Computed tomographic coronary angiography (CTCA) has been proposed as a noninvasive test for significant coronary artery disease (CAD), but only limited data are available from prospective multicenter trials. The goal of this study was to establish the diagnostic accuracy of CTCA compared to coronary angiography (CA) in a large population of symptomatic patients with clinical indications for coronary imaging. This national, multicenter study was designed to prospectively evaluate stable patients able to undergo CTCA followed by conventional CA. Data from CTCA and CA were analyzed in a blinded fashion at central core laboratories. The main outcome was the evaluation of patient-, vessel-, and segment-based diagnostic performance of CTCA to detect or rule out significant CAD (≥50% luminal diameter reduction). Of 757 patients enrolled, 746 (mean age 61 ± 12 years, 71% men) were analyzed. They underwent CTCA followed by CA 1.7 ± 0.8 days later using a 64-detector scanner. The prevalence of significant CAD in native coronary vessels by CA was 54%. The rate of nonassessable segments by CTCA was 6%. In a patient-based analysis, sensitivity, specificity, positive and negative predictive values, and positive and negative likelihood ratios of CTCA were 91%, 50%, 68%, 83%, 1.82, and 0.18, respectively. The strongest predictors of false-negative results on CTCA were high estimated pretest probability of CAD (odds ratio [OR] 1.97, p <0.001), male gender (OR 1.5, p <0.002), diabetes (OR 1.5, p <0.0001), and age (OR 1.2, p <0.0001). In conclusion, in this large multicenter study, CTCA identified significant CAD with high sensitivity. However, in routine clinical practice, each patient should be individually evaluated, and the pretest probability of obstructive CAD should be taken into account when deciding which method, CTCA or CA, to use to diagnose its presence and severity.
Computed tomographic coronary angiography (CTCA) has been proposed as a noninvasive alternative to invasive coronary angiography (CA) for diagnosing obstructive coronary artery disease (CAD). Most published studies, the results of which have been reviewed in meta-analyses, have been conducted in relatively limited numbers of patients by highly experienced investigators at single centers. These studies have found high negative predictive value (NPV) of CTCA. Although the diagnostic performance of 64-row computed tomography has now been assessed in a greater number of patients in multicenter trials, their results suggest that a further large study reflecting “real-life” management may be useful to define the precise role of CTCA in patients with suspected CAD. The Evaluation of CT Scanner (EVASCAN) study is a large, national, multicenter, multivendor, prospective study in which data were analyzed at central core laboratories. Its main goal was to establish the diagnostic accuracy of 64-slice CTCA compared to conventional CA in a large population of symptomatic patients with clinical indications for anatomic coronary imaging.
Methods
The study was designed to prospectively evaluate stable adults with chest pain who were clinically referred for nonemergent invasive CA. Eligible patients were ≥18 years of age with known or suspected stable CAD, able to undergo CTCA and then CA within 4 days. The main exclusion criteria were unstable clinical status, serum creatinine >150 μmol/L, atrial fibrillation, pregnancy, and lactation. Patients were not excluded for elevated calcium score, heart rate, high body mass index, history of myocardial infarction, previous percutaneous coronary intervention, and/or coronary artery bypass grafting >6 months before CTCA.
The pretest probability of CAD was estimated using the Duke clinical score. Patients were categorized as having low (1% to 30%), intermediate (31% to 70%), or high (71% to 99%) estimated pretest probability of significant CAD.
Protocols for patient enrollment, safety monitoring, image acquisition, and interpretation were developed by the EVASCAN steering committee. The institutional review board of Paris VI University approved the protocol, and all patients gave written informed consent.
All patients underwent 64-slice CTCA followed by conventional CA. The following systems were used: GE Healthcare (Milwaukee, Wisconsin) (69%), Philips Medical Systems (Andover, Massachusetts) (14%), Siemens Healthcare (Erlangen, Germany) (14%), and Toshiba Corporation (Tokyo, Japan) (3%). CTCA was performed using a standardized, optimized protocol for each system. All patients were in normal sinus rhythm before CTCA. A β blocker was recommended if heart rate was >65 beats/min. Patients first underwent an unenhanced prospective electrocardiographically gated acquisition for calcium scoring (Agatston score) and then a retrospective electrocardiographically gated contrast-enhanced acquisition to explore the coronary tree. The mean helical volume coverage in the z axis was 15 cm for the coronary tree. Typical parameters were a tube voltage of 100 to 120 kV for patients weighing <100 kg and 140 kV for the others, effective current intensity of 600 to 1,000 mA, slice collimation ranging from 0.5- to 0.75-mm slice thickness, and 0.35- to 0.5-second gantry rotation time, depending on the system used. Current intensity modulation was systematically applied to reduce radiation during systolic phases. The effective dose of the nonenhanced scan and CTCA was estimated from the dose-length product and a conversion coefficient (k = 0.014 mSv mGy cm) for the chest as the investigated anatomic region.
