PATIENTS

Twenty one patients (16 men, 5 women) were included, 16 in NYHA functional class I and five in class IIa. All patients completed the protocol apart from one who refused to have cardiopulmonary tests because of claustrophobia. Characteristics of the study group are given in table 1.

Sixteen patients had undergone cardiac catheterisation; each had grade IV aortic regurgitation on aortography but no other valve disease or coronary artery disease. The aetiology of aortic regurgitation had been considered to be rheumatic in 12 patients (none of whom had mitral valve disease), bicuspid aortic valve in six, and previous infective endocarditis (two years before) in one. In two patients no cause could be defined.

Drug treatment was continued at the time of testing in nine patients (42%): seven were on angiotensin converting enzyme inhibitors, five on diuretics, one on sotalol, and one on verapamil.

EXERCISE RESPONSE

Exercise response results are shown in fig 3. Ejection fraction increased on exercise, or decreased by less than 5%, in 11 patients: from mean (SD) 55 (5)% to 58 (4)%, p < 0.05 (group I; good exercise response). Ejection fraction decreased by more than 5% in 10 patients: from 54 (4)% to 42 (5)%, p < 0.001 (group II; poor exercise response). In all patients in the latter group, the ejection fraction on exercise was less than 50% (34–49%).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Left: individual rest and peak exercise ejection fraction values for patients with good exercise responses (group I) (solid lines) and poor exercise responses (group II) (dashed lines); in group I mean ejection fraction increased from 55% to 58% (solid squares), and in group II it decreased from 54% to 42% (solid circles). Middle: Distributions of individual values of systolic excursion and peak systolic velocity(right) of medial mitral annular motion, in patients with good exercise response (group I) and poor exercise response (group II). The dotted lines represent the best discriminant cut off values obtained in the analyses of sensitivity and specificity.

In these two groups we compared the resting echocardiographic variables, indices of long axis and short axis function, and exercise cardiopulmonary function.

RESTING ECHOCARDIOGRAPHIC VARIABLES

Resting echocardiographic variables are given in table 2. There were no significant differences between the groups in resting ejection fraction, left ventricular diameter indices, end systolic wall stress, and end systolic wall stress to end systolic volume index ratio. Patients with poor exercise responses had a more dilated left ventricle and an increased left ventricular mass index, but resting left ventricular volume indices had low sensitivity and specificity for predicting a poor exercise response: 40% and 80%, respectively, for an end diastolic volume index ⩾ 150 ml/m2, and 70% and 64%, respectively, for an end systolic volume index ⩾ 60 ml/m².

LONG AXIS FUNCTION

Long axis function data are shown in fig 3. Patients with poor exercise responses had reduced long axis contraction of the left ventricle. The peak systolic velocity of medial mitral annular motion by tissue Doppler echocardiography was 8.6 (0.6) cm/s, compared with 11.9 (2.2) cm/s in patients with good exercise responses (p < 0.001). The systolic excursion by M mode echocardiography was 10.5 (2.2) mm, compared with 13.9 (2.1) mm (p < 0.01). Both peak systolic velocity and systolic excursion correlated with the change in ejection fraction from rest to exercise:r = 0.67 (p < 0.01) andr = 0.66 (p < 0.01), respectively.

A systolic excursion of less than 12 mm had a sensitivity of 80% and a specificity of 82% for predicting a poor exercise response. A resting velocity of less than 9.5 cm/s was the best indicator of a poor exercise response (sensitivity 90%, specificity 100%).

SHORT AXIS FUNCTION

The peak systolic velocity of the left ventricular posterior wall, which is a measure of short axis contraction, was not significantly different between the two groups (table 2).

DIASTOLIC FUNCTION

Seven of the 10 patients with poor exercise responses had global diastolic dysfunction, assessed as an E:A ratio < 1, compared with four of the 11 patients with good exercise responses (0.05 < p < 0.10). All 21 patients had normal short axis diastolic function, but eight of the 10 patients with poor exercise responses had long axis diastolic dysfunction (E_TDE:A_TDE ratio < 1), compared with only one of the 11 patients with good exercise responses (p < 0.01). Sensitivity and specificity of diastolic variables in predicting a poor exercise response were 70% and 64%, respectively, for an E:A ratio < 1, and 80% and 91%, respectively, for an E_TDE:A_TDE ratio < 1.

