Author + information
- Received March 26, 1998
- Revision received July 21, 1998
- Accepted September 15, 1998
- Published online January 1, 1999.
- Dietrich Andresen, MD, FACCa,* (, )
- Gerhard Steinbeck, MD∗,
- Thomas Brüggemann, MSca,
- Dirk Müller, MD†,
- Ralph Haberl, MD∗,
- Steffen Behrens, MD†,
- Ellen Hoffmann, MD∗,
- Karl Wegscheider, PhD‡,
- Rüdiger Dissmann, MD§ and
- Hans-Christoph Ehlers, MDa
- ↵*Address for correspondence: Prof. Dr. Dietrich Andresen, Urban-Krankenhaus, Medizinische Klinik I, Dieffenbachstrasse 1, 10967 Berlin, Germany
Objectives. We prospectively performed a two-step risk assessment in patients in the early phase after acute myocardial infarction (MI).
Background. Noninvasive methods like Holter electrocardiographic monitoring (HM) and determination of the left ventricular ejection fraction (EF) as well as the invasive technique of programmed ventricular stimulation (PVS) have been used to identify patients in the late phase after MI as candidates for prophylactic implantation of a cardioverter/defibrillator. However, it is unclear whether these results can be transferred to patients following acute MI.
Methods. A series of 657 patients with acute MI (≤75 years) underwent HM and EF. If one of the two methods yielded abnormal findings (HM ≥20 ventricular ectopic beats/h/≥10 ventricular pairs/day/ventricular tachycardia; EF ≤40%), PVS was done (abnormal PVS: induction of monomorphic ventricular tachycardia, duration >10 s, cycle length ≥230 ms).
Results. Of 657 patients, 304 (46%) had either an abnormal HM or EF. The PVS performed in 146 of 304 patients was abnormal in 22. During a mean follow-up of 37 months, there were 106 (16%) deaths, being sudden in 24 (3.6%), nonsudden cardiac in 45 (6.8%). The incidence of arrhythmic events (sudden cardiac death, symptomatic ventricular tachycardia, cardiac arrest) was 18% (4/22) with an abnormal PVS and only 4% (5/124) with a normal PVS (odds ratio 4.0, p = 0.032).
Conclusions. The rate of arrhythmic events is low in post-MI patients in the 1990s. Nevertheless, a two-step risk stratification is helpful in selecting candidates for a defibrillator trial aiming at primary prevention of sudden cardiac death after MI.
Identification of high-risk post-myocardial infarction (MI) patients prone to sudden cardiac death (SCD) still remains an unresolved clinical problem. Several studies have shown that about 5% to 10% of patients suffer SCD within the first year after the acute event (1–4). These studies, predominantly performed in the 1980s, used 24-h Holter electrocardiographic monitoring (HM) and left ventricular ejection fraction (EF) as valuable methods of prognostic assessments (2–4). Both spontaneous ventricular arrhythmias and reduced EF have since been regarded as so-called classical risk factors. However, risk assessment by these methods was mainly performed in the prethrombolytic era. Thus, it is unclear whether a relation still exists between these risk factors and an unfavorable outcome.
More recently, programmed ventricular stimulation (PVS) has been suggested for risk stratification (5,6). Some consider PVS a better predictor of SCD than noninvasive methods (6–8). However, PVS is an invasive procedure involving risks. It would also be time-consuming and expensive if applied as a screening test for all post-MI patients. It would therefore be preferable to limit PVS to a preselected group of high-risk patients. Such an approach has been suggested (5,7,9)but not evaluated in a large prospective trial.
We therefore addressed the question of whether PVS could be limited to those high-risk patients previously identified by noninvasive methods. For this purpose, screening tests were performed with HM and radionuclide ventriculography (RNV). The PVS was then carried out only in those patients with abnormal HM or RNV findings.
A total of 657 patients with acute MI were enrolled in the study between January 1990 and March 1993. Patients included were under 76 years of age, had given informed consent, and could undergo HM and RNV. Ethical approval was obtained from the Ethics Committees of both the Universities of Berlin and Munich.
