To our knowledge, this is the first study in which the SP-L was calipered during RF ablation of AVNRT. Since it was successfully ablated strictly from the right side, without any RF delivered from inside the CS ostium, SP seems anatomically consistent with a right posterior extension, identified as a discrete posterior continuation of the compact AV node [14]. Furthermore, the fewer junctional cycles observed during RF application at the posterior end of the SP compared with the site of successful ablation (Table 2) might reflect differences in thermally mediated enhanced automaticity between the perinodal tissue and the AV nodal tissue near the compact AV node. Thus we believe that, when using a stepwise approach along the tricuspid annulus, the posterior end of the right posterior extension could be detected by triggering the junctional rhythm as a marker of AV nodal tissue ablation [27‐30]. SP-L, which in the present study was measured precisely between the most proximal H and the proximal end of the SP, might have reflected the actual conduction across the SP during tachycardia. Although whether SP-L is similar in patients with versus without AVNRT needs to be verified, the individual variations in SP-L observed in the present study may reflect the expected anatomic variations of a normal AV node.

As observed in previous anatomic and clinical investigations [9, 13], the dimensions of Koch’s triangle varied considerably among our patients, and were similar to the mean 17 ± 3 mm, which was measured post mortem and during surgery [21], suggesting that we studied a typical population. In contrast to previous studies, which used two-dimensional fluoroscopy to measure Koch-L [9, 10, 16], our calipered measurements by electroanatomic mapping were likely to reflect accurately its individual variations. It is noteworthy that our population included two patients who underwent successful ablation posterior to the superior margin of the CS ostium, whose Koch-L was <10 mm.

An accurate prediction of the anatomy of the AV node before ablation of AVNRT in each individual patient would be helpful, especially the location of the right posterior extension in Koch’s triangle. Several anatomic determinants of successful SP ablation have been identified [7, 9, 14‐16]; however, the precise location of the SP in Koch’s triangle has not been reported in a clinical setting. In the present study, SP-L was correlated with Koch-L, indicating that the latter was a predictor of SP-L. However, the location of the posterior end of the SP was not confined to the anterior region of the inferior CS ostium margin, suggesting that such an anatomic landmark cannot simply be used to predict the location of the posterior end of the SP. Inoue and Becker [13] found that, in most cases, the AV nodal bundle axis reaches the level of the anterior margin of the CS ostium, and that in one-third of cases, the posterior bundle extends beyond the anterior margin of the CS ostium, which is consistent with our observations.

In the present study, the AH interval during AVNRT without infusion of isoproterenol was positively correlated with SP-L. These findings suggest that anatomic factors, such as SP-L contribute to differences in the AH interval during tachycardia observed among patients, besides functional factors, such as cellular electrophysiologic characteristics and autonomic nervous activity. In contrast to these previous studies [24, 31, 32], the AH interval during tachycardia in our study was not correlated with the site of successful ablation. This is probably (a) because in some patients, the site of successful ablation was not consistent with the precise location of the posteroseptal atrionodal input (entrance) of SP [25], as described in detail later, and/or (b) a consequence of methodological differences in the localization of the successful ablation site. In addition, the AH interval at the first AH jump (SP-T1) was not indicative of the SP-L, probably because decremental conduction over SP at the time of jump was variable among patients. The calculated mean 4.6 ± 1.5 cm/s conduction velocity across SP without infusion of isoproterenol closely resembled the 6.9 ± 2.4 cm/s measured in the rabbit heart, using fluorescent imaging of AV nodal reentry [33].

Although one might hypothesize that our successful ablation site was in a more superior location than previously described [3, 9], we believe that it was nearly identical despite differences in (a) the calipering between fluoroscopic and EAMS guidance and (b) the starting point of the interval between the largest and the proximal His-bundle electrogram. Importantly, a positive linear correlation between SP-L and SucABL-L suggests that (a) SP-L may be one of the factors responsible for individual variations in the site of successful ablation, and (b) successful SP ablation was achieved by selectively targeting the posterior atrionodal input in most patients, as in the study of Markowitz et al. [34]. However, in a minority of patients, ablation was successful away from and anterior to the posterior end of the SP, following ineffective RF applications near its posterior end. This might have two distinct explanations. First, assuming that only part of SP, which was wider at its entrance than at other more superior sites [34], functioned as an essential component of the tachycardia circuit, a point ablation at its posterior end might have caused an incomplete lesion, allowing the persistence of AVNRT. Alternatively, in some patients, instead of being located along the tricuspid annulus near the inferior CS ostium, the SP might have been located between the CS ostium and the tricuspid annulus. In such cases, a linear ablation across the wide entrance of SP might eliminate the tachycardia [16]. Second, SP might run deep in the endocardium, not readily reached by RF applications from the right atrium. We believe that such an SP might be a variant of a right posterior instead of a left inferior extension since, in contrast to earlier studies [2, 3, 5, 15], none of our patients needed the delivery of RF from the left side or from inside the CS ostium to ablate a left inferior extension [1, 6, 7, 35, 36].

Albeit low, the risk of AV block complicating ablation of the SP is not negligible [1‐7]. One of the possible causes of this complication might be that the same conventional techniques, whether anatomically or electrogram-guided, are systematically implemented in each case, irrespective of the individual SP-L. The present study, however, identified interindividual variations in SP-L, which were correlated with Koch-L, AH interval during AVNRT, and the site of successful ablation. This new information suggests that a patient-based strategy might be more appropriate. For example, the observation of a long Koch-L or a long AH interval during AVNRT would encourage a posterior ablation, away from the His-bundle, whereas the finding of a short Koch-L or AH interval would be indicative of a heightened risk of AV nodal injury. In addition, our observations suggest that ablating anteriorly, away from the posterior end of the SP, was not a prerequisite for successful ablation procedures. This strategy, therefore, might be a means of lowering to a minimum the risk of AV block as well as the number of unnecessary RF applications.

A first limitation of our study is the definition of the posterior end of the SP. There is currently no evidence of an anatomic correlation between the sites of induction of junctional rhythm and the most atrial end of the SP. In addition, the AV nodal or perinodal tissue during RF delivery might be heated by both resistive and conductive heat. When the SP is in a deep endocardial layer, the heating effect might not reach its posterior end. Furthermore, no junctional activity might occur during RF application at the site of successful ablation [37]. In these unusual circumstances, the most posterior site associated with the development of junctional rhythm during RF application might not be the posterior end of the SP. Second, the AH interval during tachycardia was measured in only one-third of the population whose tachycardia was inducible at baseline. Third, although the actual SP-T was between the earliest His electrogram and the atrial electrogram at the sites of successful ablation, we did not map that atrial electrogram during the tachycardia. Fourth, we did not evaluate the output, duration of RF energy, and tissue temperature on the development of junctional rhythm during RF application.