Author + information
- Received November 23, 2005
- Revision received April 6, 2006
- Accepted April 11, 2006
- Published online August 1, 2006.
- Craig M. Pratt, MD, FACC⁎,⁎ (, )
- Hussein R. Al-Khalidi, PhD†,
- Jose M. Brum, MD†,
- Michael J. Holroyde, PhD†,
- Peter J. Schwartz, MD, FACC‡,
- Stephen R. Marcello, MD, FACC†,
- Martin Borggrefe, MD, FACC§,
- Paul Dorian, MD, FACC∥,
- A. John Camm, MD, FACC¶,
- Azimilide Trials Investigators
- ↵⁎Reprint requests and correspondence:
Dr. Craig M. Pratt, The Methodist DeBakey Heart Center, 6565 Fannin Street, F1001, Houston, Texas 77030.
Objectives The purpose of this study was to assess the incidence, temporal characteristics, and risk factors associated with azimilide-associated torsades de pointes (TdP) ventricular tachycardia.
Background Azimilide dihydrochloride is a class III antiarrhythmic drug possessing Ikr and Iks channel-blocking properties.
Methods Oral azimilide (75 to 125 mg/day) was taken by 5,375 patients in 19 clinical trials conducted at 775 international centers. Of 3,964 patients in double-blind studies, 1,427 had a history of atrial fibrillation or other supraventricular arrhythmia, 510 had an implantable cardioverter-defibrillator, and 2,027 were post-myocardial infarction patients with a left ventricular ejection fraction ≤35%.
Results The TdP occurred in 56 patients assigned to azimilide, was dose-related, and tended to occur earlier with an azimilide-loading regimen. Forty-three percent of TdP patients had a QT interval corrected by Bazett’s formula, for heart rate, (QTc) ≥500 ms at the time of or before the TdP occurrence. Significant risk factors using logistic regression were increasing age, female gender, diuretic use, and lack of aspirin use.
Conclusions Azimilide-associated TdP has characteristics and risk factors similar to other Ikr blockers. However, there is a distinctive temporal profile. The TdP events are not concentrated in the first week. The azimilide-associated TdP rate is 1% (95% confidence interval 0.78 to 1.35) and is not increased in patients with low left ventricular ejection fraction, even in women.
Azimilide dihydrochloride is a Vaughan-Williams class III antiarrhythmic drug that has been investigated in: 1) atrial fibrillation (AF) and other supraventricular arrhythmias (SVA); 2) adjunctive antiarrhythmic therapy in patients with implantable cardioverter-defibrillators (ICD); and 3) post-myocardial infarction (MI) patients. Azimilide has both Ikr and Iks channel-blocking properties (1). Azimilide prolongs the time to the first symptomatic recurrence of AF and atrial flutter (2,3) and decreases the frequency of ICD therapies (4,5). Azimilide did not have an adverse effect on mortality in post-MI patients with a left ventricular ejection fraction (LVEF) ≤35% (6). It has been speculated that torsades de pointes (TdP) and ventricular tachycardia (VT) occur with Ikr channel blockers in part as a consequence of the reverse rate-dependent action of these drugs, and the property of Iks channel block theoretically could be less likely proarrhythmic (7). On the other hand, combining Ikr and Iks blockade theoretically could be more likely to cause TdP compared with Ikr blockade alone because congenital malfunction of Iks channels is considered a risk factor for TdP on therapy with Ikr blockers (8).
Accordingly, we analyzed all clinical trials (published and unpublished) in the azimilide database to assess the prevalence and characteristics of azimilide-associated TdP. The purpose of this report is to describe the dose-related incidence, temporal characteristics, risk factors, and clinical consequences of azimilide-associated TdP.
This database describes 5,375 patients enrolled during the period 1994 to 2004 who received oral doses (75 to 125 mg/day) of azimilide in 19 clinical trials (10 double-blind and 9 open-label) conducted at 775 international clinical centers. Of these patients, 1,427 had a history of primarily paroxysmal or persistent AF or another SVA and were enrolled in double-blind, placebo-controlled SVA clinical trials, in which patients were administered azimilide using a 3-day, twice daily loading regimen of 150 to 250 mg/day followed by a daily maintenance regimen (75 to 125 mg/day) of half the loading dose. Of these, 947 placebo and 1,222 azimilide-treated patients continued in open-label SVA trials on daily azimilide (75 to 125 mg/day). Additionally, 510 patients with an ICD and a history of VT/ventricular fibrillation and 2,027 post-MI patients with an LVEF ≤35% were enrolled in 12-month, double-blind, placebo-controlled trials. In the ICD and post-MI trials, patients received daily azimilide doses without any loading regimen.