Within the first 5 days after hospital admission, the following clinical data were recorded: demographic data, history of ischemic heart disease or anti-ischemic therapy, risk factors for coronary artery disease, location of the index infarction, maximum level of serum creatine kinase and thrombolytic therapy, if applied.
HM was performed between days 10 and 14 after the acute event. The following data were calculated: total number of ventricular ectopic beats (VEB), ventricular couplets (VC) and ventricular tachycardia (VT) (≥3 consecutive VEB). An average of ≥20 VEB per hour and/or ≥10 VC per day and/or VT with a cycle length of ≤600 ms were predefined as abnormal. This definition was derived from results of previous risk stratification studies in which these criteria were highly predictive of subsequent cardiac events (3,4,10).
The RNV was performed between days 10 and 14 after MI. A multigated acquisition in the left anterior oblique projection (30–45°) was performed with a caudal tilt of 10° after in vivo technetium-99m labeling of the blood pool. The EF was determined semiautomatically, which correlated highly to contrast ventriculography (r = 0.85 to 0.90) (11). An EF ≤40% was predefined as abnormal.
Programmed ventricular stimulation
Those patients with either an abnormal HM or an abnormal EF ≤40% were selected to undergo PVS. The electrophysiologic study was performed between days 10 and 14 after the acute event using a programmable stimulator (Biotronik, UHS 20) delivering rectangular pulses of 2-ms duration at twice the diastolic threshold. A quadripolar catheter was percutaneously inserted in the right ventricle. One and two premature extrastimuli with increasing prematurity up to refractoriness were applied during sinus rhythm and during ventricular pacing with cycle lengths of 600, 500 and 400 ms for eight paced beats at both the right ventricular apex and outflow tract. An abnormal response was defined as the induction of a sustained monomorphic VT of at least 10-s duration and a cycle length of ≥230 ms. The protocol was also stopped if a VT with a cycle length <230 ms or ventricular fibrillation/flutter was induced (unspecific response, not regarded as abnormal).
Patients discharged were followed-up for a minimum of 2 years. Patients were asked about palpitations, syncope, resuscitation, hospital admission due to heart failure, reinfarction or changes in their medical treatment. In case of an event, the general practitioner or hospital was contacted for further detailed information. When the patient died, an eyewitness account was sought. Causes of death were established from written records and interviewing witnesses and attending physicians. Sudden cardiac death (SCD) was defined strictly as witnessed instantaneous unexpected death without any prodrome or unexpected death during sleep. If death was caused by progressive myocardial dysfunction, it was defined as non-SCD. An arrhythmic event was defined as SCD, documented sustained VT, or ventricular fibrillation.
Pharmacological treatment, including beta-blockers and antiarrhythmic drugs, was left to the discretion of the individual physician. Therapy with beta-blockers, antiarrhythmic drugs and digitalis was recorded during follow-up.
The statistical evaluation was performed using SPSS, version 6.0. Continuous variables are reported as arithmetic means ± SD. Binary variables are given as absolute and relative frequencies (rates). Univariate comparisons between rates were carried out using simple Cox regression models. Multivariate analyses were performed using stepwise Cox regressions (12). In each case, a forward-stepping algorithm was applied using likelihood ratio tests at each step. Variables were included in the model if p values were below 0.05. The reported relative risks are hazard ratios adjusted for all other variables in the final models. In addition, 95% confidence intervals and p values were likewise derived from these models. Survival curves were estimated using the Kaplan-Meier life table analysis and compared by log-rank tests (13).
The clinical characteristics of the study population are shown in Table 1. Sixty-one percent of the patients received thrombolysis, and 12% were discharged on antiarrhythmic therapy.
Prevalence of spontaneous arrhythmias and low ejection fraction
The prevalence of arrhythmias during HM is shown in Table 2. The HM was abnormal in 143/657 (22%) patients, EF in 225 (34%). Either HM or EF was abnormal in 46% (304/657) of the patients. An abnormal EF but normal HM was found in 162 patients (25%), an abnormal HM but normal EF in 80 patients (12%), and both an abnormal HM and EF in 62 patients (9%).
