|Home | About | Journals | Submit | Contact Us | Français|
Studies have shown the predictive value of inducible ventricular tachycardia and clinical arrhythmia in patients who have structural heart disease. We examined the possible predictive value of electrophysiologic study before the placement of an implantable cardioverter-defibrillator.
Our retrospective study group comprised 315 patients who had ventricular tachycardia that was inducible during electrophysiologic study and who had undergone at least 1 month of follow-up (247 men; mean age, 66.9 ± 13.5 yr; mean follow-up, 24.9 ± 14.8 mo). Recorded characteristics included induced ventricular tachycardia cycle length, atrio-His and His-ventricular electrograms, PR and QT intervals, QRS duration, and drug therapy.
Of the 315 patients, 97 experienced ventricular arrhythmia during the follow-up period, as registered by 184 of more than 400 interrogations. There were 187 episodes of ventricular arrhythmia (tachycardia, 178; fibrillation, 9) during 652.5 person-years of follow-up. Subjects with a cycle length ≥240 msec were more likely to have an earlier 1st arrhythmia than those with a cycle length <240 msec (P=0.032). A quarter of the subjects with a cycle length ≥240 msec had their 1st arrhythmia by 19.14 months, compared with 23.8 months for a quarter of the subjects with a cycle length <240 msec (P <0.032).
Among the electrophysiologic characteristics examined, inducible ventricular tachycardia with a cycle length ≥240 msec is predictive of appropriate implantable cardioverter-defibrillator therapy at an earlier time. This may have prognostic implications that warrant implantable cardioverter-defibrillator programming to enable appropriate antitachycardia pacing in this group of patients.
Implantable cardioverter-defibrillators (ICDs) have been extremely effective in limiting the incidence of fatal cardiac arrhythmias in patients who are predisposed to sudden cardiac death.1 In particular, patients with a previous myocardial infarction or a left ventricular ejection fraction (LVEF) below 0.30 have been shown to have significantly improved survival benefit from the prophylactic insertion of ICDs.2 Despite the marked improvement in outcomes in patients who receive these devices, there is still a risk of fatal tachyarrhythmias in this patient population.
Clinical predictors of appropriate ICD discharge have been widely described in the medical literature. Demographic and pharmacologic factors, including age and digitalis use, have been shown to increase the risk of ICD discharge in a wide distribution of patients.3 However, few studies to date have addressed themselves to the usefulness of specific electrophysiologic (EP) values gathered during the EP study (EPS) as predictors of subsequent spontaneous events. The Multicenter Automatic Defibrillator Implantation Trial II (MADIT-II) and Multicenter Unsustained Tachycardia Trial (MUSTT) enumerated factors that can be used as clinical indicators for therapeutic ICD firing.2-4 The MADIT-II investigators4 revealed that factors such as hospitalization for heart failure or coronary events, elevated blood urea nitrogen levels, and poor New York Heart Association functional class were reliable indicators of appropriate ICD firing. The MADIT-II trial also sought to enumerate EP predictors of subsequent ICD discharge, subsequent ventricular arrhythmias (VAs), and death. Specifically, in patients with inducible VA, the likelihood of future development of ventricular tachycardia (VT) triggering ICD discharge was higher.5
The MUSTT investigators2 tested whether an EPS was an effective predictor of risk of sudden cardiac death. It showed that those patients who were inducible for VT at the time of EPS had an increased survival benefit if an ICD was subsequently implanted. The ICDs were shown to reduce significantly the incidence of death attributed to arrhythmia (73%–76% reduction), as well as the incidence of total death (55%–60% reduction). Although EPSs were used to guide therapy, no values that were measured during EPS were specifically discussed.
The aims of our study were to determine the incidence of inducible VAs at the time of EPS in a patient population with reduced LVEF and to characterize the EP variables recorded from EPS as possible predictors of subsequent spontaneous ICD discharge.
All patients undergoing EPS before implantation of an ICD at North Shore University Hospital from 1998 through 2006 comprised the cohort included in the analysis. All patients were recipients of dual-chamber devices. Patients who did not have at least 1 ICD interrogation at follow-up and at least 1 month of follow-up were excluded. Clinical and EP variables were collected retrospectively from the chart review.
