Stress-induced Worsening of Left Ventricular Diastolic Function as a Marker of Myocardial Ischemia.

BACKGROUND: Echocardiography has been the subject of interest in diagnosing diastolic dysfunction and estimating left ventricular filling pressures (LVFPs). The present study is set to estimate the correlation between the worsening of diastolic parameters and the evidence of inducible ischemia during an exercise stress echocardiography (SE) in comparison with the results of coronary computed tomographic angiogram (CCTA).
METHODS: A total of 191 consecutive patients from the executive screening program who underwent exercise SE followed by CCTA were evaluated. Baseline demographics, heart rate, and blood pressure measurements were extracted for analysis. Standard two-dimensional and tissue Doppler imaging parameters were analyzed. Diastolic function was graded at rest and peak exercise.
RESULTS: Patients who had worsening of diastolic function by at least one grade had had 2-3-fold higher odds of having abnormal SE. In addition, patients with worsening of diastolic function had higher stress LVFPs (E/e' = 11.7 ± 2.7 vs. E/e' 8.0 ± 2.0; P < 0.0001), more E/e' change >25% (48% vs. 22%, P = 0.012), and were more likely to have obstructive coronary artery disease (CAD) on CCTA (23.8% vs. 9.2%; P = 0.045). A change in E/e' >25% (stress-rest) was highly associated with a positive stress test and abnormal CCTA result. Patients with no change or improvement in diastolic function with stress had a 92% negative predictive value of having normal SE and 91% of normal/nonobstructive CCTA.
CONCLUSION: A worsening of diastolic function and a change in E/e' >25% (stress-rest) were associated with abnormal SE, positive stress test, and obstructive CAD when compared to CCTA results.

Introduction

Stress echocardiography (SE) is one of the most commonly performed tests to assess for inducible myocardial ischemia by detecting stress-induced regional wall motion abnormalities (WMAs).[1] Although the focus in SE has been predominantly on detecting systolic WMAs, there has been a focus shift toward diastolic function and the change of myocardial relaxation which precedes systolic dysfunction in the ischemic cascade.[23] Indeed, myocardial relaxation is an adenosine triphosphate-dependent process; when supply–demand mismatch occurs, there is worsening of diastolic function and elevation of filling pressures followed by regional WMAs.[456]

Therefore, the detection of new or worsening diastolic dysfunction (DD) on stress testing is of interest as it may potentially improve the sensitivity (SN) of SE.[23] Our study aims at assessing the feasibility and added value of left ventricular (LV) diastolic function assessment at rest and peak exercise as a more objective echocardiographic tool for the early diagnosis of stress-induced myocardial ischemia.[5] This noninvasive tool is practical, less user dependent, adds minimal processing time, and does not require intravenous contrast or radiation as compared to coronary computed tomographic angiography (CCTA).

Methods

Study population

A total of 210 consecutive outpatients presenting for the executive screening program at Clemenceau Medical Center (CMC), Beirut, Lebanon, between January 2010 and December 2014, were identified. Patients older than 75 years, with abnormal resting left ventricular ejection fraction (LVEF), recent myocardial infarction within the last 6 months, significant valvular disease, arrhythmia, left bundle branch block, congenital heart disease, pericarditis of any etiology, myocarditis, and previous cardiac surgery were excluded from the study (n = 11) [Table 1]. There were eight other patients who were excluded from the study because of incomplete echocardiographic measurements and suboptimal images, leaving 191 patients for the final analysis. All patients underwent SE followed by CCTA.

Table 1

Inclusion and exclusion criteria of patients enrolled

Inclusion criteria

Patients between 18 and 75 years of age

Normal or preserved LVEF defined as LVEF ≥55%

Exclusion criteria

Recent history of myocardial ischemia within last 6 months prior to enrollment

Significant valvular heart disease defined as moderate or severe mitral or aortic valve regurgitation or stenosis

Hypertrophic cardiomyopathy

Acute pericarditis/constrictive pericarditis

Hemodynamic instability of any cause

Nonsinus rhythm

LBBB on the ECG

Acute myocarditis

Congenital heart disease

Acute or subacute endocarditis

Previous cardiac surgery

LBBB: Left bundle branch block, ECG: Electrocardiogram, LVEF: Left ventricular ejection fraction

Demographics and comorbidities were prospectively entered at the time of testing and were subsequently retrieved for analysis. The study was approved by the Institutional Board Review committee and complied with the Declaration of Helsinki.

