Overview

Cardiac amyloidosis is a progressive and potentially fatal disease requiring accurate risk stratification to guide treatment decisions and patient management. Echocardiography is the first-line imaging modality, and AI-based tools are increasingly applied for automated analysis. However, the prognostic value of AI-derived echocardiographic measurements had remained unclear, particularly whether they add value beyond established biomarker-based staging systems.

 

Study Design

This retrospective cohort study included 435 patients from two registries with transthyretin amyloid cardiomyopathy (ATTR-CM), light-chain amyloid cardiomyopathy (AL-CM), and rare forms of cardiac amyloidosis. Baseline echocardiograms were fully and automatically quantified using Us2.ai, generating LVEF, global longitudinal strain, and TAPSE measurements. Prognostic performance was assessed using Kaplan-Meier analysis, Cox regression, and ROC curves for one-year event prediction. An echo staging system was developed by combining LV-GLS and TAPSE, stratifying patients into low (both normal), intermediate (one abnormal), and high risk (both abnormal) groups.

 

Key Results

  • 435 patients included (86% male, median age 76 years); 347 ATTR-CM, 64 AL-CM, 24 rare forms
  • During median follow-up of 2.4 years, 188 patients experienced the composite endpoint of death or heart failure hospitalisation
  • GLS, TAPSE, and LVEF were all significant univariate predictors of outcome (all p<0.001)
  • Echo staging stratified risk with hazard ratios of 3.10 and 6.73 for intermediate and high vs. low risk groups (p<0.001)
  • Echo staging provided incremental value over NAC and MAYO staging (chi-squared 91 vs. 65, p<0.001)
  • One-year event prediction did not differ between AI and human measurements for TAPSE, GLS, or LVEF (all p>0.21), confirming prognostic equivalence

 

Why This Matters

Accurate risk stratification in cardiac amyloidosis is critical for timing therapy initiation and identifying patients who may benefit from more intensive monitoring or earlier intervention. This study demonstrates that Us2.ai-derived measurements are not only reproducible and efficient, but prognostically equivalent to expert manual measurements and provide incremental value over existing biomarker-based staging when combined into an echo staging system. This supports their use in routine clinical practice for risk stratification without requiring specialist echocardiographic expertise.

 

Conclusion

AI-derived echocardiographic measurements provide robust prognostic stratification in cardiac amyloidosis, with performance comparable to manual measurements and incremental value over established biomarker staging systems, supporting their clinical utility for risk stratification in this population.

 

 

Walser, A., Flammer, A. J., Hundertmark, M., Shiri, I., Schwotzer, R., Ruschitzka, F., Tanner, F. C., Gräni, C., & Benz, D. C. (2026, August 28-31). Prognostic value of artificial intelligence-derived echocardiographic measurements in ATTR cardiomyopathy [ePoster presentation]. ESC Congress 2026, Munich, Germany.