Cardiovascular Research 1992;26:237-243
237
Local activation variability during monomorphic ventricular tachycardia in the dog David S Rosenbaum, David J Wilber, Joseph M Smith, Duke Du, Jeremy N Ruskin, and Hasan Garan
T
he re-entrant pathways of monomorphic ventricular tachycardia in chronic myocardial infarction are thought to be spatially fixed relative to either an anatomical or functional zone of conduction block.'-' If so the timing of local electrical activation at a given myocardial site should also be constant from one ventricular tachycardia beat to the next. However, recent experimental studies have shown that spontaneouss or extrastimulus induced6 oscillations of local activation timing during re-entrant excitation over a constant path are common and may lead to instability and termination of the re-entry mechanism.' Similar spontaneous oscillations in local activation time during sequential beats of post-infarction sustained ventricular tachycardia in the intact beating heart have not been investigated in detail. The presence of such in vivo local activation time oscillations would suggest that the pathways which underlie electrocardiographically monomorphic appearing ventricular tachycardia are either not fixed or possess dynamic electrophysiological properties which change from one beat to the next. In a canine model of experimental myocardial infarction with inducible re-entrant ventricular tachycardia,' * we tested the hypothesis that activation during consecutive beats of sustained monomorphic ventricular tachycardia is temporally constant, ie, manifests no variability in local activation time at any myocardial site. This hypothesis was tested by comparing beat to beat local activation time variability measured during sustained monomorphic ventricular tachycardia to the beat to beat constancy observed at the ~~~
~~~
same recording sites during sinus rhythm and ventricular pacing, two non-re-entrant rhythms.
Methods Experimental model Seventeen mongrel dogs weighing 15 to 20 kg underwent left lateral thoracotomy under general anaesthesia with pentobarbitone (30 mg.kg-'). An anteroapical transmural myocardial infarction was created by ligation of the left anterior descending coronary artery immediately distal to its first diagonal branch and additional ligation of all visible epicardial collateral vessels supplying the left ventricular All procedures were camed out in compliance with the guiding principles of the American Physiological Society regarding animal experimentation. Animal preparation The animals were studied 7-12 days after recovery from experimental myocardial infarction. Anaesthetised dogs were mechanically ventilated with a constant volume ventilator, and following median sternotomy, the heart was suspended in a pericardial cradle. Aortic root pressure and central venous pressure were continuously monitored and arterial blood samples were analysed at regular intervals to maintain a physiological range of values (Poz=l.20- 1.60 kPa, Pco2=0.47-0.57 kPa, pH37.35-7.45, K'=4.0-4.5 mmol.litre-I). The mapping system consisted of a standard radial array of 62 endocardia1 and epicardial bipolar electrodes with an
~~
Cardiac Unit, Massachusetts General Hospital, Harvard Medical School, Boston, MA 02 114, USA: D S Rosenbaum, D J Wilber, J M Smith, D Du; Harvard-MIT Division of Health Sciences and Technology, Cambridge, MA, USA: J N Ruskin, H Garan. Correspondence to Dr Garan at Cardiac Unit, Massachusetts General Hospital.
Downloaded from by guest on June 6, 2016
Objective: The aim was to determine the beat to beat variability in local activation time during sustained monomorphic ventricular tachycardia in a canine model of experimental myocardial infarction. Methods: A digital template matching algorithm was developed for detecting subtle beat to beat variability in local activation timing at each of multiple ventricular sites. Ten electrically induced sustained ventricular tachycardia episodes, mean cycle length 2 1 1 (SD 40) ms, were endocardially and epicardially mapped in mongrel dogs weighing 15-20 kg. Digitised data were analysed for beat to beat local activation time variability. Similar data recorded during ventricular pacing at comparable rates and during sinus rhythm served as controls. Results: The overall mean variability of local activation time for all 10 ventricular tachycardias was 3.2(1.6) ms, range lA(1.1) ms to 4.7(2.8) ms, in contrast to the overall mean variability of 0.2(0.4) ms (p=O.OOOl) for ventricular pacing and 0.7(0.6) ms (p=O.OOOl ) for sinus rhythm. Oscillations in local activation time manifested alternans type periodicity during seven of 10 ventricular tachycardias independent of any alternans in local electrogram morphology. Conclusions: During sustained, monomorphic ventricular tachycardia, beat to beat variability and alternans type oscillations in local activation time are common and may be an intrinsic property of re-entry since they are negligibly small during ventricular pacing.
