Heart, Victor F. Froelicher, Jonathan Mayers, fifth edition

290 E X E R C I S E A N D T H E H E A R T 104. Steingart RM, Hodnett P, Musso J, Feuerman M: Exercise myocar- 117. Mody FV, Nademanee K, Intarachot V, et al: Severity of silent dial perfusion imaging in elderly patients. J Nucl Cardiol myocardial ischemia on ambulatory electrocardiographic moni- 2002;9:573-580. toring in patients with stable angina pectoris: Relation to prog- nostic determinants during exercise stress testing and coronary 105. Weiner DA, Ryan TJ, McCabe CH, et al: Significance of silent angiography. J Am Coll Cardiol 1988;12:1169-1176. myocardial ischemia during exercise testing in patients with coro- nary artery disease. Am J Cardiol 1987;59:725-729. 118. Mulcahy D, Keegan J, Crean P, et al: Silent myocardial ischemia in chronic stable angina: A study of its frequency and characteristics 106. Mark DB, Hlatky MA, Califf RM, et al: Painless exercise ST devia- in 150 patients. Br Heart J 1988;60:417-423. tion on the treadmill: Long-term prognosis. J Am Coll Cardiol 1989;14:885-892. 119. Stern S, Weisz G, Gavish A, et al: Comparison between silent and symptomatic ischemia during exercise testing in patients 107. Falcone C, de Servi S, Poma E, et al: Clinical significance of exer- with coronary artery disease. J Cardiopulm Rehabil 1988;12:507- cise-induced silent myocardial ischemia in patients with coronary 512. artery disease. J Am Coll Cardiol 1987;9:295-299. 120. Flugelman MY, Halon DA, Shefer A, et al: Persistent painless ST- 108. Visser FC, van Leeuwen FT, Cernohorsky B, et al: Silent versus segment depression after exercise testing and the effect of age. symptomatic myocardial ischemia during exercise testing: A com- Clin Cardiol 1988;11:365-369. parison with coronary angiographic findings. Int J Cardiol 1990;27:71-78. 121. Hedblad B, Juul-Moller S, Svensson K, et al: Increased mortality in men with ST segment depression during 24 h ambulatory long- 109. Weiner DA, Ryan TJ, McCabe CH, et al: Risk of developing an term ECG recording. Results from prospective population study acute myocardial infarction or sudden coronary death in patients “Men born in 1914”, from Malmo, Sweden. Eur Heart J with exercise-induced silent myocardial ischemia. A report from 1989;10:149-158. the Coronary Artery Surgery Study (CASS) Registry. Am J Cardiol 1988;62:1155-1158. 122. May O, Arildsen H, Damsgaard EM, Mickley H: Prevalence and prediction of silent ischaemia in diabetes mellitus: A population- 110. Callaham P, Froelicher VF, Klein J, et al: Exercise-induced silent based study. Cardiovasc Res 1997:34:241-247. ischemia. J Am Coll Cardiol 1989;14:1175-1180. 123. Keys A: Coronary heart disease in seven countries. Circulation 111. Weiner DA, Ryan TJ, Parsons L, et al: Significance of silent 1970;41-42: I1-I211. myocardial ischemia during exercise testing in patients with dia- betes mellitus: A report from the Coronary Artery Surgery Study 124. Caracciolo EA, Chaitman BR, Forman SA, et al: Diabetics with (CASS) Registry. Am J Cardiol 1991;68:729-734. coronary disease have a prevalence of asymptomatic ischemia during exercise treadmill testing and ambulatory ischemia moni- 112. Kang X, Berman DS, Lewin HC, et al: Incremental prognostic toring similar to that of nondiabetic patients. An ACIP database value of myocardial perfusion single photon emission computed study. ACIP Investigators. Asymptomatic Cardiac Ischemia Pilot tomography in patients with diabetes mellitus. Am Heart J Investigators. Circulation 1996;93:2097-2105. 1999;138(6 Pt 1):1025-1032. 125. Koistinen MJ: Prevalence of asymptomatic myocardial ischaemia 113. Giri S, Shaw LJ, Murthy DR, et al: Impact of diabetes on the risk in diabetic subjects. BMJ 1990;301:92-95. stratification using stress single-photon emission computed tomography myocardial perfusion imaging in patients with symp- 126. Gerson MC, Khoury JC, Hertzberg VS, et al: Prediction of coro- toms suggestive of coronary artery disease. Circulation nary artery disease in a population of insulin-requiring diabetic 2002;105:32-40. patients: Results of an 8-year follow-up study. Am Heart J 1988;116:820-826. 114. Miranda C, Lehmann K, Lachterman B, et al: Comparison of silent and symptomatic ischemia during exercise testing in men. Ann 127. Janand-Delenne B, Savin B, Habib G, et al: Silent myocardial Intern Med 1991;114:649-656. ischemia in patients with diabetes: Who to screen. Diabetes Care 1999;22:1396-1400. 115. Karnegis JN, Matts JP, Tuna N, et al: Positive and negative exercise test results with and without exercise-induced angina in patients 128. Miller TD, Rajagopalan N, Hodge DO, et al: Yield of stress single- with one healed myocardial infarction: Analysis of baseline vari- photon emission computed tomography in asymptomatic patients ables and long-term prognosis. Am Heart J 1991;122:701-708. with diabetes. Am Heart J 2004;147:890-896. 116. Tzivoni D, Gavish A, Zin D, et al: Prognostic significance of ischemic episodes in patients with previous myocardial infarction. Am J Cardiol 1988;62:661-664.

CHAPTER nine Exercise Testing of Patients Recovering from Myocardial Infarction INTRODUCTION months 2 through 6, to 4% for months 7 through 30, and to 3% for the next 3 years. Other studies Although the death rate for coronary heart dis- have suggested a mortality rate of 11% in the first ease (CHD) has been decreasing steadily since the 3 months after hospital discharge, with lower mid 1960s, it still remains the leading cause of rates thereafter. In comparison with standard death in the United States.1 Four of every 10 deaths medical therapy, thrombolytic therapy exerts a are due to cardiac disorders and 90% of these can highly significant one-fifth reduction in 35-day be attributed to CHD. The four distinct clinical mortality among patients with acute MI and ST manifestations of CHD are primary cardiac arrest, elevation, corresponding to an overall reduction stable angina pectoris, acute coronary syndromes of 21 deaths per 1000 patients treated. All of these (ACS),2 and acute myocardial infarction (MI). The statistics are probably less meaningful because of resting electrocardiogram (ECG) is critical to two changes in healthcare: (1) the use of troponin guiding therapy, with ST elevation indicating the to define MI3 and, (2) data supporting emergency prompt application of thrombolysis or percuta- PCIs over thrombolysis.4 Temporal comparison neous coronary intervention (PCI) and ST depres- studies have suggested a contemporary reduction sion requiring antiplatelet drugs (Fig. 9-1). in mortality as a result of modern therapies and prevention.5,6 The impact of the 30% reduction in Each year 900,000 people in the United States mortality with implantable defibrillators in experience acute MI. Of these, roughly 225,000 patients with history of MI with left ventricular die, including 125,000 who die before obtaining (LV) dysfunction has not even been factored in yet. medical care. The case fatality rate in MI patients is temporally related to onset. The risk of death is The pathophysiologic determinates of prognosis highest within the first 24 hours of onset of signs are (1) the amount of viable myocardium and (2) and/or symptoms and declines throughout the the amount of myocardium in jeopardy. Inferences following year. Following the onset of a first MI can be made regarding these two determinates clin- in middle-aged males, 30% to 50% die within ically if a patient has had congestive heart failure 30 days and 85% of these deaths occur within the (CHF) or cardiogenic shock and continued chest first 24 hours. Those patients with a first MI who pain or ischemia. Using cardiac catheterization, actually reach a hospital alive have a 10% to 18% they can be assessed by ejection fraction (EF) and risk of dying before discharge. The mortality the number of vessels occluded. The clinical find- thereafter falls from an annualized rate of 9% for ings manifested by abnormalities of these two 291

292 E X E R C I S E A N D T H E H E A R T THE ACUTE CORONARY SYNDROME (ACS) VS MI all patients with history of MI undergo cardiac catheterization before discharge, particularly Ruptured plaque with occlusive thrombus because PCI is superseding thrombolysis.8 Exercise testing is now rarely used to decide who needs car- Thrombolysis Q-wave MI diac catheterization because it is the clinical norm. ST elevation Furthermore, success of PCI is being promoted as Fissured or ruptured plaque superior to exercise testing and clinical risk status with subocclusive thrombus for providing criteria for early discharge after acute MI.9 Anti-platelet therapy AHA/ACC EXERCISE TESTING Non-Q wave MI Acute GUIDELINES: RECOMMENDATIONS Coronary ST depression FOR EXERCISE TESTING AFTER MYOCARDIAL INFARCTION Unstable angina Syndrome This chapter begins with a synopsis of the ■ FIGURE 9–1 American College of Cardiology (ACC)/American Heart Association (AHA) Guidelines for the Use of ACS and MI extend on a continuum and treatment is the Standard Exercise Test after MI.10 Specifically directed by the ECG. with regard to testing post MI, the issue of the effects of new therapies, particularly thrombolysis determinates are the basis for several indices that and PCI, were addressed.11 have been used to predict risk. Clinical data have also been very useful in triaging patients in regard Class I (Definitely Appropriate). Conditions for to the necessary length of stay in the hospital. The which there is evidence and/or general agreement criteria for a complicated MI are listed in Table 9-1. that the standard exercise test is useful and help- Patients without these criteria, that is, those with ful in patients recovering from an MI. uncomplicated MIs, can be discharged within 3 to 5 days, whereas those with these criteria require 1. Before discharge for prognostic assessment, longer hospitalization and closer observation. activity prescription, evaluation of medical therapy (submaximal at about 4 to 6 [rather Healthcare professionals must be able to advise than 7] days) patients with history of MI as to what they should or should not do to improve their prognosis. One 2. Early after discharge for prognostic assess- strategy has been to identify high-risk patients by ment, activity prescription, evaluation of med- using various clinical markers and test results.7 ical therapy, and cardiac rehabilitation, if the Clinical markers that have indicated high risk predischarge exercise test was not done (symp- include history of MI, CHF, cardiogenic shock, tom-limited at about 14 to 21 days) tachycardia, continued chest pain, older age, stroke or transient ischemic attack, and complicating 3. Late after discharge for prognostic assessment, illnesses. Procedures used to determine risk with activity prescription, evaluation of medical some success have included the chest x-ray scan, therapy, and cardiac rehabilitation, if the early routine ECG, ambulatory monitoring, radionuclide exercise test was submaximal cardiac tests, and exercise testing. The assump- tion has been that patients at high risk should be Exceptions are noted below under Class IIb considered for intervention; the interventions are and III. coronary artery bypass surgery (CABS) and PCI. Because of easy access to these procedures nearly Class IIa (Probably Appropriate). Conditions for which there is conflicting evidence and/or a diver- TA B L E 9 – 1 . Characteristics that classify an gence of opinion that the standard exercise test is myocardial infarction as being complicated useful and helpful in patients recovering from an MI but the weight of evidence for usefulness or Congestive heart failure efficacy is in favor of the exercise test. Cardiogenic shock Large myocardial infarction—as determined by creatine 1. After discharge for activity counseling and/or phosphokinase, troponin, and/or electrocardiogram exercise training as part of cardiac rehabilitation Pericarditis Dangerous arrhythmias, including conduction problems Concurrent illnesses Pulmonary embolus Continued ischemia Stroke or transient ischemic attack

C H A P T E R 9 Exercise Testing of Patients Recovering from Myocardial Infarction 293 in patients who have undergone a coronary clinical experience has demonstrated the benefits of revascularization procedure the treadmill test following an MI in the following two areas: Class IIb (Maybe Appropriate). Conditions for which there is conflicting evidence and/or a diver- 1. Demonstration of exercise capacity for activity gence of opinion that the standard exercise test is prescription after hospital discharge—this useful and helpful in patients recovering from an includes domestic and occupational work eval- MI and/or the usefulness or efficacy is less well uation and exercise training as part of compre- established. hensive cardiac risk reduction and rehabilitation 1. Before discharge in patients who have under- 2. Evaluation of the adequacy of medical therapy gone cardiac catheterization to identify ischemia and the need to employ other diagnostic or in the distribution of a coronary lesion of bor- treatment options derline severity METHODOLOGICAL STUDIES 2. In patients with an abnormal resting ECG as a result of left bundle branch block, interventric- Safety of Exercise Testing Early ular conduction delay, electronically paced, LV Post MI hypertrophy, digoxin therapy, or those demon- strating major ST-segment depression (>1 mm) The risk of death and major arrhythmias by per- in several leads forming an exercise test early after MI is very small. However, the major experience is based on clinically 3. Periodic follow-up exercise testing in patients selected MI patients; those without major complica- who continue to participate in exercise training tions such as heart failure, severe arrhythmia or or as part of supervised or unsupervised cardiac ischemia, LV dysfunction, or other severe diseases. rehabilitation Risk is highest in those rejected for testing for these clinical reasons. The incidence of fatal cardiac Class III (Not Appropriate). Conditions for which events, including fatal MI and cardiac rupture, is there is evidence and/or general agreement that 0.03%, the incidence of nonfatal MI and unsuc- the standard exercise test is not useful and help- cessfully resuscitated cardiac arrest is 0.09%, and ful in patients recovering from an MI, and in some complex arrhythmias, including ventricular tachy- cases may be harmful. cardia, is 1.4%. Symptom-limited protocols have an event rate that is twice that of submaximal tests, 1. Severe comorbidity likely to limit life expectancy although the overall fatal event rate is quite low.12–14 and/or candidacy for revascularization. Exercise testing after MI is safe. Submaximal test- ing can be performed at 4 to 6 days, and a symptom- 2. At any time to evaluate patients with acute MI limited test can be performed 3 to 6 weeks later. who have uncompensated CHF, arrhythmia, Alternatively, symptom-limited tests can be con- or cardiac conditions that severely limit their ducted early after discharge at about 14 to 21 days. ability to exercise (level of evidence C). Submaximal Testing 3. Before discharge to evaluate patients who have already been selected for, or undergone, cardiac The exercise test can determine the possible risk catheterization. Although a stress test may the patient may incur with exercise. It is certainly be useful before or after catheterization to safer that adverse reactions be observed in con- evaluate or identify ischemia in the distribu- trolled circumstances. Arbitrarily, a heart rate limit tion of a coronary lesion of borderline severity, of 140 beats per min and a MET level of 7 is used stress imaging tests are recommended (level of for patients younger than 40 years of age, and evidence C). 130 beats per min and a MET level of 5 for patients more than 40 years of age. Particularly for patients The following sections provide the literature on beta-blockers, a Borg perceived exertion level in support for these guidelines as well as a summary the range of 15 is used to end the test. In addition, of early studies and methodological studies that conservative clinical indications for stopping the have provided useful information regarding the test should be applied. The physician providing use of exercise testing post MI. The studies that have compared exercise test results with coronary angiography are also presented. The main part is devoted to a critique of the follow-up studies. It is good to keep in mind, though, that although our approach is evidence-based, mainly empirical

294 E X E R C I S E A N D T H E H E A R T medical care for the patient can gain valuable capacity to perform activities. Perceived confidence information about the patient by being there dur- in their husbands’ physical and cardiac capabilities ing the test and interacting with the patient. were significantly greater among those wives who also performed the test than in the other two Studies have evaluated symptom-limited pro- groups. In a similar study, Ewart et al22 demon- tocols at 5 to 7 days after an MI and have included strated that the patients’ confidence was enhanced patients treated with thrombolytic agents. These by the test also. studies demonstrate that such testing yields ischemic responses nearly twice as often as sub- Protocol Comparison maximal tests and are a better estimate of actual maximal exercise capacity.15 Thus, early symptom- Handler and Sowton23 compared the Naughton and limited tests have potential to be more useful in modified Bruce treadmill protocols in 20 patients activity prescription before discharge. 6 weeks after a MI. Estimated exercise capacity and ischemic responses were similar using both pro- Historical Methodological Studies tocols. Starling et al24 evaluated 29 patients with uncomplicated MIs with heart rate-limited and Torkelson16 reported results in 10 patients follow- symptom-limited modified Naughton treadmill ing an uncomplicated MI. During the sixth week test and 31 similar patients with a symptom-limited of an in-hospital rehabilitation program, a low- modified Naughton and standard Bruce test at level treadmill test was performed using 1.7 mph 6 weeks following an MI. Predischarge, the at a 10% grade. He concluded that the treadmill test symptom-limited Naughton test identified a greater was valuable for discerning the exercise responses number of patients with ST-segment depression or of MI patients. Ibsen et al17 reported the results of angina than did the heart rate-limited test. At a maximal bicycle test in the third week after an 6 weeks following MI, the standard Bruce test iden- MI in 209 patients. Niederberger18 presented the tified significantly more ischemic abnormalities values and limitations of exercise testing after than did the symptom-limited modified Naughton MI in a monograph published in Vienna in 1977. test. During the Bruce test, a higher double product Markiewicz et al19 studied 46 men younger than was reached in a shorter time. 70 years of age with treadmill tests at 3, 5, 7, 9, and 11 weeks after their MI. The test at 3 to 5 weeks, Reproducibility and the test at 7 to 11 weeks, appeared to provide most of the information obtained in all five tests Starling et al25 evaluated the comparative predic- performed. The Washington University group tive value of ST-segment depression or angina in evaluated 41 patients with a history of MI.20 They 93 patients with history of MI tested predischarge assessed symptoms, signs, and hemodynamic and and 36 tested again at 6 weeks. They concluded ECG responses during and after three activities: that angina alone, irrespective of the presence of sitting upright, walking to an adjacent toilet, and ST-segment depression, was a better predictor of walking on a treadmill. These activities were stud- multivessel disease than ST-segment depression ied at 3, 6, and 10 days, respectively, after infarc- alone. Handler and Sowton26 evaluated the diur- tion. They concluded that successful performance nal variation and reproducibility of abnormalities of these three activities provided useful criteria occurring during predischarge treadmill testing for discharge of a patient with an MI. in 41 patients. Each patient was exercised using a symptom-limited Naughton protocol in the Effect on Patient and Spouse morning and the afternoon on 2 consecutive days. Confidence Ischemic abnormalities were poorly reproducible in any patient but no significant diurnal variation Taylor et al21 evaluated the effects of the involve- occurred. The reproducibility of an ischemic result ment of the wife in her husband’s performance of in all four tests was 66%. Starling et al27 evaluated a treadmill test 3 weeks after an uncomplicated 89 patients with predischarge and 6 weeks’ tread- acute MI.21 They compared wives who did not mill tests to determine the importance of doing observe the test, those who observed the test, and repeat tests to identify abnormalities of known those who observed and performed the test them- prognostic value. Nineteen patients completed selves. In a counseling session after the treadmill only a predischarge exercise test, nine of whom test, wives were fully informed about the patient’s

C H A P T E R 9 Exercise Testing of Patients Recovering from Myocardial Infarction 295 experienced an early cardiac event precluding It appeared that large anterior MIs behave as if left repeat testing. ST-segment depression was highly bundle branch block was present and the ST shifts reproducible. Angina, inadequate BP response, have a very low specificity for ischemia. However, and ventricular arrhythmia have limited repro- a subsequent study by Miranda et al32 demonstrated ducibility and substantial individual variability. that severe angiographic disease could be recog- nized in spite of Q waves using ST depression. Spontaneously Improved Exercise Capacity Post MI THE RESULTS OF EXERCISE TESTING AND CORONARY Wohl et al28 studied 50 patients after an acute ANGIOGRAPHY MI. They found that in stable patients, there was an improvement at 3 weeks of the relationship Exercise testing after MI has been used to decide between myocardial oxygen supply and demand who needs PCI or CABS to improve their outcome. as detected by ST-segment changes. There was a The angiographic studies are summarized in delayed improvement between 3 and 6 months in Table 9-2 and summarized below. exercise capacity associated with increased stroke volume and cardiac output. Haskell et al29 reported Weiner33 reported 154 patients with a single MI the cardiovascular responses to repeated treadmill who had exercise testing and coronary angiography. testing at 3, 7, and 11 weeks after acute MI. Two Patients averaged being 1 year post-MI. Eighty- symptom-limited tests were performed on 24 males three patients developed ST depression only, 22 had several days apart. All test variables measured at elevation with depression in other leads, 19 had ele- maximum effort increased significantly between vation only, and 30 had no changes. ST depression 3 and 11 weeks. Other studies have documented (with or without ST elevation) predicted multi- that exercise capacity increases spontaneously after vessel disease, ST elevation alone, or no ST shift an MI, even in patients not in a formal exercise suggested single vessel involvement and elevation program. predicted LV aneurysm. Paine et al34 studied 100 consecutive patients with exercise testing and car- Effect of Q-Wave Location diac catheterization at a median of 4 months after on ST-Segment Shifts MI. Of 31 patients with 0.1 mV of ST depression, 87% had two- or three-vessel disease, whereas of Castellanet et al30 studied 97 patients with a prior 21 patients with no depression, 38% had two- or transmural MI who underwent coronary angiog- three-vessel disease. Fourteen patients had ST ele- raphy and treadmill testing. In patients with a vation, and they had more LV damage. Dillahunt previous inferior wall infarction, the ST-segment and Miller35 exercise tested 28 patients from 10 to response had a high degree of sensitivity and speci- 18 days after MI and catheterized the same patients ficity (approximately 90%) in detecting additional 4 to 20 weeks later. Among 11 patients with no coronary disease. However, in patients with a pre- symptoms, ST-segment changes, or arrhythmia vious anteroseptal MI, the ST response had much during the treadmill test, eight had single-vessel less sensitivity. In this group, a positive test sug- disease (73%) and three had two-vessel disease. gested the presence of ischemia in the lateral or In contrast, among the 17 patients with any abnor- inferior posterior region. It was thought that the mality, 14 (82%) had three- or four-vessel disease. aneurysm generated an ischemic vector canceling ST-segment changes and producing a false-negative Sammel et al36 reported the results of exer- treadmill test. If the anterior infarction extended cise testing and coronary angiography in 77 men beyond V4, the sensitivity rate of treadmill testing younger than 60 years of age studied one month dropped even further. Ahnve et al31 used thallium after MI. The 22 patients with exercise-induced scintigraphy and computerized ST-vector shifts to angina had a greater proportion of myocardium evaluate the effect of Q-wave location on the rela- supplied by significant lesions compared with the tionship of ST shifts to ischemia. Anterolateral 55 patients free of angina. The combination of ST MIs had large ST-segment spatial shifts that did segment changes and angina was 91% predictive not indicate ischemia, whereas when shifts of triple vessel disease. All four patients with sig- occurred in patients with inferior or subendocardial nificant left main disease had both angina and MIs, ischemia was detected by thallium defects. ST segment changes. Fuller et al37 performed sub- maximal exercise tests on 40 MI patients before discharge and performed catheterization 5 to 12 weeks after MI. Among the 15 patients with an

