The use of cardiac magnetic resonance imaging for the evaluation of pulmonary hypertension

Purpose
Evaluate the utility of cardiac magnetic resonance imaging to estimate the principle hemodynamic parameters that are measured by right heart catheterization in a noninvasive manner i.e. mean pulmonary artery pressure, pulmonary vascular resistance and pulmonary artery wedge pressure through cardiac magnetic resonance based numerical models.


Materials and methods
29 pulmonary hypertension patients, fitting the inclusion criteria were randomly selected and included in the study. CMR Imaging and right side heart catheter (RHC) were performed within one month. 3 Cardiac MRI based models in literature that showed high accuracy were tested. Two equations for mPAP calculation; mPAP=-231.423 + 53.8(loge inter-ventricular septal angle)+log10(right ventricular mass divided by left ventricular mass) i.e ventricular mass index X 8.708+area of pulmonary artery in diastole X 0.009 and mPAP = –4.6+(0.32*septal angle)+(ventricular mass index × 16.3). One equation for PAWP; PAWP = left atrial volume index +6.43 × 0.22.


Results
The Altman and Bland correlation between mPAP invasively measured and CMR-estimated mPAP had good correlation with r= 0.594 and r=0.599 (P<0.001) for CMR based mPAP model 1 and 2, respectively. The calculated mean bias between the RHC-derived and CMR-estimated mPAP was 7.9 (agreement interval -24.8 to 40.6 mm Hg) and mean bias -3 (agreement interval -34.8 to 28.2 mm Hg) for CMR based mPAP model 1 and 2, respectively. There was no correlation between invasively measured and CMR-estimated PAWP with (P =0.092) for CMR based PAWP model. The mean bias between the RHC-derived and CMR-estimated PAWP was 2.4 (agreement interval –13.5 to 18.2 mm Hg). The correlation between invasively calculated and CMR-estimated PVR had good correlation with r=0.703 and r=0.704 (P<0.001) for CMR based PVR model 1 and 2, respectively. The mean bias between the RHC-measured and CMR-estimated mPAP was 0.6 (agreement interval -11.6 to 12.8 mm Hg) and mean bias -1.3 (agreement interval -12.1 to 9.5 mm Hg) for CMR based mPAP model 1 and 2, respectively.


Conclusion
Our results showed good correlations between CMR findings and RHC as regard mPAP and PVR. Thus, estimation of mPAP, PAWP and PVR non-invasively using CMR is feasible but needs further studies to improve accuracy.