Using cardiac magnetic resonance imaging to evaluate cardiac function and predict outcomes in patients with valvular heart disease

• Published in: Innovations (Philadelphia, Pa.) Pa.), 2015-01, Vol.10 (1), p.63-67
• Main Authors: Emerson, Dominic A, Amdur, Richard L, Morrissette, Jason R, Mordini, Federico E, Nagy, Christian D, Greenberg, Michael D, Trachiotis, Gregory D
• Format: Article
• Language: English
• Subjects: Aged; Female; Heart Valve Diseases - diagnosis; Heart Valve Diseases - physiopathology; Humans; Magnetic Resonance Imaging - methods; Male; Middle Aged; Prognosis; Retrospective Studies; Ventricular Dysfunction, Right - diagnosis; Ventricular Dysfunction, Right - physiopathology
• ISSN: 1556-9845; 1559-0879
• DOI: 10.1097/imi.0000000000000119

In valvular heart disease, elevated left atrial and pulmonary pressures contribute to right ventricular strain and, ultimately, right ventricle failure. Elevated pulmonary artery (PAP) and left ventricular end diastolic pressures are used as markers of right ventricle dysfunction and correlate with poor outcomes. Using cardiac magnetic resonance imaging (CMR), it is possible to directly quantify both left and right ventricular ejection function (LVEF and RVEF), and here, we compare CMR with traditional markers as outcome predictors. A retrospective review of prospectively collected data was performed for patients from January 2004 to February 2008 at a single center (n = 103). Patients were divided into those receiving CMR (n = 56) and those receiving only catheterization (n = 47). Univariate and multivariate logistic regression models were applied to determine predictors of mortality. Finally, predictive models for mortality using PAP, mean PAP, and left ventricular end diastolic pressure were compared to models using LVEF and RVEF obtained from CMR. Preoperative average CMR LVEF and RVEF were 57% and 46%, respectively. Only age emerged as an isolated predictor of mortality (P = 0.01) within the univariate models. Stepwise regression models were created using the catheterization or CMR data. When compared, the CMR model has a slightly better R, c (prediction accuracy), and sensitivity/specificity (0.22 vs 0.28, 0.77 vs 0.82, and 0.63/0.62 vs 0.69/0.64, respectively). Within our population, LVEF and RVEF predict mortality as least as well as traditional catheterization values. Additionally, CMR may identify of elevated PAPs caused by right ventricle dysfunction and those due to other causes, allowing these other causes to be addressed preoperatively.