CARDIOPULMONARY FUNCTION

There were no significant differences between the two groups in maximum oxygen consumption, exercise duration, workload, anaerobic threshold, and minute ventilation against carbon dioxide production (V˙e/V˙co ₂ slope) (table2). Maximum oxygen consumption correlated with change in ejection fraction from rest to exercise (r = 0.57, p < 0.01); a value ⩽ 20 ml/kg/min had good specificity (91%) but low sensitivity (44%) for predicting a poor exercise response.

REPRODUCIBILITY

Intraobserver variability for the measurement of ejection fraction was ± 3.1% (in absolute terms), or ± 5.6% from the mean value for the resting values, and ± 2.3% (in absolute terms), or ± 4.3% from the mean value for the peak exercise values. Variability of measuring change of the ejection fraction from rest to exercise was ± 3.1% from the mean value (SD 2.2%). Variability of measuring long axis function was ± 0.51 cm/s (± 4.3%) for the peak systolic velocity, and ± 0.75 mm (± 5.8%) for the systolic excursion.

OPTIMAL TIMING OF SURGERY FOR ASYMPTOMATIC PATIENTS WITH SEVERE AORTIC REGURGITATION

If surgery was always postponed to avoid the risk of operative mortality (quoted at 2.4–4.9%) until the patients developed symptoms, long term survival would be poor.5 ,17 ,18 Accurate identification of individual patients at high risk of developing irreversible left ventricular dysfunction is desirable, but indications for surgery in asymptomatic patients are controversial.

The simple criterion of a left ventricular end systolic diameter of more than 55 mm was first proposed in 1980, based on a study of only 37 patients, and it is still widely applied.1 Others have suggested using a resting ejection fraction of less than 45%.2 Perhaps because a single dimension applied without reference to body size tends to discriminate against smaller patients, the prognosis after valve replacement for aortic regurgitation has been worse in women than in men.19

To date, the best predictors of deterioration in asymptomatic patients with aortic regurgitation have been related to left ventricular functional reserve during exercise, assessed by echocardiography or radionuclide ventriculography.3 ,4 ,6 ,7 Borer and colleagues reported that patients whose ejection fraction decreased on exercise by more than 5% had a 12.5%/year risk of developing heart failure or subclinical left ventricular dysfunction, compared with only 1.9%/year if the ejection fraction response was normal; the risk was 8.8%/year if the ejection fraction on exercise was less than 50%.4 We applied the same cut off to differentiate between patients with good and poor exercise responses; a fall in ejection fraction on exercise of less than 5% was taken as no change, as this value—used by Borer and colleagues—corresponded to more than 2 SD of our intraobserver variability for this measurement.

Dynamic exercise echocardiography cannot be performed in patients with musculoskeletal problems, and acquiring data within one minute of peak exercise is difficult in patients with poor echocardiographic windows or severe breathlessness. Confidence intervals for detecting changes in individuals by serial echocardiography are disappointing: the reported values are ± 15% for end diastolic volume, ± 25% for end systolic volume, and ± 10% for ejection fraction.20 It would therefore be helpful to describe resting variables that can identify left ventricular dysfunction.

Siemienczuk and colleagues showed that progression to surgery was earlier in patients with increased left ventricular volume indices at rest (left ventricular end diastolic index ⩾ 150 ml/m2and end systolic index ⩾ 60 ml/m²).3Although left ventricular volume indices were different in our study between patients with good and poor exercise responses, they had low sensitivity and specificity for predicting the exercise response.

It has been suggested that load adjusted variables should be more reliable for predicting outcome in asymptomatic patients with aortic regurgitation. These include the end systolic wall stress to end systolic volume index and the change in ejection fraction from rest to exercise normalised for change in end systolic wall stress.4 ,12 ,21 We did not find differences between the two groups for either end systolic wall stress or the end systolic wall stress to end systolic volume index. Change of ejection fraction from rest to exercise normalised for the change in end systolic wall stress was shown by Borer and colleagues to be a strong predictor of progression to heart failure in patients with aortic regurgitation.4 In our study, peak exercise data were acquired in less than one minute, so there was not enough time also to acquire a good quality M mode recording of the left ventricle from the parasternal view in order to calculate end systolic wall stress on exercise. Moreover, this index is complex, unlikely to be measured accurately or to be used in routine clinical practice, and its reproducibility has not been reported.4 ,12

