Exercise-induced ST depression in an asymptomatic population without coronary artery disease.

Abstract Objectives. Exercise electrocardiogram (ExECG) in low risk populations frequently generates false positive ST depression. We aimed to characterize factors that are associated with exercise-induced ST depression in asymptomatic men without coronary artery disease. Design. Cycle ergometer exercise tests from 509 male firefighters without imaging proof of significant coronary artery disease were analysed. Analysed test data included heart rate at rest before exercise, and workload, blood pressure, heart rate, ST depression and ST segment slope at peak exercise. ST depression of >0.1 mV was considered significant (STdep). With a mean follow-up of 6.1 ± 1.7 years, medical records were reviewed for cardiovascular diagnoses, hyperlipidemia and diabetes. Logistic regression analysis was used for risk assessment. Results. In total, 22% had STdep in ≥1 lead. Subjects with STdep were older than those with normal ExECG (p < .001). Downsloping STdep was more common in extremity leads (9%) than in precordial leads (2%). STdep was categorized according to location (precordial/extremity) and slope direction into eight categories. Larger age-adjusted heart rate increase predicted STdep in seven categories (p < .05). Age-adjusted peak heart rate correlated with STdep in five categories, predominantly where the ST slope was positive. Peak blood pressure and exercise capacity were both associated with STdep in few categories. We found no association between STdep and hypertension, hyperlipidemia or diabetes (all p > .05). Conclusions. In asymptomatic men with a physically demanding occupation and no coronary artery disease, both age and heart rate response were associated with ST depression, whereas common cardiovascular risk factors, blood pressure response and exercise capacity were not.


Introduction
Exercise testing has the dual purpose of assessing physical capacity and evaluating cardiovascular health. In the periodic evaluation of Swedish firefighters, exercise ECG (ExECG) to maximal exhaustion is mandatory, with the aim to detect subclinical signs of cardiovascular disease [1]. The aerobic capacity of firefighters is higher than in the general population. They are also disqualified from duty if their medical health deteriorates. Therefore a firefighter cohort presumably consists of individuals with better than average health and a lower than average risk for disease. However, we have previously found the prevalence of exercise-induced ST depression -one of the cornerstones in ExECG interpretation for identification of coronary artery disease (CAD) -to be around 20% in a firefighter cohort [2]. Yet the IHD event rate was only 2% during follow-up, which illustrates the interpretational challenge in differentiating true from false positive ST changes in this firefighter cohort as well as in any asymptomatic population with low pre-test probability [3].
The clinical significance of exercise-induced ST segment deviations in apparently healthy subjects has been under debate for several decades. An increase in long-term risk for myocardial infarction or future need for coronary artery bypass grafting has been seen with risk factor adjusted ST depression in healthy middle-aged men, acknowledging a possible prognostic value [4]. Nevertheless, due to the low prevalence of disease in asymptomatic populations, the positive predictive value of a positive test response in low-risk individuals is poor, which limits the diagnostic value [5]. Multiple studies have evaluated both the diagnostic and prognostic capacity of ST depression in relation to CAD [6,7]. The cause of ST depression in the absence of CAD as well as its relation to cardiovascular risk factors and testderived variables such as heart rate (HR), blood pressure (BP) and exercise capacity, remains to be further investigated.

Aim
We aimed to evaluate factors that characterize and potentially contribute to exercise-induced ST depression in asymptomatic men without objective evidence of CAD. Bearing those factors in mind during ExECG in low-risk individuals, an improved understanding of the test results might be possible.

Study population
Screening ExECGs performed 2004-2010 were previously studied in a cohort of 521 male firefighters, analysing their ECG response in relation to the incidence of CAD during follow-up. The recruitment procedure has been described in detail elsewhere [2]. The follow-up included review of medical records until 2015, in search for 1) cardiac imaging studies (coronary angiography, myocardial scintigraphy, cardiac magnetic resonance imaging and coronary computed tomography), and 2) registered diagnoses related to the circulatory system according to the International Classification of Diseases coding (I00-I99), as well as diabetes (E10, E11, E14) and hyperlipidemia (E78) [8]. Each diagnose was registered the first time it occurred for each individual.
From the initial cohort, 12 subjects were identified with ischemic heart disease based on significant stenosis at angiography or exercise-induced ischemia at scintigraphy, and those were excluded from further analysis in the present study. No additional cases of acute myocardial infarction or sudden cardiac death were found. Subjects assigned the diagnosis "chronic ischemic heart disease" (I25) were not excluded if both history and imaging were negative. We thereby intended to reduce work-up bias and avoid incorrect CAD classification in asymptomatic firefighters referred for cardiac evaluation due to ST depression at the screening ExECG. The study sample hence consisted of 509 male firefighters without imaging or diagnostic proof of significant CAD neither before the analysed test nor during 6.1 ± 1.7 years follow-up.

