Left ventricular end‐diastolic volume predicts exercise capacity in patients with a normal ejection fraction

Abstract

Background

Exercise capacity is a powerful predictor of all‐cause mortality. The duration of exercise with treadmill stress testing is an important prognostic marker in both healthy subjects and patients with cardiovascular disease. Left ventricular (LV) structure is known to adapt to sustained changes in level of physical activity.

Hypothesis

Poor exercise capacity in patients with a preserved LV ejection fraction (LVEF) should be reflected in smaller LV dimensions, and a normal exercise capacity should be associated with larger LV dimensions, irrespective of comorbidities.

Methods

This hypothesis was first tested in a cross‐sectional analysis of 201 patients with normal chamber dimensions and preserved LVEF who underwent a clinically indicated treadmill stress echocardiogram using the Bruce protocol (derivation cohort). The best LV dimensional predictor of exercise capacity was then tested in 1285 patients who had a Bruce‐protocol treadmill exercise stress test and a separate transthoracic echocardiogram (validation cohort).

Results

In the derivation cohort, there was a strong positive relationship between exercise duration and LV end‐diastolic volume deciles (r ² = 0.85; P < 0.001). Regression analyses of several LV dimensional parameters revealed that the body surface area–based LV end‐diastolic volume index was best suited to predict exercise capacity (P < 0.0001). In a large validation cohort, LV end‐diastolic volume was confirmed to predict exercise capacity (P < 0.0001).

Conclusions

Among patients referred for outpatient stress echocardiography who have a preserved LVEF and no evidence of myocardial ischemia, we found a strong positive association between LV volume and exercise capacity.

1. INTRODUCTION

Exercise capacity is a powerful predictor of cardiovascular (CV) and all‐cause mortality. Exercise duration as a surrogate for exercise capacity adds to an age‐based prediction model of survival in both healthy subjects and patients with cardiovascular disease who are referred for exercise testing.1, 2, 3 In a small explorative study of patients with a normal left ventricular ejection fraction (LVEF), we recently demonstrated that patients with a very poor exercise capacity or who are unable to exercise on a treadmill are much more likely to have a concentrically remodeled left ventricle (LV) with a small cavity volume when compared with patients with an excellent exercise capacity.4 These findings suggested that the ability to exercise may be reflected in the LV cavity size in a wider range of patients.

Only 1 study has directly related an echocardiographic measure to exercise capacity in a larger group of patients.5 In 2879 patients who underwent exercise echocardiography using the Bruce protocol, it was demonstrated that evidence of diastolic dysfunction was inversely associated with exercise capacity. Several small studies looked at LV dimensions in select clinical populations that can be assumed to have a low exercise tolerance.6, 7, 8, 9, 10, 11 Patients with chronic obstructive pulmonary disease were found to have smaller LV chamber dimensions, and patients with end‐stage pulmonary arterial hypertension (HTN) are reported to develop LV atrophy.7, 8, 9, 10 These cardiac adaptations are believed to be the result of a reduction in LV workload.12 Most studies that directly investigated the LV size and exercise relationship focused on small healthy populations, including athletes, rather than patients.12, 13, 14 It is not known if such a relationship exists in patients with normal LVEFs.

It was, therefore, the first objective of this study to determine if there is a principal association between LV structure and exercise capacity in an unselected patient population with a normal LVEF referred for a treadmill stress echocardiogram in whom there was no evidence of myocardial ischemia during exercise. In a multivariate analysis of 201 patients, we first investigated which LV structural parameters are best suited to predict exercise capacity (derivation cohort).

The second objective was to test whether the best structural predictor of exercise capacity can be validated in a large analysis of 1285 patients with a preserved LVEF who had both a transthoracic echocardiogram (TTE) and an independent Bruce protocol stress test (validation cohort). We also evaluated the predictive capacity of an age‐based prediction model of exercise capacity that includes the LV volume.