All patients gave written informed consent for EPS and underwent EPS in the fasting state. The initial indications for the EPS were the MADIT criteria and the MUSTT criteria. Electrode catheters were placed under fluoroscopic guidance in the high right atrium across the tricuspid valve (to record a His bundle electrogram) and in the right ventricular apex or outflow tract (for electrogram recording and programmed stimulation). Cardiac stimulation was performed with a constant-current programmable stimulator, and all patients underwent standard programmed electrical stimulation with up to 3 ventricular extrastimuli introduced during sinus rhythm and ventricular pacing. The endpoint of stimulation was either induction of VT or the achievement of a ventricular refractory state. Induction of ventricular fibrillation (VF) or polymorphic VT was considered a nonspecific finding. Only if monomorphic VT developed was a patient considered inducible.
The primary outcome variable was the time-to-occurrence of a VA (VA includes VT and VF and is referred to as the “event”). Demographic variables, clinical and medication history, and variables from the EPS before ICD implantation were evaluated separately as possible risk factors for VT or VF. These are enumerated below:
The Kaplan-Meier product-limit method was used to analyze the time-to-1st occurrence of an event; subjects were stratified according to whether the EPS cycle length was longer or shorter than 240 msec. In cases in which the endpoint of a VT or VF did not occur, the subject was considered “censored” at the time of last follow-up. The log-rank test was used to compare the 2 groups (EPS, ≥240 vs <240 msec). Whenever the median time-to-1st event was not reached, we present instead the 25th percentile of the cumulative distribution, along with the corresponding 95% confidence intervals. The Mann-Whitney test was used to compare continuous variables between groups.
The incidence rate of VA was estimated using standard methods in epidemiology (the total number of VAs divided by the total number of follow-up years) and compared using the incidence density ratio method, under the assumption that the likelihood of a VA in a patient remains constant over the entire follow-up period. The method uses a large-sample normal approximation.
All analyses were carried out using SAS version 9.1 software (SAS Institute, Inc.; Cary, NC); results of P <0.05 were considered statistically significant.
There were 315 subjects identified in the cohort (247 men; mean age, 66.9 ± 13.5 yr; mean follow-up, 24.9 ± 14.8 mo [range, 1.2–54.6 mo]). Of these, 97 experienced a VA during the follow-up period, as registered by 184 of more than 400 interogations. There were 187 episodes of VA (178 of VT and 9 of VF) during 652.5 person-years of follow-up. Baseline patient characteristics, including clinical, pharmacologic, and EP data, appear in Tables I and andIIII.
There was no significant difference between ischemic and nonischemic populations with respect to inducibility. Most patients (99%) in both groups were inducible (P >0.05). The PR was significantly longer in the slow-VT group (EPS cycle, ≥240 msec) than in the fast-VT group (EPS cycle, <240 msec). The mean PR was 183.4 ± 40.86 and 170.6 ± 40.72 in the slow- and fast-VT groups, respectively (P <0.005). Subjects with a cycle length ≥240 msec were more likely to have an earlier 1st VA than those with a cycle length <240 msec (P <0.032). A quarter of the patients with an EP cycle length ≥240 msec experienced their 1st event by 19.14 months, compared with 23.8 months for a quarter of the patients with an EP cycle length <240 msec (P <0.032) (Fig. 1). Of those with a cycle length ≥240 msec, 75% were on β-blocker therapy, 75% were on ACE-inhibitor therapy, and 47.8% were on antiarrhythmic therapy. This is comparable to those subjects with faster inducible VA, except for a lower rate of exposure to antiarrhythmic therapy, which, in both groups, consisted of either amiodarone or sotalol (34.2%; P=0.02; Table III).
The incidence density of a VT or VF in the EPS ≥240-msec group was significantly higher than in the EPS <240-msec group (121 events in 338.7 person-yr vs 61 in 306.4 person-yr; P <0.002).
In the present study, we found that subjects with inducible VT at a cycle length ≥240 msec were more likely to have an earlier 1st VA than those with inducible VT at faster cycle lengths. There were no significant differences with respect to β-blocker or ACE-inhibitor use. Patients in the slow-VT group, however, did have a higher rate of antiarrhythmic drug use than did those in the fast-VT group. The higher rate of antiarrhythmic pharmacotherapy use in this group might account for the shorter time to 1st VA, given the pro-arrhythmogenicity and ability to induce slow VT associated with these drugs.
