Incorporating coronary calcification into pretest assessment of the likelihood of coronary artery disease: validation and recalibration of a new diagnostic tool

Abstract

Abstract Background A new clinical tool was recently proposed to improve the estimation of pre-test probability of obstructive coronary artery disease (CAD) by incorporating coronary artery calcium score (CACS) with clinical risk factors. This new model (Clinical+CACS) showed improved prediction when compared to the method recommended by the 2019 ESC guidelines on chronic coronary syndromes, but was never tested or adjusted for use in our population. The aim of this study was to assess the performance of this new method in a Portuguese cohort of symptomatic patients referred for coronary computed tomography angiography (CCTA), and to recalibrate it if necessary. Methods We conducted a two-center cross-sectional study assessing symptomatic patients who underwent CCTA for suspected CAD. Key exclusion criteria were age <30 years, known CAD, suspected acute coronary syndrome, or symptoms other than chest pain or dyspnea. Obstructive CAD was defined as any luminal stenosis ≥50% on CCTA. The Clinical+CACS prediction model was assessed for discrimination and calibration. A logistical recalibration of the model was conducted in a random sample of 50% of the patients and subsequently validated in the other half. Results A total of 1910 patients (mean age 60±11 years, 60% women) were included in the analysis. Symptom characteristics were: 39% non-anginal chest pain, 30% atypical angina, 19% dyspnea and 12% typical angina. The observed prevalence of obstructive CAD was 12.9% (n=247). Patients with obstructive CAD were more often male, were significantly older, had higher prevalence of typical angina and cardiovascular risk factors, and higher CACS values. The new Clinical+CACS tool showed greater discriminative power than the ESC 2019 prediction model, with a C-statistic of 0.83 (CI 95% 0.81–0.86) versus 0.67 (CI 95% 0.64–0.71), respectively (p-value for comparison <0.001). Before recalibration, the Clinical+CACS model underestimated the likelihood of CAD in our population across all quartiles of pretest probability (mean relative underestimation of 49%), which was subsequently corrected by the recalibration procedure - Figure. Conclusions In a Portuguese cohort of symptomatic patients undergoing CCTA for suspected CAD, the new Clinical+CACS model showed better discrimination power than the 2019 ESC method. The underestimation of the Clinical+CACS model was corrected by recalibrating it for our population. This new tool might prove useful for guiding decisions on the need for further testing. Funding Acknowledgement Type of funding sources: None.