Routine Adjustment of Doppler Echocardiographically Derived Aortic Valve Area Using a Previously Derived Equation to Account for the Effect of Pressure Recovery

Methods 

We reviewed 2-dimensional and Doppler echocardiography records along with cardiac catheterization records for 166 consecutive patients who underwent evaluation for AS and were found to have AVA between 0.6 and 2.0 cm2. Patients were included if the echocardiogram was performed within 3 months before or after the cardiac catheterization. Patients with moderate or severe aortic regurgitation or prosthetic aortic valves were excluded.

Two-Dimensional and Doppler Echocardiography 

The two-dimensional and Doppler imaging studies were performed with an echocardiography unit (Sonos 5500, Philips, Andover, MA). AVADop was calculated using the continuity equation. Because measurement of the ascending aorta is often difficult on transthoracic echocardiography, measurement of the aortic root at the level of the sinuses of Valsalva was used to estimate cross-sectional area of the proximal ascending aorta in equation 1 (Figure 1). The cross-sectional area of the aortic root was calculated from the measured aortic root diameter using π × (aortic root diameter/2)2. Although it is recognized that the aortic root is generally larger than the proximal ascending aorta, diameter measurements of the aortic root are highly reproducible and, thus, commonly reported on clinical echocardiograms.3

View Large Image Download to PowerPoint

Figure 1. Echocardiogram from parasternal long-axis window. Measurement of aortic root used in this study (A). Measurement of ascending aorta intended for use in equation 1 (B). Ao, Aorta; LA, left atrium; LV, left ventricle, RV, right ventricle.

Results 

Patient demographics and measured AVA are shown in Table 1. The mean difference between AVAcath and AVADop was 0.11 ± 0.23 cm2 (P < .001), as displayed in Figure 2, A. The mean difference between AVAcath and AVApredict was −0.01 ± 0.28 cm2 (P = .74) (Figure 2, B). AVAcath was larger than AVADop in 117 patients (70%).

Table 1. Demographics and valve areas for 166 patients

		Total N = 166	M N = 94	F N = 72	P (M vs F)
	Age, y	74±10	73±10	74±10	.36
	Echo-cath delay, days	6 [1, 18]	6.5 [3, 38]	6 [3, 34]	.91
	Cardiac output, L/min⁎	4.8±1.3	5.0±1.3	4.4±1.0	.004
	Ejection fraction	62 [45, 75]	62 [30, 75]	60 [35, 75]	.88
	AVAcath, cm2	0.98±0.30	1.01±0.32	0.94±0.28	.20
	AVADop, cm2	0.87±0.26	0.86±0.30	0.88±0.22	.68
	AVApredict, cm2	0.99±0.34	0.97±0.37	1.01±0.30	.42
	Mean gradient cath, mm Hg†	26±15	28±18	23±9	.23
	Mean gradient echo, mm Hg†	33±13	32±16	34±10	.55
	Aortic root diameter, cm	3.3±0.4	3.4±0.4	3.1±0.3	<.001

F, Female; M, male. Normal data are presented as the mean ± SD with sexes compared by t test. Nonnormal data are presented as the median [interquartile range] with sexes compared by Mann-Whitney.

⁎ Cardiac output data were available for 78 men and 54 women. † Mean gradient data were available from cath and echo for 34 men and 32 women.

View Large Image Download to PowerPoint

Figure 2. Bland-Altman plots displaying mean difference between aortic valve area (AVA) values obtained by cardiac catheterization (AVAcath) and AVA calculated by Doppler echocardiography (AVADop) (A) and between AVAcath and AVA that would be measured on catheterization (AVApredict) (B) for various values of AVAcath. Dashed lines, SD of observed mean difference.

ACC/AHA classification of AS severity based on AVAcath, AVADop, and AVApredict is shown in Table 2. Using AVAcath as a reference, no significant improvement was seen in interrater agreement using AVApredict (kappa 0.29 ± 0.06 [95% CI, 0.17-0.41]) instead of AVADop (kappa 0.29 ± 0.06 [95% CI, 0.17-0.41]) to classify AS severity, as evidenced by overlapping CIs. The proportion of patients categorized with severe AS was significantly different when assessment was based on AVAcath 60% (95% CI, 53%-67%) compared with assessment based on AVADop 81% (95% CI, 75%-87%, P < .001). No significant difference was seen between the proportion of patients categorized with severe AS based on AVAcath 60% (95% CI, 53%-67%) and AVApredict 63% (95% CI, 56%-71%, P = .57).

