Article Text
PDF
Abstract
Objective To evaluate diagnostic accuracy of exercise ECG (ex-ECG) versus 64-slice CT coronary angiography (CT-CA) for the detection of significant coronary artery stenosis in a population with low-to-intermediate pre-test likelihood of coronary artery disease (CAD).
Design Retrospective single centre.
Setting Tertiary academic hospital.
Patients 177 consecutive patients (88 men, 89 women, mean age 53.5±7.6 years) with chest pain and low-to-intermediate pre-test likelihood of CAD were retrospectively enrolled.
Interventions All patients underwent ex-ECG, CT-CA and invasive coronary angiography (ICA).
Main outcome measure A lumen diameter reduction of ≥50% was considered as significant stenosis for CT-CA. Ex-ECG was classified as positive, negative or non-diagnostic. Results were compared with ICA. Diagnostic accuracy of CT-CA and ex-ECG was calculated using ICA as the reference standard. A parallel comparative analysis using a cut-off value of 70% for significant lumen reduction was also performed too.
Results ICA disclosed an absence of significant stenosis (≥50% luminal narrowing) in 85.3% (151/177) patients, single-vessel disease in 9.0% (16/177) patients and multivessel disease in 5.6% (10/177) patients. Prevalence of obstructive disease at ICA was 14.7% (26/177). Sensitivity, specificity, positive and negative predictive values at the patient level were 100.0%, 98.7%, 92.9%, 100%, respectively, for CT-CA and 46.2%, 16.6%, 8.7%, 64.1%, respectively, for ex-ECG. Agreement between CT-CA and ex-ECG was 20.9%. CT-CA performed equally well in men and women, while ex-ECG had a better performance in men. After considering the cut-off value of 70% for significant stenosis, the difference between CT-CA and ex-ECG remained significant (p<0.01), with a low agreement (21.5%).
Conclusions CT-CA provides optimal diagnostic performance in patients with atypical chest pain and low-to-intermediate risk of CAD. Ex-ECG has poor diagnostic accuracy in this population. Concerns are related to risk of radiation dose versus the benefits of correct disease stratification.
- CT coronary angiography
- invasive coronary angiography
- clinical value
- exercise ECG
- low–intermediate pre-test likelihood
- atypical chest pain
- CT scanning, exercise testing, coronary artery disease (CAD)
Statistics from Altmetric.com
- CT coronary angiography
- invasive coronary angiography
- clinical value
- exercise ECG
- low–intermediate pre-test likelihood
- atypical chest pain
- CT scanning, exercise testing, coronary artery disease (CAD)
Introduction
Sixty-four-slice CT coronary angiography (CT-CA) is a valuable tool for the non-invasive detection of significant coronary artery disease (CAD). Several studies have demonstrated a very good diagnostic accuracy of CT-CA using different scanner generations.1–4 A major advantage of CT-CA is the very high negative predictive value, which reliably permits the exclusion of CAD.1–4
Exercise electrocardiography (ex-ECG) is the most commonly applied non-invasive test for the detection of inducible ischaemia: it is a baseline tool, easy to perform, cheap, and unlike CT, it does not require radiation exposure and contrast agent administration.5 As yet, relatively few studies have been performed comparing CT-CA with other non-invasive diagnostic modalities. Although CT-CA has been shown to perform at least as well as stress single-photon emission CT and better than ex-ECG6 7 for detection of coronary artery stenosis, most of these studies have been performed in patients with an intermediate-to-high pre-test likelihood of CAD. Based on recent recommendations and guidelines CT-CA should also be performed in patients with low-to-intermediate pre-test likelihood of disease.8 9 However, validation data are very poor in this patient group.10
The aim of this study was to compare the diagnostic accuracy of CT-CA with ex-ECG in a population with atypical chest pain and a low-to-intermediate pre-test likelihood of CAD with invasive coronary angiography (ICA) as the reference standard test.
