- Email: [email protected]
Secondary prevention trials after acute myocardial infarction
SecondaryPreventionTrials After Acute MyocardialInfarction CURT
D.
FURBERG,
Of those patients who reach the hospital after an acute myocardial infarction, 18% die during their stay and 85% to 90 % of the remainder will eventually die of coronary artery disease. Several secondary preventive approaches have been made to prolong life in these patients. Long-term controlled trials involving nonsurgical measures and at least 100 patients will be reviewed. Lipid-lowering regimens have shown no demonstrable effect on survival over a 4- to 6-year period but show some benefit with respect to nonfatal infarction. Survival was not improved essentially by anticoagulants, antiarrhythmic agents or calcium channel blockers, although new trials are underway
MD
that might clarify their role. Platelet-active drugs achieved little reduction in mortality but showed benefit in nonfatal infarction (30% reduction with aspirin). Pooled data on physical exercise programs demonstrated a 15 % benefit on mortality but larger trials are required to confirm this. The data on ,8 blockers (particularly those without intrinsic sympathomimetic activity) show that these drugs improve long-term survival after myocardial infarction, reducing all-cause mortality by as much as 25% to 30 % . Larger trials are necessary to detect statistically significant reductions in mortality both overall and in selected subgroups of patients. (Am J Cardiol 1987;60:28A-32A)
A
pproximately 40% of patients who suffer a myocardial infarction die before receiving medical attention and this enormous proportion emphasizes the importance of primary prevention. Of those who reach hospitals in the U.S. 18% die during their hospital stay and, although the risk of dying decreases after discharge, mortality is still high after the convalescent period. The stabilized annual mortality rate is about 4 to 8 times that of a comparable noncoronary population after 6 months. The magnitude of the problem is enormous despite a dramatic decrease in coronary mortality due in part to the success of primary preventi0n.l In the US. alone, of almost 700,000 persons admitted annually with acute myocardial infarction as a primary diagnosis, 550,000 are discharged alive and this group is the target population for long-term secondary preventive measures.2 Between 85% and 90% of these patients will eventually die as a result of their coronary artery
disease and therefore secondary preventive intervention must decrease the incidence of coronary mortality, including reinfarction and sudden death, if it is tb be successful.
SecondaryPreventionMechanisms Several approaches have been made to try to prolong life in patients who have survived the acute phase of myocardial infarction. Figure 1 summarizes the presumed mechanisms by which patients die and the interventions designed to interfere with each pathophysiologic process. The most fundamental cause of reduced oxygen supply is coronary atherosclerosis and lipid-lowering regimens with diet or drugs may slow its rate of progression. Anticoagulants and plateletactive drugs have been given in an attempt to prevent formation or extension of coronary thrombi. Sudden death is a common cause of death in this population and antiarrhythmic agents have been used in an attempt to prevent ventricular fibrillation. Finally, (3 blockers, calcium channel blockers and physical exercise programs have all been used in the hope that they might improve the balance between myocardial oxygen supply and demand. These interventions have all been assessed in controlled clinical trials. A number of other interventions have not been adequately tested in long-term clinical trials. These in-
From the Center for Prevention Research and Biometry, The Bowman Gray School of Medicine, Wake Forest University, Winston-Salem, North Carolina. Address for reprints: Curt D. Furberg, MD, Center for Prevention Research and Biometry, The Bowman Gray School of Medicine, Wake Forest University, 300 South Hawthorne Road, Winston-Salem, North Carolina 27103. 28A
July 15, 1987
elude the use of antihypertensive agents, inotropes, afterload-reducing agents and antismoking regimens. Thrombolytics, hyaluronidase, afterload-reducing agents and intervention in the glucose-insulin-potassium system have been tested in short-term trials. This review of nonsurgical secondary preventive measures is based on randomized, controlled, longterm (generally I1 year) trials with a total sample size of at least 100 patients (described in full elsewhere3F4). The endpoint normally used was total mortality, which usually obviates any element of judgment, but the effect of treatment on the incidence of nonfatal myocardial infarction will also be discussed in relation to some of the trials. The restriction to randomized trials was considered necessary to incorporate some protection against treatment allocation bias, and the minimum sample size was chosen so that the results had sufficient statistical power to discern clinically important effects or trends.
Results Lipid-lowering agents: Several randomized controlled trials have now been undertaken to determine the effect on morbidity and mortality of diet or longterm administration of lipid-lowering agents in patients with established coronary artery disease. Ten trials met the definition for this review; the numbers of patients ranged between 152 and 8,341. Five different interventions were examined: diet, estrogen, dextrothyroxine, clofibrate and nicotinic acid. Some trials used more than 1 active treatment and some used drug combinations. Cholesterol was reduced by an average of 4% to 20%. Side effects were fairly common with pharmacologic intervention, however, and in 3 of the trials treatment was stopped because of adverse effects attributed to the intervention. The dietary studies suffered from difficulties with compliance. Of the early trials of hormones, 5 reported results in favor of control treatment, 2 in favor of active intervention and 1 was equivocal. Five of the trials of clofibrate and nicotinic acid produced equivocal results. The exception was the Newcastle trial,5 which reported significant results but cardiac mortality was used as an endpoint and total mortality data have not been released. Three clinical trials showed some benefit from diet although the effect was small with wide confidence intervals. Woodhill et al6 reported an opposite effect; overall, however, diet does seem promising in secondary prevention. Assessment of the effect of lipid lowering in nonfatal myocardial infarction showed that active intervention did better than control in all but 1 trial. These results were both consistent and convincing. As tested in these unselected groups of myocardial infarction survivors, there is no evidence of improved survival, at least over a 4- to 6-year period. There are now data from the Coronary Drug Project7 suggesting that it may take over 10 years for the benefits of lipid lowering to be reflected in prolonged survival, Anticoagulants: The rationale for the use of anticoagulants in the acute phase of myocardial infarction is based on the expectation that they would prevent ex-
THE AMERICAN
JOURNAL
OF CARDIOLOGY
Volume 60
Progressive coronary atherosclerosis
Lipid-lowering
Coronarythrombosis
Anticoagulants Platelet-active
drugs
Ventricularfibrillation
Antiarrhythmics
increased
Beta-blockers Exercise Calcium blockers
0, demand
FIGURE 1. Long-term prevention trials after myocardial morbid mechanisms and tested interventions.
