Pulmonary Hypertension: Accuracy of Detection with Left Ventricular Septal-to–Free Wall Curvature Ratio Measured at Cardiac MR

Abstract

To retrospectively evaluate the accuracy and reproducibility of the cardiac magnetic resonance (MR) imaging-derived left ventricular septal-to-free wall curvature ratio for prediction of the right ventricular systolic pressure (RVSP) in patients clinically known to have or suspected of having pulmonary hypertension (PH), with same-day right-side heart catheterization (RHC) as the reference standard. Institutional review board approval was received for this HIPAA-compliant study. Sixty-one patients clinically known or suspected of having PH underwent cardiac MR and RHC on the same day. Interventricular septal curvature (C(IVS)) and left ventricular free wall curvature (C(FW)) measured at end systole were used to derive the curvature ratio (C(IVS)/C(FW)). Effective distending transmural pressure (dP(FW)) and transseptal pressure gradient (dP(IVS)) were assumed to be equivalent, respectively, to the systolic blood pressure (SBP) and the difference between SBP and RVSP. Curvature ratio and SBP were used to noninvasively estimate RVSP. Linear regression analysis was performed to assess the difference between curvature ratio and rate of pressure rise (dP) ratio (dP(IVS)/dP(FW)). The accuracy of the dichotomized curvature ratio in PH detection was analyzed by using receiver operating characteristic (ROC) curves. PH, defined as RVSP higher than 40 mm Hg, was confirmed with RHC in 46 patients. A direct linear correlation between dP ratio and curvature ratio was observed (r = 0.85, P < .001). Bland-Altman analysis revealed moderate agreement between cardiac MR- and RHC-derived RVSPs (mean difference, -1.1 mm Hg +/- 15.9 [standard deviation]). ROC analysis of the accuracy of the curvature ratio for detection of increased RVSP revealed 87% sensitivity and 100% specificity (area under ROC curve, 0.95; P < .001). Intraobserver (r = 0.97) and interobserver (r = 0.95) curvature ratio measurements were closely correlated. In patients clinically known to have or suspected of having PH, cardiac MR-derived curvature ratio, as compared with RHC measurement, was an accurate and reproducible index for estimation of RVSP.