https://doi.org/10.1007/s00392-025-02625-4
1Universitätsklinik Schleswig-Holstein, Campus Kiel Klinik für Kardiologie und internistische Intenisvmedizin Kiel, Deutschland; 2Universitätsklinikum Schleswig-Holstein Innere Medizin III mit den Schwerpunkten Kardiologie, Angiologie und internistische Intensivmedizin Kiel, Deutschland; 3Universitätsklinikum Schleswig-Holstein Innere Medizin III mit den Schwerpunkten Kardiologie und internistische Intensivmedizin Kiel, Deutschland
Background and Aims
Automatic intracardiac pattern matching (aICPM) can be used to monitor the stability of atrial tachycardia (AT) during activation mapping or to perform an atrial pace mapping (APM) for localizing the origin of non-sustained atrial tachycardia (nsAT), or atrial fibrillation (AF) triggers. When aICPM is performed, IC pattern similarity score can fluctuate despite a stable pacing site and the score is usually less than 100%, even when paced from the site of origin (SOO). We assessed the underlying cause of the fluctuation and aimed to define an optimal cutoff, which ensures to accurately detect the SOO by aICPM.
Methods
aICPM was performed with six biatrial unipolar signals to create two sets of score maps: one for the stable rhythm from the same SOO and one for paced IC patterns outside the SOO. The dispersion of the IC pattern similarity score was assessed in 17 consecutive patients during stable focal (n=5) and reentry (n=12) AT. For the same SOO score map, an IC pattern of the first atrial beat was acquired and automatically matched to the subsequent IC morphologies without catheter manipulation to exclude the mechanically induced beats. For the score map with the beats outside the SOO, the IC pattern of the SOO was automatically matched to the paced morphologies from different locations outside the SOO.
Results
The median number of points per score map was 848 (IQR 460; 1016). The median IC pattern similarity for the same SOO was 86% (IQR 75%; 94%). The KDE curve of the IC pattern similarity distribution showed a non-gaussian form with multiple local maximums corresponding to the overlay of the p-wave with the different segments of the QRST, resulting in fluctuation of the IC pattern similarity score. (Figure 1). There is a moderate positive correlation between the median score and cycle length (r = 0.49, p=0.045), as well as a strong negative correlation between the cycle length and score standard deviation (r = -0.88, p<0.0001). That indicates a lower median IC pattern similarity score and higher variability in the rhythms with shorter cycle lengths. The ROC-curve analysis showed an optimal IC similarity threshold for the rhythm of the same SOO of 63% with a sensitivity of 95% and specificity of 98%. The inflexion point analysis using the peak detection in the KDE first derivative curve revealed an optimal IC pattern similarity cutoff of 69% for the rhythm of the same SOO with a sensitivity of 90.4% and specificity of 99%.
Conclusion
Fluctuations in IC pattern morphology can occur due to an intermittent overlay of the paced unipolar morphology with farfield electrograms of different QRST segments. The IC pattern similarity score is cycle length dependent with higher scores and less dispersion in slower tachycardias. The newly defined IC similarity cutoff allows a better applicability of aICPM in electrophysiological procedures.