Cardiac CT does not stack up to MPI for ischemia detection

 
 
 
 - MDCT perfusion
60-year-old woman with obstructive coronary artery disease in left anterior descending and left circumflex territories. Commercially available time-volume analysis tool was used to evaluate subendocardial attenuation. Automated left ventricular wall segmentation is adjusted manually to select subendocardium (blue outline).
Source: (AJR 2013;200:337-342)
 

Resting cardiac CT performed on a 64-slice system does not reliably detect myocardial ischemia compared with radionuclide myocardial perfusion imaging (MPI), according to a study published in the February issue of the American Journal of Roentgenology .

The study found resting cardiac 64-slice CT did not reliably show differential perfusion by visual or semiautomated analysis, according to Ari J. Spiro, MD, of Montefiore Medical Center, Bronx, N.Y., and colleagues.

“Although different attenuations were subjectively detected on visual inspection of some segments, the performance characteristics of visual analysis were not suitable for clinical diagnosis,” wrote the authors.

The standard is for ischemia to be assessed using radionuclide MPI. However, Spiro and colleagues explained that interest in the feasibility of CT detection of myocardial perfusion has been growing. Resting CT has the advantage of reduced imaging time, radiation dose and patient discomfort, and it has been hypothesized that CT may detect small areas of differential blood flow due to its high spatial resolution.

To assess the viability of resting CT for this purpose, the authors retrospectively identified 35 patients with myocardial ischemia found on MPI who underwent retrospectively gated CT within 90 days of MPI. Perfusion on CT was evaluated using both a visual and automated method, with a subendocardial perfusion of the standard 17 American Heart Association segments measured with a commercial tool in both systole and diastole for the automated analysis.

A total of 580 myocardial segments were evaluated, and 152 were ischemic on MPI. Compared with MPI, visual analysis had a sensitivity of 16 percent, specificity of 92 percent, positive predictive value (PPV) of 40 percent and a negative predictive value (NPV) of 75 percent in systole, reported Spiro and colleagues. In diastole, visual analysis had a sensitivity of 18 percent, specificity of 89 percent, PPV of 37 percent and NPV of 75 percent.

Results showed no significant difference in subendocardial perfusion between ischemic and nonischemic segments by the automated method. There was no significant difference in CT perfusion between patients with and without obstructive coronary artery disease on CT angiography using the visual or automated methods. “This finding suggests that visual analysis of resting CT datasets for perfusion abnormalities in areas of myocardium subtended by stenosed vessels may also be inaccurate,” wrote the authors.

Spiro and colleagues noted that while 64-slice CT is unsuitable for clinical use in revealing ischemia seen on MPI, ongoing investigations using 256- and 320-slice scanners with stress protocols or dual-energy CT may more effectively close the accuracy gap between CT and MPI.