296 E X E R C I S E A N D T H E H E A R T TA B L E 9 – 2 . Studies in which results of exercise testing were used to predict results of coronary angiography after acute myocardial infarction exercise test characteristics Year Patients Endpoints for ECG Protocol Time Angiography Investigator published tested leads after MI time after MI testing Bruce Weiner 1978 154 12LD 2−36 mo 2−36 mo SS, SBPd, >4 mm, Bruce Paine 1978 100 RVA V4–6 4 mo 4 mo Naughton Dillahunt 1979 28 90% MHR, SS, CM5, V2 10−18 days 4−20 wk IVCD, 1 mm Green Lane Samuel 1980 77 12LD Low Bruce 1 mo 1 mo Fuller 1981 40 SS, 1 mm, >3 12LD PVC/min, 5 min Naughton 9–18 days 5–12 wk 12LD Bruce Starling 1981 57 SS, 6 METs 12LD Low Bruce 9–21 days 3–12 wk Boschat 1981 65 HR 120, SS, 12LD 2–12 mo 2–12 mo Schwartz 1981 48 1 mm, >5 PVCs Bruce 18–22 days 3 wk SS, VT, SBPd, HBP 12LD Bruce De Feyter 1982 179 85% MHR, 1 mm 12LD UPR Bike 6–8 wk 6–8 wk Akhras 1984 119 SS, SBPd VT, 2 mm, 12LD Bruce/TH 2 wk 6 wk Morris 1984 110 75% MHR 12LD >6 wk <3 mo van der Wall 1985 176 SS, VT 6–8 wk 6–8 wk SS SS SS Exercise test characteristic columns: CM5, a bipolar lead; HBP, high blood pressure; HR, heart rate; IVCD, intraventricular conduction defect; MET, a maximal exercise level allowed to be reached as estimated from work load; MHR, heart rate at maximal effort; mm, amount in millime- ters of ST shift taken as an endpoint; (percent heart rate), percentage of age-predicted maximal heart rate chosen as a limit; SPBd, systolic blood pressure drop; SS, signs or symptoms, or both; 12LD, the full set of 12 leads; V5, fifth precordial lead; VT, ventricular tachycardia. Protocol, type of exercise study done: Bruce, Bruce protocol stopped at 85% of the age-predicted maximal heart rate; low Bruce, Bruce protocol with 0 and 1/2 stages, which are 0% and 5% grade at 1.7 mph before stage 1 (10% grade at 1.7 mph); Bruce/TH, Bruce protocol with thallium imaging; Green Lane, Green Lane Hospital treadmill protocol; Naughton, Naughton treadmill test; UPR, upright bicycle combined with radionu- clide testing. Time after MI, mean time after myocardial infarction that the exercise test or angiography was done. abnormal response (angina and/or ST segment ST-segment elevation was noted in patients with depression), 13 (87%) had multivessel disease ver- wall motion abnormalities in the leads facing the sus 7 of 25 patients (28%) with a negative test. In areas of infarction and was associated with a lower a subgroup of 30 patients with a first MI, 89% with EF but was a poor indicator of multivessel disease. an abnormal test had multivessel disease compared ST-segment depression was only about 60% sensi- with 19% of those with a negative test. Among the tive for multivessel disease. The occurrence of ST- 15 patients with an abnormal test, 73% later had segment elevation in the leads facing the infarcted angina compared with 16% among the 25 patients zone along with significant depression in the with a negative test. opposed leads always indicated that another major vessel was involved, but this occurred in only Boschat et al38 from France have reported their 25% of the cases presented. Patients who had both results in 65 patients who sustained their first angina and ST-segment depression usually had transmural MI and within four months had under- multivessel disease. gone coronary angiography and treadmill testing. These 65 who had a treadmill test were from a Schwartz et al39 reported 48 patients studied group of 80 patients (81%) who had coronary angio- with an exercise test and coronary angiography graphy. Approximately 33% had post-MI angina. 3 weeks after their MI. Among the 21 patients with Only half of the vessels supplying the infarcted abnormal responses, 90% had multivessel disease areas remained occluded, meaning that half had versus 55% among the 27 patients with a normal undergone spontaneous recanalization. Only 28 test. Exercise-induced ST-segment elevation in (43%) had an abnormal test by ST-segment depres- 24 patients was associated with lower EF and more sion criteria and abnormal tests were more com- abnormally contracting segments. Starling et al40 mon in the inferior MIs (54%). The clinical severity evaluated 57 uncomplicated patients with a of the angina was directly related to abnormal tests, symptom-limited Naughton treadmill test 9 to whereas exercise aerobic impairment closely 21 days after MI and with coronary angiography correlated with the number of diseased vessels. within 12 weeks. They found that ST-segment

C H A P T E R 9 Exercise Testing of Patients Recovering from Myocardial Infarction 297 depression and/or angina during the exercise test symptoms, or CABS. The latter is especially worri- had a superior sensitivity (88%) for detecting mul- some, because the results of the test can influence tivessel disease compared with ST-segment depres- who will have CABS, and CABS may affect mor- sion alone (54%). Patients with inadequate BP tality. These studies are summarized in Table 9-3. response had multivessel disease (12 of 13) and they The studies are grouped and combined for meta- had mean reduced EF (EF 39%) compared with analysis by the institution at which they were per- patients with a normal systolic blood pressure (SBP) formed. Each column is explained in the legend. response (EF 58%). Ericsson et al42 reported their results of tread- De Feyter et al41 found the prevalence of multi- mill testing 3 weeks after an acute MI in 100 of vessel disease was 63% in inferior and 42% in ante- 228 MI patients. Ventricular dysrhythmias were rior MIs. Left ventricle impairment was more severe classified as occurring during monitoring, during in anterior and prior MIs more prevalent than rest before the test, and during and after the tread- in inferior or nontransmural MIs. When they mill test. They considered premature ventricular considered an abnormal exercise response to be contractions (PVCs) if equal to or greater than five ST-segment depression and/or angina, the sensitiv- per minute and specifically as to patterns, ventric- ity and specificity for multivessel disease was low ular tachycardia, and ventricular fibrillation. for anterior and inferior transmural MIs. However, During rest before the treadmill test, two patients 80% sensitivity and 91% specificity was obtained had unifocal and multifocal PVCs. During and in 21 patients with non-Q-wave MIs. With the def- after the treadmill test, six had unifocal, eight had inition of an abnormal test as depression and/or multifocal, seven had two or three in a row, and angina and elevation, they analyzed the diagnostic one had four or more PVCs in a row. value for combined multivessel disease and wall motion abnormalities and a sensitivity of 41% and Kentala43 have reported their findings in con- a specificity of 87% was obtained. secutive male patients discharged after acute MI in 1969 from the University of Helsinki Hospital. Summary of the Angiographic During this period, 298 males younger than Studies 65 years old were treated. Forty-five died in-hos- pital and the patients were selected for follow-up These studies involve populations that are much because of their availability and willingness to par- selected, often containing a higher prevalence of ticipate in a randomized trial of cardiac rehabilita- patients with angina than the usual population tion. The prognostic power of clinical and ECG with history of MI, because they were more likely variables recorded soon after MI, and in connec- to undergo angiography. Review of the studies tion with the exercise test, were analyzed by step- demonstrates a limited sensitivity and specificity for wise multiple discriminant analysis. Patients dying multivessel disease. within 2 years had a low exercise SBP. With longer follow-up, the exercise blood pressure had a weaker PROGNOSTIC STUDIES impact. At the 4- and 6-year points, an abnormal resting terminal P wave was the best predictor of This portion is based on the analysis of reports poor prognosis, probably identifying a group with published between 1972 and 1987 of longitudinal CHF. Patients with a high level of physical activity studies using exercise testing in the early period before infarction were less prone to die suddenly. after an MI with a follow-up for cardiac events. Exercise-induced ST-segment depression did not The most commonly cited studies and those of identify a high-risk group at any point during fol- particular instructive value were chosen. These low-up. Abnormal apical impulses, T-wave inver- studies have been carefully analyzed for their: (1) sion after exercise, prior resuscitation, sedentary methodology, (2) sample selection, (3) detailed life style, and PVCs during exercise were predictive description of sample, and, (4) description of sta- of sudden death. tistical methods to permit identification of differ- ences that might be due to their lack of agreement Granath et al44 performed exercise tests at 3 or commonality. The cardiac event endpoints cho- and 9 weeks after an acute MI in 205 patients and sen are reinfarction and death. Some studies followed them for up to 5 years. The investigators combine these two endpoints to predict outcome. chose not to evaluate the ST segments because Some investigators combine reinfarction and death of the accepted difficulties of evaluating ST shifts with soft endpoints such as angina, worsening of after MI and because of medications. The appear- ance of tachycardia at low workloads, major ventricular dysrhythmias, or anginal complaints during these early exercise tests was associated with a significantly increased mortality during the

TA B L E 9 – 3 . Summary of 24 prospective studies evaluating the ability of exercise test after acute myocardial infarction to predict morbidity and mortality 298 E X E R C I S E A N D T H E H E A R T Population characteristics MI % Exercise test characteristics Investigator MI Exercise End points ECG Protocal Weeks Age/% Exclusions PR Meds pop. tested leads after of women 25 Transmural (Dig or 1 Ericsson Year size n% HR 140, SS TM MI >65 28 2 Kentala 100 54 Max PC Bike 59/7 >65, Rehab SE A IP BB) 73 184 158 53 CH1-6 3 53/0 18 75 298 HR 140,SS Tm/Bike 6-8 >65 ? ? 51 43 35%D, 205 48 60%HR 12LD GXT ? ? 13 42 58 1%BB 3 Granath 77 430 62 57 70%HR, SS 12LD Bike 3&9 59/11 No complic ? 66%D 4 Smith 79 109 56 9 7LD 3 60/? 8 ? 48 33 10%BB 5 Hunt 79 633 154 SS 7LD Naughton 6 57/11 CHF, USAP 8 5? ? ? Srinivasan 81 200 HR/SS 12LD Stanford >70, drgs, 0 47 ? 6 Sami 79 12LD 3-52 57/10 CHF ? ?? 53 8%D Davidson 80 461 195 42 3 53/0 >70, CHF, 9 29 ? None USAP 34 10 29 338 48 >70, CHF, 62 USAP 25 61 210 64 DeBusk 83 702 SS 12LD Naughton 3 54/0 CHF, drgs, ? ? ? ? 3%D 225 68 ANG 7 Theroux 79 326 5 METs, 70%HR CM5 Naughton 1.6 52/0 USAP, CHF 24 18 31 50 40%BB, 108 26 21 43 55 1%D Waters 85 330 53/16 >70 21 24 28 48 6%D, 130 68 32%BB 8 Koppes 80 410 Submax Max 12LD Bruce 3&8 52/13 Age, CHF, 10 None 236 73 ANG 9 Starling 80 190 HR130/SS V1,5,6 Naughton 2 53/14 >66, se, w 3 29 34 37 26%D, 317 78 54/12 16%BB 10 Weld 81 325 4 METs, SS V5 Low Bruce 2 >65, referrals 8 ?? ? 12%BB, 200 21 ANG, CHF 18 41%D 11 Saunamaki 81 404 SS PC Bike 3 57/20 >75, CHF, 31 ? 32 ? 20%D, 179 81 USAP 2%BB 12 Velasco 81 958 188 30 w, SS PC SupBike 2.5 60/22 >65, CHF 19 0 46 55 11%D, 456 52 9%BB 13 De Feyter 82 222 SS 12LD Bruce 6-8 52/0 >60 0 12 35 45 Stopped 14 Jelinek 82 886 140 61 Symptoms Bike 1.5 52/10 >70 23 28 29 42 ? 15 Madsen 83 SS V4-6 Bike 2.6 51/? 12%D, 315 80 9LD ? 6 35 ? 2%BB 205 91 2%D, 16 Gibson 83 229 HR 120, SS 3LD Naughton 1.6 63/13 26 35 53 61%BB 103 20 30%BB 17 Norris 84 395 SS ? 2.5 mph 4 51/13 27 29 42 16%D, 6 METs 23 33 46 90%BB 18 Williams 84 226 3LD Bruce 1.7 50/0 4%D, ? 51 49 10%BB 19 Jennings 84 503 5 METs, SS V5 2 mph 1.7 56/18

20 Fioretti 84 293 214 72 Symptoms XYZ Bike 2 54/13 >66, CHF, ? 40%BB ANG 36 85 405 300 74 54/16 CHF, ANG 27 ? 18%D, 22 22 31 42 52%BB 21 Krone 85 1417 667 47 5 METs 3LD Low Bruce 2 ?/20 >70 28%D, ? 31%BB Dwyer 85 60% <60 21 54/16 >65, CABS, 1%D, 22 Handler 85 296 222 75 5 METs, 70%HR 3LD Naughton 1.4 BBB 0 21 42 37 17%BB 58/18 MD judgment 3 18 38 44 26%D, 23 SCOR 85 1469 295 20 75%HR, SS 12LD Mixed TM 1.7 ? 36 31 33 53%BB C H A P T E R 9 Exercise Testing of Patients Recovering from Myocardial Infarction 299 57/14 >71, CHF, ? 11 32 57 13%D, 24 Jespersen 85 126 Max, SS II, V4,6 Bike 3.4 USAP 22 20%BB 12 Bike 7 50/0 >65, CHF, ?? ? 2%D, 25 Paolila 85 362 263 73 Max USAP, w ?? ? 2%BB 53/17 >66, CHF 28 ? ? 20%BB 26 Murray 86 350 300 86 Sub TM 2 58/0 >70, w, CHF 10%D, 27 Cleempoel 86 202 198 98 Sub 4 TM 1.6 50%BB 54/21 >75, USAP, 26%D, 28 Stone 86 719 473 66 Max 12 TM 24 CHF, PVCs 39%BB PVC TOTAL — 4× 7029 + 2×* Follow-up period Exercise test risk markers — 1 Investigator Mean or Range %CABS Mortality if Repeat MI if SBP ? ExCap Angina ST Statistical method median ET performed ET performed — Ericsson 3 mo-? ? yes/no yes/no NR ? ? ? NR Descriptive Kentala 3 mo ? 0% + NR NR NR +* UV; some DF Granath 6 yr 2-5 yr ? 5%/ ? NR 2× 2× NR UV Smith 2-5 yr ? ? 32%/ NR 2× NR NR 6×* UV Hunt 1.5 yr ? ? 25%/ ? NR + NR 4×* 3×* Descriptive, UV Srinivasan 1 yr 1-2 yr ? 10%/17% ? NR ? NR 3×* 7×* Not cited (UV) Sami 1.25 yr 2-51 mo 10% 14%/18% ? NR 2× 1 NR 3×* UV Davidson 19 mo 1-60 mo 10% 8% 5%/ ? 2×* +* 1 +* MV-LR, LT, K-M, est DeBusk 26 mo ? 6% 2%/ 6%/ NR 2×* NR NR 8×* UV; Cox to select 34 mo 1.5%/ 2%/ 2.1%/5.5% NR — 13×* some variables + NR 8×* UV Theroux 1 yr 1 yr 5.7% 9.5% 6%/ NR ? ? ? UV (Cox), MV-Cox Waters 2 yr 5-7 yr 16% 11%-3% NR 4× 4× UV Koppes 2 yr ? 2%/ 9%/ +* 19×* 2× 2× UV Starling 11 mo 6-20 mo ? 8%/ ? ? NR NR 1 MV-LR; UV est Weld 1 yr ? ? 9%/ LT w/in clinical Saunamaki 5.7 yr 5-6 yr ? 35.6%/ 5× subsets 5×* 3×* Continued

TA B L E 9 – 3 . Summary of 24 prospective studies evaluating the ability of exercise test after acute myocardial infarction to predict morbidity and 300 E X E R C I S E A N D T H E H E A R T mortality—cont’d Follow-up period Exercise test risk markers Mortality if RE MI if ET Mean or ET performed performed Investigator median Range %CABS yes/no yes/no SBP PVC ExCap Angina ST Statistical method Velasco 3 yr 3 mo-6 yr ? 11%/ 3%/ 3× 2× NR 3×* 4×* UV De Feyter 28 mo 13-40 mo 13% 6%/ 7%/ Jelinek 2.3 yr 10 days-62 ? 7%/ 19%/ NR 3× + 2× 1 UV mo Madsen 1 yr — NR + 2×* 1 UV 0% Gibson 1.3 yr 6.6%/28% 4%/12% +* +* +* ? 1 MV-DF, Cox; Norris 3.5 yr Williams 1 yr algoritham Jennings 1 yr Fioretti 1.2 yr 1-3 hr 14% 5%/ 6%/ NR NR NR + + UV 1 yr 1-6 yr 24% Krone 1 yr 1 yr 12% 13%/33% 12%/ NR NR ? ? 1 UV-LT; Cox cited Dwyer ? 5% Handler 1.2 yr 8% 6%/31% 6.8% 2× — 2×* 2× 1 MV-DF; UV est SCOR 1 yr 1 yr 8% Jespersen 1 yr 1 yr 12% 9%/21% 3%/ 8×* 1 8×* ? 1 UV Paolila 2.6 yr Murray 13 mo 9%/23% +* 2× + 1 2× UV Cleempoel 0.16 yr 7%/28% 4%/ +* + +* 1 — MV-DF, algorithm 5%/14% 8×* 2× 3×* 3×* 1 UV;MV-LR 5%/10% NR ? ? ? ? UV;MV-LR 6-36 mo 9% 7%/ 4%/ 5×* 1 8×* 1 2× UV ? ? 1 yr >1% 7%/15% 1% 2× 9×* 2× 3× UV, MV-DF 3-57 mo 6% 6 mo-? ? 7% 2% 1 1 1 1 3×* UV, K-M 2 mo ? 4.1%/ 8.3%/ 1 1 1 1 4× UV 18%/ 13%/ NR NR NR + + UV 5%/ ? NR NR + NR 1 UV, MV-DF

Stone 1 yr ? 2 3/16 5%/ 5× 6× 6× 1 1 UV, MV, LT PVC RR ExCap RR Ang RR ST RR SPB RR 5 95 9 *Number 9 of studies demonstrat- ing signfi- cant risk predictor C H A P T E R 9 Exercise Testing of Patients Recovering from Myocardial Infarction 301 Number 13 14 14 12 15 with 23 18 20 24 positive risk Number 18 with reported effect Investigator, the first author, SCOR, Specialized Center of Organized Research, year; year of publication; MI Pop. size, number of patients admitted to the hospital with myocardial infarction over the period of the study; Exercise tested: n, number, and %, percentage, of patients out of this MI population who underwent exercise testing. Exercise test characteristics: SS, signs or symptoms, or both; HR with a heart rate value—a heart rate limit; max, maximal effort; (percent heart rate), percentage of age-predicted maximal heart rate cho- sen as a limit; MET, a maximal exercise level allowed to be reached as estimated from work load; Symptoms, symptoms alone were the endpoint; PC, precordial leads; 12LD, the full set of 12 leads; CM5, a bipolar lead; V5, fifth precordial lead (among others); XYZ, Frank vector leads; Protocol. Type of exercise study done; TM, treadmill; GXT, Bruce protocol stopped at 85% of the age-predicted miximal heart rate; Stanford, Stanford version of the Naughton test; low Bruce, Bruce protocol with 0 and 1/2 stages, which are 0% and 5% grade at 1.7 mph before stage 1 (10% grade at 1.7 mph). The Norris study at Green Lane used a 2.5-mph tradmill protocol with increasing grade; Weeks after MI, mean time after MI that the exercise test or tests were done. Population characteristics including age, sex, exclusions, MI mix, and medications: Age/% of women, mean age of patients and the percentage of women included in the study; Exclusions, > (greater-than symbol) excludes patients above a certain age; other exclusion factors were CHF, congestive heart failure; USAP, unstable angina pectoris; drgs, cardiac drugs; ANG, angiography; se, subendocardial MI; w, women; complic, complications; Rehab, not in a rehabilitation program; PVCs, abnormal premature venticular contractions; MI%, percentage of the types of infarctions included in the study; PR, prior MI; SE, subendocardial or non-Q-wave MIs; A, transmural (Q wave) anterior wall MI; IP, transmural inferior and/or posterior MI; Meds, percentage of patients on digoxin (Dig, D) or a beta-blocker (BB) at the time of treadmill testing and often through the follow-up period. CABS, coronary artery bypass surgery; Mortality, in those patients included in the study who underwent exercise testing (ET) (yes) and in those who were excluded from exercise testing for clinical reasons (no); RE MI, recurrent MI, the percentage who had a repeat MI if exercise tested (yes, left of/) or if not exercise tested (no right of/). Exercise test risk markers: SBP, abnormal systolic blood pressure response; PVC, abnormal premature ventricular contractions seen; Excap, abnormally low exercise capacity tolerance; Angina (Ang), angina induced by test; ST, abnormal ST-segment response (usually only depression). These are the responses to exercise testing that have been most commonly reported as having prognostic value. RR, Risk ratio—univariate (UV) or multivariate (MV) analysis risk ratio. If significant statistically, the risk ratio has an asterisk.*Nonsignificant risk ratios permit trends across studies to be detected. The risk ratio means that if the cutpoint value for this abnormality was reached, those with that abnormality have a certain times (×) risk of death (high risk) as opposed to those without the abnormality. Only the hard endpoints of death (and in some studies, reinfarction) are considered. NR < Results of prediction with the exercise test marker were not reported; LT, clinical life table, usually stratified; LR, logistic regression: K-M, Kaplan-Meier; est. estimates; w/in, within; DF, discriminant function analysis; ?, insufficient data to test significance; 1, null effect; +, a positive nonsignificant association of usual high- risk with death; –, a negative nonsignificant association of usual high risk level with death; Cox, proportion hazard regression model for survival analysis; algorithm, detailed specific algorithm displayed for clinical use.

