- Email: [email protected]
A novel method for the prediction of focal wavefront origins in cardiac arrhythmias
Accepted Manuscript A novel method for the prediction of focal wavefront origins in cardiac arrhythmias Deepak Saluja, John Kassotis, William Kostis, James Coromilas PII:
S0010-4825(18)30239-7
DOI:
10.1016/j.compbiomed.2018.08.019
Reference:
CBM 3054
To appear in:
Computers in Biology and Medicine
Received Date: 31 May 2018 Revised Date:
17 August 2018
Accepted Date: 18 August 2018
Please cite this article as: D. Saluja, J. Kassotis, W. Kostis, J. Coromilas, A novel method for the prediction of focal wavefront origins in cardiac arrhythmias, Computers in Biology and Medicine (2018), doi: 10.1016/j.compbiomed.2018.08.019. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Title: A novel method for the prediction of focal wavefront origins in cardiac arrhythmias
RI PT
Corresponding Author: Deepak Saluja, MD, FHRSa,1 [email protected]
SC
Phone: 732-235-8980
Additional Authors: John Kassotis, MD, Eng Sci Da
TE D
[email protected]
M AN U
Fax: 732-235-8722
William Kostis PhD, MD, FHRSa
EP
[email protected]
James Coromilas, MD, FHRSa
a
AC C
[email protected]
Rutgers-Robert Wood Johnson Medical School
MEB 5th Floor, Division of Cardiology 1 Present Address: Columbia University College of Physicians and Surgeons 622 W 168th Street New York, NY 10032
ACCEPTED MANUSCRIPT
New Brunswick, NJ 08901
Abstract
RI PT
Background:
Current techniques for mapping and ablating cardiac arrhythmias are valuable, but have limitations. We devised a novel method of predicting the origin of a focal arrhythmia
SC
wavefront that utilizes conduction velocity (CV), the difference in electrogram timing during arrhythmia (t), and the distance between two points (z) to generate prediction
M AN U
curves which can be applied to an electroanatomic map. The intersection of two such curves predicts the origin of the wavefront.
Objective: To describe the rationale behind a novel method of arrhythmia mapping and
Methods:
TE D
assess its feasibility in a retrospective study of focal arrhythmias.
EP
We retrospectively studied 12 patients with arrhythmias with focal chamber activation that were successfully mapped and treated with ablation. CV during arrhythmia was
AC C
measured using electroanatomic mapping software. Values for z and t were calculated for two pairs of points. Two prediction curves were generated and superimposed onto the electroanatomic maps. The distance between the intersection of the two curves the wavefront origin was recorded. The shortest distance between individual curves and the wavefront origin was also measured.
ACCEPTED MANUSCRIPT
Results: Twenty-four curves were successfully generated in 12 patients. The distance from the intersection of two curves and the wavefront origin was 9.2 +/- 7.7 mm. The shortest
RI PT
distance between individual prediction curves and the wavefront origin was 5.2 +/- 5.2
SC
mm.
Conclusions:
M AN U
Wavefront origins may be predicted by a novel method utilizing a limited number of measurements. Further study of this method requires its integration with an electroanatomical mapping system.
TE D
Keywords: Arrhythmia; electroanatomic mapping; conduction velocity; arrhythmias; wavefronts; electrophysiology
EP
Funding:
AC C
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
ACCEPTED MANUSCRIPT
RI PT
Introduction
Cardiac arrhythmias are common and cause a substantial burden of symptoms and
SC
morbidity, and catheter ablation has become a commonly utilized treatment. 1
M AN U
Three-dimensional electroanatomic mapping (EAM) was developed to aid in the mapping and ablation of complex arrhythmias. EAM is performed using a computer system that records serial measurements of local cardiac activation relative to a fixed fiducial point, and the location of those measurements in space.2 With a sufficient number of
TE D
measurements, a three-dimensional reproduction of the chamber anatomy can be produced, upon which activation data can be plotted. Information obtained about the magnitude of the voltage recorded at each site may provide information about potential
EP
arrhythmia substrate.
AC C
In this method, the creation of a usable map requires 1) a fiducial electrogram (EGM) with a stable location and morphology 2) a sampled EGM that has a discreet activation for unambiguous annotation 3) persistence of the arrhythmia with a stable cycle length and 4) time to create a map of sufficient resolution.
ACCEPTED MANUSCRIPT
While this approach works for many patients, a substantial number may have multiple and/or transient arrhythmias that do not lend themselves easily to conventional mapping. Other patients may have abnormal substrate with EGMs that may be diffuse and low-
RI PT
voltage (Figure 1), hampering the accurate annotation of local activation time.
We sought to develop a new method of mapping arrhythmias that might overcome some
SC
of these limitations. We postulated that the origin of focal arrhythmia wavefronts could be mathematically extrapolated based on the relative timing of recorded EGMs during
M AN U
arrhythmia, the distance between them, and a measurement of conduction velocity (CV). In this report, we describe the derivation of the extrapolation formula and its accuracy in a retrospective feasibility study.
TE D
Methods Derivation of prediction formula
The derivation of the prediction formula is as follows. Consider two recording bipoles (A
EP
and B) located in a cardiac chamber during an arrhythmia of focal origin (Figure 2). The relative EGM activation recorded at A and B will depend on the location of the wavefront
AC C
origin (O) with respect to the two bipoles. In Figure 2A, A and B measure simultaneous activation. This situation could be seen if the origin if the tachycardia was located directly between, equidistant above, or equidistant below A and B.
In Figure 2B, B is activated before A. In this situation, O may lie between, above, or below A and B, with O oriented closer to B than A. In this case, the time required to travel
ACCEPTED MANUSCRIPT
from O to A minus the time required to travel from the O to B is equal to the difference in activation timing recorded between points A and B.
