Chronotropic Response to Exercise
Improved Performance of ST-Segment Depression Criteria After Adjustment for Heart Rate Reserve
Background Heart rate (HR) response to exercise plays an important role in the diagnosis of coronary artery disease (CAD). Adjustment of ST-segment depression for the change in HR with exercise increases the accuracy of the exercise ECG in the detection of CAD. In addition, an attenuated HR response to exercise, a manifestation of chronotropic incompetence, may have independent diagnostic value for CAD.
Methods and Results The diagnostic value of adjusting the magnitude of ST-segment depression, the ST-segment (ST)/HR index, and the ST/HR slope for chronotropic response to exercise was assessed in 283 control subjects and 337 patients with CAD by dividing each ST measurement by the fraction of HR reserve achieved. At a matched specificity of 96%, ST-segment depression of >160 μV identified CAD with a sensitivity of 52%, an ST/HR index of >1.69 μV/bpm identified CAD with a sensitivity of 90%, and an ST/HR slope of >2.96 μV/bpm identified CAD with a sensitivity of 88%. Adjustment for HR reserve improved the sensitivity of each method: adjusted ST-segment depression of >176 had a sensitivity of 87% (P<.0001), an adjusted ST/HR index of >2.14 had a sensitivity of 94% (P=.005), and an adjusted ST/HR slope of >3.47 had a sensitivity of 93% (P=.0001). In addition, the 94% and 93% sensitivities of the adjusted ST/HR index and ST/HR slope were significantly greater than the 87% sensitivity of adjusted ST-segment depression (P<.0001).
Conclusions Correction for an attenuated HR response to exercise improves performance of the simple and HR-adjusted ST-segment depression criteria for the identification of CAD. These findings support assessment of the degree of chronotropic reserve in routine evaluation of the exercise ECG.
Interpretation of the exercise ECG in the identification of CAD has traditionally focused on the magnitude and configuration of ST-segment depression present at the end of exercise or in early recovery.1 2 3 4 5 6 However, clinical usefulness of the exercise ECG has been limited by the poor sensitivity of standard ST-segment analysis,1 2 3 4 5 6 which in part reflects an inability to achieve a threshold magnitude of ST-segment depression due to submaximal effort duration in patients with CAD.6 Independent of ST-segment depression measurements, an attenuated HR response to exercise, a measure of chronotropic incompetence, has been associated with the presence and severity of CAD7 8 9 10 and has been shown to be a predictor of subsequent coronary heart disease events11 12 13 and of total mortality,13 even after adjustment for ST-segment findings.13 However, performance of a combination of the chronotropic and ST-segment response to exercise has not been directly evaluated.
Recognition that the magnitude of ST-segment depression recorded on the surface ECG in a patient with CAD represents a product of the anatomic severity of disease and the incremental myocardial workload reflected by increasing HR with exercise6 14 15 16 has resulted in the development and use of the ST/HR index and the ST/HR slope, HR-adjusted indexes of ST-segment depression that significantly improve performance of the exercise ECG.5 6 17 18 19 20 21 22 23 24 25 However, calculation of the ST/HR index and the ST/HR slope does not directly take into account the chronotropic response to exercise. Therefore, the purpose of this study was to examine the value of direct integration of the chronotropic response to exercise into the assessment of both simple and HR-adjusted ST-segment depression criteria for the identification of CAD.
There were 283 control subjects and 337 patients with known or suspected CAD who underwent exercise ECG and met previously defined criteria.21 23 Performance of standard and HR-adjusted ST-segment depression criteria for the identification of CAD in these patients has been previously described.23
There were 231 men and 52 women in the control group (mean age, 48±10 years). These asymptomatic subjects were selected to represent clinically normal ambulatory populations rather than volunteer groups or patients with normal coronary arteries found at catheterization.21 23 All subjects had no history of cardiac disease or hypercholesterolemia or family history of premature death due to CAD; they did not have diabetes mellitus; they had normal cardiac physical examinations, blood pressure, and resting ECGs; and they were not taking any cardioactive medications.