Materials and methods
Over a period of 6 months we retrospectively enrolled 177 consecutive patients (88 men and 89 women; mean age±SD 53.5±7.6 years, range 21–69 years) with chest pain and a low-to-intermediate pre-test probability of disease. All patients underwent ex-ECG, CT-CA and ICA. During the enrolment period we performed 545 CT-CA and 894 ex-ECG in our hospital. The probability of disease was calculated taking into account age, gender and chest pain characteristics using the Diamond–Forrester tabular method as recommended by the American Heart Association/American College of Cardiology exercise test guidelines.11
Men aged 30–39 years with typical angina, men and women of all ages with atypical angina and women aged 30–59 years with typical angina were included.
The demographic characteristics and estimates of coronary disease pre-test probability of the patients are presented, respectively, in table 1 and table 2.
Demographics
Estimates of coronary disease pre-test probability of overall population by symptoms, gender and age as recommended by AHA/ACC exercise test guidelines11
All patients underwent ex-ECG and subsequently CT-CA and ICA. Patients in sinus heart rhythm who were able to hold their breath for 15 s and for whom administration of iodinated contrast agents was not contraindicated were included. Exclusion criteria were previous bypass graft surgery or percutaneous coronary intervention with stent implantation. The study was retrospective and local medical ethical committee approval was waived. All patients gave written informed consent to undergo CT-CA.
Pre-scan patient preparation
Patients with pre-scan heart rates >65 bpm received, unless contraindicated (ie, asthma, aortic valve stenosis, left ventricular ejection fraction <30%), a single intravenous dose of 5 mg atenolol 10 min before the scan under ECG and blood pressure control. Immediately before the scan we administered sublingual isosorbide dinitrates (0.3 mg) to all patients.
CT scan protocol
All patients underwent CT coronary angiography using a 64-slice CT scanner (Sensation 64, Siemens, Forchheim, Germany). A bolus of 100 ml contrast material (iomeprol, Iomeron 400, Bracco, Milan, Italy) was injected into an antecubital vein at high flow rate (5 ml/s). A bolus tracking technique was used to determine the start of CT data acquisition.
The radiation exposure associated was 679 mGy (range 509–1045) corresponding to an effective dose of 11.6 mSv (range 8.7–17.8).
CT image reconstruction
Retrospective ECG gating was used for the reconstruction of images. Datasets were reconstructed during the mid-to-end diastolic phase, with reconstruction windows set at –300 to –450 ms before the next R-wave or 60–70% of the R–R interval. If images were of insufficient quality, additional reconstruction during the end-systolic phase (25–35% of the R–R interval) was performed. The reconstructed slice thickness was 0.75 mm with an increment of 0.4 mm. All CT datasets were reconstructed using a medium-soft convolution kernel.
The CT scans were evaluated by two experienced readers in consensus. The readers were both blinded to the results of conventional coronary angiography.
Maximum intensity projections were used to provide an anatomical overview of the coronary arteries and to identify side branches. Multiplanar reformatted images were used to classify lesions as significant or non-significant using a threshold of 50% (diameter stenosis), which is the reference standard at our institution. Furthermore, a parallel analysis was performed for the presence of luminal stenosis ≥70%.
Exercise ECG (ex-ECG)
Ex-ECG was performed according to current guidelines with the patient in a semi-upright position on a bicycle ergometer.12 Before stress testing, the 12-channel resting ECG was analysed. The ECG trace was monitored continuously throughout the test, and blood pressure was measured at every load increment stage. Criteria for discontinuation of exercise testing were a systolic and diastolic blood pressure of >230 and >130 mm Hg, respectively, horizontal and downsloping ST-segment depressions of >0.3 mV or ST-segment elevations of >0.1 mV, ventricular tachycardia persisting for >30 s, increasing number or polytopic premature beats, disturbed intraventricular or atrioventricular transition, exhaustion of the patient, severe dyspnoea and angina pectoris. The target heart rate was determined according to the rate–pressure product formula. After discontinuation, ECG recording and blood pressure measurements were continued until values returned to baseline levels. The stress test response was considered ‘positive’ when at least 0.1 mV horizontal or downsloping ST-segment depression or elevation was measured at least 80 ms or 60 ms after the J point during exercise or recovery or typical angina pectoris occurred. For upsloping ST-segment depression, a change of 0.15 mV noted at least 80 ms after the J point was classified as a diagnostically relevant change. The stress test response was considered non-diagnostic in patients who did not reach the reference exercise capacity normalised for age, gender and weight. For the assessment of diagnostic accuracy non-diagnostic ex-ECG examinations were considered as positive and also analysed separately.
