| JOURNAL HOME | CME HOME | THIS MONTH | PAST ISSUES | ETOC | COLLECTIONS |
| AUTHORS | REVIEWERS | EDITORIAL BOARD | FEEDBACK | RSS | HELP |
| ||||||||||||||
| ||||||||||||||
|
|
|
|
||||||||||||
|
||||||||||||||
Departments of
*Anesthesia and
Cardiology, Beth Israel-Deaconess Medical Center; and
Department of Cardiology, Mass General Hospital, Boston, Massachusetts
Address correspondence and reprint requests to Andrew Maslow, MD, Department of Anesthesia, Rhode Island Hospital, 593 Eddy St., Davol 129, Providence, RI 02903. Address e-mail to amaslow{at}lifespan.org
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionReferences |
|---|
Implications: Quantitative assessment of prosthetic aortic valve area can be performed on-line by using transesophageal echocardiography using the double envelope technique.
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Calculation of prosthetic AVA is accomplished by using the conventional continuity equation, with nonsimultaneous measurements of left ventricular outflow tract (LVOT; V1) and aortic valve (AoV; V2) velocities by using pulse wave (PW) and continuous wave (CW) Doppler, respectively. After rearranging the continuity equation, it is apparent that the AVA is directly related to the instantaneous ratio of the LVOT velocity to the AoV velocity (V1/V2). This has been referred to as the Doppler velocity index (DVI) or velocity ratio (VR) (1,5,8–10).
We have observed that a smaller flow velocity profile can be seen within the time velocity integral obtained with CW Doppler examination of the aortic valve making a double-envelope (DE). We postulated that this inner envelope represents subvalvular blood flow (V1) and can be used to calculate AVA. This has been previously described during transthoracic examination of native aortic valves (23).
The purpose of this study is to assess the feasibility of TEE to assess prosthetic AVA by using the continuity equation with data obtained from the DE (obtaining both V1 and V2 simultaneously). We hypothesize that prosthetic AVA can be accurately measured by using TEE and the DE technique.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Three calculations of AVA were performed in each patient from three separate cardiac cycles at various times during the TEE examination. These were averaged to obtain a mean AVA. To assess variability of individual measurements, five calculations of AVA were performed in a subset of 20 consecutive patients.
All patients undergoing elective aortic valve replacement, with or without additional surgical procedures, were included. Exclusion criteria included patients with contraindications to TEE.
Echocardiographers were blinded to the size of the prosthetic valve placed. It is not possible to be truly blinded to the type of valve (bioprosthetic versus bileaflet mechanical) given their differences on two-dimensional examination. All measurements were performed on-line, immediately after valve replacement (Sonos 2000 or Sonos 2500).
Patients received either Carbomedics (CM) bileaflet mechanical or Carpentier-Edward’s (CE) bioprosthetic valves. All patients had a sinus rhythm or paced atrio-ventricular rhythm. There was no evidence, by two-dimensional TEE of LVOT obstruction. As previously described, LVOT was measured during TEE examination at 135° (± 10°), while subvalvular (V1) and valvular (V2) peak velocities were obtained from deep gastric views (18–21). Transgastric examination began at 0° and continued to approximately 155°. Manipulating both flexion/extension and lateral wheels of the TEE probe, the maximal velocity across the aortic valve (V2) was sought. Adjustment of Doppler gain allowed distinction between the smaller velocity profile (V1) within the larger one (V2). Prosthetic AVA was calculated by using the continuity equation and simultaneously recorded V1 and V2 (DE) as shown in Equation 1:
|
|
|
|
AVA was also calculated by using the reported external and internal diameters of the prosthetic valve. These diameters were used to calculate the subvalvular cross-sectional area in place of a measured LVOT diameter. These data were compared with data provided by the manufacturer.
The V1/V2 ratio, or the DVI, was also calculated on-line and recorded. Previously reported Doppler velocity integrals were not available for CM valves. They were, however, available for St. Jude bileaflet mechanical prostheses (1), which may be comparable to CM valves because both are similarly designed bileaflet mechanical prostheses with two lateral orifices and a smaller central orifice. We were able to find DVI data for only the #21 bioprosthetic valve (3).
V1 was also assessed by using PW Doppler and compared with V1 measured by DE. The PW sample volume was initially placed at the level of the aortic valve and withdrawn into the subvalvular space until aliasing ceased and a clear velocity envelope was obtained (1).
Follow-up data were obtained by phone interview and review of medical records. Recorded cardiovascular events included congestive heart failure, angina, syncope, and death.
All data were averaged and presented as mean ± SD. Comparison analyses were performed by using regression and bias analyses. Comparisons were made among data from our results, data provided by the manufacturer, and data from previously reported data. These data include prosthetic valve areas, using measured LVOT diameters, and DVI where available (1,3,5). Comparisons were made between the V1 obtained from CW and PW Doppler tracings. We also compared differences in calculated AVA using different measures of subvalvular diameters, i.e., measured LVOT diameter, and internal and external diameters of the prosthetic valve. A P value < .05 was taken to be significant.
Beat-to-beat variability was assessed in a subset of patients from five measurements obtained from five separate cardiac cycles at random times during the TEE examination. Mean valve area, mean DVI, SD, and the coefficient of variation (SD/mean) were calculated for each patient. From these results, the mean valve area, DVI, SD, and coefficient of variation were calculated for this subset of patients.
Interobserver variability was assessed in a subset of 10 patients by using bias analyses.
