Reproducibility and reliability of echocardiography in a preclinical model of myocardial infarction in rodent

Abstract

Echocardiography is a useful non-invasive imaging modality to assess left ventricular remodeling and function post myocardial infarction (MI) in clinical and preclinical studies. Nevertheless, echocardiography is known to be highly observer-dependent and poorly reproducible. Coverage probability (CP) was recently defined as the best index to evaluate the echocardiography reproducibility in clinical settings [1] , [2] . We aimed to use the CP in a preclinical model of MI in rats (R) and mice (M) to assess the reproducibility of our echocardiographic measurements. MI was induced by coronary ligation in Wistar rats or C57/BL6 mice. Echocardiography was performed, with a Vevo2100 machine, 2 and 3 months post-MI in mice and rats respectively. The echocardiography exams were analyzed by two trained and independent observers with VevoLab® LV trace tool to calculate the observed absolute difference (OAD). This OAD was determined for a C P = 80% for the left ventricular diastolic volume (LVEDV; μl), left ventricular systolic volume (LVESV; μl) and ejection fraction (EF; %). Good reproducibility was confirmed if OAD < 50 μl for the volumes and OAD < 5% for the EF. Different calculated OAD were: – 29 μl in rat; – 12 μl in mice for LVEDV; – 20 μl in rat; – 10 μl in mice for LVESV; – 1% in rat; – 5% in mice for EF. Therefore, good reproducibility of the echocardiographic measurements were confirmed in rat and mice as OAD for the volumes were below 50 μl and below 5% for the ejection fraction. Coverage probability is a good tool to assess reproducibility of cardiac echocardiographic analysis in preclinical setting, at least in mice and rat. This parameter is also a good tool for training validation.