Quantitative coronary angiography
Invasive coronary angiography was performed using standard techniques through transfemoral or radial access. A single observer blinded to the results of CT-CA performed a quantitative evaluation of the coronary lumen, measured as the percentage stenosis of all diseased coronary segments, using dedicated software (CAAS, Pie Medical, Maastricht, The Netherlands). Stenosis was classified as significant for a mean lumen diameter reduction of ≥50% in two orthogonal projections. All coronary segments were included for comparison with CT-CA. Coronary segments were classified according to the 16-segment model of the American Heart Association.13 Furthermore, a parallel comparative analysis using a threshold of 70% for significant luminal stenosis was performed.
Statistics
Continuous variables are expressed as mean values±SD. Differences between groups were compared using the Student t and χ2 tests, as appropriate. The diagnostic accuracy of CT-CA for the detection of significant coronary lesions was evaluated using quantitative coronary angiography as the reference standard. Sensitivity, specificity, positive and negative predictive values, as well as positive (sensitivity/(1–specificity)) and negative ((1−sensitivity)/specificity)) likelihood ratios with the corresponding 95% central CIs (calculated with binomial expansion) were calculated.
Diagnostic accuracy was calculated on a per-segment and per-patient (presence or absence of any lesions in each patient) basis. Diagnostic accuracies were compared using the χ2 test.
All statistical tests were performed for the entire population and for the subpopulation with positive or negative ex-ECG (excluding non-diagnostic ex-ECG examinations).
Results
Characteristics of the population
Among the overall population, 151 patients (85%) had an intermediate pre-test probability, and 26 (15%) patients had a low/very low pre-test probability (table 2).
All patients but one with low/very low pre-test probability were female with statistically significant difference between gender groups (p<0.0001).
Prevalence of disease
Overall, 49 significant lesions (≥50% luminal narrowing) were detected by ICA determining a prevalence of 0.28 lesions per patient (prevalence of disease 15%; 26/177). ICA showed no significant stenosis in 151 (85%) patients, single-vessel disease in 16 (9%) patients, two-vessel disease in 9 (5%) patients and triple-vessel disease in 1 (0.6%) patient (table 3). A left main lesion was found in one patient.
Distribution of coronary artery disease (≥50% luminal narrowing) as detected by conventional coronary angiography
After considering 70% as the cut-off value for significant stenosis, 17 significant lesions were detected by ICA—that is, a prevalence of 0.1 lesions per patient (prevalence of disease 7%; 12/177).
Overall 13% (23/177) of the patients were treated; 91% (21/23) underwent PCI and 9% (2/23) underwent bypass graft. Three patients with obstructive lesions >50% did not undergo any revascularisation.
CT-CA results
Intravenous atenolol was administered to 92.1% (163/177) of the patients. As a result, the average heart rate during the scan was 61.4±7.6 bpm versus 70.5±9.4 bpm before pharmacological preparation. The average scan time was 11±1 s (range 9–13). No significant adverse reactions to contrast material were recorded.
A total of 2607 coronary segments were available for assessment (out of a total of 2716; 4% were not assessable). All (49/49) cases of significant coronary stenosis (≥50% luminal narrowing) were correctly identified by CT-CA. Twenty segments were judged false positive on CT-CA compared with ICA. The reasons for overestimation were segmental calcifications (12/20) and motion artefacts (8/20). All patients (26/26) with at least one significant stenosis were correctly identified by CT-CA.
