Diagnostic Performance of 64-Multidetector Row Coronary Computed Tomographic Angiography for Evaluation of Coronary Artery Stenosis in Individuals Without Known Coronary Artery Disease

Results From the Prospective Multicenter ACCURACY (Assessment by Coronary Computed Tomographic Angiography of Individuals Undergoing Invasive Coronary Angiography) Trial

Patients

The ACCURACY (Assessment by Coronary Computed Tomographic Angiography of Individuals Undergoing Invasive Coronary Angiography) study was designed to prospectively evaluate adult subjects with chest pain who were being clinically referred for nonemergent invasive coronary angiography (ICA). Potential study subjects were screened and enrolled by a site research coordinator if they met both inclusion and exclusion criteria. Study subjects were asked to undergo a research CCTA, as well as data and blood collection as specified by a pre-defined research protocol.

Individuals were eligible for participation in the ACCURACY trial if they were ≥18 years of age, experienced typical or atypical chest pain, and were being referred for nonemergent ICA.

Individuals were excluded from participation in the ACCURACY trial for the following reasons: known allergy to iodinated contrast; baseline renal insufficiency (creatinine ≥1.7 mg/dl); irregular cardiac rhythm; resting heart rate >100 beats/min; resting systolic blood pressure <100 mm Hg; contraindication to beta-blocker, calcium-channel blocker, or nitroglycerin; pregnancy; and known history of CAD (prior myocardial infarction, percutaneous transluminal coronary angioplasty or intracoronary stent, or coronary artery bypass surgery). Importantly, patients were not excluded for an elevated CAC score or body mass index.

The study was performed at 16 centers in the U.S. (Online Appendix). Before the study commenced, each Institutional Review Board had reviewed and approved the study protocol and patient safety monitoring plan. Protocols associated with patient enrollment, safety analysis, image acquisition, image interpretation, and statistical analysis were developed by a Steering Committee. GE Healthcare (Milwaukee, Wisconsin) performed study monitoring, data management, and quality control. Adverse and serious adverse events were determined for follow-up by a Data and Safety Monitoring Board.

Sample size

Estimation of sample sizes using a binary end point for each subject (e.g., agreement between CCTA and ICA, whether sensitivity or specificity) gave a conservative estimate of the required sample size. Based on historical data, we assumed values for sensitivity and specificity of 0.88 (standard deviation: 0.045) at the patient level, which required a minimum of 173 subjects to reject the null hypothesis that either sensitivity or specificity is ≤0.80 in favor of the alternative >0.80.

CCTA image acquisition

Study subjects underwent CCTA before conventional ICA. All CCTA scans were performed with a 64-multidetector row Lightspeed VCT scanner (GE Healthcare). All patients were in normal sinus rhythm at the time of the CCTA scan. Individuals presenting with baseline heart rates >65 beats/min were administered oral beta-blocker therapy as the preferred method for slowing down the heart rate. Intravenous administration was allowed in the protocol, using metoprolol at 5 mg increments to a total possible dose of 25 mg to achieve a resting heart rate <65 beats/min. All patients eligible for CCTA were scanned, whether or not the goal of a heart rate <65 beats/min was achieved.

Following a scout radiograph of the chest (anteroposterior and lateral), a timing bolus (using 10 to 20 ml contrast) was performed to detect time to optimal contrast opacification in the axial image at a level immediately superior to the ostium of the left main artery. Nitroglycerine 0.4 mg sublingually was administered immediately before contrast injection. During CCTA acquisition, 80-ml iodinated contrast (Visipaque, GE Healthcare, Buckinghamshire, United Kingdom) was injected using a triple-phase contrast protocol: 60-ml iodixanol, followed by 40 ml of a 50:50 mixture of iodixanol and saline, followed by a 50-ml saline flush. Retrospective electrocardiogram-gated helical contrast-enhanced CCTA was performed, with scan initiation 20 mm above the level of the left main artery to 20 mm below the inferior myocardial apex. The scan parameters were 64 × 0.625 mm collimation, tube voltage 120 mV, and effective mA 350 to 780 mA. Radiation reduction algorithms using electrocardiography modulation were used, which reduce radiation exposure (mA) during systole and end-diastole. After scan completion, multiphasic reconstruction of the CCTA scans was performed, with reconstructed images from 70% to 80% by 5% increments and 5% to 95% by 10% increments.

