Author + information
- Published online February 6, 2017.
- Dana C. Peters, PhD∗ (, )
- James S. Duncan, PhD,
- Karl Grunseich, BA,
- Mark A. Marieb, MD,
- Daniel Cornfeld, MD,
- Albert J. Sinusas, MD and
- Sudhakar Chelikani, PhD
- ↵∗Department of Radiology and Biomedical Imaging, Yale School of Medicine, Magnetic Resonance Research Center, TAC N 117, P.O. Box 20843, New Haven, Connecticut 06519
The regional relationship between left atrial (LA) strain and fibrosis is not known. Late gadolinium enhancement (LGE) cardiac magnetic resonance (CMR) identifies regions of possible atrial fibrosis (1). LA strain measurement, extracted by feature tracking using echocardiography or cine CMR, is an emerging tool for assessment of atrial function, which correlates globally with atrial LGE (2–5). Here, we report data comparing regional strain with the regional presence of fibrosis by LGE enhancement.
Eighteen patients with atrial fibrillation (AF) and 12 healthier controls (51 ± 13 years of age; 30% female) were imaged using 3-dimensional LGE and 2-dimensional cine CMR with 2- and 4-chamber views on a Siemens 1.5-T scanner (Erlangen, Germany). Controls had no AF, no cardiomyopathies, and no more than mild cardiac dysfunction or chamber enlargement. The 3-dimensional LGE sequence (1) (0.7 × 0.7 × 2 mm3 after zero-filling) used 0.2 mmol/kg gadobutrol (Gadavist, Bayer Healthcare, Leverkusen, Germany). LGE burden was measured by segmenting LGE-enhanced pixels on the atrial wall, normalized by atrial myocardial volume. LGE-enhanced pixels were defined as atrial wall pixels with a contrast-to-noise ratio (vs. blood pool) >3.5. Noise was measured as the SD of signal within a blood pool region of interest (ROI). The atrial myocardial volume was estimated using the surface area of a scalene ellipsoid (measuring 3 radii) and a myocardial thickness of 2.1 mm. LA volumes and LA ejection fraction (EF) were measured using the biplane method.
Tangential strain, largely representative of long-axis longitudinal strain, was estimated using a customized point-matching method for the LA contours at phases of maximum and minimum area on the cine MRIs. The contours were discretized into points, with points matched between phases using an iterative process that included closest point matching followed by piece-wise affine transformation. Strain was calculated as the change in distance between neighboring points, normalized by the initial distance. Regional 2- and 4-chamber strain maps were generated, and “global” strain was calculated. Continuous variables were compared using Pearson correlation. Two-tailed Student t tests were used to compare between groups.
Using 4-chamber LGE views aligned visually with corresponding cines, ROIs were drawn on the cines indicating regions with and without LGE enhancement. These ROIs were automatically propagated to the strain maps, so that all 4-chamber strains were labeled as either positive or negative for LGE enhancement in a blinded fashion. Receiver-operating characteristic curve analysis identified the optimal strain threshold for predicting LGE enhancement.
AF patients (1 with nonparoxysmal AF) had lower global strain values versus controls (17 ± 11% vs. 27 ± 10%; p = 0.02), greater LA volumes (60 ± 18 ml/m2 vs. 35 ± 10 ml/m2; p < 0.001), lower LA EF (34 ± 18% vs. 52 ± 10%; p = 0.002), but similar LGE burden (6 ± 7% with AF vs. 3 ± 3%; p = 0.17). Between groups, left ventricular EF, prevalence of hypertension, and sex were similar, but age and body mass index were greater in AF patients (p = 0.04).
Global strain correlated well with decreasing maximum LA volume and increasing LA EF (both r = ±0.57; p ≤ 0.001), but modestly with LGE burden (r = −0.36; p = 0.049). By multivariate regression (LGE burden, AF, LA volume), LA volume had a marginal association with strain (p = 0.056). Importantly (Figure 1), in atrial regions of LGE enhancement, regional strain was lower compared with regions without LGE (14 ± 11% vs. 20 ± 12%; p = 0.007). By receiver-operating characteristic analysis, strain <16% best predicted LGE enhancement (area under the curve = 0.66; p = 0.11, ns).
This study confirms (3–5) strong correlations between global strain and LA volumes and EF, and a weak correlation with LGE burden in paroxysmal AF patients (2) that have less remodeling. Additionally, there is a regional relationship between locations of atrial LGE (potentially fibrosis) and reduced strain.
The cohort size was small, and the control group included subjects with minor cardiovascular findings. Our strain methodology requires further refinement and extension to 3 dimensions. Strain mapping might be an alternative approach for detection of atrial fibrosis, serving in risk stratification of AF therapies such as ablation, but this remains undemonstrated. In conclusion, we report the new finding that lower atrial strain collocates with regional LGE enhancement.
Please note: This work was supported by in part by grants from the National Institutes of Health (National Heart, Lung, and Blood Institute R21 HL098573, R21 HL103463, and 1R01HL122560). The authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- American College of Cardiology Foundation
- Kuppahally S.S.,
- Akoum N.,
- Burgon N.S.,
- et al.
- Habibi M.,
- Lima J.A.,
- Khurram I.M.,
- et al.