OUP user menu

Temporal trends on the risk of arrhythmic vs. non-arrhythmic deaths in high-risk patients after myocardial infarction: a combined analysis from multicentre trials

Yee Guan Yap, Trinh Duong, Martin Bland, Marek Malik, Christian Torp-Pedersen, Lars Køber, Stuart J. Connolly, Bradley Marchant, John Camm
DOI: http://dx.doi.org/10.1093/eurheartj/ehi268 1385-1393 First published online: 24 May 2005

Abstract

Aims An understanding of the temporal trends on the risks of arrhythmic death (AD) vs. non-arrhythmic deaths (NAD) after myocardial infarction (MI) is crucial in deciding the optimal timing for risk stratification and treatment window for prophylactic antiarrhythmic treatment. However, contemporary data on such information is lacking.

Methods and results Individual patient data were pooled from the placebo arms of EMIAT, CAMIAT, SWORD, TRACE, and DIAMOND-MI who had a recent MI and left ventricular ejection fraction (LVEF) <40% or frequent ventricular premature beats (VPBs). Temporal trends were investigated for all studies from day 45 after acute myocardial infarction (AMI) to account for different recruitment periods between trials, and then from the onset of MI for TRACE and DIAMOND-MI that recruited patients within 2 weeks after MI. In total, 3104 patients (median age 65, range: 23–92; 2471 males) were pooled from all five studies, with a total of 487 deaths at 2-year follow-up; 220 deaths were ADs and 172 were NADs. The risks of both AD and NAD were highest in the first 6 months but the risk of AD was consistently higher than that of NAD throughout the 2-year period [rate of death/100 person-year at risk (AD/NAD): 8.09/6.07 (45 days to 6 months), 4.07/3.35 (>6–12 months), 4.34/3.60 (>12–18 months), 3.76/2.77 (>18–24 months)]. There were significant interactions between the temporal trends of mortalities and gender (P=0.03) and history of hypertension (P=0.04). A similar trend was observed when mortality was measured from time of onset of MI from the combined TRACE and DIAMOND-MI dataset.

Conclusion Our study provided the first contemporary evidence that in high-risk post-MI patients with LVEF <40% or frequent VPBs, the risk of AD was higher than that of NAD for up to 2 years although in female patients, they became increasingly more likely to die from NAD after 6 months. Therefore, risk stratification of post-MI patient at high risk of AD remains a worthwhile exercise. However, the risks of AD (and NAD) were highest in the first 6 months after AMI and level-off thereafter, suggesting that the optimal window period for risk stratification for implantable cardioverter defibrillator after AMI is in the first 6 months.

  • Myocardial infarction
  • Risk
  • Arrhythmic mortality
  • Non-arrhythmic cardiac mortality

See page 1350 for the editorial comment on this article (doi:10.1093/eurheartj/ehi314)

Introduction

Despite the decrease in the death rate after acute myocardial infarction (AMI) by 30% over the last 10 years following the introduction of contemporary medical therapy and revascularization, one-third of patients continue to die following an AMI.1 Of those who survive to hospital discharge following an MI, ∼10% will die within the first year.2 The mechanism of cardiac death following MI can be broadly divided into either arrhythmic (electrical) or non-arrhythmic (mechanical). The distinction between modes of death is important when treatments with antiarrhythmic drugs or implantable cardioverter defibrillator (ICD) are considered. Temporal influence on mode of death remains speculative.3 Most survival studies have merely reported the ‘risk’ of arrhythmic or sudden deaths as a per cent of the total deaths. As this provides some insight into the incidence rate of arrhythmic death (AD), it does not provide any information on the temporal trends and relative risk of AD vs. non-arrhythmic death (NAD) and important prognostic risk factors have not been adjusted for.

The purpose of this study was to explore the temporal trends on the risk of AD vs. NAD in high-risk post-MI patients with reduced left ventricular ejection fraction (LVEF) or ventricular arrhythmias.

