Estimation of the End of Ejection in Aortic Stenosis
An Unreported Source of Error in the Invasive Assessment of Severity
Background— All indices of aortic stenosis (AS) rely on measurements of mean transvalvular pressure gradient (ΔP) and flow rate. Because the gradient is reversed during late ejection, the late systolic left ventricular (LV)–aortic pressure crossover may be an erroneous landmark of end-ejection. The aortic incisura should be a better reference to calculate indices of AS invasively.
Methods and Results— The accuracy of the pressure crossover and the incisura to define end-ejection was assessed in a chronic AS experimental model (9 dogs) with the use of an implantable flowmeter and Doppler echocardiography as reference. In 288 hemodynamic recordings analyzed (aortic valve area [AVA]: 0.74±0.46 cm2), ejection ended 37±29 ms after the pressure crossover but almost simultaneously with the incisura (2±17 ms). Pressure crossover error accounted for significant errors in the measurement of ΔP (95% limits of agreement, +0 to +7 mm Hg) and AVA (−0.1 to +0.2 cm2). These errors were reduced to less than half with the use of the incisura to define end-ejection. Additionally, the agreement with Doppler-derived AS indices was best with use of the incisura. Pressure crossover error was maximal in situations of higher output, moderate orifice narrowing, higher arterial compliance, and lower vascular resistance. In 32 consecutive patients undergoing cardiac catheterization for AS, the pressure crossover induced a clinically important overestimation of the ΔP from +22 to +50%. Errors in AVA estimation were considerably smaller (−2% to +6%) because of simultaneous and offsetting errors in the measurements of ΔP and flow.
Conclusions— The aortic incisura and not the second pressure crossover should be used to obtain invasive indices of AS.
Received January 31, 2004; revision received March 30, 2004; accepted March 31, 2004.
Cardiac catheterization is established as the gold standard technique for quantifying aortic stenosis (AS), and noninvasive methods have been validated with catheterization used as reference.1 Invasive assessment of AS relies on accurate measurements of mean systolic ejection transvalvular pressure gradient (ΔP) and mean ejection flow rate (), from which aortic valve area (AVA) is calculated.1 Obviously, both ΔP and need to be measured throughout the full systolic ejection period (SEP).2
Surprisingly, the method for measuring SEP has not been definitely established in AS. Ejection starts when intraventricular pressure rises above aortic pressure and ends with aortic valve closure. Most textbooks recommend using the second left ventricular (LV)–aortic pressure crossover to identify end-ejection when aortic and LV pressures are obtained simultaneously.3–5 However, we6 and others7,8 have demonstrated a reversed ΔP between the LV and the aorta during end-ejection in AS. Thus, measuring the SEP from the second LV-aortic pressure crossover may be erroneous. The aortic incisura may be a more accurate landmark of end-ejection,9 and this criterion is used by some authors to calculate .10–12 However, whatever landmark is used to define end-ejection, not only flow but also ΔP should be averaged for the full SEP.
The present study was designed to clarify which landmark of end-ejection should be used for the most reliable assessment of AS during cardiac catheterization. An experimental animal model of chronic AS was used in which high-fidelity pressure, flow, and Doppler echocardiographic data were obtained. The implications of the animal findings were then assessed in 32 consecutive patients undergoing cardiac catheterization for AS.
Animals and Surgical Protocol
Nine mongrel dogs (weight, 18 to 24 kg) of both sexes underwent surgery to create a chronic model of degenerative valvular AS characterized by stiff leaflets without commissural fusion.6,13,14 Studies were approved by the University of Washington Institutional Animal Care Committee and conform with the Guide for the Care and Use of Laboratory Animals. An implantable 16-mm transit-time flow probe (Transonics) was placed immediately above the sinotubular junction. Table 1 summarizes the hemodynamic data.
Animal Hemodynamic Studies
Postoperatively, 3 cardiac catheterization studies were performed in each animal, under general anesthesia, at 2-week intervals. A high-fidelity, dual-micromanometer (5 cm apart) catheter (Millar Instruments) was inserted across the aortic valve via the carotid artery. Transvalvular was measured with a factory-calibrated T101D Transonics flowmeter. Flow, pressure signals, and a lead II ECG were digitized simultaneously at 200 Hz over a 30- to 60-second period. Cardiac output was subsequently varied with saline and dobutamine, and measurements were repeated when hemodynamics stabilized.