238
Rnsenhauni, Wilber; Smith, Du, Ruskiii, Curcrri
interpolar distance of 0.3 cm and interelectrode distances varying from 0.8- 1 .0 cni in the apical regions to 2.5-3.0 cm in the basal regions. Each epicardial electrode consisted of two copper wires (diameter, 0.05 cm) embedded in an epoxy button. Teflon coated stainless steel plunge electrodes (diameter. 0.012 cm) were inserted through a pair of predrilled holes in each epicardial button and anchored onto the endocardium, ensuring that the spatial orientation of the endocardial and epicardial electrodes remained constant relative to the surface of the heart and relative to each other for all experiments.
Local actiwtion tirnes and rictivcrtion sequence incipping A “window” of 10 technically optimal consecutive cardiac cycles recorded immediately following the onset of
LI
Teinplrrte inatchirig rrlgorithm rind v~rricihility i i i ~ic,ti~~rtion tiine Due to intraobserver variability the precision of estimating local activation time from extracellular electrograms during experimental ventricular tachycardia in this model is in the order of -C 10 ms.“’ Thus subtle beat to beat variability i n activation may not be reliably detected. To circumvent this problem, an automated template matching technique was developed to improve the sensitivity of detecting subtle oscillations i n local activation time between sequential beats of ventricular tachycardia or ventricular pacing at any given ventricular recording site. The surface ECG waveforms and multiple local electrograms recorded during the first of 10 consecutive cardiac cycles served as template signals. Within each recording channel. the template. ie, the first local electrograni, was shifted over each subsequent electrograni by 1 ms increments and a cross correlation coefficient. a measure of waveform similarity. was calculated for each time shift (see Appendix and fig I ). For paced beats, pacing
Downloaded from by guest on June 6, 2016
Electrt,llhvsiologiccil xturiy The chest was closed following implantation of all the bipolar recording electrodes and intrathoracic air was evacuated by suctioning through an indwelling chest tube. The electrodes were externalised through the sternotomy incision and connected to a multiplexer. Seven body surface electrodes were positioned in a Frank lead configuration for monitoring orthogonal (XYZ) ECG signals. Following a 30 minute period of stabilisation. the 62 extracellular local electrograms were filtered (5 to 500 Hz), amplified (200X), and digitised (1000 Hz with 12 bit precision) to a laboratory microcomputer (Masscomp 500, Westford, MA, USA). Simultaneously acquired surface ECG signals were treated similarly except for filtering (0.01 to 500 Hz). This mappin 6 system has been described in greater detail previously. Local electrograms and the surface ECG were recorded continuously for 5 s, permitting data acquisition during 15-20 sequential cardiac cycles during ventricular tachycardia or ventricular pacing at end expiration to minimise the influence of repiratory motion. The variability in local activation time during sinus rhythm and ventricular pacing served as controls for comparison to the variability measured during sustained monomorphic ventricular tachycardia. Each animal underwent bipolar programmed ventricular stimulation, first from the right ventricle, then from the left ventricle. The stimulus strength was twice diastolic threshold (Medtronic model 2346 stimulator, Minneapolis, MN, USA). The protocol included introduction of one, two, and three extrastimuli (SZ, S3, and SJ) following a drive train of eight beats at a fixed cycle length of 300 ms, with each coupling interval decremented by 10 ms until refractoriness for each S?, S3, and S4 was encountered. The endpoint was the induction of a sustained ventricular tachycardia. Most of the episodes of ventricular tachycardia could be terminated by rapid ventricular pacing. However, if external countershock was required, a minimum period of 10 minutes was allowed to elapse prior to resuming the protocol. Similarly, all local electrograms and the surface ECG were recorded during ventricular pacing at twice diastolic threshold current and at cycle lengths approximately the cycle length of ventricular tachycardia. For each separate ventricular tachycardia, the corresponding ventricular pacing was camed out from the ventricular recording site manifesting the earliest local activation time during ventricular tachycardia, henceforth called the ventricular tachycardia breakthrough site. Thus comparisons between ventricular tachycardia and ventricular pacing were controlled for site related as well as rate related factors, since the ventricular tachycardia data were compared only to the recordings of ventricular pacing from the same site as the ventricular tachycardia breakthrough site.
