COOPERATIVE
STUDIES
Events in the Cardiac Arrhythmia Suppression Trial: Baseline Predictors of Mortality in Placebo-Treated Patients ROBERT THERESE ROBEKT
I. CAPONE,
MD, FACC,*II
S. GERACI, C. SCHLANT,
YUDl
RN,§ KATHRYN
PAWITAN, HANDSHAW,
MD, FACC, ** ALBERT
PHD,t
NABIL
EL-SHERIF,
MD, FACC,B
RN,11 JOEL MORGANROTH,
L. WALDO,
MD, FACCtt
MD. FACC,n
AND THE
CAST INVESTIGATORSZ+
PatientE randomized to placebo in the enwinidr and Recainide arms of ,i,e Cardtae Arrbytbmts Supprerslan Trlnl (CAST) have beenfound to haveB relativelylow l-year mmtality rale of 3.9% in comparisontitk previousstudiesof palitnls in the postmyecardial infarction pried. To determine the comparability ol CAST with preview stodios, bs&ne vsrisbles WIR examined in the 741 pntianlr randomieed to pbwebo in the Secabide and rncainide arms of CAST. Twenty4brco basrllne cbaractrristtu
asmisted
The Cardiac Arrhythmia Suppression Trial (CAST) is a large, prospective. double-blind, placebo-controlled study undertaken to test the hypothesis that pharmacologic suppression of asymptomatic or mildly symmomatic ventricular bremature depolarizations will r&i id a reduction of the risk of arrhythmic death after myocardial infarction. Unlike other trials of anliarrhythmic therapy in the postmyocardial
infarction period, CAST was designed to mimic clinical practice by including an open label titration phase, during which a patient was randomly assigned to one of four possible titration sequences, utilizing a maximum of three drugs at two doseseach to identify one that was associated with 280% suppression of ventricular premature depolaizalions and 8% suppression of nonsustained ventricular tachycardia. If such adrug and dose combination was found. the patient was randomized to treatment with the active aeent or a matching placebo. This element of the CAST design seems to haie’had a significanl e&t on the study patients in that the estimated l-year mortality rate of 3.9% for the placebo group of the encainide and Aecainide arms of the trial appears to be relatively low compared with that of previous studies of patients recovering fmm myocardial infarction. In this study, we investigated the associalion between the baseline variables and atrhythmic death, total mortality and congestive heart failure to determine the comparability of the CAST placebo group wilh those described in previous postmyocardial infarction studies. Because these patients are not treated wirh antiarrhythmic agents, this investigation also provides an impartant natural history study of patients
with an index Q wavemyocardtt l&&on, Idstoryof heart failure, UP (I( d&it&, dbbeta and pr&mg& QRS dwation. T-1 mortality or rwusritakd cardiac arre(t WBT s&nilscantI, awinted with M index Q wave myofsrdlal infarction, diabetes,ST ugnent depressloo, high bndtonal drcJ, protonged QRS duration zmd low e&don fractloo. llw vsriabks at beeline sssorlstedwith martality Imm all uusa or urhythnic death or remtitated cardiac merestsad hear, f.8”~ in the CAST placrbwtrutcd patientsM stmttu to those idpntlded in previous pc&,yoc.rdipI infwctio” st”dt~l. Thus, the obrervntlonof lmrcnsPd mwtallty in CAST awociated with tbendmlnbtratlon atenmlnldtaud fkcalnldtBrsupprv&on ofvenlikular prematuredepdariuIians is probably sp~ikabk to any wmpnrrbly definedgroup of patimb In the wnimyccudW infarction period.
in the myocardial infarction period who have supprec$hle ventricular premature depolarizations.