Hemodynamics

Resting and peak stress heart rate (HR) and blood pressure were recorded prospectively and subsequently retrieved from the database.

Resting and stress echocardiography

The SE study consisted of performing a full resting transthoracic echocardiogram followed by exercise stress and then reacquiring images at peak stress.

The resting transthoracic echocardiogram was performed with the patient in the left lateral decubitus position using the commercially available machine (GE, Vivid E9 Vingmed Ultrasound, Horten, Norway) with the M5Sc-D probe. The images were recorded and saved on the machine, and off-line analyses were performed using EchoPAC software (GE Medical Systems, Model BT10, Horten, Norway). Image quality was labeled as good, fair, or poor.

LVEF was assessed in semi-quantitative manner using the biplane Simpson method (for most patients) or Teichholz's method (for those with suboptimal images) in case of regional WMAs as previously described.[7] Diastolic function was assessed in our institution in a standardized method and in accordance with the most recently published guidelines by the European Association of Cardiovascular Imaging using a combination of echocardiographic variables (transmitral inflow pattern, mitral annular velocities with tissue Doppler imaging, left atrial volume index [LAVI], and pulmonary venous flow pattern).[7] Two Level III trained echocardiography-certified cardiologists (EC, WAJ) reviewed all cases and graded the diastolic function (rest and stress) in a blinded manner; in case of a discrepancy, the images were reviewed by a third cardiologist. Diastolic function was labeled as normal or abnormal. DD was then categorized as mild (Grade 1, impaired relaxation), moderate (Grade 2, pseudonormal), or severe (Grade 3, restrictive).[4] A change in E/e’ (stress-rest) >25% was considered abnormal. LAVI was measured in accordance with the published guidelines.[7]

Five views were obtained during image acquisition: Parasternal long axis, parasternal short axis, apical four chamber, apical two chamber, and apical three chamber. The whole echocardiographic study was performed by the same physician and in the same echocardiography laboratory at CMC hospital.

Diastolic stress echocardiography and stress imaging

After obtaining rest images, exercise testing was performed according to a multistage, variable load, upright bicycle ergometer starting by a workload of 25 W and increasing by an increment of 25 W for every 3 min as previously published.[8]

The images were acquired at peak stress imaging within 1 min with standardized views according to the published guidelines.[8] Transmitral inflow pattern as well as tissue Doppler imaging parameters were also recorded at peak stress. SE was considered normal (no inducible ischemia) or abnormal (new regional WMAs in one or more segments).

Coronary computed tomographic angiography

CCTA was performed as a standard test on all patients for the diagnosis of coronary artery disease (CAD), using 64-slice GE Discovery 750 HD GSI scanner according to CMC protocol and as previously published.[9] All coronary scans were performed after the SE, during the same day.

Patient preparation and premedication

Resting HR was measured for all patients who underwent CCTA before the procedure with a target HR ≤60 bpm for optimal images. Accordingly, patients were prepared and often required the administration of beta-blockers with or without ivabradine.

Image acquisition and electrocardiogram gating

The CCTA scan was an electrocardiogram (ECG)-gated retrospective study performed similar to a previously published protocol.[9] Images were reconstructed using continuous acquisition of CCTA data throughout the cardiac cycle. The use of retrospective ECG-gated reconstruction allowed image reconstruction at different points of the R-R interval, allowing to choose the optimal cycle for image reconstruction.

Image analyses

CCTA was considered normal, non obstructive CAD (plaque with <70% stenosis in a major epicardial vessel or <50% stenosis in the left main coronary artery), or obstructive CAD (plaque with ≥70% stenosis in a major epicardial vessel or ≥50% stenosis in the left main coronary artery).

