1. Home>
2. Specialties>
3. Cardiology>
4. Summary and Comment

CT Angiography: High False-Positive Rate

In an international study, multislice CT had high sensitivity but low specificity for identifying significant coronary stenosis, compared with the gold standard of invasive angiography.

In a recent small study from Germany, 16-slice computed tomography had high sensitivity and specificity for predicting obstructive coronary artery disease, compared with gold-standard invasive angiography (Journal Watch Cardiology Jun 24 2005). Now, we have data from an industry-funded multinational study comparing 16-slice CT angiography (CTA) with conventional invasive angiography, conducted within 2 weeks after CTA, in 187 patients with suspected coronary artery disease (CAD) at 11 centers. About 70% of 1629 CTA-imaged segments were deemed evaluable by independent core laboratories, where all scans were assessed.

In segment-based analyses, CTA had a sensitivity of 89%, a specificity of 65%, a positive predictive value (PPV) of 13%, and a negative predictive value (NPV) of 99% for detecting >50% stenosis, compared with gold-standard angiography. In patient-based analyses, CTA had the following results: 98% sensitivity, 54% specificity, 50% PPV, and 99% NPV.

Comment: In this study, interpretable 16-slice CTA scans were effective in ruling out >50% stenosis in CAD patients (sensitivity, 98%). However, false-positives were common, and roughly 30% of CTA images were not evaluable in the first place. Whether newer 64-slice scanners will have similar results remains to be determined. The current results suggest that CTA should not be a first screening test of choice. The authors suggest limited use of multislice CTA in patients with undetermined or false-positive stress-test results.

— JoAnne M. Foody, MD

Published in Journal Watch Cardiology August 23, 2006

Citation(s):

Garcia MJ et al. Accuracy of 16-row multidetector computed tomography for the assessment of coronary artery stenosis. JAMA 2006 Jul 26; 296:403-11.

• Original article (Subscription may be required)
• Medline abstract (Free)