Methods

Data retrieval

We pooled the individual patient data from the placebo limbs of EMIAT, CAMIAT, SWORD, TRACE, and DIAMOND-MI studies. The detailed designs of these studies have already been published elsewhere.48 All of these survival trials:

  1. recruited patients with recent documented AMI;

  2. were published in the thrombolytic era;

  3. were multicentre randomized, placebo-controlled prospective survival trials;

  4. comprised high-risk patients with either LVEF ≤40% (measured by echocardiography, MUGA scan, or ventriculography) or ventricular arrhythmias, i.e. more than 10 ventricular premature beats (VPBs)/h or a run of VT (measured by three-channel Holter ECGs);

  5. have clearly defined mortality endpoints determined by an event committee.

Outcomes

The outcomes pooled across the trials were (i) all-cause mortality, (ii) arrhythmic cardiac mortality (including sudden cardiac death), i.e. AD, and (iii) non-arrhythmic cardiac mortality, i.e. NAD, regardless of which outcome was designated as the primary end-point by the trial investigators. All outcomes were analysed within 2-year follow-up.

Statistics

A total of 1113 patients from the remote MI arm of SWORD who were randomized 380 days after MI were excluded from our analysis. In order to allow for studies having different recruitment periods among all five trials, the main analysis of 2 years survival was measured from day 45 after MI. A further analysis of survival measured from the day of MI to 2-year follow-up was based on TRACE and DIAMOND-MI in which patients were recruited within 2 weeks of MI.

Overall rates of AD and NAD were compared each 6 months after MI. Patients who were censored would not contribute information to mode of death, and therefore were not included in this analysis. Cause-specific cumulative incidence of mortality was calculated using the Kaplan–Meier method. The incidence for a particular cause of death was estimated by treating deaths from other causes as censored observations (i.e. as if death has not yet occurred), therefore implying that different mortality risks acted independently. We investigated this assumption by calculating in addition, the crude cumulative incidence which allowed risk of AD and NAD to be related. In analyses of time-to-failure data with competing risks, cumulative incidence functions may be used to estimate the time-dependent cumulative probability of failure due to specific causes (i.e. it is an estimation of failure probabilities in the presence of competing risks).

The interaction between modes of cardiac death and risk factors at baseline was investigated. Comparisons between arrhythmic and non-arrhythmic cumulative incidences were made across groups of patients defined by levels of each risk variable. Variables that were found to be significantly associated to either AD or NAD were chosen: age, sex, previous MI, history of hypertension, previous angina, systolic blood pressure, heart rate, New York Heart Association (NYHA) functional class, LVEF, and Q-waves. No interactions between mode of cardiac death (AD and NAD) and smoking status, atrial fibrillation, and presence of diabetes were found. LVEF was analysed as <30% and 30–40% whereas other continuous variables were stratified using their median values (age <65 or ≥65, systolic blood pressure <120 or ≥120 mmHg, and heart rate <75 or ≥75 b.p.m.). NYHA classes were grouped into 0–I and II–IV.

Formal tests for interaction were made by fitting separate Cox multiple regression models for AD and NAD including all risk factors and adjusting for study and treatment effects. For each variable, the corresponding estimates of adjusted log hazards for AD and NAD using large sample z-test were compared. Effectively, this compares the ratio of risk of AD to NAD between the respective risk groups (see Appendix). Two-sided significance test with P<0.05 considered as a statistical significant result was used.

Results

Trends in mode of cardiac death over time

In total, 3431 patients were pooled together from the five trials. Three patients who died, but for whom no cause of death could be classified, were excluded. For the main analysis, patients who died within 45 days were also excluded from the analysis to allow for studies having different recruitment periods. The remaining 3104 patients were known to have survived at least 45 days from MI. The baseline characteristics of these patients were summarized in Table 1. Of these, 2471 (79.6%) are male and the median age was 65 years (range 23–92). There were 487 deaths within the first 2 years, 220 were of AD and 172 were of NAD. The rate of AD per 100 person-year at risk was reduced but remained consistently higher than that of NAD during the corresponding period (Tables 2 and 3).

