P165A novel cardiac magnetic resonance imaging model to predict level of mixed venous oxygen levels in pulmonary hypertension

Abstract

Background: Pulmonary hypertension (PH) is a debilitating disease characterised by a progressive increase in pulmonary arterial pressure (PAP) that leads to right ventricular (RV) failure and death. Mixed venous oxygen saturation (SVO2) represents the oxygen saturation of blood returning to the lungs before reaching the alveolo-capillary units. it is strongly associated with clinical outcomes in PH. The relationship of non-invasive cardiac magnetic resonance (CMR) metrics to this prognostically relevant parameter in patients with PH are unknown. Purpose: This study sought to develop an early understanding of which CMR volumetric and flow parameters are most associated with mixed venous saturation (SVO2). Methods: 18 patients were recruited at a tertiary PH unit. SVO2 was measured during right heart catheterisation. All patients had CMR on the 1.5 T GE scanner. CMR protocol included cines and through-plane pulmonary artery phase contrast acquisition. The velocity encoded images were analysed for stroke volume. The 4 chamber cine was used to measure end-diastolic right atrial area. RV volumes were analysed using standard methods. Results: The RV volumetric functional parameters demonstrated no association to SVO2 (P > 0.05). However, a negative association was shown between right atrial (RA) area and SVO2 (R= -0.57, p = 0.01). The only other parameter which correlated with SVO2 was MPA stroke volume (R = 0.5, p = 0.03). In stepwise multiple regression, both parameters demonstrated independent association to SVO2. The predictive values generated by a combined model demonstrated high correlation to measured SVO2 (R = 0.73, p < 0.001). The CMR model equation to predict SVO2 is as follows: predicted SVO2 = 68.2092 + -0.5173∗(RA area) + (0.1899)∗(MPA SV). Conclusion: RA area and MPA stroke volume are independently associated with SVO2. A novel CMR prediction model comprised of these two metrics, demonstrates high association to the measured SVO2 by invasive haemodynamic study. Figure 1. A: Scatter plot of measured and predicted SVO2. B: RA area just before the opening of the tricuspid valve.