A systematic reconstruction of the cardiac phases encompassing the RR interval (in 10% increments) was performed in all patients. Data were uploaded to dedicated workstations (Advantage Windows, GE Healthcare; Brilliance, Philips Medical Systems; Leonardo, Siemens Healthcare; and Vitrea, Toshiba Corporation).
Conventional CA was performed using standard techniques with a transfemoral or transradial approach. All studies were performed using digital equipment. Multiple projections were obtained as deemed necessary by the angiographer.
The results of CTCA and CA were analyzed visually in separate central core laboratories in a blinded manner by experienced readers who were unaware of patients’ clinical information or the results of the alternative imaging technique.
Coronary arteries were scored using the coronary artery classification of the American Heart Association. Each coronary segment was visually graded as individually assessable or not, normal, nonsignificant stenosis (<50%), stenosis ≥50%, or total occlusion. In case of multiple lesions in a given segment or artery, the worst lesion was considered. Importantly, nonassessable segments by CTCA were counted as positive for stenosis (≥50%).
Using a binary end point for each patient (agreement between CTCA and CA), expecting a 50% to 65% prevalence of CAD, sensitivity of 0.85, specificity of 0.9, accuracy of 4%, and a 10% dropout rate, it was considered necessary to enroll 650 to 875 patients.
Measures of diagnostic accuracy of CTCA for the detection of significant CAD on native coronary vessels (sensitivity, specificity, positive predictive value [PPV], and NPV, with their corresponding 95% confidence intervals [CIs], as well as positive and negative likelihood ratios of CTCA) were calculated on per-patient, per-vessel, and per-segment bases.
Results are expressed as point estimates and their exact 95% CIs, while continuous variables are presented as mean ± SD when normally distributed. All statistical tests were 2 sided. Chi-square tests were used to compare categorical variables. Analysis of variance or nonparametric Kruskal-Wallis tests were used to compare continuous variables depending on their normality. Interobserver reproducibility for the detection of significant CAD stenosis (<50% vs ≥50%) in computed tomographic coronary angiographic images was evaluated by κ statistics between 2 observers unaware of the results of CA who analyzed the computed tomographic scans of 30 patients randomly selected. The κ statistic was 0.58 (95% CI 0.28 to 0.88). Multivariate stepwise logistic regression analysis was used to calculate the odds ratios (ORs) and 95% CIs and to identify the independent predictors of false-negative results on CTCA as well as of nonevaluable coronary segments by CTCA. To compare the likelihood ratios between subgroups, the logistic regression models of the prior odds and posterior odds were used. Analyses were carried out using SAS version 9.2 (SAS Institute Inc., Cary, North Carolina).
Results
Between June 2006 and June 2008, 40 centers prospectively enrolled 757 patients; data from 746 were analyzed. Of the 11 patients excluded from the analysis, 8 had incomplete or canceled CTCA or CA, 2 had protocol deviations, and 1 withdrew consent.
Demographic and clinical characteristics are listed in Table 1 . The mean age was 61 ± 12 years, 71% were men, and most patients had not previously been diagnosed with CAD (n = 481 [65%]). The main cardiovascular risk factors were hypertension in 52% of patients, smoking in 26%, body mass index ≥30 kg/m² in 25%, diabetes in 25%, and hypercholesterolemia in 20%.
