392Systematic review and meta-analysis on the diagnostic accuracy for the detection of relevant coronary artery stenosis of vasodilator myocardial perfusion CMR and dobutamine stress echocardiography

Abstract Background Relevance of coronary artery stenosis in patients with stabile coronary artery disease (SCAD) is defined by myocardial ischemia due to flow limitation. While FFR-guided treatment of SCAD is a class IA recommendation. The initial risk stratification with detection of relevant CAD can be facilitated by several myocardial imaging methods without any preference mentioned in current guidelines. Objectives This study aimed to systematically assess and to compare the diagnostic accuracy of vasodilator myocardial perfusion cardiovascular magnetic resonance imaging (pCMR) and dobutamine stress echocardiography (DSE) for the non-invasive detection of relevant SCAD through a meta-analysis, to enable an evidential preference in risk stratification. In contrast to previously published work, this meta-analysis explicitly included only studies with rigorous eligibility criteria and a narrowly prespecified definition of their invasive reference tests. Selection criteria A study was included if (1) CCA or FFR was used as a reference standard for diagnosing relevant SCAD, defined as >70% stenosis or a value <0.80 on FFR recordings, respectively; (2) sufficient data to permit analysis and to reconstruct contingency tables (explicitly true-positive, false-positive, false-negative and true-negative findings) was provided; (3) there was a minimal sample size of 20 patients; (4) assessment of myocardial perfusion reserve was performed using vasodilators adenosine or regadenoson for pCMR, and dobutamine used for echocardiography; and (5) the studies were of prospective design. Data collection and analysis: From the 5,634 studies identified, 1,306 relevant articles were selected after title screening. Just 47 fulfilled all inclusion criteria on full-text review, resulting in a total sample size of 4,742 patients. Data extraction was performed for each study by two reviewers independently.Pooled analysis was performed based on a random effects models. Results The sensitivity, specificity and diagnostic odds ratio (DOR) for pCMR were 0.88 (95% confidence interval (CI): 0.85–0.90), 0.84 (95% CI: 0.81–0.87), and 38 (95% CI: 29–49), and for DSE 0.72 (95% CI: 0.61–0.81), 0.89 (95% CI: 0.83–0.93), and 20 (95% CI: 9–46), respectively. Post-test probability was augmented by positive (likelihood ratio) LR of 5.5 (95% CI: 4.7–6.5) and negative LR of 0.14 (95% CI: 0.12–0.18) based on Bayes' theorem, as compared to LR of 6.3 (95% CI: 3.8, 10.4) and negative LR of 0.31 (95% CI: 0.21, 0.46) for DSE. The size of the prediction region on the hierarchical summary receiver operating characteristic (HSROC) plot for pCMR (0.29; 95% CI 0.11–0.77) was significantly smaller compared to the one of DSE (1.07; 95% CI 0.27–4.19; p<0.01). Forrest plot pCMR Conclusion The results of this systematic review and meta-analysis show that pCMR is characterized by a superior diagnostic test accuracy of relevant SCAD compared to DSE and that it can refine the post-test probability of SCAD. Acknowledgement/Funding European Heart Academy of the European Society of Cardiology