Grid annotation vs ablation tags in predicting lesion area coverage for pulmonary vein isolation using a variable-loop circular catheter

Key Findings.

• ▪Grid-based annotation (GRID) showed stronger agreement with post-pulmonary vein isolation (PVI) low-voltage areas than conventional ablation tags for predicting lesion coverage after VARIPULSE variable-loop circular catheter (VLCC)-guided PVI.
• ▪GRID more accurately predicted first-pass isolation and adequate antralization compared with conventional ablation tags.
• ▪GRID demonstrated higher interobserver reproducibility than ablation tags, supporting a more consistent assessment of lesion distribution during VLCC-guided pulsed field ablation.

Catheter ablation is the most effective rhythm control strategy for atrial fibrillation (AF),^1,2 with pulmonary vein isolation (PVI) as the cornerstone of first-time procedures.¹ The VARIPULSE™ (Biosense Webster, Irvine, CA) variable-loop circular catheter (VLCC) enables electroanatomical mapping (EAM) and pulsed field energy delivery for AF ablation.² As the first single-shot PFA catheter fully integrated within an EAM system, the VLCC enables direct projection of the ablation footprint onto the EAM, allowing real-time visualization and quantification of lesion coverage.³ Lesion coverage can be displayed as grid-based annotation (GRID) or conventional ablation tags (ABLt). However, the relative accuracy of these annotation strategies in predicting lesion coverage has not been systematically evaluated. Therefore, we performed a prospective comparison of GRID and ABLt visualization during VLCC-guided PVI to determine which method more accurately predicts lesion coverage assessed by post-ablation EAM. We prospectively enrolled consecutive patients undergoing first-time PVI-only procedures using a VLCC between April 2024 and December 2025 at our 2 European centers, performed by 6 expert operators. The study complied with relevant ethical guidelines. Institutional review board approval and written informed consent were obtained. All procedures were performed under general anesthesia with uninterrupted oral anticoagulation, ultrasound-guided femoral venous access, and transesophageal echocardiography-guided transseptal puncture, maintaining an activated clotting time of >350 seconds. Left atrial mapping was performed with the VLCC using CARTO 3, and PVI was achieved through a standardized workflow, with both GRID and ABLt visualizations displayed throughout the procedure, consisting of 12 applications per PV, as previously described,³ and additional carina applications. After completion of the ablation workflow, post-PVI remapping was performed to confirm PVI and quantify voltage-derived low-voltage area (LVA). Mapping points were acquired using contact filtering based on tissue proximity indication. The PV antrum was defined as the region extending 5 mm from the ostium. Bipolar voltage amplitude of <0.5 mV during sinus rhythm defined LVA, and a 10-segment model was used to analyze the PV antra. When residual electrograms were present, pacing maneuvers were used when needed to differentiate far-field from local PV/antral signals. Ablation footprints were visualized as projected 1-mm³ GRID points or 2-mm projected ABLt. Post-PVI bipolar voltage maps quantified LVA per PV segment. Anatomical maps annotated with GRID or ABLt were used to estimate predicted lesion coverage by 2 electrophysiologists blinded to the voltage map. The predictive capability of GRID/ABLt was calculated using the formula GRID or ABLt Area in LVA / Total LVA and expressed as a percentage. Insufficient antral PVI area was defined by a bipolar electrogram amplitude of ≥0.5 mV within the antral PV segments. Group comparisons used t-test, Mann–Whitney, or χ² tests. Agreement between GRID or ABLt predictions and voltage-derived lesion assessment was evaluated using Cohen’s κ, correlation using Pearson’s coefficient, and agreement using Bland–Altman analysis. Interobserver reproducibility was assessed with intraclass correlation coefficients.

