Ten year insights into atrial fibrillation from the SCOT-HEART trial

Background

The Scottish Computed Tomography of the HEART (SCOT-HEART) trial randomised patients presenting with stable chest pain to either computed tomography (CT) coronary angiogram (CCTA) or standard care.

Purpose

This post hoc analysis assessed the incidence and prevalence of atrial fibrillation (AF) at baseline and at 10 years in SCOT-HEART participants. We also investigated whether CT assessment of the left atrium and left atrial appendage can predict the presence or future development of AF.

Methods

Patients were recruited as part of the SCOT-HEART trial and clinical data was obtained from the SCOT-HEART study database. A machine learning model (based on neural network framework) was trained on 104 CCTA to individually segment both the left atrium and left atrial appendage (Dice coefficient >0.87 for all). Chamber volume, wall mass and mean attenuation were extracted (Figure 1). Wall mass was indexed to chamber size.

Results

Of the 4,146 participants, 85 patients (2%) had AF at baseline and an additional 147 patients (3.5%) developed AF at 10-years. The majority of patients with AF were male (69%) and over 60 years of age (71%). There was a significantly higher prevalence of hypertension (51% vs 33%), diabetes mellitus (20% vs 10%) and hyperlipidaemia (71% vs 57%) in patients who either had AF at baseline or developed AF at 10 years.

Left atrial CT metrics were available for 1732 patients and, of these, 101 either had AF at baseline (n=33) or at 10 years (n=71). Patients with AF had significantly higher left atrial and left atrial appendage volumes compared with patients without AF (110.1±36.9 vs 88.7±21.3 ml, p<0.001 and 5.5±2.4 vs 4.8±2.0 ml, p=0.001 respectively). Both parameters predicted AF independent of age and sex (Odds ratio [OR] 1.02, 95% confidence interval [CI] 1.02 to 1.03 and OR 1.11, 95% CI 1.02 to 1.22, respectively). Patients with AF had lower left atrial chamber attenuation, left atrial wall attenuation and left atrial mass compared to patients without AF (p<0.03 for all). Both mass and attenuation of the left atrial appendage were also lower in patients with AF compared to patients without AF (p<0.04 for both).

Conclusions

CT-derived parameters, including left atrial volume, mass and attenuation, may identify early atrial remodelling in AF and help recognise patients at risk of future AF. Further large-scale prospective studies are warranted to evaluate the clinical utility of these parameters.Figure 1