Is regional atrial strain a useful surrogate of regional atrial fibrosis in atrial cardiomyopathy?

Abstract

<h4>Aims</h4>To determine whether atrial biomechanics measured using 3D regional strain, left atrial volume (LAV), and left atrial emptying fraction (LAEF) are associated with atrial fibrosis in patients with suspected atrial cardiomyopathy.<h4>Methods and results</h4>Cardiovascular magnetic resonance (CMR) was performed in atrial fibrillation (AF) patients (<i>n</i> = 47). Healthy volunteer (<i>n</i> = 41) and familial dilated cardiomyopathy (DCM) (<i>n</i> = 31) cohorts were acquired for normalization and validation, respectively. Fibrosis was quantified using late gadolinium enhancement (LGE)-CMR, and 3D regional strain was quantified using feature tracking. Machine learning classifiers were used to classify regional severe fibrosis (>30% LGE enhancement) using regional strain and global measures of atrial anatomy and function. 3D regional strain measures (peak reservoir strain or first/second strain principal component) alone were not associated with regional fibrosis (accuracies ≤ 56.0%) in the AF cohort. Severe fibrosis was found primarily in the lateral (85.1% of AF patients) and posterior (66.0%) regions. In AF patients, the classifier incorporating LAV, LAEF, and regional location was associated with severe regional fibrosis (AUC = 0.86 ± 0.06, accuracy = 79.4 ± 6.2%), while in familial DCM patients, the accuracy was lower (62.8%).<h4>Conclusion</h4>There is a distinctive pattern of fibrosis that develops with progression of atrial cardiomyopathy, irrespective of cause. Global measures reflecting overall atrial anatomy (LAV) and function (LAEF), rather than localized regional 3D strain, were associated with severe regional fibrosis. These data suggest that regional atrial 3D strain alone is not a reliable surrogate for severe regional fibrosis.