302 E X E R C I S E A N D T H E H E A R T observation period. Exercise-induced PVCs proved after an MI. At 3 weeks, 100% of those who subse- to be of greater prognostic significance than those quently had an episode of cardiopulmonary resus- recorded at rest. During exercise testing, 9 weeks citation and 60% of those who required CABS had after infarct, PVCs were seen in 23% of the patients. 0.2 mV of ST-segment depression during tread- During follow-up, 16 of them died compared with mill testing. Only 35% of those without an event 25 of 134 without arrhythmia. Tachycardia during a had a similar amount of ST-segment depression. submaximal workload (greater than 130 beats per At 5 weeks and beyond, recurrent PVCs during minute) identified a high-risk group at both periods. serial treadmill testing occurred in 90% of those who had a recurrent MI and in only 47% of those Smith et al45 from Arizona did treadmill tests on without an event. Exercise-induced PVCs or 62 patients 18 days after admission for acute MI. ischemic ST-segment depression 11 weeks after Death and MI were similarly high, both in the infarction identified patients with an increased group with elevation and in the group with depres- risk of subsequent coronary events, whereas the sion. Of the patients who developed ST-segment absence of either identified a group of patients who depression, 30% (6 of 20) either died or had another were free of problems. MI after discharge from the hospital versus only 2 (5%) of 42 patients who did not have ST-segment Davidson and DeBusk50 reported results of depression during exercise. treadmill testing in 195 men tested 3 weeks after acute MI. Stepwise logistic analysis on a subset Australia of 92 with at least 2-year follow-up showed ST- segment depression equal or greater than 0.2 mV, Hunt et al46 reported findings from the Royal angina, and a work capacity of less than 4 METS Melbourne Hospital in 75 patients younger than to be risk markers. These results were confirmed 70 years of age. They selected their patients on in the 195 men using stratified life table analysis the basis of having survived an MI complicated by with log rank tests. The patients were followed for arrhythmia and/or mechanical abnormalities. Of 1 year and had a 19% event rate; however, more 11 patients with ST depression of 1 mm or more, than half of these endpoint events were CABS. PVCs 36% died whereas 4 of 45 (11%) without depres- on a single treadmill test 3 weeks after MI had no sion died. A second study of exercise testing was independent prognostic value. performed in patients with electrical and/or mechanical complications during their acute MI.47 DeBusk and Dennis51 applied a stepwise risk stratification procedure sequentially combining Jelinek et al,48 also from Melbourne, presented historical, then clinical characteristics and finally their findings in 188 patients with an uncompli- treadmill test results in a study population of 702 cated MI. All underwent bicycle testing on the day consecutive men less than 70 years of age and alive of discharge (about day 10) and returned to work 21 days after an acute MI. Prior MI or angina, or at a median of 6 weeks post MI. They considered recurrence of pain in the cardiac care unit (CCU) the total duration of exercise, maximal heart rate, identified 10% of the patients with the highest rate maximal blood pressure, and ST-segment shifts. of reinfarction and death within 6 months (18%). Secondary risk factors for recurrence of heart Clinical contraindications to exercise testing iden- attack were found to be angina before the MI, tified another 40% with an intermediate risk (6%). angina on the exercise test, and CHF. There was Exercise test results included ST-segment shifts, no difference between the two groups for maximal the MET level, angina pectoris, peak heart rate, workload, maximal heart rate, maximal SBP, or peak SBP, exertional hypotension, and PVCs. In maximal double product. The risk factors for total the patients who underwent treadmill testing, an events were angina before MI, angina during exer- abnormal test identified a high-risk group (10%), cise testing, and x-ray findings of CHF. No other whereas those with a negative test had a 4% inci- variables were predictive, including ST depres- dence of hard medical events. No other treadmill sion, but only chi-square analysis was performed. responses were predictive. Stanford Studies Montreal Heart Institute Studies Sami et al49 studied the prognostic value of tread- Theroux et al52 studied the prognostic value of mill testing in 200 males who were tested serially a limited treadmill test performed 1 day before approximately five times each from 3 to 52 weeks hospital discharge after an MI in 210 consecu- tive patients. These patients were followed for

C H A P T E R 9 Exercise Testing of Patients Recovering from Myocardial Infarction 303 cardiovascular endpoints for 1 year. Exercise capac- rate-pressure product (HR × SBP) from rest to ity and the BP response were not considered. Sixty- maximal exercise adjusted for age was empirically five percent (28 of 43) who had angina during found to be discriminating. Mortality increased treadmill testing reported the onset of angina among patients with major PVCs. Those with a subsequently, according to the authors. In those small increase in rate pressure product and/or with a normal ECG response to exercise testing, arrhythmia had a 5-year survival of 55% versus there was 2% mortality and a 0.7% sudden death 80% in the others. In their 1982 study, they con- rate; in those with ST-segment depression, there sidered clinical parameters as well. Clinical sub- was a 27% mortality (17 of 64) and a sudden death groups were defined as (1) patients with clinical rate of 16%. heart failure during hospitalization and/or previ- ous MI, (2) patients with anterior MI versus inferior Waters et al53 reported an expansion of the initial or indefinite MI. Within each clinical group, exer- study from the same institution. During 1976 to cise tests still determined a high-risk and low-risk 1977, 12% of all patients admitted died in the hos- group. Follow-up was complete at 6 years. pital, 28% were excluded from the study and 60% were included and underwent exercise testing. Madsen and Gilpin58 reported findings from Over the 5- to 7-year follow-up of the 225 patients symptom-limited bike testing at Grostrup Hospital tested, 16% had CABS. ST elevation and ST depres- in Denmark. The study population included 886 sion were similar risk predictors, and so they were patients discharged between 1977 and 1980 after combined. Target heart rate was considered to be an MI. During the 1-year follow-up, few patients 70% of predicted maximal heart rate and the max- were on beta-blockers and no one underwent CABS. imal workload was 5 METs. In the first year, overall Madsen considered angina, ST-segment depression, mortality was 11% and it was 3% per year after- PVCs, duration of exercise, maximal heart rate, and ward. Exercise-induced ST-segment depression was maximal rate pressure product as possible risk present in 31% and generated a risk ratio of 8× for markers. The most important exercise test variables 1 year mortality; 12% had ST elevation and the were duration of exercise and PVCs. Prediction of risk ratio was slightly less than with ST-segment death was not different with clinical or exercise test depression; 28% had PVCs and 9% had a flat BP variables or their combination. For reinfarction, response. Predictors by the Cox regression model the predictive value was significantly higher for the differed from the first year to the second year of exercise test variables than the combined set. follow-up. During the first year, ST-segment shift in either direction or a flat BP response were pre- Jespersen et al59 from two Danish Hospitals dictors. During the second year, a history of MI, have reported a series of 126 consecutive patients the QRS score, or PVCs were independent risk selected because they could exercise and had no predictors. evidence of prior MI, unstable angina pectoris, or severe heart failure and were younger than 71 years Wilford Hall USAF Medical Center of age. The nine patients with ST-segment depres- sion and subsequent cardiac events did not differ in Koppes et al54 have presented their results in a any of their clinical or exercise test features from highly selected group of 108 patients with MI of a the patients without ST-segment depression. One group of 410 admitted to Wilford Hall Air Force patient who had ST-segment depression under- Medical Center from 1975 to 1978. Starling et al55 went CABS because of angina refractory to medical have reported results using treadmill testing in management. During the year of follow up, there 130 patients after an uncomplicated MI. were nine major cardiac events, six being fatal, in the 46 patients who developed ST-segment depres- Denmark sion. Only three cardiac events (all deaths) occurred in 80 patients without exercise-induced ST-segment Saunamaki and Andersen56 in Copenhagen depression. The subgroup with exercise-induced reported the prognostic value of the exercise test ST-segment depression had annual death rates 3 weeks post MI. They considered the general and reinfarction of 13% and 17%, respectively, prognostic importance of ventricular arrhythmia and the annual rate of cardiac death was 4% in the associated with the exercise test, LV function, and subgroup without ST-segment depression. The ST-segment changes. ST-segment deviation was estimates of cardiac event-free probability showed not associated with endpoints. The change of a significantly worse prognosis for patients with ST-segment depression. Exercise-induced angina pectoris was not predictive for further cardiac events. There was no significant difference for rate

304 E X E R C I S E A N D T H E H E A R T pressure product, estimated VO2 or arrhythmia in with triple-vessel disease; 1% in patients with an those with cardiac events. EF greater than 30% or with one- or two-vessel disease. Spain Fioretti et al62 from the Thorax center in Velasco et al60 reported their findings using exer- Rotterdam have evaluated the relative merits of cise testing after an uncomplicated transmural MI. resting EF by radionuclide ventriculography and From 1973 to 1978, 958 patients with a prelimi- the predischarge exercise test for predicting prog- nary diagnosis of MI were admitted to their CCU. nosis in hospital survivors of MI. The Frank leads Men younger than 66 years old with a transmural were computer processed; 43% had abnormal ST- MI, who survived, were considered for the studies. segment depression and approximately 40% were This study is flawed by the large dropout rate (over on beta-blockers. The hospital mortality was 13% 50% of those tested chose not to be followed) and and 19 additional patients of 214 died in the sub- by the use of only univariate analysis. sequent follow-up (9%). Mortality was 33% for patients with an EF less that 20%, 19% for patients Houston with EF between the 20 and 39, and 3% for patients with an EF greater than 40%. Mortality was high Weld61 reported the results of low-level exercise (23%) in 47 patients excluded from performing testing on 236 of 250 patients who had diagnosed exercise tests because of heart failure or other acute MIs. Angina was not found to be useful in limitations. The patients could be stratified fur- predicting outcome. The exercise test variables ther into intermediate, low-risk groups according ranked in the following order: (1) exercise duration, to an increase in SBP during exercise. Maximal (2) PVCs, and (3) ST-segment depression. Patients workload, angina, ST-segment changes, and PVCs unable to reach an exercise capacity of 4 METS were less predictive. After discharge, 14% of the had a relative risk of 15×. Exertional hypotension patients had clinical signs or symptoms of heart (a maximal SBP of less than 130) generated an failure and 38% had angina; 17 were treated with odds ratio of 5 but a drop in SBP was not predic- bypass surgery or angioplasty. This study was later tive. Standardized regression coefficients showed expanded to 405 patients and similar results were that all three exercise variables had a stronger obtained. Discriminant function analysis demon- association with 1-year cardiac mortality than any strated that the combination of clinical and exer- of the clinical variables. However, by this multi- cise variables gave better predictive accuracy than variate analysis, ST-segment depression was not either used alone. statistically associated with 1-year mortality. New Zealand The Netherlands Norris et al63 from Greenlane Hospital reported De Feyter et al41 from the Free University Hospital the determinants of reinfarction and sudden death in Amsterdam have reported the prognostic value in male survivors of a first MI who were younger of exercise testing and cardiac catheterization than 60 years of age. All underwent exercise test- 6 to 8 weeks after MI. Their study provides data ing and coronary angiography 4 weeks after their on a consecutive series of 179 survivors of acute MI. Between January 1977 and June 1982, 425 suit- MI who had a symptom-limited Bruce test. able men were admitted to the hospital. Of these, They considered the number of vessels, EF, LV 7% died in the hospital, leaving 395 survivors. Of end-diastolic pressure, wall motion abnormali- these 395, 315 (80%) underwent exercise testing ties, and left anterior descending coronary artery and 325 (82%) underwent coronary angiography. (LAD) involvement. Fifty-eight patients with at Exercise testing was performed at 2.5 mph start- least 10 METs had a very low risk for cardiac death ing at 0% grade and gradually increasing to 15%. or reinfarction. Patients having no treadmill Total cardiac mortality was best predicted by EF and markers resulted in a higher risk group, whereas by a coronary prognostic index dependent on age, three-vessel disease or a LV EF of 30% history of infarct, and chest x-ray scan. Neither the or less did predict high risk. The mortality rate severity of coronary artery lesions nor the results of was 22% in patients with an EF less than 30% or exercise testing predicted mortality. Reinfarction could not be predicted by any clinical or angio- graphic variable.

C H A P T E R 9 Exercise Testing of Patients Recovering from Myocardial Infarction 305 United Kingdom Thirty-two percent were readmitted (7% for CABS) with a death rate of 14%. The relative risk Jennings et al64 at Newcastle on Tyne considered of death in the first year after readmission was 1253 patients admitted over 1 year to their CCU; 2.6× greater than for patients who did not have a 503 sustained an MI but only 289 were younger readmission. Only an EF less than 40% and angina than 66 years of age. Of these 289, 18% died in the following an MI were predictive of readmission. hospital and 36% were excluded from study because Reinfarction was best predicted by predischarge of left bundle branch block, ischemic pain, or other angina that carried a risk ratio of 2.5×. Failure to complications; 49 could not be tested before dis- perform the exercise test was significantly associ- charge for logistic reasons. Using univariate analy- ated as well with reinfarction, but none of the sis, exertional hypotension generated a risk ratio treadmill variables were discriminating. of 8×, inability to complete the protocol a risk ratio of 8×, and an excessive HR response a risk ratio of Canada 4×. No survival analysis techniques were employed; only chi-square and t-tests were used. Williams et al68 from Ottawa Civic Hospital com- pared clinical and treadmill variables for the pre- Handler65 from Guy’s Hospital in London diction of outcome after MI. They considered the reported using submaximal predischarge exercise relative prognostic merits of 15 clinical and 10 pre- testing on 339 consecutive patients’ age 66 years or discharge exercise test variables in 226 patients. younger. Although abnormal ST-segment depres- A submaximal treadmill test was performed on sion generated a risk ratio of 6, which was not sig- 205 patients (88%) to a mean workload of 6 METs nificant, ST elevation and combined elevation and after an average of 12 days after MI. During the depression had risk ratios greater than 10 that were first year of observation, 3.4% of the patients devel- statistically significant. An abnormal BP response oped unstable angina, 6.8% had a recurrent infarc- and ST-segment elevation also predicted heart tion, and 6% died. Twelve percent underwent failure. coronary bypass surgery. Among those who did not have a treadmill test, there was a 31% death rate. Multicenter Post-MI Research The predictors of death were found to be resting ST- Group segment depression, a high creatine phosphokinase, a poor exercise tolerance, and a history of prior MI. Krone et al66 reported the experience of the Multicenter Post-MI Research Group using low University of California San Diego level exercise testing after MI. Fourteen hundred (UCSD) Specialized Center for and seventeen patients met their criteria and 866 Organized Research (SCOR) consented. Of those who consented to be in the study, 77% performed the treadmill test. Of those Madsen and Gilpin69 attempted to answer two who exceeded a SBP of 110 during testing, there important questions: Can an “ischemic” exercise test was 3% mortality versus 18% for those unable to do response and the exercise capacity be predicted so. In those that had an absence of couplets, there from historical and clinical data available during was 4% mortality, whereas it was 13% in those with hospitalization? Can the patients at low or high couplets. In patients with a normal exercise blood risk of death or new MI be identified by the exercise pressure and no pulmonary congestion on the test? To answer these questions, they analyzed chest x-ray scan, there was a 1% mortality versus data from 1469 patients discharged after an acute 13% in those with either abnormality. Most of the MI from four hospitals. Of these patients, 466 or results are presented in univariate form with 32% underwent a treadmill test at discharge. The Fisher’s exact test evaluation. Further analysis of exercise test was an optional part of the SCOR selected clinical and demographic variables using multicenter study protocol. The main reasons stepwise logistic regression demonstrated that for not performing an exercise test were advanced exercise results significantly improved the predic- age, poor general condition, severe cardiac dysfunc- tion model for cardiac death. In this same study tion, or complicating diseases. The 466 patients, population, Dwyer et al67 reported the experience who underwent exercise testing, had a lower fre- with nonfatal events in the year following an quency of clinical risk factors than patients that acute MI. Radionuclide ventriculography and did not undergo exercise testing. Various treadmill Holter monitoring were performed on all subjects and treadmill tests were performed in 76%.

306 E X E R C I S E A N D T H E H E A R T protocols were used but MET levels were calcu- an individual for clinical reasons from undergoing lated. Limiting conditions of exercise tests were exercise testing. Possible biases as a result of this angina in 16%, marked ST-segment changes in clinical selection process, as well as the characteris- 7%, fatigue in 44%, shortness of breath in 17%, tics associated with being admitted to the academic claudication in 4%, and severe arrhythmia in 2%. centers from which these reports come, must be If no symptoms developed the patients continued considered. Specific summaries grouped by each exercise until they approached 75% of maximal of the exercise test risk markers follow. Only stud- age-adjusted heart rate. In the 9% of patients ies reporting statistically significant results are without limiting symptoms, where the exercise explicitly cited. From the previous summaries of test was stopped at a low heart rate, the test was each study, where the definitions for an abnormal considered indeterminate. Patients taking beta- responses were given, it is apparent that often sev- blockers were included if a heart rate greater than eral different responses under each heading are 100 beats per minute were achieved above being considered together by summarizing across 6 METS. Medications taken during the testing studies (i.e., the thresholds for abnormal PVCs, time included digoxin in 26% and beta-blockers exercise capacity, or SBP response differ). In addi- in 53%. Ninety-two patients with indeterminate tion, the various investigators considered not all test results were excluded, leaving 374 patients. of the exercise predictors; such studies are indi- cated in Table 9-3 with an NR for “not reported” Four historical variables from hospitalization in the appropriate test response column. were chosen as predicting an ischemic exercise test response by discriminate analysis. These included The five exercise test variables suggested to have previous angina, ST-segment depression at rest, prognostic importance are ST-segment depression beta-blocking agents on discharge, and age; (and sometimes elevation), exercise test-induced however, prediction was poor. In the 295 patients angina, poor exercise capacity, or excessive heart followed 1 year with satisfactory exercise tests, rate response to a low workload, a blunted SBP among exercise test variables tested univariately, response (or exertional hypotension), and PVCs. only exercise capacity in METS and the occur- Because they involve the same populations and rence of exercise-induced ST-segment depression institutions and usually obtained the same results, were important for predicting death and/or new MI the following studies are grouped together: within 1 year. A discriminate analysis using all exer- Theroux and Waters (Montreal Heart Institute); cise test variables selected only the exercise capac- Sami, Davidson, and DeBusk (Stanford); Hunt and ity in METS. Total correct classification was 75%. Srinivasan (Royal Melbourne Hospital), Krone In the low-risk group of patients (72% of patients and Dwyer (Multicenter Post-MI Group), and with an exercise capacity greater than 4 METS), Fioretti (1984 and 1985, Thoraxcenter). Thus, the fewer than 2% died or had a new MI within 1 year. results from a total of 24 centers are considered. In the high-risk group of patients (29% of patients with an exercise capacity less than or equal to Exercise-Induced ST-Segment Shifts 4 METS), 18% had a cardiac endpoint. They con- cluded that an ischemic exercise test response ST Depression. Of the 28 centers, 9 found ST- could not be reliably predicted from historical or segment depression to be significantly predictive clinical variables from the hospitalization. Using of subsequent death; additional 6 centers reported a age and ST-segment changes at rest would iden- positive, but insignificant, association; and 9 centers tify patients likely to have good exercise capacity. reported a null effect with 4 of the 28 failing to Good exercise capacity is the most important report data on ST-segment depression. exercise test variable for identifying those with a very low risk of death and new MI within a year. A ST Elevation. Sullivan et al70 evaluated the prog- group of patients at relatively high risk can be nostic importance of exercise-induced ST-segment identified by a poor exercise capacity. elevation in 64 patients who underwent submaxi- mal exercise testing a mean of 11 days after an acute Summary of Prognostic infarct. Follow-up was for 1 year. The presence of Indicators from Exercise Tests exercise-induced ST-segment elevation was the only exercise test variable that predicted cardiac The inconsistencies found in these studies make death. De Feyter et al41 found that ST-segment it difficult to develop an algorithm for interven- depression indicated multivessel disease, whereas tion in patients with history of MI. One of the best ST-segment elevation indicated advanced LV wall means of selecting a high-risk group is to exclude motion abnormalities and a low EF. Both shifts indicated that both multivessel disease and

C H A P T E R 9 Exercise Testing of Patients Recovering from Myocardial Infarction 307 advanced LV wall motion abnormalities existed. In Attempts to establish risk have included scores Water’s study, ST-segment elevation generated the based on clinical features of the MI and historical same univariate risk as did depression and so they information such as the Norris and Peel indices. were considered together. However, location of the There are reasons other than prognostication for ST shift was not specified. Saunamaki and Andersen performing exercise testing, but given the need to considered ST-segment depression and elevation cost-account, all possible justification for perform- separately, but did not specify its location. In their ing a procedure is needed. study, the ST responses were found to have little prognostic value. Handler65 found ST-segment Kentala et al assessed clinical parameters, elevation and combined depression and elevation including a careful history of prior activity level. to generate significant risk ratios. Elevation was The prognostic power of clinical and ECG vari- more common in anterior MIs. ST-segment eleva- ables recorded soon after MI, and in connection tion also predicted heart failure. These results are with the exercise test, were analyzed by stepwise too inconsistent to make a conclusion. multiple discriminant analysis. They found that both clinical and exercise variables were important. Exercise-Induced Arrhythmia Patients dying within 2 years had a low exercise systolic BP. With longer follow-up, the exercise BP Only 5 of 28 centers reported exercise test-induced had a weaker impact. At the 4- and 6-year points, PVCs to indicate a significant increase in risk. Four an abnormal resting terminal P wave was the best centers did not include results regarding PVCs; predictor of poor prognosis. This probably identi- nine centers reported null or negative associations fied a group with mild heart failure. For patients of PVCs with mortality. who suddenly died after 2 years, the T-wave changes after exercise, which possibly indicated subendo- Exercise Capacity cardial injury, were common. Patients with a high level of physical activity before their MI were less Nine centers of 28 reported that a low exercise prone to die suddenly. Of the many factors consid- capacity and/or an excessive heart rate (HR) ered, an abnormal apical impulse, T-wave inver- response to exercise indicated a high-risk group. sion after exercise, prior CPR, sedentary lifestyle Five additional centers reported nonsignificant before infarction, and occurrence of PVCs during positive associations, Stanford reported a positive exercise were of discriminatory value in relation association in only one of three studies, whereas 10 to sudden death. of the 28 centers failed to report sufficient data on this variable to assess its effect. Granath et al44 found that analysis of clinical data in the CCU failed to produce any differences Exercise-Induced Angina between survivors and those who died, although there were more deaths among those patients who Only 5 of 28 centers reported exercise test-induced had a previous MI. Saunamaki and Andersen57 angina to indicate a significantly increased risk demonstrated that exercise testing variables, group. Eight centers failed to report angina data. including PVCs, and a poor SBP HR change in Seven of the remaining 11 reported nonsignifi- response to exercise still were able to predict risk cant positive associations. within the strata of CHF, prior MI, and anterior MI. The exercise variables outperformed these impor- Systolic Blood Pressure Response to Exercise tant clinical parameters. Weld61 found the exercise test variables of duration, PVCs, and ST- segment Nine of 28 centers found that inadequate or abnor- depression to be ranked in that order ahead of the mal SBP response to exercise significantly identi- clinical variables of x-ray vascular congestion, prior fied a high-risk group; 11 of the centers failed to MI, and x-ray cardiomegaly in predictive value. report data, and four of the remaining six reported a nonsignificant positive association. De Feyter et al41 were unable to identify a higher risk group from treadmill markers, whereas three- Comparison of Exercise Data to vessel disease or a LV EF of 30% or less did. Clinical Data Madsen and Gilpin58 found that in those who under- went testing, clinical variables were better able to An important question to be resolved is does predict outcome than in the nontested group. The the exercise test give more predictive informa- most important exercise test variables were exer- tion than the standard clinical risk predictors do? cise duration and PVCs; however, they improved prediction of reinfarction but not death. Although exercise test variables were selected by discriminant analysis, the correct total classification of deaths

308 E X E R C I S E A N D T H E H E A R T and survivors was not improved. The total correct these events, but none of the treadmill variables prediction was 71% for clinical data alone, 67% was discriminating. for exercise data alone, and 71% for both combined. Waters et al53 found that predictors by the Cox regression model were different in the first DeBusk et al51 found that prior MI or angina, and the second year of follow-up. During the first or recurrence of pain in the CCU identified the 10% year, ST-segment shift in either direction, a flat of patients with the highest rate of reinfarction and BP response or angina within the 48 hours after death within 6 months (18%). Clinical contradic- admission were predictors (“markers of ischemia”). tions to exercise testing identified another 40% During the second year, a history of MI, the QRS with an intermediate risk (6.4%). In those who score, or PVCs was independent risk predictors underwent treadmill testing, ST-segment depres- (“markers of LV dysfunction”). sion and low peak workload were selected before any clinical variables or ambulatory ECG data in In summary, the results are mixed regarding the logistic regression analysis. whether the exercise test gives information that can predict death and reinfarction better than the Norris et al63 found that total cardiac mortality clinical features. Remember that clinical judgment was best predicted by EF and by an index depend- to exclude patients from testing identifies the high- ent of age, history of MI, and chest x-ray scan. est risk group and that the threshold for doing so Neither the severity of coronary lesions nor the must be quite variable between locations. results of exercise testing predicted mortality. Any clinical exercise test or angiographic variable Clinical Design Features could not predict reinfarction. Williams et al68 con- sidered the relative prognostic merits of 15 clinical The column headings used in Table 9-3, and sep- and 10 predischarge exercise test variables in arately listed in Table 9-4, are the important fea- 226 patients. The predictors of death were found tures of the study design that could affect the to be resting ST depression, a high creatine phos- findings. Following is a discussion of these features. phokinase, a poor exercise tolerance, and a his- tory of MI. TA B L E 9 – 4 . Characteristics that could differ as to methodology among studies Jennings et al64 found that the Norris index score (age, prior MI, x-ray scan abnormalities) of Patients excluded less than 3 was associated with a 12% mortality Entrance criteria and a score of more than 12 with a mortality of Age range; gender 85%. Fioretti et al62 evaluated the relative merits Infarct mix (i.e., non-Q wave, inferior/anterior/lateral of resting EF by radionuclide ventriculography Q wave) and the predischarge exercise test. Mortality was Patients with prior MI and those with complications 33% for patients with an EF less than 20%, 19% included or not for patients with EF between 20% and 39%, and Prior coronary artery bypass surgery or PCI 3% for patients with an EF greater than 40%. History of congestive heart failure and angina Mortality was high (23%) in 47 patients excluded MI size from performing exercise tests because of heart Follow-up thoroughness and length failure or other limitations. Percentage of patients undergoing CABS or PCI during follow-up and whether they are censored Krone et al66 found that among those not able Cardiac events (problems with using CABS as to take a treadmill test, there was a 14% mortality an endpoint) compared with 5% in those who were able to take Mortality during follow-up (are they a high- or it. In patients with a normal exercise blood pressure low-risk group?) and no pulmonary congestion on the chest x-ray Reinfarction rate scan, there was a 1% mortality versus 13% in those Exercise protocol with either abnormality. In this same population, Time post-MI test performed Dwyer et al67 reported the experience with nonfatal Endpoints of test events in the year following an MI. Thirty-two per- Leads monitored cent were readmitted (7% for CABS) with a death Medications taken after discharge from hospital and at rate of 14% and a risk ratio of 2.6. Only an EF less time of exercise test than 40% and post-infarction angina were predic- Test responses considered (PVCs, ST segment, blood tive of readmission. Reinfarction was best pre- pressure, exercise capacity, angina) dicted by predischarge angina. Failure to perform Statistical methods the exercise test was significantly associated with CABS, coronary artery bypass surgery; MI, myocardial infarction; PCI, percutaneous coronary intervention; PVCs, premature ventricular contractions.

