Coronary Inflammation on CCTA as Predictor of Cardiac Events: the ORFAN Study

By Christina Lalani on

Key Points:

  • Out of all patients who undergo a CCTA, the highest number of cardiac events occurs in patients without obstructive CAD. 
  • In patients with non-obstructive CAD on CCTA, the use of a novel AI tool to quantify coronary inflammation results in accurate predictions of patients’ ten-year risk of cardiac mortality and MACE. 
  • In this study, patients with non-obstructive CAD on CCTA who have an inflammation score that is above the 75th percentile have an ~ 20 times higher risk of dying from a cardiac event over the next ten years.
  • The AI risk model used in this study results in the reclassification of the risk profile of ~ 40% of patients and was found to lead to changes in clinical management in roughly half the patients.

Coronary Computed Tomography Angiography (CCTA) is increasingly used as a tool for risk stratification in patients with suspected coronary artery disease (CAD). The 2021 ACC Guidelines for the diagnosis of chest pain offer a Class I recommendation to use CCTA to diagnose CAD, risk-stratify patients, and guide treatment decisions in intermediate-high risk patients with stable chest pain and no known CAD. In a late-breaking presentation at the 2023 American Heart Association conference, Dr. Charalambos Antoniades outlines the results of the Oxford Risk Factors and Non-Invasive Imaging (ORFAN) Study, which incorporates artificial intelligence (AI) into the interpretation of CCTA. 

The ORFAN study evaluated the use of an AI model that calculates a fat attenuation index score (FAI score) on CCTA to quantify coronary inflammation and predict patients’ overall risk of a cardiac event over the next ten years. The pathophysiologic basis of using fat attenuation indexing is based on the premise that arteries that have intraluminal inflammation will secrete molecules that activate lipolysis in the perivascular fat. The AI model used in this study evaluates the changes in the perivascular fat surrounding each coronary artery to calculate a FAI score that is corrected for age and gender. 

The ultimate aim of the ORFAN study is to consolidate data from 250,000 CCTA’s with lifelong outcomes data. Thus far, the study includes ~ 102,000 patients from the U.K. as well as ~ 33,000 patients from international sites. Out of the 40,091 patients that have been processed from the U.K. dataset, 18% were found to have obstructive CAD (> 70% stenosis) leading to 636 cardiac deaths and 1450 major adverse cardiac events (MACE) and 82% were found to have no obstructive CAD leading to 1118 cardiac deaths and 2857 MACE. The authors found that there was a 41% higher relative risk of cardiac mortality [HR, 95% CI: 1.41 (1.28-1.56)] and a 57% higher relative risk of MACE [HR, 95% CI: 1.57 (1.47-1.68)] in patients with obstructive CAD relative to patients with non-obstructive CAD. However, in absolute numbers, there were double as many cardiac events in patients with non-obstructive CAD compared to patients with obstructive CAD. 

In the second part of the study, the authors evaluated whether patients’ FAI scores could predict their future risk of cardiac events. They used a nested cohort of 3,666 patients from two ORFAN sites. Patients were excluded if they had congenital scans, cardiac transplant scans, poor image quality or opted out of NHS digital. In the final cohort of 3,393 patients with a median follow-up of 7.7 years, higher quartiles of FAI score in the left anterior descending (LAD) artery were associated with a higher relative risk of MACE and cardiac mortality. Patients who had a FAI score that was above the 75th percentile were found to have a greater than 20 times higher risk of dying from a cardiac event over the next 10 years [HR, 95% CI: 20.20 (11.49-35.53)] and a greater than 6 times higher risk of a MACE [HR, 95% CI: 6.76 (5.21-8.78)] compared to patients in the lowest quartile. This increase in relative risk with higher quartiles of LAD FAI score was also seen when patients were stratified by whether they had obstructive CAD and in the subset of patients with no plaque or calcium on CCTA. In patients with no plaque or calcium, having a FAI score above the 75th percentile was associated with a hazard ratio of 11.6 (95% CI: 3.51-38.21) for cardiac mortality compared to patients below the 25th percentile. 

In the third part of the study, the authors trained an AI prognostic model that included the FAI score, patient risk factors and the degree of stenosis to predict the overall risk of a cardiac event over the next ten years. In patients without obstructive CAD, there was a near perfect correlation between the predicted rate of events and actual rate of events. However, the model underestimated the risk of future cardiac events in the patients who had obstructive CAD on coronary CTA. In this AI risk model, ~ 30% of patients were classified as higher risk and ~ 10% of patients were classified as lower risk than they would otherwise be classified based on their CCTA. This change in risk stratification resulted in changes in clinical decision-making in approximately half of patients. Overall, the use of the AI risk model in patients with non-obstructive CAD on CCTA resulted in accurate predictions of the ten-year risk of MACE and cardiac mortality and had a meaningful impact on decision-making in patient care.