Our center recruited a consecutive cohort of patients who underwent catheter ablation for AT using the Rhythmia mapping system (Boston Scientific, Marlborough, MA, USA) from October 2019 to July 2022. We reviewed the electrophysiological study (EPS) files and analyzed the LA mapping points’ voltage information. While there is no universally accepted cutoff for low LA, a mean LA voltage 0.5 mV was selected as the threshold value based on prior studies [11, 12]. Only LA tachycardias with a mean LA voltage 0.5 mV, identified by high-density mapping, were included in the study. All patients provided written consent to review and include their medical records. The study was approved by Shanghai Chest Hospital Ethics Committee (IS22041) and conducted in accordance with the Declaration of Helsinki (as revised in 2013).

The ablation and high-density mapping procedure details are described in supplementary materials. Briefly, a decapolar coronary sinus (CS) catheter was inserted via femoral vein access for pacing, using a bipolar CS potential as a stable reference for local activation time. Heparin was used to maintain an active clotting time between 300-350s. An IntellaMap Orion${}^{\text{TM}}$ multipolar catheter (Boston Scientific, MA, USA) was used for electroanatomic mapping under the guidance of Rhythmia mapping system. Unipolar and bipolar electrograms were combined to obtain accurate annotation of the local activation time of each bipolar electrogram. For fragmented or multiple potential electrograms, the timing in the surrounding area was used to select potential for timing annotation.

If the atrial wave was preceding in the proximal CS lead, Right atrial (RA) mapping was conducted first. If the atrial wave preceded in distal CS leads or RA mapping results suggested LA-originated AT, LA mapping was performed via transseptal approach. This included cases where (1) a missing total cycle length (CL) of $>$10%; (2) RA activated in a centrifugal pattern and earliest at RA septum. AT mechanism and critical area were analyzed after activation mapping. This was followed by ablation with an irrigated ablation catheter (Intellatip MIFI, Boston Scientific, MA, USA; 43 ℃, 35W, irrigation rate 12 mL/min).

The AT mechanism was diagnosed by analyzing the activation map: (1) the origin of AT (LA, RA, or bi-atrial) was identified by CL coverage or earliest activation site within the atrium/atria and (2) the precise AT circuit was identified via activation propagation. The window of interest was set at the CL value, and the propagation of AT wavefront was visualized with a 5–10 ms window of activation and was advanced along the timescale step by step.

Bi-loop AT was defined as two simultaneous macro-reentrant circuits using a common isthmus. Bi-atrial tachycardia (biAT) was diagnosed if the circuit used both LA and RA through two interatrial connections. The possibility of biAT should be considered when a local breakthrough is observed near interatrial connections including the Bachmann’s bundle, fossa ovalis, posteroinferior interatrial connection, and CS ostium. Epicardia connection mediated AT (epiAT) was diagnosed if the wavefront propagation showed a ‘jump-frog’ pattern with focal activation after bypassing the atrial conduction barrier.

The Rhythmia system developed an algorithm named ‘Confidence Mask’ to assess the reliability of mapping points. This process included two factors: the amplitude of voltage and annotation time consistency. In order to avoid introducing artifacts into the mapping result, mapping points with voltage lower than set threshold or inconsistent local activation time were excluded from activation mapping, and the corresponding area was colored grey. LVTA was performed only in low-voltage mapping points without annotation time inconsistency, which usually exhibits disordered colors, to avoid yielding confusing results.

To the best of our knowledge, there is no absolute voltage cutoff to guarantee the absence of conduction. The threshold for unexcitable myocardium was limited by the BGN level of the mapping system. The Rhythmia system had a low BGN level of 0.01 mV and used a default low voltage threshold (LVT) value of 0.03 mV [4]. When a grey area was present in LA, especially if the mapping points within LVZ appeared to be uniform in color and related to the surrounding area with propagation, LVT was manually tuned to 0.02 mV and 0.01 mV to include as many points into activation mapping as possible. If inconsistent automatic annotation was observed in a few discrete points, manual annotation was performed. Then AT mechanism was re-analyzed.