Methods
A
Study design. detailed summary of CAbT study design has been previously published (IL In brief. all patients had had an acute myocardial infarction within the previous 2 years and had an average of six or more ventricular premature depolarizations per hour on 24-h ambulatory electrocardiwraohy (ECG) recordina. A left ventricular eiection fraclion of ~0.55 was required for those whose-myocardlal infarction had occurred within 90 days; those whose infarclion had occurred HO days but ?lOmgldl. There were no patients in funcrional class IV and a limlted number in class III; extrapolation of the results beyond class II must therefore be made wth c&on. Exerare ierting was not a requirement for entry into the study and IS not mcluded
among the harelms variables evaluated. The average follow-up durauon kas IO months. End points for analysis. These included: II arrhythmic death or resusc,mwd cardmc arrect, the man CAST end pmnr: 2) lolai monality or resuscilated cardiac arrest. a secondarv CAST end pomt: and 3) new or worsening congestwe heart fallwe. &rhythmic denrk was defined as mstantaneous death m the absence of severe congzstive heart failure or shock. unwtnessed death without a preceding change in symptoms for whrch no other cwse could be ascribed or documented cardiac west. Each death and resuscitated cardiac arrest was reviewed bv the Events Commmee. whxh is composed of CAST &stigaors who did not know the treatment assignment of each SubJect evaluated. NW congestive heon foilare wasdefined as the appearance of ar least two of the following symptoms or s,gns when none ,\as noted at baseline study: shortness of breath. easy f&ability. edema. orthopnea. paroxysma: nocturnal dvpnca. jugular venous distension. rales and a third heart sound t&L \Vorsrr!hg congestive hrnn fnilrw wac defined ac an tocrease m one or more functional class. The stud” orotocot was aooroved bv rhe lnstilut,onal Review Board of each participating clinic-t center and by an mdependendy appointed external study Data and Safely !donaorinx Board. Wnttcn informed consent was obtained from all oatients oarticioatine in the studv. Statis&l methods. iVe &lyzed tw&way contingency trebles for the univ.&ate analysis using the standard chisquare test All p values are two-sided and declared signifimot If 80.05. Multivariate analyses of all end points were done wth use of the Car proponional hazard regression model. The exposure time began at the ttme of randomirabon to blinded therapy: patients who were still alive on April I& 1489. rhe dare on which rhe study Data and Safety Moatoring Board recommended that Ihe encainide and Recainide arms of CAST be terminated. had exposure time censored at that point. The exposure time for congestive heart failure was censored at the time of death or cardiac arrest or on Awl 18. 1989. and only the event was considered in the ana!ysis. In the multi&ate analysis. the variables were entered into rhe model by a stepwise procedure until none of the remaining variables outside of the model was agnificant.
1
1.
I
Results A to!al of 743 patients were assigned IO placebo therapy on encamide or flecainide between June IS, 1987 and April 17. 1989. The average follow-up was about IO months. Results of the univariate analyses of the 23 baseline factors are shown on Table I. Results of the multivariate analysis are summarized m Table 2. New or worseningcongestive heart failure. There were 51 subjects who experienced new or worsening congestive heart fadure. On univariate analysis. II variables were significantly associated with this outcome. with I (namely.
histary of previous myocardial infarction) showing a trend toward significance. On mulrivariate analysis, diuretic use (hazard ratio 2.21. diabetes (hazard ratio 1.7). high functional class (hazard ratio I.‘l/class). age (hazard ratio I.Ustandard deviation [SD] of 9.8 years), pr&ged QRS complex duration (hazard ratio I.USD of 19 msl and low left ventricular ejection fraction (hazard ratio I .7/SD of 0.09) were associated with new or worsened congestive heart failure during f0ll0w-up. Arrhythmic death or resuscitated cardiac arrest. Sixteen subjects experienced arrhythmic death or were resuscitated from cardiac arrest. an occurrence rate that would tend to limit the statntical wwer of the ccmmarisons. On univariate analysis, six base& variables were’significantly associated with this end point. with one (namely, Q wave myocardial infarction1 showing a trend toward significance. On multivariate analysis, Q wave myocardial infarction (hazard ratio 7.51, history of congestive heart failure (hazard ratio 3.81, use of digitalis (hazard ratio 3. I), diabetes (hazard ratio 2.6) and prolonged QRS complex duration (hazard ratio IS/SD of 19 ms) were assuciated with sudden arrhythmic death or resuscitated cardiac arrest. Mortality from all causes or resuscitated cardiac arrest. There were 26 deaths in this group. of which 21 were cardiac. On univariate analysis. IO baseline variables were significantly associated with this end pant. On multivariate analysis. Q wave myocardial infarction (hazard ratio 7).
