (all p < 0.001) (Table 1). More than one-half of the study population (1,825 patients, 61%) had intermediate pretest probability of CAD, and 823 (28%) had low pretest probability of CAD. Ninety patients (3%) had high pretest probability of CAD. Detailed patient characteristics are summarized in (Table 1).

As demonstrated in (Table 2), 358 (12%) patients had positive XECG results, and 130 (4%) had equivocal XECG results among study cohort. By coronary CTA, 409 (13%) patients had obstructive CAD: 1-VD (274, 9%), 2-VD (92, 3%), or 3-VD (43, 1%). In terms of severity of CAD, patients were classified as having mild (782, 26%), moderate (372, 13%), and severe (155, 5%) CAD.

Prevalence of obstructive CAD (1-, 2-, or 3-VD) in patients with positive XECG results was higher than those with negative XECG results (28% vs. 11%, respectively, p < 0.001). In addition, prevalence of moderate (20%) and severe (16%) CAD was higher in patients with positive XECG results than in those with negative XECG results (11% and 7% respectively, all p < 0.001).

During a median follow-up of 3.34 years (interquartile range: 2.33 to 4.55), a total of 97 MACE were observed. These included 3 cardiac deaths, 6 nonfatal MI (4 ST-segment elevation MI and 2 non–ST-segment elevation MI), 8 cases of unstable angina requiring hospitalization, and 80 revascularizations after 90 days from the index test (Table 3). The overall 5-year cumulative event rate was 3.6% (95% CI: 3.0 to 4.3). Patients without obstructive CAD (no CAD or nonobstructive CAD) on coronary CTA experienced 22 MACE during the follow-up period, and the 5-year MACE rate was 0.9% (95% CI: 0.6% to 1.4%). In particular, only 2 MACE (2 nonfatal MI) were reported in patients without CAD (no stenosis) on coronary CTA during follow-up (5-year cumulative event rate: 0.1%, 95% CI: 0.03 to 0.50). These nonfatal MI occurred as a result of severe coronary spasm confirmed by invasive coronary angiogram and of essential thrombocytosis in the absence of epicardial obstructive CAD. However, 54 MACE were observed in patients with negative XECG results (5-year cumulative event rate: 2.4%, 95% CI: 1.8 to 3.1). Detailed MACE according to study results are summarized in (Table 3). In addition, a detailed description of revascularization procedures in relation to index coronary CTA findings are listed in (5). Most revascularization procedures (85%, 79 cases) targeted coronary lesions that had been detected by index coronary CTA: 63% had been obstructive (≥ 50%) and 22% were nonobstructive on index coronary CTA.

In a univariate Cox regression analysis, positive XECG results, detailed XECG subsets including exercise time (minutes), percentage of age-predicted heart rate, exercise capacity (metabolic equivalent), and intermediate (−10 to 4) and high (≤−11) Duke treadmill scores were predictors of MACE (all p < 0.05). Compared with patients without CAD, those with obstructive CAD including 1-VD, 2-VD, and 3-VD had significantly higher risk of MACE (all p < 0.001). For CAD severity model by coronary CTA, we used a combined category of normal and mild stenosis as a reference subset for Cox regression and C-statistics analysis, because the MACE rate was extremely low in patients in the normal category. Patients with moderate and severe CAD experienced more MACE compared with patients with normal and mild stenosis (all p < 0.001) (Table 4).

We also explored the prognostic value of coronary CTA in relation to XECG. Patients with normal or mild stenosis determined by coronary CTA had an excellent prognosis independent of the XECG results. The Kaplan-Meier 5-year event rate of patients with negative XECG results and normal or mild stenosis and patients with positive XECG results and normal to mild CAD were 0.8% (95% CI: 0.4 to 1.2) and 1.0% (95% CI: 0.2% to 3.6%), respectively. Compared with patients with negative XECG results and normal to mild stenosis, patients with all other combinations of XECG and coronary CTA models had a significantly increased risk of MACE (all p < 0.001), except for patients with positive XECG results and normal to mild CAD (p = 0.616)_.

