Variability of CMR Tissue Characterization to Identify Cardiotoxicity

Study Questions:

What is the effect of the temporal and observer variability of cardiac magnetic resonance (CMR)–measured native T1, T2, and extracellular volume fraction (ECV) and serum biomarkers for the detection of cancer-therapeutics-related cardiac dysfunction (CTRCD)?


The study cohort was comprised of 50 participants (age 48.9 ± 12.1 years) who underwent three CMR studies (1.5-T) and biomarker measurements (high-sensitivity troponin I [hs-TnI] and B-type natriuretic peptide [BNP]) at 3-month intervals: 20 with HER2-positive breast cancer (10 with and 10 without CTRCD), and 30 prospectively recruited healthy participants. CTRCD was primarily defined as a >10% reduction in left ventricular ejection fraction to <55% and subsequently also as a >15% relative reduction in global longitudinal strain (GLS). The study investigators obtained T1 and T2 maps at three left ventricular short-axis locations. They calculated temporal and observer variability as the coefficient of variation and as the standard error of the measurement (SEM) using repeated measures and two-way analysis of variance. Minimal detected difference was defined as 2 ± SEM.


Compared with the patients without CTRCD, those with CTRCD had larger temporal change in native T1 (27.2 ms [95% confidence interval {CI}, 20.8-39.3 ms] vs. 12.4 ms [95% CI, 9.5-17.9 ms]), T2 (2.0 ms [95% CI, 1.5-2.9 ms] vs. 1.0 ms [95% CI, 0.74-1.4 ms]), and ECV (2.1% [95% CI, 1.5%-3.1%] vs. 1.0% [95% CI, 0.8%-1.5%]). Moreover, the temporal changes in biomarkers overlapped. The hs-TnI values and BNP values were available for 90 and 89 CMR time points, respectively. The temporal variability measured, as coefficient of variation for hs-TnI and BNP, were 17.7% (95% CI, 14.7%-21.1%) and 42.5% (95% CI, 34.5%-52.0%), respectively. The minimal detected difference for T1 (29 ms), T2 (3.0 ms), and ECV (2.2%) in healthy participants approached the mean temporal changes in patients with CTRCD. For individual patients with CTRCD, there was overlap in the temporal changes of all three parameters, and the variability in healthy participants with the least overlap for native T1. The interobserver/intraobserver variabilities for the CMR parameters were low (coefficient of variation, 0.5%-4.3%). At the individual participant level, there was overlap between the temporal variability in healthy participants and the temporal changes in patients with both definitions of CTRCD. The least overlap was seen on native T1 measurements.


The authors concluded that temporal changes in both biomarkers and tissue characterization measures in individual patients overlap with the temporal variability in healthy participants and approach the minimal detectable temporal differences. Also, the temporal variability of these methods may pose challenges to routine clinical application in individual patients receiving cancer therapy.


The authors need to be congratulated for their diligence in conducting this study on the utility of CMR in breast cancer patients receiving therapy with anthracyclines and trastuzumab. The findings of this study suggest that CMR has limitations. It appears that it is prudent for the same CMR reader to ascertain the T1, T2, and ECV values at different time points whenever possible for serial CMR, particularly in patients suspected of chemotherapy-related cardiac dysfunction. More studies are required to determine the optimal approach to such patients (may be a combination of biomarkers and imaging techniques?).

Clinical Topics: Cardio-Oncology, Heart Failure and Cardiomyopathies, Noninvasive Imaging, Acute Heart Failure, Heart Failure and Cardiac Biomarkers, Computed Tomography, Nuclear Imaging

Keywords: Anthracyclines, Biomarkers, Pharmacological, Breast Neoplasms, Cardiotoxicity, Diagnostic Imaging, Heart Failure, Natriuretic Peptide, Brain, Neoplasms, Stroke Volume, Tomography, X-Ray Computed, Troponin I, Ventricular Function, Left

< Back to Listings