Left atrial appendage strain predicts subclinical atrial fibrillation in embolic strokes of undetermined source

Abstract

Aims: Left atrial (LA) strain is promising in prediction of clinical atrial fibrillation (AF) in stroke patients. However, prediction of subclinical AF is critical in patients with embolic strokes of undetermined source (ESUS). The aim of this prospective study was to investigate novel LA and left atrial appendage (LAA) strain markers in prediction of subclinical AF in ESUS patients.

Methods and results: A total of 185 patients with ESUS, mean age 68 ± 13years, 33% female, without diagnosed AF, were included. LAA and LA function by conventional echocardiographic parameters and reservoir strain (Sr), conduit strain (Scd), contraction strain (Sct), and mechanical dispersion (MD) of Sr were assessed with transoesophageal and transthoracic echocardiography. Subclinical AF was detected by insertable cardiac monitors during follow-up. LAA strain was impaired in 60 (32%) patients with subclinical AF compared to those with sinus rhythm: LAA-Sr, 19.2 ± 4.5% vs. 25.6 ± 6.5% (P < 0.001); LAA-Scd, -11.0 ± 3.1% vs. -14.4 ± 4.5% (P < 0.001); and LAA-Sct, -7.9 ± 4.0% vs. -11.2 ± 4% (P < 0.001), respectively, while LAA-MD was increased, 34 ± 24 ms vs. 26 ± 20 ms (P = 0.02). However, there was no significant difference in phasic LA strain or LA-MD. By ROC analyses, LAA-Sr was highly significant in prediction of subclinical AF and showed the best AUC of 0.80 (95% CI 0.73-0.87) with a sensitivity of 80% and a specificity of 73% (P < 0.001). LAA-Sr and LAA-MD were both independent and incremental markers of subclinical AF in ESUS patients.

Conclusion: LAA function by strain and mechanical dispersion predicted subclinical AF in ESUS patients. These novel echocardiographic markers may improve risk stratification in ESUS patients.

Keywords: Atrial strain; Cardiac embolism; Embolic stroke of undetermined source; Subclinical atrial fibrillation.

Graphical abstract

Figure 1

Examples of LAA six-segment strain model and LAA multiplane and 3D imaging. LA, left atrium; LAA, left atrial appendage; LV, left ventricle; LUPV, left upper pulmonic vein.

Figure 2

(A–C) LAA strain imaging, different image examples from one cardiac cycle. (A) LAA triphasic strain curve, vertical white arrows indicate the amplitudes of LAA-Sr, LAA-Scd, and LAA-Sct. (B) LAA-positive triphasic strain curves, vertical white arrow indicates peak global LAA-Sr strain. (C) LAA mechanical dispersion, horizontal white arrows indicate time to peak LAA-Sr strain. The standard deviation of time to peak LAA-Sr ewas dfined as LAA mechanical dispersion, reflecting contraction inhomogeneity. LAA, left atrial appendage; LAA-Sr, left atrial appendage reservoir strain; LAA-Scd, left atrial appendage conduit strain; LAA-Sct, left atrial appendage contraction strain. (D) LAA strain imaging of three different, independent cardiac cycles recorded in the same patient but briefly after each other.

Figure 3

By ROC analyses, LAA-Sr showed the best ability to predict subclinical AF. LAA-Sr, left atrial appendage reservoir strain; LAA-Scd, left atrial appendage conduit strain; LAA-Sct, left atrial appendage contraction strain; LAA-MD, left atrial appendage mechanical dispersion; LAVI, left atrial volume index; AF, atrial fibrillation.

Figure 4

Independent and incremental predictive value of left atrial appendage strain and mechanical dispersion. The initial model with CHA2DS2-VASc quartiles was significantly improved by the addition of LAVI and further improved by adding LAA strain and LAA mechanical dispersion to predict subclinical AF in ESUS patients. AF, atrial fibrillation; ESUS, embolic strokes of undetermined source; LAA, left atrial appendage; LAA-Sr, left atrial appendage reservoir strain; LAA-MD, left atrial appendage mechanical dispersion; LAVI, left atrial volume index.