The trusted source for
healthcare information and
Accuracy of CT Coronary Angiography
Abstract & Commentary
By Michael H. Crawford, MD
Sources: Meijboom WB, et al. Diagnostic accuracy of 64-slice computed tomography coronary angiography. J Am Coll Cardiol. 2008;52:2135-2144; Nissen SE. Limitations of computed tomography coronary angiography. J Am Coll Cardiol. 2008;52:2145-2147.
Small studies with older equipment at single centers have shown that coronary CT angiography (CTA) is promising for evaluating coronary anatomy. New 64-slice CT scanners have improved spatial and temporal resolution. Thus, these investigators from the Netherlands conducted a prospective, multi-centered, multi-vendor study to assess the accuracy of 64-slice CTA for in dentifying significant coronary artery disease (CAD) in symptomatic patients undergoing conventional coronary angiography (CCA).
Patients eligible for the study were between ages 50-70 years, clinically stable, and without prior revascularization or renal dysfunction. Among 433 eligible patients, 360 gave consent and were able to complete CTA. The three sites each used a scanner from a different vendor. Patients were not excluded because of the presence of calcium in their coronary arteries. The reference standard was quantitative coronary angiography interpreted in a blinded fashion. A coronary luminal diameter reduction > 50% was considered significant.
Results: The patient's clinical presentations included typical angina (42%), atypical chest pain (23%), unstable angina (21%), and non-ST elevation myocardial infarction (14%). The percentage of patients with at least one coronary artery with a significant lesion by CCA was 68%. Beta blockers were given to 56% of the patients to reduce heart rate for CTA. One patient developed second-degree AV block after beta blockers. All but two patients (99%) with significant CAD by CCA were detected by CTA. The two missed had one-vessel disease. All patients with left-main or three-vessel disease were detected. Among the 114 patients without significant CAD, 41 (36%) were classified as significant by CTA. On a per-patient basis, sensitivity was 99%, specificity 64%, positive predictive value (PPV) 86%, and negative predictive value (NPV) 97%. On a per-vessel basis, sensitivity was 95%, specificity 77%, PPV 59%, and NPV 98%. CTA overestimated the extent of disease in 44% and underestimated it in 3%. The presence of vessel calcification increased sensitivity, but decreased specificity. The authors concluded that, in patients referred for CCA because of symptoms suggestive of ischemic heart disease, but without ST elevation on ECG, 64-slice CTA was reliable for excluding significant CAD but tended to overestimate the presence and severity of CAD.
This is an important and well done study. It is important because of the rapid deployment of CTA in the United States and the issue of health care costs. It is a strong study for several reasons. First, it is multi-centered. Prior studies mainly have been done in single centers. Second, each center used different equipment, which is based more in reality. Third, there were very few exclusion criteria. Notably, no one was excluded for poor images or the presence of coronary calcium. Fourth, there was no referral bias where the results of one test predicate who gets the second test. Here, everyone scheduled for CCA got a CTA.
Their CTA reading approach helps understand the results. They basically read for maximal sensitivity. Calcified segments were considered significantly diseased, and poor images were assumed to be significantly diseased segments. Thus, it is not surprising that sensitivity (99%) and the NPV (97%) were high, but the specificity (64%) and PPV (86%) were modest. They believed it better not to miss anyone, since positive studies would occasion further evaluation. In fact, there are now combination CT/SPECT machines so functional data can be obtained at the same setting to weed out the false positives. Nissen, in his editorial, points out that the further evaluation of all these false positives comes at a high cost. Many of these patients may have been better served by a functional test first; a stress echocardiogram is less costly and has no radiation exposure.
This study focused on intermediate- to high-risk patients, two-thirds of whom were proven to have significant CAD. The results would likely be different in low-risk patients. There are few studies in such patients because CCA is not always indicated in low-risk patients. Presumably, there would be fewer false positives, since fewer would have coronary calcium; false negatives in one study were 16%. Given these data, sensitivity and specificity would likely be similar to stress testing. In fact, the ACC/AHA guidelines suggest that CTA be considered an alternative to stress testing in selected patients. My own use of CTA is for patients in whom knowing whether they have CAD is important therapeutically, but the risk of an invasive test is not justified and a stress test is not considered accurate enough. For example, a pre-menopausal asymptomatic active woman with several male family members with premature CAD, an LDL-cholesterol of 128 mg/dL, and no other risk factors. A positive CTA would suggest the need for drug therapy to lower her LDL cholesterol.