After re-identifying the AT mechanism following LVTA, radiofrequency catheter ablation was performed. (1) For macro-reentrant AT and bystander conduction gap, linear ablation targeted the CI, the narrowest part of the circuit with slow conduction. (2) For focal AT, the earliest activation site with fractionated electrogram was targeted or in a linear fashion from the circuit to an anatomical barrier or an area of conduction block.

Effective ablation refers to AT termination with sinus rhythm restoration or AT circuit change, the latter of which included changes in conduction path and activation sequence and could manifest as abrupt and sustained changes in (1) global propagation, (2) local conduction connection or exit without altering AT global propagation, and (3) reversal in conduction sequence without changing circuit path. The ablation endpoint included (1) AT termination and sinus rhythm restoration; (2) subsequent substrate modification targeting potential substrate facilitating other ATs, including bystander conduction gaps, slow conduction zone, or local area with complex fractionated atrial electrogram.

Electrocardiographic monitoring was applied to all patients during hospitalization. And after discharge, patients underwent regular clinic visits and were assessed with 12-lead electrocardiography and 24 h Holter monitoring.

Continuous variables were expressed as mean $\pm$ standard deviation if normally distributed or medians (range) if abnormally distributed, and their comparisons were conducted by paired-samples t-test or Wilcoxon rank sum test. Event-free survival was estimated by Kaplan-Meier method. A p 0.05 was considered statistically significant. Statistical analysis was performed by SPSS 26.0 (IBM Corp., Armonk, NY, USA).

From October 2019 to July 2022, 235 patients underwent catheter ablation for AT in our center by using Rhythmia mapping system, among which 42 patients were enrolled. Forty patients had histories of iatrogenic intervention which included catheter ablation for atrial fibrillation in 28 patients (66.7%) and cardiac surgery in 12 patients (28.6%). All 12 patients underwent mitral valve replacement following cardiac surgery. Additional procedures accompanying these replacements included tricuspid valve plasty (TVP) in 4 patients, MAZE in 2 patients, TVP plus MAZE IV in 3 patients, and TVP plus aortic valve replacement in 1 patient. The baseline characteristics of the patients are summarized in Table 1.

In total, 55 ATs were identified, with 13 patients having multiple ATs. Of these, 36 ATs underwent LA mapping, which was completed in an average time of 10.9 $\pm$ 3.4 minutes. The remaining 19 ATs were mapped with both LA and RA, with an average time of 16.8 $\pm$ 2.7 minutes. In terms of mapping points, the mean numbers were 11047.0 $\pm$ 4652.6 for LA and 5629.8 $\pm$ 1265.7 for RA. The mean total CL was 248.1 $\pm$ 52.8 ms, and manual annotation was performed in only 33 points in 5 AT maps.

AT mechanisms identified included 53 macro-reentrant ATs and 2 focal ATs. The macro-reentrant ATs were further divided into (1) 6 biATs; (2) 11 epiATs, with epicardial connections including Marshall ligament in 7 ATs, LA anterior wall lesion in 2 ATs, and LA inferior wall in 2 ATs; and (3) 36 endocardial ATs, including 10 bi-loop ATs and 26 single-loop ATs. The locations of CI included LA anterior wall in 21 ATs, LA posterior wall in 5 ATs, LA roof in 7 ATs, LA inferior wall in 4 ATs, LA lateral wall including mitral isthmus in 18 ATs, LA septum in 1 AT, and CS in 1 AT.

At default LVT (0.03 mV), the spatial relationship between LVZ (grey area) and AT circuit or CI was categorized as follows: (1) away from AT circuit, (2) adjacent to CI, (3) at the corresponding epicardial or endocardial side of CI, and (4) inside the AT circuit. Under this default LVT, the median grey area in the patients was 4.6 cm${}^2$ (range, 0.36–15.5 cm${}^2$), and decreased to 2.4 cm${}^2$ (range, 0–11.9 cm${}^2$, p 0.001) and 0.3 cm${}^2$ (range, 0–6.9 cm${}^2$, p 0.001) when LVT was set to 0.02 mV and 0.01 mV, respectively. In this adjusted setting, some points within previously grey area were assigned colors, allowing for the observation of activation propagation in the newly-colored regions (Fig. 1).

