Abstract 4145207: A Comparison Between Global Longitudinal Strain (GLS) Derived with CMR Feature-Tracking (CMR-FT) and 2D Speckle-Tracking Echocardiography (2D-STE) to Monitor Cancer Therapy-Related Cardiac Dysfunction (CTRCD)

Left ventricular global longitudinal strain (GLS) is a critical imaging biomarker for monitoring cancer therapy-related cardiac dysfunction (CTRCD) in patients treated with chemotherapy. This prospective study investigated the interchangeability between GLS monitored with cardiovascular magnetic resonance feature-tracking (CMR-FT) and 2D speckle-tracking echocardiography (2D-STE) in breast cancer patients who received antineoplastic chemotherapy. At baseline and 3- and 6-month follow-ups after chemotherapy initiation, LV images were acquired on 34 breast cancer patients with 2D echocardiography, Steady State Free Precession (SSFP) MRI, and Displacement Encoding with Stimulated Echoes (DENSE) MRI as reference (Fig. 1). For 3D strains, the DENSE images were analyzed with automated Artificial Intelligence pipelines, SSFP images with SegmentCMR, and the echocardiographic images with EchoPAC. The latter two are vendor tools. Tests of equivalence, with two-one-sided t-tests (TOST) for statistical inferences of a large effect, were conducted between pairs of the DENSE, 2D-STE, and CMR-FT data; the interaction between time and modalities was tested with 2-way repeated ANOVA (RANOVA). The results showed TOST effect sizes larger than the equivalence bounds at all time points between GLS analyzed with CMR-FT and 2D-STE (CMR-FT mean = -16.0±1.9 %, 2D-STE mean = -15.1 ± 3.3%, Equiv. bounds, [-1.6%, 1.6%], 95% CI, [0.3%, 1.8%], p = 0.05 at 3-months), and DENSE and 2D-STE (DENSE mean = -15.9 ± 1.7%, 2D-STE mean = -15.1 ± 3.3%, Equiv. bounds, [-1.5%, 1.5%], 95% CI, [0.0%, 1.8%], p = 0.1 at 3-months), but not between DENSE and CMR-FT (DENSE mean = -15.9 ± 1.7%, CMR-FT mean = -16.0 ± 1.9%, Equiv. bounds, [-1.6%, 1.6%], 95% CI, [-1.3%, 1.5%], p < 0.001 at 3-months). Pairwise comparison from the RANOVA methodology*time interaction showed differences between CMR-FT and 2D-STE-derived GLS (ΔGLS = -0.9%, 95% CI, [-1.7%, -0.1%,], p = 0.03), and DENSE and 2D-STE-derived GLS (ΔGLS = -0.8%, 95% CI, [-1.9%, -0.1%], p = 0.04) but not between DENSE and CMR-FT (ΔGLS = 0.1%, 95% CI, [-0.1%, 0.4%], p = 0.33) at 3-months and also baseline. Statistically, the significant differences between the CMR-FT and 2D-STE results show that the two methodologies should not be used interchangeably for GLS evaluation in CTCRD, and further large-scale studies are required to establish the most suitable methodology.