Cycle ergometer test
A PC-based ExECG system (Welch Allyn Cardioperfect 1.6.3) was used for all tests. This system was connected to an ergometer bicycle (Monark E839, Vansbro Sweden) to control load, and to the wearable ECG recorder. The exercise tests were carried out as incremental ramp tests with a continuous increase by 1 W every 3 seconds until exhaustion, either starting with 6 minutes at 200 W or 250 W followed by immediate conversion to ramp test, or as ramp protocol during the entire test.
Workload and 12-lead ECG were monitored continuously during all tests and BP was measured every 3 minutes. The same cardiologist supervised all tests. Firefighters with exercise tests abnormalities such as arrhythmias, pathological BP response and ST segment deviations, were routinely referred for further cardiologic evaluation.

Exercise ECG variables
Tests where the ECG had abnormalities that could hamper the interpretation of the ST analysis were excluded in a previous test selection step [2]. Exclusion criteria for resting ECGs were bundle branch block and software-identified left ventricular hypertrophy, if manual assessment verified secondary ST-T changes. Tests with exercise-induced arrhythmia or intermittent conduction disorders were also excluded, together with tests performed by pacemaker carriers, and tests with insufficient signal quality. The total exclusion rate from the main firefighter study database was 5%.
The most recent test was selected if a subject had performed multiple ExECGs during the inclusion period. From the selected tests, peak workload (P peak ; W) and peak BP (BP peak ; mmHg) were registered; the latter refers to the latest BP measured during exercise. BP peak lower than two standard deviations below the average BP peak for the cohort were manually checked and excluded if registered >3 minutes before end of exercise. From the 12-lead ECG, ST data were analysed in all leads except V1 and aVL. ST values (60 milliseconds from the J-point, ST60; lV) were collected from rest and from peak exercise, and was calculated as the median value of the first three measurements during sitting rest and the last three during pedalling respectively. ST depression was defined at peak exercise. If ST60 was negative at rest, ST depression was calculated as additional negative deflection at peak exercise compared to rest. Also, the slope of the ST segment and peak HR (HR peak ; beats/min) were evaluated at peak exercise. ST depression !0.1 mV was considered significant (ST dep ) and the ST slope was classified as downsloping if 0 lV/s. The HR span (HR span ; beats/min) was calculated as difference between HR peak and HR at rest sitting on the bicycle. We analyzed target HR achievement as percentage of age-predicted maximum HR according to the Tanaka formula [9].
Statistics SPSS statistical software (IBM SPSS Statistics for Windows, Version 23.0. Armonk, NY, USA) was used for analyses. Student's t-test was used for comparison of continuous data and Chi-squared test was used for comparison of categorical data. Binary logistic regression analysis for single and multiple variables was used to assess association between cardiovascular diagnoses, test derived variables and exerciseinduced ST dep . Odds ratios (OR) with 95% confidence interval (CI) were reported from the binary logistic regression analysis.

Results
We studied 509 asymptomatic male firefighters with no history of myocardial infarction. The average exercise capacity was 279 ± 39 W, 97% reached !85% of age-predicted maximum HR. None of the test subjects reported exertional chest pain. Anthropometrics of the study sample, and basic test data are presented in Table 1.
In this cohort, 8% had gone through cardiac imaging studies during the inclusion period or follow-up, with negative result for exercise-induced ischemia or with none or less than 50% coronary artery lumen reduction. The remaining firefighters had not presented clinical symptoms or exercise test abnormalities that motivated additional imaging studies, neither earlier in their career nor during the follow-up period of 6.1 ± 1.7 years. During follow-up there were six deaths, none of them due to CAD.
According to medical records, 7% of subjects had either a cardiovascular diagnosis, hyperlipidemia or diabetes registered prior to the examined test. Diagnoses of established risk factors for CAD (hypertension, hyperlipidemia and diabetes) were all infrequent. Hypertension was diagnosed in 2% while the diagnoses diabetes mellitus and hyperlipidemia occurred in 1% each. None of the other identified diagnoses was more frequent (Appendix 1).