Patients With CAD
There were 246 men and 91 women in the patient group (mean age, 60±10 years) with stable exertional angina and known or suspected CAD who underwent exercise ECG over a 4-year period and met study entry criteria. In this group, there were 153 patients with clinical angina (94 men and 59 women; mean age, 61±9 years) and 184 patients with CAD proved through catheterization (152 men and 32 women; mean age, 64±11 years). This combination of patients was selected to represent both ambulatory CAD populations not selected for angiography (and thus free of accelerating symptoms and post-test referral bias) and patients with angiographically proven CAD.21 23 Patients with left bundle-branch block or myocardial infarction within 8 weeks were not included. Forty patients had resting ECG evidence of previous Q-wave myocardial infarction, and 31 patients had ECG evidence of left ventricular hypertrophy. There were 130 patients who were not taking medications; among the remaining 207 patients, 152 were taking β-blocking drugs, 117 were taking long-acting nitrates, and 118 were taking calcium channel blockers at the time of exercise testing.
Exercise ECGs were performed with participants on a treadmill with the use of a computerized exercise system modified by the addition of a bipolar lead CM5, as previously described in detail.23 All patients exercised according to the Cornell protocol,26 our more gently graded modification of the Bruce protocol that produces the small HR increments between stages necessary for accurate determination of the ST/HR slope.25 The protocol divides each stage of the modified Bruce protocol into 2-minute half-stages, beginning with stage 0 at 1.7 mph and 0% grade and gradually increasing in a stepwise fashion to stage 5 and 5.0 mph at an 18% grade.26 Age-adjusted target HRs were sought as the exercise end point for all studies,21 23 but tests were also terminated for limiting chest discomfort, dyspnea, or fatigue; for a 10 mm Hg drop in systolic blood pressure; and for the development of nonsustained ventricular tachycardia. Computer-calculated ST-segment amplitudes, measured to the nearest 10 μV at a point 60 ms after the J point with the end of the PR segment as reference, were obtained in each lead after each minute of exercise and at peak exercise.21 23 Exercise tests were evaluated using standard ECG criteria based on the measured amount of ST-segment depression on the peak exercise ECG2 21 and were considered to be positive in the presence of 0.1 mV (100 μV) of additional horizontal or downsloping ST-segment depression.
Calculation of the maximal ST/HR slope was performed using linear regression analysis to relate the magnitude of ST-segment depression in each lead (except aVR, aVL, and V1, which were excluded from all analyses) to HR at the end of each stage of exercise and at peak exercise, according to methods previously reported in detail.6 21 22 23 The highest ST/HR slope with a significant coefficient of correlation among all the leads was taken as the test result. The ST/HR index was calculated by dividing the maximal additional ST-segment depression at end exercise (corrected for any ST-segment depression in that lead on the upright preexercise resting ECG) by the exercise-induced change in HR.5 6 21 22 23
Determination of Chronotropic Response
Exercise performance was examined in a number of ways. Simple measures of HR response to exercise included peak HR achieved, the percentage of age-predicted target HR achieved, and the change in HR from standing rest to peak exercise. Peak METs achieved for each stage of the Cornell protocol were calculated based on speed and grade of the treadmill.27 Additional measures of chronotropic response based on the concept that a valid measure of HR response should also take into account age, resting HR, and possibly exercise capacity28 29 were also calculated.
Because the ability to achieve target HR is in part directly related to resting HR,28 the HR response to exercise at any stage of exercise can be assessed by the fraction of the HRR achieved,28 29 calculated as:Fraction of HRR|<|=|>|(HR_|<|stage|>||<|-|>|HR_|<|rest|>|)/(100% MPHR|<|-|>|HR_|<|rest|>|)where HRstage is the HR at the stage of exercise being evaluated, HRrest is the HR on the standing preexercise ECG, and 100% MPHR is the age-predicted target HR. In analogous fashion, the fraction of MR achieved can be calculated as:Fraction of MR|<|=|>|(METs_|<|stage|>||<|-|>|1)/(METs_|<|peak|>||<|-|>|1)where METsstage is the workload at a fixed, submaximal exercise stage, 1 is the approximate workload in METs at rest,29 and METspeak is the workload in METs at peak exercise. From these two relations, a mathematical expression of the ratio of HR to metabolic response can be calculated28 29 as:Chronotropic Response Index|<|=|>|Fraction of HRR/Fraction of MR
In control subjects, the chronotropic response index is ≈1, which is a reflection of the generally linear relation between HR and metabolic work during exercise.28 However, for the present study, in which peak exercise ST changes are being evaluated, we wanted to assess chronotropic response at peak rather than at a submaximal exercise workload.10 13 Making the appropriate substitutions in Equations 1 and 2, the chronotropic response index is equal to the fraction of HRR when calculated at peak exercise. Thus, for the present study, ST-segment depression criteria were corrected for an attenuated HR response to exercise by dividing the magnitude of ST-segment depression at peak exercise, the maximal ST/HR index, and the maximal ST/HR slope by the fraction of HRR.