View this table:
Table 1

Baseline characteristics of patients from all five studies and TRACE and DIAMOND-MI only

Patients surviving 45 days from all five studiesAll patients from TRACE and DIAMOND-MI
n31041631
TRACE766873
DIAMOND-MI620758
CAMIAT594
EMIAT726
SWORD398
Median (range) no. of days from MI to randomization 10 (0–53)  3 (0–14)
M/F (%M)2470/634 (79.6)1187/444 (72.8)
Mean age (SD) 65 (23–92) 69 (30–92)
Smoker Y/N (%)2116/585 (78.3)1254/372 (77.1)
Medical history Y/N (%)
 Previous MI 981/2113 (31.7) 573/1055 (35.2)
 Hypertension 923/2160 (29.9) 379/1244 (23.4)
 Diabetes 402/2296 (14.9) 221/1402 (13.6)
 Median heart rate (range) 75 (40–135) 80 (40–138)
 Median systolic blood pressure (range) 120 (75–120) 120 (80–200)
 Q-wave2188/878 (71.4)1176/446 (72.5)
 Atrial fibrillation Y/N (%) 303/1876 (16.2) 286/1337 (17.6)
NYHA
 0–I1376 (44.7) 421 (26.7)
 II–IV1701 (55.3)1159 (73.4)
LVEF
 <30 791 (31.6) 478 (29.4)
 ≥301716 (68.5)1150 (70.6)
Concomitant medication Y/N (%)
 Thrombolytic treatment1615/1468 (52.4) 771/861 (47.3)
 Digoxin 508/2591 (16.4) 338/1293 (20.7)
 Beta-blockers1034/2065 (33.4) 232/1399 (14.2)
 Calcium antagonist 559/2540 (18.0) 429/1202 (26.3)
 ACE‐inhibitor 981/2118 (31.7) 103/1528 (6.3)
View this table:
Table 2

Rates of arrhythmic and non-arrhythmic cardiac mortality for all patients measured from day 45 after MI and patients from TRACE and DIAMOND-MI that were measured from day of MI

Up to 6 month>6–12 month>12–18 month>18–24 monthOverall
Rate per 100 person-year at risk
Survival from day 45 after MI from all five studies
AD 8.094.074.343.764.68
NAD 6.073.353.602.773.66
Survival from date of MI from TRACE and DIAMOND-MI
AD23.975.625.144.9611.01
NAD18.915.285.144.489.28

Overall mode of cardiac death did not appear to change over time, with the rate of arrhythmic death consistently higher than NAD.

View this table:
Table 3

Rate of mortality for various sub-groups that showed significant interactions with the modes of deaths

Up to 6 month>6–12 month>12–18 month>18–24 month
Rates per 100 person-year at risk. Arrhythmic Death: Non-arrhythmic Death
Survival from day 45 after MI from all five studies
Males 7.48 : 5.064.61 : 3.204.41 : 3.374.47 : 2.38
Females10.49 : 10.031.96 : 3.924.05 : 4.501.07 : 4.27
HBP+ 8.70 : 4.664.13 : 4.136.53 : 3.595.89 : 1.96
HBP− 7.78 : 6.724.08 : 3.063.38 : 3.642.80 : 3.11
Survival from date of MI from TRACE and DIAMOND-MI
Males22.85 : 15.637.07 : 4.795.31 : 4.256.04 : 3.82
Females27.24 : 28.401.35 : 6.734.63 : 7.711.83 : 6.41
HBP+36.00 : 20.007.06 : 7.858.67 : 7.705.79 : 3.48
HBP−20.47 : 18.735.25 : 4.604.25 : 4.504.47 : 4.77
Hr<7516 : 04 : 7.184.19 : 2.803.83 : 2.735.92 : 3.95
Hr≥7528.30 : 25.316.44 : 6.715.88 : 6.494.42 : 4.78
Age <6517.71 : 6.594.01 : 1.783.03 : 2.534.12 : 3.53
Age ≥6527.52 : 25.896.62 : 7.456.49 : 6.815.52 : 5.13
SBP <12031.02 : 18.827.10 : 3.954.14 : 3.684.95 : 3.85
SBP ≥12018.35 : 17.764.02 : 6.706.09 : 6.854.95 : 5.00

Rates per 100 person-year at risk are presented (AD:NAD).

