Ventricular tachycardias above the initially programmed tachycardia detection interval in patients with implantable cardioverter-defibrillators

Incidence, prediction and significance

Definitions

The occurrence of a VT, detected by the ICD, was assumed, if stored tachycardia CLs were short (<250 ms), or decreased suddenly at the onset of the tachycardia and were stable, or the morphology of stored electrograms was different from the morphology during sinus rhythm or antitachycardia pacing successfully terminated the tachyarrhythmias.

A VT above TDI was assumed, if at least a one-lead electrocardiogram (ECG) documentation of the VT was available. Nonsustained VTs above the TDI were not included. Few VTs (9 of 61) above TDI were gathered from ICD memory data. In these cases, the VT was symptomatic and lasted long enough for the patient to be hospitalized because of the VT, but the VT accelerated before an external ECG documentation could be obtained. All ECGs of VTs above TDI were reviewed by at least two cardiologists. If there was doubt about the origin of the tachycardia, it was not classified as ventricular.

ICD programming

Programming practice has changed during the last decade due to technical and scientific progress. Coherence in programming strategy was achieved in our clinic by programming rules layed down in an ICD handbook, which the implanting physician had to follow.

We have always considered the spontaneous or inducible VT for ICD programming and added a safety margin between 30 and 60 ms to the slowest sustained VT. In case no VT was documented or induced, the programming strategy has changed during recent years. Until 1994, we used a one-zone shock-only programming in these patients, if anamnestic details such as syncope did not suggest monomorphic VTs. The TDI was programmed empirically between 180 and 220 beats/min depending on the patient’s left ventricular ejection fraction (EF). In recent years, we have always programmed two zones with a VT zone starting at 170 to 190 beats/min and a VF zone between 200 and 220 beats/min if no VT was documented, because several trials suggest that patients with structural heart disease tend to have VF initiated by VTs, even if no VTs could be documented (18).

Patients

Baseline characteristics of 659 consecutive patients who received ICDs at our Institution (University of Münster) between July 1988 and August 1998 were prospectively collected in an ICD database. Indications for ICD implantation were survived cardiac arrest or polymorphic VT in 273 patients (41.4%), monomorphic VT in 188 (28.5%), monomorphic VT and VF in 141 (21.4%), nonsustained VT in 24 (3.6%) and syncope plus inducible VTs in 34 (5.2%; Table 1). Prophylactic indications for ICD implantation were excluded from analysis.

All patients underwent coronary angiography and programmed ventricular stimulation (PVS) off antiarrhythmic therapy or on amiodarone before ICD implantation, at discharge and after any change of antiarrhythmic medication. In case of amiodorone control, stimulation was performed after a loading phase of 10 to 15 g of amiodarone. Electrophysiological study before and after ICD implantation was performed according to the same protocol. Programmed ventricular stimulation was performed at three basic CLs with up to three extra stimuli until the protocol was finished or VF or very fast VT was induced, which required cardioversion (19). In 299 patients (45.4%), a single detection zone with shock therapy, and in 360 (54.6%), at least two detection zones and antitachycardia pacing were programmed. The mean TDI was 376 ± 55 beats/min.

Variables for risk prediction of VTs, shocks and VTs below the detection rate were prospectively collected in an ICD database and included patients’ characteristics such as heart disease, New York Heart Association (NYHA) class at the time of implantation, history of decompensation, previous antiarrhythmic interventions, results of left and right heart catheter, spontaneous VTs, results of PVS, baseline electrocardiograms at implantation as well as hospital discharge data and antiarrhythmic medication at discharge and during follow-up.

Follow-up

Patients visited the ICD outpatient clinic routinely every three months and were encouraged to schedule additional visits, if the first shock, clusters of shocks or syncope occurred. They were questioned about extraordinary events such as extra hospital admissions for cardiac events, and medical reports of these events were obtained. If patients died, we routinely performed a postmortem interrogation of the device. No patient has been lost to follow-up. However, patients who were no longer followed at our or an affiliated clinic and patients who underwent ICD explantation for various reasons were censored at the last visit at our or an affiliated clinic. Chart reviews and printouts were checked for VTs above the programmed TDI, causes or permissive factors and measures taken to avoid recurrences. Mean follow-up time was 31 ± 26 months.