2. METHODS

2.1. Derivation cohort

We used the University of Vermont Medical Center (UVMMC) image database of digitally recorded treadmill exercise echocardiographic studies performed between January 2015 and November 2015. Based on our previous analysis, it was determined that approximately 200 studies would be sufficient to establish the relationship between resting LV end‐diastolic chamber volume (LVEDV) and exercise capacity.4 Exercise duration on a Bruce protocol is well suited to estimate exercise capacity and workload.15

All studies were initiated by the treating physician. Patients with a normal LVEF (≥50%) without chamber dilation were included irrespective of the test indication. Exclusion criteria were age < 20 years, LVEF <50%, LV volume index (LVVI) ≥75 mL/m², baseline or stress‐induced wall‐motion abnormalities, and more than moderate valvular disease. To meet the enrollment target, 380 patient studies were screened in reverse chronological order according to the date of study. The following patient characteristics were tabulated in a de‐identified manner: test indication, CV risk factors, reason for test termination, sex, age, pretest blood pressure (BP) and heart rate, body weight and height, medications, exercise duration, and estimated metabolic equivalents of tasks (METs). The enrollment flow diagram is shown in Figure 1.

Figure 1.

Flowchart of study derivation cohort and validation cohort enrollment. Abbreviations: EF, ejection fraction; LV, left ventricle

All echocardiographic images were evaluated by a level 3 certified echocardiographer who was blinded to the clinical parameters and exercise results. LV dimensions were obtained in accordance with guideline recommendations using the parasternal short‐ and long‐axis views and the apical 2‐ and 4‐chamber views.16, 17 LV end‐diastolic chamber volumes were calculated using the biplane method of disks (modified Simpson analysis) and were used to calculate LVEF. The University of Vermont Institutional Review Board reviewed and approved this study. Only studies that were deemed to have a good imaging quality were analyzed; 201 studies were included in the final analysis. An analysis of intraobserver variability of LV volumes resulted in a correlation of r ² = 0.86 and a typical error of 7.8 ±4.4 mL.

2.2. Validation cohort

2.2.1. Stress test database

A database of 13 361 treadmill exercise stress tests performed at UVMMC between June 2011 and November 2014 was established (Figure 1). The following patient parameters could be retrieved and tabulated: sex, age, date of birth, study date, study type, weight, height, body surface area (BSA), exercise time, maximum work rate, maximum heart rate achieved, percent of predicted maximum heart rate, and rate pressure product. Studies that were determined not to follow the standard Bruce protocol were excluded. All patients in both cohorts were encouraged to exercise to peak capacity or at the minimum to 85% of maximum predicted heart rate without unwarranted or premature termination of the exercise treadmill test. Test endpoints include an increase in exercise BP with maximum of 240/115 mm Hg, any clinically significant drop in systolic BP (≥10 mm Hg), a run of ventricular tachycardia ≥4 beats, development of sustained atrial fibrillation or supraventricular tachycardia during exercise, rate‐induced left bundle branch block that is not distinguishable from possible ventricular tachycardia, heart block or symptomatic bradyarrhythmia, excessive ST‐segment depressions (>2 mm), ST‐segment elevation (1 mm in leads without diagnostic Q waves except for leads V₁ or aVR), moderate to severe angina, marked dyspnea or fatigue, ataxia, dizziness, near‐syncope, any instability, or patient desire to stop.

2.2.2. Echocardiographic database

A database covering the same time period was created. This resulted in 36 749 TTE studies. The following recorded parameters could be retrieved and tabulated: age, weight, height, BSA, and end‐diastolic and end‐systolic LV volume in the 4‐chamber and 2‐chamber view using the modified Simpson analysis. All parameters in this database were recorded and documented by a total of 27 cardiac sonographers at 5 sites (1 outpatient clinic and 4 hospital sites). Patients with an LVEF <50% (apical 4‐chamber view or 2‐chamber view) or dilated LV chamber (LVVI >75 mL/m²) were excluded, which resulted in 17 954 remaining studies. Because many studies lacked the 2‐chamber volume data, only the 4‐chamber values were used to assess the interaction between LV volume and exercise capacity. In the patients with a complete set of volume data, the volume in the 4‐chamber view was 80 ±23 mL, and in the 2‐chamber view 79 ±26 mL.