Figure 1 compared the modes of cardiac deaths (AD vs. NAD) among patients surviving 45 days from MI over time. Cause-specific cumulative incidence at 2 years was 8.7% (95% CI 7.7–9.9) for AD and 6.9% (6.0–8.0) for NAD. Crude cumulative incidences for both mortalities were similar but slightly lower than the cause-specific incidence at all times. As this remained the case in subsequent analyses, it was tenable to assume that the risks of AD and NAD were independent from each other. Only cause-specific incidence is therefore presented in subsequent results.

Figure 1 Cause-specific and crude cumulative incidences of arrhythmic and non-arrhythmic cardiac mortality in patients surviving 45 days. The risk of AD and NAD were independent from each other as the patterns of crude cumulative were very similar to cause-specific incidence curves. The risk of AD was persistently higher than that of NAD during the 2-year period after MI.

Of the 1631 patients pooled from TRACE and DIAMOND-MI, 536 died within 2 years (241 AD and 203 NAD). During the first 6 months, the rate of AD was 23.97 per 100 person-year at risk compared with 18.91 for NAD (Table 2). Both rates fell dramatically immediately thereafter and continued to fall gradually. Although a higher proportion of cardiac deaths were attributable to arrhythmic causes in the first 6 months (55.9%) than after (51.3%), this was not statistically significant (P=0.4).

Interaction between mode of cardiac death and risk factors

Table 4 compares mode of cardiac death by risk groups. Both analyses showed that male patients were significantly more likely to die of arrhythmic than non-arrhythmic causes compared with female patients, by an average of 1.7-fold (Table 4). Figure 2 showed that female patients began with similar rates of AD and NAD but by the end of first year NAD became the predominant mode of cardiac death. In contrast, AD remained higher than NAD in male patients at all times. We also found significant interaction between the mode of cardiac death and a history of hypertension for both survival periods (Table 4). The incidence of AD at 2 years was notably higher than that of NAD among patients with history of hypertension, but the difference was slight in those without (Figure 3).

Figure 2 (A) Female patients began with similar rates of arrhythmic and non-arrhythmic mortality but by the end of first year arrhythmic mortality became the predominant mode of cardiac death. In contrast, arrhythmic mortality remained higher than non-arrhythmic cardiac mortality in male patients at all time. This change occurred just before 12 months in patients from the five trials who survived at least 45 days and around 6 months in DIAMOND-MI and TRACE patients. (B) The incidence of AD during the 2-year period was notably higher than that of NAD among patients with history of hypertension (NBP+) but the difference was slight in those without hypertension (HBP−).

Figure 3 (A) In younger age group patients (<65 years), the risk of AD is significantly higher than that of NAD, which is consistent with the overall trend. However, in the older age group patients (≥65 years), the risk of AD is similar to that of non-arrhythmic cardiac death. (B) In patients with low baseline systolic blood pressure (<120 mmHg), the risk of AD is significantly higher than that of NAD, whereas the risks for both mortalities are similar in patients with high baseline systolic blood pressure (≥120 mmHg). (C) Patients with low heart rate had higher risk of AD than NAD whereas in patients with high heart rate, the risk of NAD is very close to that of AD.