In multivariate Cox regression analysis adjusted by age, sex, hypertension, diabetes, current smoking, and dyslipidemia, positive XECG results (HR: 4.53, 95% CI: 2.96 to 6.94, p < 0.001) and detailed XECG subsets, including exercise time (HR: 0.86, 95% CI: 0.77 to 0.95, p = 0.003), percentage of age-predicted maximal heart rate (HR: 0.98, 95% CI: 0.97 to 0.99, p = 0.042), and exercise capacity (HR: 0.85, 95% CI: 0.77 to 0.94, p = 0.001) were associated with future MACE. In addition, the intermediate-risk group (HR: 1.72, 95% CI: 1.13 to 2.63, p = 0.012) to high-risk group (HR: 10.51, 95% CI: 3.20 to 34.49, p < 0.001), based on Duke treadmill score, had a higher risk of MACE compared with the low-risk group.

By coronary CTA, as compared with patients without CAD, the relative HR for MACE increased proportionally to CAD extent for obstructive 1-VD (HR: 16.46, 95% CI: 9.30 to 29.12, p < 0.001), 2-VD (HR: 24.52, 95% CI: 12.55 to 47.90, p < 0.001), and 3-VD (HR: 38.20, 95% CI: 18.37 to 79.44, p < 0.001). Similarly, as compared with patients with normal to mild stenosis, those with moderate stenosis (HR: 14.10, 95% CI: 7.41 to 26.84, p < 0.001) and severe stenosis (HR: 44.86, 95% CI: 23.79 to 84.60, p < 0.001) experienced proportionally an increased risk of MACE (Table 4).

Compared with patients with negative XECG results and normal or mild stenosis, patients with all other combinations of XECG and coronary CTA models had a significantly increased risk of MACE, with the exception of patients with positive XECG results and normal to mild CAD (HR: 1.51, 95% CI: 0.34 to 6.78, p = 0.587)_. In particular, subsets of patients exhibiting a high risk of MACE (listed in increasing risk) included: 1) patients with negative XECG results and moderate CAD (HR: 11.59, 95% CI: 5.33 to 24.27, p < 0.001); 2) patients with positive XECG results and moderate CAD (HR: 30.51, 95% CI: 13.67 to 68.08, p < 0.001); 3) patients with negative XECG results and severe CAD (HR: 32.44, 95% CI: 15.28 to 68.89, p < 0.001); and 4) patients with positive XECG results and severe CAD (HR: 79.33, 95% CI: 37.28 to 168.82, p < 0.001). Risk-adjusted overall MACE-free survival according to XECG subset (positive vs. negative results), coronary CTA models, including extent and severity of CAD, and combined models of XECG and coronary CTA are plotted in (Figure 81_gr1).

As described in (Table 5), compared with clinical RF alone (C-statistics: 0.746, 95% CI: 0.695 to 0.796), summary XECG (C-statistics: 0.790, 95% CI: 0.742 to 0.839), summary XECG plus Duke treadmill score (C-statistics: 0.789, 95% CI: 0.741 to 0.838), and severity of CAD on coronary CTA (C-statistics: 0.909, 95% CI: 0.883 to 0.936) had an added benefit for prediction of future MACE (all p < 0.001 for differences). Compared with the summary XECG model (model 2), an addition of Duke treadmill score did not show increment in C-statistics (p = 0.707 for difference). In contrast, the addition of coronary CTA findings to the summary XECG model (model 5, C-statistics: 0.908, 95% CI: 0.890 to 0.937) and the summary XECG results plus Duke treadmill score (model 6, C-statistics: 0.909, 95% CI: 0.880 to 0.938) showed significant increment in C-statistics compared with their baseline models without coronary CTA findings (models 2 and 3, respectively, all p < 0.001 for difference). Compared with the baseline coronary CTA model (model 4), however, an addition of summary XECG results (model 5) or summary XECG plus Duke treadmill score (model 6) failed to discriminate future risk of MACE (p = 0.638 and p = 0.545 for difference, respectively).

