Precision MRI of Left Atrial Fibrosis for Patients with Atrial Fibrillation

Project Summary/Abstract: Catheter ablation targeting the pulmonary veins and other atrial sites has emerged as the best intervention for restoring and maintaining sinus rhythm; however, 1-year success rates are only 60- 70%. Because ablation does not benefit all atrial fibrillation (AF) patients, a personalized medicine approach is needed to avoid an unnecessary procedure (cost >$20,000, risk ~5%) for expected non-responders (30-40%). Potential predictors of AF recurrence derived from standard clinical and imaging metrics have proven to be of limited use. Left atrial (LA) fibrosis is more promising, because fibrosis plays a central role in the development of an arrhythmogenic substrate for AF and may be a marker for more extensive disease less amenable to standard pulmonary vein isolation. In fact, LA fibrosis assessed with 3D LA late gadolinium enhancement (LGE) cardiovascular magnetic resonance (CMR), pioneered by the Utah group, has shown promise for predicting AF recurrence post-ablation. However, the “Utah” classification of LA fibrosis has garnered skepticism because of a lack of independent verification and validation. This lack of reproducibility stems from two fundamental methodologic deficiencies: (a) inadequate spatial resolution (1.5 mm x 1.5 mm x 2.5 to 5 mm) and contrast-to- noise ratio (CNR) and lengthy scan time (~11 min) at 1.5 Tesla and (b) unreliable image analysis techniques for quantification of fibrosis in the thin (~2 mm) LA wall. These deficiencies preclude widespread adoption of LA fibrosis quantification in clinical practice. To push the field of forward through these obstacles, we propose to develop disruptive technologies for quantification of LA fibrosis by integrating the following advanced techniques: (1) free-breathing 3D LGE CMR balanced steady state free precession (b-SSFP) readout with stack-of-stars k-space sampling and compressed sensing (CS) or eXtra-Dimensional Golden-angle RAdial Sparse Parallel (XD-GRASP) reconstruction with self- gating respiratory motion for achieving unprecedented image quality (i.e. CNR) with high spatial resolution (1.3 mm x 1.3 mm x 1.5 mm) and acceptable scan time (6 min) at 1.5 Tesla and (2) novel signatures technique for precise quantification of LA fibrosis using stochastic analysis. Unique advantages of the proposed signatures technique over standard analysis techniques include: (2a) more precise threshold-free fibrosis definition, (2b) insensitivity to LA segmentation, (2c) self-correction for intensity inhomogeneity, (2d) standardization and patient-specific quantification, and (2e) full automation and fast (2 min) processing. The specific objectives of this multi-center study are to: a) develop and validate robust 3D LA LGE CMR acquisition and reconstruction methods for 1.5 Tesla, (b) develop and validate a novel LGE signatures technique for quantification of LA fibrosis, and c) evaluate the prediction accuracy and reproducibility of LA fibrosis signatures across two sites. This proposal has high potential impact because it addresses two fundamental methodologic deficiencies precluding widespread adoption of LA fibrosis quantification in clinical practice.