View this table:
Table 4

Interactions between risk factors and arrhythmic and non-arrhythmic cardiac mortality at 2 years from MI

Patients from all five studies survival from 45 days after MIPatients from TRACE and DIAMOND-MI survival from date of MI
Risk groupsCause-specific 2-year incidence (%) AD : NADRatio of relative hazard of AD:NADa (95% CI)P-value for interactionCause-specific 2-year incidence (%) AD : NADRatio of relative hazard of AD:NADa (95% CI)P-value for interaction
Age
 <65 5.9 : 3.61.18 (0.72–1.92)0.513.1 : 6.81.75 (1.04–2.94)0.04
 ≥6511.2 : 9.9119.8 : 19.41
Sex
 M 9.1 : 6.21.74 (1.04–2.91)0.0318.1 : 12.91.72 (1.09–2.70)0.02
 F 7.2 : 9.5115.1 : 20.81
Previous MI
 Y14.0 : 10.90.92 (0.59–1.44)0.724.1 : 19.21.23 (0.80–1.88)0.3
 N 6.4 : 5.0113.9 : 12.91
Hypertension
 Y11.0 : 6.41.65 (1.02–2.65)0.0424.1 : 16.81.90 (1.20–2.99)0.006
 N 7.7 : 7.2115.3 : 14.61
Previous anginab
 Y13.0 : 11.90.74 (0.46–1.21)0.222.1 : 21.01.10 (0.69–1.75)0.7
 N 6.1 : 4.1113.8 : 10.71
Systolic blood pressure
 <120 9.3 : 6.31.47 (0.95–2.27)0.0920.1 : 13.31.79 (1.18–2.70)0.007
 ≥120 8.1 : 7.6115.1 : 16.51
Heart rate
 <75 6.3 : 3.51.59 (0.96–2.56)0.0713.6 : 7.71.85 (1.12–3.13)0.02
 ≥7511.0 : 10.1119.4 : 18.71
NYHA
 0/I 5.0 : 4.00.98 (0.58–1.67)0.9 8.1 : 8.50.95 (0.54–1.69)0.9
 II–IV12.0 : 9.7120.1 : 16.91
Q-wave
 Y 7.5 : 5.61.02 (0.62–1.54)0.916.4 : 13.01.04 (0.67–1.61)0.9
 N11.5 : 9.5120.3 : 20.21
Smokingb
 Current/ex 9.0 : 7.10.84 (0.49–1.43)0.517.3 : 14.50.79 (0.48–1.28)0.4
 Non 8.1 : 7.3117.2 : 16.91
Atrial fibrillationc
 Y13.2 : 15.01.06 (0.54–2.10)0.921.5 : 23.91.07 (0.59–1.92)0.8
 N 9.8 : 7.6116.4 : 13.21
Diabetesb
 Y13.1 : 11.81.14 (0.66–1.94)0.626.4 : 19.51.48 (0.87–2.52)0.1
 N 8.1 : 6.4115.9 : 14.41
LVEFc
 <30%15.2 : 14.01.14 (0.71–1.81)0.628.2 : 25.31.04 (0.69–1.59)0.8
 30–40% 6.6 : 2.6113.3 : 11.31

aData not available for CAMIAT.

bData not available for SWORD.

cAdjusted for other risk factors.

When considering survival time from the time of index MI based on TRACE and DIAMOND-MI, we found significant interactions between modes of cardiac deaths and age, systolic blood pressure, and heart rate. Increase in the risk of AD compared with NAD was observed in patients with age <65, baseline systolic blood pressure <120 mmHg, or heart rate <75 b.p.m. (Figure 3). There was some indication when comparing mortality rates over time that the effect of hypertension, systolic blood pressure, heart rate, and age on mode of cardiac death may be greater in the first year for these subgroups of patients.

Discussion

We performed this study to examine specifically the relative risk of AD vs. NAD in high-risk post-MI patients with reduced LVEF or ventricular arrhythmias. We used only the placebo arms of patients because the respective drug investigated in these trials, including amiodarone, d-sotalol, and trandolapril [the only angiotensin-converting enzyme (ACE)-inhibitor shown to affect arrhythmic mortality] could affect the risk of AD. The principal finding in our study is that the overall risk of AD from either date of MI or day 45 after MI was persistently higher than that of NAD and this did not change over time in the 2-year follow-up period. Furthermore, the absolute risks of both AD and NAD were highest in the first 6 months after AMI and decreased with time. To our knowledge, this is the first report on the temporal risk relationship of AD vs. NAD in post-MI patients.