The discriminatory function of other coronary CTA models, including the binary CAD model (presence or absence of obstructive CAD) and extent of CAD (number of obstructive stenosis vessels), also were analyzed (5). The C-statistics of binary CAD model (0.885, 95% CI: 0.851 to 0.920) and extent of CAD (C-statistics: 0.888, 95% CI: 0.853 to 0.922) were slightly lower than those of the severity of CAD model, but discriminatory function compared with clinical RF and XECG results was similar.

To determine predicted power of coronary CTA in relation to XECG, we also examined the predictive value of coronary CTA in a subgroup categorized by XECG and the predictive value of XECG in a subgroup categorized by severity of CAD determined by coronary CTA (Figure 81_gr2). Severity of CAD determined by coronary CTA successfully stratified future risk of MACE in the group with positive XECG results as well as the group with negative XECG results (all p < 0.001 for trend).

However, XECG failed to stratify future risk of MACE in the group with normal to mild stenosis on coronary CTA (compared with negative XECG results, HR: 1.28, 95% CI: 0.28 to 5.82, p = 0.746). Abnormality of XECG results predicted future risk of MACE only in the subgroups of patients with moderate stenosis (compared with negative XECG results, HR: 2.58, 95% CI: 1.29 to 5.19, p = 0.008) and severe stenosis (compared with negative XECG results, HR: 2.28, 95% CI: 1.19 to 4.38, p = 0.013).

Our study aimed to investigate the roles of coronary CTA and XECG in the risk stratification of patients undergoing initial evaluation with suspected CAD. The main finding of this study is that, in a large number of patients with suspected CAD, both XECG and coronary CTA are independently predictive of MACE, but there is no overall incremental benefit for predicting future MACE when XECG is added to coronary CTA. In subgroup analyses, coronary CTA stratifies risk of future MACE in both XECG-negative and XECG-positive groups. However, XECG stratifies future risk of MACE only in coronary CTA subgroups of moderate or severe stenoses.

Because of its widespread availability, lower cost, and simplicity of its operation and interpretation, XECG has been recommended as the initial diagnostic test for suspected CAD patients with a normal resting ECG results who are able to exercise (14). However, considerable evidence exists that demonstrates the limitations of XECG as a diagnostic method because of its relatively low sensitivity and specificity ((20),21). Furthermore, it has been reported that there are some limitations regarding the use of XECG to identify high-risk patients in the evaluation of suspected CAD (22).

Coronary CTA has emerged as a novel diagnostic tool with high sensitivity and specificity that demonstrated superiority to XECG in detecting CAD in previous head-to-head comparisons (23). In recent 64-slice multidetector coronary CTA studies, the negative predictive value reached 99%, which is higher than that of any other noninvasive imaging techniques (24). Accordingly, it has been suggested that coronary CTA could be used as a first-line imaging technique to exclude CAD and to replace invasive coronary angiography in some patients (25). Prognosis prediction of patients suspected of having CAD is as important as detecting disease, because it determines the subsequent treatment plan. Recent studies have shown the potential prognostic value of coronary CTA in suspected CAD patients ((26),27). However, to date, it remains unclear whether coronary CTA adds prognostic usefulness beyond standard exercise test findings in patients who are classified according to XECG results, as well as whether XECG adds incremental prognostic value to coronary CTA.