Statistical analysis

Continuous variables were expressed as means (standard deviation) and compared with the unpaired Student's t-test or Wilcoxon rank test as appropriate. Categorical variables were expressed as frequency (percentages) and compared with Fisher's exact test or Pearson's Chi-square test as appropriate.

Multivariate regression analysis model was performed to determine the independent predictors of abnormal SE. Significant univariates (P < 0.1) or clinically relevant ones were entered into the model. Variables with collinearity were entered into the model one at a time. The final model was adjusted for age, gender, body mass index (BMI), cerebrovascular disease (CVD), diabetes mellitus (DM), hypertension (HTN), dyslipidemia (DL), smoking, resting HR, ejection fraction, image quality, and change in LV filling pressure (LVFP). For practical and more clinically relevant interpretation, significant worsening of LVFP was defined as an increase (stress-rest) by 25% or more. The incremental value of worsening of LVFP was tested using the change in Chi-square test from model 1 (without E/e’) to model 2 (with E/e’) with one degree of freedom. Similarly, another model was performed using worsening of diastolic function by ≥1 grade instead of LVFP. LAVI and other diastolic parameters were not entered into the models because of confounding effect with E/e’. For better illustration, a predicted probability plot was performed using the beta coefficients from the multivariate regression analysis to represent the probability of abnormal SE as a function of age and stratified by E/e’.

The diagnostic accuracy of SE was compared against CCTA using 2 × 2 tables for all patients and further stratified by image quality of the stress study. The SN, specificity (SP), positive predictive value (PPV), negative predictive value (NPV), and diagnostic accuracies were derived from the tables as previously described.

All statistical tests were two sided. P < 0.05 was set a priori and considered statistically significant. All statistical analyses were performed with the SPSS Statistics version 22 (IBM, Inc., Armonk, NY, USA).

Results

Baseline characteristics

The cohort consisted of 191 patients, with mean age of 52 ± 12 years, 17.3% of females, and 19.9% with DM [Table 2]. Sixty-four patients (33%) had an abnormal SE; they were older with more comorbidities as compared to those with normal SE [Table 2].

Table 2

Baseline characteristics stratified by stress echocardiography results

Variable	All patients (n=191)	Normal SE (n=127)	Abnormal SE (n=64)	P
Demographics
Age, years (SD)	52 (12)	47.0 (10.1)	61.8 (10.3)	<0.0001
Female gender (%)	33 (17.3)	23 (18.1)	10 (15.6)	0.67
BMI, kg/m2 (SD)	29 (4.9)	28.0 (4.2)	31.3 (5.5)	<0.0001
Comorbidities (%)
CVD	26 (13.6)	4 (3.1)	22 (34.4)	<0.0001
DM	38 (19.9)	12 (9.4)	26 (40.6)	<0.0001
HTN	63 (33)	23 (18.1)	40 (62.5)	<0.0001
DL	72 (37.7)	41 (32.3)	31 (48.4)	0.030
Smoking history	123 (64.4)	80 (63)	43 (67.2)	0.57
Hemodynamics
HR (rest), bpm (SD)	69.3 (10.5)	68.2 (9.4)	71.3 (12.1)	0.058
HR (peak), bpm (SD)	141 (18)	146 (16)	131 (18)	<0.0001
SBP (rest), mmHg (SD)	117 (12)	115 (11)	122 (13)	<0.0001
SBP (stress), mmHg (SD)	166 (17)	164 (16)	171 (17)	0.009

BMI: Body mass index, CVD: Cerebrovascular disease, DM: Diabetes mellitus, HTN: Hypertension, DL: Dyslipidemia, SD: Standard deviation, HR: Heart rate, SBP: Systolic blood pressure, SE: Stress echocardiography

Echocardiography parameters

The echocardiography parameters stratified by the type of stress test results are summarized in Table 3. There was no significant difference in image quality, LVEF, or LAVI between both groups [Table 3]. Patients with abnormal SE had higher prevalence of baseline DD and stress-induced worsening of diastolic function [Table 3].