29A
infarction:
tension of the coronary thrombus and reduce the risk of embolism from mural thrombus and pulmonary embolism from deep venous thrombosis. Their long-term benefit is more uncertain, however, aiming at prevention of subsequent coronary thrombus. Seven trials met the definition for inclusion in this review; there were quite a number with questionable randomization that have not been included. Numbers ranged from 159 to 912, and either coumarin or indandione derivatives were used, sometimes with initial heparin therapy. Successful anticoagulation was demonstrated in all the trials. Two of the trials favored control therapy and some were equivocal. Among those favoring active treatment, one was a small study from Norway with only 11 deaths and one was a withdrawal study that had an inappropriate design for evaluation of efficacy. None found a statistically significant difference for overall mortality between the intervention and control groups. Results showed, therefore, that long-term survival does not seem to be improved substantially by anticoagulants although these agents do reduce mortality in the acute in-hospital phase of myocardial infarction. Platelet-active drugs: The 9 trials of platelet-active drugs are more recent, which is reflected by their larger sample size (626 to 4,524).Aspirin, dipyridamole and sulfinpyrazone were the drugs studied and they all achieved modification of platelet function. Most of the trials showed benefit, with 2 exceptions-the largest aspirin trial (Aspirin Myocardial Infarction Study) and the second trial of sulfinpyrazone (Anturan Reinfarction Italian Study). If the data are pooled [thus disregarding some of the differences in population, dosage, time of initiation of therapy and length of follow-up] aspirin [3OOto 1,500 mg daily) produced a reduction in mortality of around 10%. The addition of dipyridamole to aspirin did not add any benefit, the only difference being the higher cost of this combination treatment. All the trials showed benefit in nonfatal myocardial infarction, however, and 2 of the trials, including the large AMIS trial, achieved statistical significance, When all the data on aspirin are combined, this agent shows a 30% benefit with relation to nonfatal myocardial infarction. The Food and Drug Administration has now approved aspirin for preventive use after myocardial infarction.
30A
A SYMPOSIUM:
CARDIOPROTECTION
IN ISCHEMIC
HEART DISEASE-FACT
Antiarrhythmic agents: The 7 antiarrhythmic trials that met the review criteria were all small with only 630 patients in the largest trial. Four different drugs were evaluated in these studies-dilantin, tocainide, mexiletine and aprindine-and all trials demonstrated suppression of ectopic beats. Investigators encountered a number of adverse effects leading to a high withdrawal rate. In 6 of 7 of these trials, post-myocardial infarction patients were randomized regardless of whether they had a ventricular arrhythmia. In most of the trials the control group fared better than the treated group although 2 showed a favorable trend in the actively treated group. That mortality was higher in the intervention group in 4 of these trials suggests that treatment with antiarrhythmic drugs might actually be harmful in these patients. It has not been demonstrated, therefore, that arrhythmia control in postinfarction patients improves prognosis, and adverse effects occurred frequently in patients receiving the various antiarrhythmic drugs. The relevance to survival of the proarrhythmic effect attributed to most antiarrhythmics is unclear and this factor may have contributed to these results. Physical exercise: A total of 6 prospective, randomized trials have reported the effects of physical exercise training as a secondary preventive measure in patients who have survived a myocardial infarction. Patient numbers ranged from 203 to 1,777 and intervention consisted of individualized supervised training for approximately 45 minutes 2 to 3 times a week. Some studies involved supporting educational programs. As might be expected, the exercise capacity was increased in the trained patients. Study populations were small and none of the results was statistically significant although there was suggestion of benefit in 4 of the trials. One trial from Canada showed an opposite trend and one was equivocal. There appeared to be little effect on the incidence of nonfatal reinfarction. Pooling these data, however, demonstrates a 15% benefit of exercise on mortality in spite of problems of poor long-term compliance in these programs (40% to 60% of patients discontinued training of their own accord). Trials of adequate size are needed to determine with certainty the preventive value of physical exercise suggested by these studies. Calcium antagonists: Three long-term trials of reasonable size, with 1,436 to 2,279 patients randomized, were reviewed. Three different calcium antagonistslidoflazine, verapamil and nifedipine-were used. The results of these trials, the Holland Interuniversity Nifedipine Trial of unstable angina and other acute myocardial infarction trials have shown no effect on mortality, although it may still be too soon to judge the preventive effect of this heterogeneous group of drugs. Beta blockers: p blockers possessseveral different properties that may be of potential value to the postmyocardial infarction patient; they are antiischemic and antiarrhythmic, have platelet antiaggregating effects and may possibly enhance the flow properties of erythrocytes. The assumption that the sympathetic
OR FANTASY?
