Q1997 by Excerpta Medica, Inc. 0002-9149/97/$17.00 587All rights reserved. PII S0002-9149(96)00820-X

/ av1 2w1a 0805 Mp 587 Friday Jan 24 10:29 PM EL–AJC (v. 79, no. 5 ’97) 0805

Persistent Reduction of Mortality for FiveYears After One Year of AcebutololTreatment Initiated During Acute

Myocardial InfarctionMichel Cucherat, MD, Jean-Pierre Boissel, MD, and Alain Leizorovicz, MD,

for the APSI investigators

The APSI trial was a randomized placebo-controlled trialdesigned to assess the efficacy of 1 year of treatment byacebutolol in high-risk patients who had survived anacute myocardial infarction. At 1 year there was a sta-tistically significant 48% relative reduction in total mor-tality (p Å 0.019) in favor of acebutolol. In 1995 a long-term mortality survey was undertaken through anadministrative inquiry and contacts with investigators.The vital status of 586 of the 607 (96.5%) patients en-rolled was known at the cutoff date and all these pa-tients were followed up for at least 5 years. During fol-low-up (in-trial and post-trial period), 74 deaths (24.8%)occurred in the acebutolol group and 96 (31.1%) in the

placebo group (p Å 0.10). No difference between the 2groups was observed for the number of deaths that oc-curred after the end of the trial: 55 deaths (19.6%)among the 281 survivors in the acebutolol group and59 deaths (21.7%) (p Å 0.70) among the 272 survivorsin the placebo group. The annual hazard rate (annualdeath rate), calculated year by year using the actuarialmethod, was significantly different (p õ0.01) only forthe first year and was not significantly different there-after. Thus, the initial benefit obtained in 1 year of treat-ment by acebutolol lasts for 5 years. Q1997 by Ex-cerpta Medica, Inc.

(Am J Cardiol 1997;79:587–589)

Beta-blocking drugs have been shown to reducetotal mortality, sudden death, and reinfarction

rates in patients who survive an acute myocardialinfarction whether they are at low, middle,1 or highrisk.2,5 However, these reductions were demonstratedonly for the duration of the trials (range 9 months to4 years). There are few data available on whether thebenefit on total mortality vanishes, continues, or in-creases after the treatment period. Only 2 previoustrials have followed-up patients after the end of thedouble-blind phase.3,4 This study was undertaken toexplore the longer term mortality in patients whowere alive at the end of the Acebutolol et PreventionSecondaire de l’Infarctus (APSI) trial. The APSI trialwas a randomized, placebo-controlled trial, whichaimed to assess the effectiveness of acebutolol inpreventing late death in high-risk patients who hadsurvived an acute myocardial infarction.5,6 BetweenApril 1987 and December 1988, 607 men andwomen were enrolled, but the trial was discontinuedafter the second interim analysis because 1-year mor-tality in the placebo group was much lower than ex-pected. A statistically significant reduction for totalmortality at 1 year of 48% (p Å 0.019) was observedin the acebutolol group compared with the placebogroup. In 1995, 6 years after the end of the trial, theAPSI coordinating center followed up those patients

From the APSI Coordinating Center, Service de pharmacologie cli-nique, Hopital cardiologique, Lyon France. The study was supportedby a grant from SPECIA Pharmaceuticals, Paris, France. Manuscriptreceived July 24, 1996; revised manuscript received and acceptedOctober 1, 1996.

Address for reprints: Michel Cucherat, Service de pharmacologieclinique, 162 avenue Lacassagne, BP3041, 69394 Lyon Cedex03, France.

still alive at the end of the trial, the results of whichare presented in this study.

METHODSThe protocol has been described elsewhere.5

Briefly, high-risk patients were selected using a pre-diction function in 44 French hospitals. After selec-tion, the patients were admitted to the trial°22 daysafter the myocardial infarction and were treated andfollowed up for 1 year. The protocol was approvedby an Ethics Committee and the patient consent wasobtained.

The present follow-up study was not planned inthe protocol, but the data recorded were sufficient toenable the vital status of patients to be assessed byadministrative inquiries or by phone call to the pa-tients’ family doctor. The investigators could contactthe patient or the family directly. The only data re-corded was vital status as of February 1994. Anydeaths after this date were not considered in thisstudy.

The survival rates in the 2 groups were comparedusing the log-rank test.