Table 3

Echocardiography parameters stratified by stress echocardiography results

Variable	All patients (n=191)	Normal SE (n=127)	Abnormal SE (n=64)	P
Image quality (%)
Good	29 (15.2)	17 (13.4)	12 (18.8)	0.60
Fair	116 (60.7)	78 (61.40)	38 (59.4)
Bad	46 (24.1)	32 (25.2)	14 (21.9)
LVEF (%) (SD)	65 (7)	65 (7)	64 (8)	0.31
LAVI, ml/m2 (SD)	21 (9)	21 (9)	23 (10)	0.23
Diastolic parameters (rest)
E, cm/s (SD)	75 (17)	78 (14)	68 (20)	<0.0001
e’ lateral, cm/s (SD)	10 (3)	11 (3)	8 (2)	<0.0001
E/e’ (SD)	7.9 (2.3)	7.5 (1.9)	8.8 (2.8)	<0.0001
Diastolic parameters (peak stress)
E, cm/s (SD)	100 (25)	103 (24)	96 (28)	0.096
e’ lateral, cm/s (SD)	12 (3)	13 (3)	10 (3)	<0.0001
E/e’ (SD)	8.4 (2.3)	7.9 (2.0)	11.7 (2.7)	<0.0001
Resting diastolic function (%)
Normal	122 (63.9)	116 (91.3)	6 (9.4)	<0.0001
Mild, Grade 1 DD	60 (31.4)	9 (7.1)	51 (79.7)
Moderate or severe, Grade 2-3	9 (4.7)	2 (1.6)	7 (10.9)
Stress diastolic function (%)
Normal	144 (75.4)	115 (90.6)	29 (45.3)	<0.0001
Mild, Grade 1 DD	28 (14.7)	5 (3.9)	23 (35.9)
Moderate or severe, Grade 2-3	19 (9.9)	7 (5.5)	12 (18.8)
Stress-induced worsening of diastolic function (%)	21 (11)	10 (7.9)	11 (17.2)	0.052
Increase in E/e’ ≥25% (stress-rest) (%)	48 (25.1)	31 (24.4)	17 (26.6)	0.75

LVEF: Left ventricular ejection fraction, DD: Diastolic dysfunction, SD: Standard deviation, SE: Stress echocardiography

On the other hand, patients with worsening diastolic function had higher stress filling pressures (11.7 ± 2.7 vs. 8.0 ± 2.0; P < 0.0001) and more E/e’ change >25% (48% vs. 22%, P = 0.012). The SN, SP, PPV, and NPV of worsening diastolic function as a predictor of abnormal SE were 52%, 69%, 17%, and 92%, respectively.

Independent predictors of abnormal stress echocardiography

Uni- and multi-variate models were performed to identify the independent predictors of abnormal SE. After multivariate adjustment, advanced age, increased BMI, elevated resting HR, history of CVD, and worsening of LVFP (ΔE/e’ ≥25% [stress-rest]) were associated with increased odds of having abnormal SE. Indeed, an increase in E/e’ by >25% was associated with almost 3-fold increased odds of positive stress test (odds ratio: 3.11 [1.00–9.73], P = 0.051). On the other hand, a poor image quality was associated with lower odds of having abnormal stress test and more likely a normal (or false negative) test [Table 4]. In addition, stress-induced worsening of LVFP added incremental value with Chi-square model increase from 122.1 to 126.1 (P = 0.045).

Table 4

Uni- and multi-variate model predicting abnormal stress echocardiogram

Variable	Univariate		Multivariate
	OR (95% CI)	P	OR (95% CI)	P
Age, years	1.15 (1.10-1.20)	<0.0001	1.20 (1.122-1.286)	<0.0001
Gender, female	0.84 (0.37-1.89)	0.67	0.64 (0.15-2.64)	0.53
BMI, kg/m2	1.16 (1.08-1.24)	<0.0001	1.20 (1.07-1.35)	0.003
Resting HR, bpm	1.028 (1.00-1.06)	0.062	1.05 (1.01-1.10)	0.026
CVD	16.10 (5.25-49.4)	<0.0001	7.98 (1.53-41.7)	0.014
DM	6.56 (3.02-14.25)	<0.0001	2.30 (0.74-7.19)	0.15
HTN	7.54 (3.82-14.85)	<0.0001	2.24 (0.78-6.43)	0.13
DL	1.97 (1.07-3.65)	0.031	0.546 (0.18-1.63)	0.27
Smoking history	1.20 (0.64-2.27)	0.57	1.57 (0.52-4.72)	0.42
LVEF	0.98 (0.94-1.02)	0.31	0.98 (0.91-1.05)	0.48
Image quality
Good	Reference
Fair	0.69 (0.30-1.60)	0.38	0.37 (0.077-1.77)	0.21
Poor	0.62 (0.24-1.64)	0.33	0.17 (0.029-1.06)	0.057
ΔE/e’ ≥25% (stress-rest)	1.20 (0.56-2.23)	0.75	3.11 (1.00-9.73)	0.051