Despite several studies indicating that EPS may, in some populations (particularly in patients with reduced LV function after myocardial infarction), be useful as an indicator of future risk of ventricular arrhythmias and subsequent ICD discharge, the precise worth of EPS in this regard remains uncertain. The MUSTT investigators demonstrated a statistically significant higher mortality rate among patients who had inducible ventricular arrhythmias on EPS.2
Nevertheless, as that study also showed, the negative predictive value of an EPS without inducible VT was quite low: the group with no documented VT on EPS still had a 12% death rate attributable to arrhythmia after 2 years of follow-up. As the MUSTT investigators acknowledged, several factors can negate the potentially beneficial predictive qualities of EPS—including the fact that the results of EPS can vary significantly from day to day and year to year. In addition, in order to be included in MUSTT, patients must have had coronary artery disease, decreased LV function, and a history of asymptomatic nonsustained VT. Such patients are more likely to develop new reentrant circuits, which could account for the higher-than-expected rate of VT even in those patients who displayed normal EPS at some point in time.2 As a result of the uncertainty regarding the usefulness of EPS in predicting benefit from preimplantation EPS, the MADIT-II substudy5 examined prophylactic ICD implantation (without EPS before implantation as a qualifier) in patients who had a higher risk of ventricular arrhythmias as a result of reduced LV function after myocardial infarction. Therefore, despite the wealth of information that can be obtained from EPS, the results of previous studies that examined the actual clinical benefit gained from using EPS to predict subsequent ICD therapy have been unimpressive. Clinical variables such as a diminished LVEF and a history of recurrent ventricular arrhythmias often suffice to determine which candidates will benefit from ICD implantation.
The growing number of clinical trials showing the clinical advantage of ICD therapy for the prevention of sudden cardiac death in patients with unstable VAs or heart failure has greatly increased the size of the patient population with these devices.1-6 In 2005, there were approximately 91,000 ICD implantations in the United States, compared with approximately 26,000 in 1998.7 In order to optimize device settings and thereby improve the quality of life and reduce morbidity and death in patients who have undergone ICD implantation, more studies are now examining clinical and EP variables that might be used to predict the risk of recurrence of malignant VAs. In addition, knowledge of potential predictors of ICD discharge may help stratify patients who may benefit from ICD implantation; the number of patients who undergo the procedure will certainly increase as implantation rates catch up with the rising number of indications.
Clinical variables such as age and LVEF have been shown to be reliable predictors of benefit from ICD therapy. In addition, EP variables such as a history of VF or VT are also used to predict which patients will benefit from ICD therapy. The use of more varied guidelines to determine ICD eligibility and to optimize antiarrhythmic therapy might possibly reduce the morbidity and death that are associated with the most common VAs. Although smaller studies8 of the potential usefulness of VT cycle length have not shown this easily quantifiable variable to be a reliable predictor of future VAs, our data indicate that patients who, during EPS, had inducible VT with longer cycle lengths (≥240 msec) were more likely to experience subsequent VAs and appropriate ICD discharges than were patients who had inducible VT with shorter cycle lengths.
The finding that a longer VT cycle length correlates with an increased risk of future ICD therapy may shed light on the mechanisms that give rise to ventricular arrhythmias, particularly in patients who have heart failure and LV dysfunction. Studies that examine VF cycle lengths in both human and animal models9-11 have shown that mean VF cycle length correlates relatively well with ventricular refractory states, particularly at significant rates of tachycardia, in which a certain degree of myocardial ischemia may contribute to prolongation of the cycle. The same case can probably be made for VT. Prolongation of cardiac repolarization in patients with LV dysfunction has been established and is the basis for the growing acknowledgment of the role of ventricular dyssynchrony and for the use of biventricular pacing.12-14 Long-term cardiac remodeling and the consequent alteration of calcium and potassium currents leads to prolonged action-potential durations and cardiac-myocyte refractory periods.14-16 Several in vivo animal and ex vivo human studies15,17–21 have indicated a correlation between LV systolic dysfunction and a prolonged ventricular refractory period.