ventricular tachycardia or ventricular pacing was selected for analysis. The investigators analysed the signals recorded during the first of 10 consecutive cardiac cycles. A local activation time was assigned to each local electrogram recorded during the first cardiac cycle. using the rapid intrinsic deflection of the local clectrograrn where (dV/dt( I x II first exceeded 0.5 mV.ms-, by mutual agreement between two investigators, each blinded t o the assignment by the other. The computer calculated numeric value o f the slope could be obtained for exact determination in case o f disagreement between the assignments by the two investigators. The onset of the first cardiac cyclc (fiducial point) was also determined by two investigators as the earliest surface ECG deflection present in any one of the three orthogonal surface leads. The local activation time at each recording site was normalised to this fiducial point (LAT=O) derived from the surface ECG. The patterns of epicardial and endocardial activation were represented by computer generated isochrones, calculated by linear interpolation of adjacent local activation times, iind were plotted on an apical polar projection on the mapping grid.
239
Local activation variability during ventricular tuchycnrdia
Myocardial injiirction size and several rle~.trol,hvsiologic.al iuriables ussociated with 10 induced .siistainerl t ~ ~ ~ ) t ~ ~ ) r ~i~entriciilur ior~~liic tachycardias (VT). The last column shows the root inran .squared ( R M S ) magnitudr of betrt to beat variabilih in local Lictirwtion time (LAT). a s mean (SO).Etrdo=rndoccirdial: Epi=epicardial: LV=leji ventricle; MI=nivocurdial injhrc.tion. Endo rind Epi listed iinder "Earliest site column refer to endoc.ardial or epicarditil location of the site qf earliest activation ohserr~ed.Similarly. M I find peri-Ml r+r to the location of the earliest octiwtion as within the M I zone or at a recording site udiacent to the M I zone. I'
VT
I 2
3 4
5
6 7
8 9
in
MI s i x (% LV mtrss)
VT cycle lrnRfli ( i n s )
Emliest U T ( m s )
Etrrlirst .sit(,
KMS wrriulilit~of LAT
22 22 29 29 20 20 22 22 25
202 204 296 283 204 I68 228 204 152 I63
-5 8 -56 -5 2 -44 -18 -69
Endo/peri-M I Enddperi-MI Enddpcri-MI Enddperi- M I Epi/peri-MI EpilMl Enddperi-M I Endo/peri-MI Endo/MI Epi/peri-MI
2.5iO.5) 4 . a 1.3) 3 . 3 I .2) I .8(I . I ) 1.7(2.3)
16
-25 -42 -41 4 5
artefact was intentionally excluded from the template in order not to have the paced waveforms temporally fixed by the artefact and also since digitised pacing artefacts are known to show variable and unpredictable morphology.'' At the point in each cardiac cycle when the electrogram morphology was maximally correlated with the template morphology (ie, the two signals were superimposable) the computer assigned a local activation time at the identical point of the electrogram waveform that was chosen by the
(IIIS)
4.0(3.2) 2 3 2 .I ) 3.0( I .9) 2.oi 3 2 ) 434.6)
investigators for the template signal (see Appendix, and figs 1 and 2 ) . This process was then iterated over all 62 recording sites. At each site, cross correlation functions generated from ventricular tachycardia complexes showed monophasic and unambiguous peaks and were similar in form to cross correlation functions derived from paced complexes. This process has been shown to be a reliable technique for optimally aligning electrocardiographic waveforms, provided that the cross correlated waveforms have similar
Ventricular tachycardia CL 210 rns Downloaded from by guest on June 6, 2016
ECG
SITE 1
SITE 2
"I
I
64
?$I
I
Ventricular pacing CL 205 ms
ECG
SITE 1
SITE 2
Figure 2 Surface electrocardionrams, local electronrrrms, and their activation times (. m s.) are shown durinw ventricular tac1ivcardic.i I, VT)I and ventricula; pacing petj?ormed at nearly identical cycle length (CL) in the same preparation. The ECG ,fiduciuI point (interrilpted i~erticcil line) assigned by the investigator to the j r s t complex is reproduced on the subsequent complexes by the ternplate mutching program. Siniilarly. the local activation times assigned to the electrograms recorded from two mapping sites during the first cardiac c j d e are reproduced precisely on subsequent cycles by the template matching technique. Note that at identical recording sites. beat to beat oscillritions of local activation time are evident during VT und absent during pacing.