CCTA interpretation

The CCTA images were interpreted separately by 3 readers (M.J.B., D.D., and J.K.M.) blinded to all patient characteristics and ICA results. All CCTA images were evaluated on a 3-dimensional image analysis workstation (GE Advantage Workstation, GE Healthcare, Milwaukee, Wisconsin). The CCTA readers were permitted to use any or all of the available post-processing image reconstruction algorithms, including 2-dimensional axial and 3-dimensional maximal intensity projection, multiplanar reformat, cross-sectional analysis, and volume-rendered technique.

Coronary arteries were scored using a 15-segment AHA coronary artery classification, as previously described (3). An overall assessment of image quality and coronary supply dominance was performed on the subject level (4,5). For each coronary segment, readers assessed whether coronary segments were evaluable. For any coronary artery segments considered to be nonevaluable, stenosis severity was assigned based on the outcome of the most adjacent proximal and identifiable segment, as previously described (6). A semiquantitative scale was used by the CCTA readers to grade extent of luminal stenosis as a percentage of the vessel diameter using visual estimations. Stenosis severity was recorded in the following manner: no stenosis; 1% to 29% stenosis; 30% to 49% stenosis; 50% to 69% stenosis; 70% to 99% stenosis; and 100% stenosis. For coronary artery segments considered to have 100% stenosis by CCTA, all segments distal to the occlusion were excluded from analysis (Fig. 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Computed Tomographic Angiogram and Invasive Angiogram Demonstrating Significant Left Anterior Descending Artery Disease

(A) Right anterior oblique orientation of left anterior descending artery with quantitative coronary angiography. (B) Multiplanar reformation and short-axis cross-sectional view (inset) of the left anterior descending artery demonstrating obstructive coronary artery stenosis. (C) Curved multiplanar reformat of the left anterior descending artery demonstrating obstructive coronary artery stenosis. (D) Volume-rendered view of the left anterior descending artery. Arrows indicate the significant stenosis present on the computed tomographic angiogram and corresponding invasive angiogram.

The degree of coronary artery stenosis identified by CCTA was assigned based upon a consensus of ≥2 of the 3 blinded CCTA readers who identified narrowing of the coronary artery lumen at a threshold of 50% or 70% stenosis. Consensus was achieved on a per-patient and per-vessel level. Consensus was obtained in all but 3 cases.

ICA image acquisition and interpretation

Selective ICA was performed by standard transfemoral arterial catheterization. A minimum of 8 projections were obtained (minimum of 5 views for the left coronary artery system and minimum of 3 views for the right coronary artery system). Because of differences in cardiac position, angles of projection for ICA differed slightly among study subjects.

All ICA images were interpreted by an independent ICA reader (J.G.J.) blinded to all patient characteristics and CCTA results. The ICAs were quantitatively evaluated for coronary artery stenosis with quantitative coronary angiography (QCA) software (CAAS, Pie Medical Imaging, Maastricht, the Netherlands). Any segment deemed visually to have >15% stenosis was quantified. Coronary artery segments by QCA were also evaluated using a 15-segment AHA coronary tree model and were judged as having significant stenosis at 2 levels (i.e., if ≥50% or ≥70% luminal narrowing of the coronary artery diameter was present).

Data analysis

In all analyses, all patients and all vessels were included. Analyses were performed separately for 2 distinct conditions—≥50% and ≥70% luminal diameter narrowing—that defined obstructive coronary artery stenosis. For the patient-based analysis, a true-positive was defined as the presence of ≥1 coronary artery segment considered to have an obstructive stenosis by both CCTA and ICA, irrespective of location. For the vessel-based analysis, a true-positive was defined as the presence of ≥1 coronary artery segment considered to have an obstructive stenosis by both CCTA and ICA in a single arterial system. Four arterial systems were predefined and consisted of the: 1) left main artery; 2) left anterior descending artery inclusive of diagonal branches; 3) left circumflex artery inclusive of obtuse marginal and left-sided posterolateral branches; and 4) right coronary artery inclusive of posterior descending artery and right-sided posterolateral branches. Ramus intermediate arteries were considered to be the first obtuse marginal branch for per-vessel analyses.