RI PT
Extrapolating from these examples, we can postulate that for any A and B, in order to
satisfy the observed timing between the EGMs in A and B, the origin O must lie along a defined locus of points. In Figure 1A, that locus inscribes a line perpendicular to and
SC
bisecting AB. In Figure 1B, it inscribes a curve centered around B.
M AN U
The mathematical description of locus O is as follows (Figure 2C):
Let z be the distance AB, t be the difference in electrogram timing observed between A
TE D
and B, and CV be equal to conduction velocity.
Time t is equal to the difference in the time required for an impulse to travel distance OA
EP
minus to the time required to travel distance OB:
AC C
t = OA/CV – OB/CV.
Multiplying all terms by CV yields
CV * t = OA – OB
ACCEPTED MANUSCRIPT
The distances OA and OB can be described in terms of z using the Pythagorean Theorem,
= ( − ) + =
+
We then have +
SC
-
M AN U
∗ = ( − ) +
RI PT
where x and y are unknowns.:
If z and t are measured as distance and time, respectively, and CV is either assumed or measured, the solution to the equation may be plotted where for every x, one or more points y will be calculated that would satisfy the observed time t. Stated differently, plotting the result of the prediction equation inscribes a locus of points along which the
TE D
wavefront origin must lie.
If activation measured at additional bipoles is available, multiple plots may be applied to
AC C
EP
the same wavefront, the intersection of which should indicate its origin.
In order to assess the feasibility of applying this method to the prediction of wavefront origin in human arrhythmias, we retrospectively applied it to 12 patients who had undergone successful mapping and ablation of arrhythmias with focal chamber activation at Robert Wood Johnson University Hospital.
ACCEPTED MANUSCRIPT
Patient selection Patients were included in the study if they had an arrhythmia with focal chamber activation that was mapped and successfully ablated using EAM, no endocardial scarring
RI PT
in the region mapped, and no anatomic boundaries between the wavefront origin and the bipole pair utilized to generate prediction curves. Procedures were performed under
general anesthesia or moderate sedation by a single operator (DS) using the CARTO 3
SC
system (Biosense Webster, Diamond Bar, CA). A multipolar mapping catheter
(Pentarray, Biosense Webster) and an open-irrigated ablation catheter (SmartTouch or
M AN U
SmartTouch RMT, Biosense Webster) was used for mapping and ablation. Arrhythmias were determined to be focal using the results of EAM, entrainment maneuvers 3, and the outcome of focal ablation, which was acutely successful in all patients. Bipolar EGMs were recorded using a multi-channel electrophysiologic recorder (Boston Scientific LS
TE D
Pro, Marlborough, MA) with filtering set to a bandwidth of 25 –300 Hz. Ablation success was defined as acute arrhythmia termination during ablation followed by arrhythmia noninducibility. The study was approved by the Rutgers Health Science New Brunswick
EP
Institutional Review Board.
AC C
Measurement of t, CV, and z
Electroanatomic maps and EGM recordings from included patients were retrieved. Two prediction curves were constructed in each patient using two pairs of points. Pairs of points from the chamber of activation origin were randomly chosen from among those available that minimized the effects of chamber curvature. Measurement of t for each pair of points was made offline at 200mm/sec sweep speed using digital calipers. CV was
ACCEPTED MANUSCRIPT
calculated using EAM software by dividing the difference in activation time between adjacent points during arrhythmia by the measured distance between them (Figure 3). Points were chosen such that they were parallel to the local vector of activation to within
RI PT
15 degrees. Up to 5 measurements were averaged to determine the CV. The distance
between the bipoles used to construct prediction curves was recorded as z. The locations of these bipoles with respect to the origin of arrhythmia and the bipoles used to calculate
SC
CV were chosen to minimize the effects of chamber curvature. All measurements were
Construction of prediction curves
M AN U
made with digital calipers.
Measurements were applied to the prediction equation, prediction curves were created with on-line graphical software (www.desmos.com), and the results superimposed on the
Outcome measures
TE D
electroanatomic map (Figure 4).
EP
For each arrhythmia, the origin was defined as the point at which ablation terminated the
AC C
arrhythmia. For each pair of bipoles used to generate a prediction curve, the difference in activation time (t) between the bipoles, the distance between bipoles, and the distance from each bipole to the wavefront origin was recorded. The primary outcome was the distance between the intersection of the two prediction curves and the wavefront origin. The secondary outcome was the shortest measured distance between a single prediction line and wavefront origin.
ACCEPTED MANUSCRIPT
Statistical Analysis Continuous data are reported as mean +/- SD. Linear regression was used to assess
RI PT
correlation between outcomes and measured variables.
Results
Twelve tachycardias were studied in 12 patients. Patient and rhythm characteristics are
SC
given in Table 1. Wavefront activation initiated in the atrium in 5 patients and in the ventricles in 7 patients. Three patients had orthodromic reciprocating tachycardias
M AN U
utilizing accessory pathways. Although these arrhythmias are reentrant, these patients were included because activation from the perspective of the chamber mapped proceeds from a focal origin located at the site of the accessory pathway. Nine of the patients were
TE D
women. The mean age was 52.9 +/- 15.4 years.
A total of 47 measurements of CV were made (a mean of 2.9 +/- 1.7 per patient). The average CV was 0.76 +/- 0. 39 m/s. The mean CV was 0.96 +/- 0.34 m/s (n=9) for ATs,
EP
0.48 +/- 0.19 m/s (n=8) for ORTs, and 0.78 +/- .41 m/s (n=30) for PVCs.
AC C
The characteristics of the points used to construct the prediction curves are described in Table 2. The mean distance from a bipole to the arrhythmia origin was 16.9 +/- 11.0 mm. The mean distance between bipoles was 16.1 +/- 11.0 mm. The mean difference in activation time between the two bipoles was 5.2+/- 5.2 ms.