Thus, this analysis includes 3,964 patients assigned to azimilide (75 to 125 mg/day) with a mean exposure of 214 (median 223) days in placebo-controlled trials and 1,411 taking azimilide in open-label SVA (n = 1,263) and ICD (n = 148) trials. There were 6,445 patient-years of exposure to azimilide. Of the 3,964 patients in double-blind studies, 1,427 were in 7 SVA trials, 510 were in 2 ICD trials, and 2,027 post-MI patients were in the Azimilide Post-Infarct Survival Evaluation (ALIVE) trial (6). In the placebo-controlled trials, there were 1,498 (38%) patients who had their first dose initiation in-hospital, and the remainder were outpatients.
Among a total of 3,168 placebo-assigned patients, 2 cases of TdP (0.06%, 95% confidence interval [CI] 0.01 to 0.23) were documented. The focus of this article is on the 56 azimilide-assigned cases of TdP. A total of 31 azimilide-associated TdP cases occurred in placebo-controlled trials (n = 3,964), and 25 azimilide-associated TdP cases occurred in open-label extension trials (n = 1,411).
Continuous baseline characteristics are presented as mean ± SD and were compared using the Wilcoxon rank-sum test. All p values are two-sided. Categorical variables were analyzed using the Pearson chi-square test or Fisher exact test. The log-rank statistic was used to analyze time-to-TdP occurrence (9), and the Cochran-Armitage trend test (10) was used to compare the TdP rates per 100 patient-years of exposure. Univariate risk factor analyses for TdP occurrences were performed using the Fisher exact test. Multivariate logistic regression analysis was implemented using the risk factors for TdP identified in the univariate analyses. A stepwise procedure was performed to include risk factors in the model that met the 0.05 significance level as an entry criterion. A logistic regression model with selected significant risk factors was used to estimate the adjusted odds ratios, 95% CIs, and p values.
Adjudication of TdP events
The diagnosis of torsades de pointes was appropriate for episodes of polymorphic VT comprising at least 2 cycles of ventricular complexes with alternating electrical polarity and amplitude that were preceded by pauses and excessive QT interval corrected by Bazett’s formula, for heart rate, (QTc) prolongation. The QTc reported uses the Bazett formula (11). All arrhythmic events in the double-blind trials that were documented by a 12-lead electrocardiogram (ECG) or ECG rhythm strip were adjudicated by an event committee (EC). Other cases were reported by the site investigator (SI) for studies without EC (ALIVE and ICD open-label) and included even if ECG confirmation was not available. To be conservative, all 56 cases of TdP that were reported by the EC and/or by the SI are included in the analysis. Of these 56 cases, 24 of them (12 double-blind and 12 open-label) had concurrence by both EC and SI, 17 cases (8 double-blind and 9 open-label) reported by SI alone (in those trials in which no EC was present or a strip was not available to send to the EC) and 15 cases (11 double-blind and 4 open-label) by EC alone. If available and when feasible, ICD electrograms were a source of documentation for the diagnosis of TdP, if pauses and excessive QTc prolongation preceded any polymorphic VT event.
Demographics of patients
Azimilide-associated TdP occurred in 56 of 5,375 (1%, 95% CI 0.78 to 1.35) patients (median azimilide exposure = 366 days). The baseline comparison of TdP patients to those without TdP is presented in Table 1.The summary of all double-blind study experience is listed in Table 2.
Effect of loading azimilide in SVA patients
There was no significant difference in episodes of TdP that occurred in SVA trials (34 of 2,690) versus ICD/ventricular arrhythmia trials (22 of 2,685). Although no formal statistical test was conducted for the time course comparisons, the time courses of the azimilide-associated TdP in the double-blind SVA trials tended to occur earlier than in the open-label SVA (placebo roll-over patients) trials (Fig. 1A),possibly related to the azimilide loading regimen in the double-blind SVA trials. In patients administered daily azimilide doses without loading, the time courses for TdP cases were similar for the SVA open-label, ICD, and post-MI patient populations (Fig. 1B). However, in patients administered daily azimilide doses with a loading regimen, approximately two-thirds of the TdP cases were clustered early in the double-blind studies (Fig. 1C).
