The ECG exercise stress test (ECG-EST) has long been the reference test for the management of patients with suspected or ascertained coronary artery disease (CAD), despite its suboptimal diagnostic accuracy.1
Starting in the 1980s, several studies have reported imaging stress tests (IMG-ST), mainly scintigraphic and echocardiographic, to have a higher diagnostic accuracy for identifying/excluding obstructive CAD.2 More recently, this has also been reported for coronary CT angiography (CCTA).3 Based on these data, authoritative guidelines have recently recommended IMG-ST or CCTA as the first-choice tests for the management of patients with suspected angina, discouraging the use of ECG-EST for this scope.4,5
From this perspective, I discuss whether this recommendation is justified by current scientific evidence, briefly reviewing the main articles to which the latest European Society of Cardiology (ESC) guidelines refer to support the recommendation.4
Meta-analysis of Studies
Two meta-analyses are referenced by the ESC guidelines to recommend IMG-ST or CCTA, rather than ECG-EST, as the initial methods for CAD diagnosis.4,6,7
In the meta-analysis by Knuuti et al., ECG-EST was found to have an average sensitivity and specificity for obstructive CAD of 58% and 62%, respectively, whereas the respective values were 87% and 70% for scintigraphic stress testing, 85% and 82% for echocardiographic stress testing (ECHO-ST) and 97% and 78% for CCTA.6 However, this meta-analysis presents several limitations. First, the diagnostic performance of the methods was derived from different populations, included in different studies, making any comparison among tests questionable. Second, most studies selected to assess the diagnostic accuracy of ECG-EST compared ECG-EST with scintigraphic stress tests and were likely hampered by two typical referral biases, i.e. referral to scintigraphy laboratories to undergo a second-level (rather than initial) stress test, and referral for invasive coronary angiography (ICA) based on IMG-ST results. Third, the ECG-EST studies often enrolled specific subgroups of patients only (e.g. women, older patients or those with percutaneous coronary interventions). Fourth, ECG-EST was likely not always performed in washout from anti-ischaemic therapy. Finally, in some studies, non-standard criteria for myocardial ischaemia on ECG-EST and/or IMG-ST were applied.
In the meta-analysis by Banerjee et al., only ECG-EST and ECHO-ST studies were analysed.7 Overall, ECG-EST was found to have sensitivity and specificity for obstructive CAD of 79% and 69%, respectively, whereas the respective values for ECHO-ST were 90% and 84%. This meta-analysis had limitations similar to those of Knuuti et al., including typical referral biases, results of the two tests being derived from different populations and variability in the definition of test positivity. Moreover, most studies selected for the meta-analysis included a small number of patients (over one-quarter of the included studies comprised fewer than 50 patients).
In summary, the reliability derived from meta-analyses of the estimates of the accuracy of diagnostic tests for obstructive CAD, as well as of the true differences in their diagnostic accuracy, seems very questionable. The significant difference between the two meta-analyses in sensitivity (58% versus 79%) and specificity (62% versus 69%) of ECG-EST is a clear clue to the poor reliability of the estimates.
ECG Exercise Stress Tests Versus Imaging Stress Tests
To the best of my knowledge, there is no controlled study that directly compares the diagnostic performance of ECG-EST and scintigraphic stress tests for obstructive CAD. However, the WOMEN trial randomised 824 women with suspected CAD to ECG-EST or scintigraphic stress testing to compare clinical outcomes in patients managed according to the results of the two tests.8 At 2-year follow-up, major cardiac events (CAD death, hospitalisation for acute coronary syndrome, heart failure) occurred in 2.0% and 2.3% in the ECG-EST and scintigraphic stress test groups, respectively (p=0.59). Importantly, costs were 48% lower for ECG-EST compared with scintigraphic stress testing.
Only one randomised controlled study, including 385 patients with new stable angina, has compared ECG-EST with ECHO-ST for the diagnosis of obstructive CAD.9 Of 191 patients randomised to ECHO-ST, nine (4.5%) had a positive test and all showed obstructive CAD at ICA. Of 194 patients randomised to ECG-EST, 14 (7.2%) showed a positive test, but only nine (64%) had confirmed obstructive CAD (i.e. by ICA; p=0.04).
