Comparison of Strain using Feature Tracking CMR and Speckle Tracking Echocardiography Against a Reference Standard of cine DENSE in Breast Cancer Patients Exposed to Cardiotoxic Therapy.

Presenting Author(s)

• OO

  Olivia Ondigi, MD

  Fellow of Cardiac Imaging
  University of Virginia School of Medicine
  Charlottesville, Virginia, United States

Primary Author(s)

• 

  Cristiane C. Singulane, MD, MSc

  Research Associate
  University of Virginia Health System
  Charlottesville, Virginia, United States

Co-Author(s)

• 

  Sona Ghadimi, PhD

  research scientist
  University of Virginia
  Charlottesville, Virginia, United States
• NP

  Nidhi Purohit, BA

  Medical Student
  University of Virginia School of Medicine, United States
• MM

  Maxwell Malter, BSc

  Medical Student
  University of Virginia School of Medicine, United States
• ML

  Max A. Luna, MD

  Associate Professor of Medicine
  University of Virginia School of Medicine, United States
• PD

  Patrick M. Dillon, MD

  Associate Professor of Medicine
  University of Virginia School of Medicine, United States
• TM

  Trish A. Millard, MD

  Associate Professor of Medicine
  University of Virginia School of Medicine, United States
• CB

  Christiana M. Brenin, MD

  Associate Professor of Medicine
  University of Virginia School of Medicine, United States
• 

  Frederick H. Epstein, PhD

  Professor
  University of Virginia
  Charlottesville, Virginia, United States
• 

  Amit R. Patel, MD, FSCMR

  Professor of Medicine
  Division of Cardiology, University of Virginia Health System, Charlottesville, Virginia, USA.
  Charlottesville, Virginia, United States

Background: Speckle tracking echocardiography (STE) is widely used but suffers from reduced reproducibility, particularly in patients with suboptimal acoustic windows potentially leading to higher variability. Cardiac magnetic resonance (CMR) provides accurate quantification of LVEF and strain. Feature tracking (FT) is a readily available CMR-based technique for strain analysis, though its agreement with the CMR reference standard, displacement encoding with stimulated echoes (DENSE) and STE, remains unclear. We hypothesize that FT and DENSE-derived strain demonstrates stronger correlation and lower variability than STE.

Methods: Sixteen women diagnosed with breast cancer (stages I–IV), with exposure to cardiotoxic drugs, were prospectively enrolled. They underwent CMR imaging at 3.0T, which included long- and short-axis cine balanced steady-state free precession (bSSFP) imaging, and DENSE imaging. Within 24 hours of the CMR, patients underwent echo, with acquisition of comparable apical views (4-, 2-, and 3-chamber) and short-axis views (basal, mid, and apical levels). Repeated acquisitions were performed for all three modalities.

Global circumferential strain (GCS) STE was analyzed using Cardiac Performance Analysis software, and global longitudinal strain (GLS) was measured using the AutoStrain application (TOMTEC). FT analysis was conducted using SuitHeart software (NeoSoft), while DENSE analysis was performed using the open-source DENSE analysis code (Figure 1). Agreement between DENSE and strain values obtained from FT and STE was evaluated using Bland–Altman and Pearson correlation coefficients. Reproducibility between repeated measurements was quantified using intraclass correlation coefficients (ICC).

Results: The median age was 55 years (41–65), and median CMR-derived ejection fraction was 57% (55–60). STE could not be performed in three patients due to image quality. A strong correlation was observed between GCS (ρ = 0.86, p < 0.001) and GLS (ρ = 0.90, p < 0.001) measured by DENSE and FT, whereas GCS (ρ = 0.36, p = 0.22) and GLS (ρ = 0.21, p = 0.53) derived from STE showed a weak and non-significant correlation with DENSE. Bland–Altman analysis revealed a minimal GCS and GLS bias between DENSE and FT. In contrast, DENSE vs. STE showed a larger positive bias with wider limits of agreement on GCS and GLS indicating greater variability and systematic overestimation by STE (Figure 2). Reproducibility of strain measures was higher for FT than STE. The ICC for FT-derived GCS was 0.87 (95% CI: 0.63–0.95, p < 0.001), and for GLS 0.73 (95% CI: 0.39–0.90, p < 0.001), reflecting good reliability. By comparison, STE showed moderate reproducibility, with ICCs of 0.69 (95% CI: 0.26–0.90, p = 0.003) for GLS and 0.71 (95% CI: 0.09–0.94, p = 0.015) for GCS.

Conclusion: FT showed stronger correlation, smaller bias, and better reproducibility than STE when compared with DENSE. FT may offer a more reliable approach for clinical decision-making in patients requiring cardiac strain evaluation.

Figure 1. Myocardial strain measurement across imaging modalities: Displacement Encoding with Stimulated Echoes (DENSE) (A) and Feature Tracking (FT) (B) from cardiac magnetic resonance (CMR), and Speckle Tracking (STE) from echocardiography (C). Global longitudinal strain values from FT and STE were compared with DENSE strain values of the left ventricular mid-segments, as previously validated.
Figure 2. Comparison of Myocardial Strain Measurements by Feature Tracking (FT) and Speckle Tracking Echocardiography (STE) Against Displacement Encoding with Stimulated Echoes (DENSE). Top row: Scatter plots comparing global circumferential strain (GCS) (left panel) and global longitudinal strain (GLS) (right panel) values obtained by FT (A) and STE (B) versus the DENSE reference standard. Bottom row: Bland–Altman plots assessing agreement between DENSE and FT (C) and DENSE and STE (D) for GCS and GLS.