C H A P T E R 9 Exercise Testing of Patients Recovering from Myocardial Infarction 309 Exercise Protocol. Bike protocols, especially a was more common among patients with non- supine protocol, can give different responses than Q-wave MIs. This review and other data support the a treadmill. Most protocols were continuous but concept that ST depression with exercise effectively some were not progressive in workload increments. stratify patients following a non-Q-wave MI. This The standard Bruce protocol starts at a relatively group is now considered in the ACS category with high workload (4–5 METS). The protocol as well unstable angina. as beta-blockade, fitness, and anxiety can affect heart rate responses at submaximal levels. The failure of exercise-induced ST-segment depression to consistently be associated with Endpoints of Exercise Test. If stopped at a certain increased risk in patients after MI was hard to amount of ST-segment shift, MET level, or heart explain. This failure could be a result of popula- rate, then this response could not be considered tion differences and the resting ECG. To test this as a continuous variable nor could a higher value, we studied 198 males who survived an MI, under- which might be more discriminating, be reached. went a submaximal predischarge treadmill test, and were followed-up for cardiac events for 2 years.71 ECG Leads Monitored. Use of different electrode Abnormal ST-segment depression was associated placements can make comparisons between stud- with twice the risk for death and the risk increased ies difficult, but probably does not have a great to 11 times in patients without diagnostic Q waves, impact. similar to the results by Krone et al72 in patients with an initial non-Q-wave MI. These results sug- Time Post MI When Exercise Test was Performed. gest that the difference in the prognostic value of “Stunned” myocardium and deconditioning affect the post-MI exercise ECG between studies is due predischarge testing more than they affect hemo- to variations in the prevalence of the patterns of the dynamic responses later. ST-segment responses rest ECG among study populations. Angiographic appear more labile earlypost MI. The responses studies, however, have demonstrated that exercise- differ at various times post MI as well, with a spon- induced ST depression is associated with severe taneous improvement in hemodynamics occurring coronary artery disease whether Q waves are pres- by 2 months. The spontaneous improvement in ent. The conflicting results from follow-up and both EF and exercise capacity, but their failure to angiographic studies most probably relate to the correlate with each other, makes them difficult to fact that early mortality is strongly associated with interpret. The studies that included exercise test- LV damage, whereas later mortality is associated ing at multiple times found the same responses to with ischemia and severe coronary artery disease. have a different predictive value at the specific times the tests were performed. There is a sponta- Thoroughness and Length of Follow-Up. Those neous improvement during the first year post MI lost to follow-up most likely have a higher percent- in the blunted BP response to exercise that occurs age of deaths. In addition, follow-up affects analysis particularly in large anterior MIs. if censored data cannot be handled adequately with the statistical program. Mortality changes over time MI Mix (i.e., Q-Wave Location). Each have a dif- and predictors change. ferent prognosis and different “normal” response to exercise. Exercise predictors may be different Percentage of Patient Undergoing CABS (or in each type. PCI) During Follow-Up. CABS could alter mortal- ity and affect outcome prediction. In addition, Inclusion of Non-Q-Wave MIs. After much con- patients with ischemic predictors would be troversy regarding the risk of having a “subendo- selected to have this procedure more frequently. cardial” MI, a study from Mayo clinic appears to These patients should be censored at the time of clarify the situation.70 From 1960 to 1979, 1221 res- intervention but such censoring is not random. idents of Rochester, Minnesota had an MI as the first manifestation of CHD; 784 had a transmural Cardiac Events Considered as Endpoints. The only (Q wave) and 353 had a non-Q-wave MI. The 30-day hard endpoints that should be considered, from an fatality rate was 18% among transmural and 9% epidemiological point of view, are death and rein- in subendocardial MIs. No significant difference was farction. Separation or distinction of sudden death found in the rates of reinfarction, CABS, or mortal- makes little sense and may confuse the analysis, ity over the next 5 years. CHF was more common particularly if those with sudden death are com- among patients with transmural MIs, and angina pared with all others (including nonsudden cardiac death). Noncardiac deaths are often difficult to

310 E X E R C I S E A N D T H E H E A R T distinguish and lead to biased results but may play Although beta-adrenergic blockade attenuates the a confusing role, particularly in older popula- ischemic response, two long term follow-up stud- tions. CABS is not a valid endpoint and should be ies have demonstrated that these agents do not considered as a censored outcome. It is clearly interfere with poor exercise capacity as a marker related to certain exercise test results that physi- of adverse prognosis.76,77 Patients taking beta- cians feel motivated to “fix” with that procedure. blockers after an MI should continue to do so at “Instability” or progression of symptoms (CHF or the time of exercise testing. Because patients will angina) is a soft endpoint that should not be used take these medications for an indefinite period for epidemiological purposes. after infarction, the exercise test response while on beta-blockers will provide information regarding Mortality During Follow Up. If there is a low mor- the adequacy of medical therapy in preventing tality rate, more patients are needed to find a sta- ischemia and arrhythmias as well as controlling tistical difference between those with or without the heart rate and BP response during exercise. certain variables. Some studies have compensated Moreover, discontinuation of beta-blockers solely for this by using soft endpoints and combining for the purpose of exercise testing may expose the endpoints. patient to the unnecessary risks of recurrent ischemia, arrhythmia, and exaggerated hemody- Prior MI Patients Included or Not. Prior MI is an namic responses during exercise. important predictive variable that depends on the severity of the prior MI or MIs. Patients with prior Statistical Critique of the Prognostic Studies large MIs are biased toward being admitted with non-Q-wave MIs, because another transmural MI There are several general problems that are appar- increases their likelihood of dying before hospital- ent across many of the studies. The purpose of a ization. Few studies have tried to account for the specific study is not always clear; there is confu- number or severity of prior MIs. sion evident between the desire to develop a pre- diction algorithm that will be of practical clinical Exclusion Criteria. Clearly, clinical judgment use in patient treatment and the desire to demon- applied to the population who had a prior MI, to strate an association of exercise testing responses exclude patients from exercise testing, identifies to subsequent cardiac events in any form. the highest risk group. Though this process con- Development of a prediction algorithm requires siders complicating illnesses and age, cardiac dys- an approach to validation that is quite different function and ischemia are considered as well. from the testing of the statistical significance of an Because of this, alternative testing methods that effect, as is done in many of the studies. Although have been compared favorably with exercise test- effect size estimation is probably the most clini- ing have included right atrial pacing and electro- cally relevant procedure, most of the studies report physiologic stimulation studies. only significance test results, perhaps with some means or frequency differences cited. None of Age Range and Gender. Women are thought to the studies reported effect size estimates with have a higher MI mortality and certainly are known confidence intervals, even though this is the well- to respond differently than men to exercise testing. established method of reporting estimation results. Because of this, they should be considered sepa- rately, but the studies do not contain a sufficient Finally, many of the studies reviewed failed number for valid analysis. Death rates are directly to provide enough details about the data to related to age. allow independent evaluation of the investigators’ conclusions. Such details are especially neces- Medications Taken After Discharge from Hospital sary to compare results across different studies. and at the Time of the Test. Digoxin causes ST Recompilation of effects may be required to com- depression, but it is usually taken for CHF, thus pare studies that have reported results in different implicating an ischemic etiology for a potential formats. The number of “?” appearing in the exer- death because of dysfunction. Digoxin administra- cise test risk markers column of Table 9-3 illus- tion post-MI may actually be an independent risk trates how often data reported were insufficient to predictor and act by predisposing to ventricular compute even the direction of the associations in dysrhythmias. Beta-blockers affect BP and heart the study (whether the association is “significant”). rate response and improve survival but do not seem to impact the value of the exercise test.73-75 Common areas of difficulty include selection biases, a relatively rare outcome of interest, use of multiple endpoints, and unequal follow-up times.

C H A P T E R 9 Exercise Testing of Patients Recovering from Myocardial Infarction 311 Many of the studies fail to be specific enough significance tests. It is true that multivariate tech- about the target population of interest. Selection niques often have stricter assumptions than some biases in the patients studied may be too severe of the univariate techniques available and should for the results to be considered representative of not be used without initial screening with uni- the general population. However, the limited target variate analysis. Even if univariate estimates are population is carefully designing further research, given for comparison to other studies, the multi- even if the results are not generally applicable. variate results should be reported so that the extent Evaluation of possible biases requires information of adjustment necessary for inter-relationships can on patients who were eligible for the study but be assessed. declined to participate, or who dropped out of the study after their initial entry. A few of the investi- The other major analysis issue is the problem gators have reported on such nonparticipants or of unequal follow-up. Unequal follow-up that is follow-up losses, but many do not report more than not controlled in the design of a study must be the number of individuals involved. handled in the analysis of the data. Unequal fol- low-up of patients can be treated as censored data. The most desirable endpoint for analyses in A typical approach in biomedical research for these studies is cardiovascular death because the analysis of censored time-to-response data is to use aim of the test is to identify those benefiting from survival analysis techniques. This approach was cardiac interventions. One approach to attempt used in several of the more recent studies. However, to deal with small numbers of deaths is the use a fundamental assumption of most survival tech- of multiple endpoints, often combined. However, niques is that the censoring is random with respect this practice may obscure underlying relationships to the outcome of interest. This assumption cannot for several reasons. Endpoints other than death, be evaluated without reporting on those patients such as angina, cannot be well enough defined to who were lost to follow-up either because of drop- avoid extensive misclassification errors. A poten- ping out of the study or because of lack of com- tially more serious issue when endpoints are plete follow-up due to late entry into the study. combined is independent of the precision of the Information on those who have dropped out could endpoint measurement. Different endpoints may be gathered by death certificate searches or other be related to different mechanisms and thus may techniques; reports on such persons are often miss- have different associations with the test markers. ing from the studies reviewed. Including patients Such differences confound any attempt to measure who are censored observations because of short fol- associations using combined endpoints. Perhaps low-up time must be considered carefully, because the worst pitfall is the use of an endpoint to assess the risk of subsequent cardiac events is known to associations that may be influenced by the exercise change with time. Multivariate approaches to sur- test result; studies that have included CABS or vival analysis are available using proportional haz- PCI as an endpoint have fallen prey to this trap. ard regression models or other hazard functions. However, these models may be relatively insensitive Finally, the problem of unequal follow-up peri- to modeling of interactions among the variables. ods of patients cannot be ignored. This problem In addition, the results may not be readily inter- can be circumvented in the design of a study by pretable in terms useful to clinicians. using a limited period for entry into the study, with follow-up that allows the study to be completed Other approaches to the problem of censored with sufficient events. This approach requires data are possible. One solution often used in epi- that the follow-up time be limited enough to min- demiological research is computations in the form imize loss-to-follow-up problems. Adjustments for of events/person-time or person-time incidence. unequal follow-up time can also be made in the Another approach that avoids the inclusion of analysis phase of the study, but these were not short-term follow-up patients is to stop entry into used in most of the studies. the study early enough so that all patients avail- able can be followed for a fixed time. A limited, fixed Only one fourth of the research centers reported time of follow-up can also help reduce the number any use of multivariate techniques. Computer pro- of dropouts, because the likelihood of losing a grams for such analyses were certainly widely avail- patient from the study increases with time. One able after 1980; only 5 of the 28 centers have reports approach to be avoided that was used in several of limited to before 1981, when access to such analysis the studies is to merely count events in various sub- tools may have been more difficult. None of the groups without regard to differences in follow-up studies reported multivariate estimates of effect, time. Data that is reported in such a way is essen- even though the effect estimate is at least as sensi- tially meaningless. tive to error from exclusive univariate analysis as

312 E X E R C I S E A N D T H E H E A R T Survival analysis is appropriate when outcome partitioning, and nearest neighbor. Variables used measurements represent the time to occurrence of were identified as predictive univariately from the some event (i.e., death or reinfarction). If differ- base hospital and were obtained during the first ences in important covariates or prognostic vari- 24 hours. Linear discriminant analysis assumes ables exist at entry between the groups to be normality among the predictor variables, whereas compared, the investigator must be concerned with logistic regression is based on the assumption the analysis of the survival experience as influenced that the log of the classification function is a lin- by that difference. To adjust for these differences in ear function of the fitted coefficients. Recursive prognostic variables, stratified analysis or a covari- partitioning makes no assumption regarding nor- ance type of survival analysis could be done. If there mality and can detect interactions among variables are many covariates the number of strata can and handles missing data. The nearest neighbor quickly become large, with few subjects in each. procedure is based on the concept that in the Moreover, if a covariate is continuous, it must be multidimensional space defined by the variables, divided into intervals and each interval assigned a patient would likely have the same outcome as to a score or rank before it can be used in a strati- another patient in that space. It cannot detect fied analysis. Cox proposed a regression model that interaction or assign importance. Linear discrim- allows analysis of censored survival data adjusting inant analysis, logistic regression, and recursive for continuous and discrete covariates, thus, avoid- partitioning performed similarly within a given ing these two problems. This model, also called the population, although each used the information proportional hazard model, assumes that the haz- contained in the prognostic variables differently. ard rate or “force of mortality” can be expressed as Application between different populations of pre- a product of two terms. Available statistical pack- diction schemes based on linear discriminant ages allow incomplete data; that is, there are cases analysis and logistic regression was shown to be for which the response is not observed but the feasible but prior validation is essential. data (time in study) are included in the analysis. This could occur in the study of survival where an Temporal changes in risk. It is well documented individual may remain alive at the close of the that changes in the risk of subsequent cardiac observation period or may drop out before the end. events occur within the first year post MI. Such The Cox survival analysis allows covariates that underlying changes in the hazard function suggest can be selected in a stepwise fashion. The covariates that there may be temporal changes in the effects or prognostic factors usually represent either of any related risk markers. Evaluation of this inherent differences among the study subjects or effect requires time-dependent modeling or con- constitute a set of one or more indicator variables ditional analysis with respect to time. Waters et al53 representing different groups. The covariates may are the only investigators to have addressed this also describe changes in a patient’s prognostic problem. One expected effect of not considering the status as a function of time. The Cox proportional temporal changes in risk is that estimates of effect hazards regression model presumes death rates size may be biased toward the null over intervals may be modeled as log-linear functions of the that span several risk periods. covariates. A regression coefficient is estimated, which relates the effect of each covariate to the Meta-Analysis Considerations survival function. Meta-analysis is a statistical approach to develop a The Cox model is currently favored; however, consensus from an existing body of research. It is few investigators have compared the various tech- a quantitative approach to reviewing research using niques in one data set. Madsen et al78 compared a variety of statistical techniques for sorting, classi- two software versions of the Cox multivariate analy- fying, and summarizing information from the find- sis, stepwise discriminant analysis, and recursive ings of many studies. It is also the application of partitioning. They concluded that all four tech- research methodology to the characteristics and niques gave equally precise prognostic evaluations findings of studies. This includes problem selection, but that recursive partitioning was easier to use hypothesis formulation, the definition and mea- and the Cox models were more accurate. The surement of constructs and variables, sampling, UCSD SCOR group evaluated several multivariate and data analysis. statistical methods in two different hospital popu- lations to predict 30-day mortality and survival fol- The application of meta-analysis to a body of lowing MI.79 The methods evaluated were linear research involves three stages. First, a complete lit- discriminant analysis, logistic regression, recursive erature search is conducted which is analogous to the collection of data in an experimental study.

C H A P T E R 9 Exercise Testing of Patients Recovering from Myocardial Infarction 313 Second, the important characteristics and findings or submaximal endpoints for exercise testing. Other of relevant studies are classified. Third, statistical exploratory subgroupings included examining techniques are applied to the compiled data. This American studies versus studies from other coun- last stage can involve descriptive, correlational, and tries and selection of the “best” studies. Subset inferential statistical analysis. The statistical tech- analysis by whether women or patients with prior niques applied here are sign testing, correlation, or non-Q-wave MIs were included was limited by and weighted regression analysis. Sign testing is a the small number of studies that used these fea- statistical test that evaluates the proportions of tures as exclusion criteria as well as by the surpris- findings and determines if they are related by more ing number of studies that did not make this than chance. information available. The same could be said for analyzing the data regarding cardiac medications. Although scientific truth relies on repro- Therefore, the percentages of these clinical fea- ducibility, clinical studies often do not agree tures were correlated and regressed against the because of the effect of confounding variables that risk ratios found for the exercise risk markers. at times can be accounted for by statistical tech- niques. When applying meta-analysis, it became As a result of the varied statistical treatments apparent that an electronic spreadsheet facilitated used by the studies analyzed, as well as the lack of the process. After word processing, electronic complete reporting by some investigators, exercise spreadsheets are the most common software used responses were associated with relative risk by in microcomputers. The first of these was VisiCalc both quantified univariate or multivariate analysis (1979) and its introduction was the greatest impe- or by unquantified multivariate analysis. Risk tus for the use of personal computers in the office. ratios from the former were used in regression and These programs create a matrix of cells indexed correlation analysis whether the reported value by column and row headings. The cells can be was statistically significant. All values were plotted adjusted for size and data presentation. Once data in x–y graphs. For plotting purposes, if the risk is entered, it can be moved, deleted, copied, ratios were not available, arbitrary values of posi- sorted, and subjected to mathematical manipula- tive or negative 1.6 and 1.3 are used for significant tion. However useful these programs have been and nonsignificant multivariate ratios, respectively. in business, there has been little application in medicine. Initial data analysis consisted of the construc- tion of a correlation matrix consisting of Pearson To identify the studies previously presented, product moment correlations for the risk ratios of Medline was searched using the keywords of exer- the five exercise test responses. The risk ratios were cise testing and MI. Studies were included if they correlated with each of the following clinical vari- attempted to evaluate the relationship between ables: percent tested, percent females, percent of exercise test variables and cardiac events during a prior MI, percent with each Q-wave location, per- follow-up period and were published before 1986. cent on digoxin, percent on a beta-blocking agent, The review data was entered into the spreadsheet percent subsequent mortality if tested, and percent and tables directly printed from the program. The mortality if not tested). spreadsheet program allowed very flexible data entry. Column and row headings were specified As part of meta-analysis, sign testing was without excessive care for priority, appearance, or applied to the findings in Table 9-3. Because it order, because data ranges could be easily moved, is not possible to ascertain the directions of the ordered alphabetically or numerically, copied, or nonsignificant associations listed as “?” in Table 9-3, deleted. Graphic capabilities made it possible to meta-analysis conclusions must be tentative. Some present the data in various graphic formats (pies, researchers probably did not evaluate markers that bar, and x–y plots) and to visualize the relation- are not reported, but others are likely to have ships between data in columns and/or rows. failed to report null or negative findings. The Facile identification and separation of subgroups most generous evaluation would be to omit stud- were possible; the latter being the second step in ies that did not report results for a particular the application of meta-analysis. marker; the most conservative approach would be to include these studies and to assume that any Initial analysis consisted of searching and unreported results were not positive associations. sorting findings within the spreadsheet. Studies Results are presented for both situations with upper were categorized by predischarge testing (arbi- and lower bounds on the overall published results trarily set at <3 weeks post MI) and postdischarge on exercise test markers as predictors of death. testing (≥3 weeks). The studies were then sub- grouped to see if differences were due to maximal If there were not a true underlying association of a risk marker with death, we would expect that

314 E X E R C I S E A N D T H E H E A R T 50% of studies would report positive association or published and complete data on all risk mark- based on chance. “Statistical significance” is not ers evaluated may not be reported. Often not even considered here; only the directions of the observed the direction of a possible effect can be computed associations. Using only studies with any reported for a particular exercise test result. effect as the denominator, the generous estimates of the percentages of positive associations reported Two studies from the prethrombolysis era were for BP (12 centers reporting positive associations/ not included because they were only reported 18 centers presenting any results for SBP), PVCs after a long follow-up period. Between 1979 and (14/23), ExCap (14/18), Angina (12/20), and ST 1983, 1773 consecutive patients were admitted to (15/19) are 72%, 61%, 78%, 60%, and 63%, respec- Glostrup County Hospital in Denmark with an tively. Only the SBP and exercise capacity propor- acute MI. Of 1430 patients who were alive after tions are significantly different from chance by 3 weeks, 718 performed an exercise test.80 Survival a sign test. The conservative estimates using data were available after 15 years for all patients. all 28 studies as the denominator are 46%, 50%, Performing an exercise test was associated with a 50%, 43%, and 54%, respectively. None of these are risk reduction of death of one half when adjusting different from chance. Considering probable pub- for known differences between the groups. lishing bias against negative findings, the true sit- Among patients who performed the test, most uations are likely to be closer to the conservative indicators of ischemia were without prognostic computations than to the generous ones. Other information. METs were the best predictor of exploratory subgroupings, including American future mortality. Only ST-segment depression of studies versus studies from other countries, and 2 mm or more could identify a population with an selection of the “best” studies failed to identify test increased risk of death. In the United Kingdom, performance differences. 255 consecutive patients (210 men) aged 55 years or younger (mean 48 years) admitted to hospital Under the criteria described for inclusion for for an MI (1981–1985) were eligible.81 Of these, regression analysis, 24 pairs of variables qualified 150 patients (130 men) were able to undergo an for additional analysis. Of these, the following exercise test and coronary angiography within pairs weighted regression displayed P < 0.10 and 6 months; they were followed-up for up to 15 years. P > 0.05: (1) percent mortality in those not tested Survival at a median of 16 years was 52% for the was negatively related to ST (P = 0.06) and to whole cohort, 62% for the study group, and 48% Angina (P = 0.10) risk ratios; (2) ST risk ratio was for the excluded group. From 9 years onward sur- negatively related to the percent taking Digoxin vival deteriorated significantly in the study group (P = 0.10). Additionally, ST risk ratio was nega- compared with an age matched background pop- tively related to the percent of females in the stud- ulation. Fifteen years after MI, 121 patients (81%) ies and positively related to the percent with in the study group had had at least one cardiovas- inferior-posterior MIs, both with high confidence cular event leaving 29 (19%) event-free. The levels: P = 0.03 and 0.01, respectively. Last, the number of diseased vessels was the major deter- relationship between percent tested and percent minant of time to first event and event-free sur- mortality in those tested was examined: as the vival, but exercise capacity was also important in proportion of patients tested increased, mortality the prediction of time to first event. increases in those not tested. The DUKE Meta-Analysis of Because meta-analysis tries to consider infor- Stress Testing Modalities after mation from a pool of data (but at the study level, Acute Myocardial Infarction without actually pooling data), problems arise in comparing results from studies with different Although the above meta-analysis summarizes protocols. Differences in types of exercise tests, the experience in the prethrombosis era, an excel- ECG leads used, and others increase the difficul- lent report from DUKE, summarized below, pre- ties of summarizing the research by meta-analysis, sents a meta-analysis of the exercise ECG, stress particularly because effect sizes cannot be calcu- myocardial perfusion imaging, and stress ventric- lated from the data reported in many of the stud- ular function imaging reports published from ies. Even though all of the published studies are 1980 to 1995.82 They described the predictive val- considered, there is probably a serious publishing ues of ECG, radionuclide, and echocardiographic bias both by authors and editors toward excluding markers for cardiac death or nonfatal MI. negative results. This occurs at two levels: com- pletely negative studies may not get submitted

C H A P T E R 9 Exercise Testing of Patients Recovering from Myocardial Infarction 315 Study Entry Criteria. (1) predischarge testing less of 1-mm ST-segment depression, exercise-induced than 6 weeks post MI, (2) most of patients enrolled chest pain, decrease in SBP or peak SBP to less after 1980, (3) series containing only post-MI than 120 mmHg (exertional hypotension), and less patients, (4) at least 80% complete follow-up of than 7 METs exercise capacity. For myocardial per- the patients, (5) available prevalence rates of car- fusion imaging, results included the presence of a diac death or nonfatal MI outcome data for testing reversible defect or multiple perfusion defects. results, and (6) the most current publication from For ventricular function imaging, high-risk mark- institutions with multiple reports. ers included peak EF less than 40%, change in EF less than 5%, and new ventricular wall motion Quality Assessment. Study quality was evaluated abnormalities. Outcomes included 1-year cardiac (independent of outcome assessment) according death and combined cardiac death and nonfatal MI. to criteria defined within the Congestive Heart Failure guidelines revised for use with noninvasive Sensitivity, specificity, and positive and nega- testing literature.83 Specific methodological flaws tive predictive values were calculated for each included: (1) patient selection: nonconsecutive or abnormal test criteria. A random-effects model, referral patient series, (2) study administration: which provides a more conservative range of providing a limited description of the testing pro- uncertainty about the outcome data (i.e., empiri- tocol and abnormal test/image interpretation cri- cal Bayes), was used to combine the prevalence- teria, (3) withdrawals/dropouts: no description of outcome tables.84 patient loss during follow-up or a low follow-up rate, (4) outcome measurements: use of combined Quality Assessment of the Literature. Table 9-5 “hard” and “soft” endpoints, including recurrent provides the results of quality assessment for all angina or coronary surgery or duration of less than the exercise ECG, myocardial perfusion, and ven- 2 months of follow-up, and (5) statistical analysis: tricular function-imaging reports. Only 24% of all no attempt to control for or stratify by significant reports were from prospective patient series. A confounding variables. Of the initial 115 articles limited description of the handling of withdrawals identified by literature review, the DUKE experts or a limited duration of follow-up occurred in rejected 53%. 33% of all reports. Approximately 21% of reports failed to control for or stratify by significant con- Statistical Analysis. An outcome prevalence table founding variables. was generated for each test result. Test results for exercise ECG included the dichotomous measures Baseline Characteristics. Table 9-6 provides the baseline characteristics for the exercise ECG, where TA B L E 9 – 5 . Methodologic flaws in cohort studies of risk stratification after myocardial infarction Type of study No. Prospective Patient Study Outcome Confounder studies series (%) selection administration Withdrawals measurement measurement Exercise (%) (%) (%) (%) (%) electrocardiography 28 28 Myocardial 13 10 32 53 33 perfusion imaging 8 Ventricular 00 25 14 88 25 function imaging 10 Echocardiography 20 0 0 0 40 30 Radionuclide 2 angiography 8 50 0 0 0 50 50 Pharmacologic 13 0 0 0 38 25 stress imaging 8 Echocardiography 38 25 0 50 50 0 Myocardial perfusion 4 imaging 5 25 33 0 75 50 0 TOTAL 50 40 0 20 40 0 54 24 11 9 33 56 21 Note: Some studies reported under multiple modalities. Modified from Shaw LJ, Peterson ED, Kesler K, et al: Am J Cardiol 1996;78:1327-1337.

