Patients

We performed a retrospective analysis of patients with WPW syndrome who underwent EPS and RFCA in our center between January 2000 and December 2012. Eligible patients were those who had a resting electrocardiogram (ECG) of adequate quality and standard calibration (25mm/s and 10mm/mV) which revealed an obvious preexcitation pattern with the following characteristics: 1) QRS duration ≥ 120 ms, 2) short PR interval and 3) Clear delta wave. Patients with congenital or acquired heart disease, multiple AP, bundle branch block or previous attempt of catheter ablation were excluded.

ECG analysis

The best available ECG of each patient taken before ablation, showing a clear preexcitation pattern, was analyzed by three experts who were unaware of the outcome in the RFCA. Using the QRS-Polarity algorithm the experts localized de AP in one of the five possible anatomical positions of this algorithm: Left anterolateral/left lateral (LAL/LL), Right antero septal (RAS), left postero septal/left posterior (LPS/LP), right posteroseptal/right posterolateral (RPS/RPL) and right lateral (RL) (Figure 1). To assess the concordance of the interpretation of the algorithm, the scores obtained by each researcher were compared to the average among the 3 researchers combined.

The results were regarded to be consistent when the three researchers came to the same conclusion about the location of the AP, regardless the location was correct or not, and were regarded to be jarring when they predicted different locations. The locations inferred by the algorithm were validated using the anatomical locations of AP that had already identified in the EPS with a successful RFCA. Prediction of the location of the AP employing the algorithm were considered accurate if it corresponded to the same anatomical position as the one identified in the EPS, was considered adjacent when the location was in a surrounding location and incorrect when it was far from the exact anatomical region determined in by the EPS.

Electrophysiological testing

EPS was carried out after signing the informed consent, we used general anesthesia in children and if it required sedation in adults. The presence and AP location was determined by conventional methods of mapping with multipole catheters (identifying the earliest ventricular or atrial signal during anterograde or retrograde conduction through the AP, respectively, or directly recording the potential of the AP)⁸.

The location of the AP was assessed using the right and left anterior oblique views according to the position of the ablation catheter at the successful ablation site. For the left AP, a transaortic approach was used with heparin infusion to maintain an activated clotting time (ACT) between 250 to 350 seconds. To determine the site of a successful ablation we look for sites where the atrial electrogram merged with the preexcited ventricular electrogram. These ventricular electrograms should precede the beginning of the delta wave registered in the surface ECG. The final location was confirmed in the place where the ablation was successful, attempts were considered successful when the disappearance of the delta wave occurred in < 10 seconds after applying temperature-controlled radiofrequency⁹, finally, the anatomical site was determined by the fluoroscopic position of the ablation catheter on the plane of the mitral or tricuspid annulus.

Parahisian AP have great variability in terms of electrocardiographic findings⁹. Although the QRS polarity algorithm was not designed for parahisian AP the vast majority of these are antero-septal and mid-septal pathways¹⁰, therefore the algorithm could be useful to give an approximation of the location of these AP in the right antero septal position. An AP was classified as parahisian if the potential of the His bundle was recorded simultaneously with the ventricular or atrial insertion of the pathway in the ablation catheter.¹⁰.

Statistical analysis

Baseline characteristics were presented as numbers (percentage,%) for categorical variables and mean (± SD) for continuous variables. We applicated Cohen's kappa coefficient and Pearson coefficient to determine the correlation between the locations predicted employing the surface ECG with the locations found in the EPS, after that we determinate the global accuracy. We also determined the specific accuracy for each of the five zones of the QRS-Polarity algorithm and we determined the algorithm's precision for the parahisian AP, and we compared it with the results obtained by other 5 algorithms: D' Avila¹², Fitzpatrick¹³, Arruda¹⁴, Chiang¹⁵ and Boersma¹⁶.

Global accuracy

When we correlated the location predicted by the algorithm and the true location determined by the EPS and RFCA, to determine the accuracy of the algorithm we obtained a global kappa coefficient score of 0.78 and Pearson's coefficient of 0.90. We found that for the left lateral AP the correlation was very good with a kappa coefficient close to 1; in this group, 144 patients had correct correlation (97.9%), only 3 patients had adjacent locations and no patient an erroneous location. On the other hand, the right and left posterior AP had greater variability (Table 2).

