Meta-Analysis of Diagnostic Efficacy of 64-Slice Computed Tomography in the Evaluation of Coronary In-Stent Restenosis

We sought to conduct a meta-analysis using available studies to determine the diagnostic efficacy of 64-slice computed tomography (CT) in evaluation of in-stent restenosis (ISR). Sixty-four-slice CT allows optimal noninvasive assessment of coronary artery disease. However, a variety of artifacts limit evaluation of stented coronary segments. We included studies that used 64-slice CT for evaluation of coronary ISR. We pooled efficacy estimates across studies using random-effects models. We identified 14 studies, which included 895 patients (1,447 stents, mean diameter 3.1 mm). Of these, 1,231 (91.4%) stents were adequately assessed by 64-slice CT. Overall sensitivity was 91% (95% confidence interval [CI] 86 to 94), specificity was 91% (95% CI 89 to 92), positive predictive value (PPV) was 68% (95% CI 63 to 73), and negative predictive value (NPV) was 98% (95% CI 97 to 99). The summary receiver operating characteristic curves graph showed a symmetric area under the curve of 0.96. When nonassessable segments were included, overall sensitivity and specificity decreased to 87% (95% CI 81 to 92) and 84% (95% CI 82 to 87), with a PPV of 53% (95% CI 47 to 59) and an NPV of 97% (95% CI 96 to 98), respectively. In conclusion, 64-slice CT detects (high sensitivity and specificity) or excludes ISR (high NPV) with a high degree of confidence; however, precise quantification of ISR is not accurate (low PPV). Efficacy estimates are even lower when nonassessable segments are included. Hence, at this point, stress imaging remains the most acceptable noninvasive technique for diagnosis of ISR. (C) 2009 Elsevier Inc. All rights reserved. (Am J Cardiol 2009;103:1675-1681)