https://doi.org/10.1007/s00392-025-02625-4
1Immanuel Klinikum Bernau Herzzentrum Brandenburg / Kardiologie Bernau bei Berlin, Deutschland; 2Immanuel Klinikum Bernau Herzzentrum Brandenburg Bernau bei Berlin, Deutschland; 3Universitätsklinik der Salzburger Landeskliniken Klinik für Innere Med. II, Kardiologie u. intern. Intensivmedizin Salzburg, Österreich
Background: The presence of non‐pulmonary vein arrhythmogenic substrates has been reported to be a strong predictor of recurrence of atrial fibrillation (AF) after pulmonary vein isolation (PVI). The DR-FLASH score is a risk stratification tool for patients with AF, developed to predict such arrhythmogenic substrates. The latter can be identified as low voltage areas (LVA) in the endocardial map. The biomarker Myeloperoxidase (MPO) has been shown to play an important role in atrial structural remodeling and was recently associated with LVA as a surrogate for endocardial fibrosis and AF recurrence after ablation.
Aim: In this study, we examined the predictive value of the DR‐FLASH score for LVA with and without the addition of the biomarker MPO in patients undergoing redo ablation.
Methods: In this single-center prospective cohort study 50 consecutive patients underwent ultra-high-density mapping and redo ablation after primary PVI. LVA was defined as sites with a bipolar peak-to-peak voltage of <0.5 mV with an extent of >1 cm² and was set in relation to the left atrial surface. LVA classes, analogous to DECAAF I, were defined using the following cut-offs: I) LVA <10 %, II) ≥10 % to <20 %, III) ≥20 % to <30 % and IV) ≥30 %. The DR-FLASH score was calculated and serum MPO before ablation was analyzed using a commercially available ELISA (R&D Systems Inc., Minneapolis, USA). Statistics were performed using SPSS Version 29.
Results: Fifty patients were included in the study. Of those, 15 (30 %) had paroxysmal, 33 (66 %) persistent, and 2 (4 %) long-standing persistent AF. Baseline characteristics are shown in table 1. DR-FLASH score and LVA class distribution is shown in figure 1 A and B.
Overall, the median DR-FLASH score was 3 points (IQR 2). The DR-FLASH score significantly correlated with LVA in the endocardial map (absolute LVA r = 0.390, p = 0.005 and relative LVA r = 0.365, p = 0.010). The Receiver operator characteristics (ROC) of the DR-FLASH score showed an area under the curve (AUC) of 0.688 for LVA ≥ 10 % and 0.647 for LVA ≥ 20 % (figure 2 A and B).
Adding MPO increased the AUC of the model: 0.716 for LVA ≥ 10 % and 0.784 for LVA ≥ 20 % (figure 2 C and D). However, AUC comparison was not statistically significant (z = 1,244, p = 0,629 and z = 1,323, p = 0,907).
Eleven (22%) patients showed AF recurrence after redo ablation. Higher LVA classes were associated with higher recurrence rates (HR 7.4, 95 % CI 2.5 – 21.7, p < 0.001). The DR-FLASH score alone, which was particularly designed to predict LVA, did not predict recurrence. The combination of DR-FLASH and the biomarker MPO, however, was an acceptable predictor for AF recurrence at 6 months follow-up in 7-day Holter, AUC 0.744 (figure 2 E and F).
Conclusions: The DR-FLASH score was a fair predictor for LVA ≥ 10 % in our cohort. Adding the biomarker MPO increased the model’s diagnostic ability with an AUC > 0.700 for LVA ≥ 10 %, LVA ≥ 20 %, and AF recurrence in patients undergoing redo ablation.