Exercise Tolerance and the Post Exercise Diastolic Filling Pattern in Patients With the Resting Impaired Relaxation

Background In patients with normal LV systolic function, cardiac output increases with exercise mediated by increased stroke volume early in exercise and an increase in heart rate later in exercise. Despite normal LV systolic function, patients who display an impaired relaxation pattern may have a reduced exercise tolerance. We hypothesized that the resting impaired relaxation pattern that persists during exercise results in reduced LV filling volume and reduced exercise tolerance. Methods We evaluated consecutive exercise echocardiograms performed at Harper Hospital from 1998-2000 for patients with sinus rhythm, normal resting wall motion and ejection fraction (> 55%), evidence of resting impaired relaxation, and a negative exercise echocardiogram. There were 49 patients fitting the above criteria who were compared with a group of age and sex matched patients (43 patients) with a normal rest and exercise echocardiogram with normal resting transmitral Doppler. Rest and post exercise echocardiography and Doppler parameters were obtained. Results Patients in the impaired relaxation group demonstrated shorter exercise times as compared to the normal control group (8.8 ± 1.6 versus 9.7 ± 2.0 minutes, P < 0.001). In patients with normal resting transmitral diastolic filling, there was an increased the extent of atrial contribution to LV filling volume post exercise associated with shortening of isovolumic relaxation. Two patterns were seen in the impaired relaxation group post exercise. In 1 subgroup in which E/A ratio decreased post exercise, exercise duration was reduced (7.4 ± 1.3 minutes, P < 0.001) as compared to the subgroup with E/A increase (9.6 ± 1.2 minutes) post exercise which was similar to normal controls. Forward stepwise regression indicated that exercise time was primarily related to E/A change post exercise for all patient groups (r = 0.625, P = 0.0008). Specifically, this was true for patients with E/A reversal at rest (r = 0.584, P = 0.0028). However, for patients with normal diastolic filling at rest, the diastolic velocity integral was the major predictor (r = 0.695, P < 0.0084). Conclusion We conclude that the transmitral Doppler pattern post exercise provides insight into the mechanism of reduced exercise tolerance in some patients with the resting impaired relaxation pattern. Preservation of this pattern post exercise is associated with reduced exercise tolerance.


Introduction
In patients with normal LV systolic function, cardiac output increases with exercise mediated by an increase in stroke volume and heart rate. Early in the course of exercise, stroke volume increases and plateaus while progressively increasing heart rate becomes responsible for additional cardiac output increases in the normal left ventricle [1][2]. Many patients display an impaired relaxation pattern associated with hypertension, LV hypertrophy, diabetes, increasing age, or coronary disease [3][4][5][6][7] despite normal LV systolic dysfunction. Exercise tolerance in patients with an impaired mitral Doppler relaxation pattern at rest has been variably described either as normal or reduced [8][9]. The addition of pulsed mitral annular Doppler rapid fi lling velocity (e') has not always been helpful in identifying exercise tolerance [10][11]. However, an elevated ratio of the resting rapid fi lling transmitral Doppler velocity/pulsed mitral annular Doppler rapid fi lling velocity (E/e') may be associated with reduced exercise capacity [10][11]. Unfortunately, the groups studied include patients being evaluated for dyspnea and hypertensive patients who were variably symptomatic.
More recently, the use of exercise mitral Doppler patterns and the exercise ratio of E/e' have been utilized to assess exercise capacity but only in patient groups with normal systolic and diastolic function, reduced systolic function, and in patients who develop ischemia. These studies have demonstrated that both exercise transmitral Doppler and the exercise E/e' pattern predicted reduced exercise tolerance [11][12][13].
In patients with normal systolic function and impaired relaxation, atrial compensation has already been invoked at rest to maintain fi lling volume [1]. It is not clear whether exercise in these patients will result in increases in stroke volume or whether cardiac output solely increases from heart rate. Accordingly, we hypothesize that the resting impaired relaxation pattern that persists during exercise results in reduced LV fi lling volume and consequently reduced exercise tolerance. For patients with normalization of the diastolic fi lling pattern with exercise, exercise capacity may increase despite possibly being limited by mildly elevated LV fi lling pressures [13].