Response to PVS
Programmed ventricular stimulation was performed in 146 cases (48%) but could not be performed in 158. The reasons are listed in Table 3. The response was abnormal in 22 (15%), and normal or unspecific in 124 cases (Table 4). Patients with an abnormal response had a somewhat lower EF and, interestingly, a somewhat lower rate of spontaneous arrhythmias. However, the differences were not statistically significant (Table 5). The stimulation procedure caused no vascular or neurologic complications.
During the 37 ± 17 months follow-up, 106 (16%) patients died: The cardiac mortality was 11% (69/657: SCD in 24 and heart failure in 45). Noncardiac disease was the cause of death in 21 patients (malignant disease, 7; apoplexy, 6; pneumonia, 2; kidney disease, 3; sepsis, 3). A further 16 patients died of unknown causes. Three additional patients presented with symptomatic sustained VT, and five were resuscitated from ventricular fibrillation not associated with an acute MI. Thus, arrhythmic events occurred in 32 of 657 patients (5%).
Predictive value of noninvasive tests
Predictive value of programmed ventricular stimulation
Four of 22 patients (18%) with abnormal PVS had a subsequent arrhythmic event in contrast to only 5 of 124 (4%) with a negative PVS (odds ratio 4.0, p = 0.032; Fig. 1). Interestingly, the rate of non-SCD did not differ between patients with abnormal and normal PVS results (13.6% vs. 12.1%, p = 0.65).
Analysis of other risk factors
An increased rate of arrhythmic events was associated with the presence of diabetes mellitus, a history of previous MI, the absence of thrombolysis, ≥10 VC per day, VT during HM and a low EF (≤40%) in the univariate analysis (Table 6). However, according to the multivariate analysis, the only independent significant predictors of an arrhythmic event were ≥10 VC during HM and the presence of diabetes mellitus (Table 7).
A significant relation to non-SCD was identified for hypertension, diabetes mellitus, digitalis therapy, previous MI, ≥20 VEB per hour, ≥10 VC per day during HM, EF ≤40% and advanced age in the univariate analysis (Table 6). Multivariate analysis demonstrated an independent relation to non-SCD for a reduced EF, digitalis therapy, diabetes mellitus and ≥10 ventricular couplets per day (Table 7).
Prevalence of spontaneous arrhythmias and a low ejection fraction in the thrombolytic era
This investigation differs from previous risk stratification studies in one essential point. Nearly two-thirds of our patients underwent thrombolytic therapy. However, this intervention had no appreciable effect on the prevalence of ventricular arrhythmias or a reduced EF. The EF was ≤40% in 34% of the patients, and ≤30% in 15%. Bigger et al. (3), who studied an infarction population comparable to ours from 1979 to 1980, obtained almost identical results (EF ≤40%: 32%, EF ≤30%: 13%). Likewise, the MILIS study reported a similar proportion of patients with an EF below 40% (24% of patients). Results were comparable for spontaneous VEB (4).
Predictive value of HM and RNV
The rate of arrhythmic events was about three times higher with than without VT or frequent VC (see Table 6), which is consistent with results of other studies performed in the thrombolytic era (14–16).
Both HM and RNV were highly efficient in identifying a group of MI survivors with an excellent prognosis. Less than 1% of the patients with a normal HM and RNV died suddenly during the first year after MI (Fig. 2). This fully corroborates the results of other studies unanimously indicating an approximately 1% arrhythmic death risk per year in MI patients with normal noninvasive test results (6,9,14,17).
In summary, our study clearly demonstrates that the so-called classical risk factors have lost none of their predictive value in the thrombolytic era. It appears to be of paramount clinical value that HM and RNV are effective in identifying a low-risk group that can dispense with more extensive risk stratification, the benefits being outweighed by risks and costs, and which does not require prophylactic therapy.
Predictive value of PVS
Programmed ventricular stimulation has been suggested as a risk stratification method in patients following acute MI (6–8). Richards et al. (6)compared the prognostic value of PVS with that of other noninvasive methods (HM, RNV, signal-averaged electrocardiogram and exercise testing). The authors were able to show that arrhythmic events were most reliably predicted by PVS.