nervous system is important in triggering sudden death led to the testing of these agents in myocardial infarction survivors. A total of 17 trials met the review definition, ranging in size from 111 to 3,837 patients. Eight different P blockers were tested (alprenolol, metopro101,oxprenolol, pindolol, practolol, propranolol, sota101and timolol) and in all of them evidence of ,&receptor blockade was seen. The trials were divided into those evaluating P blockers with intrinsic sympathomimetic activity and those without intrinsic sympathomimetic activity. Beta blockers with intrinsic sympathomimetic activity: Two early Swedish trials of p blockers with intrinsic sympathomimetic activity produced encouraging mortality results, but they were small studies with very wide confidence intervals. Other trials of these 6 blockers have been less promising; the last trial reported negative results. Pooling suggests an overall benefit in the range of 10% reduction in mortality from /3 blockers with intrinsic sympathomimetic activity. p blockers with intrinsic sympathomimetic activity do seem to show greater benefit in reduction of recurrent nonfatal myocardial infarction and in 1 trial the results are significant. Two others are close to significance and active treatment was superior to control in all the trials. Beta blockers without intrinsic sympathomimetic activity: Most of the trials of p blockers without intrinsic sympathomimetic activity showed substantial reductions in mortality, with 1 exception (the Baber study]. Two trials, the Norwegian Timolol Study and the P-Blocker Heart Attack Trial,8 were significant on their own. The largest metoprolol study has not yet been reported, but it is anticipated that the results will not show any benefit from treatment. The pooled data suggest that the benefit from these drugs in terms of decrease in all-cause mortality may be as much as 25% to 30%. In terms of nonfatal myocardial infarction 2 trials of p blockers without intrinsic sympathomimetic activity were significant on their own and most, except for the Baber study, clearly showed benefit of about 25% to 30%) which is similar to that seen for all-cause mortality. The decrease in sudden death (defined as sudden death within 1 hour of onset of symptoms) brought about by these drugs implies an antiarrhythmic mechanism of action. /3 blockers also reduce nonsudden cardiac death, which is assumed to be an antiischemic effect of these drugs, and this view is supported by the results regarding the incidence of nonfatal myocardial infarction.
Discussion Absolute effect on mortality: A number of issues have been raised by these trials. One relates to how we present and interpret efficacy; more emphasis should be placed on the absolute effect of treatment, i.e., how many patients must be treated to benefit one. A 25% relative effect can have very different meanings; if an event rate of 4% is reduced to 3%, then 1 event is prevented for every 100 patients treated. If the mortality rate is higher, at 2O%, and reduced to 1570, 5 events
July 15, 1987
THE AMERICAN
JOURNAL
OF CARDIOLOGY
Volume 60
31A
per 100 will be prevented. With a 10% relative effect, TABLE I Efficacy Measures and Their Clinical Usefulness and a decrease from 20% to 16% in mortality, 2 events Event Rate (%) will be prevented out of 100 [Table I). There are situaRelative Efficacy by No. Efficacy of Patients Treated Intervention tions, therefore, where a 10% relative benefit may be Placebo more important clinically than a 25% benefit. -25% l/100 3.0 4.0 This concept was applied to post-hoc analysis of -25% 5/100 15.0 20.0 data from the P-Blocker Heart Attack Trial9 using a -10% 2/100 18.0 20.0 classification system developed in Gothenburg. The patients were divided into whether they had transient evidence of electrical failure, mechanical failure, both or neither before randomization into the trial. Half of Table II All-Cause Mortality* by Risk Group and Treatment in the the patients were in the uncomplicated group and their &Blocker Heart Attack Trial placebo group mortality over 25 months was 6.6%. For Mortality Rates (% ) those with electrical problems mortality doubled and it tripled in those with mechanical problems, whether or Difference Relative Risk Risk Group Propranolol Placebo not they also had electrical problems. Looking at the .94 -0.4 6.2 6.6 None absolute difference in mortality, however, there was .48 -5.7 5.2 10.9 Electrical very little difference in the low risk group between .62 -6.4 10.4 16.8 Mechanical active and placebo treatment, proving that it is hard to .76 -4.2 12.9 17.1 Both improve on a rate that is already low. Benefit in the * Average follow-up 25 months. high risk group meanwhile, ranged from 4 to 6 lives saved per 100 patients treated [Table II). For the lower risk group, the cumulative mortality curves for placebo and propranolol overlapped. For 18 the combined high risk group, these curves diverged 16 l’ and propranolol showed substantial benefit. The - I shape of the curve also suggests that treatment should be continued for at least 3 years (Fig. 2) Adverse effects and risk: If we accept that high risk patients should be prime candidates for preventive therapy, we have to be prepared to pay a price-a greater incidence of adverse effects. Looking at adverse effects by risk group in the P-Blocker Heart Attack Trial, the incidence of congestive heart failure in the uncomplicated group and the group with electrical complications only was about 5% to 6% over 25 Months months, whether or not they received p blockers. If the FIGURE 2. @Blocker Heart Attack Trial: cumulative mortality by patients had a history of transient mechanical complications and were given placebo, their risk was twice history of complications. Ml = myocardial infarction. that (10%) and if propranolol was added the risk of developing congestive heart failure increased by another 5% (Fig. 3). Caution is required in interpreting consideration and reporting of results. Further analysubgroup findings, particularly those identified after sis of the ,&Blocker Heart Attack Trial data showed the trial was completed, but this illustrates at least the that smaller subgroups are more susceptible to variatheoretical argument for caution in treating high risk tion than the larger groups. patients. The best way of determining whether subgroup Difficulties with subgroups: The danger in sub- findings are real is to repeat the experiment, which can group analyses is that a striking effect may be found in rarely be done easily. Another method is to look for some subgroup by chance alone. This danger has been replications elsewhere. To this end investigators from illustrated by data from the P-Blocker Heart Attack 9 trials of long-term p blockers conducted subgroup Trial. A total of 146 subgroups was examined compar- analyses, using common definitions, aiming to pool the ing the absolute mortality rates for propranolol and subgroup findings as appropriate. The endpoint was placebo treatment. Few of these groups were defined standardized as l-year all-cause mortality, using an in advance. Most analyses were exploratory but some intention-to-treat analysis. were done in order to compare findings with subThe preliminary results of the P-Blocker Pooling groups of interest in other P-blocker trials. The most Project show a 25% overall decrease in mortality. High striking finding to emerge from this pooling exercise risk groups, such as those with transient mechanical was the enormous variability in absolute mortality dif- complications, showed large absolute benefits with ,f3ferences between subgroups; it ranged from a “harm- blocker therapy. These results confirm the earlier obful” effect of 5% to a “beneficial” effect of 11%. It is servation in the ,&Blocker Heart Attack Trial. Unexdifficult to ascertain which effects are real and which pectedly, patients with electrical complications did not are due to chance and there is a danger in selective show a high mortality rate, but the ,&Blocker Pooling
1
/-
32A
A SYMPOSIUM:
16 -
Congestive
CARDIOPROTECTION
IN ISCHEMIC
HEART DISEASE-FACT
heart failure
15 14 13 -
El
Placebo
@J
Propranolol
12 11 10 9 %
87 6.1 6 5 4 3 2 1 n” i No risk
Electrical complication
Mechanical complication
Electrical and mechanical
FIGURE 3. Long-term prevention trials afler myocardial infarction: adverse effects by risk group in the &Blocker Heart Attack Trial.