TA B L E 9 – 6 . Clinical and study characteristics of reports included in meta-analysis 316 E X E R C I S E A N D T H E H E A R T Testing modality Mean Male (%) Beta- Prior Mi Thrombolytic Weeks Follow-up Cardiac Death (no. of studies) Total (n) Median (n) age (years) 83 blockers (%) (%) therapy (%) post-MI (years) death (%) or MI (%) 82 Exercise 15,613 173 54 29 15 64.5 2.2 1.4 3.3 8.1 electrocardiography (28) 1247 126 55 83 9 Exercise 88 31 10 2.6 2.1 4.8 13.9 myocardial 301 51 57 85 51 perfusion 1357 106 56 85 22 10 NA 1.4 6.6 15.0 scintigraphy (8) 107 44 57 23 15 8 2.8 1.9 9.3 13.2 Pharmacologic 1338 162 57 48 24 2.1 0.8 5.6 15.9 stress perfusion 52 NA 1.2 1.7 2.5 5.0 scintigraphy (5) 48 Exercise radionuclide angiography (9) Exercise echocardiography (2) Pharmacologic stress echocardiography (4) Note: One pharmacologic stress study reported echocardiographic and scintigraphic results. Modified from Shaw LJ, Peterson ED, Kesler K, et al: Am J Cardiol 1996;78:1327-1337. MI, myocardial infarction; NA, not available.

C H A P T E R 9 Exercise Testing of Patients Recovering from Myocardial Infarction 317 28 reports met the criteria providing information The pooled values for individual markers were on 15,613 tested patients. For myocardial perfusion quite low. The sensitivity of risk markers derived imaging, eight studies reported on 1247 patients. from exercise treadmill or bicycle tests ranged Ventricular function-imaging reports included nine from 23% to 56% for cardiac death. Sensitivity radionuclide angiography studies (1357 patients) values obtained from myocardial perfusion and and two echocardiographic imaging studies (107 radionuclide angiographic imaging reports were patients). A total of eight pharmacologic stress- higher (56% to 100%), but this most likely is spu- imaging reports reported on 1550 patients; 301 rious because of their smaller numbers. The posi- for perfusion and 1338 for echocardiography. The tive predictive values for cardiac death (percentage mean ages of patients were similar (54 to 57 years); of those with an abnormal test that have the out- 80% of the patients were male. Only 18 reports come) were low for most risk markers, with values included patients receiving thrombolytic therapy, of less than 10% for exercise-induced ST depres- and a quarter of all patients had a prior MI. sion, chest pain, any reversible or multiple myocar- dial perfusion defects, and the presence of new Outcomes (see Table 9-6). The pooled 1-year stress-induced wall motion abnormality. Higher cardiac death rate was 3.3% for the 28 exercise positive predictive values were noted for the com- ECG reports; the combined cardiac death and MI bined endpoint of cardiac death or MI but remained rate was 8.1%. Pooled cardiac death and combined less than 20%. The positive predictive values of a death and repeat MI rates from the eight exercise peak exercise EF less than 40% (cardiac death myocardial perfusion reports at 1 year were higher 27%, cardiac death or MI 31%) were higher than at 4.8% and 13.9%, respectively. The cardiac death those of other noninvasive predictors. In 33 patients rate was higher yet for exercise radionuclide with a new or worsening wall motion abnormality angiography (9.3%); the combined “hard” event after exercise, the positive predictive value for car- rate of death and MI was 13.2%. For the two exer- diac death or MI was 48%. In contrast to the low cise echocardiography reports, the rates of cardiac positive predictive values for most markers, nega- death and combined events at 1 year were 5.6% tive predictive values (percentage of those with a and 15.9%, respectively. The cardiac death rate negative test result that do not experience the among the eight pharmacologic stress reports outcome during follow-up) exceeded 90% in most was 2.5% for echocardiography and 6.6% for per- cases. fusion imaging, whereas cardiac death and MI rates were 5.0% and 15.0%, respectively. Summary Odds Ratio (OR) for Cardiac Death and Death or Reinfarction Table 9-7 synthesizes all the predictive values of risk markers from the 54 noninvasive reports Exercise ECG. Figures 9-2 through 9-4 provide stratified by the total number of cardiac deaths. pooled cardiac event rates and summary OR of When the number of cardiac deaths was small, cardiac death and cardiac death or nonfatal MI for predictive values for cardiac death were often the 54 reports. The summary OR for cardiac death much larger than for cohorts with more frequent was significantly higher for patients with 1-mm events. ST depression (OR 1.7, 95% confidence interval [CI] 1.2 to 2.5), impaired SBP (OR 4.0, 95% CI 2.5 Risk Indices. Table 9-8 provides a breakdown of to 6.3), or limited exercise capacity (OR 4.0, 95% various risk indices for high-risk markers obtained CI 1.9 to 8.4). A similar pattern was noted for the during exercise or pharmacologic examination. TA B L E 9 – 7 . Predictive value of noninvasive testing for cardiac death based upon total number of observed cardiac deaths Total no. Average no. Average Sensitivity Specificity Summary deaths deaths per study sample size or (95% ci) 0.63 0.77 0–5 (21 studies) 2 89 0.46 0.62 4.92 (1.15, 21.12) 6–10 (9 studies) 7 145 0.55 0.58 1.92 (0.85, 4.35) 11–19 (9 studies) 16 328 0.43 0.73 1.63 (0.84, 3.15) ≥20 (15 studies) 39 1840 1.52 (1.05, 3.51) Note: A positive test was identified from the most predictive risk marker from each testing technique. Modified from Shaw LJ, Peterson ED, Kesler K, et al: Am J Cardiol 1996;78:1327-1337.

318 E X E R C I S E A N D T H E H E A R T TA B L E 9 – 8 . Predischarge risk stratification with noninvasive testing Sensitivity Specificity (+) Predictive Value (–) Predictive value Cardiac Cardiac Cardiac Cardiac Cardiac Cardiac Cardiac Cardiac death death/MI death death/MI death death/MI death death/MI Exercise 0.42 0.44 0.75 0.70 0.04 0.16 0.98 0.91 Electrocardiography 0.44 0.23 0.79 0.87 0.11 0.21 0.96 0.88 0.56 0.53 0.62 0.65 0.10 0.18 0.95 0.91 ST depression 0.23 0.29 0.83 0.82 0.08 0.19 0.94 0.89 Impaired systolic BP Limited exercise duration 0.89 0.80 0.38 0.48 0.07 0.16 0.98 0.95 Exercise chest pain 0.64 0.75 0.71 0.76 0.07 0.17 0.98 0.97 Exercise Myocardial 0.56 0.71 0.46 0.49 0.10 0.19 0.90 0.91 Perfusion Imaging — 0.50 — 0.64 — 0.17 — 0.90 Reversible perfusion defect 0.63 0.60 Multiple perfusion defects 0.80 0.55 0.77 0.75 0.27 0.31 0.94 0.91 — 0.78 0.67 0.74 0.15 Pharmacologic Stress — 0.50 — 0.18 0.98 0.94 Imaging — 0.56 1.00 0.62 0.17 — 0.94 Reversible perfusion defect Multiple perfusion defects 0.67 0.55 — 0.60 — 0.14 — 0.92 0.62 0.79 0.18 Exercise 0.48 1.00 0.86 Radionuclide Angiography 0.56 0.54 0.05 0.08 0.98 0.94 Peak EF ≤40% Change in EF ≤ 5% New dyssynergy Exercise Echocardiography Change in EF ≤5% New dyssynergy Pharmacologic Stress Imaging (ECHO) New dyssynergy BP, blood pressure; EF, ejection fraction; MI, myocardial infarction. Modified from Shaw LJ, Peterson ED, Kesler K, et al: Am J Cardiol 1996;78:1327-1337. combined endpoint. Although not as predictive of summary odds of cardiac death was 3.1 (95% CI cardiac death, exercise-induced chest pain was 1.6 to 4.6) and for death or reinfarction was 3.6 better in predicting death or reinfarction (OR 2.1, (95% CI 1.2 to 12.6). 95% CI 1.4 to 3.2). For pharmacologic stress perfusion imaging, Exercise and Pharmacologic Stress Myocardial the summary OR for cardiac death with a reversible Perfusion Imaging. Among the 1247 patients who perfusion defect was only 1.2 times (95% CI 0.4 to underwent exercise myocardial perfusion imag- 3.7) higher. Patients who had a dipyridamole- ing, the occurrence of a reversible defect (either induced reversible perfusion defect had a 1.8 within or remote from the infarction site) was times (95% CI 0.8 to 4.1) higher risk of 1-year car- associated with a 1-year cardiac death rate of 7.1% diac death or MI. and a death or nonfatal MI rate of 15.8% (Fig. 9-3). Similar rates were reported for multiple perfusion Exercise and Pharmacologic Ventricular Function defects. For a reversible perfusion defect, the Imaging. Rates of cardiac death (27%) and com- bined events (31%) were highest for patients who

C H A P T E R 9 Exercise Testing of Patients Recovering from Myocardial Infarction 319 Cardiac event rates by test result Cardiac Cardiac death death or Pos Neg rate Pos Neg MI rate For cardiac death For cardiac death or MI Exercise 3.3% 7.8% electrocardiography ST depression 4.6% 2.1% 15.7% 9.9% (2,735) (9,943) (1,083) (2,358) Impaired systolic BP 4.9% 1.9% 21.4% 12.3% (1,796) (7,093) (182) (1,061) Limited exercise duration 3.4% 1.5% 17.5% 9.1% (3,019) (4,557) (634) (1,074) Exercise chest pain 4.6% 2.8% 18.9% 10.9% (864) (3,889) (360) (1,502) 0.1 1 10 100 0.1 1 10 100 Summary odds ratio Summary odds ratio (x-fold) (x-fold) ■ FIGURE 9–2 Summary odds of cardiac death and combined death or reinfarction for exercise electrocardiographic risk predictors. Cardiac death or reinfarction rates are in boldface in the table at left; abnormal test rates by test result are given, as well as the number of patients with a normal or abnormal test (in parentheses). Chi-square tests for homogeneity results were non-significant except for blood pressure predicting cardiac death or myocardial infarction. Cardiac event rates by test result Cardiac Cardiac death death or Pos Neg rate Pos Neg MI rate For cardiac death For cardiac death or MI Exercise ventricular 9.3% 13.2% function imaging Peak EF ≤ 40% 26.7% 6.1% 31.1% 9.1% (195) (509) (29) (66) Change in EF ≤ 5% 14.8% 2.1% 18.2% 6.2% (27) (47) (99) (243) New dyssynergy - - 17.1% 5.6% (82) (71) Exercise echocardiography 5.6% 15.9% Change in EF ≤ 5% -- 62.5% 7.8% (16) (51) New dyssynergy 17.6% 0.0% 48.5% 13.5% Pharmacologic (17) (23) (33) (74) stress New dyssynergy 2.5% 5.0% 5.4% 2.2% 8.4% 6.0% (597) (734) (191) (216) 0.1 1 10 100 0.1 1 10 100 Summary odds ratio Summary odds ratio (x-fold) (x-fold) ■ FIGURE 9–3 Summary odds of cardiac death and combined death or reinfarction for stress myocardial perfusion scintigraphy risk predictors. Chi-square tests for homogeneity results were non-significant.

320 E X E R C I S E A N D T H E H E A R T had a peak exercise EF less than 40% (Fig. 9-4). for cardiac death or MI in patients who had Summary odds of cardiac death were 3.2, 4.2, and a reversible perfusion defect was 24% in the non- 1.2 times for EF less than 40%, EF change less thrombolytic-treated versus 6% in thrombolytic- than 5%, and new echocardiographic wall motion treated patients. abnormality, respectively. For the same markers, summary odds of cardiac death or MI were 4.4, Noninvasive measurements taken during (or at 3.6, and 1.7 times higher. peak) stress can be divided into those estimating the degree of residual ischemia and LV reserve, Rates of cardiac events were lower (5.4% to however many reflect both. The degree and extent 8.4%) for patients with a pharmacologically induced of residual ischemia correlate with the extent of new or worsening wall motion abnormality. jeopardized myocardium. Such ischemic mark- The odds of cardiac death with pharmacologic ers include exercise-induced ST-segment depres- stress-induced new wall motion abnormality were sion, angina, and reversible perfusion defects. 2.7 times higher (95% CI 1.4 to 5.2). For cardiac In the meta-analysis, exercise test-induced chest death or MI, the 95% CI included 1.0 for the sum- pain was not associated with an increased risk of mary pharmacologic echocardiography data. death. The odds of cardiac death in patients with ST- segment depression of 1 mm were half that reported Comparative Predictive Value in the for patients with hemodynamic and exercise limita- Thrombolytic Era tions. Approximately 20% of patients undergoing exercise ECG testing had an abnormal test based The average cardiac death rates were lower in upon exercise-induced ST depression or chest pain. studies including thrombolytic-treated patients than in those that did not (4% versus 7%). Single versus Multiple Reperfusion Defects. The In Figure 9-5, the positive predictive values for car- presence of a single redistribution abnormality, diac death and cardiac death or MI are illustrated which relates to poststenotic flow and infarct for patients who had ST-segment depression, a artery patency, may be insufficient to stratify reversible perfusion defect, or a peak exercise EF patients. The decrease in specificity may relate to less than 40%. Positive predictive values were usu- a lower threshold for “abnormality”; more than ally decreased in patients receiving thrombolytic half of the patients who underwent myocardial therapy. For example, the positive predictive value perfusion imaging were considered to have had an Cardiac event rates by test result Cardiac Cardiac death death or Pos Neg rate Pos Neg MI rate For cardiac death For cardiac death or MI Exercise myocardial 4.8% 13.9% perfusion imaging 71% 1.6% 15.8% 5.1% Reversible defect (437) (255) (417) (335) Multiple defects 6.9% 1.7% 16.7% 2.0% (101) (230) (36) (99) Pharmacologic stress 6.6% 15.0% Reversible defect 10.4% 9.8% 19.5% 9.1% (89) (41) (154) (132) Multiple defects -- 16.7% 10.0% (12) (20) 0.1 1 10 100 0.1 1 10 100 Summary odds ratio Summary odds ratio (x-fold) (x-fold) ■ FIGURE 9–4 Summary odds of cardiac death and combined death or reinfarction for stress radionuclide angiographic (RNA) and echocardio- graphic risk predictors. EF, ejection fraction; see Figure 9-2 for other definitions. Chi-square tests for homogeneity results were non-significant.

C H A P T E R 9 Exercise Testing of Patients Recovering from Myocardial Infarction 321 Overall 1-year cardiac death rate 7 13 Non-thrombolytic PPV for ST depression 4 Thrombolytic PPV for reversible defect 9 27 PPV for peak EF ≤ 40% 2 3 Overall 1-year cardiac death rate 10 18 PPV for ST depression 7 24 PPV for reversible defect 8 PPV for peak EF ≤ 40% 28 6 31 Abnormal test rate 41 37 0 10 20 30 40 50 ■ FIGURE 9–5 Percent Positive predictive values (PPV) of noninvasive tests in non-thrombolytic and thrombolytic-treated patients. The PPVs for cardiac death or MI are illustrated for patients who had ST-segment depression, a reversible perfusion defect, or a peak exercise ejection fraction ≤40%. PPVs were usually decreased in patients receiving thrombolytic therapy. For example, the PPV for cardiac death or MI in patients who had a reversible perfusion defect was 24% in the non–thrombolytic-treated versus 6% in thrombolytic-treated patients. EF, ejection fraction; MI, myocardial infarction. abnormal scan. Risk stratification with pharmaco- hypotension or who could not complete the exer- logic stress perfusion imaging resulted in equally cise test. This is not surprising given that these high event rates in patients with normal and abnor- markers are due to both LV dysfunction and mal test results. The size and extent of the perfusion ischemia. defect and the number of abnormal ECG leads may be better predictors of 1-year outcome, but this Effect of Low Prevalence (e.g., “The Reperfusion information was not available. Era”). The positive predictive values of noninva- sive risk markers for cardiac death and combined Risk increases with LV dysfunction and is cardiac death or nonfatal MI are low in studies largely determined by the degree of myocardial with low mortality rates. The therapy clinicians damage/dysfunction secondary to the MI. The apply, as a result of an abnormal predischarge increase in risk of death was greater than 4 times test, should subsequently lower a patient’s post- higher in patients with exertional LV dysfunction test likelihood for events. A significant proportion (approximately 30% of patients) and for patients of acute MI survivors have single-vessel disease and, with a peak EF less than 40%, the positive predic- even with an abnormal test for ischemia, have a tive value for cardiac death was 27%. good prognosis making prediction difficult. This can be seen in cohorts where sensitivity is high For patients who underwent treadmill or bicycle but specificity is low. An example of lower positive exercise with no additional imaging agents, evi- predictive values in lower-risk groups was observed dence of an impaired SBP or exercise response was in reports of patients treated with thrombolytics. more prognostic of death than ischemic markers Predischarge testing after reperfusion therapy may of risk (ST depression and angina), even in throm- have a limited predictive value for several reasons. bolytic trials. The risk of cardiac death was four times higher in patients who had exertional

322 E X E R C I S E A N D T H E H E A R T Successful reperfusion results in less myocardial often, they were highly select and had various damage and may leave patients with nonsignifi- lengths of follow-up with small samples. Thus, the cant angiographic lesions and a negative stress variation in predictive accuracy of noninvasive test who still have an increased likelihood of rein- measures could, in part, reflect the primarily obser- farction. Additionally, patients who receive throm- vational nature of these reports. Further, many of bolytic agents have a generally lower risk than the studies contained few outcome events. Early in other patients with history of MI because they the development of new imaging agents or tech- are younger and less likely to have complicating niques, small patient series may be more likely illnesses. to be published because of excitement about the possible impact of this new modality. Substantial Choice of Predischarge Stress Test concern exists when negative trials remain unpublished.86 This problem of publication bias Although sensitivity and specificity values are not has been shown to lead to significant overestima- affected by disease prevalence, the predictive value tion of treatment effect. of the test is. Adjusting the risk by the threefold higher baseline risk for patients included in the There are certain patient subsets for whom the exercise radionuclide angiography studies lessens sensitivity and specificity of noninvasive measures the predictive accuracy of ventricular function may be affected (e.g., those receiving beta-block- abnormalities. Thus, if underlying risk was equal, ers or those with pulmonary disease, resting ST– all abnormal noninvasive risk markers would be T-wave changes, obesity, inability to exercise, or equally ineffective at predicting adverse outcome, submaximal stress). There is a potential for an although the predictive estimates would not increased accuracy of exercise-induced ST depres- decrease linearly with the underlying risk in the sion in patients with non-Q-wave MI that must be population. Although adjusted values allow com- confirmed.87 For the predischarge noninvasive test parisons among the various modalities, these dif- to improve upon initial pretest risk estimates, the ferences in baseline risk and subsequent post-test statistical and clinical incremental value of non- predictive estimates may be used to guide appro- invasive measures must be established. The positive priate referral to predischarge testing. Lower-risk predictive value of clinical history and ECG mea- patients should be referred to exercise ECG whereas sures in predicting preserved LV function has been higher-risk patients should be considered for a reported as 94%,88 which could obviate the need radionuclide angiogram. Using this rule, low-risk for echocardiography or radionuclide imaging to patients with an uncomplicated MI who exercise estimate systolic function in otherwise low-risk beyond 5 METs without ECG or hemodynamic patients. abnormalities are at low risk of a recurrent cardiac event during the ensuing year. This reassurance to The DUKE researchers observed little improve- the patient and family as well as the minimal cost ment in the quality of the data compared to our of this test may be the overriding reasons for per- similar meta-analysis published 10 years earlier.89 forming this test during the predischarge phase. The impact of methodological limitations on sub- However, the role of myocardial perfusion imaging sequent predictive accuracy is difficult to quantify is unclear because of its poor specificity, similar pre- but should prompt more rigorous, well-controlled dictive values to those of exercise ECG, and fivefold studies in the future to elucidate the relative higher cost. Perfusion imaging may have a role in impact of these tests on patient outcome. Although patients subsets for whom the ECG or exercise quality-assessment tools have been devised for use capacity may not be accurately interpreted as well with randomized trial data, these are not applica- as for those whose risk of recurrent MI may be high ble to retrospective data.90 Given all of the limita- (i.e., non-Q-wave MI). tions, this scholarly meta-analysis provides the best synopsis of the knowledge regarding nonin- When assessing the literature on the prognostic vasive stress testing for risk stratification post MI. value of noninvasive tests, several methodological considerations of note were encountered. Marx THE REPERFUSION ERA and Feinstein85 published an extensive review of the literature on prognosis following an MI. Their Contemporary management of the patient with results show that prognostic studies in this setting acute MI includes one or more of the following: frequently have methodological limitations and medical therapy, thrombolytic agents, and coronary explain the variation in predictive ability. Among revascularization. The first striking improvement the 54 reports, few were prospective series; more in survival in all subsets is with beta-blockers