LONG AXIS FUNCTION IN SEVERE AORTIC REGURGITATION

Severe and chronic left ventricular volume overload increases wall stress and may cause subendocardial ischaemia. This is the suggested precursor of deposition of collagen and subendocardial fibrosis, and a mechanism of left ventricular dysfunction in aortic regurgitation.8 Thus subendocardial dysfunction might provide a better marker of incipient left ventricular damage than the variables that have already been described. The subendocardial fibres are aligned longitudinally and connected with the mitral annulus.22 Long axis contraction results in the apical displacement of the mitral annulus, which can be measured by M mode echocardiography in terms of excursion, or by tissue Doppler echocardiography in terms of velocity.9 ,23-26 Myocardial velocities are less dependent on heart rate than M mode amplitudes, data acquisition is possible when endocardial visualisation is suboptimal, and in the apical four chamber view the angle of incidence is favourable for accurate measurements.27 ,28 In our study, long axis contraction assessed by both systolic excursion and peak systolic velocity was reduced in patients with aortic regurgitation who had poor exercise responses. However, peak systolic velocity had a higher sensitivity and specificity in predicting a poor exercise response than systolic excursion.

DIASTOLIC FUNCTION

Patients with severe aortic regurgitation have diastolic dysfunction, and this correlates with the degree of subendocardial fibrosis.8 ,29 ,30 A ratio of early to late velocity of mitral annular motion < 1, measured by tissue Doppler echocardiography, has good sensitivity and specificity (> 70%) for detecting diastolic dysfunction, even in patients with a pseudonormal E:A ratio.31

In our study, patients with poor exercise responses had a higher prevalence of a reversed E:A ratio, but this is difficult to interpret because they were also older than patients with good exercise responses. The changes could be caused by impaired relaxation of the left ventricle in early diastole in subjects with poor exercise responses, related to subendocardial dysfunction. The fact that there were diastolic abnormalities of long axis function but not of short axis function would support this. However, this hypothesis could only be proven by comparison with age matched controls.

CARDIOPULMONARY TESTING IN PATIENTS WITH AORTIC REGURGITATION

Maximum oxygen consumption and anaerobic threshold are relatively direct indices of cardiac functional reserve.14 ,15 In heart failure, maximum oxygen consumption correlates with symptoms, ejection fraction, and prognosis.14 ,32 The ventilatory cost of carbon dioxide elimination, reflected by the minute ventilation against carbon dioxide production (V˙e/V˙co ₂ slope), is increased in patients with heart failure.33 Therefore it is reasonable to suppose that these cardiopulmonary tests reflect left ventricular function in patients with isolated severe aortic regurgitation who have no respiratory disease.34

We did not find significant differences between patients with aortic regurgitation who had good or poor exercise responses for any of the cardiopulmonary tests investigated. A maximum oxygen consumption of ⩽ 20 ml/kg/min had a good specificity but a low sensitivity for predicting a poor exercise response. Maximum oxygen consumption correlated with change in ejection fraction from rest to exercise, and might be a useful adjunct to conventional exercise testing in asymptomatic patients with severe aortic regurgitation.

STUDY LIMITATIONS

In our study, patients with poor exercise responses were older than patients with good exercise responses. This might affect the comparison of long axis function, but systolic velocities of the myocardium have been reported to be independent of age.35

The single plane area–length method tends to overestimate left ventricular volumes and ejection fraction.36 Its advantages are that it is reliable for assessing ejection fraction even when this is less than 50%, and it has good reproducibility,37-39 as also demonstrated by our study. It is suitable for stress echocardiography, when a short acquisition time is crucial, because it needs only an apical four chamber view.

To assess long axis function we used only the medial mitral annulus, but in our experience this site gives the best Doppler alignment and has good intraobserver variability.40 It is possible that recording myocardial velocities during stress could add to the prognostic value of resting measurements, but this is hard to assess after dynamic exercise because of the need to record all the data quickly.

This study included a relatively small number of patients, but it is difficult in a single centre to identify many asymptomatic patients with both severe aortic regurgitation and normal resting left ventricular function. Nonetheless, long axis function was clearly different between the two groups, and so it may be a powerful predictor.

CONCLUSIONS

Asymptomatic patients with severe aortic regurgitation who have an abnormal ejection fraction response to exercise are accurately identified by measurements of left ventricular long axis contraction at rest. The new echocardiographic technique of tissue Doppler echocardiography may have the potential to reveal subclinical myocardial dysfunction and thereby to improve significantly the timing of aortic valve surgery in such patients. This should now be tested in long term, prospective studies.