2.2.3. Composite validation database

Alignment of the individual treadmill stress data and TTE data was accomplished by converging data using a unique nonidentifying number. The data with the closest testing dates were chosen if a given patient had >1 stress test or TTE study. Eighty‐eight percent of the patients underwent the 2 tests within a 12‐month period; this yielded a total of 1285 composite datasets (Figure 1). Ancillary analyses of a 30‐day and 7‐day time window between stress test and TTE were also performed.

2.3. Statistical analysis

Based on our previous study, we determined that a sample size of about 200 patients would be sufficient to provide 80% power to detect a difference between patients with below‐average vs above‐average exercise capacity (5% type I error level).4

Using multivariate regression analysis, we compared the association between exercise duration and the standard recommended linear LV dimensions and volumes, as well as other clinical variables including age, resting heart rate, and BSA. This analysis was performed in both the derivation and validation cohorts. The multivariate regression algorithm formulated from the derivation cohort was tested in the validation cohort, and a computational F‐test was performed to determine the variance between the different sets of predicted values from the 2 cohorts. Furthermore, we performed a receiver operating characteristic curve analysis and calculated area under the curve (AUC) for different prediction models.

All statistical analyses were performed with SPSS software version 24.0 (IBM Corp., Armonk, NY). A P value ≤0.05 was considered statistically significant.

3. RESULTS

3.1. Derivation cohort

The derivation cohort was established to determine which standard parameter of LV structure correlated best with exercise duration in Bruce protocol stress tests that were negative for ischemia.

Demographics of the derivation cohort are shown in Table 1. The mean age of the patients was 60 ±13 years. The most common indication for the exercise stress test was chest pain, followed by dyspnea and coronary artery disease. Multivariate analysis did not reveal any significant association between test indications and exercise duration. Nearly half the patients had HTN or dyslipidemia as an established risk factor.

Table 1.

Clinical and echocardiographic characteristics, derivation cohort (n = 201)

	Value	Range
Age, y	60 ±13	20–88
Female sex	93 (46)	—
BMI, kg/m2	29 ± 8	16–57
BSA, m−2	1.97 ± 0.24	1.41–2.54
BP, mm Hg	130/77 ± 16/9	100–180/50–110
Resting HR, min−1	68 ± 12	42–102
Test indication
Dyspnea	57 (28)	—
Chest pain	91 (45)	—
CAD	30 (15)	—
Palpitations	8 (4)	—
Other	16 (8)	—
CV risk factors
Current smoker	30 (15)	—
DM	23 (11)	—
HTN	87 (43)	—
Hyperlipidemia	94 (47)	—
Medications
β‐Blocker	66 (33)	—
ACEI/ARB	40 (20)	—
CCB	31 (15)	—
ASA	113 (56)	—
Statin	86 (43)	—
Diuretic	38 (19)	—
Stress test performance
Test duration, s	502 ± 195	60–1063
METs	9 ± 3	3–16
Reason for test termination
Dyspnea/fatigue	187 (93)	—
Other	17 (7)	—
Echocardiographic parameters
Septal wall, mm	10 ±1.8	6.1–17.1
Posterior wall, mm	9.3 ± 1.4	5.1–15.5
LVEDD, mm	50.9 ± 6.1	31.3–69.8
LVEDV, mL	88 ± 23	41–169
LVEDV/BSA	45 ± 10	20–70
LVEDV/height, mL/m	52 ±12	27–92
LVEDV/height,2.7 mL/m2.7	21 ± 5	11–39
LVESV, mL	33 ± 12	11–76