C H A P T E R 9 Exercise Testing of Patients Recovering from Myocardial Infarction 323 (25% reduction in the first year post MI). The next the prognostic role of exercise testing in patients dramatic change in treatment of patients with who have received thrombolysis.93 Exercise tests acute MI was the broad use of thrombolytic therapy were performed on 6296 patients at an average of beginning in 1988. Equally important has been the 28 days after randomization for thrombolysis post widespread use of aspirin, beta-adrenergic-blocking MI. The test was not performed on 3923 patients agents, vasodilator therapy, common use of angio- (40%) because of contraindications. The test was tensin converting enzyme inhibitors, and a far positive for ischemia in 26% of the patients, neg- more aggressive use of revascularization therapy in ative in 38%, and nondiagnostic in 36%. Among patients who have clinical markers of a poor prog- the patients with an ischemic test result, 33% nosis. It is this constellation of new therapy, and had symptoms, whereas 67% had silent myocar- not solely the administration of thrombolytic ther- dial ischemia. The mortality rate was 7.1% among apy, that marks what is generally called reperfu- patients who did not have an exercise test and 1.7% sion era. This period has witnessed an impressive for those with an ischemic test, 0.9% for those who reduction in early and 1-year mortality rates for had a normal test, and 1.3% for those with nondi- patients with acute MI, which is particularly strik- agnostic tests. In an adjusted analysis, sympto- ing in patients who have received thrombolytic matic-induced ischemia, ischemia at a submaximal therapy and revascularization during hospitaliza- work load, low work capacity, and abnormal SBP tion. The ’90s brought the widespread application were independent predictors of 6-month mortal- of cardiac catheterization negating the use of the ity (relative risks of 2× for each). However, when exercise test to select patients for this procedure, these variables were considered simultaneously, as had been the situation earlier. Currently, the only symptomatic-induced ischemia and low work evidence supports the use of angiography when capacity were confirmed as independent predictors possible instead of thrombolysis and even “facili- of mortality (Cox hazard ratio of 2 and 1.8, respec- tated” PCI, where thrombolysis is only used to hold tively). The GISSI investigators concluded that the patient until angiography is possible. In any cir- patients with a normal exercise response have cumstance, once the coronary arteries are visual- an excellent medium-term prognosis and do not ized it is hard for the angiographer not to open need further investigation as shown by others.94 closed arteries with drug-eluting stents. CABS only However, evaluation must be directed to the remains for patients with difficult lesions. Most patients who cannot undergo exercise testing patients recover from their MI with minimal loss because the mortality was five to seven times of myocardium and reperfused myocardium as greater in that group. The GISSI-2 researchers 95 well. Thus, the exercise test plays a different role calculated the Duke treadmill score (DTS) and the than it did in the past. Veterans Affairs Medical Center Score (VAMCS) for each patient and used coefficients of a multivariate Shorter hospital stays, widespread use of throm- analysis to develop a simple predictive scoring sys- bolytic agents, greater uses of revascularization tem. Six-month mortality rates in the subgroups strategies, implantable cardiac defibrillators, and of each scoring system were as follows: DTS: low increased use of beta-adrenergic-blocking agents risk 0.6%, moderate risk 1.8%, high risk 3.4%; and angiotensin converting enzyme inhibitors or VAMCS: low risk 0.6%, moderate risk 2%, high angiogenesis receptor blockers continue to change risk 5%; GISSI-2 Index: low risk 0.5%, moderate the clinical presentation of the patient with history risk 2%, high risk 6%. The results of multivariate of MI. Not all patients have received each of these analysis were as follows: DTS: moderate risk 2.5×, various therapies; hence, survivors of MI are quite high risk 5×; VAMCS: moderate risk 3×, high risk heterogeneous. The CAMI study reported that 6×; GISSI-2 Index: moderate risk 3×, high risk 9×. among 3178 consecutive patients with acute MI, 45% received thrombolytic agents, 20% under- The prognosis among survivors of MI contin- went PCI, and 8% had CABS.91 Medications at the ues to improve as newer treatment strategies are time of hospital discharge included beta-blockers applied. The 1-year postdischarge mortality in the in 61%, angiotensin converting enzyme inhibitors CAMI study was 8.4% and was distinctly lower in in 24%, and aspirin in 86%. the 45% of patients who received thrombolytic therapy (4% mortality) and in the 28% who under- Although exercise testing was helpful in went coronary angioplasty (3% mortality) or CABS the management of patients with a history of (3.7% mortality).96 Data from the GUSTO trial97 MI in the prethrombolytic era, the impact over demonstrate that 57% of the 41,021 patients who the past decade of thrombolytic therapies could received thrombolytic therapy were uncompli- have decreased the value of exercise testing.92 cated (no recurrent ischemia, reinfarction, heart The GISSI-2 database has enabled reevaluation of

324 E X E R C I S E A N D T H E H E A R T failure, stroke, or invasive procedures) at 4 days demanding occupation, simulated work tests can after MI. The mortality rate at 1 month was 1% and be performed.102,103 at 1 year was 3.6%. Recurrent ischemia occurred in 7% of this group. These and other data from large Exercise testing in cardiac rehabilitation is thrombolytic trials98,99 demonstrate that those essential in the development of the exercise pre- patients unable to perform an exercise test have scription to establish a safe and effective training the highest adverse cardiac event rate, whereas intensity, in risk stratification of patients to deter- uncomplicated stable patients have a low cardiac mine the level of supervision and monitoring event rate even before undergoing further risk required during exercise training sessions, and assessment by exercise testing. in evaluation of training program outcome.104 For these reasons, symptom-limited exercise test- The two meta-analyses summarized earlier of ing before program initiation is needed for all 30 studies, including more than 20,000 patients, patients in whom cardiac rehabilitation is recom- found that exercise incapacity and abnormal SBP mended (recent MI, recent CABS, recent coronary response were more predictive of adverse cardiac angioplasty, chronic stable angina, and controlled events after MI than measures of exercise-induced heart failure).105 Although there are no available ischemia. Although most of the studies included studies to assess its value, it is the consensus of this were performed before the reperfusion era, similar committee based on practical experience that exer- results were found in the GISSI report that con- cise testing in the stable cardiac patient who contin- sidered 6000 patients who received thrombolysis. ues an exercise training program be performed after the initial 8 to 12 weeks of exercise training ACTIVITY COUNSELING and at least yearly thereafter, or sooner as needed depending on changes in symptoms or medica- Exercise testing after MI is useful in counseling tions that may affect the exercise prescription. patients and their families regarding domestic, Such testing may be useful to rewrite the exercise recreational, and occupational activities that can be prescription, evaluate improvement in exercise safely performed after hospital discharge. Exercise capacity, and provide feedback to the patient. capacity in METs derived from the exercise test can be applied to estimate an individual’s toler- SUMMARY ance for specific activities. Published charts that estimate energy requirements of various activities The benefits of performing an exercise test in are available100 but should be used only as a guide, patients with history of MI are listed in Table 9-9. realizing that the intensity at which activities per- Submitting patients to exercise testing can expedite formed directly influence the amount of energy and optimize their discharge from the hospital. required. Most domestic chores and activities require less than 5 METs, hence a submaximal test TA B L E 9 – 9 . Benefits of exercise testing post at the time of hospital discharge can be useful in myocardial infarction counseling with regard to the first several weeks after an MI. PREDISCHARGE SUBMAXIMAL TEST Setting safe exercise levels (exercise prescription) The follow-up symptom-limited testing per- Optimizing discharge formed at 3 to 6 weeks after MI can assist in further Altering medical therapy activity prescription and issues regarding return to Triaging for intensity of follow-up work. Most occupational activities require less than First step in rehabilitation—assurance, encouragement 5 METs. In the 15% of individuals in the work force Reassuring spouse whose work involves heavy manual labor,101 the Recognizing exercise-induced ischemia and dysrhythmias exercise test data should not be used as the sole criterion for recommendations regarding return MAXIMAL TEST FOR RETURN TO NORMAL ACTIVITIES to work. Energy demands of lifting heavy objects, Determining limitations temperature, environmental, and psychological Prognostication stresses are not assessed by routine exercise tests Reassuring employers and must be taken into consideration. In patients Determining level of disability with low exercise capacity, LV dysfunction, exercise- Triaging for invasive studies induced ischemia, and in those who are other- Deciding upon medications wise apprehensive about returning to a physically Exercise prescription Continued rehabilitation

C H A P T E R 9 Exercise Testing of Patients Recovering from Myocardial Infarction 325 The patients response to exercise, their work capac- test results. The qualifying reports (n = 54) included ity, and limiting factors at the time of discharge can 19,874 patients and three quarters were retro- be assessed by the exercise test. An exercise test spective (76%) and a third were small samples before discharge is important for giving patient with less than five deaths. One-year mortality in the guidelines for exercise at home, reassuring them studies ranged from 2.5% for pharmacologic stress of their physical status, and determining the risk of echocardiography to 9.3% for exercise radionuclide complications. It provides a safe basis for advising angiography studies, consistent with population the patient to resume or increase his or her activity differences. Positive predictive values (the percent- level and return to work. The test can demonstrate age of those with an abnormal test that have the to the patient, relatives, or employer the effect of outcome during follow-up) for most noninvasive the MI on the capacity for physical performance. risk markers were less than 10% for cardiac death Psychologically, it can cause an improvement in and less than 20% for death or reinfarction. ECG, the patient’s self-confidence by making the patient symptomatic, and scintigraphic markers of ischemia less anxious about daily physical activities. The test (ST-segment depression, angina, and a reversible has been helpful in reassuring spouses of patients defect) were less sensitive (average about 44%) who had an MI of their physical capabilities. The for identifying morbid and fatal outcomes than psychological impact of performing well on the markers of both LV dysfunction and ischemia exercise test is impressive. Many patients increase (exercise duration, exertional hypotension, and their activity and actually rehabilitate them- peak LVEF). The positive predictive value of pre- selves after being encouraged and reassured by discharge noninvasive testing is low. Markers of their response to this test. LV dysfunction or both dysfunction and ischemia were better predictors than markers of myocardial Exercise testing is useful in activity counseling ischemia alone. after hospital discharge. It is also an important tool in exercise training as part of comprehensive car- The two meta-analyses summarized of 30 studies, diac rehabilitation, where it can be used to develop including more than 20,000 patients, found that and modify the exercise prescription, assist in pro- exercise incapacity and abnormal SBP response viding activity counseling, and assess the patient’s were more predictive of adverse cardiac events response into, and progress in, the exercise training after MI than measures of exercise-induced program. ischemia. Although most of the studies included were performed before the reperfusion era, similar One consistent finding in the review of the results were found in the GISSI report that con- exercise test studies following an MI that included sidered 6000 patients who received thrombolysis. a follow-up for cardiac endpoints is that patients who met whatever criteria set forth for exercise The evaluation of the patient with an MI has testing were at lower risk than patients not tested. dramatically changed with the issue of who needs This finding supports the clinical judgment of the cardiac catheterization being resolved: cardiac skilled clinician. In the complete data set from the catheterization is the preferred treatment before review, only an abnormal SBP response or a low and possibly after thrombolysis (facilitated PCI). exercise capacity was significantly associated with Patients with LV dysfunction post MI can expect a a poor outcome. These responses are so powerful 30% reduction in mortality with an implantable because they can be associated with either ischemic defibrillator. The clinical value of exercise testing events or CHF events (see Chapter 8, Prognostic post MI most likely will be resolved by studies like Applications of Exercise Testing). the ROSETTA106 and PERISCOP107 study which have evaluated the value of functional testing The DUKE meta-analysis compared the available after interventions. However, the exercise test noninvasive tests results to outcomes in patients remains helpful to estimate prognosis in the recovering from acute-MI. Studies published from post-MI patient. 1980 to 1995 had to fulfill these criteria: only MI patients, most patients enrolled after 1980, tested REFERENCES within 6 weeks of MI, follow-up rates greater than 80%, and having outcome prevalence rates for 1. Hunink MG, Goldman L, Tosteson AN, et al: The recent decline in test results, and only the latest results if there mortality from coronary heart disease, 1980-1990. The effect of were multiple reports from the same institution. secular trends in risk factors and treatment. JAMA 1997;277: Sensitivity, specificity, and predictive values were 535-542. calculated for test results for 1-year outcomes (cardiac death, cardiac death, or reinfarction). 2. Braunwald E, Antman EM, Beasley JW, et al: Committee on Univariable and summary OR were calculated for the Management of Patients With Unstable Angina. ACC/AHA 2002

326 E X E R C I S E A N D T H E H E A R T guideline update for the management of patients with unstable 24. Starling MR, Crawford MH, O’Rourke RA: Superiority of selected angina and non-ST-segment elevation myocardial infarction— treadmill exercise protocols predischarge and six weeks postinfarc- summary article: A report of the American College of Cardiology/ tion for detecting ischemic abnormalities. Am Heart J 1982;104: American Heart Association task force on practice guidelines 1054-1059. (Committee on the Management of Patients With Unstable Angina). J Am Coll Cardiol 2002;40:1366-1374. 25. Starling MR, Crawford MH, Kennedy GT, O’Rourke RA: Exercise 3. Dargie H. Myocardial infarction: Redefined or reinvented? Heart testing early after myocardial infarction: Predictive value of subse- 2002;88:1-3. quent unstable angina and death. Am J Cardiol 1980;46:909-914. 4. Dalby M, Bouzamondo A, Lechat P, Montalescot G: Transfer for pri- mary angioplasty versus immediate thrombolysis in acute myocardial 26. Handler CE, Sowton E: Diurnal variation in symptom-limited exer- infarction: A meta-analysis. Circulation 2003;108:1809-1814. Epub cise test responses six weeks after myocardial infarction. Eur Heart 2003 Oct 6. J 1985;6:444-450. 5. Arciero TJ, Jacobsen SJ, Reeder GS, et al: Temporal trends in the incidence of coronary disease. Am J Med 2004;117:228-233. 27. Starling MR, Crawford MH, Kennedy GT, O’Rourke RA: Treadmill 6. Goldman L, Phillips KA, Coxson P, et al: The effect of risk factor exercise tests predischarge and six weeks post-myocardial infarction reductions between 1981 and 1990 on coronary heart disease inci- to detect abnormalities of known prognostic value. Ann Intern Med dence, prevalence, mortality and cost. J Am Coll Cardiol 2001;38: 1981;94:721-727. 1012-1017. 7. Guidelines for risk stratification after myocardial infarction. 28. Wohl AJ, Lewis HR, Campbell W, et al: Cardiovascular function dur- American College of Physicians. Ann Intern Med 1997;126:556-560. ing early recovery from acute myocardial infarction. Circulation 8. Cucherat M, Bonnefoy E, Tremeau G: Primary angioplasty versus 1977;56:931-937. intravenous thrombolysis for acute myocardial infarction. Cochrane Database Syst Rev 2003(3):CD001560. 29. Haskell WL, Savin W, Oldridge N, DeBusk R: Factors influencing 9. Heggunje PS, Harjai KJ, Stone GW, et al: Procedural success ver- estimated oxygen uptake during exercise testing soon after sus clinical risk status in determining discharge of patients after myocardial infarction. Am J Cardiol 1982;50:299-304. primary angioplasty for acute myocardial infarction. J Am Coll Cardiol 2004;44:1400-1407. 30. Castellanet MJ, Greenberg PS, Ellestad MH: Comparison of S-T seg- 10. Gibbons RJ, Balady GJ, Beasley JW, et al: ACC/AHA Guidelines for ment changes on exercise testing with angiographic findings in Exercise Testing. A report of the American College of Cardiology/ patients with prior myocardial infarction. Am J Cardiol 1978;42:29-35. American Heart Association Task Force on Practice Guidelines (Committee on Exercise Testing). J Am Coll Cardiol 1997;30: 31. Ahnve S, Savvides M, Abouantoun S, et al: Can ischemia be recog- 260-311. nized when Q waves are present on the resting electrocardiogram? 11. Antman EM, Anbe DT, Armstrong PW, et al: ACC/AHA guidelines Am Heart J 1986;110:1016-1020. for the management of patients with ST-elevation myocardial infarction—executive summary. A report of the American College 32. Miranda C, Herbert W, Dubach P, et al: Post MI exercise testing: of Cardiology/American Heart Association Task Force on Practice Non Q wave vs Q wave. Circulation 1991;84:2357-2365. Guidelines (Writing Committee to revise the 1999 guidelines for the management of patients with acute myocardial infarction). 33. Weiner DA: Prognostic value of exercise testing early after myocar- J Am Coll Cardiol 2004;44:671-719. dial infarction. J Cardiac Rehabil 1983;3:114-122. 12. Juneau M, Colle SP, Theroux P, et al: Symptom-limited versus low level exercise testing before hospital discharge after myocardial 34. Paine TD, Dye LE, Roitman DI, et al: Relation of graded exercise infarction. J Am Coll Cardiol 1992;20:927-933. test findings after myocardial infarction to extent of coronary 13. Hamm LF, Crow RS, Stull A, Hannan P: Safety and characteristics artery disease and left ventricular dysfunction. Am J Cardiol of exercise testing early after myocardial infarction. Am J Cardiol 1978;42:716-723. 1989;63:1193-1197. 14. Jain A, Myers GH, Sapin PM, O’Rourke RA: Comparison of symp- 35. Dillahunt PH, Miller AB: Early treadmill testing after myocardial tom-limited and low level exercise tolerance tests early after infarction. Chest 1979;76:150-155. myocardial infarction. J Am Coll Cardiol 1993;22:1816-1820. 15. Jespersen CM, Hagerup L, Hollander N, et al: Exercise-provoked ST 36. Sammel NL, Wilson RL, Norris RM, et al: Angiocardiography and segment depression and prognosis in patients recovering from exercise testing at one month after a first myocardial infarction. acute myocardial infarction. Significance and pitfalls. J Intern Med Aust NZ J Med 1980;10:182-187. 1993;233:27-32. 16. Torkelson LO: Rehabilitation of the patient with acute myocardial 37. Fuller CM, Raizner AE, Verani MS, et al: Early post-myocardial infarction. J Chronic Dis 1964;17:685-704. infarction treadmill stress testing. An accurate predictor of multi- 17. Ibsen H, Kjoller E, Styperek J, Pedersen A: Routine exercise ECG vessel coronary disease and subsequent cardiac events. Ann Intern three weeks after acute myocardial infarction. Acta Med Scand Med 1981;94:734-739. 1975;198:463-469. 18. Niederberger M: Values and limitations of exercise testing after 38. Boschat J, Rigaud M, Bardet J, et al: Treadmill exercise testing and myocardial infarction (monograph). Wien, Verlag Bruder Hollinek, coronary cineangiography following first myocardial infarction. J 1977; pp 3-45. Cardiac Rehab 1981;1:206-211. 19. Markiewicz W, Houston N, DeBusk RF: Exercise testing soon after myocardial infarction. Circulation 1977;56:26-31. 39. Schwartz KM, Turner JD, Sheffield LT, et al: Limited exercise testing 20. Sivarajan ES, Bruce RA, Lindskog BD, et al: Treadmill test soon after myocardial infarction. Correlation with early coronary responses to an early exercise program after myocardial infarction: and left ventricular angiography. Ann Intern Med 1981;94:727-734. A randomized study. Circulation 1982;65:1420-1428. 21. Taylor CB, Bandura A, Ewart CK, et al: Exercise testing to enhance 40. Starling MR, Crawford MH, Richards KL, O’Rourke RA. Predictive wives’ confidence in their husbands’ cardiac capability soon after value of early postmyocardial infarction modified treadmill exer- clinically uncomplicated acute myocardial infarction. Am J Cardiol cise testing in multivessel coronary artery disease detection. Am 1985;55:635-638. Heart J 1981;102:169-175. 22. Ewart CK, Taylor CB, Reese LB, DeBusk RF: Effects of early post- myocardial infarction exercise testing on self-perception and sub- 41. De Feyter PJ, van den Brand M, Serruys PW, Wijns W: Early angiog- sequent physical activity. Am J Cardiol 1983;51:1076-1080. raphy after myocardial infarction: What have we learned? Am Heart 23. Handler CE, Sowton E: A comparison of the Naughton and modi- J 1985;109:194-199. fied Bruce treadmill exercise protocols in their ability to detect ischaemic abnormalities six weeks after myocardial infarction. 42. Ericsson M, Granath A, Ohlsen P, et al: Arrhythmias and symptoms Eur Heart J 1984;5:752-755. during treadmill testing three weeks after myocardial infarction in 100 patients. Br Heart J 1973;35:787-790. 43. Kentala E: Physical fitness and feasibility of physical rehabilitation after myocardial infarction in men of working age. Ann Clin Res 1972;4(suppl9):1-84. 44. Granath A, Sodermark T, Winge T, et al: Early work load tests for evaluation of long-term prognosis of acute myocardial infarction. Br Heart J 1977;39:758-763. 45. Smith JW, Dennis CA, Gassmann A, et al: Exercise testing three weeks after myocardial infarction. Chest 1979;75:12-16. 46. Hunt D, Hamer A, Duffield A, et al: Predictors of reinfarction and sudden death in a high-risk group of acute myocardial infarction survivors. Lancet 1979;1:233-236. 47. Srinivasan M, Young A, Baker G, et al: The value of postcardiac infarction exercise stress testing. Identification of a group at high risk. Med J Aust 1981;2:466-467. 48. Jelinek VM, Ziffer RW, McDonald IG, et al: Early exercise testing and mobilization after myocardial infarction. Med J Aust 1977; 2:589-593.

