Author + information
- Received June 8, 2011
- Revision received August 2, 2011
- Accepted August 18, 2011
- Published online November 1, 2011.
- Sagit Ben Zekry, MD⁎,
- Robert M. Saad, MD⁎,
- Mehmet Özkan, MD†,
- Maie S. Al Shahid, MD‡,
- Mauro Pepi, MD§,
- Manuela Muratori, MD§,
- Jiaqiong Xu, PhD∥,
- Stephen H. Little, MD⁎ and
- William A. Zoghbi, MD⁎,⁎ ()
- ↵⁎Reprint requests and correspondence:
Dr. William A. Zoghbi, Cardiovascular Imaging Institute, The Methodist DeBakey Heart and Vascular Center, 6550 Fannin Street, SM 677, Houston, Texas 77030
Objectives We sought to evaluate whether ejection dynamics, particularly acceleration time (AT) and the ratio of AT to ejection time (ET), can differentiate prosthetic aortic valve (PAV) stenosis from controls and prosthesis–patient mismatch (PPM).
Background Diagnosing PAV stenosis, especially in mechanical valves, may be challenging and has significant clinical implications.
Methods Doppler echocardiography was quantitated in 88 patients with PAV (44 mechanical and 44 bioprosthetic; age 63 ± 16 years; valve size range 18 to 25 mm) of whom 22 patients had documented PAV stenosis, 22 had PPM, and 44 served as controls. Quantitative Doppler parameters included ejection dynamics (AT, ET, and AT/ET) and conventional PAV parameters.
Results Patients with PAV stenosis had significantly lower effective orifice area (EOA) values and higher gradients compared with controls and PPM. Flow ejection parameters (AT and AT/ET) were significantly longer in the stenotic valves compared with PPM and controls (respective values for AT: 120 ± 24 ms, 89 ± 16 ms, and 71 ± 15 ms; for AT/ET: 0.4, 0.32, and 0.3, p ≤ 0.001). Patients with PPM had gradients and ejection dynamics that were intermediate between normal and stenotic valves. Receiver-operator characteristic (ROC) curve analysis showed that AT and AT/ET discriminated PAV stenosis from PPM and controls (area under ROC curve = 0.92 and 0.88, respectively). Combining AT with the conventional Doppler velocity index gave the highest area under the curve of 0.98 but was not statistically different from that of AT alone (p = 0.12). A cutoff of AT = 100 ms had a sensitivity and specificity of 86% for identifying PAV stenosis; for an AT/ET = 0.37, the sensitivity and specificity were 96% and 82%, respectively. Analysis by valve type (mechanical and biological) revealed similar results; however, biological valves had slightly higher areas under the curve for all systolic time intervals.
Conclusions Ejection dynamics through PAV, particularly AT and AT/ET, are reliable angle-independent parameters that can help evaluate valve function and identify PAV stenosis.
The assessment of prosthetic aortic valve (PAV) function remains a challenge. This stems from the variability in the measured pressure gradients, which depends on flow, in addition to the wide variety of prosthetic valve types and sizes (1-14). Derivation of an effective orifice area (EOA) provides a parameter that is less dependent on flow but still relies on knowledge of valve size and type (1,2,9,12). Clinically, knowledge of valve size and type may not be readily available during an echocardiographic examination, a lingering challenge in evaluation of prosthetic valve function. A simple parameter that complements assessment of PAV function and that could overcome these limitations would be desirable.
In native aortic stenosis, early studies have described changes in ejection flow dynamics across the valve that occur with worsening severity of stenosis: a prolongation of ejection time (ET), delayed peak velocity, or acceleration time (AT) with a resultant rounded contour of the Doppler jet velocity profile. These systolic time intervals correlated with the severity of the stenotic valves (7,8,11). We postulated that PAV ejection dynamics, particularly acceleration time, and the ratio of AT to ET can differentiate PAV stenosis from patients with normal prosthetic valve function and those with prosthesis–patient mismatch (PPM).
