Comparing High-Density Electroanatomic Mapping and MRI: A Step Forward in Myocardial Characterization in Nonischemic...

Risk stratification for sudden cardiac death (SCD) in nonischemic cardiomyopathy (NICM) relies on identification of myocardial scars, the substrate for ventricular arrhythmias. Scar assessment is typically performed with late gadolinium enhancement cardiac magnetic resonance (LGE-CMR). LGE-CMR images can be postprocessed by dedicated software to generate 3D scar maps that can be integrated with electroanatomic voltage mapping (EVM).

This study sought to assess the concordance and reliability between LGE-CMR-derived scar maps and EVM performed with a high-density catheter in myocardial tissue characterization.

In this multicenter, observational, cross-sectional study, we included 36 NICM patients who underwent left ventricular EVM for arrhythmic and SCD risk stratification. Each patient underwent LGE-CMR, processed with the use of dedicated software to delineate scar core and border zones. EVM depicted scar regions with the use of omnipolar technology (OT), bipolar (BIP), and unipolar (UNI) acquisition modes. Cohen's kappa coefficient and Bland-Altman methods were used to assess agreement and reliability, respectively.

The UNI configuration showed the best agreement with subepicardial LGE-CMR maps (κ: 0.93; positive predictive value [PPV]: 97.2%; negative predictive value [NPV]: 86.5%), and OT outperformed BIP for subendocardial layers (κ: 0.97 vs 0.74; PPV: 96% vs 76%; NPV: 100% vs 96%). In reliability analysis, OT showed the least systematic bias and the narrowest limits of agreement with LGE-CMR-derived scar maps, and was associated with higher point density maps. Whereas BIP mapping had the best agreement with the 10%-20% myocardial layer, OT best aligned with the 20%-40% myocardial layer.

EVM had high agreement with LGE-CMR images and accurately delineated endo-mid-myocardial substrates, supporting its potential role in substrate characterization for NICM patients unfit for LGE-CMR.