In the patients who underwent catheterization, selective coronary cineangiography was performed as previously reported.21 22 23 The degree of obstruction was defined as the greatest percentage reduction in luminal diameter in any view compared with the nearest control segment. According to 50% luminal diameter obstruction criteria, there were 41 patients with one-vessel CAD, 61 with two-vessel CAD, and 82 with three-vessel CAD. Seventeen patients had left main CAD, including four with additional two-vessel disease and 13 with additional three-vessel disease.
Data Analysis and Statistical Methods
Mean and SD values are reported for each variable by group and by sex and were compared statistically with the use of two-way ANOVA. Simple proportions were compared with the use of χ2 analysis. The independent predictive power of chronotropic and ST-segment depression variables for the presence of CAD was determined with logistic regression analysis. Definitions of test sensitivity and specificity conform to standard use.30 Test specificity of each method for the identification of CAD was first determined in the 283 control subjects, and test sensitivity of each method was then assessed in the 337 patients with known or suspected CAD. Partitions with matched specificity of 96% found for standard test criteria were used to compare sensitivity of the simple magnitude of ST-segment depression at end exercise, the ST/HR index, and ST/HR slope, with and without adjustment for fraction of HRR achieved, with the use of McNemar's modification of the χ2 method for paired proportions. Because test sensitivity and specificity of a test are dependent on the partition value chosen for test positivity, test accuracy was also compared with the use of ROC curve analysis. ROC curves were compared statistically by means of a univariate Z test of the difference between the areas under two ROC curves.31 For all comparisons, a value of P<.05 was required for rejection of the null hypothesis.
Exercise Test Performance
Effort tolerance, HR, and chronotropic response to exercise according to sex and the presence or absence of CAD are shown in Table 1⇓. Both men and women with CAD exercised for a shorter time and to lower peak METs than did subjects without CAD, and they had lower resting and peak exercise HRs and a smaller change in HR with exercise and achieved a lower mean percentage of maximal age-predicted HR and a significantly lower fraction of HRR. Significant sex differences were found for exercise duration, peak METs achieved, rest HR, and change in HR with exercise, but there were no significant differences in peak HR, percentage of maximal predicted HR, or the fraction of HRR achieved between men and women. These differences and similarities persisted after further adjustment for β-blocker use. Of note, there were modest correlations between exercise duration and peak HR and fraction HRR in control subjects (r=.30 and .25, each P<.0001) and in patients with CAD (r=.33 and .39, each P<.0001), with no significant differences between men and women.
ST-segment depression, ST/HR index, and ST/HR slope findings, with and without adjustment for fraction of HRR achieved, are shown in Table 2⇓. Patients with CAD had significantly greater mean ST-segment depression and higher ST/HR index and ST/HR slope findings than did control subjects, independent of adjustment for fraction of HRR. However, sex differences in ST/HR index and ST/HR slope findings were minimized by further HR adjustment for the fraction of HRR.
Identification of CAD
The performance of various measures of the chronotropic response to exercise for the identification of CAD is examined in Table 3⇓. At matched specificity of 96% found for standard ST-segment depression criteria, simple measures of HR response to exercise had sensitivities ranging from 66% to 75%; exercise duration, 73%; peak METs, 72%; and attenuated HR response alone, defined by a fraction of HRR <0.588, 71%. Comparison of ROC curve areas confirmed the similar overall test accuracy of these measures for the detection of CAD.