RESULTSFollow-up: The vital status on the cutoff date was

obtained for 586 of the 607 patients (96.5%) enrolledin APSI. The number of patients lost to follow-upwas not statistically different between the groups(acebutolol: 15 of 298, 5.0%; placebo: 6 of 309,1.9%, p Å 0.06, chi-square test).

The follow-up time for the last patient enrolled inApril 1987 was 5.12 years. The first patient enrolledin December 1988 was followed 7.6 years. The me-dian duration of follow-up was 6 years (5 years after

588 THE AMERICAN JOURNAL OF CARDIOLOGYT VOL. 79 MARCH 1, 1997

/ av1 2w1a 0805 Mp 588 Friday Jan 24 10:29 PM EL–AJC (v. 79, no. 5 ’97) 0805

FIGURE 1. Survival curves for the acebutolol and the placebogroups.

TABLE I Annual Hazard Rates Observed in the Acebutolol and the Placebo Group for Each Year

Year

Acebutolol group (n Å 298)

Deaths inthe Year*

EffectiveSample Size†

Hazard Rate% (SD)

Placebo Group (n Å 309)

Deaths inthe Year

EffectiveSample Size†

Hazard Rate% (SD)

Comparisonbetween the2 Groups

1 17 291.5 5.38 (1.37) 37 307.5 12.03 (1.86) p õ0.012 9 267.5 3.36 (1.10) 9 267.5 3.36 (1.10) NS3 9 257.0 3.50 (1.15) 12 256.5 4.68 (1.32) NS4 11 247 4.45 (1.31) 12 244 4.92 (1.38) NS5 13 235.5 5.52 (1.49) 10 231.5 4.32 (1.34) NS6 7 189.5 3.69 (1.37) 7 191.5 3.66 (1.36) NS7 6 100 6.00 (2.37) 9 102 8.57 (2.73) NS

*The exact dates for 2 deaths are not known. These deaths were counted in the global mortality, but could not be used to estimate the annual hazard rate.†The effective sample size represents the mean number of patients at the risk during the year considered.

the end of the trial), and 25% of the patients werefollowed upú6.75 years. The distribution of the du-ration of follow-up was similar in both the acebutololand placebo groups. The dates of the deaths for 2patients in the acebutolol group were not preciselyknown and were therefore not taken into considera-tion for the log-rank test.

Mortality: Over the total follow-up period (in-trialand post-trial periods), 74 of the 298 patients(24.8%) allocated to the acebutolol groups died com-pared with 96 of the 309 patients (31.1%) allocatedto the placebo group (pÅ 0.10, chi-square test). Dur-ing the first 5 years, 59 patients (19.8%) in the ace-butolol group died compared with 80 (25.9%) in theplacebo group (p Å 0.075, Fisher’s exact test). Nodifference was observed for the number of deathsoccurring after the end of the trial; in the acebutololgroup 55 of the 281 survivors (19.6%) died com-pared with 59 of 272 survivors (21.7%) in the pla-cebo group (p Å 0.70).

Figure 1 shows the survival curves for the 2groups. The log-rank test over the whole follow-upperiod was not significant with p Å 0.066 althoughit was significant up to the end of the 4th year.

Instantaneous risk of death: The number of deathsand the instantaneous risk of death (the hazard rate)for each year calculated using actuarial proceduresare presented in Table I. The instantaneous risk ofdeath is the probability of dying during a given year,

for the patients still alive at the beginning of thatyear.

DISCUSSIONIt is important to note that these results concern

only the long-term effect of 1 year of treatment byb-blocking drug immediately following acute myo-cardial infarction. The controlled intervention wasonly applied during the trial period, but after thisperiod we cannot rule out the possibility that patientsin both groups received b blockers. Also, an un-known number of patients in the acebutolol groupprobably stopped taking the drug. The influence ofthese unknown treatments on the survival after theend of the trial could not be taken into account inthe analysis of the long-term effects of the first yearof treatment with acebutolol. This study thereforeprovides no evidence about the optimal duration forb-blocker treatment after acute myocardial infarc-tion. This important question can only be answeredby further trials in which patients are randomly as-signed to different durations of treatment.

Graphically, it appears that the survival curvesdiverge during the first year, witnessing the benefitsof the b blockade during the trial, and thereafter thecurves remained parallel. This pattern is confirmedby the evolution of the annual hazard rate (Table I).The hazard rate was only significantly different forthe first year and was similar for years 2 to 7, cor-responding to the parallel survival curves over thisperiod. The number of deaths occurring after the trialstopped were also the same in both groups. The ini-tial benefit, with a significant log-rank test, remainsuntil the 4th year. No evidence of a late increase inmortality in the treated group was observed, thus theinitial reduction was not counterbalanced by a re-bound effect at the end of acebutolol treatment. Thereversal to a nonsignificant difference after the 5thyear could be explained by the reduction in the ef-fective groups’ sizes and the reduced relative risk,leading to a lower statistical power for this compar-ison which was not initially planned.