After using the cut-off value of 70% for significant stenosis, CT-CA correctly detected significant CAD in all coronary lesions (17/17) and in all patients (12/12) with at least one significant stenosis. Five segments were judged false positive on CT-CA owing to extensive segmental calcifications.
Ex-ECG results
Forty-nine (28%) ex-ECG examinations were classified as non-diagnostic, 39 (22%) were classified as negative and 89 (50%) were classified as positive. No significant adverse events were recorded.
Diagnostic accuracy
The sensitivity, specificity, positive predictive value and negative predictive value of CT-CA and ex-ECG for the detection of significant CAD (≥50% luminal stenosis) are presented in table 4 and figures 1 and 2. On a per-patient basis, in the entire population sensitivity, specificity, positive and negative predictive values of ex-ECG for the detection of significant CAD were 46.2%, 16.6%, 8.7% and 64.1%, respectively. For CT-CA the corresponding values were 100%, 98.7%, 92.9% and 100%, respectively.
Diagnostic accuracy of CT-CA and ex-ECG versus ICA using a cut-off value of 50% for significant lumen reduction
Bar graph of diagnostic accuracy. The bar graph shows diagnostic accuracy for CT-CA and ex-ECG in comparison with conventional coronary angiography using a cut-off value of 50% for significant lumen reduction. CT-CA is displayed with per-segment and per-patient analysis, while ex-ECG is displayed with per-patient analysis. CT-CA, CT coronary angiography; ex-ECG, exercise ECG; NPV, negative predictive value; PPV, positive predictive value.
Graphs of conditional probabilities. The graph shows the conditional probabilities for CT-CA and ex-ECG as the relationship between pre-test and post-test probability of disease. Arrowheads highlight the points corresponding to the prevalence of disease (14.7%) for the test positive (arrowhead on dark grey line) and test negative (arrowhead on the light grey line). CT-CA performs very nicely for the entire population and for the subgroup without non-diagnostic ex-ECG. Ex-ECG, instead, shows a very poor diagnostic performance in both populations. CT-CA, CT coronary angiography; ex-ECG, exercise ECG; All, entire population; w/o non-diagnostic ex-ECG, population excluding non-diagnostic ex-ECG.
The difference between CT-CA and ex-ECG was significant (p<0.01) and agreement was low (20.9%). CT-CA resulted in two patients being considered false positive for CAD. Conversely, ex-ECG led to 126 patients being considered false positive and 14 patients being considered false negative.
When we excluded patients with non-diagnostic ex-ECG examinations from the entire population, the performance of CT-CA was not significantly modified (sensitivity, specificity, positive and negative predictive values of 100%, 98.1%, 92.6% and 100%, respectively) while the performance of ex-ECG was slightly improved (sensitivity, specificity, positive and negative predictive values of 44%, 24.3%, 12.4% and 64.1%, respectively). The difference between CT-CA and ex-ECG was significant (p<0.01) and the agreement remained low (28.1%). CT-CA assessed two patients as false positives while ex-ECG determined 78 false positives and 14 false negatives.
The sensitivity and a negative predictive value of CT-CA in the subgroup of patients with non-diagnostic ex-ECG examinations (49/177) was 100%, with a prevalence of disease of 2% (1/49).
In a parallel model for diagnostic accuracy using the threshold of 70% for significant luminal stenosis, table 5, on a per-patient basis in the entire population sensitivity, specificity, positive and negative predictive values of CT-CA and ex-ECG for the detection of significant CAD were 100%, 99.4%, 92.3%, 100% and 41.7%, 19.4%, 3.6%, 82.1%, respectively. Notably, the negative predictive value of ex-ECG moderately improved owing to a halving of the false-negative rate (and of prevalence of significant CAD), whereas sensibility and specificity remained stable; however, the difference between CT-CA and ex-ECG remained significant (p<0.01), and the agreement remained poor (21.5%). Even after excluding patients with non-diagnostic ex-ECG, difference in diagnostic accuracy remained significant (p<0.01) with a low agreement (28.9%).
