Methods

Study population

Between September 2006 and December 2008, 471 consecutive patients with stable chest pain and no history of CAD were evaluated at our 1-day chest pain clinic. All patients with a pre-test probability >5%6 were intended to undergo XECG and contrast-enhanced cardiac CT, in addition to a clinical examination, blood analysis and echocardiography. Angina was categorised as typical: retrosternal discomfort, precipitated by exercise or emotion, relieved within 10 min after rest or nitroglycerin; atypical: with only two of the previous characteristics; or non-anginal: with none or only a single characteristic (table 1). The study was approved by the institutional ethical committee and informed consent was obtained from all patients.

Computed tomography

Contraindications to CTA were pregnancy, known allergy to iodine contrast media, impaired kidney function. Contrast-enhanced, 64-slice dual-source CT (Siemens Definition, Forchheim, Germany) was performed with the following parameters: spiral mode, collimation 32×0.6 mm, 64-channel acquisition using Z-axis alternation of the focal spot, rotation time 330 ms, temporal resolution 83 ms, tube voltage 120 kV, tube current 380–412 mA depending on the patient size, variable pitch depending on the heart rate. Prospectively ECG-triggered tube modulation was used in all patients with a regular heart rate. A 70–100 ml bolus was injected at 5.0–5.5 ml/s through a peripheral vein in the arm, followed by 40 ml of saline at the same injection rate. A bolus tracking technique was used to synchronise the data acquisition with contrast enhancement. Patients received a sublingual dose of nitroglycerin just before the scan. No additional β blockers were administered. Retrospective ECG-gated image reconstruction was performed using a slice thickness of 0.75 mm, with an overlap of 0.4 mm. End-systolic and/or diastolic datasets were created depending on the heart rate and the ECG-triggered tube modulation protocol.

The coronary arteries, the right coronary artery, left main stem, left anterior descending coronary artery and left circumflex branch, were assessed using the axial images, multiplanar reformations and maximum intensity projections. Vessels were qualitatively scored as significantly stenosed (>50% diameter narrowing), less than significantly stenosed (<50%) or normal.

Exercise ECG

XECG on a bicycle ergometer was performed by standardised protocol. XECG was not performed if the patient was technically or physically unable to perform the test, or in patients with a significantly abnormal resting ECG. During continuous ECG registration and 12-lead prints at 1 min intervals, the workload was increased from 40 W by 20 W increments at 2 min intervals. Blood pressure was measured every 2 min. Criteria for discontinuation were systolic blood pressure >230 mm Hg, diastolic blood pressure >130 mm Hg, a >10 mm Hg systolic blood pressure drop, ST depression >0.3 mV, ST elevation >0.1 mV, sustained ventricular tachycardia, increasing frequency of polymorphic ventricular complexes, altered atrioventricular or intraventricular conduction, exhaustion, severe dyspnoea, angina or other discomfort. Criteria for myocardial ischaemia included horizontal or downsloping ST depression or elevation >0.1 mSv measured at 80 ms from the J point during or after exercise, or typical, increasing angina during exercise. The XECG was considered non-diagnostic if the test was discontinued without evidence of myocardial ischaemia before the 85% target heart rate was reached.

Invasive angiography

Clinically indicated cardiac catheterisation was performed in 98 patients using standard techniques. Semiautomatic quantification angiography (QCA) of luminal obstruction was performed by an independent, blinded observer. Maximum lumen diameter stenosis ⩾50% was considered moderate, ⩾70% was considered severely stenosed.

Quantitative coronary angiography

For patients who underwent invasive coronary angiography, coronary obstructions were also quantitatively assessed on CTA, using validated semiautomatic quantification software (Circulation; Siemens, Forchheim, Germany).7 8 After vessel segmentation and centre-lumen line reconstruction, short-axis images of the vessel at submillimetre increments were reconstructed. The lumen area was automatically quantified based on attenuation thresholds, with the option for manual adjustments. The diameter stenosis was calculated from the minimal lumen area and the average of the proximal and distal reference area, with the assumption of a circular shape of the respective lumen areas.

