Systematic overview of clinical trials of antiarrhythmic drugs
BACKGROUND Arrhythmia is a cardiovascular disorder which can lead to several complications. Over the past decade the introduction of many new drugs has raised concerns about their questionable benefits and cost-effectiveness. Classification of antiarrhythmic drugs has not been fully resolved. Although numerous clinical trials have been conducted, the value of antiarrhythmic drugs in many indications remains controversial. Two meta-analyses of clinical trials addressing the indication of quinidine (Class I) for maintenance of sinus rhythm after cardioversion have suggested high efficacy rates but increased mortality relative to placebo. Several overviews which were conducted to evaluate the impact of antiarrhythmic therapy on improving survival post acute myocardial infarction, have defined a turning point in the management strategy from Class I to Class III drugs, particularly amiodarone and sotalol, due to the unfavourable mortality outcome with the former Class. MAJOR AIMS This thesis was conducted with three major aims: 1) To assess both qualitatively and quantitatively the benefits and risks associated with flecainide (Class Ic), amiodarone (Class III), and sotalol (Class III & II) in treatment of chronic atrial fibrillation, acute medical or surgical supraventricular arrhythmias, and life-threatening ventricular arrhythmias developing post acute myocardial infarction; 2) To produce an overall summary estimate of effectiveness and probabilities of incidence of adverse effects, which can be useful for subsequent incorporation in cost-effectiveness analysis; 3) To validate the usefulness of various therapeutic outcomes implemented by general treatment guidelines. OVERVIEW OF THESIS A meta-analysis was carried out to compare the efficacy and safety of three antiarrhythmic agents (flecainide, sotalol, and amiodarone) in maintaining sinus rhythm after cardioversion of chronic atrial fibrillation. 42 of 119 clinical trials retrieved satisfied the predefined inclusion criteria. Data from 17 amiodarone trials (5 randomised, and 12 uncontrolled), 8 sotalol trials (6 randomised, and 2 nonrandomised), and 19 flecainide trials (8 randomised, 4 nonrandomised controlled, and 6 uncontrolled) were pooled separately after testing for homogeneity of treatment effect across the trials. Although the pooled rate difference in proportion of patients remaining in sinus rhythm between amiodarone and placebo (2 trials) was statistically nonsignificant (RD3mon = 16.1 %, 95% CI = -29.7 to 61.7, P>0.05), the pooled effect compared to Class IA drugs (3 trials) demonstrated significant differences at all time intervals (RDs were 20.5%, 31 %, and 28.8% at 3, 6, and 12 months respectively). Aggregating sotalol efficacy data in randomised or nonrandomised controlled trials has yielded highly significant effect in favour of sotalol as compared to placebo and equal effect as compared to Class IA and Class IC at all time points. Furthermore, comparison of flecainide to placebo or Class IA has revealed a highly superior effect in favour of flecainide. The calculated summary statistics (ORpeto, ORMH, RD, and RR) for the incidence of mortality and pro arrhythmia in the full-exposure group in amiodarone and sotalol trials were not significant, affirming the safety of those two drugs. In flecainide placebo-controlled trials, the ORMH for mortality and proarrhythmia were 1.8 (95% CI, 1.2-2.7, P=0.002), and (95% CI, 4.23-10.6, P<0.00l) respectively, thus indicating low benefit-risk ratio for flecainide as compared to amiodarone. The validity of this meta-analysis was examined by assessment of publication bias using funnel-plots. A funnel-plot of the amiodarone clinical trials displayed the shape of an 'inverted funnel', thus suggesting an evidence of low retrieval bias. However, due to the small sample size identified (18 trials only), a firm conclusion with regard to absence of publication bias could not be drawn. Evolving strategies for management of newly occurring supraventricular arrhythmias were reviewed. A meta-analysis was undertaken to determine the most effective agent for prompt cardioversion to sinus rhythm. Flecainide efficacy relative to placebo was confirmed by pooling data from 5 placebo-controlled trials (OR3hrs, 7.2; 95% CI, 4.7 to 11.1; Z=8.9; and OR8hrs, 5.5; 95% CI, 3.6 to 8.4; Z=7.85). However, pooling the data from three amiodarone, placebo-controlled trials at 3 and 8 hour-intervals demonstrated a nonsignificant effect (OR3hrs, 1.3; 95% CI, 0.7-2.4; Z=0.85; and OR8hrs, 1.03; 95% CI, 0.6-1.8, Z=0.12). All individual odds ratios for intravenous sotalol compared to placebo were highly significant with pooled OR at 1 hour of 8.8 (95% CI, 4.7-16.5; Z=6.8). The effect sizes of the three agents on mean ventricular response rate was estimated for both converted and unconverted patients. Whilst the effect size of flecainide versus placebo was not statistically significant at any time point, those of sotalol and amiodarone were statistically and clinically meaningful for both converted and unconverted patients. It is suggested that for acute cardioversion, intravenous flecainide or sotalol should be initially implemented. Intravenous amiodarone can be subsequently introduced for controlling the ventricular rate in persistent unconverted patients. Recent meta-analyses of randomised controlled trials of secondary prevention of myocardial infarction by antiarrhythmic agents have questioned the validity of using arrhythmia suppression as a substitutive end point for mortality. A meta-analysis examining the effect of sotalol and amiodarone for prevention of death post acute myocardial infarction was undertaken. In addition to single point estimates of pooled odds ratios of total mortality and sudden death, a meta-analysis of survival data which included censored end points was employed. An attempt was made to reconstruct the life tables in individual trials of amiodarone. The Kaplan-Meier percentages were recalculated and pooled at specific time points to reproduce the final meta-analytic survival curves of total mortality and sudden death. The meta-analysis confirmed the clinical efficacy of amiodarone for prolonging the survival in patients with congestive heart failure or myocardial infarction. The nonparametric log-rank odds ratio method was applied to raw actuarial data deduced from published Kaplan-Meier graphs as well as data generated by curve fitting. Pooling each set of data separately has yielded highly significant log-rank ORs for total mortality in the first set of four trials with censoring (log-rank OR at 102 months, 0.598; 95% CI, 0.43 to 0.83; Z = -3). However, log-rank ORs from data generated by curve fitting of data from a further three trials, were nonsignificant up to 48 months (log-rank OR, 0.87; 95% CI, 0.72 to 1.06, Z = -1.4). Merging of the two data sets has suggested strong evidence of efficacy for improving survival in terms of both total mortality and sudden death.