ACCEPTED MANUSCRIPT
A total of 48 points were used to construct 24 prediction curves in 12 patients. The average distance between the intersection of the two curves and origin of tachycardia was 9.2 +/- 7.7 mm. The corresponding measure was 10.4 +/- 4.7 mm (n=2) among ATs, 3.9
RI PT
+/- 5.4 mm (n=3) among ORTs, and 11.1 +/- 8.8 mm (n=7) among PVCs.
The mean minimum distance from the prediction curve to the wavefront origin was 5.2
SC
+/- 5.2 mm. The corresponding measure was 7.1 +/- 4.2 mm (n=4) among ATs, 2.6 +/3.2 mm (n=6) among ORTs, and 5.8 +/- 5.9 mm (n=14) among PVCs. There was no
M AN U
significant correlation between this outcome and the mean distance from the bipoles to the arrhythmia origin, the distance between the bipole pairs, or the difference in
TE D
activation times between bipole pairs.
Discussion
EP
We propose a novel method of mapping arrhythmia wavefront origins that uses a limited number of easily obtained measurements and was successfully implemented in an initial
AC C
feasibility study.
This method overcomes some of the limitations of conventional mapping. Stability of neither the fiducial point morphology nor the arrhythmia cycle length is required. Prediction curves may be generated in a limited period of time, allowing for the mapping of short-lived or unstable rhythms. While conventional mapping may require annotation
ACCEPTED MANUSCRIPT
of fractionated EGMs (Figure 1) that may be present in scarred substrate, measurements of z and t may be made in tissues remote from site of arrhythmia origin, where discreet
RI PT
EGMs may be found.
Although this method has advantages, it has limitations as well. Our method presumes
that a wavefront proceeds in a circular, centrifugal manner from the point of origin to the
SC
recording bipoles, and that CV is constant in all (relevant) parts of the chamber mapped.
M AN U
In reality, wavefront direction can be altered by anatomic boundaries or regional CV variability which can result in a “bending” of the wavefront. Further, heterogeneity of CV, both in tissue type (ventricular vs atrial myocardium) as well as in direction of conduction relative to myofibril orientation (longitudinal vs transverse), has been
TE D
established in normal hearts.4 One would therefore expect this method to produce inaccurate results if applied over areas of significant anisotropy. However, the accuracy of this method does not depend on global homogeneity of CV in all cardiac tissues.
EP
Rather, accuracy will be optimal when the CV seen in the local environment is consistent
AC C
and approximates the CV used in constructing the prediction curves.
Since the opportunity for the advancing wavefront to encounter anisotropy should increase with increasing 1) endocardial scarring, 2) distance to the sampling bipoles, and 3) distance between bipole pairs, this method is expected to be most accurate when the individual bipole pairs and wavefront origin are in closer proximity, and when the intervening myocardium contains limited scar. Although we did not find a correlation
ACCEPTED MANUSCRIPT
between accuracy and the distance between the sampled bipoles in this study, these data are underpowered to detect such differences. It is possible that a larger sample would
RI PT
have yielded different results.
Although the formula used in this method contains CV as a single term, utilizing separate terms (i.e. CV1 and CV2 ) for conduction between the arrhythmia origin and each bipole
SC
would allow for anisotropy in the model and might improve its accuracy.
M AN U
The value of t has implications for the accuracy of the prediction curves with respect to other measurements. As t approaches 0 (indicating simultaneous activation between two sampled bipoles), the term t*CV approaches 0, and the prediction curve will be a straight line independent of CV. Conversely, when the value of t is large, the prediction curves
TE D
will have the greatest curvature. Under these conditions, as the distance between the sampled bipoles and the wavefront origin increases, accuracy will become increasingly
EP
sensitive to values of t and CV, as well as errors in their measurement.
In our retrospective study, CV was measured using completed EAMs. This may not be
AC C
feasible clinical practice, as this method would most likely be applied prior to the completion of such a map. Clinically, an alternative method of measuring CV may therefore be required.
It is possible to measure CV in a background of sinus rhythm by measuring the conduction time between two bipoles, provided conduction is unimpeded and the
ACCEPTED MANUSCRIPT
wavefront is parallel to the two bipoles. Practically, these conditions are easiest to achieve when pacing from one bipole and recording activation time on the other. By knowing the distance between the two bipoles, either by utilizing catheters with fixed
RI PT
interspacing or by measuring the distance with an EAM system, CV may be calculated.
As CV is known to decrease with decreasing cycle-length, pacing is ideally performed at
SC
a cycle-length relevant to the arrhythmia in question.5
CV may be measured in the presence of arrhythmia as well. For example, during a focal
M AN U
tachycardia, pacing from a bipole at a rate faster than the tachycardia cycle length will temporarily result in overdrive pacing. Recording the time to activation at a second site and the distance between the sites allows CV to be calculated.
TE D
The feasibility analysis was done by constructing prediction curves in two dimensions and projecting them onto a three-dimensional surface. Although EAM projections and sampled points were chosen to minimize chamber curvature, there is an inherent
EP
limitation in using planar projections in a three-dimensional structure which may result in inaccurate results.6 Therefore, any conclusions about the accuracy of this method based
AC C
on these data should be approached cautiously.
In this report, we considered only rhythms with focal chamber activation. Reentrant rhythms that appear to be coming from a focal source, such as in the case of microreentry or macroreetry in a protected area with a discreet exit point, could be treated in a similar manner. Macroreentrant mechanisms with continuous activation (such typical
ACCEPTED MANUSCRIPT
atrial flutter) cannot be described as having a discreet wavefront origin. From the perspective of a bipole pair, a macroreentrant wavefront will appear as if originating tangentially to the circuit. Multiple applications of this algorithm could outline the course
RI PT
of a macroreentrant circuit, but would not be expected to reveal an ablation target.