Abbreviations: ACEI, angiotensin‐converting enzyme inhibitor; ARB, angiotensin II receptor blocker; ASA, acetylsalicylic acid (aspirin); BMI, body mass index; BP, blood pressure; BSA, body surface area; CAD, coronary artery disease; CCB, calcium channel blocker; CV, cardiovascular; DM, diabetes mellitus; HR, heart rate; HTN, hypertension; LVEDD, left ventricular end‐diastolic dimension; LVEDV, left ventricular end‐diastolic volume; LVESV, left ventricular end‐systolic volume; METs, metabolic equivalents of tasks; SD, standard deviation. Data are presented as n (%) or mean ±SD.

A multivariate regression analysis of the standard recommended linear LV dimensions and volumes was performed (see Supporting Information, Table 1a, in the online version of this article). The LVEDV was best at predicting exercise capacity (P = 8.6 × 10⁻⁵), whereas the end‐systolic volume was not predictive. The only other parameter that predicted exercise capacity was the septal wall thickness (P = 0.005). We compared LVEDV with other clinical variables (see Supporting Information, Table 1b, in the online version of this article); this comparison revealed that age was better at predicting exercise capacity, whereas resting heart rate and BSA were inferior.

We next evaluated whether the recommended anthropometric indexing of LV volumes using BSA, height, and the allometric height index (H^2.7) would improve the capacity of LVEDV to predict exercise duration (see Supporting Information, Table 2, in the online version of this article). This analysis suggested that BSA‐based indexing had a better predictive capacity than uncorrected volumes, and both height‐based volume indices were inferior to the uncorrected volumes. Analysis of LVEDV/BSA deciles demonstrated a positive linear correlation between LVEDV/BSA and exercise duration (r ² = 0.85; P = 0.0002) as shown in Figure 2.

Table 2.

Clinical and echocardiographic characteristics, validation cohort, n = 1285

	Value	Range
Age, y	63 ±10	45–92
Female sex	483 (38)	—
BMI, kg/m2	29 ± 6	12–58
BSA, m−2	2.01 ± 0.27	1.05–3.51
Stress test performance
Test duration, s	460 ± 170	32–1095
Baseline echocardiographic parameters
LVEDV, mL	80 ± 26	23–170
LVEDV/BSA	39 ± 11	9–74

Abbreviations: BMI, body mass index; BSA, body surface area; LVEDV, left ventricular end‐diastolic volume; SD, standard deviation.

Data are presented as n (%) or mean ± SD.

Figure 2.

Average exercise time in seconds plotted against decile LVEDV divided by BSA of the derivation cohort (n = 201). Abbreviations: BSA, body surface area; LVEDV, left ventricular end‐diastolic volume

3.2. Validation cohort

This principal relationship was tested in a cohort of 1285 patients who had a TTE independent from a Bruce protocol exercise stress test. The baseline patient characteristics were similar to those of the derivation cohort as shown in Table 2. Multivariate regression analysis confirmed a strong positive association between exercise duration and LVEDV (see Supporting Information, Table 3, in the online version of this article). Expectedly, the relationship between age and exercise capacity was strongest. Analysis of the BSA‐based LVEDV deciles confirmed a linear relationship between the LVVI and exercise duration in the validation cohort (r ² = 0.82; P < 0.001). An analysis of shorter time windows between the echocardiogram and the exercise stress test (30 days and 7 days) continued to demonstrate a strong association between exercise duration, age, and end‐diastolic LVVI, but it did not improve the significance levels.

3.3. LV volume–based calculation of exercise capacity

The above association allowed us to formulate an age‐and‐LVVI–based algorithm to predict exercise duration. The multivariate regression algorithm from the derivation cohort was tested in the validation cohort. A computational F‐test revealed that the residual variances were not different between both cohorts, which suggests that the following equation is generally applicable: Exercise time (s) = 501 – (5.7 × age) + (7.6 × [LVEDV/BSA]).

