Prognostic Utility of Cardiac MRI Myocardial Strain Parameters in Patients With Ischemic and Nonischemic Dilated Cardiomyopathy: A Multicenter Study

Abstract

Background: Prior small single-center studies have yielded conflicting results regarding the prognostic significance of myocardial strain parameters derived from feature tracking (FT) on cardiac MRI in patients with dilated cardiomyopathy (DCM). Objective: To evaluate the prognostic utility of FT parameters on cardiac MRI in patients with ischemic and nonischemic DCM and to determine the optimal strain parameter for outcome prediction. Methods: This retrospective study included 471 patients (median age, 61 years; 365 men, 106 women) with ischemic (n=233) or nonischemic (n=238) DCM and left ventricular (LV) ejection fraction (EF) <50% who underwent cardiac MRI at any of four centers from January 2011 to December 2019. Cardiac MRI parameters were determined by manual contouring. In addition, software-based FT was used to calculate six myocardial strain parameters [LV and right ventricular (RV) global radial, global circumferential, and global longitudinal strain (GLS)]. Late gadolinium enhancement (LGE) was also evaluated. Patients were assessed for a composite outcome of all-cause mortality and/or heart-failure hospitalization. Cox regression models were performed to determine associations between strain parameters and the composite outcome. Results: Mean LV EF was 27.5%, and mean LV GLS was -6.9%. Median follow-up period was 1328 days. The composite outcome occurred in 220 patients (125 deaths, 95 heart-failure hospitalizations). All six myocardial strain parameters were significant independent predictors of the composite outcome [hazard ratio (HR)=0.92-1.16; all p<.05]. In multivariable models that included age, corrected LV and right ventricular (RV) end-diastolic volume, LV and RV EF, and presence of LGE, the only strain parameter that was a significant independent predictor of the composite outcome was LV GLS (HR=1.13, p=.006); LV EF and presence of LGE were not independent predictors of the composite outcome in the models (p>.05). A LV GLS threshold of -6.8% had sensitivity of 62.6% and specificity of 62.6% in predicting the composite outcome rate at 4.0 years. Conclusion: LV GLS, derived from FT on cardiac MRI, is a significant independent predictor of adverse outcomes in patients with DCM. Clinical Impact: This study strengthens the body of evidence supporting the clinical implementation of FT when performing cardiac MRI in patients with DCM.