| Characteristic | Value |
|---|---|
| Age (yrs) | 61 ± 12 |
| Men | 529 (71%) |
| Body mass index ≥30 kg/m 2 | 187 (25%) |
| Hypertension | 388 (52%) |
| Diabetes mellitus | 189 (25%) |
| Total cholesterol ≥220 mg/dl | 150 (20%) |
| Smokers | 191 (26%) |
| Family history of CAD | 228 (31%) |
| Previous myocardial infarction | 152 (20%) |
| Previous percutaneous coronary intervention | 206 (28%) |
| Previous coronary bypass | 32 (4%) |
| Chest pain at presentation | |
| Typical angina pectoris | 418 (56%) |
| Atypical chest pain | 234 (31%) |
| Suspected CAD | 481 (65%) |
| Previously known CAD | 259 (35%) |
| Creatinine (μmol/L) | 88 ± 34 |
| Heart rate during CTCA (beats/min) | 63 ± 11 |
| Agatston calcium score | 396 ± 827 |
CA was performed 1.7 ± 0.8 days after CTCA. The estimated radiation dose for CTCA was 17.2 ± 5.9 mSv. The mean heart rate during CTCA was 63 ± 11 beats/min. Additional β blockers were administered before scanning in 20% of patients. The mean Agatston calcium score was 396 ± 827, and 11% of patients had calcium scores ≥600.
From the coronary angiographic analysis, the prevalence of ≥1 coronary lesion ≥50% was 54% (403 patients). Among these, 41% had single-vessel disease, 34% had 2-vessel disease, 23% had 3-vessel disease, and 2% had significant coronary lesions in the left main coronary artery.
Most patients with significant coronary stenosis on CA were identified by CTCA (367 of 403, a 91% true-positive rate). In 114 of 125 patients (91%) with left main or 3-vessel disease, CTCA detected ≥1 significant lesion. Thirty-six patients with CAD were missed by CTCA (a 9% false-negative rate). CTCA correctly ruled out significant CAD in 171 of 343 patients without significant stenosis using CA (a true-negative rate of 50%). ( Tables 2 and 3 ) However, because nonassessable segments were considered as stenosed, the false-positive rate was 50%. If nonassessable segments were assumed to be negative for coronary stenosis, sensitivity would decrease (from 91% to 84%), as would NPV (from 83% to 80%), but specificity would increase (from 50% to 74%), as would PPV (from 68% to 79%).
| Variable | Per Patient (n = 746) | Per Vessel (n = 2,969) | Per Segment (n = 10,767) |
|---|---|---|---|
| Stenoses by CA( ∗ ) | 403 (54%) | 790 (27%) | 1,125 (10%) |
| Stenoses by CTCA | 539 | 963 | 1,604 |
| False-positives | 172 | 410 | 1,098 |
| False-negatives | 36 | 237 | 619 |
| Sensitivity (95% CI) | 91% (88–93) | 70% (67–73) | 45 (42–48) |
| Specificity (95% CI) | 50% (45–55) | 81% (79–83) | 88 (88–89) |
| PPV (95% CI) | 68% (64–72) | 57% (54–61) | 32 (29–34) |
| NPV (95% CI) | 83% (77–89) | 88% (87–90) | 93 (93–94) |
| Positive likelihood ratio (95% CI) | 1.82 (1.63–2.03) | 3.72 (3.37–4.11) | 3.95 (3.63–4.30) |
| Negative likelihood ratio (95% CI) | 0.18 (0.13–0.25) | 0.37 (0.33–0.41) | 0.62 (0.59–0.65) |
| Variable | n | CAD Prevalence | Sensitivity (95% CI) | Specificity (95% CI) | PPV (95% CI) | NPV (95% CI) | Positive Likelihood Ratio (95% CI) | Negative Likelihood Ratio (95% CI) |
|---|---|---|---|---|---|---|---|---|