Echocardiographic and Doppler studies of patients with PAV were reviewed from the database of The Methodist Hospital, Houston, Texas; Koşuyolu Kartal Heart Training and Research Hospital, Istanbul, Turkey; King Faisal Specialist Hospital and Research Center, Riyadh, Saudi Arabia; and Centro Cardiologico Monzino, IRCCS, Milan, Italy, to identify patients with suspected PAV stenosis, as the incidence of this entity is uncommon, particularly in mechanical valves. Patients were identified as having PAV stenosis if they had a high pressure gradient, a reduced EOA below 1 cm2 (11,12), as well as confirmation of PAV stenosis at surgery or, in the very few cases of patients with mechanical valves not undergoing surgery, resolution of the high gradient and improvement in EOA after thrombolytic therapy. Another cohort with a similar number of patients was identified from The Methodist Hospital database with severe PPM. These were defined as patients with normal functioning PAV, high peak velocity (>3 m/s), and mean gradient (>25 mm Hg), normal EOA for the valve type and size but significantly reduced indexed EOA to body surface area (BSA), <0.65 cm2/m2 (12). Patients with PPM also had further diagnostic studies (either with transesophageal echocardiography or fluoroscopy) to confirm no obstruction of the valve. Last, a third cohort of individuals with normal PAV function served as controls. These patients were chosen to match those with PAV stenosis (2:1) with respect to age, sex, and valve type. Normal valve function was defined as having a normal pressure gradient and EOA in the reported range for the valve type and size (9,12), without evidence of regurgitation. To further ensure normality of function, these patients had to have their echo/Doppler studies within 3 months after their aortic valve replacement and with no signs or symptoms of PAV dysfunction. Patients with more than mild PAV regurgitation were excluded from analysis, as were those with concomitant prosthetic mitral valves or poor image quality.
Two-dimensional transthoracic echocardiographic and Doppler studies were obtained with clinical ultrasound machines equipped with 2.5 to 3.5 MHz transducers using standard views. Pulsed Doppler in the left ventricular outflow tract (LVOT) from the apical window was used to evaluate flow. Continuous wave Doppler recording of flow through the PAV was performed from the apical, right parasternal, suprasternal, and subcostal windows to minimize the effect of Doppler angulation with flow. Doppler recordings were performed at a sweep speed of 100 mm/s.
The studies from all centers were sent to the Methodist Hospital, and then reviewed and quantitated by a single observer (S.B.Z.) on an offline station (Digisonics Inc., Houston, Texas). All measurements represent an average of 3 cardiac cycles for patients in sinus rhythm and at least 6 cardiac cycles for patients in atrial fibrillation.
Standard echocardiographic and Doppler measurements
Parasternal long-axis view was used for measuring the aortic annulus diameter in early systole, just below the insertion of the PAV. Using the pulsed wave Doppler recording in the LVOT, the time-velocity integral (TVI) was calculated. Stroke volume in the LVOT was derived using the TVILVOT, assuming a circular geometry of the LVOT. Left ventricular function was estimated by the reader. From the continuous wave jet recording, peak velocity was measured and mean gradient was calculated using the modified Bernoulli equation. The TVI of the aortic jet by continuous wave was also measured to derive an effective orifice area (EOA) using the continuity equation (1,2) as (stroke volume)/TVIjet. An indexed EOA to BSA was calculated as EOA/BSA. A Doppler velocity index (DVI), a simplification of the continuity equation (1) was calculated as TVILVOT/TVIjet.
Systolic time intervals
The systolic time intervals of flow through the PAV were measured using the velocity curve from the continuous wave Doppler recording (Fig. 1). ET was measured as the time from onset to end of systolic flow across the prosthetic valve. The time interval from the beginning of systolic flow to its peak velocity was defined as AT. The ratio of AT/ET was then calculated.
Intraobserver and interobserver variability
To assess the intraobserver and interobserver variability of systolic time intervals, a total of 15 studies were quantitated by the same observer on 2 different occasions (from the same echocardiographic study, not necessarily the same beats) in a random order and by another observer. The 15 cases included 5 controls, 5 patients with PPM, and 5 with prosthetic valve obstruction; to have a wide range of values. Variability was expressed as the difference between the measurements and percent error (difference in measurement divided by the mean value of the 2 observations, expressed as percent). Actual values were also correlated using linear regression analysis.