OR: Odds ratio, CI: Confidence interval, BMI: Body mass index, CVD: Cerebrovascular disease, DM: Diabetes mellitus, HTN: Hypertension, DL: Dyslipidemia, LVEF: Left ventricular ejection fraction, HR: Heart rate

Worsening of diastolic function as a predictor of abnormal stress echocardiography

Patients with obstructive CCTA had more worsening of diastolic function (23.8% vs. 9.2%; P = 0.045). Patients with no change or improvement in diastolic function with stress had a 92% NPV of having normal SE and 91% of normal/nonobstructive CCTA.

Diastolic grade as opposed to change in E/e’

Patients with worsening of diastolic function by 1 or more grade had 2–3-fold higher odds of having abnormal SE and obstructive CAD on CCTA.

Probability plot of having abnormal SE stratified by age and by exercise-induced increase in left ventricular filling pressure

Plot was performed using the data from the multivariate regression analysis [Table 4]. The mean values for continuous variable and median values for categorical variables were used (male = 1, BMI = 29.1 kg/m2, CVD = 0, DM = 0, HTN = 0, DL = 0, smoking = 0, LVEF = 65%, resting HR = 69.3 bpm, and image quality = good). Figure 1 shows an increased probability of abnormal study with age. For each age group, the probability significantly increased in E/e’ by at least 25%. At age 65 and higher, however, the curves merge. E/e’ is helpful for younger patients.

Figure 1

Predicted probability of abnormal stress echocardiography.

The effect of image quality on SE result compared to coronary computed tomographic angiogram

With good or fair images obtained during SE, all patients (14) who had a positive SE had a positive CCTA. In other words, none of the patients had a negative SE while they were tested positive for CCTA (SN: 100%). Nevertheless, 118 patients had a negative CCTA. Of them, 87 had a negative SE (SP: 74%). Forty-five patients had a positive SE, but only 14 patients tested positive in CCTA (PPV: 31%). In addition, all patients who had a negative SE had a negative CCTA (NPV: 100%). Diagnostic accuracy was 77%.

With poor quality images, there was a higher chance of obtaining normal SE, i.e., a lower chance for detecting abnormal findings. The SN of the study decreased to 71% (P = 0.10), the NPV decreased to 93% (P = 0.057), and the diagnostic accuracy decreased to 76% as well when CCTA was used as gold standard. However, there was no significant change in SP or PPV. All results are shown in Figure 2.

Figure 2

Plot of 2 × 2 tables of diagnostic accuracy of stress echocardiogram versus coronary computed tomographic angiogram, stratified by image quality of the stress echocardiogram study.

Discussion

We investigated the worsening of diastolic function at exercise and its relationship with myocardial ischemia detected by abnormal SE and obstructive CCTA. Stress-induced worsening of diastolic function is a more reliable and objective marker than nonspecific symptoms to assess patients with myocardial ischemia. Argulian et al.[10] found that, although not specific, dyspnea could be an angina equivalent in patients with known CAD and abnormal diastolic function during stress. Dyspneic patients with known CAD had a higher risk of stress-induced ischemia. However, dyspneic patients without known CAD and normal resting wall motion seemed to have rates of positive stress test results comparable with those of asymptomatic patients, i.e., a decreased likelihood of angina.