Unfortunately, this study had serious flaws. First, ECG-EST was considered positive not only based on typical ST-segment changes, but also on the occurrence of chest pain only, hypotension or unspecified arrhythmias. Second, ECHO-ST was considered positive even when wall motion abnormalities were present at rest only. Third, there was a very low rate (0.5%) of inconclusive ECHO-ST, suggesting selection of patients with good echocardiographic windows, but a high rate of inconclusive ECG-EST (38%), which might have biased its diagnostic accuracy. Interestingly, the only two cardiac events recorded during follow-up occurred both in patients with a negative ECHO-ST.9
ECG Exercise Stress Tests Versus Coronary CT Angiography
Three main studies are referenced in the ESC guidelines to support the superiority of CCTA over ECG-EST as the initial test for diagnosis of obstructive CAD.4,10–12
The CRESCENT trial, in fact, did not compare the diagnostic accuracy of ECG-EST and CCTA. The trial randomised 350 patients with suspected CAD to a CT scan or functional testing.10 While 95% of patients in the latter group underwent ECG-EST, only patients with coronary calcium score of 1–400 were referred for CCTA in the CT scan group. Nevertheless, no differences were found between the two groups in the rates of ICA, coronary revascularisation, diagnostic accuracy and clinical events at follow-up.
The prospective CAPP trial, instead, randomised 500 patients with suspected angina to CCTA or ECG-EST.11 The primary endpoint was the change in scores of the Seattle Angina Questionnaire domains. At 3- and 12-month follow-ups, the CCTA group showed a greater improvement in angina stability and quality of life compared with the ECG-EST group. However, this finding is unreliable as it resulted from significantly lower values of the scores at baseline in the CCTA group, whereas there were no differences between the two groups in absolute score values at both 3 and 12 months. Moreover, no differences between the two groups in major cardiac events (death, acute MI) were observed.
Finally, the SCOT-HEART trial randomised 4,146 patients referred for suspected angina to standard care (possibly including ECG-EST and/or IMG-ST) or standard care plus CCTA.12 Although CCTA significantly improved the certainty of CAD diagnosis, the primary endpoint of the study (the diagnosis of definite angina due to CAD after 6 weeks) did not significantly differ between the two groups (p=0.1289). Furthermore, the only significant difference in clinical outcome between the two groups (after 1, 5 and 10 years) was a slightly lower rate of non-fatal acute MI in the CCTA group compared with the standard care group (4.3% versus 6.0% at 10 years).13
The SCOT-HEART trial cannot be taken as proof of superiority of CCTA over ECG-EST for the initial assessment of patients with suspected CAD for two main reasons. First, it did not compare the two tests but only investigated the additional value of CCTA to standard care. Second, ECG-EST was performed as standard care in most patients (85%),12 and its results were therefore also available for clinical decisions in patients in the CCTA group. Of note, the small reduction in non-fatal acute MI deriving from the addition of CCTA to standard care appears of questionable clinical relevance and cost–benefit ratio.
Concluding Remarks
While recent authoritative guidelines propose that ECG-EST should be abandoned in favour of IMG-ST or CCTA as the initial test for the diagnosis of obstructive CAD, this recommendation does not seem to be based on clear scientific evidence.4,5
Uncontrolled studies and meta-analyses reporting the superiority of imaging methods over ECG-EST present several limitations and are significantly influenced by referral biases, making them of poor reliability. Only randomised or controlled head-to-head comparisons of the various diagnostic methods and an unbiased referral to ICA might give valid estimates of the differences in their diagnostic performance for obstructive CAD and form the base for valid recommendations.
Importantly, the recommendation of a non-invasive test as the first-choice test for CAD diagnosis should not simply be based on statistical differences in diagnostic accuracy between methods but should instead result from a balanced and comprehensive assessment of statistical results considering various other characteristics, including availability, costs, repeatability, risks and – importantly – the impact on clinical outcome.
Data derived from the few available controlled studies, in fact, show that the initial management of patients with suspected CAD based on ECG-EST is not associated with a worse outcome compared with that based on imaging methods.8,9,13
While not based on scientific evidence, the reasons why ECG-EST tends to be discarded as a first-choice method for CAD diagnosis by recent guidelines are likely multiple. They certainly include the current predominant professional/scientific/economic interests in imaging methods and the declining interest in clinical ECG and ECG methods by contemporary cardiologists.
However, in my opinion, current scientific evidence still suggests that, in light of its lower costs, easy availability and repeatability for follow-up assessment, ECG-EST remains the ideal test for the initial assessment of patients with suspected or ascertained obstructive CAD. Imaging tests should be reserved for patients who actually need second-level assessment for a correct diagnosis and management. Notably, this would also allow the allocation of imaging tests to more specific clinical indications, with a reasonable and sustainable distribution of resources.