Non-Invasive Estimation of Pulmonary Vascular Resistance With Cardiac Magnetic Resonance

Study Questions:

What is the feasibility of creating a model based on cardiac magnetic resonance (CMR) for noninvasive estimation of absolute pulmonary vascular resistance (PVR) values?


The study comprised 100 consecutive patients with known or suspected pulmonary hypertension (PH; 53 ± 16 years, 73% women) who underwent same-day right heart catheterization (RHC) and CMR. Increased PVR was defined from RHC as >3 WU (n = 66, 66%). From CMR cine and phase-contrast images, right ventricular (RV) volumes and ejection fraction (RVEF), pulmonary artery (PA) flow velocities and areas, and cardiac output were quantified. The best statistical model to estimate PVR was obtained from a derivation cohort (n = 80) based on physiological plausibility and statistical criteria. Validity of the model was assessed in the remaining 20 patients (validation cohort). Diagnostic accuracy of the model for PVR estimation in the derivation cohort was assessed using Spearman correlation, Bland–Altman analysis, and receiver operator characteristics (ROC) curve.


The CMR-derived model was: estimated PVR (in WU) = 19.38 − [4.62 × Ln PA average velocity (in cm/s)] − [0.08 × RVEF (in %)]. In the validation cohort, the correlation between invasively quantified and CMR-estimated PVR was 0.84 (p < 0.001). The mean bias between the RHC-derived and CMR-estimated PVR was −0.54 (agreement interval −6.02 to 4.94 WU). The CMR model correctly classified 18 (90%) patients as having normal or increased PVR (area under the ROC curve, 0.97; 95% confidence interval, 0.89-1.00).


The authors concluded that noninvasive estimation of PVR using CMR is feasible.


The study suggests that noninvasive estimation of PVR using CMR is feasible and the model allowed the detection of increased PVR with high accuracy when compared with RHC. This method may have clinical applications such as being an initial step in the diagnosis of PH, and potentially evaluating serial changes related to disease course or therapeutic interventions, but needs further validation. It should be noted that PVR was underestimated in patients with severely increased resistances, and the model would probably not be able to detect changes within ranges of very severe PVR elevation.

Clinical Topics: Heart Failure and Cardiomyopathies

Keywords: Cardiac Catheterization, Hydrogen-Ion Concentration, Cardiac Output, Vascular Resistance, Magnetic Resonance Spectroscopy

< Back to Listings