Practical usage and proper testing of this method requires its integration with EAM
SC
systems, which have the capability of making geodesic measurements and generate
prediction lines in three dimensions. Future work should involve studying the accuracy
AC C
EP
TE D
M AN U
and behavior of this method in a three-dimensional system in a larger number of patients.
ACCEPTED MANUSCRIPT
Figure 1. Conventional mapping of a repetitive premature ventricular contraction originating from the right ventricular outflow tract (right lateral projection) A) Accurate
RI PT
mapping requires a stable fiducial point (the maximal positive deflection in aVF) and a
discreet potential to measure relative activation (measured in Map 1-2). Marking relative activation over successive beats and integrating this information with three-dimensional
SC
positional data generates an electroanatomic map activation map (middle panel). B)
Fractionated activation (*), as may be seen in scarred substrate, may make it difficult to
M AN U
annotate local activation discreetly, diminishing the accuracy of the map. The location of successful ablation is marked (blue dot). In this and subsequent figures, local activation time (LAT) is color coded, with the earliest activation in red and the latest in violet. The orientation of the map in three-dimensional space is noted using an idealized heart model
AC C
EP
TE D
(panel B, bottom right).
ACCEPTED MANUSCRIPT
Figure 2. (2A) Simultaneous activation of A and B could be the result a wavefront origin located anywhere on a line perpendicular to A and B. (2B) Relative activation of A and B could be explained by a wavefront origin located anywhere on a curve centered around
AC C
EP
TE D
M AN U
SC
and B (z) and the time elapsed between EGMs on A and B.
RI PT
point B. (2C) The equation describing the relationship between the distance between A
ACCEPTED MANUSCRIPT
Figure 3: Measurement of CV in a focal left atrial tachycardia (AT, modified left lateral view). AT is originating from the lateral aspect of the left atrium. Location of successful
RI PT
ablation is noted (blue dot). Bipoles (A and B) were chosen to be as parallel as possible to the direction of the local propagation vector (arrow), as determined by visual analysis of the electroanatomic activation mapping sequence. The distance between bipoles A and B
SC
divided by the difference in activation time between the two points was recorded as the
AC C
EP
TE D
M AN U
conduction velocity.
ACCEPTED MANUSCRIPT
Figure 4. Mapping of a repetitive premature ventricular contraction arising from the
RI PT
inflow of the right ventricle (modified left lateral view). A) A prediction curve is
generated utilizing the calculated CV and measured t and z between two selected bipoles (blue and green dots) and superimposed onto the electroanatomic map. B) Applying this
SC
method for a second set of points generates another prediction line. C) The intersection of
AC C
EP
TE D
ablation is marked with a red X.
M AN U
the two lines (open circle) predicts the origin of the wavefront. The location of successful
ACCEPTED MANUSCRIPT
1
Cappato R, Calkins H, Chen SA, Davies W, Iesaka Y, Kalman J, Kim YH, Klein
RI PT
G, Natale A, Packer D, Skanes A, Ambrogi F, Biganzoli E. Updated worldwide survey on the methods, efficacy, and safety of catheter ablation for human atrial fibrillation. Circ Arrhythm Electrophysiol. 2010 Feb;3(1):32-8. Epub 2009 Dec 7. Varanasi S, Dhala A, Blanck Z, Deshpande S, Akhtar M, Sra J.
SC
2
Electroanatomic mapping for radiofrequency ablation of cardiac arrhythmias. J
3
M AN U
Cardiovasc Electrophysiol. 1999 Apr;10(4):538-44.
Waldo AL, Maclean WAH, Karp RB, Kouchoukos NT, James TN. Entrainment and
interruption of atrial flutter with atrial pacing: studies in man following open heart
4
TE D
surgery. Circulation. 1977; 56:737-745.
Valderrábano M. Influence of anisotropic conduction properties in the propagation
5
AC C
2007 Mar 24.
EP
of the cardiac action potential. Prog Biophys Mol Biol. 2007; 94(1-2): 144–168. EPub
Grossi S, Grassi F, Galleani L, Bianchi F, Masi AS, Conte MR. Atrial Conduction
Velocity Correlates with Frequency Content of Bipolar Signal. Pacing Clin Electrophysiol. 2016 Aug;39(8):814-21.
6
Pathik B, Kalman JM, Walters T, Kuklik P, Zhao J, Madry A, Sanders P, Kistler PM,
Lee G. Absence of rotational activity detected using 2-dimensional phase mapping in
ACCEPTED MANUSCRIPT
the corresponding 3-dimensional phase maps in human persistent atrial fibrillation.
AC C
EP
TE D
M AN U
SC
RI PT
Heart Rhythm. February 2018 Volume 15, Issue 2, Pages 182–192
ACCEPTED MANUSCRIPT
Gender F M F F F F F M M F F F
Chamber RV LA RV LA RV LA RA RV LA LV RV RV
Location Anterior Lateral Septal Lateral Septal Lateral Lateral Septal Lateral Septal Septal Septal
Arrhythmia PVC AP Tachycardia PVC AP Tachycardia PVC AT AT PVC AP Tachycardia PVC PVC PVC
RI PT
Age 26 41 59 57 49 59 58 66 43 50 54 73
SC
Patient # 1 2 3 4 5 6 7 8 9 10 11 12
AC C
EP
TE D
M AN U
Table 1. Clinical characteristics of the study population and arrhythmias.
ACCEPTED MANUSCRIPT
Distance between bipoles (mm) 16.1 11.0
Difference in activation time between bipoles (ms) 9.8 10.7
Minimum distance of prediction line to origin (mm) 5.2 5.2
Distance of intersection to origin (mm) 9.2 7.7
RI PT
Distance of bipoles from origin (mm) Mean 16.9 SD 11.0
AC C
EP
TE D
M AN U
SC
Table 2. Characteristics of the sampled bipoles and accuracy of the prediction curves.