240
Rosetibaunt, Wilber; Smith, Du, Rirskin, Gartin
morphology,” a restriction clearly adhered to in this study due to precise and quantitative definition of “monomorphic” (see below). This technique then avoids spurious detection of signal artefacts which are often association with completely automated algorithms and at the same time reproduces exactly the electrophysiologist’s estimate of the instant of activation on each subsequent electrogram recorded at the same site (figs 1 and 2 ) , thereby eliminating the error from intraobserver variability.
Necropsy exarnintitioti At the end of each study the animal was killed by induction of ventricular fibrillation. The heart was excised immediately, taking care not to dislocate the electrodes in situ, and washed with isotonic saline. A catheter was placed in the left main coronary artery with a ligature around it and the vessel was perfused for 3 min with 1 % triphenyltetrazolium chloride solution.” The precise positions of all recording electrodes were confirmed under direct visual inspection. The heart was sectioned transversely into 1 cm slices which were then weighed and their surfaces traced for myocardial infarct size estimation by planimetry.
Results Characteristics of induced ventriculur tachycardici In nine of the 17 animals, a total of 42 sustained ventricular tachycardias were induced by programmed stimulation. Ten of these 42 ventricular tachycardias in six animals satisfied the stringent quantitative criteria (see Methods) for monomorphic ventricular tachycardia and were analysed in detail. The mean cycle length was 211 (SD 40) ms. The aortic root pressure was stable and without evidence of pulsus alternans throughout the course of ventricular tachycardia, and returned to its preinduction value following termination of each ventricular tachycardia episode. The local activation times of the breakthrough sites, relative to the fiducial point, were all presystolic, ranging from -18 to -69 ms, with a mean value of -47 ms (table).
Alternating versu,s iioii-alterticiting vcrricrbility Alternans periodicities, characterised by local activation time oscillations with a period equal to twice ventricular tachycardia cycle length, were observed during seven of the 10 ventricular tachycardia episodes, and were present at all recording sites manifesting oscillations during four episodes of ventricular tachycardia. There was no pulsus alternans in the systemic arterial pressure during any of these ventricular tachycardias or their pacing controls. In order to exclude the possibility that alternans in local activation time was simply due to beat to beat alternations in local electrogram morphology, the correlation coefficients C,,,, (see Appendix),
Downloaded from by guest on June 6, 2016
Definitions and statistical analysis This study analysed local activation time variability from one ventricular tachycardia beat to the next at the same recording site, not local activation time variability between different sites. At a given ventricular recording site the magnitude of beat to beat fluctuations in local activation time was expressed as the root mean square (RMS) of local activation time shifts (henceforth referred to as RMS-LAT) measured during 10 consecutive cardiac cycles (see Appendix). The average RMS-LAT measured from ventricular recording sites during 10 beats of ventricular tachycardia was compared to the average RMS-LAT observed during 10 beats of sinus rhythm and 10 beats of ventricular pacing, using an unpaired Student’s t test with appropriate Bonferroni correction for multiple comparisons. I 4 For all comparisons, values are given as means (SD) and p
237
Local activation variability during monomorphic ventricular tachycardia in the dog David S Rosenbaum, David J Wilber, Joseph M Smith, Duke Du, Jeremy N Ruskin, and Hasan Garan
T