| Overall | 746 | 54% | 91% (88%–93%) | 50% (45%–55%) | 68% (64%–72%) | 83% (77%–89%) | 1.82 (1.63–2.03) | 0.18 (0.13–0.25) |
| By gender | ||||||||
| Male | 529 | 61% | 91% (88%–94%) | 47% (40%–53%) | 73% (69%–77%) | 77% (70%–85%) | 1.71 (1.50–1.95) | 0.19 (0.13–0.27) |
| Female | 217 | 36% | 90% (83%–96%) | 55% (46%–63%) | 53% (44%–61%) | 92% (84%–97%) | 1.98 (1.63–2.41) | 0.19 (0.10–0.37) |
| Chi-square | 0.21 | 2.17 | 18.84 | 5.75 | 37.87 | |||
| p value | 0.65 | 0.14 | <0.001 | 0.016 | <0.001 | |||
| By heart rate (beats/min) | ||||||||
| <65 | 458 | 53% | 90% (86%–94%) | 49% (42%–56%) | 67% (62%–72%) | 82% (75%–88%) | 1.78 (1.55–2.04) | 0.20 (0.13–0.30) |
| ≥65 | 288 | 56% | 93% (88%–97%) | 51% (42%–59%) | 70% (63%–76%) | 84% (76%–93%) | 1.88 (1.57–2.25) | 0.15 (0.08–0.26) |
| Chi-square | 0.67 | 0.07 | 0.67 | 0.28 | 15.23 | |||
| p value | 0.04 | 0.79 | 0.41 | 0.60 | <0.001 | |||
| By body mass index (kg/m 2 ) | ||||||||
| <25 | 221 | 53% | 88% (82%–93%) | 57% (46%–65%) | 69% (62%–77%) | 81% (71%–90%) | 1.99 (1.59–2.49) | 0.21 (0.13–0.36) |
| 25–30 | 318 | 56% | 92% (88%–86%) | 49% (41%–57%) | 69% (63%–75%) | 83% (75%–91%) | 1.80 (1.53–2.13) | 0.17 (0.08–0.34) |
| ≥30 | 187 | 45% | 92% (87%–97%) | 46% (36%–57%) | 68% (60%–76%) | 83% (72%–94%) | 1.72 (1.40–2.12) | 0.17 (0.08–0.34) |
| Chi-square (trend) | 1.64 | 1.82 | 0.05 | 0.14 | 39.68 | |||
| p value | 0.20 | 0.18 | 0.82 | 0.71 | <0.001 | |||
| Diabetes mellitus | ||||||||
| No | 557 | 53% | 91% (87%–94%) | 49% (43%–55%) | 67% (62%–72%) | 84% (76%–88%) | 1.79 (1.58–2.03) | 0.19 (0.13–0.28) |
| Yes | 189 | 57% | 93% (86%–97%) | 49% (40%–62%) | 71% (63%–79%) | 84% (71%–93%) | 1.90 (1.51–2.238) | 0.15 (0.07–0.29) |
| Chi-square | 0.38 | 0.08 | 0.85 | 0.09 | 15.43 | |||
| p value | 0.78 | 0.78 | 0.36 | 0.77 | <0.001 | |||
| By risk level ∗ | ||||||||
| Low | 102 | 21% | 86% (77%–100%) | 49% (38%–60%) | 31% (19%–42%) | 93% (85%–100%) | 1.69 (1.28–2.23) | 0.29 (0.11–0.84) |
| Intermediate | 201 | 44% | 87% (79%–99%) | 55% (46%–56%) | 61% (52%–69%) | 84% (75%–92%) | 1.94 (1.55–2.42) | 0.264 (0.14–0.42) |
| High | 402 | 69% | 92% (89%–96%) | 45% (36%–54%) | 79% (74%–83%) | 73% (63%–83%) | 1.67 (1.42–1.97) | 0.17 (0.11–0.27) |
| Chi-square (trend) | 0.35 | 0.02 | 51.65 | 7.49 | 15.43 | |||
| p value | 0.55 | 0.91 | <0.001 | 0.006 | <0.001 | |||
| By calcium score | ||||||||
| 0 | 147 | 21% | 81% (67%–95%) | 61% (52%–71%) | 36% (24%–47%) | 92% (86%–98%) | 2.08 (1.56–2.77) | 0.32 (0.15–0.66) |
| 1–100 | 110 | 45% | 82% (71%–93%) | 56% (43%–68%) | 60% (48%–71%) | 79% (67%–91%) | 1.84 (1.35–2.52) | 0.33 (0.18–0.62) |
| 100–600 | 70 | 63% | 93% (86%–100%) | 50% (31%–69%) | 76% (65%–87%) | 81% (62%–100%) | 1.86 (1.26–2.76) | 0.14 (0.04–0.43) |
| ≥600 | 79 | 81% | 93% (93%–100%) | 27% (4%–49%) | 85% (77%–94%) | 67% (29%–100%) | 1.32 (0.97–1.80) | 0.12 (0.02–0.58) |
| Chi-square (trend) | 4.25 | 4.17 | 15.08 | 1.36 | 48.79 | |||
| p value | 0.04 | 0.04 | <0.0001 | 0.24 | <0.001 | |||
∗ Low, 1% to 30%; intermediate, 31% to 70%; high, 71% to 100%.