The uncovered propagation impacted AT map interpretations differently, and the ATs were grouped into four categories based on the impact of LVTA (from 0.03 mV to 0.01 mV). The AT mechanism identification and subsequent ablation strategy will be described in the following section.

Type 1:

In Type 1, the AT mechanism and CI underwent complete changes, leading to the redesign of the subsequent ablation strategy. This included 9 ATs in 9 patients. The LVZ was located either within the predefined AT circuit or adjacent to the predefined CI. The characteristics of this type of AT have been summarized in Table 2. This Type 1 AT can also be subcategorized into different subtypes, based on the interpretation of activation mapping both before and after LVTA.

(1) Type 1.1 consists of uniAT/ biAT, including 2 ATs. The LVZ was located adjacent to inter-atrial connections including LA anterior wall (Bachmann’s bundle branches) and right pulmonary vein antrum (posterior interatrial connections) (Fig. 2, Supplementary Videos 1 and 2). (2) Type 1.2 is characterized by epiAT/ non-epiAT, including 3 ATs. The LVZ was found adjacent to the LA roof corresponding to the distribution of Bachmann’s bundle branches or mitral isthmus (including Marshall ligament and CS) (Fig. 3, Supplementary Videos 3 and 4). (3) Type 1.3 consists of a single-loop AT/ bi-loop AT, and includes 3ATs. The LVZ was located adjacent to the predefined circuit (Fig. 4, Supplementary Videos 5 and 6). (4) Finally, Type 1.4 represents a single-loop AT/ focal AT or a single-loop AT (circuit change), including one AT. The mapping result suggested focal AT at default LVT, but changed to a single-loop epiAT after LVTA (Fig. 5, Supplementary Videos 7 and 8).

Type 2:

Type 2 cases are characterized by activity following LVTA. While AT circuit and the location of CI remained unchanged, the activation propagation (within the grey area adjacent to predefined CI) was revealed, potentially facilitating activation conduction (Fig. 6, Supplementary Videos 9 and 10). In response, the subsequent ablation strategy was tailored by increasing ablation to cover the complete CI, avoiding conduction gap. The extents of CI before and after LVTA were 14.1 $\pm$ 4.9 mm and 28.6 $\pm$ 7.3 mm, respectively (p 0.001). The width of the conduction corridor within the LVZ measured 14.5 $\pm$ 7.0 mm. Type 2 included 16 ATs in 14 patients. The characteristics are summarized in Table 3.

Type 3:

In Type 3, the AT circuit and CI remained unchanged, but after adjusting the LVTA to 0.01mV, a bystander conduction gap was discovered (Fig. 7, Supplementary Videos 11 and 12). Consequently, after AT termination and sinus rhythm restoration, additional ablation targeting the bystander conduction gap was performed. This type included 3 ATs in 3 patients, and the LVZs were away from AT circuit.

Type 4: After LVTA, AT circuit and CI remained unchanged, and subsequent ablation strategy stayed unaltered. This type included 27 ATs in 20 patients. The following characteristics of LVZ could be observed in this type of AT after LVTA. Type 4.1: complete conduction block was observed within LVZ, regardless of its spatial relation with AT circuit, including 12 ATs in 10 patients (Fig. 8, Supplementary Videos 13 and 14). Type 4.2: passive activation propagation could run through LVZ which was located away from the predefined AT circuit at default LVT, including 15 ATs in 11 patients (Fig. 9, Supplementary Videos 15 and 16).

