BALTIMORE – CT angiography is not accurate enough to assess in-stent restenosis, compared with invasive coronary angiography, based on the results of a meta-analysis.
In 11 studies with data on the patient level (804 patients with 1,295 stents), the pooled sensitivity of CT was 94% and the specificity was 92%. In six studies using 64-row or greater CT, the pooled sensitivity of CT was 94% and specificity of CT was 94%. However, the difference compared with all 11 patient-level studies was not significant.
They also looked at the influence of stent size on CT accuracy in eight studies with 1,246 stents. Sensitivity was 81% for stents less than 3 mm and 92% for stents at least 3 mm in size. Specificity was 84% for stents less than 3 mm and 98% for stents at least 3 mm in size.
"In other words, the type of stent and the characteristics that the patient brings with him or her is much more important than the type of scanner that you have sitting there," said Dr. Marc Dewey of Charité – Universitätsmedizin Berlin. The findings were presented at the annual meeting of the Society of Cardiovascular Computed Tomography.
The researchers included prospective studies using at least 12-row CT scanners. Patients were assessed for 50% diameter in-stent restenosis as clinically relevant. All patients had to have coronary angiography.
The researchers included several databases: MEDLINE, EMBASE, ISI’s Web of Science (now Thomson Reuters Web of Knowledge) and the Cochrane Library. Search terms included "computed tomography," "coronary angiography," and "stent." Overall, they identified 33 studies with stent-level data.
Based on likelihood ratios, "we couldn’t conclude that CT was reliable in ruling in and ruling out coronary stent restenosis," said Dr. Dewey.
The investigators also looked at nondiagnostic studies. There were 30 studies with information on the stent level and 17 studies with information on the patient level. They used a missing data technique to assess for less-than-perfect reporting on the study level. The nondiagnostic rate on the patient level was 6% on average, based on the reported results, but the confidence intervals went up to more than 13%. "So almost every six or seven patients would be nondiagnostic."
"Coronary stent imaging by CT is slightly better using 64-row CT and above, though this was not significant. On the other hand, patients with stents that were at least 3 mm or larger could be significantly better assessed. Specificity in particular was significantly improved," said Dr. Dewey.
On the basis of these findings, CT should not be recommended for the assessment of coronary stent restenosis.
The study was funded by the German Research Foundation. The investigators did not report whether they have any significant conflicts of interest.