Diagnostic Accuracy of CT-CA and ex-ECG versus ICA using a cut-off value of 70% for significant lumen reduction
Gender influence on diagnostic performance
In a subanalysis performed on gender (ie, male vs female), we observed that CT-CA had comparable sensitivity (100% vs 100%), specificity (97.2% vs100%), positive (88.9% vs 100%) and negative (100% vs 100%) value (p>0.05; prevalence 18.2% (16/88) in men and 11.2% (10/89) in women). For ex-ECG instead we observed higher sensitivity (53% vs 33%), specificity (35% vs 1%), positive (16% vs 4%) and negative (76% vs 14%) values for men (p<0.05).
Excluding non-diagnostic ex-ECG from the analysis the subanalysis within gender (86 male; 42 women) showed for CT-CA comparable sensitivity (100% vs 100%), specificity (97.2% vs 100%), positive (88.2% vs 100%) and negative (100% vs 100%) values. For ex-ECG we observed again a higher sensitivity (50% vs 33%), specificity (36% vs 3%), positive (15% vs 9%) and negative (76% vs 14%) values for men (p<0.05).
Discussion
Exercise ECG is widely used as a baseline tool for the assessment of inducible ischaemia. The diagnostic value of this modality is mainly validated in patients with typical symptoms (ie, stable angina/chronic chest pain) and intermediate-to-high likelihood of CAD (11). However, for practical reasons, ex-ECG is extensively used as a screening tool for all types of patient ranging from high-risk to asymptomatic athletes.14 This wide application of the test is questionable since previous studies have already shown that the lower the pre-test likelihood of CAD, the lower the diagnostic performance of ex-ECG.15 16 In addition, the accuracy of ex-ECG is reduced in older patients because they are frequently unable to achieve their maximum predicted cardiac workload owing to comorbid states or to medical treatment.17 18 Widespread agreement exists about the reduction of ex-ECG diagnostic power for detection of underlying CAD in patients taking β-blocking agents owing to the influence of such agents on both ventricular function and myocardial oxygen demand.19 20 Moreover, is not unusual to see women presenting with atypical chest pain associated with non-specific ECG changes and totally normal coronary arteries.21 These limitations result in a high rate of patients with a non-obstructed coronary artery tree undergoing unnecessary ICA procedures. Depending on countries or investigating centres, other stress tests can be performed (ie, stress SPECT, stress echocardiography, stress MR). However, the rate of non-obstructed coronary artery trees remains high (30–50%).22 The large number of unnecessary ICA procedures can be explained by technical limitations and inappropriate use of stress testing.
CT-CA has been widely validated for the detection of significant coronary artery stenosis.2–4 However, most of the validation literature focuses on high risk/high pre-test likelihood patients. This reflects the ethical need to enrol patients who would undergo ICA anyway in order to compare CT-CA with the reference ‘gold standard’ procedure. However, recommendations and guidelines state that CT-CA would be more appropriate in patients at low-to-intermediate risk because of the high negative predictive value of the modality.8 9
In our study CT-CA showed very high sensitivity and negative predictive value (100% for both), as previously reported.1 23 Conversely, ex-ECG performed very poorly with a sensitivity of 44–46% and a negative predictive value of 64%. These results suggest that ex-ECG is an inadequate diagnostic modality to rule out CAD ≥50% for patients with chest pain and low risk/pre-test probability of disease. On the other hand, the high radiation exposure (15–20 mSv) does not allow us to propose CT-CA as the best substitute. In addition, the results of CT-CA were not affected by gender while ex-ECG performed better in men than in women.
The performance of ex-ECG in our study is low as compared with previous reports both in the overall population and even more so in women. This is probably the effect of a selection bias due to the retrospective design of the study.