Project definition was vague and different from that adopted in the P-Blocker Heart Attack Trial analysis. Recently reported findings from the ,&Blocker Heart Attack TriallO showed that the relative benefit of propranolol treatment was similar in patients with complex ventricular arrhythmias at baseline and those with no such arrhythmia (31% and 2570, respectively]. The former subgroup made up 40% of the patients in the trial and had a much higher mortality. In fact, by treating patients with complex arrhythmias 5 lives are saved for every 100 patients treated. This compares with the group without complex arrhythmia, in which 1.6 lives would be saved per 100 treated. Again, this demonstrates that absolute differences in mortality are an important consideration in the decision regarding which patients should be treated with propranolol after a myocardial infarction.
Conclusions This has been a review of 57 trials involving over 61,000patients. Some trials made more than 1 comparison and so, effectively, the total of trials becomes 67. Simple statistics show that, by chance alone, 3 findings should be significant at the level of p <0.05. There were, in fact, 3 results that were significant in this review: 2 in favor of ,f3blockers and one in favor of a lipid-lowering regimen; however, the former p values (e.g., 0.0003 and 0.005)were extremely impressive and this cannot be due to chance. The available data seem to indicate that p blockers, particularly those without intrinsic sympathomimetic activity, improve long-term survival after a myocardial infarction and that the
OR FANTASY?
prime candidates for this type of therapy are patients with complicated infarcts. Platelet-active drugs may have a 10% beneficial effect on survival but a 30% reduction in nonfatal myocardial infarction. It is not known whether this apparent benefit would be additive to that of ,f3 blockers, but these drugs may be useful for those patients with contraindications or poor tolerance of p blockers. Lipid-lowering regimens have so far had no demonstrable effect on survival but it could be that patients have not been treated for long enough and there does appear to be some benefit with respect to nonfatal myocardial infarction. A large trial is needed to determine whether physical exercise might improve survival but the evidence so far is certainly very suggestive of positive benefit. Anticoagulants and antiarrhythmics have not shown demonstrable beneficial effects in the populations studied and in the dosage used but there are new trials underway that may clarify the position of these drugs after a myocardial infarction. It may also be too soon to judge the effect of the calcium antagonists on survival but there is no evidence, so far, that calcium channel blockers have any effect on mortality after myocardial infarction. Many of these interventions may prove to be beneficial in selected patients but we have not been successful in identifying such subgroups. Patients with severe ventricular arrhythmias, for example, may show benefit from antiarrhythmic treatment that is not apparent in a more unselected postinfarction population (which may, in fact, demonstrate an increase in mortality with these drugs]. Using current methodology, reductions of less than 20% are difficult to demonstrate in clinical trials and, in defense of these studies, many were not designed to assessthe effect of treatment on mortality and a large proportion had insufficient statistical power to detect even a true benefit of as much as 30% to 40%.
References 1. Proceedings of the Conference on the Decline of Coronary Heart Disease Mortality. 1979; US DHEW, PHS, NIH publication no. 79: 1610. 2. Final Mortality Statistics. 1978, from the National Center for Health Statistics. 1980; DHEW publication no. [PHS) 29, no. 6, supplement 2:80-1120. 3. Furberg CD, May GS. Effect of long-term prophylactic treatment on surviva1 after myocardiol infarction. Am r Med 1984; suppI 6A:76-83. 4. May GS, Eberlein KA. Furberg CD, Passamani ER, De Mets DL. Secondary prevention after myocardial infarction: a review of long-term trials. Prog Cardiovasc Dis 1982;24:331-352. 5. Five-year study by a group of physicians of the Newcastle upon l’yne region. Trial of clofibrate in the treatment of ischemic heart disease. Br Med J 1971;4:767. 6. Woodhill JM, Palmer AJ, Leelarthaepin B, McGilchrist C, Blackett RB. Low fat, low cholesterol diet in secondary prevention of coronary heart disease. Adv Exp Med Biol 1978;109:317-330. 7. Canner PL, Berge KG, Wenger NK, Stamler J, Friedman L, Prineas RI, Friedewald W, for the Coronary Drug Project Research Group. FifteenYeaT mortality in Coronary Drug Project patients: long-term benefit with niacin. TACC l&6:8:1245-1255. 8. Beta-Blocker Heart Attack Trial Research Group. A randomized trial of propranolol in patients with acute myocardial infarction. I. Mortality results. JAMA 1982;247:1707-1714. 9. Furberg CD, Hawkins CM, Lichstein E. Effect of propronoIoI in postinfarction ootients with mechanical or electrical complications. Circulation 1984;69:76i-765.
10. Friedman LM, Byington RP, Capone RJ, Furberg CD, Goldstein S, Lichstein E, Writing Group for the Beta-Blocker Heart Attack Trial Research Group. Effect of propranolol in patients with myocardial infarction and ventricular arrhythmia. fACC 1986;7:1-8.