For the quantitative comparison of CTA and QCA all vessels were included, and the most severe lesion for each vessel was used in the analysis. Both for CTA and QCA normal vessels and vessels with minor wall irregularities or mildly obstructive plaque were visually scored as 0% and 20% stenosis, respectively. Severely calcified obstructions that could not be processed automatically, required qualitative assessment, classified as moderately stenosed (50–70%), severely stenosed (>70–99%), or completely obstructed (100%), and were recorded as 60%, 80% and 100%, respectively, in comparison with quantitative invasive angiography. Quantitative analysis was possible for CTA in 150, and for QCA in 143 vessels.

Statistical analysis

Summary data are presented as numbers (proportions) or means (SD), unless otherwise indicated. Diagnostic performance is expressed as sensitivity, specificity, positive predictive value and negative predictive value, with 95% confidence intervals. Association between quantitative CT and catheter angiography measurements were analysed using the Pearson coefficient. Differences between groups were compared using a two-sided unpaired t test, a χ² test or analysis of variance test. p Values <0.05 were considered significant. For lesions where quantitative CT analysis could not be applied (inferior image quality, extensive calcification) visually moderate and severe stenosis were classified as 60% or 80% diameter stenosis. Segments with wall irregularities on QCA or non-obstructive plaque on CT were semiquantified as 20% stenosis. Statistical analysis was performed using SPSS, version 15.0 (SPSS, Chicago, USA).

Results

CTA could not be performed in 16 patients (3.4%), for reasons of renal failure, known contrast allergies, patient preferences, Parkinson’s disease, patient non-cooperation, lack of venous access and severe obesity. Three scans (0.7%) failed because of patient movement (two) and premature scan initiation (one). Interpretable CT scans were available in 452/471 (96%) patients (table 2). Three patients (0.7%) had mild to moderate allergic reactions.

XECG could not be performed in 48 (10.2%) patients because of orthopaedic restraints (13), neurological restraints (four), severe obesity (two), resting ECG abnormalities (six), pulmonary disease (two), combination/unspecified (21). For the 423 XECGs performed the mean (SD) maximum workload was 143 (47) W (1.00 (24)% of the predicted workload), and the maximum heart rate was 87 (13)% of the target heart rate. Results were considered inconclusive in 140/423 (33%) patients, mostly because the target heart rate was not reached.

CTA and XECG test agreement

The percentage of CT angiograms with >50% stenosis in at least one vessel: 30% (140/452), was comparable to the proportion of abnormal XECGs: 33% (93/283, p = 0.63)(table 2). CTA and XECG results matched in 185/274 (68%) patients with interpretable results (fig 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

CT angiography (CTA; three-dimensional volume rendered and curved reconstructions), invasive coronary angiography (QCA), and maximum ST depression on exercise ECG (XECG, insert) of four patients: (1) A 54-year old woman with a symptomatic exercise test, though modest ST depression (insert), and severe obstructive disease in the left anterior descending coronary artery (LAD) on both CTA and QCA. (2) A 62-year-old man with atypical complaints, a negative XECG, a moderate stenosis with calcification on CTA, but less than significant QCA. (3) A 55-year-old man with atypical angina, a negative XECG, but three-vessel disease on CTA and QCA, including the LAD, the intermediate branch, a marginal branch and the left circumflex branch (CX). (4) A 47-year-old man with atypical angina and a positive XECG, but without significant obstruction on CTA or QCA.

Diagnostic performance compared with quantitative coronary angiography

Clinically driven invasive angiography was performed in 98 patients (21%), which showed significant CAD in 57 (58%). Including patients with both positive XECG and CTA, without invasive angiography for confirmation, the total number of patients with significant CAD was 89 (19%). Only 59 (13%) patients underwent percutaneous (46) or surgical revascularisation (13).

