1396057 - Breathing-Enhanced Oxygenation-Sensitive CMR for Detecting Significant Coronary Artery Stenosis – First Multi-Center Data

The clinical assessment for the presence of regional or global coronary vascular dysfunction (both epicardial and “microvascular”) typically uses surrogate markers based on anatomy (degree of stenosis) or flow (perfusion, Fractional Flow Reserve) instead of markers for the actual mismatch of oxygen supply and demand. Oxygenation-Sensitive CMR (OS-CMR), utilizing the specific impact of tissue oxygenation and blood flow on myocardial T2*, has been validated to assess dynamic changes in myocardial oxygenation (1). As recently shown, the myocardial oxygenation changes triggered by a vasoactive breathing maneuver allow for safely measuring global and regional coronary vascular function in patients with coronary artery disease (CAD) (2). The Breathing-induced Myocardial Oxygenation REserve (B-MORE), reflecting the vascular capacity of the assessed myocardium, has since been applied in several clinical scenarios. So far, multi-center data, however, are missing.



Methods:

We analyzed data from two short-axis OS-CMR slices obtained on a clinical 1.5T or 3T MRI scanner (Siemens, GE Healthcare, or Philips) in 62 healthy volunteers and 48 patients being clinically investigated for obstructive CAD enrolled in the international BMORE study. CAD was defined as having ≥50% stenosis in the left main or ≥75% stenosis in a single vessel on quantitative coronary angiography. The images were analyzed using a prototype tissue oxygenation module (Area19 Medical, Montreal, Canada). OS-CMR data was derived from end-systole and end-diastole. To minimize the impact of confounding effects of vendor, field strength, and participant dilutional state BMORE biomarkers were derived after normalizing the OS-CMR signal intensity to both the left and right ventricular blood pools and by a trained machine learning algorithm.



Results:

Demographics are shown in Figure 1. Of the 48 patients, 13 had no obstructive CAD, 16 had single vessel disease, 11 had two-vessel disease, and 8 had triple-vessel disease. There was a significant difference between global B-MORE normalized to left and right ventricular blood pools of healthy volunteers (13.9±15.1) and all patient participants (3.7±5.6). While there was no significant difference between myocardial coronary territories downstream of coronary arteries with vs without significant stenosis, a machine-learning informed analysis yielded an 89% diagnostic accuracy in detecting coronary territories exposed to significant stenosis. Fig. 2 shows typical findings from the assessed groups.



Conclusion:

In a preliminary analysis of the international B-MORE trial, we have compared B-MORE to coronary angiography. While the global oxygenation response was significantly different from that of healthy volunteers, correct identification of a coronary territory subject to significant stenosis was not possible using a simple signal intensity measurement. A novel machine-learning-based algorithm, however, could identify such affected coronary artery territories with high diagnostic accuracy.