D.
FURBERG,
Of those patients who reach the hospital after an acute myocardial infarction, 18% die during their stay and 85% to 90 % of the remainder will eventually die of coronary artery disease. Several secondary preventive approaches have been made to prolong life in these patients. Long-term controlled trials involving nonsurgical measures and at least 100 patients will be reviewed. Lipid-lowering regimens have shown no demonstrable effect on survival over a 4- to 6-year period but show some benefit with respect to nonfatal infarction. Survival was not improved essentially by anticoagulants, antiarrhythmic agents or calcium channel blockers, although new trials are underway
MD
that might clarify their role. Platelet-active drugs achieved little reduction in mortality but showed benefit in nonfatal infarction (30% reduction with aspirin). Pooled data on physical exercise programs demonstrated a 15 % benefit on mortality but larger trials are required to confirm this. The data on ,8 blockers (particularly those without intrinsic sympathomimetic activity) show that these drugs improve long-term survival after myocardial infarction, reducing all-cause mortality by as much as 25% to 30 % . Larger trials are necessary to detect statistically significant reductions in mortality both overall and in selected subgroups of patients. (Am J Cardiol 1987;60:28A-32A)
A
pproximately 40% of patients who suffer a myocardial infarction die before receiving medical attention and this enormous proportion emphasizes the importance of primary prevention. Of those who reach hospitals in the U.S. 18% die during their hospital stay and, although the risk of dying decreases after discharge, mortality is still high after the convalescent period. The stabilized annual mortality rate is about 4 to 8 times that of a comparable noncoronary population after 6 months. The magnitude of the problem is enormous despite a dramatic decrease in coronary mortality due in part to the success of primary preventi0n.l In the US. alone, of almost 700,000 persons admitted annually with acute myocardial infarction as a primary diagnosis, 550,000 are discharged alive and this group is the target population for long-term secondary preventive measures.2 Between 85% and 90% of these patients will eventually die as a result of their coronary artery
disease and therefore secondary preventive intervention must decrease the incidence of coronary mortality, including reinfarction and sudden death, if it is tb be successful.
SecondaryPreventionMechanisms Several approaches have been made to try to prolong life in patients who have survived the acute phase of myocardial infarction. Figure 1 summarizes the presumed mechanisms by which patients die and the interventions designed to interfere with each pathophysiologic process. The most fundamental cause of reduced oxygen supply is coronary atherosclerosis and lipid-lowering regimens with diet or drugs may slow its rate of progression. Anticoagulants and plateletactive drugs have been given in an attempt to prevent formation or extension of coronary thrombi. Sudden death is a common cause of death in this population and antiarrhythmic agents have been used in an attempt to prevent ventricular fibrillation. Finally, (3 blockers, calcium channel blockers and physical exercise programs have all been used in the hope that they might improve the balance between myocardial oxygen supply and demand. These interventions have all been assessed in controlled clinical trials. A number of other interventions have not been adequately tested in long-term clinical trials. These in-
From the Center for Prevention Research and Biometry, The Bowman Gray School of Medicine, Wake Forest University, Winston-Salem, North Carolina. Address for reprints: Curt D. Furberg, MD, Center for Prevention Research and Biometry, The Bowman Gray School of Medicine, Wake Forest University, 300 South Hawthorne Road, Winston-Salem, North Carolina 27103. 28A
July 15, 1987
elude the use of antihypertensive agents, inotropes, afterload-reducing agents and antismoking regimens. Thrombolytics, hyaluronidase, afterload-reducing agents and intervention in the glucose-insulin-potassium system have been tested in short-term trials. This review of nonsurgical secondary preventive measures is based on randomized, controlled, longterm (generally I1 year) trials with a total sample size of at least 100 patients (described in full elsewhere3F4). The endpoint normally used was total mortality, which usually obviates any element of judgment, but the effect of treatment on the incidence of nonfatal myocardial infarction will also be discussed in relation to some of the trials. The restriction to randomized trials was considered necessary to incorporate some protection against treatment allocation bias, and the minimum sample size was chosen so that the results had sufficient statistical power to discern clinically important effects or trends.
Results Lipid-lowering agents: Several randomized controlled trials have now been undertaken to determine the effect on morbidity and mortality of diet or longterm administration of lipid-lowering agents in patients with established coronary artery disease. Ten trials met the definition for this review; the numbers of patients ranged between 152 and 8,341. Five different interventions were examined: diet, estrogen, dextrothyroxine, clofibrate and nicotinic acid. Some trials used more than 1 active treatment and some used drug combinations. Cholesterol was reduced by an average of 4% to 20%. Side effects were fairly common with pharmacologic intervention, however, and in 3 of the trials treatment was stopped because of adverse effects attributed to the intervention. The dietary studies suffered from difficulties with compliance. Of the early trials of hormones, 5 reported results in favor of control treatment, 2 in favor of active intervention and 1 was equivocal. Five of the trials of clofibrate and nicotinic acid produced equivocal results. The exception was the Newcastle trial,5 which reported significant results but cardiac mortality was used as an endpoint and total mortality data have not been released. Three clinical trials showed some benefit from diet although the effect was small with wide confidence intervals. Woodhill et al6 reported an opposite effect; overall, however, diet does seem promising in secondary prevention. Assessment of the effect of lipid lowering in nonfatal myocardial infarction showed that active intervention did better than control in all but 1 trial. These results were both consistent and convincing. As tested in these unselected groups of myocardial infarction survivors, there is no evidence of improved survival, at least over a 4- to 6-year period. There are now data from the Coronary Drug Project7 suggesting that it may take over 10 years for the benefits of lipid lowering to be reflected in prolonged survival, Anticoagulants: The rationale for the use of anticoagulants in the acute phase of myocardial infarction is based on the expectation that they would prevent ex-
THE AMERICAN
JOURNAL
OF CARDIOLOGY
Volume 60
Progressive coronary atherosclerosis
Lipid-lowering
Coronarythrombosis
Anticoagulants Platelet-active
drugs
Ventricularfibrillation
Antiarrhythmics
increased
Beta-blockers Exercise Calcium blockers
0, demand
FIGURE 1. Long-term prevention trials after myocardial morbid mechanisms and tested interventions.