Standard ST-segment depression criteria based on the magnitude and configuration of ST-segment depression at end exercise identified CAD with a specificity of 96% (272 of 283) and a sensitivity of 62% (210 of 337). At matched specificity of 96% with standard test criteria, simple ST-segment depression had a sensitivity of 52%; the ST/HR index, 90%; and the ST/HR slope, 88% (Table 4⇓). When the independent value of ST-segment depression and exercise performance criteria, including simple ST-segment depression, the ST/HR index, the ST/HR slope, exercise duration, METs achieved, HR responses, and fraction of HRR for the diagnosis of CAD were assessed using stepwise logistic regression analyses, only the ST/HR slope (β=1.35, χ2=82.6, P<.0001) and the fraction of HRR achieved (β=−9.14, χ2=53.0, P<.0001) provided independent and significant diagnostic information for the identification of CAD.
As a result of these findings, the impact of adjusting ST-segment depression criteria for fraction of HRR achieved on test performance was further examined (Table 4⇑ and Fig 1⇓). Correction for chronotropic response to exercise using the fraction of HRR significantly improved sensitivity of ST-segment depression criteria compared with unadjusted criteria. At a matched specificity of 96%, adjusted ST-segment depression had a sensitivity of 87% (P<.0001); an adjusted ST/HR index, 94% (P=.005); and an adjusted ST/HR slope, 93% (P=.0001). Comparison of ROC curves demonstrated that the improvement in overall performance provided by adjustment for an attenuated HR response to exercise for each method was independent of partition value selection (Fig 1⇓). Of note, using the best-fit logistic regression model including the unadjusted ST/HR slope and fraction of HRR achieved produced similar test sensitivity (94%) and overall performance compared with the adjusted ST/HR index and ST/HR slope.
β-Blocker therapy had a significant effect on test sensitivity of the fraction of HRR and of ST-segment depression criteria adjusted for fraction of HRR achieved for the detection of CAD (Fig 2⇓). As a consequence of the blunted HR response to exercise, sensitivity of the fraction of HRR achieved was significantly higher in patients taking β-blockers than in those not on these medications (89% versus 56%, P<.0001). A less dramatic, but significant, difference in sensitivity was observed for adjusted ST-segment depression (93% versus 82%, P=.005). In contrast, there were only trends toward increased sensitivity for the adjusted ST/HR index (97% versus 91%, P=.07) and the adjusted ST/HR slope (96% versus 90%, P=.07) among patients taking β-blockers.
Because other cardiac medications could potentially affect HR adjusted criteria and the adjustment for chronotropic response, performance of ST-segment depression criteria was further examined in the subset of 130 patients with CAD who were on no cardiac medications at the time of exercise testing. In this subset, similar increases in sensitivity were observed after adjustment for fraction of HRR for simple ST-segment depression (53% versus 80%, P<.001) and the ST/HR slope (84% versus 89%, P<.05), with a small, but statistically nonsignificant, increase in sensitivity for the ST/HR index (88% versus 90%, P=NS).
ST/HR Index and ST/HR Slope
The effect of standard HR adjustment of ST-segment depression using the ST/HR index and the ST/HR slope on test performance for the detection of CAD is demonstrated in Table 4⇑ and Figs 2 and 3⇑⇓. Using unadjusted ST-segment depression measurements, sensitivity and overall performance of the ST/HR index and ST/HR slope were significantly greater than those for simple ST-segment depression criteria. In addition, although the sensitivity of simple ST-segment depression criteria was significantly improved by adjustment for chronotropic response, the 94% and 93% sensitivities of the ST/HR index and ST/HR slope after further adjustment for fraction of HRR remained significantly greater than the 87% sensitivity of adjusted ST-segment depression criteria. In addition, improved performance of the ST/HR index and ST/HR slope adjusted for fraction of HRR was independent of β-blocker use (Fig 2⇑). Comparison of ROC curves (Fig 3⇓) demonstrated the significantly greater overall accuracy of both the adjusted ST/HR index and the adjusted ST/HR slope relative to adjusted ST-segment depression criteria.