These results did not remain unchallenged, however (18,19). Marchlinski et al. (18)found inducible sustained monomorphic VT to be of low predictive value. Only 1/10 inducible, but 3/13 noninducible post MI-patients developed SCD during follow-up. Similar results were also obtained in other studies either using less aggressive protocols or including considerably smaller numbers of patients (20–22). It remains unclear whether these discrepant findings are due to differences in the stimulation protocols, different patient populations or timing of PVS after MI. In any case, the Australian study (5)has the advantage of having recruited by far the largest patient population.
The positive predictive value of 18% that we obtained is relatively low compared to the results of other stratification studies (5,6,17). Possible causes include:
1 Study protocol
At the beginning of our trial, there was little consensus as to what constitutes the optimal protocol in MI survivors. Because our aim was to stimulate a patient cohort that had never experienced an arrhythmic event, we wanted to obtain as specific a result as possible with an acceptable sensitivity. Thus, we applied only two extrastimuli. However, the results of studies published later (5,23)suggest that future risk-stratification studies should use three extrastimuli. Especially the results of the Multicenter Automatic Defibrillator Implantation Trial (24,25)seem to justify a more aggressive stimulation protocol.
2 Timing of stimulation
The PVS in our study was performed between days 10 and 14 after the acute event. If obtained in the later postinfarction period PVS might be somewhat more predictive (8,23). However, as arrhythmic events have a particularly high mortality during the first months after MI, some patients may already die before the risk stratification can be performed. For this reason and for other practical considerations, the method should be applied before hospital discharge.
Two-step risk stratification
Our study is the first to assess a two-step risk stratification strategy with a prospective design and also to examine a relatively large patient population by PVS.
Investigating patients preselected noninvasively, we demonstrated an almost threefold higher arrhythmic event rate in those with an abnormal PVS than in those whose risk assessment was based solely on noninvasive tests. In a retrospective analysis of their data, Bourke et al. (5)reported similar findings. Spontaneous arrhythmic events occurring in 28% of the patients who had an EF below40% and were inducible by PVS would only have been predicted in 6% if risk assessment had been based solely on a low EF.
Thus, a two-step procedure has two major advantages: 1) It spares many patients with an excellent prognosis a stressful and expensive examination that also can cause complications, and 2) it improves the diagnostic accuracy to such an extent as to enable specific prophylactic therapy of selected patients.
Low incidence of arrhythmic events
Within the first year, only 2% of our patients died suddenly and two patients were resuscitated from ventricular fibrillation. Only three (0.4%) patients developed documented sustained VT during the 3-year follow-up.
What are the possible explanations for the low rate of arrhythmic events?
In contrast to patients in previous studies (1–4), about two-thirds of our population received thrombolytic therapy, and almost 20% underwent angioplasty within the first 14 days after the MI. A reduction of the overall cardiac mortality in the thrombolytic era has been substantiated by the major studies (26–28). Our study is the first to indicate that this reduction also applies to SCD and other nonfatal arrhythmic events.
The exact way in which thrombolysis acts in this situation has not been conclusively clarified. The beneficial effect can hardly be explained by a reduction of spontaneous ventricular arrhythmias or an improvement of the EF, as the prevalence of both parameters has essentially remained unchanged in the course of time (5,6). It may be conjectured, however, that the infarction vessel assumed to be patent after thrombolysis leads to a more homogeneous area of infarcted tissue less prone to reentrant arrhythmias. This effect would also explain why a reduction of late potentials assessed by signal-averaged electrocardiogram (ECG) was observed after thrombolysis in several studies (29,30).
2 Definition of SCD
Several studies defined SCD as death within 1 h after the onset of symptoms (17,31). However, this probably does not exclude secondary arrhythmic deaths due to prolonged ischemia (e.g., reinfarction) or acute heart failure. Our study gave exclusive consideration to the presumed arrhythmic-death incidence by only documenting SCD in cases of a sudden fatal collapse without prior circulatory distress or unexpected death during sleep.