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Conflict of interest statement:
AC C
EP
TE D
M AN U
SC
RI PT
None of the authors report any relevant conflict of interest related to the work submitted.
S0010-4825(18)30239-7
DOI:
10.1016/j.compbiomed.2018.08.019
Reference:
CBM 3054
To appear in:
Computers in Biology and Medicine
Received Date: 31 May 2018 Revised Date:
17 August 2018
Accepted Date: 18 August 2018
Please cite this article as: D. Saluja, J. Kassotis, W. Kostis, J. Coromilas, A novel method for the prediction of focal wavefront origins in cardiac arrhythmias, Computers in Biology and Medicine (2018), doi: 10.1016/j.compbiomed.2018.08.019. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Title: A novel method for the prediction of focal wavefront origins in cardiac arrhythmias
RI PT
Corresponding Author: Deepak Saluja, MD, FHRSa,1 [email protected]
SC
Phone: 732-235-8980
Additional Authors: John Kassotis, MD, Eng Sci Da
TE D
[email protected]
M AN U
Fax: 732-235-8722
William Kostis PhD, MD, FHRSa
EP
[email protected]
James Coromilas, MD, FHRSa
a
AC C
[email protected]
Rutgers-Robert Wood Johnson Medical School
MEB 5th Floor, Division of Cardiology 1 Present Address: Columbia University College of Physicians and Surgeons 622 W 168th Street New York, NY 10032
ACCEPTED MANUSCRIPT
New Brunswick, NJ 08901
Abstract
RI PT
Background:
Current techniques for mapping and ablating cardiac arrhythmias are valuable, but have limitations. We devised a novel method of predicting the origin of a focal arrhythmia
SC
wavefront that utilizes conduction velocity (CV), the difference in electrogram timing during arrhythmia (t), and the distance between two points (z) to generate prediction
M AN U
curves which can be applied to an electroanatomic map. The intersection of two such curves predicts the origin of the wavefront.
Objective: To describe the rationale behind a novel method of arrhythmia mapping and
Methods:
TE D
assess its feasibility in a retrospective study of focal arrhythmias.
EP
We retrospectively studied 12 patients with arrhythmias with focal chamber activation that were successfully mapped and treated with ablation. CV during arrhythmia was
AC C
measured using electroanatomic mapping software. Values for z and t were calculated for two pairs of points. Two prediction curves were generated and superimposed onto the electroanatomic maps. The distance between the intersection of the two curves the wavefront origin was recorded. The shortest distance between individual curves and the wavefront origin was also measured.
ACCEPTED MANUSCRIPT
Results: Twenty-four curves were successfully generated in 12 patients. The distance from the intersection of two curves and the wavefront origin was 9.2 +/- 7.7 mm. The shortest
RI PT
distance between individual prediction curves and the wavefront origin was 5.2 +/- 5.2
SC
mm.
Conclusions:
M AN U
Wavefront origins may be predicted by a novel method utilizing a limited number of measurements. Further study of this method requires its integration with an electroanatomical mapping system.
TE D
Keywords: Arrhythmia; electroanatomic mapping; conduction velocity; arrhythmias; wavefronts; electrophysiology
EP
Funding:
AC C
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
ACCEPTED MANUSCRIPT
RI PT
Introduction
Cardiac arrhythmias are common and cause a substantial burden of symptoms and
SC
morbidity, and catheter ablation has become a commonly utilized treatment. 1
M AN U
Three-dimensional electroanatomic mapping (EAM) was developed to aid in the mapping and ablation of complex arrhythmias. EAM is performed using a computer system that records serial measurements of local cardiac activation relative to a fixed fiducial point, and the location of those measurements in space.2 With a sufficient number of
TE D
measurements, a three-dimensional reproduction of the chamber anatomy can be produced, upon which activation data can be plotted. Information obtained about the magnitude of the voltage recorded at each site may provide information about potential
EP
arrhythmia substrate.
AC C
In this method, the creation of a usable map requires 1) a fiducial electrogram (EGM) with a stable location and morphology 2) a sampled EGM that has a discreet activation for unambiguous annotation 3) persistence of the arrhythmia with a stable cycle length and 4) time to create a map of sufficient resolution.
ACCEPTED MANUSCRIPT
While this approach works for many patients, a substantial number may have multiple and/or transient arrhythmias that do not lend themselves easily to conventional mapping. Other patients may have abnormal substrate with EGMs that may be diffuse and low-
RI PT
voltage (Figure 1), hampering the accurate annotation of local activation time.
We sought to develop a new method of mapping arrhythmias that might overcome some
SC
of these limitations. We postulated that the origin of focal arrhythmia wavefronts could be mathematically extrapolated based on the relative timing of recorded EGMs during
M AN U
arrhythmia, the distance between them, and a measurement of conduction velocity (CV). In this report, we describe the derivation of the extrapolation formula and its accuracy in a retrospective feasibility study.
TE D
Methods Derivation of prediction formula
The derivation of the prediction formula is as follows. Consider two recording bipoles (A
EP
and B) located in a cardiac chamber during an arrhythmia of focal origin (Figure 2). The relative EGM activation recorded at A and B will depend on the location of the wavefront
AC C
origin (O) with respect to the two bipoles. In Figure 2A, A and B measure simultaneous activation. This situation could be seen if the origin if the tachycardia was located directly between, equidistant above, or equidistant below A and B.
In Figure 2B, B is activated before A. In this situation, O may lie between, above, or below A and B, with O oriented closer to B than A. In this case, the time required to travel
ACCEPTED MANUSCRIPT
from O to A minus the time required to travel from the O to B is equal to the difference in activation timing recorded between points A and B.
RI PT
Extrapolating from these examples, we can postulate that for any A and B, in order to
satisfy the observed timing between the EGMs in A and B, the origin O must lie along a defined locus of points. In Figure 1A, that locus inscribes a line perpendicular to and
SC
bisecting AB. In Figure 1B, it inscribes a curve centered around B.