To our knowledge, this is the first registry analysis to suggest that coronary CTA interpretations of CAD improve classification above and beyond XECG findings in a large cohort with suspected CAD. The current study demonstrated an incremental prognostic value of coronary CTA on XECG findings in the overall population (Table 5). In subgroup analyses, coronary CTA stratifies future risk of MACE in groups with both negative and positive XECG results ((Figure 81_gr2)A and Figure 81_gr2B). Therefore, coronary CTA could be used to stratify future risk of MACE in patients with both negative and positive XECG results. In contrast, the added prognostic benefit of XECG to coronary CTA models is not observed in the overall population (Table 5). In subgroup analyses, positive XECG findings are associated significantly with future risk of MACE in patients with moderate to severe stenosis ((Figure 81_gr2)D and Figure 81_gr2E), but not in patients with normal to mild stenosis (Figure 81_gr2C). Thus, XECG can be of value only in patients with moderate to severe stenosis on coronary CTA. The clinical implications of the current study are summarized in (5).

Furthermore, the current study also generated the hypothesis that coronary CTA may be used as a first-line test in patients with suspected CAD in lieu of XECG. There was a proportional increase in future risk of MACE according to the extent and severity of CAD detected by coronary CTA. In particular, coronary CTA demonstrated its clinical usefulness by identifying patients without significant stenosis who had an excellent prognosis: the 5-year cumulative event rate of patients with normal to mild stenosis on coronary CTA was less than 1.0%. Therefore, we can identify patients with excellent prognosis accurately by coronary CTA, so that unnecessary additional testing can be avoided in this population. In our present cohort, among 2,977 symptomatic patients, coronary CTA identified 2,568 (86%) patients without obstructive CAD (≥50%) and 2,450 (82%) patients without moderate to severe stenosis (≥40%) to be at low risk (Table 3). On the contrary, a non-negligible proportion of patients with negative XECG results are found to have moderate to severe stenosis on coronary CTA (15%) and to experience higher 5-year cumulative event rates of 2.4% (95% CI, 1.8% to 3.1%) compared with those patients without obstructive CAD on coronary CTA. These findings emphasize an important limitation of nonimaging exercise-induced ST-segment depression for risk stratification, as has been reported in prior studies (28).

However, despite the powerful prognostic value, coronary CTA has potential limitations, including radiation hazard, use of iodinated contrast agents, and higher cost compared with that of XECG. In addition, most cardiac events in the current study were revascularization procedures, so the prognostic value of coronary CTA to predict a so-called hard event was not evaluated fully. Moreover, the cumulative events rate of 2.4% in patients with negative XECG results is a fairly good prognosis, although it is higher than the cumulative events rate in patients without obstructive CAD detected by coronary CTA. Therefore, the decision to use coronary CTA as a first-line diagnostic method in patients with suspected CAD should be deferred until the potential risk-to-benefit ratio, cost effectiveness, and clinical efficiency are weighed by prospective randomized trials. Further, future comprehensive studies also are warranted to determine the most cost-effective, safe, and clinically efficient strategy to diagnose CAD and to predict future risk of cardiac events in low- to intermediate-risk symptomatic patients without known CAD covering various diagnostic methods, such as coronary CTA, XECG, myocardial perfusion imaging, and stress echocardiography.

The present study was retrospective and may have been influenced by unobserved confounders and selection or referral biases, or both. In addition, the effect of post-test medical treatments or risk factor control was not considered. Especially given the potential advantage of coronary CTA to identify nonobstructive CAD for prediction of future cardiac events (29), further studies are warranted to assess the impact of medical therapies in patients with nonobstructive CAD on coronary CTA.

Although we considered only revascularizations more than 90 days after coronary CTA as outcome events, revascularizations from the index test may have been included. Moreover, the pretest probability of the study population was relatively low, which limits the number of clinical events at follow-up. However, to our knowledge, the population size of this study is the largest to date reporting concurrent XECG and coronary CTA findings in relation to downstream clinical outcomes, and we plan to continue to follow through with our investigation to understand the long-term nature of these study findings.

In patients with suspected but without known CAD, coronary CTA demonstrates added prognostic benefit in patients with both positive and negative XECG results. In particular, the clinical usefulness of coronary CTA is realized by identifying patients with normal or mild stenosis (<40%) and accurately predicting the very low risk of future cardiac events. Conversely, XECG has additive value for risk stratification on coronary CTA only in patients with moderate to severe stenosis.