Continuous variables are presented as mean ± SD, and categorical variables as numbers (percentages). One-way analysis of variance was used with post hoc Tukey honestly significant difference (HSD) test adjustment for multiple comparisons among the different groups. Categorical variables were compared using chi-square test. A receiver-operator characteristic (ROC) curve was plotted to determine the best AT, ET, and AT/ET cutoff value for identifying PAV stenosis. This cutoff was determined as the value providing a balance between sensitivity and specificity, or at most, a slightly higher sensitivity because of the clinical implication of detecting valve stenosis. The area under the ROC curve was calculated. The areas under 2 or more correlated ROC curves were compared (15). Multiple linear regression models were developed to study the relationships between systolic time intervals (AT, ET, AT/ET) and the following independent variables: EOA, heart rate, stroke volume, ejection fraction, age, and sex. Statistical analysis was performed with SPSS version 15.0 (SPSS Inc., Chicago, Illinois) and STATA version 10 (College Station, Texas). A p value <0.05 was considered significant.
Baseline patient characteristics are summarized in Table 1. The 3 groups were similar in age, sex, systemic hemodynamics, and rhythm. All mechanical valves were bileaflet valves. A wide range of valve sizes was observed in all groups; compared with controls, patients with mismatch had smaller valve sizes, larger stroke volumes, and larger body surface areas.
Conventional parameters of valve function are shown in Figures 2, 3, and 4⇓⇓. In comparison with patients with normal valves and those with PPM, patients with PAV stenosis had significantly lower EOA and higher pressure gradients across the prosthetic valve (Figs. 2 and 3). Similarly, the lowest DVI values were noted in patients with PAV stenosis (Fig. 4). Using a cutoff of 0.25, DVI had a sensitivity, specificity, positive and negative predictive values, and accuracy of 59%, 100%, 100%, 88%, and 90%, respectively.
Systolic time intervals
Ejection parameters for patients with normal PAV function, PPM, and stenotic valves are presented in Figure 5. Stenotic valves had significantly longer AT and higher AT/ET compared with PPM and normal valves. Of interest is that AT/ET was similar between patients with normal valves and those with PPM; both significantly lower than in PAV stenosis. In contrast, ejection time was longer in PPM and stenotic PAV compared with normal valves, but was not different between PPM and stenotic PAVs (Fig. 5). Similar results were obtained if patients with PAV stenosis and very high gradients (peak gradient >100 mm Hg) were excluded. ROC analysis (Fig. 6) showed that AT, ET, and AT/ET could discriminate PAV stenosis from controls and those with PPM. The largest area under the ROC curve for systolic time intervals was seen with AT (0.92) followed by AT/ET (0.88). ET had the lowest area under the ROC curve (0.73). The area under the ROC curve for DVI was 0.96. Combining AT with the conventional DVI gave the highest area under the curve of 0.98 but was not statistically different from that of AT alone (p = 0.12). The AT/ET ratio did not add incremental value over the combination of AT and DVI. Table 2 summarizes the analysis and lists the derived best cutoff values for systolic time intervals, balancing sensitivity and specificity for valve stenosis as well as those cutoff values with 100% specificity of PAV stenosis. A cutoff of AT = 100 ms had a sensitivity and specificity of 86% for PAV stenosis. An AT/ET ratio of 0.37 allowed the identification of stenotic PAV with 96% sensitivity and 82% specificity. An AT >128 ms and AT/ET of >0.58 were 100% specific for PAV stenosis.
A ROC curve analysis of systolic time intervals was also performed for the detection of PAV stenosis by valve type (mechanical and biological) (Table 3). In general, similar results were observed, and biological valves had slightly higher area under curve for all systolic time intervals.
Multiple linear regression analysis
Multiple linear regression analysis was performed to evaluate the determinants of each of the systolic time intervals (AT, ET, and AT/ET) and their relation to EOA, stroke volume, heart rate, ejection fraction, age, and sex. Results of the models are shown in Tables 4, 5, and 6.⇓⇓⇓ The AT and ET models had both an R2 of 0.64, accounting for 64% of the observed variability in these parameters. The AT/ET model was less strong, with an R2 of 0.35. For all systolic time intervals, the most significant determining variable was EOA (largest coefficient; quadratic relation), followed by stroke volume; ejection fraction was less of a determinant and only for AT.
