NASCI 2015 San Diego, CA Evidence Based Radiology in Practice Stress Perfusion MRI vs. CTA in Recent Diagnostic Accuracy Studies Andrew E. Arai, MD, FAHA Chief, Advanced Cardiovascular Imaging National Heart, Lung and Blood Institute National Institutes of Health, DHHS Bethesda, MD Disclosures: Research Agreements with Siemens, Bayer, & Toshiba Views do not necessarily reflect those of the US Government FDA: Off-label uses of Gadolinium-based contrast agents
Aims of this presentation Review the diagnostic accuracy of CT Coronary Angiography (CTA) in various types of clinical trials. I will not address the accuracy of CTP or CT-FFR although I will use CTA data from those studies. Review the diagnostic accuracy of CMR stress perfusion studies relative to other stress perfusion modalities.
What is the sensitivity and specificity of CTA for significant coronary artery disease? a. Sensitivity ~93% Specificity ~80% b. Sensitivity ~93% Specificity ~96% c. Sensitivity 83-99% Specificity 64-91% d. Sensitivity 91-96% Specificity 21-88% e. Sensitivity 84-94% Specificity 25-42%
Diagnostic Accuracy of 64 slice CTA for CAD Sensitivity: 86 to 93% Specificity: 96%
Sensitivity (%) 2008 Diagnostic Accuracy of CTA for CAD 100 80 60 2008 AHA Statement 64-slice CT scanners 40 20 0 0 20 40 60 80 100 100 - Specificity (%) Sensitivity 73-100% meta-analysis 93% meta-analysis 86% Specificity 96-98% meta-analysis 96%
Multicenter Clinical Trials CTA in 2008 CORE64 Miller JM et al. NEJM 2008 Sensitivity 83% Specificity 91% ACCURACY Budoff MJ et al. JACC 2008 Sensitivity 95% Specificity 83% Multicenter, Multivendor Meijboom WB et al. JACC 2008 Sensitivity 99% Specificity 64%
Diagnostic Accuracy of CTA in studies of CTP Feuchtner G et al. Circ CVI 2011 Sensitivity 96% Specificity 88% Blankstein R et al. JACC 2009 Sensitivity 91% Specificity 73% Core320 Rochitte C et al. Eur Heart J 2013 Sensitivity 92% Specificity 51% Ko BS et al. Eur Heart J 2012 Sensitivity 93% Specificity 21%
Diagnostic Accuracy of CTA in a CT-FFR Study DISCOVER-FLOW Koo BK et al. JACC 2011; 58(19): 1989-97 CTA Per Vessel Sensitivity 91% Specificity 40% CTA Per Patient Sensitivity 94% Specificity 25%
Diagnostic Accuracy of CTA in studies of CT-FFR DISCOVER-FLOW Koo BK et al. JACC 2012 Sensitivity 94% Specificity 25% DEFACTO Min JK et al. JAMA 2012 Sensitivity 84% Specificity 42% NXT Trial Norgaard BL et al. JACC 2014 Sensitivity 94% Specificity 34%
Sensitivity (%) Diagnostic Performance of CTA in Clinical Trials 100 80 60 40 20 2008 AHA Statement Multicenter CTA Trials CTA in CTP Trials CTA in CT-FFR Studies 0 0 20 40 60 80 100 100 - Specificity (%)
Factors that could explain changes in CTA diagnostic accuracy for CAD Prior to 2008, many studies focused on less severe CAD. Include lower risk patients (easier to justify for early studies) Exclude high calcium Exclude stents Exclude CABG Exclude bad image quality Strategically, a read for high sensitivity works well since many studies are clearly normal & CTA was recognized as easy to read when normal (i.e. high negative predictive value).
Factors that could explain changes in CTA diagnostic accuracy for CAD COURAGE, FAME, And FAME II lead to a situation where coronary intervention was better managed by the physiological significance rather than angiographic appearance of stenosis. Invasive FFR as a reference standard proved many angiographic stenoses are not physiologically significant. Past true positives became false positives (i.e. decreased specificity)
Factors that could explain changes in CTA diagnostic accuracy for CAD Stress CT Perfusion might be useful in patients with intermediate stenoses, high calcium, or stents Stress CTP might also determine which stenoses are physiologically significant CT FFR might determine physiological significance of stenosis without a stress test Including patients with high calcium scores in the CTP and CT-FFR trials could have eroded specificity of CTA. Including invasive FFR as part of the reference standard can also be predicted to lower the specificity of anatomic assessment of stenosis
Visual interpretation of CTA should evolve to understand the factors that make CT-FFR differ from stenosis severity Wang R et al. JCCT 2015
What is the sensitivity and specificity of CTA for significant coronary artery disease? a. Sensitivity ~93% The best Specificity answer ~80% is: b. Sensitivity ~93% Specificity ~96% a. Sensitivity 93% and Specificity 80%. c. Sensitivity 83-99% Specificity 64-91% d. Weighted Sensitivity combined 91-96% average Specificity of (n=878): 21-88% Miller JM et al. NEJM 2008 e. Sensitivity 84-94% Specificity 25-42% Budoff MJ et al. JACC 2008 Meijboom WB et al. JACC 2008 but the reference standard is a 50% stenosis by QCA and a per patient analysis. There is strong need for trials versus invasive FFR.
