Advertisement

CA-529-03 NOVEL DETECTION OF ATRIAL TACHYCARDIA-LIKE ISLANDS DURING ATRIAL FIBRILLATION PREDICTS RESPONSE TO ABLATION

      Background

      A major limitation of identifying organized activity in atrial fibrillation (AF) has been the use of analytics that are difficult to confirm visually by experts. An intuitive metric of AF organization that does not require frequency, phase, or complex mathematical constructs may better guide therapy.

      Objective

      To test the hypothesis that AF exhibits spatial regions of repetitive 1:1 electrograms (EGMs), analogous to ‘islands of AT’ and, further, that these islands are more predictive of response to ablation than clinical features or traditional metrics of AF organization.

      Methods

      We recruited N=224 patients (64±10 Y, 29.5% women) with global 64 pole AF recordings (Abbott, IL), in whom ablation terminated AF in N=122 (“Term”) or did not (“Non-term”; demographics p=NS). In a development cohort (N=60), repetitive EGMs in 2x2 areas was quantified over 4 sec by correlation. Maps of repetitive EGM activity (REACT; Fig B) indicate high (red) or low (blue) repetition. We now applied REACT to the independent cohort (N=164), and compared its predictive value for Term to 48 clinical variables, dominant frequency width (DF), and SD of cycle length (CL).

      Results

      Repetitive islands (Fig B) were larger in Term than Non-term patients (68.3±25.4% vs 45.8±26.9%, p<0.001). Cluster analysis using 48 clinical variables alone failed to separate groups. However, adding REACT yielded 4 clusters (Fig. C) with progressively greater likelihood for termination (p<0.001). REACT provided AUC for termination which was higher than for DF or CL (p<0.001 ANOVA). Combining indices provided AUC 0.80 for AF termination.