he re-entrant pathways of monomorphic ventricular tachycardia in chronic myocardial infarction are thought to be spatially fixed relative to either an anatomical or functional zone of conduction block.'-' If so the timing of local electrical activation at a given myocardial site should also be constant from one ventricular tachycardia beat to the next. However, recent experimental studies have shown that spontaneouss or extrastimulus induced6 oscillations of local activation timing during re-entrant excitation over a constant path are common and may lead to instability and termination of the re-entry mechanism.' Similar spontaneous oscillations in local activation time during sequential beats of post-infarction sustained ventricular tachycardia in the intact beating heart have not been investigated in detail. The presence of such in vivo local activation time oscillations would suggest that the pathways which underlie electrocardiographically monomorphic appearing ventricular tachycardia are either not fixed or possess dynamic electrophysiological properties which change from one beat to the next. In a canine model of experimental myocardial infarction with inducible re-entrant ventricular tachycardia,' * we tested the hypothesis that activation during consecutive beats of sustained monomorphic ventricular tachycardia is temporally constant, ie, manifests no variability in local activation time at any myocardial site. This hypothesis was tested by comparing beat to beat local activation time variability measured during sustained monomorphic ventricular tachycardia to the beat to beat constancy observed at the ~~~
~~~
same recording sites during sinus rhythm and ventricular pacing, two non-re-entrant rhythms.
Methods Experimental model Seventeen mongrel dogs weighing 15 to 20 kg underwent left lateral thoracotomy under general anaesthesia with pentobarbitone (30 mg.kg-'). An anteroapical transmural myocardial infarction was created by ligation of the left anterior descending coronary artery immediately distal to its first diagonal branch and additional ligation of all visible epicardial collateral vessels supplying the left ventricular All procedures were camed out in compliance with the guiding principles of the American Physiological Society regarding animal experimentation. Animal preparation The animals were studied 7-12 days after recovery from experimental myocardial infarction. Anaesthetised dogs were mechanically ventilated with a constant volume ventilator, and following median sternotomy, the heart was suspended in a pericardial cradle. Aortic root pressure and central venous pressure were continuously monitored and arterial blood samples were analysed at regular intervals to maintain a physiological range of values (Poz=l.20- 1.60 kPa, Pco2=0.47-0.57 kPa, pH37.35-7.45, K'=4.0-4.5 mmol.litre-I). The mapping system consisted of a standard radial array of 62 endocardia1 and epicardial bipolar electrodes with an
~~
Cardiac Unit, Massachusetts General Hospital, Harvard Medical School, Boston, MA 02 114, USA: D S Rosenbaum, D J Wilber, J M Smith, D Du; Harvard-MIT Division of Health Sciences and Technology, Cambridge, MA, USA: J N Ruskin, H Garan. Correspondence to Dr Garan at Cardiac Unit, Massachusetts General Hospital.
Downloaded from by guest on June 6, 2016
Objective: The aim was to determine the beat to beat variability in local activation time during sustained monomorphic ventricular tachycardia in a canine model of experimental myocardial infarction. Methods: A digital template matching algorithm was developed for detecting subtle beat to beat variability in local activation timing at each of multiple ventricular sites. Ten electrically induced sustained ventricular tachycardia episodes, mean cycle length 2 1 1 (SD 40) ms, were endocardially and epicardially mapped in mongrel dogs weighing 15-20 kg. Digitised data were analysed for beat to beat local activation time variability. Similar data recorded during ventricular pacing at comparable rates and during sinus rhythm served as controls. Results: The overall mean variability of local activation time for all 10 ventricular tachycardias was 3.2(1.6) ms, range lA(1.1) ms to 4.7(2.8) ms, in contrast to the overall mean variability of 0.2(0.4) ms (p=O.OOOl) for ventricular pacing and 0.7(0.6) ms (p=O.OOOl ) for sinus rhythm. Oscillations in local activation time manifested alternans type periodicity during seven of 10 ventricular tachycardias independent of any alternans in local electrogram morphology. Conclusions: During sustained, monomorphic ventricular tachycardia, beat to beat variability and alternans type oscillations in local activation time are common and may be an intrinsic property of re-entry since they are negligibly small during ventricular pacing.