Prevalence and characteristics of ST depression
At the end of exercise, ST dep in at least one lead was present in 22%. The slope of the ST segment and the location of the affected lead(s) were used for further sub-classification of the ECG response into eight categories, each labelled with a Roman numeral, I-VIII ( Figure 1).
In 15%, ST dep was found in a precordial lead and in 11%, ST dep was present in two or more consecutive precordial leads. A negative slope of precordial ST dep was however less frequent than 2%. ST dep in one or two leads was less common in extremity leads than in precordial leads, but extremity lead ST dep more often had a negative slope (Figure 1).
One out of ten subjects (10%) had at least one lead with downsloping ST dep by the end of exercise.
Subjects with ST dep in any lead at the end of exercise were significantly older compared to those who did not develop ST dep (49 ± 9 vs. 45 ± 11 years, p < .001). Similarly, subjects who developed ST dep in any precordial lead or in two consecutive precordial leads were significantly older compared to those who did not (p < .001). Among all other categories, the mean age of subjects who developed ST dep was consistently higher than in those who did not, although not significantly so (Figure 2).
Age was an independent predictor of any-lead ST dep as well as one-and two-lead precordial ST dep (all OR > 1.0, p < .001).

Test related variables and ST depression
Relations between test-derived parameters and exerciseinduced ST dep were analysed for precordial ( Table 2) and extremity (Table 3) leads. A large HR span predicted ST dep in the majority of ST depression categories, both in absolute numbers and after adjustment for age, p < .05 or lower. High age-adjusted HR peak was also associated with ST dep in several categories, whereas unadjusted HR peak consistently lacked such association.
In general, low BP peak tended to be associated with ST dep , but risk estimates were not significant. However, in precordial leads there was a significant association between high end-exercise BP and downsloping ST dep (Table 2).
Low end-exercise workloads were associated with precordial ST dep , but it was not significantly so after adjustment for age (Table 2).
Multivariable analysis, including combination of factors did not add significant predictors. BMI: body mass index; P peak : peak workload; HR peak : peak heart rate; %HR pred : achievement of age-predicted maximum HR; BP peak : peak blood pressure. Downsloping ST dep in at least one lead, independent of its location, was significantly associated with age-adjusted high P peak (OR 1.01, CI 1.00-1.02), high HR peak (OR 1.03, CI 1.00-1.06) as well as a large HR span (OR 1.03, CI 1.01-1.05), all p < .05.

Cardiovascular risk factors
Neither diagnosed hypertension, hyperlipidemia, nor diabetes was associated with exercise-induced ST dep in this study sample, for any of the ST dep categories, neither raw nor age-adjusted. Analysis of combination of risk factors did not add significant predictors either.

Discussion
Exercise-induced ST depression of at least 0.1 mV in one or several leads was found in every five subjects in this cohort of asymptomatic firefighters with no CAD. In about 50% of these, the ST dep was horizontal or downsloping. Both age and HR response were associated with ST dep , whereas common cardiovascular risk factors, BP response and exercise capacity were not.
In a clinical setting, evaluation of the exercise test usually includes a compound of exercise capacity, BP and HR responses, subjective symptoms and ECG analysis during exercise and recovery. Out of those, the ST segment reaction is the most important and objectively verifiable parameter for evaluation of ischemic heart disease, and by far the most assessed parameter in studies of the diagnostic accuracy of the ExECG. Exercise-induced ST depression in the absence of ischemic heart disease is a false positive test result. The present study was designed to characterize exercise-induced ST depression in subjects without CAD.  Roman numerals refer to patterns of ST depression as categorized in Figure 1. Only pure ramp tests are included in the analysis of P peak . HR peak : peak heart rate; HR span : heart rate span between sitting rest and peak exercise; BP peak : peak blood pressure; P peak : peak workload. Ã p < .05; ÃÃ p < .01; ÃÃÃ p < .001.
In asymptomatic subjects and athletes, screening with ExECG typically yields a relatively low frequency of positive test results. In a review analysing athletes, the mean prevalence of positive ExECG (defined as 0.1 mV horizontal or downsloping ST depression, in most studies analysis restricted to precordial leads) was 5% in a subgroup of athletes aged 35-60 years. The calculated positive predictive value (PPV) was poor (6%), reflecting a low prevalence of underlying cardiac disease [10]. The prevalence of horizontal or downsloping ST depression in the present study was similar, 2% in precordial leads and 10% when all leads were considered. Although firefighters are not athletes by definition, physical training is necessary to maintain the required aerobic capacity. Therefore, the fitness of firefighters is generally better than in the general population and comparison to recreational athletes is motivated. In fact, smaller studies comparing master athletes and sedentary subjects have indicated a higher incidence of false positive ST depression in athletes [11,12]. The underlying mechanisms are not fully understood, but increased left ventricular mass with secondary repolarization abnormalities has been suggested [10].