This study demonstrates that adjustment of ST-segment depression for the chronotropic response to exercise improves performance of both simple and HR-adjusted ST-segment depression criteria for the identification of CAD and that performance of the ST/HR index and the ST/HR slope remains superior to ST-segment depression criteria, even after adjustment of these criteria for an attenuated HR response to exercise. These findings support assessment of the chronotropic reserve in routine evaluation of the exercise ECG.
Numerous studies have demonstrated an attenuated HR response to exercise to be associated with an increased prevalence and severity of CAD.7 8 9 10 When chronotropic incompetence was defined as failure of the peak HR response to exercise to rise within 2 SDs of the expected increase based on age and sex, Chin et al7 found a higher prevalence of CAD in patients with chronotropic incompetence than in patients with control HR responses to exercise, both in patients with significant ST-segment depression (91% versus 70%) and in patients with negative exercise tests by standard ST-segment depression criteria (72% versus 27%). Kansal et al8 found that failure to achieve target HR during exercise correlated with both the presence and severity of CAD and was independently associated with the presence of disease when exercise variables were analyzed with the use of a multivariate discriminant function. When exercise HR response was adjusted for workload achieved in a study of 312 patients by Wiens et al,9 all 13 patients who had no significant ST-segment shift but had an inappropriately low HR response for the workload performed had CAD. In a study of 475 consecutive patients who underwent exercise ECG and coronary angiography, Brenner et al10 observed that both low peak HR and low percentage of target HR achieved were independent predictors of the presence and severity of CAD. However, the chronotropic response index calculated at the end of stage 2 of the Bruce protocol predicted severe CAD only.10
Independent of ST-segment depression findings, chronotropic incompetence has also been shown to be a predictor of subsequent coronary heart disease events11 12 13 and of total mortality.13 In a retrospective review of 2700 subjects who had undergone exercise testing, Ellestad and Wan11 found that subjects with a control ST-segment response but a blunted HR response to exercise had the same incidence of coronary events as patients with ischemic ST-segment changes on their exercise ECGs. In a study of 1472 subjects who underwent both exercise testing and coronary angiography, McNeer et al12 observed a low maximum HR achieved to be associated with decreased 48-month survival, independent of the presence of ischemic ST-segment depression on the exercise ECG. More recently, Lauer et al13 found that failure to achieve target HR, a smaller increase in HR with exercise, and an abnormal chronotropic response index (calculated at stage 2) were each predictive of total mortality and incident coronary heart disease in 1575 male participants in the Framingham Offspring Study. Both an attenuated HR response to exercise and the chronotropic response index remained predictive of total mortality, even after adjustment for age, ST-segment response, and traditional CAD risk factors.13
Results of the present study further confirm the diagnostic value of an attenuated HR response to exercise for the presence of CAD, ranging from 66% to 75% for various measures of HR response to exercise (Table 3⇑), with similar sensitivities and overall accuracy of simple exercise duration and peak METs achieved. However, the chronotropic response index, taking into account both fraction of HRR and fraction of MR achieved, could not be calculated at the peak exercise stage necessary for the present study. Analysis of Equation 3 provides insights into the differences in performance between the chronotropic response indexes calculated at peak exercise and at fixed, submaximal exercise. When chronotropic response index is calculated at peak exercise, the fraction of MR becomes 1; as a consequence, only differences in the fraction of HRR are being compared. In contrast, when the chronotropic response index is determined at fixed, submaximal exercise stages, both the fraction of HRR and the fraction of MR will vary. Thus, the diagnostic and prognostic accuracies of the chronotropic response index calculated at submaximal exercise in previous studies10 13 reflect the importance of both the fraction of HRR and the fraction of MR achieved. The present study suggests that the fraction of HRR alone calculated at peak exercise may prove adequate for assessment of chronotropic response.
In comparison to previous reports in which chronotropic response was evaluated independent of ST response13 or combined with ST response in a logistic model,10 the present study demonstrates the diagnostic advantage of simple adjustment of ST-segment depression criteria for an attenuated HR response to exercise. Compared with unadjusted criteria, correction for fraction of HRR achieved improved sensitivity and overall accuracy as measured by ROC curve areas of simple ST-segment depression, the ST/HR index, and the ST/HR slope. The concept of further correcting the ST/HR index and ST/HR slope with an additional HR term is supported by the independent additive power of the fraction of HRR to the ST/HR slope when included in a logistic regression analysis. In addition, improved performance of the ST/HR index and ST/HR slope relative to simple ST-segment depression criteria persisted even after adjusting for chronotropic response and was independent of β-blocker therapy. Of note, the similar sensitivity of the unadjusted ST/HR index (90%) and ST-segment depression adjusted for fraction of HRR achieved (87%) is not surprising since both criteria include a term adjusting ST-segment depression for the total change in HR with exercise.