diabetes (hazard ratio 4.71, ST segment depression @ward ratio 2.4), high functional class (hazard ratio I.J/class), prolonged QRS complex duration (bawd ratio U/SD of 19 ms) and low ejection fraction (hazard ratio I.4/SDof 0.09) were associated with total mortality. Discussion Mortalily in lhr varlats CAST study gnupa. We (I) ereviously demonstrated that the mortality rate for Datients with qua&ins ventricular ectapic rhythm enrolled into CAST is comparable with that of other groups studied after acute myocsrdial infarction. This rate is in contrast to the mortality rate for patients in CAST randomized to plsceba after demonstrating suppression of their ventricular arrhythmia during apen label titration. The dieience in mortality rate. comparing those with qualifying witnessed ectopic rhythm with those randomized to placebo, may relate to the modifications in the composition of the originally qualified group of patients during the open label drug titration period (Fig. 1). Initially. all qualified patients in CAST were sub jetted to a sequential scheme of antiarrhythmic treatment (2). Only those with ventricular ectopic rhythm that was suppressed by one of the agents included in CAST were randomized. During this open label titration period, death occurred in 63 (2.7%) of patients eligible for study (I patient was successfully resuscitated from cardiac arrest and subse-
quently randomized). Deaths were related to factors at baseline study associated with increased risk: previous myocardial infarction, more extensive left ventricular dysfunction and a prolonged interval from the index myacardial infarction. addition to 78 patients who were in the open label titration Dhase on April IS. 1990 at the Iermination of the encabtid; and Recainide arms of the study. there were 318 patients not randomized (because of adverse effects. lack of ventricular premature depolarization suppressibility or withdrawal). This group had a mortality rate of 8.5% as described inanolherreportfrom this series(3). Thus. there were I.913 patients with suppressible ventricular ectopic rhythm who were eligible for randomization. Of these, 277 were assigned to participate in the moricizine arm (which was temtinated in August 1991); I38 subjects demonstrated xl% hut in patients recovering from myacardiai infarction as described by ihgger et al. (20) and to the
exclusion in CAST of patients with few or no ventricular premature depolarizations. Thus, in the presence of a high level of ventricular ectopic activity, the level of risk remains relatively flat and little variation in risk can be anticipated with increases in ventricular premature dep&rizatiott frequency. A low statistical power due to the small number of events may have also been responsible for this lack of correlation. It is also possible, however, that ventricular ectopic activity is not a risk factor in this group of patients with suppressible but untreated ventricular premature depokwizations. Canclusionri. We examined 23 baseline characteristics of patients randomized to placebo in the encainide and flecainide arms of CAST. We identified those characteristics associated with congestive heart failure, arhythmic death or resuscitated cardiac wrest and total (all cause) mortality or resuscitated cardiac west outcomes and found lhem to be similar to baseline predictors identified in other postmyocardial infarction studies. Thus, despite the relatively low mortality rate found in this group. WE anticipate that the increased mortality rate found in CAST patients treated with encainide and tlecainide is probably applicable to any camparably defined group of patients in the postmyocardial infarction period.
STUDIES
Events in the Cardiac Arrhythmia Suppression Trial: Baseline Predictors of Mortality in Placebo-Treated Patients ROBERT THERESE ROBEKT
I. CAPONE,
MD, FACC,*II
S. GERACI, C. SCHLANT,
YUDl
RN,§ KATHRYN
PAWITAN, HANDSHAW,
MD, FACC, ** ALBERT
PHD,t
NABIL
EL-SHERIF,
MD, FACC,B
RN,11 JOEL MORGANROTH,
L. WALDO,
MD, FACCtt
MD. FACC,n
AND THE
CAST INVESTIGATORSZ+
PatientE randomized to placebo in the enwinidr and Recainide arms of ,i,e Cardtae Arrbytbmts Supprerslan Trlnl (CAST) have beenfound to haveB relativelylow l-year mmtality rale of 3.9% in comparisontitk previousstudiesof palitnls in the postmyecardial infarction pried. To determine the comparability ol CAST with preview stodios, bs&ne vsrisbles WIR examined in the 741 pntianlr randomieed to pbwebo in the Secabide and rncainide arms of CAST. Twenty4brco basrllne cbaractrristtu
asmisted
The Cardiac Arrhythmia Suppression Trial (CAST) is a large, prospective. double-blind, placebo-controlled study undertaken to test the hypothesis that pharmacologic suppression of asymptomatic or mildly symmomatic ventricular bremature depolarizations will r&i id a reduction of the risk of arrhythmic death after myocardial infarction. Unlike other trials of anliarrhythmic therapy in the postmyocardial
infarction period, CAST was designed to mimic clinical practice by including an open label titration phase, during which a patient was randomly assigned to one of four possible titration sequences, utilizing a maximum of three drugs at two doseseach to identify one that was associated with 280% suppression of ventricular premature depolaizalions and 8% suppression of nonsustained ventricular tachycardia. If such adrug and dose combination was found. the patient was randomized to treatment with the active aeent or a matching placebo. This element of the CAST design seems to haie’had a significanl e&t on the study patients in that the estimated l-year mortality rate of 3.9% for the placebo group of the encainide and Aecainide arms of the trial appears to be relatively low compared with that of previous studies of patients recovering fmm myocardial infarction. In this study, we investigated the associalion between the baseline variables and atrhythmic death, total mortality and congestive heart failure to determine the comparability of the CAST placebo group wilh those described in previous postmyocardial infarction studies. Because these patients are not treated wirh antiarrhythmic agents, this investigation also provides an impartant natural history study of patients
with an index Q wavemyocardtt l&&on, Idstoryof heart failure, UP (I( d&it&, dbbeta and pr&mg& QRS dwation. T-1 mortality or rwusritakd cardiac arre(t WBT s&nilscantI, awinted with M index Q wave myofsrdlal infarction, diabetes,ST ugnent depressloo, high bndtonal drcJ, protonged QRS duration zmd low e&don fractloo. llw vsriabks at beeline sssorlstedwith martality Imm all uusa or urhythnic death or remtitated cardiac merestsad hear, f.8”~ in the CAST placrbwtrutcd patientsM stmttu to those idpntlded in previous pc&,yoc.rdipI infwctio” st”dt~l. Thus, the obrervntlonof lmrcnsPd mwtallty in CAST awociated with tbendmlnbtratlon atenmlnldtaud fkcalnldtBrsupprv&on ofvenlikular prematuredepdariuIians is probably sp~ikabk to any wmpnrrbly definedgroup of patimb In the wnimyccudW infarction period.