Our results show that coronary arterial calcification (calcium score) significantly alters the diagnostic performance of CTCA. There was a statistically significant trend toward increased sensitivity (p <0.04) and PPV (p <0.0001) and decreased specificity (p <0.04) with increased calcium score ( Table 3 ). Interestingly, the absence of coronary calcium alone was not sufficient to exclude CAD (21% CAD prevalence in 147 patients with calcium scores of 0). In such patients, CTCA was highly effective to rule out significant CAD (NPV 92%).
The analysis comprised 102 patients (14%) with low, 201 (28%) with intermediate, and 402 (57%) with high estimated pretest probability for CAD (incomplete data in 41 patients). No significant differences were noted in sensitivity and specificity. However, as expected, PPV increased with risk level, and in contrast, NPV was higher in patients with low pretest likelihood of CAD ( Table 3 ).
Among false results of CTCA, false-negative results may be detrimental to patient management, because clinicians may be reassured about patients’ coronary artery anatomy. We performed a logistic regression analysis to identify clinical parameters predictive of false-negative results ( Table 4 ). On multivariate analysis, factors that increased the likelihood of false-negative results on CTCA were age (OR 1.2, p <0.001), male gender (OR 1.5, p <0.002), diabetes (OR 1.5, p <0.0001), and high estimated pretest probability of CAD (OR 1.9, p <0.001).
| Variable | Univariate Analysis | Multivariate Analysis | ||||
|---|---|---|---|---|---|---|
| OR | (95% CI) | p Value | OR | (95% CI) | p Value | |
| False-negative results | ||||||
| Age | 1.2 | (1.1–1.3) | <0.0001 | 1.2 | (1.1–1.3) | <0.001 |
| Gender (male vs female) | 2.0 | (1.6–2.4) | <0.0001 | 1.5 | (1.16–1.94) | 0.0023 |
| Diabetes (yes vs no) | 1.6 | (1.3–1.9) | <0.0001 | 1.5 | (1.26–1.82) | <0.0001 |
| Body mass index (≥30 vs <30 kg/m 2 ) | 1.0 | (0.9–1.3) | 0.66 | 1.0 | (0.80–1.19) | 0.8407 |
| Heart rate (≥65 vs <65 beats/min) | 1.1 | (0.9–1.3) | 0.49 | 1.1 | (0.92–1.31) | 0.3058 |
| Pretest probability groups | ||||||
| Intermediate vs low | 1.4 | (0.9–2.0) | 0.10 | 1.1 | (0.69–1.66) | 0.8697 |
| High vs low | 3.2 | (2.3–4.4) | <0.0001 | 1.9 | (1.28–3.05) | 0.0013 |
| Nonassessable coronary segments | ||||||
| Age | 1.00 | (0.99–1.01) | 0.96 | 1.0 | (0.98–1.01) | 0.67 |
| Gender (male vs female) | 1.38 | (1.17–1.62) | <0.0001 | 1.14 | (0.85–1.5) | 0.39 |
| Diabetes (yes vs no) | 1.30 | (1.10–1.54) | <0.002 | 1.04 | (0.76–1.42) | 0.82 |
| Artifact (yes vs no) | 2.72 | (2.32–3.20) | <0.001 | 2.53 | (1.88–3.41) | <0.001 |
| Body mass index (≥30 vs <30 kg/m 2 ) | 1.92 | (1.63–2.25) | <0.001 | 2.72 | (2.07–3.66) | <0.001 |
| Heart rate (≥65 vs <65 beats/min) | 2.06 | (1.77–2.41) | <0.001 | 1.43 | (1.08–1.89) | 0.01 |
| Segment diameter (<1.5 vs ≥1.5 mm) | 19.26 | (16.13–23.00) | <0.001 | 18.66 | (13.93–24.97) | <0.001 |
| Calcium score (≥600 vs <600) | 1.61 | (1.24–2.08) | 0.0003 | 1.74 | (1.23–2.47) | <0.002 |
| Segment (distal vs proximal/medial) | 2.70 | (2.29–2.32) | <0.001 | 1.98 | (1.46–2.68) | <0.001 |
| Location of stenosis (in circumflex coronary artery vs left anterior descending coronary artery or right coronary artery) | 2.09 | (1.79–2.44) | <0.001 | 1.64 | (1.24–2.16) | 0.0005 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