This allowed calculation of the receiver operating characteristics of patients in the validation cohort who could not reach stage 3 of the Bruce protocol (partition value: 360 s). This defines a patient population with a reduced functional capacity (METs ≤7; New York Heart Association class ≥2). The composite of the age‐and‐LVVI–based prediction model resulted in an AUC of 0.67 (P < 0.001; Figure 3). In comparison, the age‐only prediction model had an AUC of 0.59 (P < 0.001).

Figure 3.

Age‐based ROC analysis with and without LV volume index of patients in the validation cohort unable to reach stage 3 of the Bruce protocol. Abbreviations: AUC, area under the curve; LV, left ventricular; ROC, receiver operating characteristic

As an example, this would predict that a 65‐year‐old subject with a small end‐diastolic LVVI of 27.5 (LVEDV, 55 mL; BSA, 2 m²) would not be able to complete stage 2 of the Bruce protocol. A patient with the same age but a normal LVEDV of 110 mL would be predicted to exercise into stage 4 of the Bruce protocol.

4. DISCUSSION

Among patients referred for stress testing who had preserved LVEF and no evidence of myocardial ischemia, we found an association between LV dimensions and exercise capacity. The main findings can be summarized as follows: (1) There is a strong positive relationship between end‐diastolic LV chamber volume and exercise duration in a standard Bruce protocol stress test; and (2) the composite of age and end‐diastolic LVVI can predict exercise duration by the following equation: exercise duration (s) = 501 – (5.7 × age) + (7.6 × [LVEDV/BSA]).

4.1. Exercise capacity and mortality

Several studies have examined exercise capacity as a predictor of mortality in populations such as the Framingham cohort.18 All‐cause mortality was evaluated in 6213 male patients with and without a history of cardiovascular disease who were referred for treadmill exercise testing for clinical reasons.1 After adjustment for age, exercise capacity was found to be the strongest predictor of the risk of death and each 1‐MET increase in exercise capacity conferred a 12% improvement in survival.1

In a select Minnesota population, Roger et al. sought to examine the prognostic value of treadmill exercise testing in a cohort of 2193 male and female patients.2 They found that the ability to exercise was strongly associated with all‐cause mortality and cardiac events, and the strength of this association was similar in both male and female patients. A 1‐MET increase in exercise capacity conferred a 20% to 25% reduction in all‐cause mortality and cardiac events.2 In a subsequent study, Goraya et al. evaluated whether the prognostic value of treadmill exercise testing was equivalent in elderly patients (age > 65 years) compared with younger patients.19 In elderly patients, each 1‐MET increase in exercise capacity was still associated with an 18% reduction in cardiac events, compared with a 14% reduction in younger patients.

Another longitudinal study found that low levels of fitness in middle‐aged subjects were associated with increased rates of hospitalization for heart failure later in life.20 In an intervention study, Fujimoto et al. reported that 1 year of walking‐based exercise training in initially sedentary seniors increased CV fitness.21 These findings demonstrate the strong relationship between exercise capacity and cardiac events.

4.2. LV dimensions and exercise capacity

Changes in cardiac workload affect LV dimensions through adaptive remodeling of the myocardium.12 Sustained changes in cardiac workload due to a sedentary lifestyle and HTN can result in concentric remodeling and smaller LV chamber dimensions. In contrast, endurance exercise has been associated with larger LV dimensions as a component of adaptive eccentric remodeling.13, 14, 22 Pluim et al. reviewed the literature on athletes in a meta‐analysis examining this concept and conclude that although there is an “endurance trained heart” and a “strength trained heart” adapted to handle high‐volume loads vs pressure loads, this concept is not absolute but exists on a continuum.22

In contrast, sudden extreme physical inactivity in otherwise‐healthy subjects results in regression of LV chamber volumes and mass.13, 14 In the general patient population, it is less clear how differences in workload affect the structure of the LV. Smaller LV dimensions and LV atrophy have been reported in patients with severe chronic obstructive pulmonary disease and pulmonary arterial HTN, which suggests that the same mechanisms are at play.7, 8, 9, 10 It can therefore be argued that LV dimensions may reflect the ability to exercise.