Statistical analysis

Categorical variables are presented as frequency and percentage, continuous variables as mean ± SD. The area under the receiver operating characteristic curve (AUC) was calculated for CCTA to identify obstructive coronary artery stenosis at either 50% or 70% threshold. All statistical analyses were performed using SAS Proprietary Software, version 9.1 (SAS Institute, Cary, North Carolina).

Patient characteristics

There were 245 subjects initially enrolled into the study. Fifteen subjects did not complete either CCTA or ICA (withdrew after CCTA and did not undergo ICA or opted out of the CCTA and only underwent ICA) and were therefore excluded from the final efficacy analyses. Of the patient study cohort, 147 patients underwent stress testing by a range of different tests, including exercise treadmill testing (n = 17), exercise echocardiogram stress testing (n = 28), exercise nuclear stress testing (n = 96), pharmacological echocardiographic stress testing (n = 2), and pharmacological nuclear stress testing (n = 4). The remainder of the test cohort was referred on the basis of symptoms and/or testing not performed at the primary site.

One subject experienced a coronary artery dissection at the time of ICA involving 2 coronary artery segments (proximal right coronary artery, mid-right coronary artery). For this patient, efficacy analyses were performed based upon a total of 13 rather than 15 coronary segments, excluding comparison of these 2 dissected segments from the ICA and CCTA. Patient-level data were available for efficacy analysis for 230 subjects (mean age: 57 ± 10 years; 59.1% male) (Table 1). The mean inter-test interval between CCTA and ICA was 5.9 ± 4.3 days.

Patient-based evaluation

The CCTA test characteristics and performance for patient-based evaluation are listed in Table 2. There were 57 (24.8%) and 32 (13.9%) subjects found to have ≥50% or ≥70% stenosis, respectively, by QCA (Figs. 1 and 2).⇓ Discordance—specifically, the first reader scoring a subject's examination nonevaluable, a second reader scoring an examination without obstructive stenosis, and a third reader scoring an examination with obstructive coronary stenosis—occurred for 3 subjects (1.3%), which were not included in the patient-level efficacy analysis. Among the 55 patients with ≥50% stenosis by QCA, 52 were correctly identified as having ≥50% stenosis by CCTA. Among the 31 patients with ≥70% stenosis by QCA, 29 were correctly identified as having ≥70% stenosis by CCTA. The AUC for identification of patients with ≥50% coronary artery stenosis by QCA was 0.96 (95% CI: 0.94 to 0.98). Applying CCTA stenosis thresholds to identify ≥70% stenosis by QCA resulted in a similarly high AUC of 0.95 (95% CI: 0.92 to 0.97) (Fig. 3). Sensitivity at both the 50% and 70% thresholds of disease was 95% and 94%, respectively, specificity was 83%, and negative predictive value (NPV) was 99% (Table 2). Given the low to intermediate prevalence of disease, positive predictive value (PPV) was low (64% and 48% for 50% and 70% thresholds of disease, respectively). Adding in the 3 unevaluable (due to lack of agreement) patients, the results remained statistically similar (at 50% threshold: sensitivity 94.7%, specificity 82.1%, PPV 64.2%, and NPV 99.3%; at 70% threshold: sensitivity 93.8%, specificity 81.8%, PPV 47.6%, and NPV 98.8%).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Computed Tomographic Angiogram Demonstrating Obstructive Disease of the Left Circumflex Artery With Quantitative Angiography Correlation

(A) Right anterior oblique orientation of the left circumflex artery with quantitative coronary angiography. (B) Curved multiplanar reformation and short-axis cross-sectional view (inset) of the left circumflex artery demonstrating obstructive coronary artery stenosis. (C) Multiplanar reformation and short-axis cross-sectional view (inset) of the left circumflex artery demonstrating obstructive coronary artery stenosis. Arrows indicate significant stenosis present on the computed tomographic angiogram and corresponding invasive angiogram.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

ROC Curve Evaluating Diagnostic Accuracy of CCTA Compared With Invasive Angiogram

Receiver-operating characteristic (ROC) curve for identification of patients by coronary computed tomographic angiography (CCTA) with ≥70% coronary artery stenosis by quantitative coronary angiography. The points on the plot represent the 6 categories of interpretation for CCTA used in this study: 0 = 100% stenosis; 1 = 70% to 99% stenosis; 2 = 50% to 69% stenosis; 3 = 30% to 49% stenosis; 4 = <30% stenosis; and 5 = no stenosis. The ROC shows us the degree to which we predict disease/no disease of 70% stenosis by invasive angiography. AUC = area under the receiver-operating characteristic curve; CI = confidence interval.