238
Rnsenhauni, Wilber; Smith, Du, Ruskiii, Curcrri
interpolar distance of 0.3 cm and interelectrode distances varying from 0.8- 1 .0 cni in the apical regions to 2.5-3.0 cm in the basal regions. Each epicardial electrode consisted of two copper wires (diameter, 0.05 cm) embedded in an epoxy button. Teflon coated stainless steel plunge electrodes (diameter. 0.012 cm) were inserted through a pair of predrilled holes in each epicardial button and anchored onto the endocardium, ensuring that the spatial orientation of the endocardial and epicardial electrodes remained constant relative to the surface of the heart and relative to each other for all experiments.
Local actiwtion tirnes and rictivcrtion sequence incipping A “window” of 10 technically optimal consecutive cardiac cycles recorded immediately following the onset of
LI
Teinplrrte inatchirig rrlgorithm rind v~rricihility i i i ~ic,ti~~rtion tiine Due to intraobserver variability the precision of estimating local activation time from extracellular electrograms during experimental ventricular tachycardia in this model is in the order of -C 10 ms.“’ Thus subtle beat to beat variability i n activation may not be reliably detected. To circumvent this problem, an automated template matching technique was developed to improve the sensitivity of detecting subtle oscillations i n local activation time between sequential beats of ventricular tachycardia or ventricular pacing at any given ventricular recording site. The surface ECG waveforms and multiple local electrograms recorded during the first of 10 consecutive cardiac cycles served as template signals. Within each recording channel. the template. ie, the first local electrograni, was shifted over each subsequent electrograni by 1 ms increments and a cross correlation coefficient. a measure of waveform similarity. was calculated for each time shift (see Appendix and fig I ). For paced beats, pacing
Downloaded from by guest on June 6, 2016
Electrt,llhvsiologiccil xturiy The chest was closed following implantation of all the bipolar recording electrodes and intrathoracic air was evacuated by suctioning through an indwelling chest tube. The electrodes were externalised through the sternotomy incision and connected to a multiplexer. Seven body surface electrodes were positioned in a Frank lead configuration for monitoring orthogonal (XYZ) ECG signals. Following a 30 minute period of stabilisation. the 62 extracellular local electrograms were filtered (5 to 500 Hz), amplified (200X), and digitised (1000 Hz with 12 bit precision) to a laboratory microcomputer (Masscomp 500, Westford, MA, USA). Simultaneously acquired surface ECG signals were treated similarly except for filtering (0.01 to 500 Hz). This mappin 6 system has been described in greater detail previously. Local electrograms and the surface ECG were recorded continuously for 5 s, permitting data acquisition during 15-20 sequential cardiac cycles during ventricular tachycardia or ventricular pacing at end expiration to minimise the influence of repiratory motion. The variability in local activation time during sinus rhythm and ventricular pacing served as controls for comparison to the variability measured during sustained monomorphic ventricular tachycardia. Each animal underwent bipolar programmed ventricular stimulation, first from the right ventricle, then from the left ventricle. The stimulus strength was twice diastolic threshold (Medtronic model 2346 stimulator, Minneapolis, MN, USA). The protocol included introduction of one, two, and three extrastimuli (SZ, S3, and SJ) following a drive train of eight beats at a fixed cycle length of 300 ms, with each coupling interval decremented by 10 ms until refractoriness for each S?, S3, and S4 was encountered. The endpoint was the induction of a sustained ventricular tachycardia. Most of the episodes of ventricular tachycardia could be terminated by rapid ventricular pacing. However, if external countershock was required, a minimum period of 10 minutes was allowed to elapse prior to resuming the protocol. Similarly, all local electrograms and the surface ECG were recorded during ventricular pacing at twice diastolic threshold current and at cycle lengths approximately the cycle length of ventricular tachycardia. For each separate ventricular tachycardia, the corresponding ventricular pacing was camed out from the ventricular recording site manifesting the earliest local activation time during ventricular tachycardia, henceforth called the ventricular tachycardia breakthrough site. Thus comparisons between ventricular tachycardia and ventricular pacing were controlled for site related as well as rate related factors, since the ventricular tachycardia data were compared only to the recordings of ventricular pacing from the same site as the ventricular tachycardia breakthrough site.