C H A P T E R 9 Exercise Testing of Patients Recovering from Myocardial Infarction 327 49. Sami M, Kraemer H, DeBusk RF: The prognostic significance of 73. Krone R, Dwyer E, Greenberg H, et al: Risk stratification in serial exercise testing after myocardial infarction. Circulation patients with first non-Q wave infarction: Limited value of the early 1979;60:1238-1246. low level exercise test after uncomplicated infarcts. J Am Coll Cardiol 1989;14;31-37. 50. Davidson DM, DeBusk RF: Prognostic value of a single exercise test 3 weeks after uncomplicated myocardial infarction. Circulation 74. Ades PA, Thomas JD, Hanson JS, et al: Effect of metoprolol on the 1980;61:236-241. submaximal stress test performed early after acute myocardial infarction. Am J Cardiol 1987;60:963-966. 51. DeBusk RF, Dennis CA: “Submaximal” predischarge exercise test- ing after acute myocardial infarction: Who needs it? Am J Cardiol 75. Curtis Jl, Houghton JL, Patterson JH, et al: Propranolol therapy 1985;55:499-500. alters estimation of potential cardiovascular risk derived from sub- maximal postinfarction exercise testing. Am Heart J 1991;121: 52. Theroux P, Marpole DGF, Bourassa MG: Exercise stress testing 1655-1664. in the post-myocardial infarction patient. Am J Cardiol 1983;52: 664-667. 76. Krone RJ, Miller JP, Gillespie JA, et al: Usefulness of low level exercise testing early after acute myocardial infarction in patients taking 53. Waters DA, Bosch X, Bouchard A, et al: Comparison of clinical vari- beta blocking agents. Am J Cardiol 1987;60:23-27. ables and variables derived from a limited predischarge exercise test as predictors of early and late mortality after myocardial infarc- 77. Ronnevik PK, VonderLippe G: Prognostic importance of predis- tion. J Am Coll Cardiol 1985;5:1-8. charge exercise capacity for long term mortality and nonfatal myocar- dial infarction in patients admitted for suspected acute myocardial 54. Koppes GM, Kruyer W, Beckmann CH, Jones FG: Response to infarction and treated with Metoprolol. Eur Heart J 1992;13: exercise early after uncomplicated acute myocardial infarction in 1468-1472. patients receiving no medication: Long-term follow-up. Am J Cardiol 1980;46:764-769. 78. Murray DP, Tan LB, Salih M, et al: Does beta adrenergic blockade influence the prognostic implications of post-myocardial infarction 55. Starling MR, Kennedy GT, Crawford MH, O’Rourke RA: exercise testing? Br Heart J 1988;60:474-479. Comparative predictive value of ST-segment depression or angina during early and repeat postinfarction exercise tests. Chest 79. Madsen EB, Gilpin E, Henning H: Short-term prognosis in acute 1984;86:845-849. myocardial infarction: Evaluation of different prediction methods. Am Heart J 1984;107:1241-1251. 56. Saunamaki KI, Andersen JD: Early exercise test in the assessment of long-term prognosis after acute myocardial infarction. Acta Med 80. Gilpin E, Olshen R, Henning H, Ross J: Risk prediction after Scand 1981;209:185-191. myocardial infarction. Comparison of three multivariate method- ologies. Cardiology 1983;70:73-84. 57. Saunamaki KI, Andersen JD: Early exercise test vs clinical variables in the long-term prognostic management after myocardial infarc- 81. Dominguez H, Torp-Pedersen C, Koeber L, Rask-Madsen C: tion. Acta Med Scand 1982;212:47-52. Prognostic value of exercise testing in a cohort of patients followed for 15 years after acute myocardial infarction. Eur Heart J. 2001;22: 58. Madsen EB, Gilpin E: Prognostic value of exercise test variables 300-306. after myocardial infarction. J Cardiac Rehabil 1983;3:481-488. 82. Awad-Elkarim AA, Bagger JP, Albers CJ, et al: A prospective study of 59. Jespersen CM, Kassis E, Edeling CJ, Madsen JK: The prognostic long term prognosis in young myocardial infarction survivors: The value of maximal exercise testing soon after first MI. Eur Heart J prognostic value of angiography and exercise testing. Heart 1985;6:769-772. 2003;89:843-847. 60. Velasco J, Tormo V, Ferrer LM, et al: Early exercise test for evalua- 83. Shaw LJ, Peterson ED, Kesler K, et al: A meta-analysis of predis- tion of long-term prognosis after uncomplicated myocardial infarc- charge risk stratification after acute myocardial infarction with tion. Eur Heart J 1981;2:401-407. stress electrocardiographic, myocardial perfusion, and ventricular function imaging. Am J Cardiol 1996;78:1327-1337. 61. Weld FM: Exercise testing after myocardial infarction. J Cardiac Rehabil 1985;5:20-27. 84. Baker DW, Jones R, Hodges J, et al: Management of heart failure. III. The role of revascularization in the treatment of patients with 62. Fioretti P, Deckers JW, Brower RW, et al: Predischarge stress test moderate or severe left ventricular systolic dysfunction. JAMA after myocardial infarction in the old age: Results and prognostic 1994;272:1528-1534. value. Eur Heart J 1984;5:101-104. 85. Eddy DM, Hasselblad C, Shachter R: An introduction to a Bayesian 63. Norris RM, Barnaby PF, Brandt PWT, et al: Prognosis after recovery method for meta-analysis: The confidence profile method. Med from first acute myocardial infarction: Determinants of reinfarc- Decis Making 1990;10:15-23. tion and sudden death. Am J Cardiol 1984;53:408-413. 86. Marx BE, Feinstein AR: Methodologic sources of inconsistent prog- 64. Jennings K, Reid DS, Hawkins T, Julian DJ: Role of exercise testing noses for post-acute myocardial infarction. Am J Med 1995;98:537- early after myocardial infarction in identifying candidates for coro- 550. nary surgery. Br Med J 1984;288:185-187. 87. Simes RJ: Confronting publication bias: a cohort design for meta- 65. Handler CE: Exercise testing to identify high risk patients analysis. Stat Med 1987;6:11-29. after myocardial infarction. J R Coll Physicians Lond 1984;18: 124-127. 88. Silver MT, Rose GA, Paul SD, et al: A clinical rule to predict preserved left ventricular ejection fraction in patients after myocar- 66. Krone RJ, Gillespie JA, Weld FM, et al: Low-level exercise testing dial infarction. Ann Intern Med 1994;121:750-756. after myocardial infarction: Usefulness in enhancing clinical risk stratification. Circulation 1985;71:80-89. 89. Froelicher VF, Perdue S, Pewen W, Risch M: Application of meta- analysis using an electronic spreadsheet to exercise testing in patients 67. Dwyer EM, McMaster P, Greenberg H: Nonfatal cardiac events and with myocardial infarction. Am J Med 1987;83:1045-1054. recurrent infarction in the year after acute myocardial infarction. J Am Coll Cardiol 1984;4:695-702. 90. Chalmers TC: Problems induced by meta-analyses. Stat Med 1991; 10:971-979. 68. Williams WL, Nair RC, Higginson LA, et al: Comparison of clinical and treadmill variables for the prediction of outcome after myocar- 91. Rouleau JL, Talajic M, Sussex B, et al: Myocardial infarction dial infarction. J Am Coll Cardiol 1984;4:477-486. patients in the 1990’s - Their risk factors, stratification and survival in Canada: The Canadian Assessment of Myocardial Infarction 69. Madsen EB, Gilpin E. How much prognostic information do exer- (CAMI) study. J Am Coll Cardiol 1996;27:1119-1127. cise test data add to clinical data after acute myocardial infarction. Int J Cardiol 1983;4:15-27. 92. Stevenson R, Muachandran V, Ranjadayalan K, et al: Reassessment of treadmill stress testing for risk stratification in patients with 70. Sullivan ID, Davies DW, Sowton E: Submaximal exercise testing acute myocardial infarction treated by thrombolysis. Br Heart J early after myocardial infarction: Difficulty of predicting coronary 1993;70:415-420. anatomy and left ventricular performance. Br Heart J 1985;53: 180-185. 93. Villella A, Maggioni AP, Villella M, et al: Prognostic significance of maximal exercise testing after myocardial infarction treated with 71. Connolly DC, Elveback LR: Coronary heart disease in residents of thrombolytic agents: The GISSI-2 data-base. Lancet 1995 Aug Rochester, Minnesota. VI. Hospital and posthospital course of 26;346(8974):523-529. patients with transmural and subendocardial myocardial infarction. Mayo Clin Proc 1985;60:375-381. 94. Piccalo G, Pirelli S, Massa D, et al: Value of negative predischarge exercise testing in identifying patients at low risk after acute 72. Klein J, Froelicher V, Detrano R, et al: Does the resting electrocar- myocardial infarction treated by systemic thrombolysis. Am J diogram after myocardial infarction determine the predictive value Cardiol 1992;70:31-33. of exercise-induced ST depression? A two year follow-up in a vet- eran population. J Am Coll Cardiol, 1989;14:305-311.

328 E X E R C I S E A N D T H E H E A R T 95. Villella M, Villella A, Santoro L, et al: Ergometric score systems 101. Wilke NA, Sheldahl LM, Dougherty SM, et al: Baltimore after myocardial infarction: Prognostic performance of the DUKE Therapeutic Equipment work simulator: Energy expenditure of Treadmill Score, Veterans Administration Medical Center Score, work activities in cardiac patients. Arch Phys Med Rehabil and of a novel score system, GISSI-2 Index, in a cohort of sur- 1993;74:419-424. vivors of acute myocardial infarction. Am Heart J. 2003 Mar; 145:475-483. 102. Sheldahl LM, Wilke NA, Tristani FE: Exercise prescription for return to work. J Cardiopulm Rehabil 1985;5:567-575. 96. Newby LK, Califf RM, Guerci A, et al: Early discharge in the thrombolytic era: an analysis of criteria for uncomplicated infarc- 103. Balady GJ, Fletcher BJ, Froelicher ES, et al: American Heart tions from the GUSTO trial. J Am Coll Cardiol 1996;27:625-632. Association Scientific Statement Cardiac Rehabilitation Programs. Circulation 1994;90:1602-1610. 97. Chaitman BR, McMahon RP, Tarrin M, et al: Impact of treatment strategy on predischarge exercise tests in the Thrombolysis in 104. American College of Sports Medicine. Guidelines for exercise test- Myocardial Infarction (TIMI) II trial. Am J Cardiol 1993:71:131-138. ing and prescription. Philadelphia, Williams & Wilkins, 1995. 98. Volpi A, DeVita C, Franzosi MG, et al: Predictors of nonfatal rein- 105. Mak KH, Eisenberg MJ, Tsang J, et al: Clinical impact of func- farction in survivals of myocardial infarction after thrombolysis. tional testing strategy among stented and non-stented patients: Results of the GISSI-2 database. J Am Coll Cardiol 1994;24:608-615. Insights from the ROSETTA Registry. Int J Cardiol 2004;95: 321-327. 99. Fletcher GF, Balady GJ, Froelicher VF, et al: American Heart Association Exercise Standards. Circulation 1995;91:580-615. 106. Sellier P, Chatellier G, D’Agrosa-Boiteux MC, et al: Use of non- invasive cardiac investigations to predict clinical endpoints after 100. U.S. Department of Health and Human Services. Clinical practice coronary bypass graft surgery in coronary artery disease patients: guideline #17: Cardiac rehabilitation. AHCPR Publication No. 96- Results from the prognosis and evaluation of risk in the coronary 0672. 1995. operated patient (PERISCOP) study. Eur Heart J 2003;24:916-926.

CHAPTER ten Exercise Testing in Patients with Heart Failure and Left Ventricular Dysfunction PATHOPHYSIOLOGY in both conditions is the same. This is fortunate since they can be difficult to distinguish clinically Myocardial damage or dysfunction is the pathophys- without echocardiography. However, it appears iologic basis of heart muscle disease. Myocardial that treatment and prognosis with diastolic dys- damage or dysfunction can be divided into sys- function is more related to the conditions under- tolic and diastolic dysfunction. Systolic function lying it and the treatment of these conditions,2 relates to the emptying characteristics of the left while the treatment of systolic dysfunction has been ventricle, and diastolic function relates to its filling clarified by numerous randomized trials. An issue properties. Systolic dysfunction due to myocar- requiring further clarification is whether ischemic dial damage is most common in clinical practice systolic dysfunction can be improved by revascu- and usually leads to left ventricular dilation. The larization. Several randomized trials are in progress ventricle dilates as a compensatory mechanism to comparing percutaneous coronary intervention take advantage of the Frank-Starling relationship and coronary artery bypass grafting versus medical (i.e., increased contractility with stretching of the management in such patients. sarcomeres), which can eventually worsen ventric- ular performance over time. Anything that causes Definition of Heart Failure ventricular damage or scarring (e.g., muscle loss) usually leads to systolic dysfunction. Congestive heart failure can be defined as a syn- drome consisting of: Approximately 70% of patients with the syn- drome of chronic heart failure (HF) have systolic • Signs and symptoms of intravascular and inter- dysfunction, while the remainder has diastolic stitial volume overload (hypervolemia), includ- dysfunction. In patients with the latter, systolic ing shortness of breath, rales, hepatomegaly and function and ejection fraction (EF) can be nor- edema mal, but filling pressure is usually elevated due to a stiff, noncompliant ventricle.1 Usually, diastolic • Manifestations of inadequate tissue perfusion, dysfunction is secondary to hypertension, patho- such as fatigue and poor exercise tolerance logical hypertrophy, infiltrative diseases of the myocardium and, at times, ischemia. All patients Chronic heart failure can be defined as the with systolic dysfunction have some degree of dias- same syndrome that is either well compensated or tolic dysfunction, and when systolic dysfunction is appropriately treated so that the manifestations of compensated, diastolic dysfunction often remains. acute hypervolemia are minimized. Currently, the treatment for acute congestive HF 329

330 E X E R C I S E A N D T H E H E A R T Key Points less than 40% are living 4 years after diagnosis. We will address the issue of whether exercise test- • HF is the major manifestation of left ventricu- ing can improve risk stratification beyond clinical lar damage caused by systolic dysfunction and variables.5,6 a dilated cardiomyopathy. Patients with systolic dysfunction usually have diastolic dysfunction Clinical Risk Markers and the latter often remains after the systolic component is compensated. Despite important advances in therapy for patients with chronic HF, the mortality rate for this • Left-sided failure can lead to right-sided failure. condition remains high and continues to be one of the important challenges facing the clinician who • Diastolic dysfunction can exist independently manages these patients. Cardiac transplantation and is frequently associated with a stiff, hyper- has evolved into an important treatment option for trophied (but normal-sized) ventricle caused patients with severe HF, but this option remains by chronic high blood pressure and/or congen- limited to a relatively small number of patients with ital abnormalities. end-stage disease because there continues to be a severe shortage of donor hearts. The high mortality • Abnormalities in the periphery (anemia, rate and widening gap between patients listed for beriberi heart disease, A-V fistulas, thyrotoxi- transplantation and available donor hearts have cosis) can cause high-output HF. magnified the need for reliable prognostic markers in HF. In addition, revascularization techniques for PREVALENCE AND PROGNOSIS IN ischemic cardiomyopathies carry a risk that must HEART FAILURE be balanced against the benefits. HF (when due to dilated cardiomyopathy) has a To direct the limited number of donor hearts 15% to 25% annual cardiac mortality. Analysis of to patients who need them the most, a great deal 34 years of follow-up of Framingham Study data of effort has been directed toward stratifying risk provides clinically relevant insights into the preva- among patients with severe HF through the use of lence, incidence, secular trends, prognosis, and clinical, hemodynamic, and exercise test data. modifiable risk factors for the occurrence of HF in Consensus statements from the American Heart a general population sample.3 HF was found in Association and American College of Cardiology7 about 1% of persons in their fifties and 10% of per- and a Bethesda Conference position statement8 sons in their eighties. The annual incidence also have helped establish guidelines for selection crite- increased with age, from about 0.2% in persons ria among patients considered for transplantation. 45 to 54 years to 4.0% in men aged 85 to 94 years, The major risk markers in HF include New York with the incidence approximately doubling with Heart Association functional class, reduced EF, each decade of age. Women had a lower incidence reduced cardiac index, renal insufficiency (creati- at all ages. Male predominance was due to coronary nine clearance <60 mL/min), persistent signs of heart disease, which conferred a fourfold increased congestion (orthopnea, jugular venous distension, risk of HF. Once HF was present, one third of men edema, weight gain, or increased need for diuret- and women died within 2 years of diagnosis. The ics), persistent elevated filling pressure, and reduced 6-year mortality rate was 82% for men and 67% for exercise capacity. Interestingly, in many studies women, which corresponded to a death rate four- performed over the last decade, exercise capacity to eightfold greater than that of the general popu- has been demonstrated to be the most important lation of the same age. Sudden death was common, component of the risk profile among patients with accounting for 28% and 14% of the cardiovascular HF. Since the early 1990s, more than 100 studies deaths in men and women, respectively, with HF. have demonstrated that peak VO2 is a significant Hypertension and coronary disease were the pre- univariate or multivariate predictor of outcomes dominant causes of HF and accounted for more in patients with HF. Some of the larger studies are than 80% of all clinical events. Factors reflecting outlined in Table 10-1. deteriorating cardiac function were associated with a substantial increase in risk of overt HF. These Increased reliance on the role of exercise testing include low vital capacity, sinus tachycardia, and for decision-making in HF has occurred for several left ventricular hypertrophy by ECG. In 2003, more reasons. The recognition that exercise capacity, than 550,000 cases of HF were diagnosed in the U.S., expressed simply as workload achieved (i.e., METs) but only 2000 heart transplants were performed.4 or exercise time, was a significant prognostic For the remainder, quality of life decreases and marker in patients with cardiovascular disease

C H A P T E R 1 0 Exercise Testing in Patients with Heart Failure and Left Ventricular Dysfunction 331 TA B L E 1 0 – 1 . Summary of major studies using ventilatory gas exchange to predict outcomes in chronic heart failure Investigator Year No. of Mean age Mean Annual Szlachcic 1985 subjects (years) follow-up mortality 27 56 (months) (%) Findings 12 Likoff 1987 201 62 ± 10 28 40 Peak VO2 ≤ 10 mL/kg/min had 77% mortality; peak VO2 > 10 mL/kg/min Willens 1987 30 58 6 had 21% mortality Stevenson 1990 42 48 ± 9 14 ± 6 23 Peak VO2 > 13 mL/kg/min was Stevenson 1990 107 53 ± 11 6±5 independent predictor of increased Mancini 1991 122 50 ± 11 11 ± 9 mortality 127 55.1 ± 9.3 14.6 Parmeshwar 1992 24.6 Peak VO2 was best independent predictor 12 ± 14 of survival by multivariate analysis Saxon 1993 528 50 ± 12 60 — Patients who survived more than 6 Cohn 1993 V-HEFT 59.5 ± 8 I = 642, 58 ± 10 12 months on sustained medical therapy V-HEFT 12 II = 804 achieved peak VO2 comparable to that 75 14 of patients surviving after cardiac 24 ± 18 transplantation — 3 Ability to increase peak VO2 by ≥ 2 mL/ 25 ± 15 kg/min to a level ≥ 12 mL/kg/min) 10 ± 6 was an indication to defer transplantation in favor of more compromised candidates — Peak VO2 > 14 mL/kg/min had 6% 1-year mortality versus 53% in patients with peak VO2 ≤ 14 mL/kg/min 18 By both univariate and multivariate analysis, peak VO2 (13.7 mL/kg/min) was one of several independent predictors of outcome 24 By both univariate and multivariate analysis, peak VO2 < 11 mL/kg/min was independent predictor of heart failure death but not of sudden death 8 Peak VO2 was highly significant univariate and multivariate predictor of survival Roul 1994 265 52 ± 13 12 Peak VO2 (threshold value 14 mL/kg/min) Stevenson 1995 32 was independent prognostic factor and 24 best predictor of risk of death DiSalvo 1995 67 51 ± 10 33 ± 3 Peak VO2 ≤ 10 mL/kg/min was one of — several predictors of death or urgent Aaronson 1995 272 52 ± 12 transplantation in patients with Class 64 49 ± 10 2 IV symptoms Wilson 1995 17 Percent VO2 rather than peak VO2 predicted survival. RVEF was more Rickenbacher 1996 116 46.6 ± 10 potent predictor than peak VO2 or Chomsky 1996 185 51.4 ± 10 percent VO2 Peak VO2 ≥ 14 mL/kg/min predicted survival. Peak VO2 was better predictor than percentage-predicted VO2 Low peak exercise VO2 (< 14 mL/kg/min) could not be used to accurately identify patients with heart failure who had severe hemodynamic dysfunction during exercise Peak VO2 predicted subsequent heart transplantation, but not cardiac death Peak VO2 (dichotomized at 10 mL/kg/min) was independent predictor of survival both by univariate analysis and multivariate analysis Continued

332 E X E R C I S E A N D T H E H E A R T TA B L E 1 0 – 1 . Summary of major studies using ventilatory gas exchange to predict outcomes in chronic heart failure—contd Investigator Year No. of Mean age Mean Annual subjects (years) follow-up mortality (months) (%)* Findings 60 50 ± 9 27 ± 11 Levine 1996 17 Peak VO2 (≥16 mL/kg/min) was used 12 as criteria for delisting patients from Haywood 1996 141 — 36 the waiting list for transplantation Aaronson 1997 12 ± 3 Consequent improvement was Kao 1997 Derivation 51 ± 10 observed in exercise performance and sample = 268; 51 ± 10 12 ± 3 hemodynamic parameters in these validation 20 ± 14 patients after the 27 ± 11 month sample = 199 32 follow-up period 76 25 ± 17 — All deaths among patients on a transplant waiting list occurred in those 17 ± 13 with cardiac index <2 L/min/m2 or 47 ± 28 peak VO2 < 12 mL/kg/min 19 ± 25 47 ± 28 20 Noninvasive multivariate model outperformed invasive model in predicting risk 11.8 In patients at extremes of exercise performance spectrum (VO2 max < 12 or >17 mL/kg/min), VO2 max related to mortality. In patients with moderate to severe exercise intolerance (VO2 max 12–17 mL/kg/min), prognostic value of Richards 1997 76 51 ± 1 Women VO2 max was limited 102 58 ± 10 10.5; Percent of predicted peak VO2 achieved Ponikowski 1997 178 52 described degree of functional Cohen Solal 1997 500 50 ± 10 men 14.5 impairment in women more accurately 653 52 ± 9 than peak VO2 644 48 ± 11 12 Peak VO2 < 14 mL//kg/min was one of 219 55 ± 10 644 48 ± 11 several independent predictors of death 12 Both peak VO2 > 17 mL//kg/min and age-predicted peak VO2 (>63%) were predictors of survival by univariate analysis, but only age-predicted peak VO2 was independent predictor of survival in multivariate analysis Osada 1998 15 Peak VO2 ≤ 14 mL//kg/min was univariate and multivariate predictor of mortality. Peak exercise SBP < 120 mmHg and percent predicted Opasich 1998 peak VO2 ≤ 50% predicted mortality in Myers 1998 patients with peak VO2 ≤ 14 mL/kg/min Metra 1999 24 Peak VO2 stratified by <10, 10–18, and >18 mL/kg/min identified high, medium, and low risk 5.3 Peak VO2 was better predictor of survival than clinical, hemodynamic, or other exercise variables 14.5 Peak exercise stroke work index was most powerful marker of 1-year survival, peak VO2 was most powerful marker of 2-year survival Myers 2000 5.3 Peak VO2 was strongest predictor of survival among clinical and exercise test variables. Different cutoffs for peak VO2 (between 10 and 17 mL/kg/min) all had roughly 20% differences in survival

C H A P T E R 1 0 Exercise Testing in Patients with Heart Failure and Left Ventricular Dysfunction 333 TA B L E 1 0 – 1 . Summary of major studies using ventilatory gas exchange to predict outcomes in chronic heart failure—cont’d Investigator Year No. of Mean age Mean Annual Cohen-Solal 2002 subjects (years) follow-up 53 ± 10 (months) mortality 175 25 ± 10 60 ± 10 (%)* Findings Mezzani 2003 570 20 ± 14 407 57 ± 11 Peak circulatory power (the product of deGroote 2004 26 systolic blood pressure and VO2) was the only multivariate predictor of prognosis Patients who achieve peak RER >1.15 have markedly better survival even when peak VO2 is ≤ 10 mL/kg/min ≈8 B-natriuretic peptide, in combination with % age-predicted peak VO2 achieved, were strong predictors of survival CHF, congestive heart failure; RER, respiratory exchange ratio; RVEF, right ventricular ejection fraction. was made in the early 1970s. Expired gas analysis relative indication for transplantation in the guide- techniques are now much more widespread, in part lines. However, as discussed later in this chapter, because of computerization and increased automa- there are a number of caveats that must be consid- tion, but also due to an appreciation for their appli- ered when applying specific cutpoints to assess risk. cations to various cardiovascular and pulmonary disorders. Justification for their use in patients Questions That Remain with HF has been strengthened by studies describ- to Be Clarified ing clinical applications of ventilatory and gas- exchange abnormalities in HF.9 Cardiopulmonary The questions that remain to be clarified include exercise testing is now part of the standard workup the following: of the patient with HF, and the guidelines on trans- plantation consider this procedure an integral com- • What is the place of cardiopulmonary exercise ponent of the decision-making process regarding testing relative to clinical, hemodynamic, and transplantation. The widespread use of cardiopul- other data in the risk paradigm in patients monary exercise testing in patients with HF over the with HF? past 15 years has provided many groups the oppor- tunity to evaluate the role of peak VO2 in prognosis. • What is the optimal cutpoint for peak VO2 when selecting patients for transplantation listing? Although previous studies assessing risk using exercise testing have varied widely in terms of • Should peak VO2 be expressed as an absolute severity of HF, the use of different outcomes for value or corrected for age or body weight? assessing risk, application of different cutpoints for peak VO2, and inclusion or exclusion of other • How well do other ventilatory gas exchange clinical, exercise, and hemodynamic variables, peak responses (e.g., the VE versus VCO2 slope, venti- VO2 is clearly one of the more robust markers of latory threshold, rate of recovery of VO2) predict risk in HF. Directly measured peak VO2 has been risk? shown to outperform clinical, hemodynamic, and other exercise test data in predicting 1- to 2-year Each of these issues is discussed in this chap- mortality. Several investigators have reported ter relative to risk stratification and decision- that patients who achieve a peak VO2 greater than making in patients with HF. 14 mL/kg/min appear to have a prognosis similar to that among patients who receive transplanta- Exercise Tolerance and Selection tion (approximately 90% survival at 1 year). This of Transplant Recipients finding implies that transplantation can be safely deferred among these patients. This cutpoint has Because there are only approximately 5000 donor emerged as a clinically practical prognostic marker hearts available each year in the U.S., recipients in HF; a value less than 14 mL/kg/min is a must be carefully selected. In this regard, factors