Patients
This study was approved by the Wayne State University Human Investigation Committee (IRB) as an exempt study. We evaluated consecutive exercise echocardiograms performed at Harper Hospital from 1998-2000 for patients with sinus rhythm, normal resting wall motion and ejection fraction (> 55%), evidence of resting impaired relaxation on diastolic fi lling (peak rapid fi lling velocity/peak atrial fi lling velocity < 1 and deceleration time > 240 msec), no signifi cant valvular regurgitation greater than mild of any of the 4 valves, and a negative ECG response to exercise associated with an improvement in exercise ejection fraction without evidence of a wall motion abnormality. As a routine since 1993, transmitral Doppler was recorded during recovery from treadmill exercise. The exercise echocardiograms were reviewed to ensure that diastolic fi lling pattern could be evaluated during the recovery period (within 2 minutes post exercise) consisting of visibly separate peak rapid fi lling (E) and peak atrial fi lling (A) waves. We performed 861 exercise echocardiograms over this 2-year period. After applying the inclusion criteria, there were 49 patients who fi t these criteria. The indication for the exercise echocardiogram was a chest pain syndrome in 45 and dyspnea in 4. A second group of age and sex matched patients (43 patients) with a normal rest and exercise echocardiogram and normal resting transmitral Doppler pattern were selected from the same days that the impaired relaxation group was studied. The indication for the exercise echocardiogram was chest pain in 39 and dyspnea in 2, and palpitations in 2.

Rest and Exercise Echocardiography
Rest and exercise echocardiography were obtained with a HP 2500 echocardiograph (Hewlett Packard, Andover, MA) interfaced to an offl ine digital acquisition and display system for LV recordings (Microsonics, Bothell, WA) and stress ECG recording system (Marquette-GE, Milwaukee, WI). Standard parasternal, apical, and subcostal views were obtained along with pulsed wave, continuous wave, and color Doppler across each valve. Transmitral Doppler was obtained at rest with a 5 x 5 mm sample volume just beyond the tips of the mitral leafl ets and recorded at 100 mm/s. Similarly, transaortic Doppler was recorded from the apical 5 chamber or 3 chamber view with 5 x 5 mm sample volume just beyond the aortic leafl ets and recorded at 100 mm/s. All recordings were made on ½ inch VCR tape, and rest LV images in the parasternal long, mid ventricular short axis and in the apical 2 and 4 chamber views were digitally obtained.
Exercise was performed using a treadmill employing the Bruce protocol. All patients were exercised to their symptom limited maximum (impaired relaxation group: fatigue in 45 and chest pain in 4; normal resting transmitral Doppler group: fatigue in 39 and chest pain in 4). Immediately post exercise, each patient was rapidly imaged in the supine position within 45-60 seconds in the apical 4 and 2 chamber and parasternal long and short axis views. After obtaining the above images both on VCR tape and digitally, transmitral Doppler was obtained at a sweep speed of 100 mm/second as above on VCR tape at held end expiration within 2 minutes post exercise when distinct E and A waves of transmitral spectral Doppler had separated.

Demographic Variables
The incidences of hypertension (blood pressure > 140/90 or on anti-hypertensive medications), diabetes (fasting blood sugar > 126 mg or on medications for glycemic control), and LV hypertrophy (LV mass index < 96 g/m 2 in women and < 115 g/m 2 in man) were determined for both the normal transmitral fi lling group and the impaired relaxation group.

Exercise Testing Variables
Heart rate at rest, peak exercise, and at the time of transmitral Doppler recording was obtained. Resting and peak systolic and diastolic pressures were obtained. Double product at rest and peak exercise was obtained by multiplying heart rate by peak systolic pressure.

Left Ventricular Volumes and Mass
LV volumes at end diastole and end systole at rest and with exercise were determined using biplane Simpson's rule. LV mass was calculated from resting 2 dimensional echo data using the recommendations of the American Society Echocardiography [14] and indexed to body surface area.