Arrhythmias and sudden cardiac death are responsible for a large proportion of cardiac deaths after MI.2 A conceptual model based on mortality data suggested that arrhythmic or sudden cardiac deaths occur most frequently in the first year after an MI.3 However, as infarct-related cardiomyopathy develops, death attributable to arrhythmia decreases and NAD from pump failure dominates in the ensuing period. However, this was not the case. We found that the risk of AD was persistently higher than NAD and they both decreased almost proportionally over time in the 2-year follow-up period after MI despite the majority of our patients having significant left ventricular dysfunction. One explanation is that left ventricular dysfunction is such a powerful predictor that it essentially pre-selected patients at the very high risk of ventricular arrhythmias and sudden cardiac death as previously demonstrated (i.e. high positive predictive value).9,10

Another possibility is that with modern therapy in the thrombolytic era, the development of infarct-related cardiomyopathy that commonly happened in the ensuing period after MI might not have increased the risk of NAD as previously speculated, nor did it alter the temporal risk relationship between AD and NAD. This is supported by our data that the risks of AD vs. NAD were not statistically different between MI patients with LVEF <30% and those with LVEF 30–40% (i.e. no interaction between LVEF on relative risk of AD and NAD), indicating that the degree of left ventricular dysfunction, at least on patients with LVEF ≤40%, has no bearing on the ratios of the rate of deaths with time between AD and NAD. Our method of analysing the risk of mortality is conservative and a more vigorous statistical analysis is probably not necessary because the risks of AD and NAD were independent from each other. Our observation was supported by the result from The Multicentre UnSustained Tachycardia Trial (MUSTT), which also demonstrated that there was no interaction between LVEF and modes of death on patients with documented coronary artery disease.11 Among the patients who did not receive antiarrhythmic therapy in the MUSTT study (i.e. the control group), the proportions of AD were similar whether the LVEF was <30% or between 30% and 40%, irrespective of inducible tachyarrhythmias. Approximately half of all deaths in both LVEF groups were caused by AD during the 5-year follow-up, which is comparable to ours. Of interest is the finding that the incidence of non-arrhythmic mortality in patients with a relative preserved LVEF (30–40%) is 2.6 or 11.3%, respectively, if the follow-up period exclude or include the first 45 days after the acute event, suggesting that the risk of NAD, probably from cardiomyopathy, is a significant mode of death in the medium period after MI despite a relative preserved LVEF. The proportion of patients receiving percutaneous reperfusion therapy instead of thrombolytic therapy was unknown in our cohort of patients. Although percutaneous reperfusion therapy has an advantage over thrombolytic therapy in mortality benefit, thrombolysis is as equal effective to primary percutaneous coronary intervention if it could be started within the first 3 h of symptoms.12 Thus, the findings in our study remain relevant in modern reperfusion therapy.

The risk of AD (and NAD) in the first 6 months after MI was approximately twice of that in the subsequent 6 months intervals when survival was measured from 45 days after MI in all studies. The risk was even higher when the survival was measured from date of index MI in TRACE and DIAMOND-MI dataset, with the risk of AD in the first 6 months after MI almost four-fold of that in the subsequent 6 months intervals. Furthermore, the AD rate in TRACE and DIAMOND-MI trials during the first 6 months is almost three times higher than in the analysis of the survival data from day 45 after MI from all five studies, which suggests that two-thirds of the ADs occur within the first 45 days after MI. The very fast increasing divergence of the survival curves during the first few months also suggests this. Thus it appears that the AD rate shows a peak of 16% of the first 45 days, followed by a much lower rate of 8% between day 45 and 6 month and 5% thereafter. The time dependence of the risk of mortality does not appear to have changed compared with pre-thrombolytic reports.9,13,14 The data from Multicentre Post-infarction Programme demonstrated that the shape of survival curves and the risk of death were influenced by the presence and magnitude of risk after the index infarct. Sub-grouping patients according to interactions between frequency of VPBs and LVEF after they have survived AMI resulted in a progressively increasing risk as the number and cumulative power of risk factors increased.9 The low-risk subgroups generated a linear survival curves, whereas the added mortality in higher-risk subgroups tended to be expressed early post-MI. Similarly, in patients with chronic coronary artery disease who survived a cardiac arrest, the risk of recurrent cardiac arrest was highest during the first 6 months and decreased dramatically thereafter with time. LVEF <35% was the strongest predictor for this early phase recurrence.13 This temporal property of risk as shown in our study is vital when planning the strategy and window period for risk stratification and intervention in these patients or designing a survival trial. For instance, when treating an individual patient, the probability of benefit from a preventive antiarrhythmic intervention will be highest if the treatment is carried in the first 6 months after MI when the risk of AD is greatest. Similarly, controlled antiarrhythmic intervention trials that enrol patients more than 6 months after MI to which the study is indexed might be confounded by a lower than anticipated event rate if such entrants are heavily represented in the study group, unless the treatment is highly effective in preventing AD.

