19 Cardiac magnetic resonance to identify raised left ventricular filling pressure

BackgroundNon-invasive imaging is routinely used to estimate left ventricular (LV) filling pressures (LVFP) in heart failure (HF), as an alternative to right heart catheterisation (RHC). Transthoracic echocardiography (TTE) estimates of LVFP are frequently deployed but produce largely dichotomised data limiting flexible clinical use and perform less well in patients with heart failure with preserved ejection fraction (HFpEF). Cardiovascular magnetic resonance (CMR) is emerging as an important imaging tool for sub-phenotyping HF. However, currently we cannot estimate LVFP from CMR. This study sought to investigate if CMR can estimate LVFP in patients with suspected HF, whether this has increased diagnostic power beyond TTE and if CMR modelled LVFP has prognostic power.MethodsSuspected HF patients underwent RHC, TTE and CMR within 24 hours of each other. RHC measured pulmonary capillary wedge pressure (PCWP) was used as a reference for LVFP. CMR included left/right heart volumetric assessment and left atrial area. Patients were split into derivation (85%) and validation (15%) cohorts (figure 1). In the derivation cohort, multivariate regression was used to determine predictors of LVFP. The CMR-derived model was then applied to the validation cohort and diagnostic accuracy was compared with TTE. Association of CMR modelled LVFP with mortality was determined using Kaplan-Meier (KM) survival analysis.ResultsWe enrolled 835 patients (mean age 66±13 years, 38% male). Two CMR metrics were incorporated in the final model; LV mass and left atrial area. When applied to the validation cohort, CMR modelled PCWP had good correlation with RHC PCWP (R=0.6). The diagnostic accuracy of CMR modelled PCWP to predict elevated filling pressures (RHC PCWP > 14 mmhg) was 73%. TTE was non-diagnostic in 75% of cases (incorrect classification or indeterminate result). Of these, 71% were reclassified to a correct diagnosis by CMR (figure 2). CMR modelled PCWP was identified as an independent predictor of death on KM analysis (HR 2.18 (95% CI 1.1 to 4.3), P=0.02) (figure 3).Abstract 19 Figure 1Study flow diagramAbstract 19 Figure 2Comparison of diagnostic accuracy of TTE and CMR derived estimates of LVFP. In the validation cohort, 75% patients had non diagnostic TTE (incorrect classification or indeterminate) compared to 27% with CMR (p