A total of 65 patients (63.6 ± 10.2 years, 50% male, 28.1% persistent AF) and 260 PVs were analyzed. PVI was achieved in all patients with a median of 22 applications (interquartile range, 20–24), and post-PVI voltage maps included a median of 550 mapping points per map (interquartile range, 410–730). First-pass isolation rate was 97.7% (left superior PV. 96.9%, left inferior PV, 96.9%, right superior PV, 98.5%, right inferior PV, 98.5%), whereas insufficient antralization occurred in 8.5% of veins (left superior PV, 10.8%, left inferior PV, 7.8%, right superior PV, 9.3%, right inferior PV, 6.1%). In cases of lack of first-pass isolation or insufficient antralization identified by post-PVI bipolar remapping, additional touch-up applications were performed. Agreement between GRID predictions and voltage-derived lesion assessment was higher than with ABLt for both first-pass isolation (κ 0.796 vs 0.743; P = .031) and antralization (κ 0.867 vs 0.625; P = .0001). Both GRID and ABLt correlated with pulmonary vein lesion area (r = 0.90 and r = 0.84, respectively; P < .0001). GRID demonstrated higher predictive capability than ABLt (95.8% vs 77.6%; P < .0001) (Figure 1). Bland–Altman analysis showed lower bias and narrower limits of agreement for GRID (bias 0.007; 0.592 to −0.577) compared with ABLt (bias −0.337; 0.578 to −1.252). Interobserver agreement was excellent for GRID (ICC: 0.984 [95% CI, 0.927–0.999]; P < .001) and lower for ABLt (ICC: 0.743 [95% CI, 0.653–0.834]; P < .001). Our study provides the first comparison of GRID and ABLt for predicting lesion coverage after VLCC-guided PVI, demonstrating that GRID achieved stronger agreement with post-PVI LVA, more accurate prediction of first-pass isolation and antralization, and higher interobserver reproducibility. In VLCC single-shot PFA workflows, lesion completeness depends on delivering sufficiently overlapped applications around the antrum, making intraprocedural decisions highly reliant on footprint visualization within the integrated EAM.^2,3 In this context, GRID provides a more faithful representation of lesion coverage than ABLt, enabling identification of residual gaps and targeted “bonus” applications. Mechanistically, GRID displays lesion coverage as a dense projection of small points approximating a continuous surface, whereas ABLt represents discrete markers with a predefined display radius that may obscure small uncovered gaps between applications or electrodes. Accordingly, the higher spatial resolution of GRID allows more accurate visualization of lesion distribution and facilitates detection of incomplete coverage. A prior randomized clinical study with radiofrequency ablation has similarly shown that grid-based annotation provides direct visual feedback on lesion continuity and catheter stability, improving identification of gaps along ablation lines.⁴ Consistent with this mechanistic advantage, GRID visualization also demonstrated substantially higher interobserver reproducibility than ABLt, supporting a more operator-independent assessment of lesion distribution. Some limitations of our study should be acknowledged. As this was a prospective exploratory analysis focused on lesion coverage assessed by post-ablation EAM, larger multicenter studies with systematic follow-up are needed to further validate these findings and assess their clinical impact. The conventional <0.5 mV bipolar voltage threshold may not be directly interchangeable between VLCC-based mapping and closely spaced multielectrode catheters; however, the same catheter, workflow, and voltage cut-off were consistently used for both GRID and ABLt comparisons. Although post-PVI voltage maps included a median of 550 mapping points, VLCC-based voltage mapping remains subject to limitations related to catheter contact, electrode spacing, and point distribution, potentially affecting absolute LVA estimation but not the within-procedure comparison between GRID and ABLt. Although GRID demonstrated superior acute voltage-derived lesion visualization, the present study did not assess lesion durability or long-term clinical outcomes. Therefore, these findings should not be interpreted as evidence of improved durable PVI or clinical efficacy. Moreover, they are specific to the VLCC PFA catheter integrated with CARTO 3 and should not be extrapolated to other PFA platforms. In conclusion, GRID-based lesion visualization demonstrated superior accuracy and reproducibility compared with ABLt in predicting effective lesion coverage during VLCC-guided PVI. GRID may therefore improve intraprocedural assessment of lesion distribution and guide more precise lesion delivery during PFA.

Figure 1.

Comparison between grid-based annotation (GRID) and ablation tag (ABLt) visualization for prediction of lesion coverage during VLCC-guided pulmonary vein isolation. LIPV =left inferior pulmonary vein; LSPV = left superior pulmonary vein; LVA = low voltage area; PV = pulmonary vein; PVI = pulmonary vein isolation; RIPV = right inferior pulmonary vein; RSPV = right superior pulmonary vein; VLCC = variable-loop circular catheter.

Disclosures

All authors have reported that they have no relationships relevant to the contents of this paper to disclose.