Doppler Variables
All calculations were from the average of 3 consecutive cycles. Color fl ow assessment of all 4 valves was performed. Any patient with valvular regurgitation greater than mild was excluded. Mild mitral regurgitation was defi ned as the ratio of the maximal color fl ow jet area/corresponding left atrial area < 20% in all apical views. Mild aortic regurgitation was defi ned as height of the aortic regurgitation jet in the LV outfl ow tract/LV outfl ow tract < 25% [15]. Mild tricuspid regurgitation was defi ned in a similar fashion to mild mitral regurgitation as a jet area of < 20% of the right atrial area in the apical 4 chamber view. Mild pulmonic regurgitation was assessed by the length of the jet in the RV outfl ow tract < 10 mm [15].
For both rest and post exercise, all Doppler indices were measured from the average of 3 consecutive cycles. From transmitral Doppler indices, peak rapid fi lling velocity (E) and peak atrial fi lling velocity (A) were measured. The rapid fi lling deceleration time was calculated as the time interval from the peak rapid fi lling velocity to the time mitral fl ow decelerated to the zero baseline. The tracing was extrapolated to the zero baseline if atrial fi lling commenced prior to mitral fl ow fully decelerating to zero. Diastolic fi lling, rapid fi lling, and atrial fi lling velocity integrals were determined. The length of the diastolic fi lling period was obtained as the interval from beginning to the end of transmitral spectral tracing. The atrial fi lling period and its integral were determined from the onset to the end of atrial fi lling. When rapid and atrial fi lling velocity spectra demonstrated any degree of merging, the onset of atrial fi lling was defi ned at the point of the end of the p wave on the ECG. The time from the R wave to the onset of the mitral time velocity spectrum was obtained. The time from the R wave to the end of rapid fi lling was obtained. If the rapid fi lling velocity had not decelerated to the baseline, then the point at which the velocity began to increase was use as the end of the rapid fi lling period. The rapid fi lling period and its integral were calculated from the onset of mitral infl ow to the end of the rapid fi lling period. Isovolumic relaxation time was calculated as the time interval from the end of aortic velocity spectrum to the onset of the mitral velocity spectrum. in 10 randomly selected patients for E, A, and the diastolic time velocity integral at rest and following exercise 3-7 weeks apart. For E, the difference for intraobserver and interobserver measurements were 2 ± 2 cm/s and 3 ± 2 cm/s at rest and 3 ± 3 cm/s and 3 ± 4 cm/s post exercise. For A, the difference for intraobserver and interobserver measurements were 2 ± 2 cm/s and 2 ± 3 cm/s at rest and 3 ± 3 cm/s and 3 ± 5 cm/s post exercise. For the diastolic time velocity integral, the difference for intraobserver and interobserver measurements were 0.09 ± 0.11 cm and 0.13 ± 0.11 cm at rest and 0.17 ± 0.19 cm and 0.20 ± 0.17 cm post exercise.   Table 1 summarizes demographics and exercise testing variables for patients with E/A reversal (impaired relaxation) at rest and patients with normal diastolic fi lling. Patients with impaired relaxation exercised a shorter period of time, achieved a lower peak heart rate, and had an increased incidence of hypertension and LV hypertrophy. Diabetes was more frequent in patients with impaired relaxation. The effect of exercise on hemodynamic and transmitral and transaortic Doppler variables in patients with and without E/A at rest are summarized in Table 2. Peak exercise heart rate was higher in the group with normal diastolic fi lling at rest. Blood pressure response to exercise was similar in both groups. The impaired relaxation group demonstrated an increased peak atrial fi lling velocity, prolongation of atrial fi lling, increased atrial fi lling velocity integral, prolonged deceleration time, and a prolonged isovolumic relaxation time. Immediately post exercise (Table 2 and Fig. 1), the normal diastolic fi lling group demonstrated increases in the peak rapid fi lling and atrial fi lling velocities with a reduction in the E/A due to a greater increment in the peak atrial fi lling velocity. The diastolic fi lling period as a function of cycle length prolonged, and the isovolumic relaxation period as a function of cycle length shortened. Immediately post exer-  cise, the impaired relaxation group increased peak rapid fi lling velocity, increased E/A compared to rest, and prolonged the atrial fi lling period. When comparing the impaired relaxation group vs the normal diastolic fi lling group post exercise, there was a lower peak rapid fi lling velocity, a smaller rapid fi lling velocity integral, a lower E/A, a larger atrial velocity integral, and a longer atrial fi lling period as a function of cycle length. There was also a longer deceleration time and isovolumic relaxation time as a function of cycle length despite a shorter diastolic fi lling period as function of cycle length. A graphic depiction of the transmitral fi lling pattern at rest and post exercise is shown in Fig. 1 for the 2 groups. Table 3 summarizes the changes in diastolic fi lling from rest to post exercise in patients with normal and impaired diastolic fi lling at rest. Patients with normal diastolic fi lling demonstrated a greater increase in peak atrial fi lling velocity and integral, diastolic velocity integral, but associated with a reduction in E/A as compared to patients with impaired relaxation at rest. In patients with impaired relaxation fi lling pattern at rest, there were 2 separate and distinct responses in the diastolic fi lling pattern post exercise (Table 4). In 17 patients there was no change or a decrease in E/A post exercise (Fig.  2a) and in 32 patients, E/A increased post exercise (Fig. 2b). Exercise time was longer in the subgroup with an increase in E/A and a higher percentage of males were noted. The increase in E/A was primarily due to an increase in rapid fi lling as characterized by a greater change in the peak rapid fi lling velocity and in the rapid fi lling velocity integral than in the subgroup with no change or decrease in E/A.