Effect of azimilide dose and patient characteristics on the occurrence of TdP
The incidence of azimilide-associated TdP is dose related (Table 3).For azimilide doses of 75, 100, and 125 mg, the TdP rate per 100 patient-years exposed was 0.28, 0.52, and 0.84, respectively (p = 0.001, Cochran-Armitage trend test) (10). In patients administered 75 mg/day without loading, the incidence of TdP in ICD patients, SVA (open-label) patients, and post-MI patients was 0.44, 0.14, and 0.39 per 100 patient-years, respectively. The increase in TdP rates with higher azimilide doses is more obvious in female than male patients (Fig. 2A).The relationship of age to TdP incidence is shown in Figure 2B. Women have a higher TdP incidence than men at all ages. For example, women older than 75 years have approximately double the risk of TdP compared with men in the same age group (4.0% of 250 vs. 2.2% of 318, p = 0.2253). Women ≥65 years tend to have double the risk of women <65 years (2.6% of 912 vs. 1.0% of 602; p = 0.0363). In the post-MI patient population with low LVEF (≤35%), the incidence of azimilide-associated TdP is very low (0.35%, 95% CI 0.14 to 0.71), even among women (0.7%, 95% CI 0.14 to 1.98).
Temporal pattern of TdP
The 56 azimilide-associated TdP events occurred at a median time of 69 days (range 2 to 2,155 days) after initial dosing, more than two-thirds (40 of 56) of which occurred more than 14 days after azimilide initiation. Only 7 of 56 (13%) of the TdP events occurred in the first 3 days of azimilide exposure, 4 in women and 3 in men. The time course of azimilide-associated TdP cases was not different for the 3 treatment indications (Fig. 1B).
The median time to TdP among placebo patients (n = 947) who started azimilide in the SVA open-label trials was 171 days (mean 277 days) after starting open-label trials as compared with patients in the double-blind AF trials (Fig. 1A), with a median of 6 days (mean 40 days). This difference was statistically significant with a log-rank p value = 0.003.
Relationship of QTc to TdP
Patients with a QTc ≥440 ms at baseline (before randomization) were excluded from entry into the azimilide trials. During the study, patients with a QTc ≥525 ms were withdrawn from the study. In addition, patients in the ICD trials who had a wide QRS (QRS > 120 ms) were excluded from entry or withdrawn if their JT interval corrected by Bazett’s formula, for heart rate, (JTc) >320 ms or JTc >400 ms, respectively. Baseline QTc of the 56 azimilide-associated TdP patients (mean 412 ms) was similar to the QTc of the remaining 5,319 patients (mean 411 ms); however, the mean maximal QTc intervals observed were significantly different (532 vs. 489 ms; p = 0.0013). The QTc outlier analysis is more revealing in that 43% of TdP patients had at least 1 ECG with a QTc ≥500 ms before TdP, compared with 22% of non-TdP patients (p = 0.0014). Furthermore, 27% of azimilide-assigned patients in whom TdP developed had at least 1 ECG with a QTc ≥550 ms, compared with 6% of patients without TdP (p < 0.0001). There were 276 (5%) patients (189 in double-blind and 87 in open-label studies) who were withdrawn because of QTc prolongation. In none of these patients did TdP develop.
There is a clear relationship between maximum QTc values >500 ms and the risk of a TdP developing. Figure 2C shows the difference in percentage of patients with and without TdP across several QTc cuts, and Figure 2D illustrates a 5-fold increase in the risk of TdP with a maximum QTc in the range of 580 ms or higher.
A univariate analysis of risk factors to the azimilide-associated TdP events is presented in Table 4.A multivariate analysis of TdP risk factors (excluding QTc because data were not available for 7 patients who experienced TdP) using a logistic regression model is presented in Table 5.The final model resulted in 4 significant risk factors: increasing age, female gender, diuretic use, and lack of aspirin use.
A total of 42 (75%) azimilide-associated TdP events occurred in-hospital or resulted in hospitalization. Of the 56 TdP patients, 4 died within 48 h (2 documented nonsustained cases and 2 investigator-reported “possible” TdP without ECG documentation). A total of 1,329 (25%) patients (246 [16%] women and 1,038 [28%] men) were taking statins and aspirin as concomitant medications. Among the 56 patients (30 women and 26 men) who experienced TdP, only 7 (13%) patients (1 woman and 6 men) were taking both a statin and aspirin as concomitant medications at the time of the event.
