BACKGROUND. Echocardiography is the primary noninvasive technique for left ventricular (LV) strain measurement. MRI has potential advantages, although reference ranges and thresholds to differentiate normal from abnormal left ventricular global longitudinal strain (LVGLS), left ventricular global circumferential strain (LVGCS), and left ventricular global radial strain (LVGRS) are not yet established. OBJECTIVE. The purpose of our study was to determine the mean and lower limit of normal (LLN) of MRI-derived LV strain measurements in healthy patients and explore factors potentially influencing these measurements. EVIDENCE ACQUISITION. PubMed, Embase, and Cochrane Library databases were searched for studies published through January 1, 2020, that reported MRI-derived LV strain measurements in at least 30 healthy individuals. Mean and LLN measurements of LV strain were pooled using random-effects models overall and for studies stratified by measurement method (feature tracking [FT] or tagging). Additional subgroup and meta-regression analyses were performed. EVIDENCE SYNTHESIS. Twenty-three studies with a total of 1782 healthy subjects were included. Pooled means and LLNs for all studies were -18.6% (95% CI, -19.5% to -17.6%) and -13.3% (-13.9% to 12.7%) for LVGLS, -21.0% (-22.4% to -19.6%) and -15.6% (-17.0% to -14.3%) for LVGCS, and 38.7% (30.5-46.9%) and 20.6% (15.1-26.1%) for LVGRS. Pooled means and LLNs for LVGLS by strain measurement method were -19.4% (95% CI, -20.6% to -18.1%) and -13.1% (-14.2% to -12.0%) for FT and -15.6% (-16.2% to -15.1%) and -13.1% (-14.1% to -12.2%) for tagging. A later year of study publication, increasing patient age, and increasing body mass index were associated with more negative mean LVGLS values. An increasing LV end-diastolic volume index was associated with less negative mean LVGLS values. No factor was associated with LLN of LVGLS. CONCLUSION. We determined the pooled means and LLNs, with associated 95% CIs, for LV strain by cardiac MRI to define thresholds for normal, abnormal, and borderline strain in healthy patients. The method of strain measurement by MRI affected the mean LVGLS. No factor affected the LLN of LVGLS. CLINICAL IMPACT. This meta-analysis lays a foundation for clinical adoption of MRI-derived LV strain measurements, with management implications in both healthy patients and patients with various disease states.
Keywords: MRI; feature tracking; heart; meta-analysis; strain; tagging.