Atherosclerosis, Cardiovascular Diseases, Coronary Artery Calcification
Conditions
Keywords
coronary artery calcium, chest x-ray, atherosclerotic cardiovascular disease, primary prevention, risk prediction, artificial intelligence
Brief summary
The AI-CAC model is an artificial intelligence system capable of assessing the presence of subclinical atherosclerosis on a simple chest radiograph. The present study will provide prospective validation of its diagnostic performance in a primary prevention population with a clinical indication for coronary artery calcium (CAC) testing.
Detailed description
The AI-CAC-PVS project is a prospective, multicenter, single-arm clinical study, with enrollment at 5 Radiology Units in Piedmont (Italy). Consecutive individuals without prior reported cardiovascular events referred for a non-contrast chest CT for the assessment of coronary artery calcium (CAC) score for cardiovascular risk stratification purposes will be considered for inclusion in the study. Individuals who agree to participate in the study will undergo a standard chest radiograph, as the only deviation from clinical practice. The CAC score will be calculated on chest CT scans according to international standards, and the result will be provided to the patient. Any subsequent changes in behavioral habits, lipid-lowering, antiplatelet, antihypertensive, and antidiabetic therapies prescribed by the attending physician will be collected in a dedicated dataset, along with the occurrence of cardiovascular events at the last available follow-up. The AI-CAC model will be applied to the chest radiograph, yielding an AI-CAC value as output. The patient, radiologist, and attending physician will not be informed of the AI-CAC value until the end of the study. The primary outcome will be the accuracy of the AI-CAC model to detect the presence of subclinical atherosclerosis on chest x-ray as compared to the CT scan (i.e. CAC \>0). The ability to predict clinical outcomes at follow-up (ASCVD, atherosclerotic cardiovascular disease events comprising myocardial infarction, ischemic stroke, coronary revascularization and cardiovascular death) will be assessed as exploratory secondary outcome.
Interventions
DIAGNOSTIC_TESTAI-CAC score
Deep-learning based prediction of the coronary artery calcium score with a plain chest x-ray
Sponsors
Compagnia di San Paolo
A.O.U. Città della Salute e della Scienza
Study design
Allocation
NA
Intervention model
SINGLE_GROUP
Primary purpose
PREVENTION
Masking
NONE
Eligibility
Sex/Gender
ALL
Age
40 Years to 75 Years
Healthy volunteers
No
Inclusion criteria
* Consent to participate in the study * Age between 40 and 75 years * Clinical indication from the treating physician to undergo chest CT for CAC score evaluation
Exclusion criteria
* Prior cardiovascular events (myocardial infarction, coronary revascularization, transient ischemic attack, stroke, symptomatic peripheral vascular disease, arterial revascularization of peripheral districts) * Cancer or other chronic diseases with an estimated prognosis of less than five years * Technical contraindications to the execution of chest CT with electrocardiographic gating (highly penetrant atrial fibrillation, frequent ventricular extrasystoles)
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| Diagnostic accuracy of the AI-CAC score to identify the presence of subclinical atherosclerosis on chest x-ray | Through study completion (anticipated average follow-up of 1 year). | Diagnostic accuracy of the AI-CAC score to identify the presence of subclinical atherosclerosis (i.e. AI-CAC \>0) on chest x-ray as compared to CAC measured on a non-contrast ECG-gated CT scan (i.e. CAC \>0). The area under the curve (AUC) method will be used to evaluate the primary outcome. |
Secondary
| Measure | Time frame | Description |
|---|---|---|
| Percentage of individuals with a therapeutic management change by the attending physician based on the CAC score, with concordant AI-CAC. | Through study completion (anticipated average follow-up of 1 year). | Potential impact on the implementation of primary prevention strategies: i.e. percentage of individuals with a therapeutic management change by the attending physician (increase or decrease in lipid-lowering therapy, initiation or discontinuation of antiplatelet therapy, behavioral measures) based on the CAC score, with concordant AI-CAC. |
| Comparison of ASCVD events occurring in patients without (AI-CAC=0) vs. with subclinical atherosclerosis (AI-CAC >0) based on the AI-CAC score, as assessed by Kaplan Meier estimates of ASCVD events occurring until study completion. | Through study completion (anticipated average follow-up of 1 year). | Predictive ability of the AI-CAC score for the incidence of adverse cardiovascular events (myocardial infarction, stroke, cardiovascular death, or coronary revascularization) at the last available follow-up. |
Contacts
Primary ContactFabrizio D'Ascenzo, MD
Outcome results
None listed