Results
Using forward stepwise multiple regression (Table 5), the independent determinants of E/A change post exercise demonstrated that exercise time and rapid fi lling velocity integral change were independent determinants of E/A change (r = 0.701, R 2 = 0.491, P < 0.0001). Restricting the analysis to patients with impaired relaxation at rest demonstrated the same determinants (r = 0.727, R 2 = 0.529, P < 0.0001). Similarly, the only independent determinant of exercise time for all groups (r = 0.625, P = 0.0008) and specifi cally in patients with impaired relaxation at rest was E/A change (r = 0.584, P = 0.0028). In patients with normal diastolic fi lling pattern at rest, exercise time was predicted by the diastolic velocity interval change (r = 0.695, P = 0.0084) which is an estimate of stroke volume.

Discussion
In the normal patient, cardiac output increases with exercise mediated by an initial increase in stroke volume early in exercise with further increases in cardiac output resulting from increased heart rate [1][2]. However, many patients with normal pump function will display an impaired relaxation pattern on transmitral Doppler often related to hypertension, diabetes, coronary disease, age [3][4][5][6][7], and has been ascribed by clinicians as a cause of dyspnea. However, the E/e' ratio and its correlation with brain natiuretic peptides has further cemented the relation of abnormal diastolic fi lling and elevated LV fi lling pressures as a cause of dyspnea and limitation of exercise [9][10]. More recently, the use of exercise mitral Doppler patterns and the exercise ratio of E/e' have been utilized to assess exercise capacity but in patient groups with normal systolic and diastolic function, reduced systolic function, and in patients who develop ischemia. These studies have demonstrated that both the exercise mitral Doppler pattern and the E/e' predict reduced exercise tolerance [9][10][11][12][13] based on increased E/e' ratios. However, the role of increases in cardiac output with exercise has not been addressed in this patient group as a cause of "fatigue and dyspnea" as a limiting symptom.
In patients with normal systolic function and impaired relaxation, atrial compensation has already been invoked at rest to maintain fi lling volume. It is not clear whether exercise in these patients will result in increases in stroke volume or whether cardiac output solely increases from heart rate. Accordingly, we hypothesized that the resting impaired relaxation pattern that persists during exercise may result in reduced LV fi lling volume and contribute to reduced exercise tolerance.
In this study, we demonstrated that patients with normal resting transmitral diastolic fi lling patterns increased the extent of atrial contribution to LV fi lling volume post exercise associated with shortening of isovolumic relaxation, a fi nding previously demonstrated [16] suggesting the atrial systole is an important contributor to cardiac reserve during exercise. However, in patients with impaired diastolic fi lling, exercise time was reduced in this group as compared to the normal diastolic fi lling group. Subgroup analysis demonstrated that if an increased rapid fi lling contribution to fi lling volume occurred, a similar exercise tolerance was noted as compared to patients with normal diastolic fi lling at rest. Exercise tolerance was reduced in the subgroup whose diastolic fi lling pattern post exercise did not change and rapid fi lling contribution was similar to the resting pattern. Forward stepwise regression indicated that exercise time was primarily related to E/A change for all patients. Specifi cally, this was true for patients with E/A reversal at rest. However, for patients with normal diastolic fi lling at rest, the diastolic velocity integral was the best predictor. The strengths of these relations were moderate (r = 0.584-0.695). As we did not measure annular velocity with tissue Doppler (not available at the time of the study), we can not comment on the contribution of elevated LV fi lling pressures to exercise tolerance. For patients with normal resting fi lling, LV fi lling volume equivalent (or stroke volume) appeared to be a major determinant. Dyspnea and fatigue were a common cause of stopping exercise and may be related to either elevated LV fi lling pressures or limitations in cardiac output.