Our data are particularly relevant in light of the recent publication of the Multicentre Automatic Defibrillator Implantation Trial II (MADIT II),14 which demonstrated that prophylactic ICD therapy was associated with a 31% reduction in the risk of death compared with conventional medical therapy among patients with a prior-MI and a LVEF of ≤30%. The cohort of patients in MADIT II is similar to our study population: both studies recruited post-MI patients with reduced ejection fraction (LVEF of ≤30% in MADIT II vs. LVEF ≤40% in our study except the subgroups of patients from CAMIAT trial) and both studies had similar though not identical recruitment periods (1 month in MADIT II vs. 45 days in our study from the index MI). In the MADIT II study, although ∼88% of the patients were recruited >6 months from the last MI, prophylactic ICD treatment confers a survival benefit. This is probably explained by our study that the risk of AD was consistently higher than that of NAD throughout the 2-year period though the risk of AD dropped by four-fold after 6 months. Thus, both studies substantiate the claim that risk stratification and prophylactic treatment with ICD for the prevention of AD after AMI remains a worthwhile exercise although one may speculate that a larger survival benefit may have been observed if majority of the patients from the MADIT II study were recruited within the first 6 months after AMI.

In our study, the rate of AD at 2 years was 8.7% vs. a higher rate of 18% in the control (i.e. no antiarrhythmic treatment) group in the MUSTT study15 despite the longer time interval between the most recent MI and enrolment in MUSTT [a median period of 9 days (0–53 days) in our study vs. >3 years in 52% of patients in MUSTT]. This is because in MUSTT, the post-MI patients recruited were of higher risk (LVEF ≤35%, NSVT, and inducible and non-suppressible ventricular tachycardia at electrophysiological study) compared with our patients who had LVEF ≤40% only except for the subgroup from CAMIAT that had ventricular arrhythmias (i.e. >10 VPDs/h or a run of VT).

There is a significant interaction between gender and temporal trend of the risks of AD vs. NAD. As female patients had a similar risk of AD to NAD during the initial period after MI, they became increasingly more likely to die from NAD subsequently. In male patients, the risk of AD is persistently higher than NAD throughout the 24-months period. Thus, female patients have a lower risk of AD but this effect only appeared between 6 and 12 months. We found this cross-over of mortality risks among female patients intriguing and the mechanism is unclear. It is our speculation that such discrepancy may be explained by the more pronounced vagal recovery and inhibition of sympathetic activity in the women than men following AMI,16 which may have protected against early coronary occlusion-induced ventricular arrhythmias17 but further study is required to examine the interaction between gender and AD.