Statistical analysis

Mean ± standard deviation was used for continuous variables with normal distribution and t test was applied to compare means. Median and range or 95% confidence interval was used to describe continuous variables without normal distribution and Mann Whitney U test was applied to compare ranks. Frequency distributions between groups of patients were tested with the chi square test. A two-tailed p value of equal or less than 0.05 was regarded as significant. The probability of major events was calculated according to the Kaplan-Meier method and measured from the date of implantation, first VT or first VT above TDI to the event (20). Differences between pairs of actuarial curves were tested by Mantel-Haenszel log-rank test. Cox regression analysis was performed on patients’ baseline characteristics to investigate the influence of different variables. For continuous variables, the hazard for an increase of one in the value of the covariate was calculated from the regression coefficient b: exp(b). For binary variables, the hazard ratio for the presence of a certain characteristic was calculated as exp(b), where b is the regression coefficient. Statistical analysis was done with SPSS (Statistical Package for Social Sciences, version 6.0) for Windows.

Incidence of tachycardias below the detection rate

Patients’ baseline characteristics are presented in Table 1. Three hundred seventy-seven patients (57.2%) had a recurrent VT or VF. The per-year risk to suffer a first VT decreased during the first four years and was 59.2%, 23.9%, 13.7% and 8.3% in the first, second, third and fourth year after ICD implantation, respectively. Forty-seven patients (7.1%) suffered 61 VTs below the detection rate. Forty-eight VTs above TDI (78.8%) were documented with 12-channel ECGs, four (6.6%) during holter monitoring and nine (14.8%) could be gathered from ICD memory data. In these patients, tachycardia lasted long enough that hospital admission was initiated, but the VTs accelerated into the programmed detection zone before an ECG could be obtained. The CLs of VTs above TDI were 441 ± 60 ms. The difference between the VT above TDI and the slowest VT, spontaneous or induced, before ICD implantation was 108 ± 58 ms. The difference between the initially programmed TDI and the first VT above TDI was 39 ± 34 ms.

The risk of a VT above TDI was about 3% (range 2.7% to 3.5% per year) and remained fairly constant during the first four years after ICD implantation (Table 2, Fig. 1). The relative risk for a first VT being a VT above TDI, therefore, increased with time (see risk of any first VT), and was 5.1%, 11.3%, 25.5% and 32.5% during the first, second, third and fourth year after ICD implantation in 281, 62, 20 and 12 patients with first VTs, respectively.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Risk of VTs above the programmed tachycardia detection interval.

In NYHA class I, the risk of a VT below the detection rate was below or equal to 1% per year during the first 4 years, whereas the risk in NYHA class II and III ranged between 3.2% and 4.8% (p = 0.021; Table 2). Patients with EFs above 0.40 had a risk between 0.9% and 2.7% per year during the first four years, as opposed to 3.3% to 6.2% in patients with an EF below 0.40 (p = 0.027; Table 2).

The risk of VTs above TDI ranged between 4.6% and 5.9% per year during the first four years in patients with spontaneous monomorphic VTs, in contrast to 0.9% to 1.4% in patients without monomorphic VTs (Table 2; Fig. 2). Patients without inducible monomorphic VTs, patients with inducible VF and polymorphic VTs included, had a significantly lower risk to suffer a VT above TDI during the first four years (0% to 1.3% per year during the first four years) than patients with inducible monomorphic VTs (3.7% to 5.3% per year, p < 0.001; Table 2, Fig. 3).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Risk of VTs above the programmed tachycardia detection interval and spontaneous ventricular tachycardias before ICD implantation. pres. mVT = presenting monomorphic ventricular tachycardia.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Risk of VTs above the programmed tachycardia detection interval and inducible tachycardias. ind. MVT = inducible monomorphic VT; VF = ventricular fibrillation; VT = ventricular tachycardia.

Of interest, there was no significant difference of the induced VT CL between patients with (299 ± 65 ms) and without VTs (294 ± 84 ms) above TDI. However, patients with VTs above TDI showed a tendency for an increase of CL from initial PVS to discharge PVS by 24 ± 94 ms, whereas patients without VTs above TDI rather showed a decrease in induced VT CL by 20 ± 71 ms (p = 0.020, Mann Whitney U test).

The risk of experiencing a VT above TDI was most prominently related to the use of class III antiarrhythmic drugs. Although all patients underwent PVS after any change in antiarrhythmic medication and ICDs were programmed according to the results (in the case of amiodarone PVS was performed after a loading phase), the risk of a VT above TDI was 25% during the first year in patients on amiodarone (p < 0.001 in comparison with patients without antiarrhythmic drugs), and ranged between 4.7% and 11.1% per year during the first four years after prescription of either d- or d,l-sotalol (p = 0.004 in comparison with patients on no antiarrhythmic drugs (Table 2; Fig. 4).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Risk of VTs above the initially programmed tachycardia detection interval and presence of risk predictors (inducible monomorphic VTs, NYHA class II + III, amiodarone).

