Reproducibility of echocardiographic assessment of 2D-derived longitudinal strain parameters in a population-based study (the STANISLAS Cohort study)

Global peak systolic longitudinal strain (PLS) derived from speckle tracking echocardiography (STE) is a widely used left ventricular deformation parameter. Modern ultrasound systems with improved temporal resolution and new software now allow automated multilayer analysis; however, there is limited evidence regarding its reproducibility. We performed intra- and inter-observer analyses within a population-based cohort study using conventional quantitative strain analysis (GE Healthcare). Fifty patients (49 ± 14 years) were randomly selected among the fourth visit of the STANISLAS Cohort. Multilayer PLS (transmural, subendocardial, and subepicardial), and strain rate (peak systolic, early and late diastolic) were evaluated. Peak systolic shortening (PSS) and early positive systolic strain (EPS) were calculated, as well as post-systolic index (PSI) and pre-stretch index (PST), two additional strain-derived parameters. Intra-observer intraclass correlation coefficients (ICC) were >0.75 for all analyzed parameters. The mean relative intra-observer differences were <5% for all considered parameters, and their 1.96 SDs were <15% for multilayer PLS, strain rate and PSS, but not for EPS, PSI and PST. Inter-observer ICCs were >0.70 (the majority being >0.80). The mean relative inter-observer differences were <7.5% for all considered parameters, with 1.96 SDs of relative differences being <21% for multilayer PLS, strain rate and PSS, but not for EPS, PSI and PST. In this population-based study, in subjects without or with a limited number of cardiovascular risk factors and no previous cardiovascular events, deformation parameters were found to be highly reproducible, except for EPS, PSI and PST, which showed moderately higher variability. Quantitative strain analysis appears to be an effective clinical and research tool, providing insights regarding longitudinal deformation using a simple three-step post-processing procedure.