in the myocardial infarction period who have supprec$hle ventricular premature depolarizations.
Methods
A
Study design. detailed summary of CAbT study design has been previously published (IL In brief. all patients had had an acute myocardial infarction within the previous 2 years and had an average of six or more ventricular premature depolarizations per hour on 24-h ambulatory electrocardiwraohy (ECG) recordina. A left ventricular eiection fraclion of ~0.55 was required for those whose-myocardlal infarction had occurred within 90 days; those whose infarclion had occurred HO days but ?lOmgldl. There were no patients in funcrional class IV and a limlted number in class III; extrapolation of the results beyond class II must therefore be made wth c&on. Exerare ierting was not a requirement for entry into the study and IS not mcluded
among the harelms variables evaluated. The average follow-up durauon kas IO months. End points for analysis. These included: II arrhythmic death or resusc,mwd cardmc arrect, the man CAST end pmnr: 2) lolai monality or resuscilated cardiac arrest. a secondarv CAST end pomt: and 3) new or worsening congestwe heart fallwe. &rhythmic denrk was defined as mstantaneous death m the absence of severe congzstive heart failure or shock. unwtnessed death without a preceding change in symptoms for whrch no other cwse could be ascribed or documented cardiac west. Each death and resuscitated cardiac arrest was reviewed bv the Events Commmee. whxh is composed of CAST &stigaors who did not know the treatment assignment of each SubJect evaluated. NW congestive heon foilare wasdefined as the appearance of ar least two of the following symptoms or s,gns when none ,\as noted at baseline study: shortness of breath. easy f&ability. edema. orthopnea. paroxysma: nocturnal dvpnca. jugular venous distension. rales and a third heart sound t&L \Vorsrr!hg congestive hrnn fnilrw wac defined ac an tocrease m one or more functional class. The stud” orotocot was aooroved bv rhe lnstilut,onal Review Board of each participating clinic-t center and by an mdependendy appointed external study Data and Safely !donaorinx Board. Wnttcn informed consent was obtained from all oatients oarticioatine in the studv. Statis&l methods. iVe &lyzed tw&way contingency trebles for the univ.&ate analysis using the standard chisquare test All p values are two-sided and declared signifimot If 80.05. Multivariate analyses of all end points were done wth use of the Car proponional hazard regression model. The exposure time began at the ttme of randomirabon to blinded therapy: patients who were still alive on April I& 1489. rhe dare on which rhe study Data and Safety Moatoring Board recommended that Ihe encainide and Recainide arms of CAST be terminated. had exposure time censored at that point. The exposure time for congestive heart failure was censored at the time of death or cardiac arrest or on Awl 18. 1989. and only the event was considered in the ana!ysis. In the multi&ate analysis. the variables were entered into rhe model by a stepwise procedure until none of the remaining variables outside of the model was agnificant.
1
1.