Catheter ablation was performed to target the CI for macro-reentrant AT or earliest activation site for focal AT. Effective ablation was achieved in 2 focal ATs, with termination in one AT and circuit change in the other. Out of 53 macro-reentrant ATs, effective ablation was obtained in 48, leading to sinus rhythm restoration in 32 ATs and AT circuit change in the other 16 ATs. In all three Type 3 ATs, additional linear ablation was performed targeting bystander conduction gaps, after which bidirectional block of linear ablation was confirmed.

Catheter ablation was ineffective in treating five macro-reentrant ATs, including one Type 2 AT, one Type 3 AT, and three Type 4 ATs. The reasons for this ineffectiveness included (1) constant changing total CL in 3 ATs, which made mapping unfeasible, and (2) ablation at LA roof where Bachmann’s bundle was distributed with thick myocardium in 2 ATs. Cardioversion was performed to restore sinus rhythm.

At a median follow-up of 16.5 months, the cumulative freedom from AT was 69.3%. After a follow-up of 18.0 $\pm$ 9.6, recurrent AT was observed in 16 patients.

Previous studies have examined the characteristics of scar-related ATs from multiple perspectives. For example, Macro-reentrant bi-atrial ATs after AF ablation or cardiac surgery have been reported in multiple studies [3, 13, 14]. Takigawa et al. [15] found a close correlation between macro-reentrant AT and ablation-induced LVZ in patients after AF ablation. Tsai et al. [16] found that LVA with conduction slowing was potentially predictive of CI for AT. Our previous study also observed that the characteristics of left atrial anterior wall AT correlated with catheter ablation or surgical incision [17]. While the significant role of LVZ as a substrate for AT has been verified, the impact of activation propagation within LVZ on AT mechanism analysis remains unclear.

In our study, 40 patients had a history of cardiac interventions. This included 28 patients (66.7%) treated with radiofrequency catheter ablation and 12 patients (28.6%) who underwent cardiac surgery. Both treatments are capable of inflicting considerable iatrogenic injury to atrial myocardium leading to myocardial scars. The presence of these scars likely explains the extensive low voltage zone and reduced average atrial voltage observed in our study group. In patients with low LA voltage, usually due to physiological atrial remodeling or iatrogenic intervention, conduction challenges can arise. Anatomical structures including the mitral annulus and iatrogenic scars may lead to AT by creating conduction obstacles and slow conduction areas. This complex landscape can make the mapping and analysis of AT a challenging task. Additionally, LVZ may contain bundles of viable myocardium of variable size separated by fibrosis, leading to fractionated, split or late potentials, resembling near-field abnormalities [18]. Differentiating these from BGN can be challenging, resulting in potentially confusing and misleading mapping outcomes. In our study, activation propagation within LVZ was observed in almost half cases (Type 1, 2, and 3), which aided in accurately identifying the true CI and in designing subsequent ablation strategies.

In the activation map of AT, the points and surrounding areas with bipolar voltage lower than LVT are colored in grey. This coloring is meant to exclude these areas from visualized activation propagation analysis and prevent low-voltage points from affecting the color of activation map. However, such exclusion may result in the loss of crucial information for AT analysis, particularly in patients with an unsatisfactory LA substrate, and therefore may influence the physician’s judgement. Hence, its reasonable to provided sufficiently low and reliable LVT, improving the accuracy of AT activation mapping in this subgroup of patients. This can be achieved through a comprehensive analysis of as many electrograms within the LVZ as possible.

When analyzing AT, LVT plays a critical role in preserving the integrity of the activation mapping. The LVT can be adjusted as needed, but should remain above the BGN level of the mapping system, approximately 0.01 mV for Rhythmia [4]. This setting allows for an extremely low LVT, which preserves the necessary accuracy for timing annotation. Additionally, it avoids introducing artifacts into the map, a factor that is highly beneficial for AT mechanism analysis in patients with extensive LVZ.

In our study, we found the use of LVTA significantly enhanced the interpretation of AT mapping in nearly half of AT cases. This included identifying the true AT mechanism (Type 1), tailoring the extent of ablation (Type 2), and uncovering bystander conduction gaps (Type 3). And among all the cases of Type 1 and Type 2 ATs, effective ablation was obtained in all but one instances of Type 2 AT, suggesting the efficacy of LVTA in pinpointing the AT mechanism identification and achieving acute ablation success.