Intraobserver and interobserver variability
Measurements of AT, ET, and AT/ET correlated well when repeated by the same observer: correlation coefficients r = 0.98, 0.99, and 0.94, respectively. Correlations between the 2 observers for AT, ET, and AT/ET were 0.91, 0.94, and 0.88, respectively. Mean differences between measurements of AT, ET, and AT/ET for the same observer were 0 ± 7 ms, 1 ± 8 ms, and 0.00 ± 0.02, respectively, and between observers were 2 ± 15 ms, −9 ± 17 ms, and 0.01 ± 0.04, respectively. Mean percent error in measurements of AT, ET, and AT/ET for the same observer were −0.5 ± 7%, 0.5 ± 3%, and 1 ± 7%, respectively, and between observers were 1.9 ± 11.6%, −3.8 ± 6.4%, and 5.8 ± 9.6%, respectively.
In the current study, we have shown that ejection parameters can differentiate patients with stenotic PAV from those with normal prosthetic valve function and those with PPM. An AT >100 ms and AT/ET over 0.37 can identify PAV stenosis with good accuracy. These simple measurements complement conventional parameters of prosthetic valve function, offering an angle-independent Doppler measure of valve function.
Prosthetic aortic valve stenosis
Similar to native valve stenosis, the velocity and gradient across PAVs increase with worsening severity of obstruction, along with prolongation of the ejection time and delay in reaching the maximal velocity across the valve (prolonged AT), leading to a rounded jet velocity contour (Fig. 1). These ejection dynamics, in contrast to assessment of gradients, are angle-independent and relatively simple to acquire and measure. Earlier, Rothbart et al. (11) reported a significantly longer AT (116 ± 15 ms) in patients with stenotic bioprosthetic PAV. Minimal correlation between AT and valve size was found; the study included mostly valve sizes above 23 mm. The authors offered 4 variables to identify normal PAV function: AT (≤100 ms), peak gradient (<48 mm Hg), mean gradient (<25 mm Hg), and EOA (>1.1 cm2). Two abnormal measurements were consistent with stenotic bioprosthetic PAV (11). The present investigation corroborates these findings and extends these observations to mechanical prosthetic valves. Our results provide further support to the new guidelines for evaluation of prosthetic valves—both mechanical and bioprosthetic—which were based on the mentioned observations in bioprosthetic valves and our preliminary observations in mechanical valves (12). Similar to what the guidelines postulated (12) and now hereby validated, a combination of AT and DVI gives an excellent assessment of valve function, differentiating valve stenosis from PPM and normal prostheses (area under the ROC curve = 0.98). In addition to AT, the ratio AT/ET was also found to be a helpful parameter to discriminate stenotic PAV from both normal and PPM valves. PAV with better hemodynamic profiles were previously reported to have lower AT/ET values (3). The present cohort included a wide range of valve sizes (19 to 25 mm), making these determinations applicable in a wide range of clinically implanted valves. Last, an advantage of systolic time intervals is that they are independent of ultrasound beam angulation with jet velocity, an inherent factor or possible limitation in all other measures of prosthetic valve function, such as velocity, gradient, EOA, and DVI.
PPM is an important cause of elevated velocity and gradients across normally functioning prosthetic valves. Calculated EOA is in the normal range for the type and size of the particular prosthetic valve, but is too small for the selected patient (reduced EOA/BSA). In bileaflet mechanical valves, EOA may be smaller than that derived hemodynamically because of localization of velocity and gradient between the 2 leaflets (12). However, normal values already incorporate this phenomenon. It is crucial to distinguish PPM from stenotic PAV. In the present study, patients with PPM had several parameters (gradients, DVI, EOA) intermediate between controls and those with severe stenosis. Similar observations were seen with AT measurements, probably because of higher stroke volume through the prosthesis. Of interest is that AT/ET correctly identified PAV function as normal in patients with PPM, because AT/ET was similar in controls and in PPM. This most likely is because both AT and ET increase with the high flow observed in patients with PPM. A ratio of AT to ET may thus still be in the normal range. A support for this rationale is that AT/ET had the lowest significant relation to stroke volume in the 3 models. Thus, a combination of AT and AT/ET may be the most helpful in the overall evaluation of prosthetic valve function using systolic time intervals. Although EOA determination is crucial in evaluating PPM, it becomes problematic when the valve size and type are not known, which is not an infrequent situation. Systolic time interval parameters, particularly AT and AT/ET would be helpful in evaluating valve function in these circumstances.
