Making left atrial 4D flow imaging ready for clinical use: a novel shortened sequence for reproducible quantification of LA flow parameters

Background Left atrial (LA) blood flow can be quantified using four-dimensional (4D) flow contrast-free cardiac magnetic resonance imaging (CMR).1 Disturbed LA flow parameters are associated with embolic brain infarcts, suggesting their potential utility as a novel biomarker for cardioembolic stroke risk prediction.2 However, existing LA 4D flow research sequences are long and largely confined to 3T systems, limiting their application to clinical practice. Purpose We developed a novel shortened LA 4D flow sequence (henceforth ‘short 4D flow’) and assessed whether this: i. provides reproducible quantification of LA flow at different field strengths; and ii. detects perturbations of LA flow associated with atrial fibrillation (AF). Methods 15 healthy volunteers underwent same-day CMR on a 3T and 1.5T system (Figure 1A). At 3T, subjects underwent: i. short 4D flow, ii. removal from the scanner with repeat isocentre positioning, iii. short 4D flow, and iv. the research long LA 4D flow sequence used in previous studies. Volunteers were then scanned at 1.5T with short 4D flow with repositioning. Blinded CMR analysis was performed as previously described with measurement of LA vorticity (describing the rotational component of LA blood flow) and peak velocity.1 We compared sequence acquisition times using one-way ANOVA. We assessed, using Bland-Altman plots: i. scan-rescan reproducibility of short 4D flow; ii. comparison of short 4D flow between 1.5T and 3T; iii. comparison between the research long sequence and short 4D flow. 17 patients with advanced atrial myopathy (i.e. history of AF and heart failure with preserved ejection fraction) underwent 3T CMR with short 4D flow (Figure 1A). Patients were stratified by rhythm status at the time of CMR (in sinus rhythm: AF-SR; in AF: AF-AF). LA vorticity and peak velocity were measured and compared with healthy volunteers using one-way ANOVA. Results Baseline characteristics are shown in Table 1. Short 4D flow at both 1.5T and 3T was substantially shorter than the original sequence (mean 1.5T 178s, 3T 180s, vs original 526s, p<0.0001; Figure 1B). Short 4D flow showed good scan-rescan reproducibility at 1.5T and 3T (Figure 1C-F) and was consistent across field strengths. When compared with the original sequence, short 4D flow displayed varying negative bias across the range of LA vorticity, without significant bias in measurement of LA peak velocity (Figure 1G-J). However, short 4D flow remained sensitive to differences in LA flow parameters between healthy controls and patients with advanced atrial myopathy (Figure 1K, L). Conclusion We have developed a novel LA 4D flow CMR sequence that i. provides reproducible quantification of LA flow parameters in 2-3 minutes; and ii. distinguishes LA flow perturbations associated with disease. This may allow application of LA 4D flow imaging to prospective large-scale studies investigating stroke risk prediction and evaluating novel stroke prevention strategies.