The sensitivity of CTA to identify patients with any >50% CAD was significantly better than XECG (table 3): 96% (95% CI 86% to 99%) compared with 71% (52% to 84%), respectively. After exclusion of non-diagnostic cases specificity was significantly better for XECG than for CTA: 76% (54% to 90%) versus 37% (23% to 53%), respectively. When an intention-to-diagnose approach was used, which includes non-diagnostic tests as positive results, the specificity of XECG decreased to the same level as CTA. Combined CTA and XECG (positive results for both) improved specificity, at a loss of sensitivity. CTA was not significantly more sensitive in patients with a diagnostic exercise test than in those who could not perform XECG or had an inconclusive result.

Quantitatively CTA correlated fairly with QCA, with a slight overestimation of diameter stenosis severity (per vessel): 6 (21)% (R = 0.71). Quantitative measurements between CT and QCA correlated best for the right coronary artery (R = 0.79) and least for the left main coronary artery (R = 0.22).

Potential value of complementary imaging

CTA did not show significant stenosis in 312 patients (66%), of whom 44 (14%) had a positive XECG. Seventeen patients (5%) had a clinical need for invasive angiography, which was positive in two patients (0.6%), of whom only one had a positive XECG. XECG was negative in 190 patients (40%), of whom 45 (24%) had a positive CTA. Of the 29 patients who underwent QCA, 10 had significantly obstructive coronary disease, all with positive CTAs (fig 2).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Virtual flow chart of exercise ECG (XECG) followed by coronary CT angiography (CTA) (A), and CTA followed by XECG (B). Unavailable tests (N/A) include tests that could not be performed, or produced equivocal results.

Among patients with a positive CTA (140), for those with a positive XECG (47, 34%) QCA showed more, and more severe CAD: 22/25 significant, 15/25 severe stenosis; compared with those with a negative XECG (45, 32%): 10/23 significant (p = 0.001), 4/23 severe CAD (p<0.003).

Within the group of patients without interpretable XECGs, CAD was ruled out by CTA in 130/188 (69%) patients, contradicted by QCA in only a single patient out of seven catheterised. Most patients with a non-diagnostic XECG and positive CTA underwent QCA (31/48, 65%), which confirmed significant disease in 21 (68%).

Discussion

Diagnostic evaluation of chest pain of recent onset

Chest pain is a common symptom, and a potential manifestation of coronary heart disease. Despite a modest diagnostic accuracy of merely 70%, combined with a thorough clinical assessment, XECG remains the most widely used test to assess the presence of ischaemic heart disease.1 Rapid-access chest pain clinics, which have emerged over the past decade, appear effective to assess patients early after the onset of symptoms and to identify patients at increased risk for adverse cardiac events.8 9 In patients diagnosed with angina, based on clinical assessment, resting ECG and XECG, death to coronary heart disease or acute coronary syndrome was reported in 16.5% at 3 years’ follow-up, compared with 2.7% in patients with non-cardiac complaints.10 Nevertheless, nearly one-third of all major adverse coronary events occurred in the (larger) group of patients thought not to have coronary related complaints, which is explained by the fact that luminal obstruction is not required before these acute events. Furthermore, Sekhri, et al, demonstrated that after clinical evaluation, the XECG contributed only modestly to the prediction of outcome.11 Taylor et al, reported that patients who were unable to perform an exercise test, or produced uninterpretable results, experienced major adverse events at a rate similar to that of patients classified as having stable angina pectoris.9 Considering these observations, XECG may be insufficient to stratify patients with undifferentiated chest pain according to risk. Stress imaging, which may be more accurate for myocardial ischaemia detection, shares the limitation of XECG with respect to the detection of less obstructive atherosclerotic disease.

