Robust Approach to Measure Time-Resolved Septal Curvature in Pediatric Pulmonary Hypertension Patients Using Standard Short-Axis Ventricular Volume Contours: A Retrospective, Cross-Sectional Study.

Abstract

Background: Right heart catheterization is the gold standard for diagnosis and monitoring of pediatric pulmonary hypertension (PH). Noninvasive cardiovascular magnetic resonance (CMR) and echocardiography can additionally provide volumetric and functional assessment of the right ventricle and possibly serve as a non-invasive method for surveillance of PH severity using ventricular septal curvature (SC), interventricular septal (IVS) angle, or eccentricity index (EI). However, these septal wall measurements require manual interaction and can be observer dependent. This study presents the results of a robust, contour-based SC computation and assesses interobserver variability and performance as a candidate marker of RV afterload as compared to manually measured SC, IVS angle and EI.

Methods: This cross-sectional, retrospective study enrolled PH patients with mixed etiology (idiopathic/heritable and congenital heart/connective tissue disease-associated PH) and similar-age controls who underwent clinical CMR and catheterization (catheterization only for patients). A standard short-axis CMR stack was processed on commercial image-processing software to semi-automatically extract the relevant contours for the endo- and epicardial ventricular walls. SC was computed based on the contours and the minimum value in the cardiac cycle was obtained (SC_contour). For comparison, SC (SC_manu), IVS angle and EI were manually obtained from the CMR images. The relationship between septal wall measurements and invasive pressure measurements by catheterization was evaluated via Spearman correlation. Association with adverse events (death, lung or heart and lung transplant, and Potts surgery) was evaluated via univariate Cox proportional hazard model. Interobserver agreement was analyzed using intraclass correlation coefficient (ICC).

Results: Analysis was performed on 24 patients and 14 controls (median age 16.4 [12.3 - 18.9] and 15.9 [14.7 - 17.4] years respectively). SC_contour demonstrated moderate-strong correlations with RVSP (ρ=-0.79), RVSP/SBP (ρ =-0.65), and mPAP (ρ=-0.80) and had the strongest correlation in idiopathic/heritable PH patients (ρ = -0.74 - -0.83). SC_contour demonstrated a higher ICC (0.96 (95% CI 0.93 - 0.98; p<0.001) versus SC_manu (0.89 (95% CI 0.80 - 0.94; p<0.001)). In the exploratory prognostic analysis, significant associations with adverse events were not observed for any septal wall measurements.

Conclusions: The contour-based SC approach demonstrated strong correlation with invasive pressure measurements, with less manual interaction and observer dependency than manual measurements.