ventricular tachycardia or ventricular pacing was selected for analysis. The investigators analysed the signals recorded during the first of 10 consecutive cardiac cycles. A local activation time was assigned to each local electrogram recorded during the first cardiac cycle. using the rapid intrinsic deflection of the local clectrograrn where (dV/dt( I x II first exceeded 0.5 mV.ms-, by mutual agreement between two investigators, each blinded t o the assignment by the other. The computer calculated numeric value o f the slope could be obtained for exact determination in case o f disagreement between the assignments by the two investigators. The onset of the first cardiac cyclc (fiducial point) was also determined by two investigators as the earliest surface ECG deflection present in any one of the three orthogonal surface leads. The local activation time at each recording site was normalised to this fiducial point (LAT=O) derived from the surface ECG. The patterns of epicardial and endocardial activation were represented by computer generated isochrones, calculated by linear interpolation of adjacent local activation times, iind were plotted on an apical polar projection on the mapping grid.
239
Local activation variability during ventricular tuchycnrdia
Myocardial injiirction size and several rle~.trol,hvsiologic.al iuriables ussociated with 10 induced .siistainerl t ~ ~ ~ ) t ~ ~ ) r ~i~entriciilur ior~~liic tachycardias (VT). The last column shows the root inran .squared ( R M S ) magnitudr of betrt to beat variabilih in local Lictirwtion time (LAT). a s mean (SO).Etrdo=rndoccirdial: Epi=epicardial: LV=leji ventricle; MI=nivocurdial injhrc.tion. Endo rind Epi listed iinder "Earliest site column refer to endoc.ardial or epicarditil location of the site qf earliest activation ohserr~ed.Similarly. M I find peri-Ml r+r to the location of the earliest octiwtion as within the M I zone or at a recording site udiacent to the M I zone. I'
VT
I 2
3 4
5
6 7
8 9
in
MI s i x (% LV mtrss)
VT cycle lrnRfli ( i n s )
Emliest U T ( m s )
Etrrlirst .sit(,
KMS wrriulilit~of LAT
22 22 29 29 20 20 22 22 25
202 204 296 283 204 I68 228 204 152 I63
-5 8 -56 -5 2 -44 -18 -69
Endo/peri-M I Enddperi-MI Enddpcri-MI Enddperi- M I Epi/peri-MI EpilMl Enddperi-M I Endo/peri-MI Endo/MI Epi/peri-MI
2.5iO.5) 4 . a 1.3) 3 . 3 I .2) I .8(I . I ) 1.7(2.3)
16
-25 -42 -41 4 5
artefact was intentionally excluded from the template in order not to have the paced waveforms temporally fixed by the artefact and also since digitised pacing artefacts are known to show variable and unpredictable morphology.'' At the point in each cardiac cycle when the electrogram morphology was maximally correlated with the template morphology (ie, the two signals were superimposable) the computer assigned a local activation time at the identical point of the electrogram waveform that was chosen by the
(IIIS)
4.0(3.2) 2 3 2 .I ) 3.0( I .9) 2.oi 3 2 ) 434.6)
investigators for the template signal (see Appendix, and figs 1 and 2 ) . This process was then iterated over all 62 recording sites. At each site, cross correlation functions generated from ventricular tachycardia complexes showed monophasic and unambiguous peaks and were similar in form to cross correlation functions derived from paced complexes. This process has been shown to be a reliable technique for optimally aligning electrocardiographic waveforms, provided that the cross correlated waveforms have similar
Ventricular tachycardia CL 210 rns Downloaded from by guest on June 6, 2016
ECG
SITE 1
SITE 2
"I
I
64
?$I
I
Ventricular pacing CL 205 ms
ECG
SITE 1
SITE 2
Figure 2 Surface electrocardionrams, local electronrrrms, and their activation times (. m s.) are shown durinw ventricular tac1ivcardic.i I, VT)I and ventricula; pacing petj?ormed at nearly identical cycle length (CL) in the same preparation. The ECG ,fiduciuI point (interrilpted i~erticcil line) assigned by the investigator to the j r s t complex is reproduced on the subsequent complexes by the ternplate mutching program. Siniilarly. the local activation times assigned to the electrograms recorded from two mapping sites during the first cardiac c j d e are reproduced precisely on subsequent cycles by the template matching technique. Note that at identical recording sites. beat to beat oscillritions of local activation time are evident during VT und absent during pacing.
