CT technology lacks robustness in restenosis

Though the latest generation of 64-slice CT scanners often excels, the technology is still not good enough for practitioners to confidently assess in-stent restenosis, according to Dr. Stephan Achenbach, a professor of medicine at the University of Erlangen in Germany.

Speaking in July at the 2007 meeting of the Society of Cardiovascular Computed Tomography held in Washington, DC, Achenbach pointed to what he calls a surprising lack of research using CT to image stents, despite their widespread use. About 615,000 stent implants were performed in the U.S. in 2005 alone, he said.

Achenbach showed several images with the stents nicely outlined but cautioned that they are exceptions, not the rule. Artifacts from calcium and from the stent itself can lead to false-positive diagnoses, but the main reason for nondiagnostic images is the still limited temporal resolution of CT scanners and consequent artifacts caused by motion.

"If you have a data set that is entirely free of motion artifacts, you will often be able to evaluate stents," Achenbach said.

According to appropriateness criteria that SCCT members helped to develop, stent evaluation with CT is not an appropriate indication. Achenbach referred to a lecture presented at the same meeting by Dr. Allen Taylor, chief of radiology at Walter Reed Army Medical Center in Washington, DC. Taylor said these types of criteria are dynamic and tend to change as the technology advances.

Achenbach outlined select situations in which it may be possible to get diagnostic images of in-stent restenosis. The first consideration is to use a scanner with high temporal resolution and to do everything possible to control the heart rate. A second consideration is the patient's body mass index; images get harder to interpret as body mass increases. The size of the stent is another major area of concern. Smaller stents, 3 mm or less, tend not to image as well as those of 3.5 mm or more. Most stents used in the U.S. are on the smaller side, Achenbach said.

To maximize spatial resolution, he recommended image reconstruction with the smallest possible slice thickness. Using a sharp reconstruction kernel will often be helpful. Sixty-four-slice scanners create fewer blooming artifacts than 16-slice machines.

Stent type also plays a role, but so many types exist that it is difficult to point to one that is the best for imaging, Achenbach said.

Several studies testing the efficacy of 64-slice scanners to evaluate in-stent restenosis have shown sensitivities in the high 80% to low 90% range and specificity in the high 90%s. But the size of the stents varied from study to study, possibly skewing results one way or the other. One study that evaluated only stents in the left main coronary artery-often larger than 4 mm-clocked a sensitivity of 98%.

The other problem with many of these studies is that, while sensitivity and specificity were high, the positive predictive value ranged between 50% and 60%. In other words, if the cardiologist or radiologist sees in-stent restenosis, there is an up to 50% chance that it is not really there.

In closing, Achenbach asked the audience, tongue slightly in cheek, to remember that a good nuclear stress test or invasive catheter angiography is a very good "alternative" to cardiac CT.

Commenting after the session to Diagnostic Imaging, Dr. Andre J. Duerinckx, clinical director of radiology at MetroHealth Medical Center in Cleveland, said he appreciated Achenbach's balanced approach to the topic. Most significant was the meta-analysis that showed the moderate positive predictive value of cardiac CT to evaluate stents.

"It's an important message because many people might erroneously believe that you can evaluate stents, at least those implanted beyond two years," he said. "I'm sure we'll get there with CT technology, but right now you have to be so careful."