Table 2. Categorization of aortic stenosis severity based on two noninvasive measures of aortic valve area (AVADop and AVApredict) compared with invasively derived categorization using AVAcath

	Using AVAcath	Total	Using AVADop			Using AVApredict
			Mild	Mod	Severe	Mild	Mod	Severe
	Mild	11	4	6	1	5	5	1
	Moderate	55	1	14	40	7	22	26
	Severe	100	1	5	94	1	21	78
	Total	166	6	25	135	13	48	105

The mean aortic valve gradient was measured both by catheterization and echocardiography in 66 patients (Table 1). The mean aortic valve gradient was significantly higher when measured by echocardiography (P = .006). This discrepancy between modalities was mainly attributable to the difference seen in women (P < .001), because there was no significant difference seen in male patients (P = .37).

Discussion 

As predicted, the application of ACC/AHA criteria for AS severity using echocardiographically determined AVA overestimated the prevalence of severe AS compared with AVAcath. Routine adjustment of echocardiographically derived AVA to account for pressure recovery significantly improved the overall agreement between invasive and noninvasive measurement of AVA. This improved overall agreement was accomplished at the expense of increased variability in the catheter-echocardiography discrepancy in AVA. Stated differently, a simple shift in the data by adding 0.11 cm2 (the observed mean difference between AVAcath and AVADop) to the echocardiographically derived valve areas improves the agreement between these techniques better than applying the more complicated equation 1. This finding can be seen visually in Figure 1, as the wider SD shown in Figure 1, B, compared with Figure 1, A.

The overall agreement between noninvasive and invasively derived assessment of AS severity was not improved with routine application of equation 1, however, there was improved agreement in the number of patients assessed to have severe AS. This result is important because the distinction between moderate and severe disease is more clinically important than the distinction between mild and moderate disease. Consequently, understanding the impact of equation 1 on AVADop is important for clinical assessment of AS. A plot of AVApredict against AVADop for a range of theoretic AVA measurements for patients with differently sized aortic roots is shown in Figure 3. The theoretic values were chosen to reflect the range of aortic sizes and AVAs encountered in clinical practice. As shown, the impact of equation 1 on AVADop is minimal for patients with very small valve areas. Adjustment of echocardiographically derived AVA to account for the effect of pressure recovery has the greatest impact and should, therefore, be considered for patients with moderate or moderately severe AS.

View Large Image Download to PowerPoint

Figure 3. Relationship between aortic valve area (AVA) calculated by Doppler echocardiography (AVADop) and AVA that would be measured on catheterization using range of theoretic values for AVADop and varying aortic sizes (diameters 1.8-5.0 cm2).

Our study has several important limitations. Invasive and noninvasive assessment was not simultaneous. Agreement between values obtained by these two modalities has been shown to be worse when measurements were not acquired simultaneously.3, 4 Another limitation was that multiple operators were involved with the performance and interpretation of both the catheterization and echocardiographic data. Slight variation in practice patterns may have influenced the outcomes. For example, no data were available describing the methods used for measuring the aortic valve gradients. Some operators performed this measurement on a catheter pullback from the ventricle to the aorta, whereas others measured the gradient between the ventricle and the femoral artery. In addition, myocardial oxygen consumption was frequently estimated and not measured during the Fick calculation of cardiac output.

As previously mentioned, measurements of the aortic root were substituted for those of the ascending aorta into equation 1 because of improved reproducibility of measurements. Because the aortic root tends to be wider than the ascending aorta, application of equation 1 using this measurement may, therefore, underestimate and underadjust the echocardiographically derived valve area. Our results show, however, that the overall difference between AVAcath and the adjusted echocardiographically derived AVA was near zero. Therefore, had the smaller ascending aorta measurement been used, worse agreement between these techniques would have resulted. In addition, data regarding patient body size were not available and, therefore, the effect of pressure recovery could not be assessed relative to individual body habitus. Lastly, there are several other known factors, unrelated to pressure recovery, which are responsible for discrepancy between invasive and noninvasively derived AVA. Improper assumptions about the size and shape of the left ventricular outflow tract in the continuity equation and improper alignment of the Doppler beam with the AS jet are among the list of factors shown to cause such a discrepancy.3, 12 Assessment of the relative impact of each of these factors on agreement with invasive AVA measurement could not be assessed in the current investigation.

Although the aforementioned limitations may cast some degree of doubt about the accuracy of the data acquired, it is important to remember that equation 1 has been previously validated under more idealized experimental conditions.2 The intent of the current investigation was to test the ability of equation 1 to improve agreement between invasive and noninvasive measures of AVA in clinical practice. In this regard, our findings are supportive of its use in the appropriate clinical context. In general, the decision to perform surgical aortic valve replacement is made based on clinical symptoms. For patients in whom symptoms are equivocal for severe AS and discrepancy exists between measured AVAcath and AVADop, assessment of the size of the ascending aorta and impact of pressure recovery should be entertained.