Robin F Gohmann1, Konrad Pawelka2, Patrick Seitz3, Nicolas Majunke4, Linda Heiser3, Katharina Renatus2, Steffen Desch4, Philipp Lauten4, David Holzhey5, Thilo Noack5, Johannes Wilde4, Philipp Kiefer5, Christian Krieghoff3, Christian Lücke3, Sebastian Gottschling3, Sebastian Ebel2, Michael A Borger6, Holger Thiele7, Christoph Panknin8, Matthias Horn9, Mohamed Abdel-Wahab4, Matthias Gutberlet10. 1. Department of Diagnostic and Interventional Radiology, Heart Center Leipzig, Leipzig, Germany; Medical Faculty, University of Leipzig, Leipzig, Germany. Electronic address: robin.gohmann@gmx.de. 2. Department of Diagnostic and Interventional Radiology, Heart Center Leipzig, Leipzig, Germany; Medical Faculty, University of Leipzig, Leipzig, Germany. 3. Department of Diagnostic and Interventional Radiology, Heart Center Leipzig, Leipzig, Germany. 4. Department of Cardiology, Heart Center Leipzig at University of Leipzig, Leipzig, Germany. 5. Department of Cardiac Surgery, Heart Center Leipzig at University of Leipzig, Leipzig, Germany. 6. Department of Cardiac Surgery, Heart Center Leipzig at University of Leipzig, Leipzig, Germany; Leipzig Heart Institute, Leipzig, Germany. 7. Department of Cardiology, Heart Center Leipzig at University of Leipzig, Leipzig, Germany; Leipzig Heart Institute, Leipzig, Germany. 8. Siemens Healthcare GmbH, Erlangen, Germany. 9. Institute for Medical Informatics, Statistics and Epidemiology (IMISE), University of Leipzig, Leipzig, Germany. 10. Department of Diagnostic and Interventional Radiology, Heart Center Leipzig, Leipzig, Germany; Medical Faculty, University of Leipzig, Leipzig, Germany; Leipzig Heart Institute, Leipzig, Germany.
Abstract
OBJECTIVES: The purpose of this study was to analyze the ability of machine-learning (ML)-based computed tomography (CT)-derived fractional flow reserve (CT-FFR) to further improve the diagnostic performance of coronary CT angiography (cCTA) for ruling out significant coronary artery disease (CAD) during pre-transcatheter aortic valve replacement (TAVR) evaluation in patients with a high pre-test probability for CAD. BACKGROUND: CAD is a frequent comorbidity in patients undergoing TAVR. Current guidelines recommend its assessment before TAVR. If significant CAD can be excluded on cCTA, invasive coronary angiography (ICA) may be avoided. Although cCTA is a very sensitive test, it is limited by relatively low specificity and positive predictive value, particularly in high-risk patients. METHODS: Overall, 460 patients (age 79.6 ± 7.4 years) undergoing pre-TAVR CT were included and examined with an electrocardiogram-gated CT scan of the heart and high-pitch scan of the vascular access route. Images were evaluated for significant CAD. Patients routinely underwent ICA (388/460), which was omitted at the discretion of the local Heart Team if CAD could be effectively ruled out on cCTA (72/460). CT examinations in which CAD could not be ruled out (CAD+) (n = 272) underwent additional ML-based CT-FFR. RESULTS: ML-based CT-FFR was successfully performed in 79.4% (216/272) of all CAD+ patients and correctly reclassified 17 patients as CAD negative. CT-FFR was not feasible in 20.6% because of reduced image quality (37/56) or anatomic variants (19/56). Sensitivity, specificity, positive predictive value, and negative predictive value were 94.9%, 52.0%, 52.2%, and 94.9%, respectively. The additional evaluation with ML-based CT-FFR increased accuracy by Δ+3.4% (CAD+: Δ+6.0%) and raised the total number of examinations negative for CAD to 43.9% (202/460). CONCLUSIONS: ML-based CT-FFR may further improve the diagnostic performance of cCTA by correctly reclassifying a considerable proportion of patients with morphological signs of obstructive CAD on cCTA during pre-TAVR evaluation. Thereby, CT-FFR has the potential to further reduce the need for ICA in this challenging elderly group of patients before TAVR.
OBJECTIVES: The purpose of this study was to analyze the ability of machine-learning (ML)-based computed tomography (CT)-derived fractional flow reserve (CT-FFR) to further improve the diagnostic performance of coronary CT angiography (cCTA) for ruling out significant coronary artery disease (CAD) during pre-transcatheter aortic valve replacement (TAVR) evaluation in patients with a high pre-test probability for CAD. BACKGROUND: CAD is a frequent comorbidity in patients undergoing TAVR. Current guidelines recommend its assessment before TAVR. If significant CAD can be excluded on cCTA, invasive coronary angiography (ICA) may be avoided. Although cCTA is a very sensitive test, it is limited by relatively low specificity and positive predictive value, particularly in high-risk patients. METHODS: Overall, 460 patients (age 79.6 ± 7.4 years) undergoing pre-TAVR CT were included and examined with an electrocardiogram-gated CT scan of the heart and high-pitch scan of the vascular access route. Images were evaluated for significant CAD. Patients routinely underwent ICA (388/460), which was omitted at the discretion of the local Heart Team if CAD could be effectively ruled out on cCTA (72/460). CT examinations in which CAD could not be ruled out (CAD+) (n = 272) underwent additional ML-based CT-FFR. RESULTS: ML-based CT-FFR was successfully performed in 79.4% (216/272) of all CAD+ patients and correctly reclassified 17 patients as CAD negative. CT-FFR was not feasible in 20.6% because of reduced image quality (37/56) or anatomic variants (19/56). Sensitivity, specificity, positive predictive value, and negative predictive value were 94.9%, 52.0%, 52.2%, and 94.9%, respectively. The additional evaluation with ML-based CT-FFR increased accuracy by Δ+3.4% (CAD+: Δ+6.0%) and raised the total number of examinations negative for CAD to 43.9% (202/460). CONCLUSIONS: ML-based CT-FFR may further improve the diagnostic performance of cCTA by correctly reclassifying a considerable proportion of patients with morphological signs of obstructive CAD on cCTA during pre-TAVR evaluation. Thereby, CT-FFR has the potential to further reduce the need for ICA in this challenging elderly group of patients before TAVR.
Authors: Hendrik Wienemann; Marcel C Langenbach; Victor Mauri; Maryam Banazadeh; Konstantin Klein; Christopher Hohmann; Samuel Lee; Isabel Breidert; Alexander Hof; Kaveh Eghbalzadeh; Elmar Kuhn; Marcel Halbach; David Maintz; Stephan Baldus; Alexander Bunck; Matti Adam Journal: J Cardiovasc Dev Dis Date: 2022-04-14
Authors: Robin Fabian Gohmann; Patrick Seitz; Konrad Pawelka; Nicolas Majunke; Adrian Schug; Linda Heiser; Katharina Renatus; Steffen Desch; Philipp Lauten; David Holzhey; Thilo Noack; Johannes Wilde; Philipp Kiefer; Christian Krieghoff; Christian Lücke; Sebastian Ebel; Sebastian Gottschling; Michael A Borger; Holger Thiele; Christoph Panknin; Mohamed Abdel-Wahab; Matthias Horn; Matthias Gutberlet Journal: J Clin Med Date: 2022-02-28 Impact factor: 4.241