4.3. Echocardiography as a predictor of exercise capacity

The assessment of LV chamber volumes using the biplane method of disks is a relatively recent addition to the recommended measurements in TTEs.17 This measurement of LV volume integrates LV chamber size and the presence of concentric remodeling, which may explain why we found that end‐diastolic volumes are a good predictor of exercise capacity. Parasternal linear dimensions were not found to be predictive of exercise capacity, with the notable exception of septal wall thickness. This is in line with previous findings in patients with septal hypertrophy and concentric LV hypertrophy.4, 23 The most likely explanation for why BSA‐based indexing was superior to height‐based indexing is the inclusion of weight. This result also provides a functional argument in support of the most recent chamber volume guidelines, which favor BSA‐based normalizations.16

At our institution, the number of patients who had echocardiograms performed during the time frame of this study exceeded the number of patients who had treadmill exercise tests by a ratio of about 5:1, which is similar to the reported Medicare population at large.24 Because the present findings suggest that standard TTE can provide a “fitness estimate,” it may be reasonable to discuss the beneficial effects of physical exercise in patients with small cavity volumes. Consideration of these findings could also help to better refer patients to the most appropriate stress‐test modality, thereby improving diagnostic yield and possibly reducing costs.

It has been demonstrated that normal aging is associated with a reduction in LV volumes. However, individuals who exercise regularly throughout their lives maintain normal LV dimensions.25, 26 This finding provides additional support to the concept that LV volumes portend functional and prognostic information.

As maximal exercise capacity is tightly linked to cardiac output reserve, our findings also provide insights into the relationship between ventricular structure and function. Like the maximal displacement volume of an engine, end‐diastolic volume is a major determinant of functional reserve. This is plausible because small LV cavity volumes will inevitably introduce a physical limitation to systolic reserve and maximum recruitable stroke volume. At rest this is compensated for by an increase in heart rate, as demonstrated in our previous study.4 However, elevated resting heart rates combined with chronotropic incompetence, which is frequently present in these patients, will further compromise the usually potentiating effect of heart rate on cardiac output. Because some of these variables can be positively influenced by exercise and even small improvements in exercise capacity translate in improved survival, patients should always be encouraged to exercise.

4.4. Study limitations

As discussed, inferences on the effects of exercise on LV volumes in patients cannot be made from the current study. Functional echocardiographic parameters such as diastolic function have a predictive capacity; this study demonstrates an association between LV structure and exercise capacity.

5. CONCLUSION

Our study reveals a prominent relationship between LVEDV and exercise capacity in patients with a normal LVEF. Consideration of the LVVI can help to predict exercise capacity.

Author contributions

Lakshmi Nambiar, MD, and Anita Li, BS, contributed equally.

Conflicts of interest

The authors declare no potential conflicts of interest.

Supporting information

Supplemental Table 1a Multiple regression table. LVEDD, LV end‐diastolic diameter; SW, septal wall thickness; PW, posterior wall thickness; LVEDV, LV end‐diastolic volume; LV end‐systolic volume.

Supplemental Table 1b Multiple regression table. LVEDV, LV end‐diastolic volume; HR, heart rate; BSA, body surface area.

Supplemental Table 2 Multiple regression table. LVEDV, LV end‐diastolic volume; BSA, body‐surface area; H, Height; H^2.7, allometric height index.

Supplemental Table 3 Multivariate regression table. LVEDV, LV end‐diastolic volume; BSA, body‐surface area.

Nambiar L, Li A, Howard A, LeWinter M, Meyer M. Left ventricular end‐diastolic volume predicts exercise capacity in patients with a normal ejection fraction. Clin Cardiol. 2018;41:628–633. 10.1002/clc.22928