29A
infarction:
tension of the coronary thrombus and reduce the risk of embolism from mural thrombus and pulmonary embolism from deep venous thrombosis. Their long-term benefit is more uncertain, however, aiming at prevention of subsequent coronary thrombus. Seven trials met the definition for inclusion in this review; there were quite a number with questionable randomization that have not been included. Numbers ranged from 159 to 912, and either coumarin or indandione derivatives were used, sometimes with initial heparin therapy. Successful anticoagulation was demonstrated in all the trials. Two of the trials favored control therapy and some were equivocal. Among those favoring active treatment, one was a small study from Norway with only 11 deaths and one was a withdrawal study that had an inappropriate design for evaluation of efficacy. None found a statistically significant difference for overall mortality between the intervention and control groups. Results showed, therefore, that long-term survival does not seem to be improved substantially by anticoagulants although these agents do reduce mortality in the acute in-hospital phase of myocardial infarction. Platelet-active drugs: The 9 trials of platelet-active drugs are more recent, which is reflected by their larger sample size (626 to 4,524).Aspirin, dipyridamole and sulfinpyrazone were the drugs studied and they all achieved modification of platelet function. Most of the trials showed benefit, with 2 exceptions-the largest aspirin trial (Aspirin Myocardial Infarction Study) and the second trial of sulfinpyrazone (Anturan Reinfarction Italian Study). If the data are pooled [thus disregarding some of the differences in population, dosage, time of initiation of therapy and length of follow-up] aspirin [3OOto 1,500 mg daily) produced a reduction in mortality of around 10%. The addition of dipyridamole to aspirin did not add any benefit, the only difference being the higher cost of this combination treatment. All the trials showed benefit in nonfatal myocardial infarction, however, and 2 of the trials, including the large AMIS trial, achieved statistical significance, When all the data on aspirin are combined, this agent shows a 30% benefit with relation to nonfatal myocardial infarction. The Food and Drug Administration has now approved aspirin for preventive use after myocardial infarction.
30A
A SYMPOSIUM:
CARDIOPROTECTION
IN ISCHEMIC
HEART DISEASE-FACT
Antiarrhythmic agents: The 7 antiarrhythmic trials that met the review criteria were all small with only 630 patients in the largest trial. Four different drugs were evaluated in these studies-dilantin, tocainide, mexiletine and aprindine-and all trials demonstrated suppression of ectopic beats. Investigators encountered a number of adverse effects leading to a high withdrawal rate. In 6 of 7 of these trials, post-myocardial infarction patients were randomized regardless of whether they had a ventricular arrhythmia. In most of the trials the control group fared better than the treated group although 2 showed a favorable trend in the actively treated group. That mortality was higher in the intervention group in 4 of these trials suggests that treatment with antiarrhythmic drugs might actually be harmful in these patients. It has not been demonstrated, therefore, that arrhythmia control in postinfarction patients improves prognosis, and adverse effects occurred frequently in patients receiving the various antiarrhythmic drugs. The relevance to survival of the proarrhythmic effect attributed to most antiarrhythmics is unclear and this factor may have contributed to these results. Physical exercise: A total of 6 prospective, randomized trials have reported the effects of physical exercise training as a secondary preventive measure in patients who have survived a myocardial infarction. Patient numbers ranged from 203 to 1,777 and intervention consisted of individualized supervised training for approximately 45 minutes 2 to 3 times a week. Some studies involved supporting educational programs. As might be expected, the exercise capacity was increased in the trained patients. Study populations were small and none of the results was statistically significant although there was suggestion of benefit in 4 of the trials. One trial from Canada showed an opposite trend and one was equivocal. There appeared to be little effect on the incidence of nonfatal reinfarction. Pooling these data, however, demonstrates a 15% benefit of exercise on mortality in spite of problems of poor long-term compliance in these programs (40% to 60% of patients discontinued training of their own accord). Trials of adequate size are needed to determine with certainty the preventive value of physical exercise suggested by these studies. Calcium antagonists: Three long-term trials of reasonable size, with 1,436 to 2,279 patients randomized, were reviewed. Three different calcium antagonistslidoflazine, verapamil and nifedipine-were used. The results of these trials, the Holland Interuniversity Nifedipine Trial of unstable angina and other acute myocardial infarction trials have shown no effect on mortality, although it may still be too soon to judge the preventive effect of this heterogeneous group of drugs. Beta blockers: p blockers possessseveral different properties that may be of potential value to the postmyocardial infarction patient; they are antiischemic and antiarrhythmic, have platelet antiaggregating effects and may possibly enhance the flow properties of erythrocytes. The assumption that the sympathetic
OR FANTASY?