M AN U
The mathematical description of locus O is as follows (Figure 2C):
Let z be the distance AB, t be the difference in electrogram timing observed between A
TE D
and B, and CV be equal to conduction velocity.
Time t is equal to the difference in the time required for an impulse to travel distance OA
EP
minus to the time required to travel distance OB:
AC C
t = OA/CV – OB/CV.
Multiplying all terms by CV yields
CV * t = OA – OB
ACCEPTED MANUSCRIPT
The distances OA and OB can be described in terms of z using the Pythagorean Theorem,
= ( − ) + =
+
We then have +
SC
-
M AN U
∗ = ( − ) +
RI PT
where x and y are unknowns.:
If z and t are measured as distance and time, respectively, and CV is either assumed or measured, the solution to the equation may be plotted where for every x, one or more points y will be calculated that would satisfy the observed time t. Stated differently, plotting the result of the prediction equation inscribes a locus of points along which the
TE D
wavefront origin must lie.
If activation measured at additional bipoles is available, multiple plots may be applied to
AC C
EP
the same wavefront, the intersection of which should indicate its origin.
In order to assess the feasibility of applying this method to the prediction of wavefront origin in human arrhythmias, we retrospectively applied it to 12 patients who had undergone successful mapping and ablation of arrhythmias with focal chamber activation at Robert Wood Johnson University Hospital.
ACCEPTED MANUSCRIPT
Patient selection Patients were included in the study if they had an arrhythmia with focal chamber activation that was mapped and successfully ablated using EAM, no endocardial scarring
RI PT
in the region mapped, and no anatomic boundaries between the wavefront origin and the bipole pair utilized to generate prediction curves. Procedures were performed under
general anesthesia or moderate sedation by a single operator (DS) using the CARTO 3
SC
system (Biosense Webster, Diamond Bar, CA). A multipolar mapping catheter
(Pentarray, Biosense Webster) and an open-irrigated ablation catheter (SmartTouch or
M AN U
SmartTouch RMT, Biosense Webster) was used for mapping and ablation. Arrhythmias were determined to be focal using the results of EAM, entrainment maneuvers 3, and the outcome of focal ablation, which was acutely successful in all patients. Bipolar EGMs were recorded using a multi-channel electrophysiologic recorder (Boston Scientific LS
TE D
Pro, Marlborough, MA) with filtering set to a bandwidth of 25 –300 Hz. Ablation success was defined as acute arrhythmia termination during ablation followed by arrhythmia noninducibility. The study was approved by the Rutgers Health Science New Brunswick
EP
Institutional Review Board.
AC C
Measurement of t, CV, and z
Electroanatomic maps and EGM recordings from included patients were retrieved. Two prediction curves were constructed in each patient using two pairs of points. Pairs of points from the chamber of activation origin were randomly chosen from among those available that minimized the effects of chamber curvature. Measurement of t for each pair of points was made offline at 200mm/sec sweep speed using digital calipers. CV was
ACCEPTED MANUSCRIPT
calculated using EAM software by dividing the difference in activation time between adjacent points during arrhythmia by the measured distance between them (Figure 3). Points were chosen such that they were parallel to the local vector of activation to within
RI PT
15 degrees. Up to 5 measurements were averaged to determine the CV. The distance
between the bipoles used to construct prediction curves was recorded as z. The locations of these bipoles with respect to the origin of arrhythmia and the bipoles used to calculate
SC
CV were chosen to minimize the effects of chamber curvature. All measurements were
Construction of prediction curves
M AN U
made with digital calipers.
Measurements were applied to the prediction equation, prediction curves were created with on-line graphical software (www.desmos.com), and the results superimposed on the
Outcome measures
TE D
electroanatomic map (Figure 4).
EP
For each arrhythmia, the origin was defined as the point at which ablation terminated the
AC C
arrhythmia. For each pair of bipoles used to generate a prediction curve, the difference in activation time (t) between the bipoles, the distance between bipoles, and the distance from each bipole to the wavefront origin was recorded. The primary outcome was the distance between the intersection of the two prediction curves and the wavefront origin. The secondary outcome was the shortest measured distance between a single prediction line and wavefront origin.
ACCEPTED MANUSCRIPT
Statistical Analysis Continuous data are reported as mean +/- SD. Linear regression was used to assess
RI PT
correlation between outcomes and measured variables.
Results
Twelve tachycardias were studied in 12 patients. Patient and rhythm characteristics are
SC
given in Table 1. Wavefront activation initiated in the atrium in 5 patients and in the ventricles in 7 patients. Three patients had orthodromic reciprocating tachycardias
M AN U
utilizing accessory pathways. Although these arrhythmias are reentrant, these patients were included because activation from the perspective of the chamber mapped proceeds from a focal origin located at the site of the accessory pathway. Nine of the patients were
TE D
women. The mean age was 52.9 +/- 15.4 years.
A total of 47 measurements of CV were made (a mean of 2.9 +/- 1.7 per patient). The average CV was 0.76 +/- 0. 39 m/s. The mean CV was 0.96 +/- 0.34 m/s (n=9) for ATs,
EP
0.48 +/- 0.19 m/s (n=8) for ORTs, and 0.78 +/- .41 m/s (n=30) for PVCs.
AC C
The characteristics of the points used to construct the prediction curves are described in Table 2. The mean distance from a bipole to the arrhythmia origin was 16.9 +/- 11.0 mm. The mean distance between bipoles was 16.1 +/- 11.0 mm. The mean difference in activation time between the two bipoles was 5.2+/- 5.2 ms.