The mechanisms that underlie the association between VF cycle length and prolonged ventricular refractoriness and action-potential duration are probably relevant to VT as well. Our results indicate that patients with longer cycle lengths are at greater risk of appropriate ICD therapy than are patients with shorter cycle lengths. The difference in average LVEF between these groups was not significant. Previous studies22-24 on the usefulness of inducible VT cycle length in post-myocardial infarction patients indicated that a shorter VT cycle length was less likely to positively correlate with future risk of VAs. In MADIT-II patients, Daubert and colleagues5 analyzed the effect of induced VT cycle length on the predictability of VA and found that, in common with our results, inducible VT with a cycle length of greater than 240 msec was associated with an increased likelihood of subsequent VT, thereby ascribing a positive predictive value to slow VT.
The clinical value of acknowledging all episodes of slow VT in patients with ICDs may be underappreciated. The multicenter prospective report by the Slow VT Study Group in 200523 indicated that the incidence of slow VT in ICD recipients is relatively high, approaching 30%. Nevertheless, the investigators concluded that the efficacy of antitachycardia pacing in terminating the arrhythmia rendered the clinical value of slow VT relatively low, particularly because most slow VT episodes are not life-threatening. Therefore, they concluded that reprogramming the device at a lower tachycardia-detection rate would be more deleterious than beneficial for these patients.25 This indeed has been the consensus regarding the programming of shock thresholds and tachycardia-detection rates, primarily due to the increased likelihood of inappropriate shocks at lower VT rates and to the physiologic and psychosocial ramifications of ICD discharge. Nevertheless, counting the frequency of episodes of slow VT may help indicate the risk of ventricular arrhythmia and therefore may be a clinically relevant tool.
In addition to its potential use as an indicator of VA recurrence, the ability of VT cycle lengths to predict the symptomatology associated with future ICD therapy for VA has been analyzed. A study by Freedberg and colleagues26 reported that, in addition to various clinical variables such as depressed EF, a VT cycle length below 250 msec in inducible VT during EPS was a reliable indicator of syncope and impaired consciousness during subsequent arrhythmic episodes. However, VT cycle length was not a statistically significant variable in predicting VA recurrence.26 More studies to analyze VT cycle-length patterns in patients with ICDs may provide yet another indicator of symptomatology and prognosis in patients who develop VAs.
Despite varied findings regarding the usefulness of VT cycle length as an indicator of risk of arrhythmia recurrence, the physiology surrounding longer cycle lengths—including longer action-potential duration and ventricular refractory states—should be investigated further. More large-scale studies examining the association between longer VT cycle lengths should be performed, because acknowledgment of a firm association (as in our study) might lead to alterations in antiarrhythmic pharmacotherapy and device settings, particularly in thresholds for antitachycardia pacing. Our data indicate that occurrences of slow VT with longer cycle lengths may be a more clinically relevant variable than was previously recognized.
Our data were collected retrospectively. Therefore, the results depend upon the accuracy of the medical records and clinical database. Some patients may have received ICD discharges after they were lost to follow-up, which could not be accounted for. In addition, our study encompassed the records of 315 patients. Because this number was not derived from a formal power calculation, it might have impeded some of the trends that we observed from reaching statistical significance.
Our study demonstrates the predictive value of EPS variables obtained before the implantation of an ICD. This certainly does not signify the necessity of EPS before ICD implantation, but only the potential usefulness of variables that might have been evaluated in patients who underwent EPS at some point before device implantation. Of the variables studied, a cycle length of ≥240 msec is associated with earlier time to 1st event and the resultant appropriate ICD discharge. Therefore, a cycle length of ≥240 msec on EPS was associated both with earlier events and with more events overall. This knowledge can be applied to optimize device programming by more accurately setting VA detection zones and antitachycardia pacing.
Address for reprints: Stuart Beldner, MD, North Shore University Hospital, 300 Community Dr., Manhasset, NY 11030
This manuscript is dedicated to the memory of Dr. Daniel R. Gold, who passed away prior to its publication. His wit, intellect, and presence will be missed by all.