Value of fRagmented atrial electrOgrams and intracardiac echocardiography for identification of optiMal sites for radio-frequency delivery during cArdioneuroablation for asystolic reflex syNcope: the ROMAN 5 study

Cardioneuroablation (CNA) is a promising method for treatment of asystolic reflex syncope1; however, optimal methods for identification of presumed sites of ganglionated plexuses (GPs) have not been established. Anatomical approach uses computed tomography2,3 or intracardiac echocardiography (ICE),4,5 combined with 3D electro-anatomical mapping (EAM). Electrophysiological methods include high-frequency stimulation or recording of fragmented atrial electrograms (FAEGMs).6–9 The performance of FAEGM for identification of effective sites for radio-frequency (RF) applications during CNA has not been established. The study group consisted of consecutive patients who gave informed written consent to undergo CNA and participate in this prospective ROMAN 5 study (NCT06472102) (Ethics Committee approval 42/2024). Inclusion criteria were as follows: (i) severe, recurrent symptoms due to reflex syncope with documented asystole >3 s; (ii) ineffective prior non-pharmacological treatment; and (iii) positive baseline atropine test [sinus rate acceleration >30% and no atrio-ventricular block (AVB) following 2 mg of intravenous atropine]. The procedure was described in detail elsewhere.4 In brief, CNA was performed under general anaesthesia with muscle relaxation using RF energy, EAM system CARTO 3, and ICE. Our strategy is first to ablate paraseptal GPs from the left atrium, followed by ablation of these GPs from the right atrium. If vagal denervation is not achieved, RF applications close to the left pulmonary veins, end of Marshall tract, floor of the left atrium, and within coronary sinus are performed. Efficacy of CNA is evaluated using extra-cardiac vagal stimulation (ECVS), performed using two catheters positioned bilaterally in internal jugular veins, utilizing neurostimulator designed by Prof. Pachon (pulse amplitude of 1 V/kg body weight up to 70 V, 50 ms width, 50 Hz frequency, delivered over 5 s). Extra-cardiac vagal stimulation was performed after each RF application to assess the effect on vagal nerve. Radio-frequency applications were defined as ineffective (<10% change in sinus pause or AVB still present) or effective (complete or partial elimination of sinus pause and AVB). During the procedure, operator was blinded to the signals from ablation electrodes (signal amplitude was truncated to zero). Bipolar endocardial potential recordings were recorded before each RF application and analysed at 400 mm/s speed and filter setting 100–500 Hz. Three FAEGM parameters were measured: number of deflections, maximal amplitude, and total FAEGM duration.6–9 The analysis of FAEGM was performed offline by an investigator who was blinded to the procedural details. The mean value of three consecutive FAEGM recordings was taken as a final parameter. The results are presented as mean ± standard deviation or median and inter-quartile values. Continuous variables were compared using Student’s t-test or Mann–Whitney U test. Sensitivity, specificity, positive predictive accuracy (PPA), negative predictive accuracy (NPA), and total predictive accuracy (TPA) values were calculated. The reliability and agreement for the analysed parameters were evaluated using intra-class correlation coefficients (ICCs) and Bland–Altman analysis. The study group consisted of 40 patients (mean age 40.2 ± 12.7 years, 25 females) who had a lifetime history of a median 5.0 ± 8.0 syncopal episodes. Intraprocedural efficacy of CNA was 100%—in 34 (85%) of them ECVS became negative after ablation of paraseptal GPs and in the remaining 6 patients—after additional ablation of other GP areas. No complications were noted. The total number of RF applications was 367, out of which 126 were effective and 241 were ineffective. Fragmented atrial electrogram at effective sites were characterized by significantly greater number of deflections (7.00 ± 3.0 vs. 6.00 ± 3.00, P = 0.002), longer duration (47.00 ± 15.25 vs. 46.67 ± 13.67 ms, P = 0.045), and tended to have higher amplitude (0.55 ± 0.52 vs. 0.49 ± 0.54 mV, P = 0.085). The inter- and intra-observer reproducibility was high (ICCs 0.924–1.00), confirmed by biases for the number of deflections (0.05), amplitude (0.004 mV), and duration (1.8 ms). Performance of the number of deflections in predicting effective sites, calculated for various cut-off values, is shown in Graphical Abstract. The PPA values ranged between 34 and 43% and TPA between 35 and 60%. The highest sum of sensitivity and specificity values reached 117% for the FAEGM greater than or equal to six deflections. The present study showed that visual assessment of FAEGM is characterized by low positive predictive value (PPV), which limits its clinical use. The differences between effective and ineffective sites were statistically significant, but the absolute separation was relatively small, which calls into question their clinical significance. The concept that increased fractionation of the atrial signal may identify sites where extensions from epicardial GPs are sought to enter the myocardium was introduced by Pachon et al.1 Others also used FAEGM to guide CNA, using spectral analysis.6,7 However, there are several other potential sources for FAEGM, including areas of changes of fibre orientation such as posteroseptal region, pulmonary vein, or superior vena cava ostia. Also, the FAEGM-identified areas might represent neuro-myocardial interaction sites that are not involved in functional loops being addressed by ECVS or may be affected by the presence of other neural structures such as ventricular GPs.10 Thus, atrial signal fragmentation is probably a complex phenomenon. One recent report called into question the usefulness of FAEGM for guiding CNA.3 The present study is unique because the effect of each RF application was evaluated by ECVS. This enabled true assessment of which RF application was effective and which was not. Previous studies did not assess individually the effectiveness of each RF by ECVS. Although prospective, this is a single-centre study using EAM and ICE-based approach; thus, our findings are applicable only for such an approach. At sites where the very first RF application occurred effective, the NPA and specificity of FAEGM could not be calculated. Our results are confined to the paraseptal GPs areas only. We used visual inspection of FAEGM, which may be less accurate than automatic spectral analysis using dedicated software. Frequently, multiple observations were derived from the same procedure and therefore may not be fully independent. No adjustment for within-patient clustering was considered. Finally, real-time quick assessment of FAEGM during procedure may not be as accurate as offline detailed analysis; however, high reproducibility is reassuring and suggests that small differences are due to intrinsic biological discriminative value of FAEGM rather than measurement inconsistency. We would like to acknowledge the expertise and passion of our colleague and friend—the late Roman Piotrowski, who was one of the pioneers in cardioneuroablation and with whom we took care of more than 400 patients with reflex syncope who underwent cardioneuroablation in our institution. Funding by The Centre of Postgraduate Medical Education - Project no 17.