nervous system is important in triggering sudden death led to the testing of these agents in myocardial infarction survivors. A total of 17 trials met the review definition, ranging in size from 111 to 3,837 patients. Eight different P blockers were tested (alprenolol, metopro101,oxprenolol, pindolol, practolol, propranolol, sota101and timolol) and in all of them evidence of ,&receptor blockade was seen. The trials were divided into those evaluating P blockers with intrinsic sympathomimetic activity and those without intrinsic sympathomimetic activity. Beta blockers with intrinsic sympathomimetic activity: Two early Swedish trials of p blockers with intrinsic sympathomimetic activity produced encouraging mortality results, but they were small studies with very wide confidence intervals. Other trials of these 6 blockers have been less promising; the last trial reported negative results. Pooling suggests an overall benefit in the range of 10% reduction in mortality from /3 blockers with intrinsic sympathomimetic activity. p blockers with intrinsic sympathomimetic activity do seem to show greater benefit in reduction of recurrent nonfatal myocardial infarction and in 1 trial the results are significant. Two others are close to significance and active treatment was superior to control in all the trials. Beta blockers without intrinsic sympathomimetic activity: Most of the trials of p blockers without intrinsic sympathomimetic activity showed substantial reductions in mortality, with 1 exception (the Baber study]. Two trials, the Norwegian Timolol Study and the P-Blocker Heart Attack Trial,8 were significant on their own. The largest metoprolol study has not yet been reported, but it is anticipated that the results will not show any benefit from treatment. The pooled data suggest that the benefit from these drugs in terms of decrease in all-cause mortality may be as much as 25% to 30%. In terms of nonfatal myocardial infarction 2 trials of p blockers without intrinsic sympathomimetic activity were significant on their own and most, except for the Baber study, clearly showed benefit of about 25% to 30%) which is similar to that seen for all-cause mortality. The decrease in sudden death (defined as sudden death within 1 hour of onset of symptoms) brought about by these drugs implies an antiarrhythmic mechanism of action. /3 blockers also reduce nonsudden cardiac death, which is assumed to be an antiischemic effect of these drugs, and this view is supported by the results regarding the incidence of nonfatal myocardial infarction.
Discussion Absolute effect on mortality: A number of issues have been raised by these trials. One relates to how we present and interpret efficacy; more emphasis should be placed on the absolute effect of treatment, i.e., how many patients must be treated to benefit one. A 25% relative effect can have very different meanings; if an event rate of 4% is reduced to 3%, then 1 event is prevented for every 100 patients treated. If the mortality rate is higher, at 2O%, and reduced to 1570, 5 events
July 15, 1987
THE AMERICAN
JOURNAL
OF CARDIOLOGY
Volume 60
31A
per 100 will be prevented. With a 10% relative effect, TABLE I Efficacy Measures and Their Clinical Usefulness and a decrease from 20% to 16% in mortality, 2 events Event Rate (%) will be prevented out of 100 [Table I). There are situaRelative Efficacy by No. Efficacy of Patients Treated Intervention tions, therefore, where a 10% relative benefit may be Placebo more important clinically than a 25% benefit. -25% l/100 3.0 4.0 This concept was applied to post-hoc analysis of -25% 5/100 15.0 20.0 data from the P-Blocker Heart Attack Trial9 using a -10% 2/100 18.0 20.0 classification system developed in Gothenburg. The patients were divided into whether they had transient evidence of electrical failure, mechanical failure, both or neither before randomization into the trial. Half of Table II All-Cause Mortality* by Risk Group and Treatment in the the patients were in the uncomplicated group and their &Blocker Heart Attack Trial placebo group mortality over 25 months was 6.6%. For Mortality Rates (% ) those with electrical problems mortality doubled and it tripled in those with mechanical problems, whether or Difference Relative Risk Risk Group Propranolol Placebo not they also had electrical problems. Looking at the .94 -0.4 6.2 6.6 None absolute difference in mortality, however, there was .48 -5.7 5.2 10.9 Electrical very little difference in the low risk group between .62 -6.4 10.4 16.8 Mechanical active and placebo treatment, proving that it is hard to .76 -4.2 12.9 17.1 Both improve on a rate that is already low. Benefit in the * Average follow-up 25 months. high risk group meanwhile, ranged from 4 to 6 lives saved per 100 patients treated [Table II). For the lower risk group, the cumulative mortality curves for placebo and propranolol overlapped. For 18 the combined high risk group, these curves diverged 16 l’ and propranolol showed substantial benefit. The - I shape of the curve also suggests that treatment should be continued for at least 3 years (Fig. 2) Adverse effects and risk: If we accept that high risk patients should be prime candidates for preventive therapy, we have to be prepared to pay a price-a greater incidence of adverse effects. Looking at adverse effects by risk group in the P-Blocker Heart Attack Trial, the incidence of congestive heart failure in the uncomplicated group and the group with electrical complications only was about 5% to 6% over 25 Months months, whether or not they received p blockers. If the FIGURE 2. @Blocker Heart Attack Trial: cumulative mortality by patients had a history of transient mechanical complications and were given placebo, their risk was twice history of complications. Ml = myocardial infarction. that (10%) and if propranolol was added the risk of developing congestive heart failure increased by another 5% (Fig. 3). Caution is required in interpreting consideration and reporting of results. Further analysubgroup findings, particularly those identified after sis of the ,&Blocker Heart Attack Trial data showed the trial was completed, but this illustrates at least the that smaller subgroups are more susceptible to variatheoretical argument for caution in treating high risk tion than the larger groups. patients. The best way of determining whether subgroup Difficulties with subgroups: The danger in sub- findings are real is to repeat the experiment, which can group analyses is that a striking effect may be found in rarely be done easily. Another method is to look for some subgroup by chance alone. This danger has been replications elsewhere. To this end investigators from illustrated by data from the P-Blocker Heart Attack 9 trials of long-term p blockers conducted subgroup Trial. A total of 146 subgroups was examined compar- analyses, using common definitions, aiming to pool the ing the absolute mortality rates for propranolol and subgroup findings as appropriate. The endpoint was placebo treatment. Few of these groups were defined standardized as l-year all-cause mortality, using an in advance. Most analyses were exploratory but some intention-to-treat analysis. were done in order to compare findings with subThe preliminary results of the P-Blocker Pooling groups of interest in other P-blocker trials. The most Project show a 25% overall decrease in mortality. High striking finding to emerge from this pooling exercise risk groups, such as those with transient mechanical was the enormous variability in absolute mortality dif- complications, showed large absolute benefits with ,f3ferences between subgroups; it ranged from a “harm- blocker therapy. These results confirm the earlier obful” effect of 5% to a “beneficial” effect of 11%. It is servation in the ,&Blocker Heart Attack Trial. Unexdifficult to ascertain which effects are real and which pectedly, patients with electrical complications did not are due to chance and there is a danger in selective show a high mortality rate, but the ,&Blocker Pooling
1
/-
32A
A SYMPOSIUM:
16 -
Congestive
CARDIOPROTECTION
IN ISCHEMIC
HEART DISEASE-FACT
heart failure
15 14 13 -
El
Placebo
@J
Propranolol
12 11 10 9 %
87 6.1 6 5 4 3 2 1 n” i No risk
Electrical complication
Mechanical complication
Electrical and mechanical
FIGURE 3. Long-term prevention trials afler myocardial infarction: adverse effects by risk group in the &Blocker Heart Attack Trial.