Digital Signal Analysis of Hemodynamic Data
Signal processing was performed with the use of custom-developed algorithms in Matlab (version 6.2, The Mathworks, Inc). Measurements were performed on an individual beat basis and then averaged for the whole 30- to 60-second run. Beats with a cardiac cycle variation >10% were rejected.
Onset of ejection was established as the first pressure crossover of the LV and aortic pressure signals.6 Reference end-ejection was established as the second zero crossover of the flow signal (Figure 1). Thus, reference SEP was calculated as the difference between these 2 time instants. Against this reference, we ascertained the accuracy of 3 end-ejection criteria: (1) the second crossover of the LV-aortic pressures (detected automatically); (2) the valley of the aortic incisura (detected manually, blinded to the flow and LV pressure signals); and (3) the time of the peak negative first temporal derivative of the pressure signal (T−dP/dt min, detected automatically).15 Interobserver variability of the identification of the incisura (75 randomly selected beats) was 3±23 ms. Stroke volume was obtained as the positive time integral of flow during ejection, and was obtained as stroke volume divided by the SEP. Similarly, instantaneous ΔP (LV−aortic pressure) was averaged for each SEP to obtain ΔP. AVA was computed according to the Gorlin formula with the use of the following: (1) ΔPflow probe and flow probe (reference); (2) ΔPpressure crossover and pressure crossover; (3) ΔPincisura and incisura; and (4) ΔPpressure crossover and incisura. This latter method for measuring AVA10,12,16 was designated AVAcombined.
Total systemic arterial compliance was calculated from the aortic pressure signal by the area method.17 The Strouhal number was calculated as follows:
where Qmax=peak ejection flow rate.6,18 The Strouhal number is a dimensionless index used to describe unsteady flow systems and accounts for the relative contribution of inertial and convective forces.8,18 In AS, the DP is the consequence of adding the forces related to (1) frictional energy losses (dissipated as heat); (2) local inertial acceleration (the change in velocity with time in a given point, due to ventricular pulsatility); and (3) convective acceleration (the change in velocity with space at a given instant, due to lumen tapering caused by the valvular stenosis). The Strouhal number accounts for the ratio between the latter 2 components. Thus, a value >1 reflects predominance of inertial forces (eg, a normal subject), whereas a value close to 0 reflects almost exclusive convective forces (eg, a critical orifice narrowing; see Discussion).8,18
In 7 animals, pulsed- and continuous-wave Doppler spectrograms of the LV outflow tract (LVOT) and AS (Ao) jets were obtained from the apical 5-chamber view.13,14 The ΔP at the level of the vena contracta was obtained from mean velocities () as ΔP=4 (2Ao−2LVOT). Doppler AVA was obtained with the use of the continuity equation as stroke volume divided by the AS time-velocity integral. To allow comparison with invasive parameters, Doppler indices of AS were corrected for the effect of pressure recovery, calculating the net ΔP (LV−ascending aortic pressure) and the energy-loss coefficient (physiologically equivalent to Gorlin-derived AVA).19,20
Hemodynamic data from 38 consecutive patients undergoing cardiac catheterization for AS were studied. Simultaneous LV and aortic pressures were recorded with the use of fluid-filled, 8F double-lumen pigtail catheters (Cordis Corp). Cardiac output was obtained by the Fick and thermodilution methods and averaged. High-quality paper pressure tracings were automatically digitized (Digitize-Pro Software, version 4.1) at 1000 Hz and processed in a manner identical to that of the animal signals. Six patients were excluded because of suboptimal superimposition of the 2 waveforms seen while both ports resided within the aorta, and therefore 32 patients (31 male; aged 73±7 years; LV ejection fraction=0.54±0.18) are the basis of this report. The aortic incisura could not be identified clearly in 19 subjects because of overdamping. In 9 patients, the clear point of slope change of the aortic pressure recording was used to identify end-ejection.21 In the remaining 10 patients, the T−dP/dt min of the LV signal was used to estimate zero systolic flow. Because T−dP/dt min is close but not simultaneous to end-ejection,21 this landmark was first calibrated by linear regression in the subset of patients showing a clear visualization of the incisura (Tincisura=1.232 · T−dP/dt min −0.038; R=0.96, n=36 beats).