Vessel-based evaluation

The CCTA test characteristics and performance for vessel-based evaluation are listed in Table 3. For the vessel-based evaluation, 3 vessels were not evaluable by QCA and 7 vessels were scored discordantly by CCTA readers, resulting in 910 vessels in 229 patients. In total, 89 (9.7%) and 39 (4.3%) vessels reached the 50% or 70% stenosis threshold, respectively, by QCA. Among the 89 vessels with ≥50% stenosis by QCA, 73 vessels were correctly identified as having ≥50% stenosis by CCTA. Among the 39 vessels with ≥70% stenosis by QCA, 32 were correctly identified as having ≥70% stenosis by CCTA. For both 50% and 70% stenosis thresholds, sensitivity was 84%, specificity was 90% and 92%, respectively, and NPV was 99% (Table 3).

Calcium score

We examined diagnostic performance of CCTA based on baseline CAC score, stratified by ≤400 versus >400 Agatston units. Stratifying the diagnostic test performance characteristics of CCTA by these CAC score thresholds resulted in no change in diagnostic sensitivity but a reduction in specificity. The patient-based sensitivity for the presence of ≥50% stenosis for patients with calcium scores ≤400 versus >400 or ≤600 versus >600 Agatston units was 95.8% versus 93.6% (p = 0.71) and 96.9% versus 91.3% (p = 0.37), respectively. The specificity of CCTA-diagnosed obstructive disease at the 50% stenosis threshold was reduced in patients with coronary artery calcium scores ≤400 versus >400 Agatston units (86.3% vs. 52.6%; p > 0.0003).

Body mass index and heart rate

Diagnostic performance of CCTA was examined in nonobese versus obese study participants, stratified by body mass index ≤30 kg/m² or >30 kg/m². For CCTA detection of ≥50% coronary artery stenosis at the patient level in nonobese versus obese subjects, sensitivity was 94.4% and 94.7% (p = 0.96), respectively, and specificity was 76.4% and 87.0% for ≤30 kg/m² or >30 kg/m², respectively (p = 0.07).

Diagnostic performance of CCTA for detection of ≥50% coronary artery stenosis at the patient level did not differ in individuals with heart rates ≤65 beats/min versus >65 beats/min, with a diagnostic sensitivity of 92.9% versus 100%, respectively, and a diagnostic specificity of 82.3% versus 82.8%, respectively (p > 0.35 for both comparisons).

Study limitations

We enrolled subjects in outpatient settings without known CAD who were being referred for nonemergent conventional coronary angiography. Therefore, whether these results can be extrapolated to individuals with known CAD or to other settings (e.g., emergency department) requires further study. Furthermore, only 1 blinded reader interpreted the ICAs. Nevertheless, the reference standard used in the present study was QCA rather than semiquantitative assessment of luminal diameter stenosis by ICA, a technique more prone to interobserver variability (17,18). We elected to implement tube current modulation to reduce radiation dose. Tube current modulation, however, limits the ability to interpret the coronary anatomy in cardiac phases other than the modulated phase and may result in artifacts in patients with irregular heart rates. Finally, no interpreting format was pre-specified for the CCTA readers, who used a variety of interpretative 3-dimensional post-processing algorithms, which precludes definitive comparison of 1 CCTA interpretation technique to another (19).

The benefit of this study is that it is a multicenter investigation using established methods widely used by practitioners and under less than ideal conditions (variable heart rates, obesity, high CAC scores, and dose modulation) that mimic current clinical practice. Despite these limitations, the performance of 64-MDCT was quite good. The analysis methods focused on the diagnostic accuracy of determining obstructive disease, not on establishing the value in defining atherosclerotic plaque burden. CCTA is clearly not as accurate as angiography in finding obstructive CAD (the findings of multiple studies put the PPV at ∼60%). However, this study, as well as earlier single-center evaluations, demonstrate that CCTA provides a valuable NPV as well as a nearly 100% PPV for presence of atherosclerotic plaque. The high NPV for obstructive disease and the high PPV for atherosclerotic plaque makes CCTA unique in evaluating patients in which the suspicion for “obstructive” is intermediate.