Diagnostic performance of clinical likelihood models of obstructive coronary artery disease to predict myocardial perfusion defects.

Aims: Clinical likelihood (CL) models are designed based on a reference of coronary stenosis in patients with suspected obstructive coronary artery disease. However, a reference standard for myocardial perfusion defects (MPDs) could be more appropriate. We aimed to investigate the ability of the 2019 European Society of Cardiology pre-test probability (ESC-PTP), the risk-factor-weighted (RF-CL) model, and coronary artery calcium score-weighted (CACS-CL) model to diagnose MPDs.
Methods and Results: Symptomatic stable de novo chest pain patients (n = 3374) underwent coronary computed tomography angiography and subsequent myocardial perfusion imaging by single-photon emission computed tomography, positron emission tomography, or cardiac magnetic resonance. For all modalities, MPD was defined as coronary computed tomography angiography with suspected stenosis and stress-perfusion abnormality in ≥2 segments. The ESC-PTP was calculated based on age, sex, and symptom typicality, and the RF-CL and CACS-CL additionally included a number of risk factors and CACS. In total, 219/3374 (6.5%) patients had an MPD. Both the RF-CL and the CACS-CL classified substantially more patients to low CL (<5%) of obstructive coronary artery disease compared with the ESC-PTP (32.5 and 54.1 vs. 12.0%, P < 0.001) with preserved low prevalences of MPD (<2% for all models). Compared with the ESC-PTP [area under the receiver-operating characteristic curve (AUC) 0.74 (0.71-0.78)], the discrimination of having an MPD was higher for the CACS-CL model [AUC 0.88 (0.86-0.91), P < 0.001], while it was similar for the RF-CL model [AUC 0.73 (0.70-0.76), P = 0.32].
Conclusion: Compared with basic CL models, the RF-CL and CACS-CL models improve down classification of patients to a very low-risk group with a low prevalence of MPD.
Competing Interests: Conflict of interest: L.D.R. acknowledges support from the Danish Cardiovascular Academy (grant number PD5Y-2023001-DCA), which is funded by the Novo Nordisk Foundation (grant number NNF20SA0067242) and The Danish Heart Foundation. S.W. acknowledges support from the Novo Nordisk Foundation Clinical Emerging Investigator grant (NNF21OC0066981). S.E.P. provides consultancy to Circle Cardiovascular Imaging Inc. M.B. discloses advisory board participation for Novo Nordisk, Astra-Zeneca, Pfizer, Boeringer Ingelheim, Bayer, Sanofi, Novartis, Amgen, CLS-Behring Acarix, and MEDtrace. All other authors declare no conflicts of interest.
(© The Author(s) 2023. Published by Oxford University Press on behalf of the European Society of Cardiology. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.)