Project definition was vague and different from that adopted in the P-Blocker Heart Attack Trial analysis. Recently reported findings from the ,&Blocker Heart Attack TriallO showed that the relative benefit of propranolol treatment was similar in patients with complex ventricular arrhythmias at baseline and those with no such arrhythmia (31% and 2570, respectively]. The former subgroup made up 40% of the patients in the trial and had a much higher mortality. In fact, by treating patients with complex arrhythmias 5 lives are saved for every 100 patients treated. This compares with the group without complex arrhythmia, in which 1.6 lives would be saved per 100 treated. Again, this demonstrates that absolute differences in mortality are an important consideration in the decision regarding which patients should be treated with propranolol after a myocardial infarction.
Conclusions This has been a review of 57 trials involving over 61,000patients. Some trials made more than 1 comparison and so, effectively, the total of trials becomes 67. Simple statistics show that, by chance alone, 3 findings should be significant at the level of p <0.05. There were, in fact, 3 results that were significant in this review: 2 in favor of ,f3blockers and one in favor of a lipid-lowering regimen; however, the former p values (e.g., 0.0003 and 0.005)were extremely impressive and this cannot be due to chance. The available data seem to indicate that p blockers, particularly those without intrinsic sympathomimetic activity, improve long-term survival after a myocardial infarction and that the
OR FANTASY?
prime candidates for this type of therapy are patients with complicated infarcts. Platelet-active drugs may have a 10% beneficial effect on survival but a 30% reduction in nonfatal myocardial infarction. It is not known whether this apparent benefit would be additive to that of ,f3 blockers, but these drugs may be useful for those patients with contraindications or poor tolerance of p blockers. Lipid-lowering regimens have so far had no demonstrable effect on survival but it could be that patients have not been treated for long enough and there does appear to be some benefit with respect to nonfatal myocardial infarction. A large trial is needed to determine whether physical exercise might improve survival but the evidence so far is certainly very suggestive of positive benefit. Anticoagulants and antiarrhythmics have not shown demonstrable beneficial effects in the populations studied and in the dosage used but there are new trials underway that may clarify the position of these drugs after a myocardial infarction. It may also be too soon to judge the effect of the calcium antagonists on survival but there is no evidence, so far, that calcium channel blockers have any effect on mortality after myocardial infarction. Many of these interventions may prove to be beneficial in selected patients but we have not been successful in identifying such subgroups. Patients with severe ventricular arrhythmias, for example, may show benefit from antiarrhythmic treatment that is not apparent in a more unselected postinfarction population (which may, in fact, demonstrate an increase in mortality with these drugs]. Using current methodology, reductions of less than 20% are difficult to demonstrate in clinical trials and, in defense of these studies, many were not designed to assessthe effect of treatment on mortality and a large proportion had insufficient statistical power to detect even a true benefit of as much as 30% to 40%.
References 1. Proceedings of the Conference on the Decline of Coronary Heart Disease Mortality. 1979; US DHEW, PHS, NIH publication no. 79: 1610. 2. Final Mortality Statistics. 1978, from the National Center for Health Statistics. 1980; DHEW publication no. [PHS) 29, no. 6, supplement 2:80-1120. 3. Furberg CD, May GS. Effect of long-term prophylactic treatment on surviva1 after myocardiol infarction. Am r Med 1984; suppI 6A:76-83. 4. May GS, Eberlein KA. Furberg CD, Passamani ER, De Mets DL. Secondary prevention after myocardial infarction: a review of long-term trials. Prog Cardiovasc Dis 1982;24:331-352. 5. Five-year study by a group of physicians of the Newcastle upon l’yne region. Trial of clofibrate in the treatment of ischemic heart disease. Br Med J 1971;4:767. 6. Woodhill JM, Palmer AJ, Leelarthaepin B, McGilchrist C, Blackett RB. Low fat, low cholesterol diet in secondary prevention of coronary heart disease. Adv Exp Med Biol 1978;109:317-330. 7. Canner PL, Berge KG, Wenger NK, Stamler J, Friedman L, Prineas RI, Friedewald W, for the Coronary Drug Project Research Group. FifteenYeaT mortality in Coronary Drug Project patients: long-term benefit with niacin. TACC l&6:8:1245-1255. 8. Beta-Blocker Heart Attack Trial Research Group. A randomized trial of propranolol in patients with acute myocardial infarction. I. Mortality results. JAMA 1982;247:1707-1714. 9. Furberg CD, Hawkins CM, Lichstein E. Effect of propronoIoI in postinfarction ootients with mechanical or electrical complications. Circulation 1984;69:76i-765.
10. Friedman LM, Byington RP, Capone RJ, Furberg CD, Goldstein S, Lichstein E, Writing Group for the Beta-Blocker Heart Attack Trial Research Group. Effect of propranolol in patients with myocardial infarction and ventricular arrhythmia. fACC 1986;7:1-8.