This review concluded that 64-slice computed tomography detected or excluded in-stent restenosis with a high degree of confidence, but precise quantification of in-stent restenosis was not accurate. Small sample sizes, clinical variability across studies, potential for missed studies and lack of reporting of some important information make the reliability and generalisability of the results uncertain.

To determine the diagnostic accuracy of 64-slice computed tomography (CT) for evaluating coronary in-stent restenosis.

MEDLINE and The Cochrane Library were searched for English-language publications to November 2008; search terms were reported. Eligible studies were cross-referenced in Science Citation Index.

Any study that evaluated 64-slice CT exclusively for coronary in-stent restenosis (luminal diameter reduction over 50%) compared to conventional coronary angiography was eligible for inclusion if it reported sufficient data to extract or calculate diagnostic outcomes. Studies of patients who underwent coronary artery bypass grafting were included if stents were used in native coronary arteries only. Mean age of participants ranged from 54 to 68 years. The proportion of men was 64% to 92%. Mean stent diameter was 3.1mm. Where reported, volume of contrast agent used, scanner type and radiation dose varied across studies. The most commonly used rotation time was 330ms and collimation 0.6mm or 0.625mm.

The authors did not state how many reviewers screened studies for inclusion.

Study quality was assessed by two independent reviewers using 14-item QUADAS tool; discrepancies were resolved by a third reviewer.

Data to complete a 2x2 table of test performance were extracted on a per segment basis, from which sensitivity, specificity and positive and negative predictive values (PPV and NPV), with 95% confidence intervals (CI) were calculated. Positive and negative likelihood ratios (LR+ and LR-), and the diagnostic odds ratio were calculated.

Two reviewers independently extracted data; discrepancies were resolved by a third reviewer.

Pooled estimates of sensitivity, specificity, positive and negative predictive values, positive and negative likelihood ratios, and diagnostic odds ratios, with 95% confidence intervals (CI) were calculated using a random-effects meta-analysis. Heterogeneity was assessed using the Cochran Q and I² statistics. Analyses were conducted separately for all segments and assessable segments only. Summary receiver operating characteristic (sROC) curves were presented, using the Moses model. Meta-regression was used to assess the impact of potential (undefined) covariates. Publication bias was was assessed using the Deeks regression method.

Fourteen studies met the inclusion criteria (n=895 patients, 1,447 stents; range 25 to 182 patients). All studies scored at least 12 on QUADAS and four studies scored 14. The failed criteria related to patient spectrum (three studies), progression bias (three studies), blinding to index test results (one study), reporting of uninterpretable results (three studies) and withdrawals (three studies).

The number of assessable stents was 1,231 (proportion ranged from 58% to 100%), duration between stenting and CT ranged from 0.8 to 25.8 months and between CT and coronary angiography ranged between 2.5 and 58 days.

When including only assessable stents, sensitivity was 91% (95% CI 86% to 94%), specificity was 91% (95% CI 89% to 92%), positive predictive value was 68% (95% CI 63% to 73%), negative predictive value was 98% (95% CI 97% to 99%), positive likelihood ratio was 10.37 (95% CI 6.65 to 16.15), negative likelihood ratio was 0.14 (95% CI 0.09 to 0.21) and diagnostic odds ratio was 95.67 (95% CI 50.6 to 180.89). Significant heterogeneity was observed for the analyses of specificity and positive likelihood ratio. The area under the sROC curve was 0.96.

When including all stents (nine studies), sensitivity was 87% (95% CI 81% to 92%), specificity was 84% (95% CI 82% to 87%), positive predictive value was 53% (95% CI 47% to 59%), negative predictive value was 97% (95% CI 96% to 98%), positive likelihood ratio was 6.38 (95% CI: 3.31 to 12.29), negative likelihood ratio was 0.18 (95% CI 0.06 to 0.49) and diagnostic odds ratio was 51.61 (95% CI 10.24 to 259.89). Significant heterogeneity was observed for all analyses.

None of the variables used in the meta-regression had a significant impact on the diagnostic odds ratio. There was no evidence of publication bias.

64-slice CT detected (high sensitivity and specificity) or excluded (high negative predictive value) in-stent restenosis with a high degree of confidence. However, precise quantification of in-stent restenosis was not accurate (low positive predictive value); accuracy estimates were even lower when non-assessable stents were included.

The authors addressed a clear review question, supported by appropriate inclusion criteria. The search was not extensive, and non-English language publications and unpublished data were not sought. Therefore, studies could have been missed. Data extraction and the quality assessment was conducted in duplicate; it was unclear whether similar methods to reduce error and bias were used during study selection. Appropriate criteria were used to assess study quality and the results were reported for each criterion for each study. The proportion of segments in the meta-analyses that were occluded was unclear; occlusions are generally easier to diagnose and if the proportion is high, diagnostic accuracy may be overestimated. There was substantial statistical heterogeneity observed for most of the meta-analyses, which made reliability and generalisability of these pooled estimates uncertain. The number of participants in the included studies was generally small. The proportion of non-assessable stents generally ranged 0 and 15%; one study reported 42% as non-assessable. The generalisability of the primary results and conclusion of the review article may be limited by the lack of information on the type of stent used and stent diameters larger that those of the drug-eluting stents commonly used in clinical practice. In addition, a high sensitivity, specificity and negative predictive value along with low positive predictive value, indicated a low prevalence population. Manufacturers of the technology provided funding for the clinic who undertook the review. Stress imaging was stated as the most acceptable noninvasive technique for the diagnosis of in-stent restenosis; stress testing was not assessed in the review and, therefore, not compared directly to 64-slice CT. Given the limitations of both the included studies and the review, the conclusions should be interpreted with caution.

Practice: The authors stated use of 64-slice CT in an unselected population with stents was likely to have a low utility for the detection of in-stent restenosis and that stress imaging remained the most acceptable noninvasive technique for the diagnosis of in-stent restenosis.

Research: The authors stated that trials were required to assess the diagnostic value of multi-slice CT in patients after previous stenting, with outcomes recorded on a per patient basis and imaging of stented and non-stented segments undertaken.

Siemens Medical Solutions and Philips Medical Systems provides funding for the Cleveland Clinic where the authors are employed.

This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn.