Primary Ventricular Fibrillation Is Associated With Increased Paced Right Ventricular Electrogram Fractionation
Background The mechanisms of spontaneous ventricular fibrillation (primary VF) in patients without structural heart disease are obscure. A new technique has shown that in patients with hypertrophic cardiomyopathy conduction of fractionated ventricular paced beats, recorded at several right ventricular sites, is prolonged in individuals who have suffered a VF arrest, and this may reveal one component of a reentrant substrate. Patients with primary VF were studied with the same methods to determine whether similar abnormalities are present in this group.
Methods and Results Nine patients with primary VF were studied by pacing one right ventricular (RV) site by use of a constant drive train with an extrastimulus inserted every third beat and reducing the extrastimulus coupling interval (S1S2 interval) by 1 ms on each occasion while recording at three other sites. The delay of each fractionated potential in the high-pass–filtered electrograms in response to the extrastimulus was determined and used to form conduction curves of delay versus the S1S2 interval. These curves were repeated by pacing each RV site in turn and recording from the other three sites. The curves were characterized by determining the S1S2 interval at which electrogram components increased in delay by 0.75 ms/20 ms reduction in S1S2 interval and the increase in electrogram duration between a coupling interval of 350 ms and 1 ms above refractoriness. Seven control patients were studied using the same method. The mean increase in electrogram duration in VF patients was 13 ms (range, 3 to 23 ms) compared with 4 ms (range, −2 to 14 ms) in unaffected control patients. The extrastimulus coupling interval at which delay increased was 318 ms (range, 293 to 334 ms) in VF patients and 274 ms (range, 265 to 284 ms) in control patients (P<.01). There was no difference between the number of fractionated potentials in VF patients and control patients.
Conclusions In primary VF patients, the individual potentials within fractionated electrograms have increased delays when compared with control patients. This may identify one component of a reentrant arrhythmic substrate.
The development of portable defibrillators has led to the survival of patients who have developed VF without detectable heart disease (primary VF). Since these patients have no structural or conventional EP abnormality that would explain the development of VF, it is not clear how to investigate them to assess their risk of another episode of VF or how to understand the mechanisms of their arrhythmias. This article describes the use of a new technique to measure inhomogeneity of intraventricular conduction1 in nine patients who have survived primary VF. The technique is based on the hypothesis that there are a number of potential conduction pathways within the ventricle with different conduction velocities. These pathways may be exposed by applying premature paced stimuli in the RV and analyzing the increase in delayed, small amplitude potentials of electrograms (fractionation of electrograms) measured elsewhere in the RV. Fractionated potentials correspond to asynchronous depolarization of fiber bundles close to the recording electrodes2 and indicate abnormal ventricular activation that may form part of a reentrant arrhythmic substrate. Thus, the presence of fractionated potentials in patients with diseases that can cause sudden arrhythmic death may identify those individuals who are at risk of VF.
The technique was originally developed to detect the possible EP effects of myocardial disarray in HCM. Patients with HCM who had suffered documented VF showed an increase in electrogram fractionation when compared with HCM patients who had not suffered VF and with unaffected control patients.1 This suggested that intraventricular conduction delay and electrogram lengthening reveal part of an arrhythmogenic substrate that may also be present in patients with other diseases that cause VF without myocardial infarction or structural abnormalities. Therefore, this hypothesis has been tested by measuring intraventricular conduction delays in patients with primary VF.
The patients, after giving written informed consent, were studied in a postabsorptive, drug-free state; all antiarrhythmic agents were stopped at least 6 half-lives before the study. Patients were premedicated with lorazepam (2 to 4 mg) and were administered 12.5 mg prochlorperazine at the start of the study. Intravenous diazepam and diamorphine were administered as required for sedation. A 1-cm–spacing 6F bipolar catheter was inserted into the RV apex, a 1-cm tripolar catheter was inserted at the His bundle recording site, and a 0.5-cm quadrapolar catheter was inserted into the high right atrium via the right femoral vein with 5 to 10 mL of 2% lidocaine for local anesthesia. Conventional atrioventricular and ventriculoatrial conduction curves were obtained. Programmed electrical stimulation was performed from the RV apex and RVOT with a drive train of eight stimuli delivered at cycle lengths of 600, 500, and 400 ms with up to three extrastimuli according to the Wellens protocol.3 The tripolar catheter was then advanced to the septum, and two additional 1-cm bipolar catheters were positioned in the RV inferoposterior wall and RVOT. Care was taken to position the catheters in standard positions and to place the septal and inferior wall catheters at approximately the same distance from the apical catheter (Fig 1A⇓). One patient had very tortuous pelvic veins, and a catheter could not be positioned in the inferior RV. This patient, therefore, was studied with only three intraventricular electrodes.
A computer-generated pacing sequence was delivered from the apical catheter, and electrograms were recorded from the remaining catheters. The high right atrium was paced simultaneously with the ventricle to avoid fusion between paced ventricular and sinus beats. The pacing threshold before each run was determined with constant rate pacing at 120 beats per minute and the sequence delivered with a stimulus voltage twice that of the threshold. Once the sequence was finished, an identical sequence was issued from the septal catheter, and electrograms were recorded from the other ventricular catheters. This sequence was repeated twice more by pacing from the inferior wall and then the RVOT while recording from the other three catheters.
The pacing sequence consisted of a drive train at a basic cycle length of 480 ms with an extrastimulus inserted every third beat. The extrastimulus coupling interval was reduced by 1 ms on each occasion, from 450 ms down to the ventricular effective refractory period (VERP). Thus, during the study, approximately 250 electrograms were collected in response to increasingly premature extrastimuli at three ventricular sites, and this was repeated with stimulation at each ventricular catheter, yielding 12 sets of electrograms.
The electrograms were recorded with band limitation by a single-pole 50-Hz (−3-dB) high-pass filter and a 280-Hz (−3-dB) four-pole Bessel low-pass antialiasing filter and were digitized at 1 KHz with 12-bit accuracy. All unsatisfactory electrograms that showed fusion with atrial beats or failure to capture were rejected from the analysis. The first phase of the analysis was performed to distinguish physiologically significant potentials from those related to background noise. The electrograms in response to an extrastimulus were processed with a digital, zero-phase high-pass filter with a −3-dB point of 150 Hz1 4 (Fig 2⇓). This technique emphasizes small, sharply defined potentials in the electrogram related to local depolarizations and rejects slowly moving components, such as the intracardiac T wave of the last drive cycle, while preserving the relative timings of each potential. The amplitudes of the samples between 150 and 200 ms after the extrastimulus were measured to determine the signal noise. The detection threshold was set at twice the maximum noise amplitude. The length of the electrogram was determined by fitting separate interpolating cubic splines to both the maxima and the minima in the electrogram as a function of time. The earliest and latest points at which these interpolating functions decreased to below the detection threshold were taken as the initial and final limits of the electrogram. All potentials within these limits that were greater than twice the detection threshold were extracted, and their delays after the pacing spike were determined (Fig 1B⇑). These data were used to construct a conduction curve (Fig 1C⇑) of the delay of individual potentials at each extrastimulus coupling interval, and examples of these curves are shown in Figs 3⇓, 4⇓, and 5⇓. The number of potentials in each electrogram was determined and the increase in number over a drive-cycle electrogram (S1S1=480 ms) calculated. The maximum increase in the number of potentials was taken as another index of fractionation that is distinct from the delays of each potential.
The second stage of the analysis reduces the conduction curves to simple parameters for statistical analysis. The envelopes of the curves are defined by identifying the first and last components of the electrogram at each S1S2 coupling interval. These envelopes are then smoothed by fitting functions, cubic splines,1 to the upper and lower boundaries of the envelopes (Figs 1C⇑ and 4⇑). These functions are designed to make a smooth approximation to the envelope and to minimize the effects of small differences between successive electrograms and thus make measurements of the shape of the curve more reliable. The curves were then quantified by use of the cubic splines fitted to the envelope to extract two parameters that are designed to reflect the arrhythmic potential of the tissue. The first parameter is the difference between the electrogram duration at an S1S2 interval of 350 ms and at 1 ms greater than VERP.1 This reflects the ability of the tissue to create dispersion of activation and hence an arrhythmic substrate. The second parameter is the S1S2 coupling interval at which delays start to increase to greater than 0.75 ms/20-ms decrease in S1S2 interval (which is the lowest stable threshold in the presence of signal noise), and this is shown as point I in Fig 1C⇑. This reflects the ease with which a substrate, activation delay, can be exposed with a stimulus and so may relate to the vulnerability of the tissue to arrhythmogenic stimuli. Thus, patients with electrograms that widen near VERP and start to become delayed at long S1S2 coupling intervals are more vulnerable to arrhythmia than those with delays that occur at short S1S2 intervals. These parameters are averaged for all the curves for a particular patient, thus creating a single statistically independent observation of mean increase in electrogram duration between an S1S2 interval of 350 ms and at VERP and a mean S1S2 interval at which delays start to increase (Fig 1D⇑).
Nine patients were studied who had been resuscitated from ECG-proven VFs. Seven patients were studied within 14 days, one at 6 months, and one at 3 years after a VF arrest. Their ages ranged from 16 to 54 years, and seven were men. Four patients had developed VF during exercise and one during “emotional stress” while watching a soccer match. There was no family history of sudden death in any patient and no history of alcohol or drug abuse. All patients had normal resting ECGs and QTc intervals. All patients had normal exercise tests, except one patient who developed ST-segment depression in the anterolateral ECG leads during exercise in the absence of chest pain or coronary artery disease. An ergonovine test in this patient was negative. One patient had a 40% stenosis of the circumflex artery with no ST-segment change on exercise, and the remainder had normal coronary arteries. All had normal left ventriculograms except for one patient who had mild anterior wall motion abnormalities. All patients had normal right RV echocardiograms and biopsies. One patient had mild, nonregurgitant, mitral valve prolapse that was not associated with a click or a murmur. One patient (No. 4) had an abnormal signal-averaged ECG, whereas the rest were normal. Seven patients were studied as control patients (3 men; age range, 17 to 67 years), and all had normal resting ECGs and echocardiograms. Two of these control patients had AV nodal reentrant tachyarrhythmia (AVNRT), two were studied to diagnose palpitations but had no demonstrable abnormality, one had paroxysmal atrial fibrillation, and two were studied in the investigation of syncope and had normal EP studies.
No patient had an accessory pathway or dual AV nodal conduction. No patient had monomorphic VT or VF in response to programmed ventricular stimulation. Two patients developed nonsustained polymorphic VT in response to three extrastimuli. No ventricular arrhythmias occurred during the intraventricular conduction pacing protocol, but one 16-year-old patient developed common atrial flutter, and one 53-year-old patient developed atrial fibrillation.
Fig 3⇑ shows a conduction curve from a 37-year-old control patient. There is a gradual reduction in the delay with premature stimulation until it increases sharply at S1S2 intervals below 250 ms. This is a characteristic pattern seen in low-risk patients with HCM or in control patients.1 4 5 Figs 4⇑ and 5⇑ show conduction curves from VF survivors. They have features in common with curves from HCM VF survivors: Fig 4⇑ shows a gradual increase in delay with an increased duration of the electrogram1 4 5 (cf Fig 4⇑ in Reference 1), and Fig 5⇑ shows a curve in which there is an abrupt increase in the electrogram duration at 320 ms (cf Fig 5⇑ in Reference 1).
The Table⇓ shows the mean increases in electrogram width and the S1S2 interval at which delay increases for the patient population and control group. The mean increase in electrogram duration was 13 ms in VF survivors and 4 ms in control patients. The mean S1S2 interval at which delay increased was 318 ms in VF survivors and 274 ms in control patients. The control and VF groups can be completely separated with these variables by constructing a discriminant line,6 with the latter having increased electrogram duration and a longer S1S2 interval at which delay increases. The probability of a line separating the two groups is <.01 (see “Statistical Appendix”). These data are shown as a scattergram in Fig 6⇓ together with the discriminant line that separates the two populations (see Fig 1D⇑). However, the median difference in the number of potentials measured at a cycle length of 480 ms and the maximum number recorded at any S1S2 during the entire sequence was 16 potentials in the VF patients (range, 3 to 27) and 13 in the control group (range, 3 to 21). These results are consistent with the hypothesis that delayed activation will be present in patients with VF, but the number of potentials themselves does not discriminate between VF patients and control patients.
Finally, two patients showed striking regional fractionation, which suggests a localized abnormality. Fig 7⇓ shows electrograms recorded at the septum, inferior wall, and RVOT while pacing the apex in patient No. 2. This suggests a regional abnormality in conduction confined to the septum. One other patient also showed a similar pattern of activation; however, in the remaining seven patients, the fractionation is more generalized, occurring at the majority of the recording electrodes.
The objective of this work was to see whether patients with primary VF had greater intraventricular conduction delays than control patients and the delays are in fact greater in VF patients. Although separation between VF patients and control patients is unlikely to occur as a random event, the line that separates the two groups is one of many lines that can be constructed using particular assumptions about distributions of the data in the two groups. Therefore, the discriminant line in Fig 6⇑ is itself a statistical hypothesis that needs prospective testing with new sets of control patients and VF patients.
The intraventricular conduction delays, revealed by fractionated potentials, would be expected to be arrhythmogenic, since they might act as one component of a reentrant substrate. While conduction delay discriminates between VF patients and control patients, the increase in the number of fractionated potentials does not distinguish between them. In control patients these potentials occur within the bulk of the electrogram and do not extend its duration (Fig 2B⇑), unlike the delayed potentials in VF patients, which supports the premise that delayed conduction forms part of a reentrant substrate. This finding is consistent with a study of post–myocardial infarction patients, in which conduction delay in response to premature stimulation was associated with the induction of VF during programmed stimulation7 and a canine ventricular infarct study that showed that dogs with inducible arrhythmias had increased electrogram duration in the region of the infarct when premature extrastimuli were introduced by pacing the His bundle.8
The analysis uses the mean values from 12 curves of electrogram duration and point of increasing delay, and thus the myocardium is treated as though it were uniformly diseased, although in two patients there was a striking regional abnormality. This distinction is of great importance in the pathology of the underlying disease because it is likely that primary VF is caused by several diseases, some of which may be patchy and some of which may be diffuse. While Fig 7⇑ shows an example of localized conduction disturbance in the septum, this disturbance was more generalized in other patients. Finally, the pattern of increased delay with premature stimulation is different in the examples shown in Figs 4⇑ and 5⇑. In Fig 4⇑ there is a gradual increase in delay with reduction of S1S2 interval, whereas in Fig 5⇑ there is an abrupt increase in the number of delayed fractionated potentials, which suggests different characteristics of myocardial conduction between the stimulating and recording electrodes. Therefore, additional studies are needed with a larger number of multielectrode catheters to map the conduction disturbances in greater detail and to determine their anatomic extent.
This study raises two major concerns. The first is that the technique originated from a hypothesis about the EP effects of disarray in HCM. Although biopsies from these primary VF patients do not show disarray, there are functional EP abnormalities similar to those seen in HCM patients with VF in that there is delayed intraventricular activation in response to premature stimulation. Fractionated potentials arise from discontinuous and anisotropic propagations that have been studied at the microscopic level in excised human atrial appendage and in canine myocardium.2 9 10 11 12 13 One mechanism of fractionation demonstrated in human atrial tissue is discontinuous side-to-side conduction between fibers, which becomes more pronounced with premature stimulation.2 9 These effects are determined by passive properties of the tissue and are associated with increasing age and the presence of small collagenous septae between fibers. Discontinuous conduction may also occur axially with premature stimulation, resulting in blocked conduction, whereas discontinuous transverse conduction may occur in a reentrant pattern, strengthening the idea that fractionated potentials may yield information about potential reentry circuits.10 Finally, anisotropic conduction in the subendocardial layers of experimental canine infarcts also results in fractionated potentials that stem from discontinuous propagation and slowed conduction associated with reduced transverse cell-to-cell connections despite normal action potentials in the myocytes involved.11 12 13 Our results are obtained from measurements made over several centimeters, which is far larger than these experimental preparations. It is therefore difficult to compare our results with those obtained from isolated tissue studies, but anisotropic conduction could account for our observations, and this may be detectable clinically using special-purpose catheters with four electrodes arranged around a “pigtail,” which will measure the directions of activation from the delays in each potential. Since the VF patients’ biopsies were normal, one possible mechanism of anisotropy and fractionation is that abnormalities of gap junctions reduce transverse conduction velocities. The density of gap junctions has been shown to be reduced in hypertrophied and ischemic myocardium14 distant from fibrotic scarring as well as in the peri-infarction zone of human and canine infarcts.15 16 Therefore, particular attention should be paid to numbers and structure of gap junctions and side-to-side connections in biopsies from future patients.
The second concern is that the conduction disturbance is less than that seen in HCM patients who have had VF, with a mean point of increased delay of 363 ms (range, 342 to 386 ms1) in HCM versus 318 ms (range, 293 to 334 ms) in primary VF. One possible interpretation of this result is that patients with HCM have major conduction disturbances due to disarray that may require a relatively minor trigger to produce VF, whereas primary VF patients with a relatively minor substrate require a major, but unknown, trigger. This model could explain why VF is relatively common in HCM but is rare in patients with structurally normal hearts.
There may be several triggering factors in primary VF.17 Coronary artery spasm has been described as leading to ventricular arrhythmia18 in survivors of primary VF. This trigger was not examined in this study, except in one patient who had ST-segment changes on exercise. Four patients developed VF during exercise, which is recognized as being associated with sudden cardiac death.19 20 Interestingly, these patients showed an apparent substrate despite being studied while sedated, when presumably the sympathetic drive to the heart was low. The effects of sympathetic drive on measurements of intraventricular conduction are unknown; nevertheless, this is an important question in the investigation of exercise-induced arrhythmias. One patient developed VF during emotional stress. The incidence of sudden cardiac death rises in populations exposed to stresses such as earthquakes21 or missile bombardment,22 but the links between stress and sudden death are uncertain23 and may have been coincidental in this patient. In the remaining patients, no trigger could be identified.
Conventional EP testing in these patients with programmed RV stimulation did not induce an arrhythmia except for nonsustained polymorphic VT in two cases. This is consistent with other studies of primary VF24 25 and highlights a major problem in the investigation of ventricular arrhythmias in noncoronary heart disease. Induction of monomorphic VT indicates the presence of a substrate for that arrhythmia; however, induction of polymorphic VT or VF may simply reflect aggressive stimulation and cannot be interpreted as evidence of a specific arrhythmia substrate. Also, failure to induce arrhythmias in primary VF survivors does not imply the absence of a substrate and does not imply a good prognosis.24 Two patients (Nos. 1 and 4) who had no inducible arrhythmia with programmed electrical stimulation both have had several episodes of nonsustained VT recorded by their intracoronary devices, implying the presence of a specific substrate. Therefore, methods of exposing components of an arrhythmic substrate, without induction of a non-specific tachycardia, may be more useful than programmed electrical stimulation in the investigation of VF patients without structural heart disease.
Although the fractionation results distinguish between primary VF patients and control patients, the technique can probably be simplified. The use of two S1 stimuli was devised as a compromise between collecting a large number of S2 complexes and keeping the pacing run relatively short. Since the S1 beat before the S2 stimulus does not vary during the run, the protocol does not cause a gradual change in myocardial conduction, and in HCM patients the results appear to be independent of the length of the pacing run. This will simplify the protocol by reducing the number of S2 beats and increasing the rate at which the S1S2 interval is reduced. Finally, it may be possible to average the surface QRS complex in response to an extrastimulus, although if there are small areas of myocardial involvement direct multielectrode intracardiac recordings may be necessary to detect local conduction disturbances.
Another concern is that primary VF patients, unlike the control patients, have been defibrillated, and their results may be due to recent defibrillation rather than a true electrophysiological disturbance. Our pattern of referral for defibrillator implantation makes study of these patients at long intervals after VF impractical. However, patients No. 7 and 9 were studied 6 months and 3 years after their arrests, and these patients do not show the least disturbance, as would be expected if the results were due to recent defibrillation. Also, our experience with HCM suggests that this is unlikely.26 Nine HCM patients with VF have been studied, and they cluster in a group with maximum electrogram duration and the longest S1S2 interval at which delay starts to increase. Four of these patients were studied within 3 weeks of defibrillation, whereas two were studied 9 months and 5 weeks, respectively, before VF, and three were studied at 18, 36, and 60 months after VF. This suggests that paced electrogram fractionation is relatively stable and not caused by defibrillation. A second concern is that the changes in activation are due to myocardial damage during prolonged cardiac arrest. This is extremely difficult to test experimentally in humans, and animal experiments with controlled periods of ischemia are necessary to establish the relation between fractionated electrograms and small degrees of ischemic damage.
Primary VF, by definition, occurs in the absence of recognized cardiac disease. However, several patients had minor abnormalities that suggest more widespread myocardial disease. One 16-year-old patient developed sustained common atrial flutter during the study, which is an unusual finding in the absence of congenital heart disease. One patient had ST-segment abnormalities on exercise that, in the absence of coronary artery disease and with a negative ergonovine test, may point to repolarization abnormalities despite a normal QTc interval. One patient had a minor left ventricular anterior wall motion abnormality in the absence of infarction, myocarditis, or recognizable cardiomyopathy. Finally, mitral valve prolapse, which was present in one patient without regurgitation, a click, or a small, hyperdynamic left ventricle, has an uncertain relation to ventricular arrhythmias27 28 and might have been related to VF in this patient. Nevertheless, this study shows that this group of patients with primary VF, irrespective of their primary and unidentified underlying diseases, has an abnormality of intraventricular activation that can be related to a reasonable model of reentrant arrhythmogenesis and that this abnormality can be exposed without induction of an arrhythmia.
The probability of the line separating the two groups of patients cannot be estimated conventionally because the data are not normally distributed and the two groups have different variances. The null hypothesis is that the points in Fig 6⇑ are randomly assigned to one of either the seven control patients or nine VF patients. Given these combinations, what is the probability of being able to construct a line that completely discriminates VF patients from control patients? The ability to construct this line can be established for any combination of patients by computing all positions of a line that passes through any two control patients and determining whether the other control patients lie on one side of that line. The lines that form the convex polygonal boundaries of the group fulfill this condition, provided a sample from the VF patients is not enclosed within this boundary. Any line that crosses this boundary cannot separate the two groups because control patients will lie on either side of it. This process is repeated for the VF patient group. Provided that the boundaries of the two groups do not intersect, a separating line can be constructed between them; if they intersect, a separating line that avoids both boundaries cannot be constructed. This condition is computed for every possible combination of exchange between the control patients and the VF patients. There are 16C7=11 440 possible combinations of the data, and only 68 of these combinations are separable into two distinct groups, making the probability of a random grouping being separable approximately .006.
Selected Abbreviations and Acronyms
|RVOT||=||RV outflow tract|
This work was supported by the UK Medical Research Council and The British Heart Foundation.
- Received June 14, 1994.
- Revision received April 17, 1995.
- Accepted May 25, 1995.
- Copyright © 1995 by American Heart Association
Saumarez RC, Camm AJ, Panagos A, Gill JS, Stewart JT, de Belder MA, Simpson IA, McKenna WJ. Ventricular fibrillation in hypertrophic cardiomyopathy is associated with increased fractionation of paced right ventricular electrograms. Circulation. 1992;86:467-474.
Spach MS, Dolber PC, Heidlage F. Influence of the passive anisotropic properties on directional differences in propagation following modification of the sodium conductance in human atrial muscle: a model of reentry based on anisotropic discontinuous propagation. Circ Res. 1988;62:811-832.
Wellens H, Brugada P, Stevenson WG. Programmed electrical stimulation of the heart in patients with life threatening arrhythmias: what is the significance of induced arrhythmias and what is the correct stimulation protocol? Circulation. 1985;72:1-7.