3 Withdrawal of antiarrhythmic drugs
Twelve percent of our patients were discharged on antiarrhythmic drugs, the majority on amiodarone or sotalol. Only 3% received class I agents. The Cardiac Arrhythmia Suppression Trial (CAST) has been instrumental in substantiating the speculation that refraining from class I antiarrhythmic treatment may favorably influence the sudden death rate (32).
Therapy of acute MI in the thrombolytic era does not appear to have exerted a major effect on the prevalence of spontaneous ventricular arrhythmias and reduced EF. Conversely, cardiac mortality and especially the incidence of arrhythmic events have substantially decreased in the postinfarction period. Risk stratification using HM and RNV is still helpful in identifying patients at risk. A two-step stratification using noninvasive methods first, followed by PVS, clearly increases the positive predictive value for subsequent arrhythmic events. However, this stratification strategy has a positive predictive value of only 18% and can thus only serve as a basis for an intervention trial if the prophylactic therapy applied is highly effective and has a low rate of side effects. The implantable cardioverter/defibrillator is the only treatment available today that prevents sudden death with overwhelming efficacy, has a low perioperative mortality and offers an acceptable quality of life. Thus, we believe that the two-step risk stratification described in this study can be usefully applied in a large-scale defibrillator trial examining acute MI patients for primary prevention of sudden death.
We are indebted to Gerlinde Langenscheidt, Dominique Rumor, Petra Nimmermann and Ruth Brennfleck for managing the patient screening and follow-up; to Klaus Wudel, MD, and C. M. Kirsch, MD, for performing and analyzing RNV; to Ursula Kropeit and Eva Hadad for technical assistance, and to Gisela John for secretarial assistance. In addition to the university hospitals of Berlin and Munich, the following hospitals recruited study patients: The Privatklinik Dr. Schindelbeck, Herrsching (Brigitte Cabell, MD) and Kreiskrankenhaus Fürstenfeldbruck, (Wolfgang Decker, MD).
☆ This work was supported in part by a grant from the Deutsche Forschungsgemeinschaft.
- left ventricular ejection fraction
- 24-h Holter electrocardiographic (ECG) monitoring
- myocardial infarction
- programmed ventricular stimulation
- radionuclide ventriculography
- sudden cardiac death
- ventricular ectopic beats
- ventricular couplets
- ventricular tachycardia
- Received March 26, 1998.
- Revision received July 21, 1998.
- Accepted September 15, 1998.
- American College of Cardiology
- Ruberman W,
- Weinblatt E,
- Goldberg D.J,
- Frank C.W,
- Chaudhary B.S,
- Shapiro S
- Multicenter Postinfarction Research Group,
- Bigger T.J,
- Fleiss J.L,
- Kleiger R,
- Miller J.P,
- Rolnitzky L.M
- Bourke J.P,
- Richards D.A.B,
- Ross D.L,
- et al.
- Richards D.A,
- Byth K,
- Ross D.L,
- et al.
- Cripps T,
- Bennett E.D,
- Camm A.J,
- et al.
- Farell T.G,
- Bashir Y,
- Cripps T,
- et al.
- Andresen D, Bethge KP, Boissel JP, et al. for the European Infarction Study Group
- Anderson S,
- Augier A,
- Hanck W,
- Oaks D,
- van Dale W,
- Weisberg N
- ↵Norusis MJ. SPSS for Windows, Advanced Statistics, Release 6.0 SPSS, Chicago 1993.
- McClements B.M,
- Adgey J
- Maggioni A.P,
- Zuanetti G,
- Franzosi M.G,
- et al.
- Kuchar D.L,
- Thorburn C.W,
- Sammel N.L
- Marchilinski F.E,
- Buxton A.E,
- Waxman H.L,
- et al.
- Roy D,
- Marchand E,
- Theroux P,
- et al.
- Kersschot I.E,
- Brugada P,
- Ramentol M,
- et al.
- Wilber D.J,
- Olshansky B,
- Moran J.F,
- Scanlon P.J
- ISIS-4 (Fourth International Study of Infarct Survival) Collaborative Group
- Hinkle L.E,
- Thaler J.T