Cox regression analysis revealed that amiodarone at discharge increased the risk of a VT above TDI 1.7-fold (p = 0.022); patients with inducible monomorphic VTs had a 2.7-fold increased risk and patients in NYHA class III had a 1.5-fold increased risk to suffer a VT above the TDI (Table 3). As expected, of these variables, only inducible VTs (relative risk 1.5, p < 0.01) and prior heart failure also predicted the recurrence of any VT.

However, the relative risk of a slow VT in patients with recurrent VTs increased with the presence of risk predictors: 2 of 50 patients with recurrent VTs and no risk predictor suffered a slow VT during follow-up (4%), 14 of 138 with one risk predictor (10%), 22 of 151 with two risk predictors (14%) and 8 of 36 with three risk predictors (22%) (Fig. 4).

Clinical significance

Only 7 VTs above TDI were asymptomatic and were detected by chance in routine ECGs or Holter ECGs during follow-up visits; 54 (88.5%) caused clinical symptoms and admission to the hospital, 39 (63.9%) caused palpitations or angina, 4 (6.6%) were associated with cardiac decompensation, 5 (8.2%) caused syncope and during 6 VTs (9.8%) above TDI patients had to be resuscitated.

Ventricular tachycardias above TDI that caused no symptoms and palpitation or angina were similar in CL (447 ± 46.8 ms, range 400 to 520 ms, vs. 445 ± 61 ms, range 330 to 550 ms). Ventricular tachycardias that were associated with cardiac decompensation (CL 505 ± 72 ms, range 330 to 550 ms) tended to be slower (NS). Ventricular tachycardias that caused syncope (358 ± 48 ms, range 340 to 460 ms) or cardiopulmonary resuscitation (CL 407 ± 36 ms, range 350 to 450 ms) tended to be faster than asymptomatic VTs (NS). All but one patient were successfully resuscitated, however, four died within one week of resuscitation, three of end-stage heart failure and one with intractable incessant VT.

Sixteen VTs above TDI (26.2%) terminated spontaneously, and nine VTs accelerated into the programmed detection zone on hospital admission and thus required no additional intervention. Thirty-six VTs above TDI required external intervention: in 14 cases (23%), the detection rate was decreased and the ICD terminated the VT automatically, in two cases (3.3%), VTs were terminated with manual overdrive stimulation via the ICD, 9 VTs (14.8%) were terminated with intravenous antiarrhythmic drugs and 11 (18%) with external overdrive stimulation after intracardiac placement of a stimulation catheter or external emergency cardioversion.

Intervention to prevent recurrence

After all but three VTs were above the TDI, the TDI was adjusted using a safety margin of at least 30 ms. In three patients, the TDI was not increased adequately, because the VT CL ranged between 500 and 600 ms and was well tolerated (n = 2) or this VT was successfully treated with radiofrequency ablation (n = 1). In one patient without inducible or documented VTs before ICD implantation, a one-zone shock-only device had to be replaced by a multi-zone device to allow for antitachycardia pacing.

Despite an adjusted TDI, 9 of 47 patients had at least one recurrent VT above TDI, 3 patients had at least two and 2 patients had three recurrences. The risk of recurrent VTs above TDI was 11.8% during the first, 12.5% during the second and 26.6% during the third year after first VT above TDI. In all cases, a further increase in VT CL caused the second VT above TDI. The CL of the first VT above TDI was 424 ± 52 ms and the CL of the second VT was 487 ± 39 ms (p < 0.002).

If we had intended to address any VT of 50% of patients with slow VTs, TDI would have to be increased to 25 to 27 ms, for 80%, TDI would have to be increased to 50 ms and for 90%, by 90 to 100 ms.

Patients without risk predictors

The risk of VTs above TDI in patients after survived cardiac arrest without spontaneous or inducible monomorphic VTs and preserved left ventricular function was below 1% per annum. Thus, the current practice to program a TDI of 280 to 330 ms for the detection interval in these patients seems to be adequate. The risk of a monomorphic VT during follow-up has been reported to be 18% in these patients during a follow-up of 14 months (18). This favors a two-zone device with the possibility of antitachycardia pacing even in patients without documented VTs.

Patients with risk predictors of VTs above the TDI

Patients with single risk predictors presented with a 5% risk to have a VT above the TDI during the first two years. This risk is almost tripled in patients with all three risk predictors and reaches 30% after four years.

The risk of a VT above TDI was significantly increased in patients with spontaneous (4.6% to 5.9% per year) or inducible (3.7% to 5.3% per year) monomorphic VTs. This finding was most surprising because we expected that in patients in whom we had information about spontaneous tachycardias, the actual risk should have been lower. Of interest, those patients with VTs above TDI did show a trend to slower VTs during predischarge PVS. The opposite was true for patients with VTs below the TDI.

