Fractional flow reserve (FFR) is an invasive procedure evaluating the functional significance of an intermediate coronary stenosis in patients with coronary artery disease (CAD)¹. Quantification of pressure gradient (ΔP) across a particular stenosis is the key to the determination of FFR, where FFR<0.80 is considered a functionally significant stenosis. Noninvasive ΔP measurement (ΔP_MR) using phase-contrast (PC)-MRI in conjunction with Navier-Stokes (NS) equations has been attempted in large to medium sized vessels^2-4. Our previous work has shown the feasibility of deriving ΔP_MR in small caliber phantom models and healthy coronary arteries^5-6. To ensure the accuracy of the proposed method, this study aimed to investigate the reproducibility of ΔP_MR in stenotic phantom models at various diameters and its correlation with measured ΔP values via a pressure transducer (ΔP_PT). The feasibility of the method was investigated in diseased coronary arteries and compared to healthy controls.

Phantom studies: 11 phantom models (0%-85% area stenosis, reference diameter=4.8mm) were individually connected to a flow pump (gadolinium-doped water, constant volume velocity=250mL/min) while 2D cross-sectional PC-MRI images were acquired. Contiguous slices (10-20) were consecutively collected across each narrowing (fig.1a/b). Imaging parameters were: FA=15^o; in-plane resolution=~0.55x0.55mm²; slice thickness=3.2mm; Venc=z(40-260cm/s) and x,y(40-80cm/s), depending on the degree of narrowing. Repeat scans were performed in 7/11 phantom models. Immediately following the PC-MRI scans, pressure was measured using an arterial catheter connected to a pressure transducer before and after the maximum narrowing.

Human studies: 11 healthy controls (47.3±14.6 years) and 9 patients (67.3±7.3 years, four with known invasive FFR) were studied. Patient inclusion criteria: known/suspected CAD, ≥1 coronary lesion (proximal stenosis ≥30%) detected by CTA and/or invasive coronary angiography (ICA). Coronary PC-MRI acquisitions were ECG-triggered (mid-diastole) and navigator-gated (end-expiration)^5-6. Fat-suppression pre-pulses were applied to avoid chemical shift effects and increase vessel contrast^7-8. Contiguous slices (4-10) were consecutively collected across the proximal coronary segment (healthy controls) or stenotic lesion (patients). Imaging parameters were: Venc=35-65cm/s in all 3 directions, cardiac phase=2(~70ms/phase), in-plane resolution=0.5-0.6x0.5-0.6mm², slice thickness=3.2mm and TA=2-4min/slice.

Data Analysis: Eddy-current correction was done offline followed by NS calculations^5,6,9 to obtain ΔP_MR. Reproducibility of the velocity values from PC-MRI and the derived ΔP_MR of the phantom studies were assessed using intra-class correlation coefficient (ICC) and Bland-Altman plot. Correlation of ΔP_MR and ΔP_PT was assessed via linear regression analysis.

Phantom studies: Bland-Altman plots of peak velocities and ΔP_MR are shown in fig.2a. For velocity measurements, excellent correlation was seen in the through plane peak velocities (Vz, ICC=0.90) and lower in Vx (ICC=0.57) and Vy (ICC=0.58). For ΔP_MRs, overall ICC=0.87; When observed individually, higher correlation was seen at smaller stenosis degrees and weaker as stenosis increased (fig.1c). This could be due to the increased velocity in larger stenoses, causing minor turbulence distal of the narrowing, thus, inconsistent velocity and ΔPMR between the two scans. Furthermore, ΔP_MR and ΔP_PT were highly correlated (fig.2b). We also observed that as %area stenosis increased, ΔP_MR also increased (fig.2c).

Human studies: A significant (p<0.01) increase in ΔP_MR was seen in the patient group (5.26±3.99mmHg) vs. healthy controls (0.70±0.57mmHg) (fig.3a). CTA/ICA reports in all patients showed a range of stenoses from 30%-50% at the left main or proximal left anterior descending coronary artery (pLAD). ICA/FFR was performed in 4/9 patients where 3/4 had a functionally non-significant lesion (FFR=0.93±0.70), corroborating with the proposed method (ΔP_MR≈3.0±1.70mmHg, low pressure drop). In one of the four patients who underwent ICA/FFR, a diffused, 50% lumen narrowing at the pLAD was observed (fig3b-c) with FFR=0.56, suggesting a functionally significant lesion. The same patient showed a ΔP_MR of ~14mmHg, likewise suggesting a functionally significant lesion (relatively high pressure drop).

Preliminary results suggest that noninvasive quantification of ΔP_MR in coronary arteries is feasible. In phantom studies, excellent correlation was found between the derived ΔP_MR and measured ΔP_PT. In human studies, patients with 30-50% stenoses were found to have a higher ΔP_MR than healthy volunteers. In patients who underwent invasive FFR, high FFR (low pressure drop) and low FFR (high pressure drop) both corroborated the ΔP_MR results. More patient studies with invasive FFR comparison are underway to further investigate the sensitivity of the approach in differentiating between a functionally non-significant and significant lesion. In addition, further technical improvements in terms of spatial, temporal resolutions and reduction of noise are also being developed to further improve the accuracy of the ΔP_MR calculations. Our preliminary studies demonstrated the feasibility of using PC-MRI to measure pressure gradient across coronary lesions. This approach has the potential to serve as a gatekeeper for unnecessary invasive catheterization procedures in patients with coronary artery disease.