Toward discerning the mechanisms of atrial fibrillation from surface electrocardiogram and spectral analysis

Abstract

Atrial fibrillation (AF) is the main cause of stroke and the most common sustained arrhythmia, afflicting about 2.3 million Americans. Clinical treatment and management of AF would benefit from a noninvasive and global assessment of the arrhythmia; however, that avenue seems currently limited in part by our poor understanding of arrhythmia itself. Experimental studies of AF in the isolated sheep heart demonstrated that high-frequency sources in the posterior wall of the left atrium drive the fibrillatory activity throughout both atria. Motivated by those results and by a growing body of work investigating how measurements of the cycle length of activity in patients during AF can contribute to its treatment, we focused our analysis on the dispersion of dominant frequency (DF) of the activity during AF in humans. Using electroanatomic mapping and Fourier methods, we generated 3-dimensional intracardiac DF maps of the atria in patients before AF ablation procedures and identified relatively small high-DF (HDF) sites. In patients with paroxysmal AF, the HDF sites are often localized to the posterior left atrium near the ostia of the pulmonary veins. In contrast, patients with permanent AF demonstrate HDF sites that are more often localized to the atria than the posterior left atrium-pulmonary vein junction. In our study, ablation at HDF sites resulted in significant slowing of the arrhythmia and termination of sustained AF in 87% of patients with paroxysmal AF. Furthermore, we found that abolishing, by ablation, preexisting left atrium to right atrium DF gradients predicted long-term freedom of AF in both paroxysmal and persistent AF patients. Overall, the analysis of intracardiac electrical recordings in the frequency domain has greatly enhanced our understanding of its underlying mechanisms and may contribute to monitoring drug effects and guide ablation procedures aiming at its termination. On the other hand, current body surface mapping methods have also suggested better correlations between surface AF frequency and intracardiac local DFs as compared with spatiotemporal activation patterns. Therefore, further study of the correlation of spectral observables obtained from the atria and from the surface electrocardiogram during AF seems to have the potential to advance our ability to diagnose and discern mechanisms of AF noninvasively.

Figure 1

Bipolar electrograms and corresponding power spectra obtained from 4 points in a patient with spontaneous paroxysmal AF. Each site shows distinct DF. Panel A clearly demonstrates the utility of spectral analysis. The bipolar recording shows low amplitude complex signals that make accurate determination of frequency of activation difficult. The power spectrum clearly demonstrates the DF at 8.1 Hz. The DF map of this patient is shown in Figure 2A. Panel A is from the site of maximal DF at the RIPV; Panels B and C are from other HDF sites in the right superior and left superior PVs; and Panel D is from the posterior RA. Reproduced from Sanders et al, Circulation. 2005;112:789–797.

Figure 2

Dominant frequency maps of human AF. A, DF map in a patient with paroxysmal AF (6 hours). Note HDF sites in each of the PVs. Ablation sequence in this patient was LSPV, LIPV RSPV and RIPV (site of AF termination); AF CL increased by 10 ms, 25 ms, 9 ms and 75 ms, respectively, before termination. B, DF map in a patient with permanent AF (24 months). The maximal DF and atrial frequency are higher than the patient in panel A. In addition, many of the HDF sites are located outside the PVs. Ablation sequence in this patient was RIPV, RSPV, LSPV and LIPV; AFCL increased by 5 ms, 2 ms, 0 ms and 5 ms respectively. The electrogram of the site of maximal DF in the LA and RA is presented showing significant fractionation. LSPV, left superior PV; LIPV, left inferior PV; RSPV, right superior PV; RIPV, right inferior PV; SVC, superior vena cava; MA, mitral annulus; TA, tricuspid annulus. Color-bar, scale in Hz. Reproduced from Sanders et al, Circulation. 2005;112:789–797.