Coronary artery disease cannot be reliably evaluated by 16-slice multidetector spiral computed tomography

Abstract

Objectives. Angiography by selective catheterization is the standard method for coronary artery imaging but carries a risk of rare, but serious complications. We investigated whether 16-slice multidetector spiral computed tomography (MDCT) could substitute for selective angiography for evaluation of coronary artery disease in surgically revascularized patients. Design. In a setting closely resembling routine clinical practice, 45 patients who had been operated with coronary artery bypass grafting 508–1135 (mean 811) days before were examined with MDCT and conventional selective angiography on the same day. The interpreters were blinded to the results of the parallel imaging modality. Results. Significant pathology (stenosis ≥50% or occlusion) in the larger coronary artery segments was detected by MDCT with a sensitivity of 70–98% (mean 87%) and a specificity of 0–37% (mean 21%). MDCT failed to identify three of ten left main stem stenoses. Conclusion. Sixteen-slice MDCT cannot routinely replace selective angiography for evaluation of coronary artery disease.

• Coronary artery disease
• multidetector computed tomography
• imaging

Angiography by selective catheterization is the standard method for coronary artery imaging. The procedure carries a risk of serious complications, i.e. death, 0.1–0.3%; myocardial infarction, 0.2%; stroke, 0.2%; vascular injury, 0.5% 1–4. High resolution imaging without cardiac catheterization has been obtained by multidetector spiral computed tomography (MDCT) 5, 6. This opens the possibility that coronary artery disease may be evaluated entirely by non-invasive means.

However, it is not clear whether the promising results reported for some patient series 7–9 will be replicated when MDCT is implemented in routine clinical practice. The Radial Artery BypAss GrAft Study of Tromsø (RABAGAST) provided an opportunity to address this question. The study was designed to evaluate our results with radial artery coronary bypass grafts (Bartnes et al., in preparation) and the feasibility of using MDCT to evaluate graft patency 10 as well as native vessels. Since the study conditions closely resembled the clinical setting, the merits of this method for routine diagnostics could be assessed.

Material and methods

The study was approved by the regional committee for medical research ethics, and the patients gave written informed consent. The patient population, imaging protocols and exclusion criteria have been described 10. Briefly, 45 patients were examined by ECG-gated 16-slice MDCT (Siemens’ Sensation 16 scanner) and conventional selective angiography (Philips’ Integris Allura Biplane) on the same day, 508–1135 (mean, 811) days after coronary artery bypass grafting. Visipaque 320 (GE Healthcare) was the contrast agent used in all imaging studies. The MDCT and selective angiography were interpreted by two experienced radiologists and cardiologists, respectively. The recorded results represented a consensus between each pair of investigators. Except from being blinded to the results of the parallel imaging modality, they had all relevant information available to them equal to a routine clinical setting. Each coronary artery segment was classified as open (without stenosis), stenotic (≥50% lumen diameter reduction), or occluded. Sensitivity, specificity, likelihood ratio, and positive and negative predictive values were calculated based on conventional selective angiography as the standard of reference.

Results

Evaluation of the proximal coronary artery segments and the left anterior descending branch of the left coronary artery (LAD) by MDCT and conventional angiography, respectively, are compared in Table I. MDCT correctly classified less than half of these segments as open, stenotic or occluded, respectively. Thus, in the 45 patients examined, a correct result was obtained for not more than 18 left main stems, 20 LADs, 19 proximal circumflex artery (Cx) segments, and 23 right main stems. MDCT failed to detect three of ten left main stem stenoses (30%), and five of 65 proximal arterial segments without significant pathology (8%) were deemed occluded by MDCT. Two of 57 occluded proximal segments (4%) were classified by MDCT as open and non-stenotic. For detection of significant pathology (stenosis ≥50% or occlusion) in proximal arterial segments, MDCT had a sensitivity of 70–98% and a specificity of 0–37%) (Table II). The likelihood ratios for MDCT-detected disease in these coronary artery segments were close to 1, which is the value at which a diagnostic test is useless (post-test likelihood does not exceed pre-test likelihood). Positive and negative predictive values with respect to detection of significant pathology for the various segments were 25–93% (mean 66%) and 43–81% (mean 66%), respectively.

Table I.  Assessment of the proximal coronary artery segments and LAD* by MDCT versus conventional selective angiography.

	Conventional selective angiography
Segment	Open	Stenotic	Occluded	Not assessable	All
MDCT
Left main
Open	13	3	0	0	16
Stenotic	19	5	0	0	24
Occluded	2	2	0	0	4
Not assessable	1	0	0	0	1
All	35	10	0	0	45
LAD
Open	0	0	0	0	0
Stenotic	2	14	18	0	34
Occluded	1	3	6	0	10
Not assessable	0	0	1	0	1
All	3	17	25	0	45
Proximal Cx^†
Open	3	3	1	0	7
Stenotic	11	14	5	1	31
Occluded	1	4	2	0	7
Not assessable	0	0	0	0	0
All	15	21	8	1	45
Right main^‡
Open	3	0	1	0	4
Stenotic	8	9	11	0	28
Occluded	1	0	11	0	12
Not assessable	0	0	1	0	1
All	12	9	24	0	45

^* The left anterior descending artery.

^† The segment of the circumflex artery which is proximal to the origin of the 1^st marginal branch.

^‡ The right coronary artery proximal to the crux.

Table II.  Test characteristics for MDCT detection of significant pathology (stenosis or occlusion) in the proximal coronary artery segments and LAD^§.

	Segment
	Left main	LAD	Proximal Cx	Right main
Sensitivity (%)	70	98	86	94
Specificity (%)	37	0	20	25
Likelihood ratio	1,1	0,98	1,1	1,3
Positive predictive value (%)	25	93	66	78
Negative predictive value (%)	81	-	43	75

^§ The left anterior descending artery.

The results for other coronary branches (diagonal, marginal, acute, posterolateral and posterior descending) largely resembled those of LAD and the proximal segments except that they more often were non-assessable due to smaller diameters. These data are not shown since the results for LAD and the proximal segments are sufficient to disqualify MDCT for routine coronary artery imaging.

Discussion

We find that the predictive power of 16-slice MDCT imaging of major coronary artery segments is poor. This mainly reflects a low specificity with respect to detection of significant disease. Whereas the observed specificity of <40% is considerably lower than the 94–99% reported in several studies 7–9, 11, 12, the observed sensitivity of 70–98% is in agreement with those reports. Coronary calcification is by far the major cause of false-positive findings with 16-slice MDCT 11 and might strongly contribute to the high false positive rate in our study population, since it includes 35 patients (78%) with triple vessel disease.

The study was conducted under conditions which closely resemble the routine clinical setting. The observed poor predictive capacity of MDCT demonstrates that at the current level of technology, this method should not be adopted for routine clinical decision-making except in centers able to consistently document a high correspondence between MDCT and conventional angiography. This contention is supported by the highly variable results reported in other studies 7–9, 11, 12, were e.g. the sensitivity ranged from 37% 14 to 96% 9.

Notably, in some patient series, high resolution MDCT angiography does permit non-invasive detection of coronary artery stenoses with high accuracy 12. Since agreement between MDCT and conventional angiography is better in patient subsets with low heart rates 9 or with limited coronary calcification 11, the variable merits of MDCT among radiological centers may reflect study population differences.

Although still largely unvalidated, technical improvements, especially the advent of 64-slice CT scanning, might bring MDCT into the realm of practical diagnostics 15.

Acknowledgements

The statistical advice of associate professor Bjørn Straume is greatly appreciated.