Despite the limited size of our study, the results suggest potential correlations between LVZ location and inadequate AT mechanism analysis. It is suggested that LVTA be conducted under the following conditions. (1) When LVZ is located at inter-atrial connections: if LVZ is found in inter-atrial connections including CS ostium, fossa ovalis, posterior interatrial connections, and LA roof/anterior wall corresponding to distribution of Bachmann’s bundle, there may be a risk of misinterpreting biAT/uniAT. (2) When LVZ overlaps with predefined CI: if LVZ coincides with the predefined CI at corresponding endocardial/epicardial locations, usually at mitral isthmus or LA/RA septum, the possibility of misinterpreting epicardial/non-epicardial AT should be evaluated. (3) When LVZ is adjacent to or inside the predefined AT circuit: in situations where the LVZ is adjacent to or within the predefined AT circuit, potential misinterpretations such as bi-loop/single-loop AT, single loop/focal AT, or single-loop AT, and underestimated CI extent should be considered.

During follow-up, recurrent AT was observed in some patients, highlighting a complex issue that may be attributed to several factors. First, this study included patients with a low mean LA voltage of less than 0.5 mV, suggesting extensive LA fibrosis and areas of slow conduction. Second, rheumatic heart disease was present in almost a quarter of patients, a factor contributing to ongoing cardiac remodeling. Moreover, multiple ATs with different mechanisms were identified in 13 patients, underscoring the complexity of the atrial substrate. Although the follow-up result were acceptable, these findings stress the need for further studies to develop ablation strategies that could reduce the risk of recurrence.

With the help of high-density mapping, AT mechanism analysis could be conducted in a fast and detailed manner, achieving satisfactory accuracy in activation mapping under most conditions. The default LVT settings led to misinterpretations in nearly one-sixth of Type 1 cases with low LA voltage, particularly in patients with extensive LVZ due to the complex AT mechanism.

Accurate identification of unipolar and bipolar signals during AT is the cornerstone of reliable potential measurement and activation timing annotation. This can be challenging when mapping extensively scarred myocardium, resulting in yielding dubious mapping results. By using a basket-shaped mapping catheter with 64 unidirectional electrodes (0.4 mm${}^2$ and 2.5 mm spacing), the Rhythmia provides a higher resolution and is less influenced by BGN and far-field signals compared to conventional ring-electrodes. The close inter-electrode spacing design also facilitates recordings of a higher bipolar voltage amplitude, improving the signal to noise ratio. Overall, this significantly improves the details of activation mapping, allowing for meticulous illustration in activation mapping interpretation. Furthermore, based on detailed mapping information with adequate accuracy, prudent LVTA can recognize inconspicuous conduction within LVZ by aquiring previously unaccessible voltage data. This extends the comprehension of AT mechanisms and reveals the true CI. Under its guidance, the efficacy of LVTA was partially verified by subsequent effective ablations.

There are several limitations to this study. Firstly, the power of our study is limited by its small sample size, impacting the generalizability of the findings. Secondly, the term ‘low atrial voltage’ remains undefined in this context. While we evaluated the global atrial substrate with mean atrial voltage using an exploratory cutoff value of 0.5 mV, this might not be universally applicable. The data for recorded bipolar voltage may differ depending on the mapping system. Consequently, while the study results are pertinent to the Rhythmia mapping system, extrapolation to other ATs should be done with caution. One of Rhythmia’s strengths lies in the accurate and reliabile identification of unipolar and bipolar potentials. The unidirectional and densely arranged electrodes enable indirect assessment of the catheter contact. However, the catheter doesn’t provide information on contact force, leaving a possibility of low contact during mapping. Our current findings do not fully endorse the benefit of LVTA guided ablation strategy for Type 3 ATs. Further randomized controlled studies are warranted for verification.