There were 56 cases of TdP in the 5,375 patients (1%, 95% CI 0.78 to 1.35) assigned to azimilide (75 to 125 mg). Although we acknowledging the heterogeneity issue in comparing 2 databases, it is instructive to compare the clinical features of azimilide-associated TdP with the published experience with other Ikr blockers (quinidine, sotalol, dofetilide). Azimilide-associated TdP is similar in profile to Ikr blockers (12–19) in that: 1) it occurs more frequently in women and in elderly patients; 2) it occurs more frequently with digitalis or diuretic use; 3) there is an increased TdP risk with increased dose and with the use of a loading regimen; and 4) there is an association between TdP and maximum QTc.
Azimilide-associated TdP differs from published reports of other Ikr blockers in that: 1) the time course of TdP cases in patients administered azimilide without a loading regimen is not concentrated early after azimilide initiation, with only 14% of such TdP cases occurring <10 days after initiation, whereas 41% of TdP cases occurred >180 days after initial azimilide assignment; 2) the overall azimilide-associated TdP rate of 1% is low, given that >50% of the patients receiving azimilide had an LVEF <35%; 3) patients with TdP were less likely to be taking aspirin; and 4) AF patients assigned to azimilide (most had a normal LVEF) had a low (1.3%, 95% CI 0.87 to 1.76) TdP rate, yet it was 4-fold higher than in a post-MI population with depressed LVEF (0.35%, 95% CI 0.14 to 0.71).
These general comparisons are primarily focused on sotalol, dofetilide, and quinidine; the first two because they are used clinically, the latter a historical comparison. Amiodarone has a very low prevalence of TdP in post-MI or congestive heart failure patients (20).
Like other Ikr blockers, azimilide-associated TdP is more common in women, especially in the AF population, in which women represented 37% of the patients but 71% of the azimilide-associated TdP events (13). That the time course of azimilide-associated TdP cases is similar in women and men is a unique analysis not previously reported.
In the ALIVE, the azimilide-associated TdP rate was 0.35% in a post-MI population with LVEF ≤35%. Sotalol-associated TdP in doses of 160 to 320 mg/day is reported to be more prevalent in patients with a lower LVEF (12,13). In contrast to azimilide, dose adjustment for progressive degrees of renal failure is required for both sotalol and dofetilide or the TdP rate increases. The time course of the appearance of TdP is reported to be front-loaded with sotalol and dofetilide as well as with quinidine-related TdP, occurring in the first week in approximately three-fourths of the cases (12,14–19,21). Noncardiac drugs may show sporadic torsades because of occasional drug interactions (e.g., terfenadine), thus not showing a front-loaded proarrhythmic profile (22). Temporal profiles of TdP may partially be explained by different degrees of early telemetry monitoring. Dofetilide trials required mandatory inpatient initiation, including 3 days of telemetry monitoring, facilitating early detection (14–15).
An additional observation of azimilide-associated TdP is that patients in the supraventricular arrhythmia population (double-blind), despite preserved LVEF, have a higher TdP risk compared with patients in the post-MI (17 of 1,427 vs. 7 of 2,027, odds ratio 3.5, 95% CI 1.44 to 8.41, p = 0.0056). One possible explanation is that AF studies required loading (23). Another probable explanation is that the sudden heart rate slowing associated with termination of paroxysms of AF by DC cardioversion in these studies (i.e., conversion to sinus rhythm) may increase the risk for TdP in this population (24). In the ALIVE study, there was no difference in all-cause mortality between azimilide and placebo (hazard ratio = 1.0, 95% CI 0.82 to 1.22). This finding is robust in that there were a total of 488 deaths in all of the azimilide studies with hazard ratio = 1.0 and 95% CI 0.83 to 1.19.
In the 2 trials with an ICD, all symptomatic or asymptomatic arrhythmic events detected by the devices that could be considered an episode of TdP were counted. The incidence of TdP in the ICD studies in which these patients are continuously monitored was 1.4% (95% CI 0.55 to 2.81). There were similar mortality rates in azimilide-treated (6.9%, 95% CI 5.7 to 7.6) and placebo-treated (6.8%, 95% CI 5.6 to 7.8) patients in all placebo-controlled studies (7.4% vs. 7.1%, respectively, in placebo-controlled studies without ICDs).
