Acarix has announced the results from a new multicentre trial of its handheld CADScor system for non-invasive, non-radiation acoustic detection of coronary artery disease. The results were presented at the 2017 scientific sessions of the American College of Cardiology (ACC; 17-19 March, Washington, DC) and showed that the handheld CADScor system can rule out coronary artery disease with 97% negative predictive value.
The trial involved 1,675 patients from two Danish hospitals with a low to intermediate likelihood of coronary artery disease referred for Cardiac computed tomography (CT) angiography. Acarix has argued in a company release that this opens the possibility of use for the system as a frontline test, reducing patient anxiety and waiting times, as well as improving triage for patients needing more expensive and invasive diagnostic modalities.
The CADScor system combines acoustic detection of turbulent arterial flow and myocardial movement with advanced algorithms in a handheld device to provide a patient specific coronary artery disease score in less than 10 minutes. The research was led by principal investigator Morten Böttcher and by Simon Winther, Department of Cardiology, Aarhus University Hospital, Denmark.
“Despite the availability of improved risk stratification algorithms, the incidence of normal investigations such as nuclear or CT imaging remains high,” says Winther. “We therefore tested the diagnostic accuracy of the CADScor system for ruling out coronary artery disease to see if it could be used to reduce demand for more advanced diagnostic modalities. We have concluded that…this advanced, easy to use, stethoscope-like device could indeed be deployed as a frontline test.”
A coronary artery disease score was recorded in all 1,675 patients enrolled in the trial, with the coronary artery disease score algorithm including both acoustic features and clinical risk factors (gender, age and hypertension). Low risk was indicated by a coronary artery disease score value ≤20.
The third version of the coronary artery disease score algorithm was developed using recordings from 711 patients from previous studies and a training cohort of 589 patients from the present study. The remaining 1086 patients were used as the validation cohort. The coronary artery disease score was successfully analysed in 1464 (87%) patients. Haemodynamic significant coronary artery disease was present in 134 (9.3%) patients. There were no differences in the performance of the coronary artery disease score algorithm in the training vs. validation cohorts. In the entire cohort, the coronary artery disease score differed between coronary artery calcium score (CACS) groups CACS=0 (n=745), CACS 1-400 (n=550) and CACS > 400 (n=142); coronary artery disease score: 17 ±11, 24 ±12, and 30 ±12 (p< 0.001).
Coronary artery disease score was also significant lower for patients without vs. with haemodynamic significant stenosis 20 ±12 vs 30 ±12 (p< 0.001). Diagnostic performance evaluated by receiver operating characteristic curve showed an accuracy of: 72% (CI: 67% – 77%). Coronary artery disease score cut-off ≤20 had an accuracy of:
- Sensitivity: 81% (CI: 74% to 88%)
- Specificity: 53% (CI: 50% to 56%)
- PPV: 15% (CI: 13% to 18%)
- NPV: 97% (CI:95% to 98%)