The results of this follow-up survey are in con-trast with the findings of a previous similar studyusing oxprenolol,4 which suggested that the initialbenefits achieved with oxprenolol during the first 3years after myocardial infarction might not be main-

CORONARY HEART DISEASE/APSI LONG-TERM FOLLOW-UP 589

/ av1 2w1a 0805 Mp 589 Friday Jan 24 10:29 PM EL–AJC (v. 79, no. 5 ’97) 0805

tained with more prolonged follow-up. However, thepremature interruption of a subsequent trial of ox-prenolol because of increased mortality with thetreatment7 has suggested that a b-adrenergic block-ing drug with marked intrinsic sympathomimetic ac-tivity may be harmful in this situation. This was notconfirmed by a metanalysis which did not show sig-nificant heterogeneity between trials.2

In contrast, our results are similar to the findingsof the long-term survey of the Norwegian Multicen-ter Study on Timolol after Acute Myocardial Infarc-tion.3 In that trial, patients were followed up initiallyfor 2 years but a later follow-up was conducted over6 years. This later analysis indicated that the bene-ficial effect of timolol therapy on total mortality re-mained significant with the survival curves becom-ing parallel 2 years after randomization.

Several hypotheses are possible to characterizethe mechanisms of the sustained benefit observed inthis study with acebutolol.

One possibility is that acebutolol acts on a processthat exists permanently after a myocardial infarctionwith a constant hazard rate. In this case, the parallelsurvival curves may be explained by an equal num-ber of patients receiving b blockers in the 2 groupsafter the end of the trial. Our data are also consistentwith an effect of acebutolol on a process existingonly during a short period after the infarction. Theyear of treatment confers a benefit only while thisprocess exists, and thus the initial divergence of thesurvival curves is accounted for. After this, the treat-ment has no impact on survival and so the curvesremain parallel.

The fact that the curves do not continue to divergeafter the end of the treatment does not support an

action of acebutolol on a process that conditions thefuture survival (such as the size of the infarct or theleft ventricular remodeling) of the patient. If this wasthe case, acebutolol would permanently influence thehazard rate and the survival curves would constantlydiverge.

However, identification of the mechanism is ham-pered by the small number of patients in this survey,it does not allow a more precise analysis of the sur-vival curve, and also by the fact that we do not knowwhat treatment was received after the end of the trial.

Acknowledgment: We acknowledge all the inves-tigators who participated in the APSI trial. The com-prehensive list of the members of the APSI groupwas published in the appendix of the study present-ing the trial results.5 We are grateful to the reviewersfor the useful comments and suggestions.

1. Yusuf S, Peto R, Lewis J, Collins R. Sleight T. Beta blockade during andafter myocardial infarction: an overview of the randomized trials. Progr Car-diovasc Dis 1985;27:335–371.2. Beta-Block Pooling Project Research Group. The Beta-Blocker Pooling Pro-ject (BBPP) research group. Subgroup findings form randomized trials in postinfarction patieints. Eur Heart J 1988;9:8–16.3. Pedersen TR, for the Norwegian Multicenter Study Group. Six year follow-upof the Norwegian Multicenter Study on Timolol after acute myocardial infarc-tion. N Engl J Med 1985;313:1055–1058.4. Taylor SH, Silke B, Ebbutt A, Sutton GC, Prout BJ, Burley DM. A long-term prevention study with oxprenolol in coronary heart disease. N Engl J Med1982;307:1293–1301.5. Boissel JP, Leizorovicz A, Piclet H, Peyrierux JC, for the APSI Invesigators.Secondary prevention after high risk acute myocardial infarction with low-doseacebutolol. Am J Cardiol 1990;66:251–260.6. Boisel JP, Leizorovicz A, Picolet H, Ducruet T, and the APSI Investigators.Efficacy of acebutolol after acute myocardial infarction (The APSI trial). Am JCardiol 1990;66:24C–31C.7. European Infarction Study Group. European Infarction Study (EIS): a sec-ondary prevention study with slow release oxprenolol after myocardial infarc-tion: morbidity and mortality. Eur Heart J 1984;5:189–202.