Granger Causality–Based Analysis for Classification of Fibrillation Mechanisms and Localization of Rotational Drivers

Abstract

Background: The mechanisms sustaining myocardial fibrillation remain disputed, partly due to a lack of mapping tools that can accurately identify the mechanism with low spatial resolution clinical recordings. Granger causality (GC) analysis, an econometric tool for quantifying causal relationships between complex time-series, was developed as a novel fibrillation mapping tool and adapted to low spatial resolution sequentially acquired data. Methods: Ventricular fibrillation (VF) optical mapping was performed in Langendorff-perfused Sprague-Dawley rat hearts (n=18), where novel algorithms were developed using GC-based analysis to (1) quantify causal dependence of neighboring signals and plot GC vectors, (2) quantify global organization with the causality pairing index, a measure of neighboring causal signal pairs, and (3) localize rotational drivers (RDs) by quantifying the circular interdependence of neighboring signals with the circular interdependence value. GC-based mapping tools were optimized for low spatial resolution from downsampled optical mapping data, validated against high-resolution phase analysis and further tested in previous VF optical mapping recordings of coronary perfused donor heart left ventricular wedge preparations (n=12), and adapted for sequentially acquired intracardiac electrograms during human persistent atrial fibrillation mapping (n=16). Results: Global VF organization quantified by causality pairing index showed a negative correlation at progressively lower resolutions (50% resolution: P =0.006, R 2 =0.38, 12.5% resolution, P =0.004, R 2 =0.41) with a phase analysis derived measure of disorganization, locations occupied by phase singularities. In organized VF with high causality pairing index values, GC vector mapping characterized dominant propagating patterns and localized stable RDs, with the circular interdependence value showing a significant difference in driver versus nondriver regions (0.91±0.05 versus 0.35±0.06, P =0.0002). These findings were further confirmed in human VF. In persistent atrial fibrillation, a positive correlation was found between the causality pairing index and presence of stable RDs ( P =0.0005, R 2 =0.56). Fifty percent of patients had RDs, with a low incidence of 0.9±0.3 RDs per patient. Conclusions: GC-based fibrillation analysis can measure global fibrillation organization, characterize dominant propagating patterns, and map RDs using low spatial resolution sequentially acquired data.

Toward Mechanism-Directed Electrophenotype-Based Treatments for Atrial Fibrillation

“…As we have demonstrated for VF, it is possible that the degree of remodeling of the atrial substrate, including remodeling of ionic currents (Samie et al, 2001;Kneller et al, 2005;Muñoz et al, 2007), calcium handling (Samie et al, 2000), connexins, and fibrosis, determines the specific AF electrophenotype and the predominant AF mechanism. Our own analysis of human persistent AF would appear to support this hypothesis (Handa et al, 2020a).…”

“…There are currently several candidate approaches to identify the AF electrophenotype to guide tailored treatments. We recently adapted Granger causality analysis, originally an econometric tool for quantifying causal relationships between complex time series data (Nobel Prize Economics 2003), for use on AF data (Handa et al, 2020a). By testing for causal relationships of electrical signals between neighboring regions of the heart over time and quantifying the number of causal relationships above a set "causality threshold, " we can reliably measure global fibrillation organization and characterize the AF electrophenotype, even when applied to low spatial resolution clinical AF data (Figure 2).…”

“…In this example, CPI using downsampled data correlates negatively with the full-resolution l ps , which measures the number of short-lived phase singularities and is a measure of fibrillation disorganization. Figure adapted and reproduced from Handa et al (2020a) , in line with Creative Commons Attribution 4.0 (CC-BY 4.0).…”

Classification of Fibrillation Organisation Using Electrocardiograms to Guide Mechanism-Directed Treatments

“…We recently showed that there is a range of AF and VF mechanisms, with varying degrees of the global organisation, using ex vivo optical mapping of explanted perfused hearts and invasive intracardiac mapping in patients undergoing AF ablation (Handa et al, 2020 ). Only some forms of AF are globally organised and driven by stable RDs, and these would be potentially amenable to ablation targeting RDs, while other forms of AF are globally disorganised with no clear drivers and as such may respond to compartmentalisation of the atria.…”

A Multiscale in Silico Study to Characterize the Atrial Electrical Activity of Patients With Atrial Fibrillation. A Translational Study to Guide Ablation Therapy

“…During persistent AF, the electrical activity is chaotic, and the identification of the wavefront propagation is not trivial. Granger causality has been proposed as a measurement that globally characterizes the organization of the wavefront propagation and maps rotational drivers using low spatial resolution sequentially acquired data [21]. However, it requires sequentially recorded segments longer than 2.5 seconds, which are not commonly found in clinical electrophysiology studies.…”