Clinical likelihood models calibrated against observed obstructive coronary artery disease by computed tomography angiography

Abstract Background Models predicting the likelihood of obstructive coronary artery disease (CAD) on invasive coronary angiography are available.(1,2) However, as stable de novo chest pain patients frequently have lower clinical likelihood and preferably undergo index testing by non-invasive tests such as coronary computed tomography angiography (CCTA)(3,4), clinical likelihood models calibrated against observed obstructive CAD at CCTA are warranted. Purpose To develop and validate CCTA-calibrated risk-factor-weighted and coronary artery calcium-score-weighted clinical likelihood models (i.e., RF-CL-CCTA and CACS-CL-CCTA models, respectively). Methods Based on age, sex, symptoms and cardiovascular risk factors, an advanced machine learning algorithm utilized a training cohort (n=38,269) of symptomatic outpatients with suspected obstructive CAD to develop both a RF-CL-CCTA and a CACS-CL-CCTA model to predict observed obstructive CAD at CCTA. The models were compared to a currently European Society of Cardiology-endorsed basic pre-test probability (Basic PTP) model(5), and findings were validated widely in a pooled validation cohort (n=26,141), including cohorts from United States of America, Brazil, Finland, United Kingdom, Singapore, and Denmark. Results Obstructive CAD at CCTA was present in 5,984 (22.9%) patients in the pooled validation cohort. While the Basic PTP underestimated the prevalence of observed obstructive CAD at CCTA, the RF-CL-CCTA and CACS-CL-CCTA models showed superior calibration (Figure 1). Compared to the Basic PTP model, the RF-CL-CCTA and CACS-CL-CCTA models showed superior discrimination (area under the receiver-operating curve (AUC) 0.71 (95% confidence interval (CI) 0.70-0.72) vs. 0.74 (95%CI 0.73-0.75) and 0.86 (95%CI 0.85-0.87), p<0.001 for both comparisons) (Figure 2). Conclusions CCTA-calibrated clinical likelihood models improve calibration and discrimination of observed obstructive CAD at CCTA.Figure 1:Calibration. Figure 2:Discrimination.