Coronary CTA can be used to rule out (obstructive) coronary artery disease, particularly in patients at intermediate risk with non-conclusive stress test results.2 Additional prognostic value may be found in the detection of coronary atherosclerosis by CT, whether obstructive or not.12 13

Performance of CTA versus XECG

In this prospective registry of real-world, symptomatic patients with a low–intermediate prevalence of ischaemic heart disease, CT could be performed in more patients (p<0.001), and interpretable results were more often achieved (p<0.001). Using strict criteria, the number of non-diagnostic XECGs was 33%, which is high in comparison with previous reports (18–31%).14 15 With quantitative invasive angiography as reference, the sensitivity of CTA to detect significant CAD was higher than XECG. Only after exclusion of non-diagnostic tests did XECG showed better specificity than CTA. Sensitivity and specificity of XECG for the detection of significant CAD on QCA was 71% and 76%, comparable to a large meta-analysis that reported a sensitivity and specificity of 68% and 77%.1 Differentiation between severe stenosis and total occlusion can be difficult by CTA.16 Catheterisation showed 14 occluded vessels, which CTA assessed as occluded in eight, severely stenosed in three, moderately stenosed in two, and less than significantly stenosed in one patient (CT diameter stenosis 41%). There were no false-positive occlusions by CTA.

To date, two studies in populations of moderate size and high disease prevalence have compared the diagnostic performance of 16-slice CTA and XECG. Using an intention-to-diagnose approach Dewey et al found that CTA outperformed XECG for both sensitivity: 91% vs 73%, and specificity: 83% vs 31%, and a substantial number of non-assessable XECGs, as well as CTAs.17 In a population with a 74% disease prevalence Mollet et al, found a sensitivity and specificity of 100% and 87% for CTA, compared with 78% and 67% for XECG.18 Verification bias is assumed at least partially responsible for the lower specificity of CTA in our study—that is, patients with perceived stenosis (as a result of excessive calcification) were more likely to undergo clinically driven invasive investigation.

This study further confirmed earlier reported discrepancies between angiographic CTA and myocardial perfusion imaging.19 20 21 Some authors reported better agreement between CTA and nuclear imaging using a 70% stenosis cut-off point.22

Clinical use of CTA in stable angina

CT could be performed with diagnostic quality in nearly all patients and could rule out obstructive CAD in 66% of patients, compared with 40% by XECG. Based on these results CT may be considered as initial test in patients with chest pain and reason to suspect CAD. A negative CTA virtually rules out ischaemic heart disease, requiring no further testing. In patients with obstructive CAD on CT, functional testing would be required to assess haemodynamic significance, localise ischaemia and determine the need for revascularisation. In our study not all patients with an abnormal exercise test were catheterised. In the absence of severe inducible ischaemia and prognostically important CAD—that is, left main or three-vessel disease, the caring doctors were more confident initiating and continuing medical treatment instead of referral for revascularisation.

Drawbacks of cardiac CTA include the lack of prognostic information from the patient’s aerobic performance, the vulnerability of the technique to cardiac arrhythmia and the need for potentially nephrotoxic contrast media and radiation (10–15 mSv). Over the past few years a number of robust techniques have been (re-)introduced to decrease the radiation dose associated with cardiac CT, including optimised ECG-triggered tube modulation for spiral acquisition protocols, variable table speed, ECG-triggered sequential acquisition protocols, anatomy-based dose modulation, lower tube output settings in smaller patients, tighter scan ranges, etc. Awareness of increased radiation dose by doctors, combined with contemporary technology, allows for routine cardiac CT at an effective dose below 5 mSv.23 24

Study limitations

Referral to cardiac catheterisation and coronary angiography was clinically driven and not available in the majority of patients without non-invasive evidence of (severe) CAD. Many patients with positive CTAs and/or XECGs were treated medically, at least initially. Selective referral to invasive angiography probably had a positive effect on sensitivity and a negative effect on the specificity of CTA, in comparison with QCA. The results of this study have generated hypotheses for the use of CTA in patients with stable chest pain, which require testing in prospective, randomised trials.

Conclusion

Contrary to XECG, CT can nearly always be performed with diagnostic results. CTA was normal or showed less than significantly obstructive coronary disease in the majority of patients, in which case an abnormal exercise test or catheter angiogram was unlikely. With invasive angiography as a reference, CTA is more sensitive but less specific than XECG for the detection of significant CAD. Randomised trials are needed to define how CTA might be used (cost-) effectively in real-world patients with stable angina.