Saumarez RC, Slade AKB, McKenna WJ. Arrhythmias in hypertrophic cardiomyopathy. In: Podrid PJ, Kowey PR, eds. Cardiac Arrhythmia. Baltimore, Md: Williams & Wilkins; 1102-1110.
Saumarez RC. Electrophysiological investigation of hypertrophic cardiomyopathy: evidence that abnormal intraventricular conduction in HCM is associated with an increased risk of sudden death. Br Heart J. 1994;72(suppl):s19-s23.
Hoel PG. Introduction to Mathematical Statistics. New York, NY: John Wiley & Sons; 1971:181-186.
Avitall B, McKinnie J, Jazayeri M, Akhtar M, Anderson AJ, Tchou P. Induction of ventricular fibrillation versus monomorphic ventricular tachycardia during programmed stimulation. Circulation. 1992;85:1271-1278.
Spear JE, Richards DE, Blake GJ, Simson MB, Moore EN. The effect of premature stimulation of the His bundle on epicardial activation and body surface potentials in dogs susceptible to sustained ventricular tachyarrhythmias. Circulation. 1985;72:214-224.
Spach MS, Dolber PC. Relating extracellular potentials and their derivatives to anisotropic propagation at a microscopic level in human cardiac muscle: evidence for electrical uncoupling of side-to-side fiber connections with increasing age. Circ Res. 1986;58:356-371.
Spach MS, Dolber PC, Heidlage F. Interaction of inhomogeneities of repolarisation with anisotropic propagation in dog atria: a mechanism for both preventing and initiating reentry. Circ Res.. 1989;65:1612-1631.
Gardner P, Ursell PC, Fenoglio JJ, Wit AL. Electrophysiological and anatomical basis for fractionated electrograms recorded from healed myocardial infarcts. Circulation. 1985;72:591-611.
Dillon SM, Allesie MA, Ursell PC, Wit AL. Influences of anisotropic tissue structure on re-entrant circuits in the epicardial border zone of subacute canine infarcts. Circ Res. 1988;63:182-191.
Ursell PC, Gardiner PI, Albala A, Fenoglio JJ Jr, Wit AL. Structural and electrophysiological changes in the epicardial border zone of canine myocardial infarcts during infarct healing. Circ Res. 1985;56:436-451.
Peters NS, Green CR, Poole-Wilson PA, Severs NJ. Reduced content of connexin-43 gap junctions in ventricular myocardium from hypertrophied and ischemic human hearts. Circulation. 1993;88:864-875.
Luke RA, Saffitz JE. Remodelling of ventricular conduction pathways in healed canine infarct border zones. J Clin Invest. 1991;87:1594-1602.
Willich SN, Maclure M, Mittleman M, Arntz HR, Muller JE. Sudden cardiac death: support for a role of triggering in causation. Circulation. 1993;87:1442-1449.
Meyerburg RJ, Kessler KM, Mallon SM, Cox MM, De Marchena E, Internan A, Castellanos A. Life threatening arrhythmias in patients with silent myocardial ischaemia due to coronary artery spasm. N Engl J Med. 1992;236:1451-1455.
de Silva RA. Psychological stress, cardiac arrhythmias and sudden cardiac death. In: Podrid PJ, Kowey PR, eds. Cardiac Arrhythmia. Baltimore, Md: Williams & Wilkins; 1995:1282-1289.
Wever EFD, Hauer RNW, Oomen A, Peters RHJ, Bakker PFA, Robles de Mina EO. Unfavorable outcome in patients with primary electrical disease who survived an episode of ventricular fibrillation. Circulation. 1993;88:1021-1029.
Saumarez RC, Slade AKB, Grace AA, Sadoul N, Camm AJ, McKenna WJ. The significance of paced electrogram fractionation in hypertrophic cardiomyopathy: a prospective study. Circulation. 1995;91:2762-2768.
Kuo CS. Arrhythmias associated with mitral valve prolapse. In: Surawicz B, Pratap Reddy C, Prystowsky EN, eds. Tachycardias. Boston, Mass: Martinus Nijhoff; 1984:179-183.