Changes over time in AI-based echocardiographic metrics before and after therapy in ATTR cardiomyopathy

Introduction Transthyretin amyloid cardiomyopathy (ATTR-CM) is a progressive and fatal disease. Although TTR-specific therapies reduce mortality, up to 30% of patients continue to progress, underscoring the need for reliable treatment monitoring. Echocardiography is widely available and cost-effective, yet real-world data on echocardiographic changes over time and their response to therapy are limited. Purpose To investigate whether AI-based echocardiograms can detect an attenuation of disease progression in response to disease-modifying therapy for ATTR-CM. Methods In this retrospective cohort study, patients with wild-type ATTR-CM from two multisite registries were included. Echocardiograms from multiple time points were fully quantified using the artificial intelligence (AI) software US2AI, and NT-proBNP values were extracted from clinical records. The time scale was aligned with initiation of disease-modifying therapy at time 0. Untreated patients were modeled as pre-treatment and referenced as their last follow-up occurring just before time 0. To avoid outlier effects, only data from 6 years before to 6 years after treatment initiation were included. Changes over time were analyzed using linear mixed-effects models with restricted cubic splines. Results Among 335 participants with ATTR-CM (93% male, median age 78 years), a median of 4 echocardiograms (interquartile range (IQR) 2 – 6) were available per participant and 244 (73%) received disease-modifying therapy. At diagnosis, median interventricular septum thickness (IVST) was 13.1 mm (IQR 11.6 – 14.9), LV global longitudinal strain (LV GLS) -14.5% (-17.0 – -11.9), E/e’ 14.4 (11.0 – 17.9), TAPSE 19.2 mm (15.8 – 23.3), left atrial (LA) volume 39 ml/m2 (30 – 47), LA reservoir strain 12% (8 – 19) and NT-proBNP 1767 ng/mL (951 – 3449). All metrics worsened before treatment (p < 0.05). After treatment initiation, we observed significant changes in trajectories with attenuation of changes: IVST (p for non-linearity = 0.001), LV GLS (p=0.008), TAPSE (p < 0.001), LA volume (p < 0.001), LA reservoir strain (p < 0.001) and NT-proBNP (p = 0.005), while E/e’ even improved (p < 0.001), as shown on the Figure. Conclusion In ATTR-CM, AI-based echocardiography captures early changes in the trajectory of multiple key metrics after initiation of disease-modifying therapy, demonstrating attenuation of disease progression. These findings support echocardiography as a sensitive and accessible tool to monitor disease progression and treatment response in patients with ATTR-CM.