We also demonstrated a significantly higher ratio of AD to NAD in patients with a history of hypertension. It is speculative to suggest that the mechanism for such an interaction may be related to infarct expansion and ventricular remodelling. Antecedent hypertension, being a major determinant of ventricular afterload, is a strong predictor of infarct expansion18 and ventricular hypertrophy. Following an MI, further hypertrophy could develop in the non-infarcted muscle as part of the remodelling process and ventricular hypertrophy is a strong risk factor of ventricular arrhythmia and sudden death.19 Furthermore, patients with antecedent hypertension have a particularly heightened level of sympathetic activity20 which is exacerbated further after an MI21 as a consequence of infarct-induced denervation hypersensitivity22 and increases specifically the risk of AD relative to NAD in this subgroup of patients.

There were significant interactions between modes of deaths with heart rate, age, and systolic blood pressure but only when considering survival time from date of MI in TRACE and DIAMOND-MI studies.

Clinical implications

This study has shown that even in the current thrombolytic era, AD remains the major cause of death in patients with reduced LVEF or frequent VPBs after MI. The risk of dying from AD is highest in the first 6 months after AMI. Thus, when treating an individual patient, the probability of benefit from a preventative antiarrhythmic intervention will be highest if the treatment is given in the first 6 months after MI when the risk of AD is greatest. However, although the risk of AD decreased significantly after this period, patients remained more likely to die of arrhythmic than non-arrhythmic causes. Thus, antiarrhythmic treatment such as ICD may still offer benefit 2 years after MI. As discussed previously, both our study and the MADIT II trial substantiated the claim that risk stratification and prophylactic treatment for antiarrhythmic death after AMI remains a worthwhile exercise although future controlled trial of antiarrhythmic interventions should enrol patients within the first 6 months after AMI in order to recruit population of patients with the highest arrhythmic event rate.

Conclusions

AD remains the major cause of death in patients with reduced LVEF or frequent VPBs after MI. The risk of dying from AD (or NAD) is highest in the first 6 months after AMI. Prophylactic antiarrhythmic treatment should be targeted during this vulnerable period when the risk is highest. However, although the risk of AD decreased significantly after this period, patients remained more likely to die of AD than non-arrhythmic causes. Thus, antiarrhythmic treatment may still offer benefit up to 2 years after MI. Both our study and the MADIT II trial substantiated the claim that risk stratification and prophylactic treatment for AD after AMI remains a worthwhile exercise although future controlled trial of antiarrhythmic interventions should enrol patients within the first 6 months after AMI in order to recruit population of patients with the highest arrhythmic event rate.

Limitations

The mortality endpoints in our study were not quite identical between the trials but were close enough to allow data pooling. The modes of death were determined by event committees. Although the exact mode of death cannot be determined for certain, such method of classifying the mode of death is the only method that is available and proved to be consistent and sufficiently accurate as demonstrated in the ICD trials. The patients in our study were pre-selected MI patients with reduced LVEF or ventricular arrhythmias on Holter monitoring. Therefore, the results in our study could not be applied across the general MI population. The proportion of patients receiving thrombolytic therapy was low in our study compared with currently expected standard and no information was available on the use of percutaneous reperfusion therapy and novel antiplatelet treatments such as glycoprotein IIb/IIIa receptor inhibitors or clopidogrel, which may limit to some degree the applicability of the findings to modern practice. Furthermore, the limited use of ACE-inhibitors and beta-blockers in the study population compared with current practice may be relevant particularly in relation to ventricular remodelling and affect the outcome after the index event. Finally, the follow-up period in our study is relatively short with cut-off period at 2 years, the mortality outcomes after this period is unknown.

Acknowledgements

This study was supported by a British Heart Foundation project grant (No.PG/98006). Y.G.Y. was a British Heart Foundation Research Fellow in Cardiology. A.J.C. is British Heart Foundation Professor of Clinical Cardiology.

Appendix

Let hA and hC denote the hazard for AD and NAD, respectively. Suppose HRA and HRC denote the hazard ratios for AD and NAD between patients in group 1 (risk factor present) and group 2 (risk factor absent), respectively. Then, Math Hence, if we calculate the ratios of incidence of AD to the incidence of NAD for those with the chosen risk factors and those without, the ratio of these incidence ratios is the ratio of the two hazard ratios.

References

View Abstract