I
Results A to!al of 743 patients were assigned IO placebo therapy on encamide or flecainide between June IS, 1987 and April 17. 1989. The average follow-up was about IO months. Results of the univariate analyses of the 23 baseline factors are shown on Table I. Results of the multivariate analysis are summarized m Table 2. New or worseningcongestive heart failure. There were 51 subjects who experienced new or worsening congestive heart fadure. On univariate analysis. II variables were significantly associated with this outcome. with I (namely.
histary of previous myocardial infarction) showing a trend toward significance. On mulrivariate analysis, diuretic use (hazard ratio 2.21. diabetes (hazard ratio 1.7). high functional class (hazard ratio I.‘l/class). age (hazard ratio I.Ustandard deviation [SD] of 9.8 years), pr&ged QRS complex duration (hazard ratio I.USD of 19 msl and low left ventricular ejection fraction (hazard ratio I .7/SD of 0.09) were associated with new or worsened congestive heart failure during f0ll0w-up. Arrhythmic death or resuscitated cardiac arrest. Sixteen subjects experienced arrhythmic death or were resuscitated from cardiac arrest. an occurrence rate that would tend to limit the statntical wwer of the ccmmarisons. On univariate analysis, six base& variables were’significantly associated with this end point. with one (namely, Q wave myocardial infarction1 showing a trend toward significance. On multivariate analysis, Q wave myocardial infarction (hazard ratio 7.51, history of congestive heart failure (hazard ratio 3.81, use of digitalis (hazard ratio 3. I), diabetes (hazard ratio 2.6) and prolonged QRS complex duration (hazard ratio IS/SD of 19 ms) were assuciated with sudden arrhythmic death or resuscitated cardiac arrest. Mortality from all causes or resuscitated cardiac arrest. There were 26 deaths in this group. of which 21 were cardiac. On univariate analysis. IO baseline variables were significantly associated with this end pant. On multivariate analysis. Q wave myocardial infarction (hazard ratio 7).
diabetes (hazard ratio 4.71, ST segment depression @ward ratio 2.4), high functional class (hazard ratio I.J/class), prolonged QRS complex duration (bawd ratio U/SD of 19 ms) and low ejection fraction (hazard ratio I.4/SDof 0.09) were associated with total mortality. Discussion Mortalily in lhr varlats CAST study gnupa. We (I) ereviously demonstrated that the mortality rate for Datients with qua&ins ventricular ectapic rhythm enrolled into CAST is comparable with that of other groups studied after acute myocsrdial infarction. This rate is in contrast to the mortality rate for patients in CAST randomized to plsceba after demonstrating suppression of their ventricular arrhythmia during apen label titration. The dieience in mortality rate. comparing those with qualifying witnessed ectopic rhythm with those randomized to placebo, may relate to the modifications in the composition of the originally qualified group of patients during the open label drug titration period (Fig. 1). Initially. all qualified patients in CAST were sub jetted to a sequential scheme of antiarrhythmic treatment (2). Only those with ventricular ectopic rhythm that was suppressed by one of the agents included in CAST were randomized. During this open label titration period, death occurred in 63 (2.7%) of patients eligible for study (I patient was successfully resuscitated from cardiac arrest and subse-
quently randomized). Deaths were related to factors at baseline study associated with increased risk: previous myocardial infarction, more extensive left ventricular dysfunction and a prolonged interval from the index myacardial infarction. addition to 78 patients who were in the open label titration Dhase on April IS. 1990 at the Iermination of the encabtid; and Recainide arms of the study. there were 318 patients not randomized (because of adverse effects. lack of ventricular premature depolarization suppressibility or withdrawal). This group had a mortality rate of 8.5% as described inanolherreportfrom this series(3). Thus. there were I.913 patients with suppressible ventricular ectopic rhythm who were eligible for randomization. Of these, 277 were assigned to participate in the moricizine arm (which was temtinated in August 1991); I38 subjects demonstrated xl% hut in patients recovering from myacardiai infarction as described by ihgger et al. (20) and to the
exclusion in CAST of patients with few or no ventricular premature depolarizations. Thus, in the presence of a high level of ventricular ectopic activity, the level of risk remains relatively flat and little variation in risk can be anticipated with increases in ventricular premature dep&rizatiott frequency. A low statistical power due to the small number of events may have also been responsible for this lack of correlation. It is also possible, however, that ventricular ectopic activity is not a risk factor in this group of patients with suppressible but untreated ventricular premature depokwizations. Canclusionri. We examined 23 baseline characteristics of patients randomized to placebo in the encainide and flecainide arms of CAST. We identified those characteristics associated with congestive heart failure, arhythmic death or resuscitated cardiac wrest and total (all cause) mortality or resuscitated cardiac west outcomes and found lhem to be similar to baseline predictors identified in other postmyocardial infarction studies. Thus, despite the relatively low mortality rate found in this group. WE anticipate that the increased mortality rate found in CAST patients treated with encainide and tlecainide is probably applicable to any camparably defined group of patients in the postmyocardial infarction period.