Heart rate
We found that a large HR span was associated with ST dep . In healthy subjects, a progressive decrease of the J-point amplitude is often seen with increasing exercise, concurrent with a marked rise of the ST segment slope [13]. At high HR, depression of the ST junction may be due to atrial repolarization occurring slightly after ventricular depolarization, instead of simultaneously, thus coinciding with the beginning of the ST segment, with a depressing effect. An isolated rapidly upsloping ST segment (without significant ST depression at ST80), has even shown to be associated with reduced long-term risk for ischemic heart disese compared to a normal exercise ST segment morphology [14].
Interestingly, unadjusted HR peak per se was not associated with ST dep in this study, but only so after adjustment for age. Healthy subjects with a high HR peak 'despite' higher age, might hence be more prone to develop ST dep than those who reached a lower HR peak . Speculatively, this might be due to a reduced coronary flow reserve at peak exercise with high HR in the older subjects. In young subjects, high HR was not as likely to explain any ST dep as in older individuals.

Cardiovascular risk factors
Firefighters who presented with precordial ST dep in one or several leads were significantly older than their colleagues who did not. That may indicate that pathological processes accumulate over time and contribute to an abnormal ST segment response, although not necessarily caused by significant luminal reduction of the coronary arteries. In this cohort, neither hyperlipidemia, diabetes, hypertension nor combinations of them were significantly associated with ST dep , suggesting that other mechanisms could be relevant in this type of population.
In one study of patients with exertional angina but angiograms with no or less than 50% stenosis, a pathological ExECG was more common in those who had coronary microvascular dysfunction [15]. Although coronary microvascular dysfunction is a phenomenon more often seen in women [16], it could be a possible cause of ST segment depression in the asymptomatic subjects in our cohort, however microvascular dysfunction was not assessed with the main imaging modalities analyzed during follow-up in this study [17].
Massie et al. found that exercise-induced ST depression in the absence of CAD was more common in hypertensive men than in normotensive, risk factor-matched controls, especially in case of left ventricular hypertrophy [18]. That study was conducted in high risk men, as opposed to the Table 3. Association between test-derived variables and !0.1 mV ST depression in extremity leads, expressed as odds ratio (OR) with 95% confidence interval (CI Roman numerals refer to patterns of ST depression as categorized in Figure 1. Only pure ramp tests are included in the analysis of P peak . HR peak : peak heart rate; HR span : heart rate span between sitting rest and peak exercise; BP peak : peak blood pressure; P peak : peak workload. Ã p < .05; ÃÃ p < .01; ÃÃÃ p < .001. low prevalence of diagnosed cardiovascular risk factors in our cohort, which may explain why no such association was seen in our cohort. ST depression in patients with severe left ventricular hypertrophy has been explained by e.g. reduced coronary artery flow reserve [19]. An exaggerated BP response during exercise is a known risk factor for future resting hypertension in healthy asymptomatic persons [20,21]. A hypertensive response at moderate exercise intensity is furthermore shown to be an independent risk factor for cardiovascular events and mortality in patients without significant CAD [22]. At the same time, athletes tend to reach higher peak exercise BP compared to healthy controls [11]. Thus, high end-exercise BP can be seen as both a presumably healthy pattern and one that is not. A high BP peak was in our study associated with ST dep in two of the sub classification categories, including precordial lead(s) downsloping ST dep ( Table 2).

Other causes to ST depression
In addition to ischemia and parameters already mentioned above, some additional factors not assessed in this study have also been suggested to affect the ST segment response, including hyperventilation, serum electrolyte concentrations, anaemia and certain therapeutic drugs (digoxin, tricyclics, antiarrhythmic drugs etc.) [23].

Limitations
The current study is based on data from a clinical database, following clinical guidelines for exercise testing but not strictly adhered to a study protocol. This might have implications on the collection of data, such as the time point for measurement of BP peak and how close to maximal exhaustion the tests are driven. Medical record derived diagnoses for hypertension, diabetes and hyperlipidemia instead of measurements/blood samples could potentially have underestimated the actual frequency of disease.
As in nearly all studies on myocardial ischemia, coronary angiography was not done in all subjects in our cohort. Instead, referral was most likely related to the result of the ExECG, hence constituting a possible work-up bias. To reduce its potential effect we have included cardiac imaging studies from a longer follow-up and not only directly related to the ExECG.
The associations between ST dep and test related variables found in the logistic regression analysis in the present study were, although statistically significant, in general modest.

Conclusion
Age-adjusted HR peak was associated with end-exercise ST dep in physically active men without CAD. Notably, neither high BP response nor established cardiovascular risk factors were associated with ST dep in this cohort. Whether the effect of age and HR on exercise-related changes is electrophysiological or secondary to microvascular or structural myocardial variations remains to be further elucidated.