It should be mentioned that the angiographic threshold of 50% lumen reduction for defining significant stenosis is suboptimal and was adopted based on CT-CA studies. It is of note that some intermediate stenosis in the range 50–70% on the conventional angiogram may have been haemodynamically non-significant (ie, not all cases of ≥50% stenosis determine significant reduction of coronary flow reserve),24 and that a negative stress test in these patients may be incorrectly classified as false negative. This could partially explain the moderate improvement of the negative predictive value of ex-ECG to 82% after raising the threshold for significant luminal stenosis to 70%, thereby emphasising that the cut-off values of anatomy and function do not always result in concordant outcomes. For this purpose more advanced tools should be used, such as fractional flow reserve.
Recently, newer CT hardware and software solutions have been introduced which enable the coronary arteries to be scanned with much lower radiation exposure. In particular, the implementation of prospective ECG triggering for CT-CA allows the investigation to be performed with 1–5 mSv or below, depending on the technology applied.25–28
A particular subgroup of patients in our population was the group with non-diagnostic ex-ECG. This group comprised mostly women, with a younger age and very low risk profile (average Agatston score =11±39; range 0–146). As previously stated, this is one of the most difficult categories of patient to assess with stress testing.29 Even though this population is the one in which radiation exposure should be avoided when not absolutely necessary (especially for those aged <40 years), the need for appropriate diagnostic testing should be satisfied. Newer low radiation dose CT-CA technology will probably meet the need for a robust and accurate test in low-risk patients.30 31
For patients with a low prevalence of CAD, diagnostic strategies including CT as a baseline tool are cost effective compared with ex-ECG in ruling out CAD, despite slightly higher costs for the CT-CA strategy.32 33 Our results support a diagnostic algorithm based on non-invasive anatomical evaluation before functional tests.32 33 Even though there is no strong evidence supporting this approach, it seems reasonable that with the introduction of low-dose CT-CA, this imaging modality may represent a more reliable tool to rule out obstructive CAD, especially in patients with low-to-intermediate pre-test probability.
Limitations
The study has a number of limitations. First, we acknowledge that there may have been a selection bias towards patients with non-diagnostic ex-ECG in our population. However, a subanalysis of these excluded patients demonstrated that the diagnostic accuracy of CT-CA remains robust while ex-ECG does not improve significantly. Second, the difference between CT-CA and ex-ECG might have been inflated by considering non-diagnostic ex-ECG examinations as positive in the analysis of the entire population. On the other hand, this is a very common situation in CT-CA. When a stenosis appears borderline (around 50%) the operator defines it as >50% which can be translated into a positive finding. This is the reason why CT-CA has a high sensitivity and negative predictive value while it has lower specificity and positive predictive value. Third, the performance of ex-ECG was so poor that we did not assess the incremental value of CT-CA compared with ex-ECG and vice versa. Fourth, we decided to use the angiographic threshold of 50% for a positive ICA. This is perhaps a limitation because it is not necessarily true that all stenoses of ≥50% cause symptoms. On the other hand, the quantification of CT-CA is limited to a semiquantitative approach. Finally, the radiation dose applied in this population (15–20 mSv) is comparatively high and raises concerns. However, newer CT equipment enables the radiation dose to be reduced to 3–5 mSv and below without impairing diagnostic accuracy.25–28
Conclusion
CT-CA provides optimal diagnostic performance in patients with atypical chest pain and a low pre-test likelihood of CAD. Even though ex-ECG is inexpensive and widely available, it appears a less than satisfactory examination in this population. Concerns are related to the risk of radiation dose versus the benefits of correct disease stratification. Newer hardware and software improvements bringing a consistent reduction in radiation dose will make the option of CT-CA more realistic in this group of patients.
References
Footnotes
Competing interests None.
Patient consent Obtained.
Provenance and peer review Not commissioned; externally peer reviewed.