ACCEPTED MANUSCRIPT
A total of 48 points were used to construct 24 prediction curves in 12 patients. The average distance between the intersection of the two curves and origin of tachycardia was 9.2 +/- 7.7 mm. The corresponding measure was 10.4 +/- 4.7 mm (n=2) among ATs, 3.9
RI PT
+/- 5.4 mm (n=3) among ORTs, and 11.1 +/- 8.8 mm (n=7) among PVCs.
The mean minimum distance from the prediction curve to the wavefront origin was 5.2
SC
+/- 5.2 mm. The corresponding measure was 7.1 +/- 4.2 mm (n=4) among ATs, 2.6 +/3.2 mm (n=6) among ORTs, and 5.8 +/- 5.9 mm (n=14) among PVCs. There was no
M AN U
significant correlation between this outcome and the mean distance from the bipoles to the arrhythmia origin, the distance between the bipole pairs, or the difference in
TE D
activation times between bipole pairs.
Discussion
EP
We propose a novel method of mapping arrhythmia wavefront origins that uses a limited number of easily obtained measurements and was successfully implemented in an initial
AC C
feasibility study.
This method overcomes some of the limitations of conventional mapping. Stability of neither the fiducial point morphology nor the arrhythmia cycle length is required. Prediction curves may be generated in a limited period of time, allowing for the mapping of short-lived or unstable rhythms. While conventional mapping may require annotation
ACCEPTED MANUSCRIPT
of fractionated EGMs (Figure 1) that may be present in scarred substrate, measurements of z and t may be made in tissues remote from site of arrhythmia origin, where discreet
RI PT
EGMs may be found.
Although this method has advantages, it has limitations as well. Our method presumes
that a wavefront proceeds in a circular, centrifugal manner from the point of origin to the
SC
recording bipoles, and that CV is constant in all (relevant) parts of the chamber mapped.
M AN U
In reality, wavefront direction can be altered by anatomic boundaries or regional CV variability which can result in a “bending” of the wavefront. Further, heterogeneity of CV, both in tissue type (ventricular vs atrial myocardium) as well as in direction of conduction relative to myofibril orientation (longitudinal vs transverse), has been
TE D
established in normal hearts.4 One would therefore expect this method to produce inaccurate results if applied over areas of significant anisotropy. However, the accuracy of this method does not depend on global homogeneity of CV in all cardiac tissues.
EP
Rather, accuracy will be optimal when the CV seen in the local environment is consistent
AC C
and approximates the CV used in constructing the prediction curves.
Since the opportunity for the advancing wavefront to encounter anisotropy should increase with increasing 1) endocardial scarring, 2) distance to the sampling bipoles, and 3) distance between bipole pairs, this method is expected to be most accurate when the individual bipole pairs and wavefront origin are in closer proximity, and when the intervening myocardium contains limited scar. Although we did not find a correlation
ACCEPTED MANUSCRIPT
between accuracy and the distance between the sampled bipoles in this study, these data are underpowered to detect such differences. It is possible that a larger sample would
RI PT
have yielded different results.
Although the formula used in this method contains CV as a single term, utilizing separate terms (i.e. CV1 and CV2 ) for conduction between the arrhythmia origin and each bipole
SC
would allow for anisotropy in the model and might improve its accuracy.
M AN U
The value of t has implications for the accuracy of the prediction curves with respect to other measurements. As t approaches 0 (indicating simultaneous activation between two sampled bipoles), the term t*CV approaches 0, and the prediction curve will be a straight line independent of CV. Conversely, when the value of t is large, the prediction curves
TE D
will have the greatest curvature. Under these conditions, as the distance between the sampled bipoles and the wavefront origin increases, accuracy will become increasingly
EP
sensitive to values of t and CV, as well as errors in their measurement.
In our retrospective study, CV was measured using completed EAMs. This may not be
AC C
feasible clinical practice, as this method would most likely be applied prior to the completion of such a map. Clinically, an alternative method of measuring CV may therefore be required.
It is possible to measure CV in a background of sinus rhythm by measuring the conduction time between two bipoles, provided conduction is unimpeded and the
ACCEPTED MANUSCRIPT
wavefront is parallel to the two bipoles. Practically, these conditions are easiest to achieve when pacing from one bipole and recording activation time on the other. By knowing the distance between the two bipoles, either by utilizing catheters with fixed
RI PT
interspacing or by measuring the distance with an EAM system, CV may be calculated.
As CV is known to decrease with decreasing cycle-length, pacing is ideally performed at
SC
a cycle-length relevant to the arrhythmia in question.5
CV may be measured in the presence of arrhythmia as well. For example, during a focal
M AN U
tachycardia, pacing from a bipole at a rate faster than the tachycardia cycle length will temporarily result in overdrive pacing. Recording the time to activation at a second site and the distance between the sites allows CV to be calculated.
TE D
The feasibility analysis was done by constructing prediction curves in two dimensions and projecting them onto a three-dimensional surface. Although EAM projections and sampled points were chosen to minimize chamber curvature, there is an inherent
EP
limitation in using planar projections in a three-dimensional structure which may result in inaccurate results.6 Therefore, any conclusions about the accuracy of this method based
AC C
on these data should be approached cautiously.
In this report, we considered only rhythms with focal chamber activation. Reentrant rhythms that appear to be coming from a focal source, such as in the case of microreentry or macroreetry in a protected area with a discreet exit point, could be treated in a similar manner. Macroreentrant mechanisms with continuous activation (such typical
ACCEPTED MANUSCRIPT
atrial flutter) cannot be described as having a discreet wavefront origin. From the perspective of a bipole pair, a macroreentrant wavefront will appear as if originating tangentially to the circuit. Multiple applications of this algorithm could outline the course
RI PT
of a macroreentrant circuit, but would not be expected to reveal an ablation target.
Practical usage and proper testing of this method requires its integration with EAM
SC
systems, which have the capability of making geodesic measurements and generate
prediction lines in three dimensions. Future work should involve studying the accuracy
AC C
EP
TE D
M AN U
and behavior of this method in a three-dimensional system in a larger number of patients.
