Reproducibility of late gadolinium enhancement quantification techniques in ischemic and nonischemic heart diseases (ReLate study)

Jonathan Gavrysh^1,2,3Philine Reisdorf^1,2,3Thomas Hadler^1,2,3Thomas Mayr¹Clemens Ammann^1,2,3,4Jan Gröschel^1,2,3Johanna Kuhnt^1,2,3Florian von Knobelsdorff-Brenkenhoff⁵Fabian Muehlberg^1,2,4,6Carsten Schwenke⁷Maximilian Fenski^1,2,3Jeanette Schulz-Menger^1,2,3,4*

• ¹Working Group on CMR, Experimental and Clinical Research Center, a cooperation between Max Delbrück Center for Molecular Medicine in the Helmholtz Association and Charité – Universitätsmedizin Berlin, Berlin, Germany
• ²Charité – Universitätsmedizin Berlin, corporate member of Freie Universität Berlin and Humboldt-Universität zu Berlin, Berlin, Germany
• ³DZHK (German Centre for Cardiovascular Research), partner site Berlin, Berlin, Germany
• ⁴Department of Cardiology and Nephrology, HELIOS Hospital Berlin-Buch, Berlin, Germany
• ⁵KIZ – Kardiologie im Zentrum and Ludwig-Maximilians-University Munich, Munich, Germany
• ⁶MSB Medical School Berlin, Berlin, Germany
• ⁷SCO:SSiS Statistical Consulting, Bleekstr. 5, 32427 Minden, Minden, Germany

Background: Late gadolinium enhancement (LGE) assessed by cardiovascular magnetic resonance (CMR) is an established metric for risk stratification and therapeutic guidance. However, consensus on the optimal technique for quantifying left ventricular (LV) LGE extent remains lacking. This study aimed to identify the most reliable method for quantifying LGE in chronic myocardial infarction (CMI), hypertrophic cardiomyopathy (HCM), and inflammatory heart disease (IHD). Methods: A retrospective analysis was conducted on 285 prospectively enrolled patients (CMI: n = 98; HCM: n = 92; IHD: n = 96). LV LGE images in short-axis orientation were analyzed twice by the same reader. The most reliable LGE quantification technique was defined as the one achieving the highest intra-observer reproducibility. A two-step study design was implemented: in the pilot phase (n = 90), three quantification methods were compared: full width at half maximum (FWHM), signal threshold versus reference mean using 2 to 6 standard deviations (n-SD), and manual thresholding. Techniques demonstrating the lowest variability were then applied in a validation cohort (n = 195). A mixed model for repeated measures was used to estimate mean differences. Equivalence was confirmed if the 95% confidence interval (CI) for the mean difference remained within predefined margins. Results: In CMI, FWHM demonstrated the highest reproducibility, with a mean difference of 0.47% (95% CI: -0.40 to 1.35). In HCM, both the 5-SD and 6-SD techniques showed the highest reproducibility, with mean differences of 0.06% (95% CI: -1.28 to 1.39) and -0.16% (95% CI: -1.50 to 1.17), respectively. In IHD, the 5-SD and 6-SD techniques achieved the highest reproducibility, with mean differences of -0.72% (95% CI: -1.54 to 0.11) and -0.71% (95% CI: -1.54 to 0.11). Conclusion: The distribution and pattern of LGE influence the reproducibility of its quantification. FWHM provided the highest intra-observer reproducibility for sharply demarcated scars, as seen in CMI. For more diffuse fibrosis patterns, such as in HCM and IHD, both the 5-SD and 6-SD techniques offered similarly reproducible performance.