The mean QTc at baseline before initiation of treatment was not a predictor of TdP during treatment. This is not surprising because baseline QTc prolongation was an exclusion criterion. During treatment, 43% of patients having TdP had at least 1 measurement of QTc ≥500 ms, suggesting that routine ECG surveillance is useful. The standard cutoff points of QTc of 500 and 550 ms used in clinical practice and previous reports were significant predictors of azimilide-associated TdP risk, consistent with published data on Ikr blockers (7,13,25).
An additional observation from these analyses was the possible identification of 2 novel TdP risk factors. Admittedly, both may represent the play of chance. Patients not taking aspirin nor receiving statins had a significantly higher risk of TdP. Only 7 of 56 (13%) of patients with azimilide-associated TdP were taking both aspirin and statins, compared with 1,322 of 5,319 (25%) non-TdP patients. Although this may well be a chance finding, it may be related to the cardiovascular benefits of these drugs, such as the anti-inflammatory effects (26–29) or the antiplatelet effects of aspirin (30). An alternative (practical) explanation is that the use of aspirin was lower in patients with AF because of the use of warfarin.
The incidence of TdP in azimilide-treated patients is dose proportional, with a wide temporal distribution of TdP events. Significant risk factors for TdP included female gender, increasing age, diuretic use, and lack of aspirin use. The TdP rate is low (1%) even when the upper 95% confidence limit (1.35%) is used and is unique in that there was no increase azimilide-associated TdP risk in patients with a low LVEF. This is true in the context that all azimilide studies had exclusion criteria of baseline QTc ≥440 ms and patients were withdrawn when their QTc prolongation ≥525 ms during drug therapy. If all patients with QTc >525 ms were withdrawn, then the incidence of TdP could theoretically be reduced to 0.8% (95% CI 0.58 to 1.08). The predominance of early TdP cases seen with other Ikr blockers does not characterize azimilide-associated TdP, except when loading doses of azimilide are administered.
The authors thank Judith M. Pepin and Elizabeth Hankel for their assistance.
Drs. Pratt, Schwartz, Borggrefe, Dorian, and Camm have all served as consultants for the scientific and regulatory development of azimilide and served on steering committees including chairman and co-chairman for multiple azimilide clinical trials. Drs. Al-Khalidi, Brum, Holroyde, and Marcello are employees of Procter & Gamble Pharmaceuticals. All authors played major roles in the design, conduct, and analysis of the azimilide clinical trials. This study was supported by a Grant-in-Aid from the Health Care Research Center, Procter & Gamble Pharmaceuticals, Inc., Cincinnati, Ohio.
- Abbreviations and Acronyms
- atrial fibrillation
- Azimilide Post-Infarct Survival Evaluation
- confidence interval
- event committee
- implantable cardioverter-defibrillator
- JT interval corrected by Bazett’s formula, for heart rate
- left ventricular ejection fraction
- myocardial infarction
- QT interval corrected by Bazett’s formula, for heart rate
- site investigator
- supraventricular arrhythmias
- Received November 23, 2005.
- Revision received April 6, 2006.
- Accepted April 11, 2006.
- American College of Cardiology Foundation
- Salata J.J.,
- Brooks R.R.
- Connolly S.J.,
- Schnell D.J.,
- Page R.L.,
- et al.,
- Azimilide Supraventricular Arrhythmia Program Investigators
- Singer I.,
- Al-Khalidi H.,
- Niazi I.,
- et al.
- Dorian P.,
- Borggrefe M.,
- Al-Khalidi H.R.,
- et al.
- Camm A.J.,
- Pratt C.M.,
- Schwartz P.J.,
- et al.,
- AzimiLide post Infarct surVival Evaluation (ALIVE) Investigators
- Arking D.E.,
- Chugh S.S.,
- Chakravart A.,
- Spooner P.M.
- Collett D.
- Bazett H.C.
- Julian D.G.,
- Camm A.J.,
- Frangin G.,
- et al.,
- European Myocardial Infarct Amiodarone Trial Investigators
- Lehmann M.H.,
- Hardy S.,
- Archibald D.,
- Quart B.,
- MacNeil D.J.
- Nguyen P.T.,
- Scheinman M.M.,
- Seger J.
- Waldo A.L.,
- Pratt C.M.
- Kumagai K.,
- Nakashima H.,
- Saku K.
- Baker K.E.,
- Curtis M.J.