240
Rosetibaunt, Wilber; Smith, Du, Rirskin, Gartin
morphology,” a restriction clearly adhered to in this study due to precise and quantitative definition of “monomorphic” (see below). This technique then avoids spurious detection of signal artefacts which are often association with completely automated algorithms and at the same time reproduces exactly the electrophysiologist’s estimate of the instant of activation on each subsequent electrogram recorded at the same site (figs 1 and 2 ) , thereby eliminating the error from intraobserver variability.
Necropsy exarnintitioti At the end of each study the animal was killed by induction of ventricular fibrillation. The heart was excised immediately, taking care not to dislocate the electrodes in situ, and washed with isotonic saline. A catheter was placed in the left main coronary artery with a ligature around it and the vessel was perfused for 3 min with 1 % triphenyltetrazolium chloride solution.” The precise positions of all recording electrodes were confirmed under direct visual inspection. The heart was sectioned transversely into 1 cm slices which were then weighed and their surfaces traced for myocardial infarct size estimation by planimetry.
Results Characteristics of induced ventriculur tachycardici In nine of the 17 animals, a total of 42 sustained ventricular tachycardias were induced by programmed stimulation. Ten of these 42 ventricular tachycardias in six animals satisfied the stringent quantitative criteria (see Methods) for monomorphic ventricular tachycardia and were analysed in detail. The mean cycle length was 211 (SD 40) ms. The aortic root pressure was stable and without evidence of pulsus alternans throughout the course of ventricular tachycardia, and returned to its preinduction value following termination of each ventricular tachycardia episode. The local activation times of the breakthrough sites, relative to the fiducial point, were all presystolic, ranging from -18 to -69 ms, with a mean value of -47 ms (table).
Alternating versu,s iioii-alterticiting vcrricrbility Alternans periodicities, characterised by local activation time oscillations with a period equal to twice ventricular tachycardia cycle length, were observed during seven of the 10 ventricular tachycardia episodes, and were present at all recording sites manifesting oscillations during four episodes of ventricular tachycardia. There was no pulsus alternans in the systemic arterial pressure during any of these ventricular tachycardias or their pacing controls. In order to exclude the possibility that alternans in local activation time was simply due to beat to beat alternations in local electrogram morphology, the correlation coefficients C,,,, (see Appendix),
Downloaded from by guest on June 6, 2016
Definitions and statistical analysis This study analysed local activation time variability from one ventricular tachycardia beat to the next at the same recording site, not local activation time variability between different sites. At a given ventricular recording site the magnitude of beat to beat fluctuations in local activation time was expressed as the root mean square (RMS) of local activation time shifts (henceforth referred to as RMS-LAT) measured during 10 consecutive cardiac cycles (see Appendix). The average RMS-LAT measured from ventricular recording sites during 10 beats of ventricular tachycardia was compared to the average RMS-LAT observed during 10 beats of sinus rhythm and 10 beats of ventricular pacing, using an unpaired Student’s t test with appropriate Bonferroni correction for multiple comparisons. I 4 For all comparisons, values are given as means (SD) and p