Determinants of systolic time intervals
Various factors were evaluated to assess determinants of systolic time intervals in PAVs. The major determinant for AT, ET, and AT/ET was EOA of the valve. Stroke volume through the valve was also a determinant, but to a much lesser degree, as judged by the coefficient value (Tables 3, 4, and 5): for example, for any increase of 10 ml in stroke volume, AT and AT/ET values will have a minor increase of 4.6 ms and 0.01, respectively. An even lesser influence of ejection fraction was noted for AT (Table 3). To demonstrate this relation, we can assess, for example, the difference in AT values among patients with ejection fractions of 30%, 50%, and 70% (assuming all other parameters in the model are kept the same). Compared with a patient with an ejection fraction of 50%, a patient with an ejection fraction of 30% will have a shorter AT by 5.6 ms, whereas for a patient with an ejection fraction of 70%, the AT will be 8 ms longer.
For the ET model, heart rate was a significant determinant, in addition to EOA and stroke volume (Table 4). Using the model, any increase in heart rate by 10 beats/min results in a decrease in ET by 13 ms, whereas an increase in stroke volume by 10 ml results in an increase in ET by 29 ms. The influence of heart rate and stroke volume on ET in PAVs is not surprising because it has been documented in native aortic valve stenosis (16). Similarly, ET has shown a relationship to cardiac index, stroke volume, and ejection fraction in PAV (7). Thus, ET as a sole parameter of function should be used carefully in PAVs.
The number of observations in the stenotic PAVs may be relatively small. However, this condition, particularly in mechanical valves, is uncommon. The present series stems from collaboration among the mentioned institutions to achieve this goal. Furthermore, we required convincing documentation of prosthetic valve obstruction. Because the study is a retrospective study and the study participants were identified by institutional staff, there exists the potential for a sampling bias; the generalization of the current findings may therefore be tempered pending further validation and experience. The systolic time intervals are complementary parameters for assessment of PAV function and should be used in conjunction with other indexes such as pressure gradients, EOA, and DVI. For patients where valve obstruction is in question, confirmation of the abnormality in valve motion is undertaken with transesophageal echocardiography and/or fluoroscopy or computer tomography (12). Although systolic time intervals are related predominantly to EOA, ventricular function (stroke volume/ejection fraction), as well as heart rate, may carry an influence, albeit much smaller.
The present findings over a wide range of conditions show that systolic time intervals, particularly AT and the ratio of AT/ET, are reliable, angle-independent parameters that can enhance the evaluation of prosthetic valve function and help identify PAV stenosis.
The authors thank Hen Hallevi, MD, for his assistance.
Dr. Little has received research support from St. Jude Medical and Siemens Medical Imaging. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose. Jeroen J. Bax, MD, PhD, served as Guest Editor for this paper.
Presented in part at the Annual Scientific Sessions of the American Heart Association, November 2008, New Orleans, Louisiana.
- Abbreviations and Acronyms
- acceleration time
- body surface area
- Doppler velocity index
- effective orifice area
- ejection time
- left ventricular outflow tract
- prosthetic aortic valve
- prosthesis–patient mismatch
- receiver-operator characteristic
- time-velocity integral
- Received June 8, 2011.
- Revision received August 2, 2011.
- Accepted August 18, 2011.
- American College of Cardiology Foundation
- Chafizadeh E.R.,
- Zoghbi W.A.
- Burstow D.J.,
- Nishimura R.A.,
- Bailey K.R.,
- et al.
- Pibarot P.,
- Dumesnil J.G.
- Rothbart R.M.,
- Castriz J.L.,
- Harding L.V.,
- Russo C.D.,
- Teague S.M.
- Zoghbi W.A.,
- Chambers J.B.,
- Dumesnil J.G.,
- et al.
- Hatle L.,
- Angelsen B.A.,
- Tromsdal A.