The improved risk stratification found with HR adjustment of ST-segment depression24 25 taken together with the current findings suggests that correction of the ST/HR index and ST/HR slope for fraction of HRR should further improve risk assessment in large populations. Further evaluation of adjustment of ST-segment depression criteria for chronotropic response in larger study populations and in relation to left ventricular function will be necessary to more clearly delineate the clinical role of these criteria.
Selected Abbreviations and Acronyms
|CAD||=||coronary artery disease|
|HRR||=||heart rate reserve|
|MPHR||=||maximal age-predicted heart rate|
|ROC||=||receiver operating characteristic|
- Received March 27, 1996.
- Revision received July 16, 1996.
- Accepted July 31, 1996.
- Copyright © 1996 by American Heart Association
Sheffield LT. The exercise test in perspective. Circulation. 1977;55:681-685.
Goldschlager N, Selzer A, Cohn K. Treadmill stress tests as indicators of the presence and severity of coronary artery disease. Ann Intern Med. 1976;85:277-286.
Hollenberg M, Budge WR, Wisneski JA, Gertz WE. Treadmill score quantifies electrocardiographic response to exercise and improves test accuracy and reproducibility. Circulation. 1980;61:276-285.
Ellestad MH, Wan MKC. Predictive implications of stress testing: follow-up of 2700 subjects after maximal treadmill stress testing. Circulation. 1975;51:363-369.
McNeer JF, Margolis JR, Lee KL, Kisslo JA, Peter RH, Kong Y, Behar VS, Wallace AG, McCants CB, Rosati RA. The role of the exercise test in the evaluation of patients for ischemic heart disease. Circulation. 1978;57:64-70.
Lauer MS, Okin PM, Larson MG, Evans JC, Levy D. Impaired heart rate response to graded exercise: prognostic implications of chronotropic incompetence in the Framingham Heart Study. Circulation. 1996;93:1520-1526.
Detry JMR, Piette F, Brasseur LA. Hemodynamic determinants of exercise ST-segment depression in coronary patients. Circulation. 1970;42:593-599.
Holland RP, Brooks H. Precordial and epicardial surface potentials during myocardial ischemia in the pig: a theoretical and experimental analysis. Circ Res. 1975;37:471-480.
Berenyi I, Hajduczki S, Baszormenyi E. Quantitative evaluation of exercise-induced ST-segment depression for estimation of degree of coronary artery disease. Eur Heart J. 1984;5:289-294.
Deckers JW, Rensing BJ, Tijssen JG, Vinke RVH, Azar AJ, Simoons ML. A comparison of methods of analysing exercise tests for diagnosis of coronary artery disease. Br Heart J. 1989;62:434-444.
Kligfield P, Ameisen O, Okin PM. Heart rate adjustment of ST-segment depression for improved detection of coronary artery disease. Circulation. 1989;79:245-255.
Okin PM, Kligfield P. Gender-specific criteria and performance of the exercise electrocardiogram. Circulation. 1995;92:1209-1216.
Okin PM, Anderson KM, Levy D, Kligfield P. Heart rate adjustment of ST-segment depression: improved risk stratification in the Framingham Offspring Study. Circulation. 1991;83:866-874.
American College of Sports Medicine. Guidelines for Graded Exercise Testing and Exercise Prescription. Philadelphia, Pa: Lea & Febiger; 1988:138-140.
Vecchio TJ. Predictive value of a single diagnostic test in unselected populations. N Engl J Med. 1966;274:1171-1173.
Metz CE, Wang P, Kronman HB. A new approach for testing the significance of differences between ROC curves measured from correlated data. In: Deconick F, ed. Information Processing in Medical Imaging. The Hague: Martinus Nijhoff Publishing; 1984:432-445.