Data are presented as median±interquartile range. Agreement between indices was assessed by the intraclass correlation coefficient (Ric) and Bland-Altman analysis. Errors are reported as mean±SD (limits of agreement). The analysis of the determinants of SEP error was based on prior evidence suggesting a relationship with the Strouhal number,6,8,18 arterial compliance, and vascular resistance.22 Hence, this association was assessed by multivariate regression accounting for factor interaction where the Strouhal number was first fitted to a 3-knot restricted cubic spline function. All analyses were performed with S-plus software (Insightful, version 2000), expanded by public-domain libraries.23 A probability value <0.05 was considered significant.
Accuracy of End-Ejection Criteria and Impact on AS Indices
End-ejection took place 37±29 ms (range, 0 to 146 ms) after the second LV-aortic pressure crossover (288 hemodynamic data sets, 2532 beats; Figure 1). The pressure crossover was a very inaccurate criterion to measure the SEP, and agreement with the reference was far better for the SEPincisura (Figure 2). Reference SEP could also be closely predicted by the T−dP/dt min. Compared with Doppler-derived SEP, SEPincisura was also more accurate (error=−6±8%, Ric=0.82) than the SEPpressure crossover (error=+14±11%, Ric=0.53).
The SEPpressure crossover error caused significant bias and imprecision in the estimation of ΔP and (Table 2). These errors in ΔPpressure crossover and pressure crossover partially balanced each other in the calculation of AVApressure crossover. The incisura criterion highly improved the accuracy of ΔP and measurements, yielding a more accurate AVAincisura. The most inaccurate method of measuring AVA was AVAcombined. Compared with Doppler-derived indices, agreement was also better for the incisura than for the pressure crossover (ΔP: error=−5±49%, Ric=0.74, versus −21±47%, Ric=0.68; AVA: error=−3±20%, Ric=0.92, versus −12±21%, Ric=0.88, respectively).
Hemodynamic Predictors of Error
Relative SEPpressure crossover error correlated with a higher Strouhal number (total, interaction, and nonlinear factors: all P<0.0001), a higher arterial compliance (total and interaction factors: P<0.0001), and a lower vascular resistance (total and interaction factors: P<0.0001; adjusted R2 for the final model=0.70). The slope of SEP error was very steep at low Strouhal numbers, particularly when arterial compliance was high and vascular resistance was low (Figure 3).
Differences in AS indices due to the method used to estimate end-ejection are summarized in Table 3 and Figure 4. As shown, AVAcombined may cause a >33% underestimation of AVA. Errors in the 13 patients showing a clear identification of the incisura were similar to those of the full group (P>0.4 for all AS indices).
To assess the severity of AS, current textbooks recommend measuring ΔP by averaging the area inside the LV and aortic pressure tracings.3–5,11 Mean flow rate is calculated by dividing stroke volume by the SEP, measured as the time between the first and second LV-aortic pressure crossover. Our study demonstrates potential errors in this methodology. First, during end-ejection, the ΔP between the LV and the aorta is frequently reversed. The planimetry method for measuring ΔP will therefore overestimate the total ejection gradient in such cases (up to +31%). Second, for the same reason, SEP measured from the pressure crossover may be shorter than actual ejection period, thus also overestimating . Because the errors in ΔP and partially compensate each other in the Gorlin formula, the inaccuracy in estimating AVA is smaller (+6% error) but may increase to as much as −33% error when real ejection is combined with the ΔPpressure crossover to calculate AVA.11 These errors in quantifying the severity of AS may become clinically relevant in situations in which there are discrepant noninvasive findings or associated coronary heart disease or in the setting of low-flow AS (see below).