334 E X E R C I S E A N D T H E H E A R T associated with 1 to 2-year survival among poten- heart rates in these patients (100 to 110 beats per tial candidates are critical. Historically, the major minute) and the relatively slow adaptation of the factors associated with poor short-term outcome heart to a given amount of submaximal work. As without transplantation have included an EF less a result, the delivery of oxygen to the working tis- than 15%, complex ventricular ectopy, sympa- sue is slower, contributing to earlier than normal thetic nervous system activation, and impaired metabolic acidosis and hyperventilation during exercise capacity, although there are many other exercise. Maximal heart rate is lower in transplant clinical markers that have been associated with patients than in normal subjects, which con- risk in HF (Table 10-2). With advances in the tributes to a reduction in cardiac output and exer- treatment for HF, many patients once thought to cise capacity. have end-stage HF can be stabilized by aggressive medical therapy. Although predicting the clinical Cardiopulmonary Exercise Testing course in individual patients is imprecise, trans- and Prognosis in HF plantation has been safely deferred in many patients by combinations of angiotensin-convert- Early Studies ing enzyme (ACE) inhibition or ACE-II blockade, diuretics, beta-blockade, and careful monitoring Several small studies were published in the early of patient status, including weight, electrolytes, to mid-1980s that addressed factors associated and renal function. Other patients will deteriorate with the risk of death in HF, in which the inclu- despite intensive medical management. Multi- sion of directly measured VO2 was seemingly disciplinary HF management programs have been incidental. Among the earliest studies was that set up to manage and monitor patients, and these of Szlachcic et al11 who performed resting and programs appear to improve survival.10 For this exercise hemodynamic measurements, ventila- reason, many heart transplant centers have evolved tory gas exchange, and radionuclide ventriculog- into “heart failure management” clinics. raphy in 27 patients with HF and observed them for 1 year. Patients were dichotomized by those Increasing numbers of patients have undergone achieving a peak VO2 less than or equal to 10 cardiac transplantation for end-stage HF, and today versus those achieving a peak VO2 greater than approximately three quarters of these patients 10 mL/kg/min. The group of patients achieving remain alive after 5 years. Because the transplant less than or equal to 10 mL/kg/min were found patient’s heart is denervated, some intriguing to have worse hemodynamic responses, including hemodynamic responses to exercise are observed. higher pulmonary capillary wedge pressures, lower The heart is not responsive to the normal actions left ventricular and right ventricular EFs, and of the parasympathetic and sympathetic systems. lower exercise heart rates and cardiac indexes, and The absence of vagal tone explains the high resting the mortality rate over the subsequent year was higher among those with limited peak levels of TA B L E 1 0 – 2 . Variables associated with risk in VO2 (77% vs. 21%, P < 0.001). chronic heart failure Likoff et al12 evaluated 201 patients with HF Reduced ejection fraction and prospectively observed them for 28 months. Poor exercise capacity: Fifteen clinical, hemodynamic, and exercise - NYHA Functional Class III or IV variables were entered into a Cox proportional - Dyspnea on exertion hazards model. Three characteristics at study - Peak VO2 < 14 mL/kg/min entry predicted an increased mortality risk: the Heightened neurohormonal markers (BNP, ANP, presence of a third heart sound, low peak VO2, and diagnosis of ischemic cardiomyopathy. For patients endothelins, norepinephine) with a third heart sound, ischemic cardiomyopa- Complex ventricular ectopy thy, and peak VO2 less than the sample mean of Reduced cardiac index (<2.0 L/min/m2) 13 mL/kg/min, the 1-year mortality rate was 36%, Renal insufficiency (creatinine clearance <60 mL/min) whereas among patients without any of these risk Persistent signs of congestion (orthopnea, jugular venous markers the 1-year mortality rate was only 10% distension, edema, weight gain, increased need for (P < 0.001). diuretics) Left ventricular end-diastolic dimension >80 mm Willens et al13 studied 30 patients whose baseline Duration of heart failure evaluation included radionuclide angiography, Hyponatremia (serum sodium <134 mEq/L) echocardiography, 24-hour Holter monitoring, High pulmonary capillary wedge pressure ANP, atrial natriuretic peptide; BNP, brain natriuretic peptide; NYHA, New York Heart Association.

C H A P T E R 1 0 Exercise Testing in Patients with Heart Failure and Left Ventricular Dysfunction 335 and exercise testing and observed them for a mean 2-year survival rates, respectively. Patients with of 15 months. Univariate predictors of survival included peak VO2, age, estimated VO2 (from exer- an initial peak VO2 greater than 10 mL/kg/min had cise time), presence of left bundle branch block, left a 2-year event (death or transplant) rate of 28%, ventricular end-diastolic dimensions, and frequency of ventricular arrhythmias. Importantly, the best whereas among patients with a peak VO2 less than multivariate predictor of survival was peak VO2. 10 mL/kg/min, the event rate was 52%. These early studies were the first to suggest Opasich et al16 evaluated predictors of survival that peak VO2 may have an important place in risk stratification among patients with HF, and they 6 months, 1 year, and 2 years after consideration for provided an important springboard for the routine use of gas exchange techniques in the evaluation of transplantation among 653 patients. The presence transplant candidates today. The study of Szlachcic et al11 was the first to suggest that a single cutpoint, of a contraindication to exercise testing identified 10 mL/kg/min, can have considerable importance in risk-stratifying patients with severe HF. However, very high-risk patients, which confirmed observa- the 12 deaths in that study, like the eight deaths in tions made by Stevenson et al15 and many studies in the study carried out by Willens et al13 made them too small to perform valid multivariate analyses. patients with coronary disease. Peak VO2 stratified into three levels (≤10, 10 to 18, and >18 mL/kg/min) Multivariate Studies in the 1990s identified groups at high, medium, and low risk, In a landmark 1991 study that provided an impetus respectively. However, in patients in New York for many others, Mancini et al14 observed three groups of patients referred for transplantation over Heart Association Class III or IV, peak VO2 did not 2 years. One group comprised patients accepted for have prognostic power. transplant on the basis of achieving a peak VO2 less than 14 mL/kg/min; a second group comprised Haywood et al17 studied patients accepted for patients considered too well for transplant (peak VO2 >14 mL/kg/min); and a third group comprised heart transplantation listing between 1986 and patients with a peak VO2 less than 14 mL/kg/min but rejected from transplantation for noncardiac 1994 at Stanford University. Of 141 consecutive reasons. Patients with preserved exercise capacity (>14 mL/kg/min) had 1- and 2-year survival rates patients accepted for cardiac transplant, all deaths of 94% and 84%, respectively, roughly equivalent to those observed after transplantation. This was and 88% of patients who deteriorated to status in contrast to patients with poor exercise capacity (peak VO2 <14 mL/kg/min) who were rejected one while on the waiting list had either a cardiac for transplantation, among whom 1- and 2-year index less than 2.0 L/min/m2 or a peak VO2 less survival rates were only 47% and 32%, respec- than 12 mL/kg/min. In those with a cardiac index tively. By both univariate and multivariate analysis, less than 2.0 L/min/m2 and a peak VO2 less than peak VO2 was the best predictor of survival. This 12 mL/kg/min, 38% died or deteriorated to status study fostered the concept that a single cutpoint, 14 mL/kg/min, provides a clinically applicable one during the first year on the waiting list. cutpoint between patients who require transplan- tation for survival benefit and those who do not. Conversely, all patients with a cardiac index equal or greater than to 2.0 L/min/m2 and peak VO2 Stevenson et al15 studied 500 patients who equal or greater than 12 mL/kg/min survived were discharged on tailored medical therapy after evaluation for transplantation, and the risk of death throughout the follow-up. These investigators or need for urgent transplantation was studied over the subsequent 2 years. Low cardiac index later studied patients referred for heart transplan- and high filling pressures did not confer a higher risk, but serum sodium levels less than 133 mEq/L tation but selected for medical management. One and left ventricular diastolic dimensions greater than 80 mm were associated with 34% and 23% hundred sixteen patients were observed for a mean of 25 ± 15 months. In this comparatively healthy group (mean peak VO2 17.4 ± 4.3 mL/kg/min, mean pulmonary capillary wedge pressure 16 ± 9 mmHg), there were only eight cardiac deaths, and no clinical, exercise, or hemodynamic variable significantly predicted death by logistic regression. By multivariate regression, only pulmonary artery systolic pressure and duration of HF predicted the need for later transplantation. Saxon et al18 studied 528 consecutive patients hospitalized for advanced HF. Predictors of death or hemodynamic deterioration requiring transplan- tation were evaluated over the subsequent year; a total of 129 patients (24%) experienced one of these outcomes. A serum sodium level equal to 134 mEq/L, pulmonary arterial diastolic pressure greater than 19 mmHg, left ventricular diastolic dimension greater than 44 mm/m2, peak VO2 less than 11 mL/kg/min, and the presence of a perma- nent pacemaker were independent predictors of

336 E X E R C I S E A N D T H E H E A R T hemodynamic deterioration or death. In the 10 and 14 mL/kg/min.37 Patients whose peak VO2 was less than 14 mL/kg/min but greater than 50% absence of any of these risk factors, the risk of a of their age and gender-predicted value had a 3-year survival rate similar to patients who achieved a negative outcome was only 2%. The presence of peak VO2 greater than 14 mL/kg/min (93% versus 91%). Patients with limited exercise capacity had hyponatremia and any two additional risk factors a particularly poor survival if they were unable to raise peak exercise systolic blood pressure to at least raised the risk to greater than 50%. 120 mmHg; the 3-year survival rate among these Cohn et al19 studied 1446 patients prior to patients was 55%, compared to an 83% survival rate among patients with a measurement less than randomization in a vasodilator multicenter trial. 14 mL/kg/min whose peak exercise blood pressure was greater than 120 mmHg. Patients were followed up to 5 years. EF, peak VO2, cardiothoracic ratio, and plasma norepinephrine Roul et al22 prospectively studied 75 patients with clinical, radionuclide, and right heart catheter- were independent predictors of mortality. An inter- ization data and observed them for 1 year.5 The cohort was divided into two groups based on peak esting interaction between EF and peak VO2 was VO2 greater or less than 14 mL/kg/min. Patients observed; EF was more influential as a prognostic with preserved exercise capacity had lower left ven- tricular filling pressures, lower total peripheral factor among patients whose peak VO2 was above resistance, lower creatinine and blood urea nitro- the median (14.5 mL/kg/min). Likewise, peak VO2 gen levels, and higher exercise duration. During was a significant additional prognostic marker only the 1-year follow-up, nine patients died in the group with peak VO2 levels less than 14 mL/kg/min, among patients whose EF was above the median whereas there were no deaths in the group with lev- els more than 14 mL/kg/min. Seven major events (28%). The increase in risk for patients with EFs requiring hospitalization occurred in the limited exercise capacity group versus only three in the below the median more than doubled when peak preserved group. VO2 was above the median (risk ratio 2.43) com- Kao et al23 studied survival rates among 178 pared to when peak VO2 was below the median patients who underwent exercise testing at a base- (risk ratio 1.43). Similarly, the increase in risk for line evaluation. Patients whose peak VO2 levels were less than 12 mL/kg/min had a higher mortality rate patients with peak VO2 values below the median when compared to patients with peak VO2 levels more than doubled when EF was above the median greater than 17 mL/kg/min. However when patients were compared by tertiles of peak VO2 within the (risk ratio 2.17) compared to when EF was below intermediate range (12 to 17 mL/kg/min), no dif- ferences in survival were observed between the the median (risk ratio 1.27). tertiles. These investigators suggested that although peak VO2 differentiates patients who do A comprehensive evaluation of clinical, hemo- and do not survive at the extremes of the exercise performance spectrum, the prognostic value of dynamic, and exercise variables was performed dur- peak VO2 in the intermediate range (in which most patients fall) is limited. ing a 10-year period among 644 patients referred for evaluation of HF at Stanford.20 The longer fol- Cardiopulmonary Markers of Risk Other than Peak VO2 low-up period (mean, 4 years), large number of Although peak VO2 defines the limits of the car- deaths (187), and the inclusion of both measured diopulmonary system, there are other cardiopul- monary responses which are important in and predicted VO2 made it unique among the defining the severity of HF and prognosis. These multivariate studies, and one of the more robust responses are to one extent or another related to the ventilatory response to exercise, the capacity data sets to evaluate prognosis. Univariately, the most powerful predictors of death were from the exercise test; peak VO2, VO2 at the ventilatory threshold, VO2 expressed as a percentage of the predicted value, peak systolic blood pressure lower than 130 mmHg, and watts achieved were signifi- cant predictors of death. Age was the only predictor of death among clinical variables, and hemody- namic variables, including EF, pulmonary capil- lary wedge pressure, and left ventricular dimensions were not important predictors of outcome. By mul- tivariate analysis, peak VO2 was the only signifi- cant predictor of death. This study provided the strongest evidence to date that directly measured peak VO2 not only outperforms clinical and hemo- dynamic data but also was a better predictor of death than exercise duration or watts achieved. Osada et al21 reported results from 500 patients observed for a mean of 25 months. Patients who achieved a peak VO2 greater than 14 mL/kg/min had a 3-year survival rate of 93%, compared with 68% among patients whose peak VO2 was between

C H A P T E R 1 0 Exercise Testing in Patients with Heart Failure and Left Ventricular Dysfunction 337 of the cardiopulmonary system to adapt to the these responses have been demonstrated to have greater prognostic value than peak VO2, and these demands of a given work rate, or the ability of the studies are summarized in Table 10-3. cardiopulmonary system to recover from a bout of Ventilatory Threshold exercise. Responses such as the anaerobic threshold The ventilatory threshold, one important submax- imal marker of cardiopulmonary function with a (AT), the VE/VCO2 slope, oxygen uptake kinetics, long history, has been employed in surprisingly rate of recovery of VO2, and the oxygen uptake effi- few multivariate models to predict risk in HF. ciency slope (OUES) have been used with greater frequency to classify functional limitations and stratify risk in patients with heart disease. Examples of these are illustrated in Figure 10-1. Some of Peak VO2 VE/VCO2 Slope 3500 180 160 3000 Normal 140 2500 120 100 Oxygen uptake (ml/min) VO2 2000 VE L/min CHF 80 Slope = 39.0 60 1500 CHF 40 Normal 1000 20 Slope = 25.1 500 0 0 0 123456 7 0 1 2 3 4 5 6 7 8 9 10 VCO2 L/min Time (minutes) VO2 Recovery 4000 Oxygen uptake efficiency slope 4000 3000 Normal T1/2 VO2 VO2 ml/min 2000 Normal 2000 1000 CHF 15 20 0 CHF 100 150 200 T1/2 VO2 10 50 0 5 10 0 Time (minutes) Minute ventilation (L/min) ■ FIGURE 10–1 Examples of four different cardiopulmonary exercise test methods that have been used to estimate prognosis in patients with cardiovascular disease. The peak VO2 responses (upper left) are taken from a normal subject and a typical patient with chronic heart failure the same age. The VE/VCO2 slope (upper right) is derived from the slope of the regression line between VE and VCO2, excluding data points beyond the ventilatory threshold. VO2 in recovery (lower left) shows a more graded recovery response in the CHF patient (i.e., longer recovery time), despite the lower exercise capacity. T1/2 represents the time required for a 50% fall from the peak VO2 value. The OUES (lower right) is derived by plotting VO2 against the log of VE; a steeper slope reflects a lower VE for any given VO2, that is, more efficient ventilation. From Myers J: Applications of cardiopulmonary exercise testing in the management of cardiovascular and pulmonary disease. Int J Sports Med 2005;26:S49-S55.

338 E X E R C I S E A N D T H E H E A R T TA B L E 1 0 – 3 . Prognostic studies on ventilatory gas exchange responses other than peak Vo2 Mean follow-up Study (Ref) Year Subjects, (N) Mean age, Y Period, Mo Findings VE/VCO2 slope 1997 CHF (173) 59±12 – Chua 2000 CHF (303) 59 ± 11 47 VE/VCO2 slope (>34) provided stronger Francis 1999 CHF (470) 52 ± 11 prognostic information than peak VO2 Robbins CHF (223) 63 ± 11 18 2002 CHF (142) 52 ± 10 21 Peak VO2 and VE/VCO2 slope Gitt CHF (213) 57±13 16* similar in prognostic power 2000 CHF (600) 57 ± 9 32 Kleber 26 VE/VCO2 slope and low 2003 CHF (72) 63 ± 12 chronotropic index most powerful Arena 2002 DCM (153) 50 ± 12 – Corra 15* multivariate predictors of death Bol 2000 VO2 at the anaerobic threshold <11 mL/kg/min and VE/VCO2 de Groote (15) 1996 slope best predictors of risk VE/VCO2 slope outperformed peak VO2 as predictor of death, Tx, or LVAD VE/VCO2 slope stronger predictor of cardiac mortality than peak VO2 VE/VCO2 slope was strongest predictor of death or Tx. Peak VO2 ≤10mL/kg/min and VE/VCO2 slope ≥35 had similar mortality rate VE/VO2 slope was more powerful predictor of mortality than clinical variables or peak VO2 in recovery VO2 recovery significantly delayed in DCM versus normals; ratio of exercise and recovery VO2 independently predicted survival VO2 Kinetics 2003 CHF (202) 52 ± 11 29 Mean response time >50 sec was Rickli 2003 CHF (146) 52 ± 10 strongest predictor of death or Tx, Schalcher followed by predicted VO2 <50% 25 Mean response time was strongest Brunner-LaRocca 1999 CHF (48) 55 ±10 Oxygen Uptake Efficiency Slope predictor of survival or freedom from Tx or hospitalization, followed by VE/VCO2 slope 22 Mean response time >60 sec was significant predictor of mortality, and was more powerful than peak VO2 Pardaens 2000 CHF (284) 52 ± 11 16* Peak VO2 was stronger predictor of death or cardiovascular events than OUES or VE/VCO2 slope *Median DCM, dilated cardiomyopathy; LVAD, left ventricular assist device implantation; OUES, oxygen uptake efficiency slope; Tx, transplantation. Studies that have included the ventilatory thresh- stronger predictor of death.20 Gitt et al24 recently old have demonstrated that VO2 at this point sig- tested 223 consecutive patients with HF in nificantly predicts outcome. This point has the Germany. They compared the prognostic power of potential to be a particularly useful marker of out- peak VO2, VO2 at the ventilatory threshold, and come, since for many patients with HF, “maximal” the VE/VCO2 slope in predicting all-cause death. Cutpoints for VO2 less than or equal to 14 mL/kg/ exercise is not achieved for various reasons or is min, VO2 at the ventilatory threshold (VO2AT) less than 11 mL/kg/min, and a VE/VCO2 slope more difficult to define. In the Stanford study, VO2 at than 34 were used as threshold values for high the ventilatory threshold was a significant uni- risk of death. Patients with a peak VO2 less than or variate predictor of death in patients evaluated for HF, but in a multivariate analysis, peak VO2 was a

C H A P T E R 1 0 Exercise Testing in Patients with Heart Failure and Left Ventricular Dysfunction 339 equal to 14 mL/kg/min had a greater than three- SurvivalCox regression analysis both to be significant fold increased risk while a VO2AT less than 11 mL/ predictors of cardiac-related mortality and hospi- kg/min or a VE/VCO2 slope greater than 34 had talization (P < 0.01). Multivariate analysis revealed fivefold increased risks for early death. In patients that peak VO2 added additional value to the VE/VCO2 with both VO2AT less than 11 mL/kg/min and slope in predicting cardiac-related hospitalization, VE/VCO2 slope greater than than 34, the risk of but not cardiac mortality. Patients who exhibited early death was 10-fold higher. After correction for a VE/VCO2 slope greater than or equal to 34 had a age, gender, EF, and New York Heart Association particularly high probability of hospitalization class in a multivariate analysis, the combination (> 50% 1-year following evaluation, Figure 10-2). of VO2AT less than 11 mL/kg/min and VE/VCO2 The VE/VCO2 slope was demonstrated with receiver slope greater than 34 was the best predictor of operating characteristic curve analysis to be signif- 6-month mortality (relative risk = 5.1). icantly better than peak VO2 in predicting cardiac- related mortality (P < 0.05). Although area under VE/VCO2 Slope the receiver operating characteristic curve for the VE/VCO2 slope was greater than peak VO2 in pre- There is an impressive body of recent data demon- dicting cardiac-related hospitalization (0.77 strating the role of the VE/VCO2 slope in predict- versus 0.73), the difference was not statistically ing prognosis in HF. These studies have shown significant (P = 0.14). that the VE/VCO2 slope predicts mortality at least as well as, and independent from, peak VO2. This Kleber et al29 evaluated the cardiopulmonary response is usually expressed as the slope of the response to exercise in 142 patients with HF and best-fit linear regression line relating VE and VCO2 followed then for a mean of 16 months. Forty-four below the ventilatory compensation point for exer- events (37 deaths and seven instances of heart cise lactic acidosis (see Figure 3-1). While the slope transplantation, cardiomyoplasty, or left ventricular of this relationship is normally between 20 and assist device implantation) occurred. Among peak 30, values in the thirties are common in patients VO2, NYHA class, EF, total lung capacity, and age, with mild-to-moderate HF, and values in the for- the most powerful predictor of event-free survival ties are often observed in patients with more was the VE/VCO2 slope; patients with a VE/VCO2 severe HF.6,25-27 An elevated VE/VCO2 slope is a slope greater than or equal to 130% of age-and gen- reflection of the pathophysiology of the abnormal der-adjusted normal values had a significantly bet- ventilatory response to exercise in HF.6,26-28 Thus, ter 1-year event-free survival (88.3%) than patients the VE/VCO2 slope is elevated in the presence of with a slope greater than 130% (54.7%; P < 0.001). early lactate accumulation, ventilation/perfusion mismatching in the lungs (e.g., poor cardiac out- Robbins et al30 studied 470 consecutive patients put response to exercise), or the deconditioning with HF who were not taking beta-blockers and that is commonly observed in HF. Cardiac hospitalization (1 year) Corra et al27 performed cardiopulmonary exer- cise testing in 600 patients with HF and followed 1.1 them for major cardiac events (death or urgent transplantation) over a 2-year period. The VE/VCO2 1.0 slope was the strongest independent predictor of a cardiac event (outperforming peak VO2, EF, and .9 other clinical and exercise test variables). The best < 34 cutpoint for predicting risk was 35 (relative risk = 3.2 for a VE/VCO2 slope >35). The total mortality .8 rate in patients with a VE/VCO2 slope greater than or equal to 35 was 30% versus 10% in patients .7 with a VE/VCO2 slope less than 35. Patients with a VE/VCO2 slope greater than or equal to 35 had a .6 similar mortality rate as those with a peak VO2 less than or equal to 10 mL/kg/min. .5 ≥ 34 Arena et al25 from our laboratory compared the .4 prognostic power of peak VO2 and the VE/VCO2 slope in 213 patients with HF. Peak VO2 and the –2 0 2 4 6 8 10 12 14 VE/VCO2 slope were demonstrated with univariate Months ■ FIGURE 10–2 Kaplan-Meier survival curves for 1-year cardiac-related hospi- talization using a VE/VCO2 slope threshold <34 versus ≥34 (P < 0.0001). From Arena R, Myers J, Aslam S, et al: Peak VO2 and VE/VCO2 slope in patients with heart failure: A prog- nostic comparison. Am Heart J 2004;147:354-360.