VTs above TDI and heart failure

Most VTs above TDI caused clinical symptoms and hospital admission (88.5%); 24.6% of VTs above TDI were even associated with devastating events such as heart failure, syncope or cardiopulmonary resuscitation.

The association between cardiac decompensation and slowing of VTs may first be explained by a progressive scarring of the myocardium, which again may cause slowing of conduction (21–23).

Second, slow VTs caused cardiac arrest and heart failure in some patients, although VTs were very slow (407 ms for cardiac arrest, 505 ms for heart failure), because patients with impaired left ventricular function often do not feel any palpitations and VTs are not documented until these patients are hospitalized for other symptoms such as angina and heart failure. The fact that heart failure was reversible in all but three patients favors this view.

Antiarrhythmic medication

Although all patients underwent PVS in our clinic, if antiarrhythmic therapy was changed (in case of amiodarone after a loading phase), to detect an increase in defibrillation threshold and a slowing of VTs above the TDI, the risk of VTs above the TDI was significantly increased on amiodarone (25% during the first year after initiation of therapy) or sotalol (4.7% to 11.1% per year during the first four years) (24). This suggests an increase of TDI after the initiation of antiarrythmic therapy irrespective of EPS results.

Furthermore, the initiation of antiarryhthmic therapy should be restricted to otherwise untractable patients with many episodes that require shocks, because we may just replace the problem of frequent tachycardias with the problem of VTs above the TDI with significant clinical impact. With the favorable results of radiofrequency ablation both in patients with coronary artery disease, reported by Strickberger et al. (25) and in patients with dilated cardiomyopathy, reported by Kottkamp et al. (22), radiofrequency ablation should always be considered first line therapy.

Which TDI for which patient

Only 50% of the VTs below the detection rate could have been detected if the detection interval had been increased by 25 ms beyond the usual 30 to 60 ms (7). Ninety percent detection would have required an increase in detection interval by 87 ms, and a 95% detection would have made an increase of the TDI by 123 ms necessary. The need for a much larger safety margin in some patients is in conflict with the fact that an increase in TDI is strongly associated with inappropriate episodes. Our report on the same cohort, published elsewere, revealed that an increase of TDI by 10 beats/min increased the risk of inappropriate episodes by 20% (26).

The clinical effects of the VTs above the TDI may suggest an increase in TDI in patients who are prone to highly symptomatic slow VTs. However, it is not yet clear whether such slow VTs can be addressed by ICD therapy at all with an acceptable risk of inadequate therapies and whether this translates into a survival benefit.

Another solution to the dilemma between a mandatory decrease of detection rate in patients with presenting or inducible monomorphic VTs or on antiarrhythmic therapy and the increasing risk of inappropriate episodes may be detection algorithms, which allow for more independence from detection rate. Such algorithms may include atrioventricular dissociation (12–14) or QRS width (10,11) as first-line detection criterion, especially in patients prone to very slow VTs.

Conclusions and outlook

If the TDI of an ICD is programmed as the CL of the slowest spontaneous or inducible VT plus 30 to 60 ms, the risk of a VT below the detection rate was about 3% per year. The risk of a recurrent VT above the TDI ranged between 11% and 13% per year during the first two years after the first VT above the TDI, although the TDI was adapted to the slow VT.

The clinical effects of VTs above the TDI may suggests an increase in TDI in selected patients. However, an increase in TDI increases the risk of inappropriate therapies. The questions whether slow VTs should be addressed by ICDs at all and whether this is feasible with current ICD technology have to be answered by a prospective trial, which is underway: the “1+1−Trial.”

Study limitations

A retrospective analysis of VTs above TDI is naturally based on reports of clinical emergencies. Therefore, the number of VTs and the risk of VTs above TDI may have been underestimated. Additionally, some sudden deaths despite ICDs might be due to VTs above TDI, as some of our patients with VTs above TDI had to be resuscitated. Grubman et al. (27) reported that in 46% of patients dying suddenly despite ICDs, no VT had been recorded.

Supraventricular tachycardias with bundle-branch-block morphology may have been erroneously taken for VTs. However, ECGs were reviewed with care by at least two cardiologists. If there were doubts about the ventricular origin of the tachyarrhythmia, the episode was not supposed to be of ventricular origin. Furthermore, atrial fibrillation and sinus tachycardias during treadmill ECG, which we have reported to be the strongest predictors and major causes of inappropriate episodes, were not predictors of an increased risk of VTs above TDI (Table 1). Therefore, we assume that differentiation of VTs from other kinds of episodes was sufficiently achieved (24).

Finally, all generations of ICDs were included in the study. With current programming options, the detection rate may be programmed less restrictively and the risk of VTs above the TDI may be lower.