Systolic Ejection Period in AS
Although never considered in the clinical assessment of AS, the fact that forward flow continues in the systemic circulation despite a negative ΔP was recognized several years ago.24,25 The term hangout interval was coined to designate the time interval between the LV pressure (measured not at the time of pressure crossover but at the pressure value of the incisura) and the aortic incisura.7 The basis of the reverse ejection gradients was further established in humans with the use of high-fidelity micromanometers.26 End-ejection ΔPs have been also characterized in AS. Clark8 demonstrated reverse ΔPs in an experimental model of supravalvular AS, and Shaver22 reported a wide hangout interval in a patient with severe AS. The latter author further suggested a relationship of the hangout interval with aortic compliance.22 The present study demonstrates that these concepts of basic ejection hemodynamics need to be incorporated in the clinical assessment of AS.
As expected,6,8,18 the Strouhal number was the most important variable related to SEP error. In very tight orifice stenoses, convective forces predominate, the Strouhal number is close to 0.05,1 and SEP-related error is small. However, in AS there may be situations in which inertial forces cannot be neglected. If inertial acceleration increases, and orifice narrowing is moderate, the Strouhal number rises to 0.1 to 0.2. In this range, SEP error is dramatically augmented, particularly in patients with high total systemic arterial compliance and low systemic resistance (Figure 3). It is precisely these patients with an “intermediate” degree of valve narrowing who require an accurate determination of AS severity, because arriving at the correct clinical decision on the necessity of valve surgery can be difficult and challenging.
Clinical Implications: Moderate Disease and Low-Flow AS
A mild or moderately reduced valve orifice associated with a hyperdynamic state constitutes a typical scenario resulting in a relatively high Strouhal number. Inotropic stimulation during cardiac catheterization has been advocated to ascertain the severity of AS in patients with impaired systolic function.1 In the presence of contractile reserve, dobutamine markedly increases inertial forces.6,27 Consequently, the Strouhal number rises, and transvalvular pressure reversal is expected. Accounting only for the positive component of ΔP may therefore induce significant overestimation of disease severity in the hypercontractile state. An illustrative example of this phenomenon is demonstrated in a recent study by Nishimura et al28 on the value of dobutamine challenge for assessing the severity of low-flow AS. In the example shown in Figure 2B of their article, the authors display the tracings of a patient with only mild AS at the time of operation. Reported values of ΔP at rest and peak stress are 17 and 20 mm Hg, respectively. Of note, the baseline tracing shows almost no SEP error. However, with dobutamine infusion, the pressure crossover anticipates end-ejection by >50%, as determined by the aortic incisura. Consequently, the “actual” ejection ΔP during dobutamine infusion is reduced to only 10 mm Hg. This drop in ΔP is inconsistent with “fixed” severe AS and clearly identifies the patient as having pseudosevere AS. Thus, we believe that this example illustrates the advantages of using the aortic incisura.
The absence of an independent gold standard method for assessing AS severity is a limitation of all in vivo studies of AS hemodynamics. Pulse wave propagation from the distal to the proximal measuring stations may cause a certain delay in pressure recordings between the transducers. The fact that the first LV-aortic pressure crossover closely matches flow probe–defined ejection in our experimental model6 corroborates that correction for pulse wave propagation is not necessary. As found in our clinical study, the incisura may be difficult to identify from aortic pressure tracings obtained with the use of fluid-filled catheters. In those cases in which the incisura remains impossible to recognize despite careful catheter damping, high-fidelity catheters and/or the T−dP/dt min may be a suitable alternative. Whether an exact identification of end-systole is amenable by processing LV-aortic pullback pressure recordings deserves further investigation. Unfortunately, the methods proposed in our study are not suitable if the femoral artery sheath pressure is used as a surrogate of central aortic pressure.
The potential existence of reverse pressure gradients during end-ejection should be taken into consideration when the severity of AS is assessed invasively, particularly in patients with moderate orifice narrowing and increased cardiac output. Therefore, end-ejection landmarks such as the aortic incisura, and not the LV-aortic pressure crossover, should be used to measure the SEP and calculate ; additionally, the ΔP must be averaged up to the aortic incisura to measure the full ejection ΔP. These considerations improve the invasive assessment of disease severity.