CTP Takx RAP et al. Circ CV Imaging 2015
CTP Takx RAP et al. Circ CV Imaging 2015
Post-Test Probability of Disease vs Pre-Test Probability Takx RAP et al. Circ CV Imaging 2015
CTP Takx RAP et al. Circ CV Imaging 2015
Summary ROC : Per Patient Analysis Takx RAP et al. Circ CV Imaging 2015
Douglas PS et al. NEJM 2015 Apr 2;372(14):1291-300
Conclusions CTA has high sensitivity for CAD (~93%) CTA specificity for CAD depends on many factors but I d quote 80% unless your readers over-call in which case it might be lower. Of the various stress tests, CMR and PET are the most sensitive (~84-88%) and specific (~87%) for detecting physiologically significant coronary stenoses.?time for a case?
Clinical Case of Atypical Chest Pain 49-yr-old male with exertional chest discomfort and dyspnea. The chest pain resolves with rest. However, his chest discomfort sometimes resolves even if he keeps walking. He never had nitroglycerin. Actively exercises for 30 minutes > 3 times per week. No prior history of heart disease Hyperlipidemia Positive family history Ex-smoker (quit 9 years ago)
Stress/Rest Gated Sestamibi SPECT
49-yr-old male with exertional chest discomfort and dyspnea has an equivocal stress SPECT What would you do next? Reassure the patient Stress Echo Stress MRI CT Coronary Angiography Invasive Coronary Angiography
CT Coronary Angiography - LAD CAC = 116 mostly in the LAD Marcus Chen, MD NHLBI/NIH
CT Coronary Angiography 1 st Diagonal Marcus Chen, MD NHLBI/NIH
CT Coronary Angiography - Circumflex Marcus Chen, MD NHLBI/NIH
CT Coronary Angiography - RCA Marcus Chen, MD NHLBI/NIH
CMR: cine MRI long axis views
Stress/Rest Myocardial Perfusion Dipyridamole Rest
Stress/Rest Myocardial Perfusion Dipyridamole Rest 9 of 16 segments had moderate or severe perfusion defects
SAX Delayed Enhancement
Cath correlation
Cath correlation
CMR: cine MRI short axis views
Concordance and diagnostic accuracy of vasodilator stress cardiac MRI and 320-detector row coronary CTA Chen MY et al. Int J Cardiovasc Imaging 2013
NXT Trial Excluded 141 of 365 subjects Screened (365) CTA Accepted by Core Lab (310) Final sample size (251) Excluded After CTA (8) BMI >35 (4) No CTA stenosis (2) Afib (1) CTA rejected by Core Lab (47) Image artifacts (44) Prior coronary stent (1) Incomplete data (2) Excluded after CTA Acceptance (56) No invasive FFR (11) By ICA/FFR Core Lab (22) Other (23) Missing coronary stenosis by CTA (3) At least 49 Excluded for reasons that might be common in routine CTA
Diagnostic Accuracy of CTA & CT-FFR in the NXT Trial NXT Trial Per Patient Analysis Sensitivity 0.94 FFR 0.80 FFR > 0.80 Specificity 0.34 CTA > 50% 75 115 PPV 0.39 CTA 50% 5 59 NPV 0.92 Accuracy 0.53 NXT Trial Per Vessel Analysis Sensitivity 0.83 FFR 0.80 FFR > 0.80 Specificity 0.60 CTA > 50% 83 154 PPV 0.35 CTA 50% 17 230 NPV 0.93 Accuracy 0.65 Norgaard BL et al. JACC 2014
Diagnostic Accuracy of CTA & CT-FFR in Core320 Core320 Sensitivity 0.92 Specificity 0.51 PPV 0.53 NPV 0.92 AUC 0.84 DEFACTO Sensitivity 0.84 Specificity 0.42 PPV 0.61 NPV 0.72 AUC 0.64 Rochitte C. et al. Eur Heart J 2013 Min JK et al. JAMA 2012
Sensitivity (%) 2008 Diagnostic Accuracy of CTA for CAD 100 80 60 CTA Multi-center Clinical Trials 40 20 2008 AHA Statement 0 0 20 40 60 80 100 100 - Specificity (%)