We also looked for changes in LVFP between stress and rest and the association with higher grade of DD. E/e’ is accurate and widely used in demonstrating LV end-diastolic pressure. Ha et al.[5] conducted a study to prove the efficacy of diastolic SE in detecting LV delayed relaxation on 45 patients referred for the evaluation of exertional dyspnea. In this series of patients, exercise was the major factor to induce dyspnea and mostly affected the E/e’ ratio. As a consequence, symptoms of DD occurred more commonly during exercise because of decreased diastolic filling and thus increased LVFP. Another study performed on 219 patients by Nakajima et al.[11] validated the importance of our observation. It demonstrated the evidence of an inverse correlation between the number of ischemic segments and peak s’ and e’ velocities; in other words, patients who developed dyspnea and angina had a lower e’ velocity at rest, during dobutamine SE and during recovery, and a higher E/e’ ratio compared with patients who did not develop ischemia. Nevertheless, Podolec et al.[12] designed a study on fifty consecutive patients with severe systolic heart failure to assess DD by measuring E/e’ ratio at baseline and peak stress. Data analysis of this study suggested that elevated LVFP using E/e’ ratio at peak stress was highly indicative of poor exercise performance in patients with ischemic heart failure. Finally, using exercise-induced changes of the E/e’ ratio, Tsougos et al.[13] detected flow-limiting CAD when they studied 359 patients referred for chest pain; ΔE/e’ had a 87.3% SN and 75.2% SP.

In the present study, CAD was evaluated using CCTA as a gold standard test instead of coronary angiography. In a study performed on 73 patients, Leschka et al.[14] proved that 64-slice multidetector CT scan had a high diagnostic accuracy and 94% SN when compared to invasive coronary angiography.

The image quality was also analyzed in correlation with SE results. A better image was strongly associated with a better detection of abnormal SE. This underlines the importance of the study done by Nagel et al.[15] when they showed that the SN of dobutamine stress magnetic resonance was superior to dobutamine SE. The difference was mainly due to moderate echocardiographic images.

Finally, it can be concluded that patients who are older than 65 years are more prone to have abnormal SE. Studer Bruengger et al.[16] concluded that, in healthy young individuals aged <40 years, although E/e’ increased during exercise, it remained normal and did not show a major difference between endurance-trained athletes and nonathletes.

Study limitations

The current study has several limitations. First, although measurements were carried out by two experienced cardiologists and reviewed by a third one in case of discrepancy, potential errors in measuring epicardial and endocardial borders at peak exercise were unavoidable, especially in case of suboptimal image quality. The addition of LV opacification contrast for poor quality images would have been a good alternative, but it was not feasible given its cost and restricted availability.

Second, detailed results of positive SE (i.e., the occurrence of chest pain, ECG changes, and wall motion score indices) were not added to the analysis because they were lacking in the database. Furthermore, frame rate dependency was also a limiting factor in eight patients and its use diminished the accuracy of standard echocardiographic measurements, especially when E and A waves fused at peak stress. For this reason, we attempted to minimize the influence of poor images because they were associated with a worse diagnostic accuracy, especially when the HR at peak exercise was above 140 bpm and when patients were unable to hold their breath due to fatigue.

Third, the mean global LVEF was not different between groups and this is not unusual. Among patients with positive SE, regional WMAs do occur, but the global LVEF which was reported might still be within normal levels, particularly with the remaining hyperdynamic segments. Although there were differences in E/e’ and diastolic grades between normal and abnormal SE results, the change in E/e’ ≥25% (stress-rest) was not strongly significant.

Fourth, statistical analyses did not show that worsening of diastolic function was a predictor of abnormal SE. However, the trend is present and most likely, the study was not powered enough to detect the difference.

Fifth, CCTA was used as gold standard rather coronary angiogram.

Finally, the cohort included patients from a single center with referral and selection bias.

Conclusions

We demonstrate a good agreement between the worsening of LV diastolic function and obstructive CAD on CCTA. SE is a noninvasive method and a good predictor of ischemia, particularly when diastolic function is integrated in the test interpretation. The importance of the change in E/e’ ≥25% and the worsening of diastolic function grade between rest and peak exercise are highly suggestive of positive stress test, increase the likelihood of obstructive CCTA, and should become the standard of practice during routine SE.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.