This study was supported in part by the SCIT 2000-2003 Contract Program of the Comunidad de Madrid to Dr Antoranz and by a research grant (BF03/00031) of the Fondo de Investigación Sanitaria, Instituto Carlos III, Madrid, Spain, to Dr Yotti.
Presented in part at the 2001 Scientific Sessions of the American Heart Association, Anaheim, Calif, November 11–14, 2001, and published in abstract form (Circulation. 2001;104[suppl II]: II-361).
Bonow RO, Carabello B, de Leon AC Jr, et al. Guidelines for the management of patients with valvular heart disease: executive summary: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Committee on Management of Patients with Valvular Heart Disease). Circulation. 1998; 98: 1949–1984.
Wilson RF. Catheterization of patients with aortic valve disease. In: Emery RW, Arom KV, eds. The Aortic Valve. Philadelphia, Pa: Hanley and Belfus, Inc; 1991: 57–78.
Lange RA, Hillis LD. Cardiac catheterization and hemodynamic assessment. In: Topol EJ, ed. Textbook of Cardiovascular Medicine. 2nd ed. Philadelphia, Pa: Lippincott Williams & Wilkins; 2002: 1957–1976.
Kern MJ, Deligonul U, Donohue EC, et al. Hemodynamic data. In: Kern MJ, ed. The Cardiac Catheterization Handbook. 2nd ed. St Louis, Mo: Mosby; 1995: 108–207.
Shaver JA, Nadolny RA, O’Toole JD, et al. Sound pressure correlates of the second heart sound: an intracardiac sound study. Circulation. 1974; 49: 316–325.
Laniado S, Yellin E, Terdiman R, et al. Hemodynamic correlates of the normal aortic valve echogram. Circulation. 1976; 54: 729–737.
Bache RJ, Wang Y, Greenfield JC Jr. Left ventricular ejection time in valvular aortic stenosis. Circulation. 1973; 47: 527–533.
Carabello BA, Grossman W. Calculation of stenotic valve orifice area. In: Grossman W, Baim DS, eds. Cardiac Catheterization, Angiography and Intervention. 4th ed. Philadelphia, Pa: Lea & Febiger; 1991: 152–165.
Kligfield P, Goldberg H, Kline SA, et al. Effect of additional valve lesions on left ventricular ejection time in aortic stenosis. Br Heart J. 1977; 39: 1259–1264.
Burwash IG, Thomas DD, Sadahiro M, et al. Dependence of Gorlin formula and continuity equation valve areas on transvalvular volume flow rate in valvular aortic stenosis. Circulation. 1994; 89: 827–835.
Garcia D, Dumesnil JG, Durand LG, et al. Discrepancies between catheter and Doppler estimates of valve effective orifice area can be predicted from the pressure recovery phenomenon: practical implications with regard to quantification of aortic stenosis severity. J Am Coll Cardiol. 2003; 41: 435–442.
Brickner ME, Starling MR. Dissociation of end systole from end ejection in patients with long-term mitral regurgitation. Circulation. 1990; 81: 1277–1286.
Harrell FE. Hmisc and Design S-Plus function library. Public-domain software. Available at: http://biostat.mc.vanderbilt.edu/twiki/bin/view/Main/Design. Accessed August 4, 2004.
Noble MI. The contribution of blood momentum to left ventricular ejection in the dog. Circ Res. 1968; 23: 663–670.
Spencer MP, Greiss FC. Dynamics of ventricular ejection. Circ Res. 1962; 10: 274–279.
Murgo JP, Altobelli SA, Dorethy JF, et al. Normal ventricular ejection dynamics in man during rest and exercise. Am Heart Assoc Monogr. 1975; 46: 92–101.
Nishimura RA, Grantham JA, Connolly HM, et al. Low-output, low-gradient aortic stenosis in patients with depressed left ventricular systolic function: the clinical utility of the dobutamine challenge in the catheterization laboratory. Circulation. 2002; 106: 809–813.