Previous Literature
In patients with normal LV function, immediately post ex- ercise, there is evidence of increased atrial contribution to fi lling volume suggesting that atrial systole is important for increasing stroke volume [1,17,18]. Impaired relaxation is seen in patients with coronary disease, hypertension, and LVH. These patients often demonstrate the transmitral pattern of impaired relaxation with prolonged isovolumic relaxation, E/A reversal, and prolonged deceleration time [3][4][5][6][7]19]. Post exercise, the pattern noted may be variable. However, at rest, E/A reversal correlated with maximal O 2 consumption in some patients with elevated LV fi lling pressures, decreased increment in cardiac index, and often LV systolic dysfunction [20]. In patients with hypertension and impaired relaxation, exercise capacity was reduced if isovolumic period did not shorten and if there was continued E/A reversal post exercise [21,22]. As further diastolic dysfunction in hypertensive patients become evident with elevated LV fi lling pressures, workload correlated with natiuretic peptides and E/E' [11]. Nagueh demonstrated that a reduced peak annular fi lling velocity correlated with impairment of relaxation, and that the E/e' ratio correlated with mean pulmonary capillary pressure [23]. In patients with heart failure, increasing E/e' ratios correlated with increasing pulmonary capillary pressures and decreased O 2 consumption [24][25][26]. Using a combination of transmitral fi lling and peak annular diastolic velocity at rest and post exercise in 179 patients referred for exercise echocardiography, exercise capacity was reduced in patients with impaired relaxation at rest and post exercise if there was an E/e' > 11. Otherwise, a restrictive fi lling pattern at rest or post exercise resulted in reduced exercise capacity [12].
Our data indicates that E/A reversal at rest is associated with reduced exercise tolerance occurring in the subgroup with continued E/A reversal post exercise. This data is consistent with the LIFE sub-study [16]. Other studies indicate that E/A is inversely related to exercise tolerance, natiuretic peptides and the E/e' ratios [8,10,13]. These patients often develop greater increases in LV fi lling pressures than likely in the LIFE sub-study. This is consistent with data from the diastolic stress testing in which post exercise increases in E/e' correlate with reduced exercise capacity and increased natiuretic peptides [13] Limitations This is a retrospective, single center study with limited numbers in each group. There was referral bias in our institution for exercise echocardiography as patients with lower coronary risk were more often studied using exercise echocardiography. Patients with a history of coronary disease, previous revascularization, or at high risk based on symptoms and risk profi le more often-received nuclear perfusion studies. The inclusion and exclusion criteria for this study identifi ed a cohort of patients in which the effect of impaired relaxation on exercise tolerance would have less confounding infl uences of a high cardiovascular risk profi le. Despite the above limitations, our study demonstrates that patients with impaired relaxation by transmitral diastolic fi lling have reduced exercise tolerance and, specifi cally, when the transmitral Doppler fi lling profi le post exercise continues to demonstrate E/A reversal. Finally, we did not record the transmitral fi lling profi le during early exercise when the transmitral fi lling pattern was not fused. Different results may be possible.

Clinical Implications
Recently, the use of E/e' at rest and post exercise has been employed to determine whether dyspnea with exercise is related to elevated LV fi lling pressures [10][11]13] The E/e' ratio appears to predict elevated LV fi lling pressures at least at rest in patients with both reduced and preserved LV ejection fraction. The data post exercise appears to support the further use of this parameter as to a guide as to whether dyspnea as a limiting factor is due to elevation of LV fi lling pressures [10][11]. Ischemia with exercise can also manifest itself as dyspnea. This study specifi cally excluded these patients. Exercise tolerance may also be reduced due to fatigue due to a lesser increment in cardiac output with exercise. Fatigue may be diffi cult for patients to discern from dyspnea. Most patients in this study ceased exercise due to fatigue. We are unable to differentiate specifi cally whether LV fi lling pressures may have increased in these patients. However, the observation that atrial contribution to LV fi lling may be invoked in normals in this study suggest a limitation in the ability to increase stroke volume in patients with impaired relaxation at rest resulting in a heart rate increases to augment cardiac output. Not surprising, there are 2 distinct responses with preservation of the impaired relaxation pattern with reduced exercise tolerance as compared to the subgroup who utilized more rapid fi lling.

Conclusion
We conclude that the transmitral Doppler pattern at rest and post exercise provides insight into the importance of the impaired relaxation pattern. Preservation of this pattern post exercise is associated with reduced exercise tolerance in a patient group with E/A reversal on transmitral Doppler referred for chest or dyspnea that continue to manifest this pattern post exercise. Further evaluation of this observation is indicated in groups with evidence of impaired relaxation: hypertension, valve disease, and diabetes.