The algorithm had a better diagnostic performance for the AP in the LL/LAL zone, with 90% of sensitivity, 98% of specificity, 98% of positive predictive value (PPV) and 93% of negative predictive value (NPV). The sensitivity and PPV for the other areas were lower, but all of them had specificity and NPV above 90% (Table 3).

Accuracy for the parahisian AP

There were 26 patients with a parahisian AP, which represents an incidence of 7.14%. In these patients the mean age was 24 ± 17 years, and 61% were male, the mean of fluoroscopic time was 27 ± 20 minutes, and the mean time since the beginning of the radiofrequency to the loss of the preexcitation was 3.6 ± 2.5 seconds, similar values comparing with the other locations (Table 4).

Patients with parahisian AP had a greater heterogenicity in the electrocardiographic features. We found a great variability in the polarity of the QRS complex in DIII, V1 and V2 leads, which are employed in the QRS-Polarity algorithm. The QRS polarity in DIII were positive in 34.6% patients and negative in 61.5%, in V1 7.6% were positive and 88.4% were negative and in V2 57% were positive and 38.4% were negative (Table 5).

Another important finding is that in patients with parahisian AP, the QRS polarity in the DI was always positive, finding that had been previously reported by Haïssaguerre et al¹⁰. We also noted that these patients had a great variability in the concordance of the QRS polarity in inferior leads, 50% had concordance in leads DIII and DII, while 34.6% showed discordance in the same leads (Table 5).

Finally, we made a comparison of the accuracy between the QRS-Polarity algorithm with the others algorithm which are much more complex. Comparing with the rest of the algorithms, the QRS-Polarity has an accuracy as high as the Fitzpatrick's algorithm, which is probably the most complex.

Employing the QRS-Polarity algorithm, 34.6% patients had a correct anatomical location, 42.3% had an adjacent location and only 23% an incorrect location. While, with the Fitzpatrick's algorithm, 30.7% had a correct anatomical location, 50.0% had an adjacent location, and 19.2% an incorrect location, on the other hand employing Arruda's algorithm, only 11.5% had a correct anatomical location, 73.1% had an adjacent location, and 15.4% an incorrect location. The others (d'Avila, Chiang and Boersma) do not achieve comparable levels of accuracy (Figure 3).

Figure 3 Accuracy of the different algorithms for parahisian accessory pathways. 

Accuracy for parahisian accessory pathways

There are many difficulties for all algorithms, one of them is the location of the AP in the parahisian zone, since the proximity with the normal conduction system, can make the preexcitation patter not as evident as other locations also generates much heterogeneity in ECG findings¹⁰,¹¹. Wren et al. demonstrated that in children with WPW syndrome, no algorithm has a good accuracy for mid septal AP and right antero septal AP, areas where the parahisian AP are located¹. Nonetheless, we demonstrated that if we considered useful the exact and the adjacent anatomical location, the QRS-polarity algorithm has an accuracy of 76.9% for parahisian AP.

Our results are different from those reported by Texeira et al, they also considered the exact and the adjacent positions as good locations for determinate the accuracy, they found that for septal AP the accuracy of the QRS-polarity algorithm was 68.6% and for Fitzpatrick's algorithm was 69.6%⁵. These differences could be explained because the anteroseptal AP are not necessarily Parahisian.

We must clarify that, although the QRS-polarity algorithm does not include specifically the mid septal and parahisian pathway, it does not reduce its power. These anatomical positions were not considered due to the fact, that in these locations it is less common to see a clear preexcitation pattern, this affects the precision of any algorithm. A recent publication considers that the anteroseptal and parahisian zones are the same⁷ but the AP will be considered as parahisian only if the His bundle potential is recorded at the same time as the ventricular or atrial insertion. Therefore, a correct location for the parahisian AP according to the QRS-Polarity algorithm would be the right anteroseptal region.

Limitations

Like all algorithms there is an inherently limitation produced by biologic variability in anatomy, presence of multiple pathways, intrinsic ECG abnormalities and variation in degree of pre-excitation. Most algorithms attempt to determine the AP location using the resting ECG, but the degree of preexcitation at rest varies from one patient to another²². Interobserver variation also plays an important role.

This trial has the inherent limitations of any retrospective analysis. It is also unicentric. The relatively small number of parahisian AP in our population is probably one of the more important limitations, so our results about the parahisian AP need to be confirmed by another trial or meta-analysis.