ACCEPTED MANUSCRIPT
Figure 1. Conventional mapping of a repetitive premature ventricular contraction originating from the right ventricular outflow tract (right lateral projection) A) Accurate
RI PT
mapping requires a stable fiducial point (the maximal positive deflection in aVF) and a
discreet potential to measure relative activation (measured in Map 1-2). Marking relative activation over successive beats and integrating this information with three-dimensional
SC
positional data generates an electroanatomic map activation map (middle panel). B)
Fractionated activation (*), as may be seen in scarred substrate, may make it difficult to
M AN U
annotate local activation discreetly, diminishing the accuracy of the map. The location of successful ablation is marked (blue dot). In this and subsequent figures, local activation time (LAT) is color coded, with the earliest activation in red and the latest in violet. The orientation of the map in three-dimensional space is noted using an idealized heart model
AC C
EP
TE D
(panel B, bottom right).
ACCEPTED MANUSCRIPT
Figure 2. (2A) Simultaneous activation of A and B could be the result a wavefront origin located anywhere on a line perpendicular to A and B. (2B) Relative activation of A and B could be explained by a wavefront origin located anywhere on a curve centered around
AC C
EP
TE D
M AN U
SC
and B (z) and the time elapsed between EGMs on A and B.
RI PT
point B. (2C) The equation describing the relationship between the distance between A
ACCEPTED MANUSCRIPT
Figure 3: Measurement of CV in a focal left atrial tachycardia (AT, modified left lateral view). AT is originating from the lateral aspect of the left atrium. Location of successful
RI PT
ablation is noted (blue dot). Bipoles (A and B) were chosen to be as parallel as possible to the direction of the local propagation vector (arrow), as determined by visual analysis of the electroanatomic activation mapping sequence. The distance between bipoles A and B
SC
divided by the difference in activation time between the two points was recorded as the
AC C
EP
TE D
M AN U
conduction velocity.
ACCEPTED MANUSCRIPT
Figure 4. Mapping of a repetitive premature ventricular contraction arising from the
RI PT
inflow of the right ventricle (modified left lateral view). A) A prediction curve is
generated utilizing the calculated CV and measured t and z between two selected bipoles (blue and green dots) and superimposed onto the electroanatomic map. B) Applying this
SC
method for a second set of points generates another prediction line. C) The intersection of
AC C
EP
TE D
ablation is marked with a red X.
M AN U
the two lines (open circle) predicts the origin of the wavefront. The location of successful
ACCEPTED MANUSCRIPT
1
Cappato R, Calkins H, Chen SA, Davies W, Iesaka Y, Kalman J, Kim YH, Klein
RI PT
G, Natale A, Packer D, Skanes A, Ambrogi F, Biganzoli E. Updated worldwide survey on the methods, efficacy, and safety of catheter ablation for human atrial fibrillation. Circ Arrhythm Electrophysiol. 2010 Feb;3(1):32-8. Epub 2009 Dec 7. Varanasi S, Dhala A, Blanck Z, Deshpande S, Akhtar M, Sra J.
SC
2
Electroanatomic mapping for radiofrequency ablation of cardiac arrhythmias. J
3
M AN U
Cardiovasc Electrophysiol. 1999 Apr;10(4):538-44.
Waldo AL, Maclean WAH, Karp RB, Kouchoukos NT, James TN. Entrainment and
interruption of atrial flutter with atrial pacing: studies in man following open heart
4
TE D
surgery. Circulation. 1977; 56:737-745.
Valderrábano M. Influence of anisotropic conduction properties in the propagation
5
AC C
2007 Mar 24.
EP
of the cardiac action potential. Prog Biophys Mol Biol. 2007; 94(1-2): 144–168. EPub
Grossi S, Grassi F, Galleani L, Bianchi F, Masi AS, Conte MR. Atrial Conduction
Velocity Correlates with Frequency Content of Bipolar Signal. Pacing Clin Electrophysiol. 2016 Aug;39(8):814-21.
6
Pathik B, Kalman JM, Walters T, Kuklik P, Zhao J, Madry A, Sanders P, Kistler PM,
Lee G. Absence of rotational activity detected using 2-dimensional phase mapping in
ACCEPTED MANUSCRIPT
the corresponding 3-dimensional phase maps in human persistent atrial fibrillation.
AC C
EP
TE D
M AN U
SC
RI PT
Heart Rhythm. February 2018 Volume 15, Issue 2, Pages 182–192
ACCEPTED MANUSCRIPT
Gender F M F F F F F M M F F F
Chamber RV LA RV LA RV LA RA RV LA LV RV RV
Location Anterior Lateral Septal Lateral Septal Lateral Lateral Septal Lateral Septal Septal Septal
Arrhythmia PVC AP Tachycardia PVC AP Tachycardia PVC AT AT PVC AP Tachycardia PVC PVC PVC
RI PT
Age 26 41 59 57 49 59 58 66 43 50 54 73
SC
Patient # 1 2 3 4 5 6 7 8 9 10 11 12
AC C
EP
TE D
M AN U
Table 1. Clinical characteristics of the study population and arrhythmias.
ACCEPTED MANUSCRIPT
Distance between bipoles (mm) 16.1 11.0
Difference in activation time between bipoles (ms) 9.8 10.7
Minimum distance of prediction line to origin (mm) 5.2 5.2
Distance of intersection to origin (mm) 9.2 7.7
RI PT
Distance of bipoles from origin (mm) Mean 16.9 SD 11.0
AC C
EP
TE D
M AN U
SC
Table 2. Characteristics of the sampled bipoles and accuracy of the prediction curves.
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
ACCEPTED MANUSCRIPT
Conflict of interest statement:
AC C
EP
TE D
M AN U
SC
RI PT
None of the authors report any relevant conflict of interest related to the work submitted.