An example of the DE is shown in Figure 1.
|
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
mild aortic insufficiency), and no patient had evidence of fixed or dynamic subvalvular obstruction. Data are presented in Tables 1 and 2, as mean (±SD). Bias analyses are displayed in Figures 2 and 3.
|
|
|
|
The AVA for the CE’s valve (#19 to #27), measured by using the DE technique, by using the measured LVOT agreed with that reported by the manufacturer (range 1.05 to 2.30 cm2, r2 = 0.73; mean bias -0.02 cm2 [SD 0.27]; 95% confidence intervals -0.15 to 0.12 cm2) (Table 2 and Figure 3). Compared with the AVA reported by the manufacturer, the calculated AVA using the annular (mounting) diameter of the bioprosthetic valve had a mean bias of 0.45 cm2 (SD 0.30 cm2; 95% confidence interval 0.30 to 0.60 cm2; r2 = 0.92). Using the internal diameter for the calculation of AVA resulted in a mean bias of 0.29 cm2 (SD 0.26 cm2; 95% confidence interval 0.16 to 0.41 cm2; r2 = 0.93).
Left ventricular outflow velocity (V1) obtained by using the DE-V1 method was nearly identical to the V1 obtained by using PW (DE-V1 1.09 m/s [SD 0.40] versus PW-V1 1.08 m/s [SD 0.36], respectively; r2 = 0.95; mean bias 0.02 m/s [SD 0.04 m/s; 95% confidence intervals -0.01 to 0.02]).
DVI data obtained by using the DE agreed with available data reported in the literature (Tables 1 and 2) (1,3). The mean bias was 0.03 (SD 0.05; 95% confidence intervals 0.01 to 0.05; Figure 4). Prosthetic valve area did not correlate well with DVI (r2 = 0.27).
|
Beat-to-beat variability was measured in 16 patients. Five measurements were performed for each patient. The mean DVI for this subset was 0.45 (SD 0.06). The mean AVA was 1.42 (SD 0.10). The mean coefficient of variation was 6.42% (95% confidence intervals 4.10 to 8.70%).
Interobserver variability was assessed in 10 patients. Agreement between separate echocardiographers was excellent. The mean bias for DVI was 0.02 (SD 0.05). The mean bias for DE AVA was 0.06 cm2 (SD 0.12 cm2).
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
The DE technique allows simultaneous measurement of LVOT and AoV velocities from a single CW velocity profile. Although it is accepted that the peak of the CW velocity trace is generated by blood flow across the narrowest point or stenotic aortic valve, it has not been clearly determined where the lower velocity profile originates. We showed excellent agreement with blood flow velocity obtained from the LVOT by using PW Doppler, which was placed in the immediately subvalvular area. Although we did not map out the changes in velocity within the LVOT, it is evident that the lower velocity profile of the CW profile comes from the immediate subvalvular area at or near the site of the PW sample volume. We suspect that this is the result of convergence of blood flow from the left ventricular cavity to the narrower LVOT, much the same way that blood flow converges at the site of the relatively narrower orifice of the prosthetic aortic valve, resulting in higher flow velocity.
According to the continuity equation, the calculated AVA is directly related to the instantaneous ratio of the LVOT to the AoV velocities, which has been referred to as the DVI (V1/V2) (1,5,8,10). Potential errors are introduced when calculating AVA using nonsimultaneous measurements of LVOT and AoV (C/TTE). Typically, three to five measurements of the LVOT and transvalvular peak velocities (total 6–10 measurements from 6–10 different cardiac cycles) are obtained and averaged (24). The sampling is increased in patients with irregular rhythms. Errors may arise from misplacement of the PW sample volume or when the direction of the ultrasound beam is not aligned with the path of blood flow. In addition, because velocities are obtained from different cardiac cycles, beat-to-beat variability in blood flow is likely and will increase errors. When considering the effect these errors produce, the calculation of AVA values may vary by 10%–30% (1,10,24,25). Because AVA is directly related to the instantaneous ratio of V1 to V2, simultaneous measurement of these two velocities should improve accuracy. A conceivably accurate AVA can be calculated from a single CW velocity profile by using the DE technique. We demonstrated little beat-to-beat variability of the VR (V1/V2) and calculated AVA from one cardiac cycle to another.
The literature displays significant variability in the quantitative Doppler evaluation of both mechanical and bioprosthetic aortic valves (1,3–7,14,16,17,26). Several factors may explain this. These include variability in blood flow, nonsimultaneous measurement of V1 and V2, and calculation of the subaortic cross-sectional area (1–3,6,7,26–28). Errors in the measurement of V1, V2, and LVOT diameter can each result in approximately 10%–30% error in the calculation of the AVA (1,10,25). Substituting the LVOT diameter with the prosthetic valve annular diameter is reported to improve correlation statistics with mechanical prostheses, although the opposite result was reported when evaluating bioprosthetic valves (1,2). In the present study, using the LVOT diameter resulted in underestimation of AVA for CM valves, whereas using the reported annular diameter resulted in a similar degree of overestimation. Use of either the annular or internal orifice diameter decreased agreement for the CE valves. Because of this variability, Chambers et al. (3) suggested that the continuity equation "should not be used alone as a measure of the valve opening behavior," but instead it should be viewed as a part of a "balance of information derived from imaging, color flow mapping, and continuous wave Doppler." Calculation of the prosthetic valve area may be considered a "semi-quantitative estimate of function."
Other sources of error in the calculation of prosthetic valve area may include a phenomenon called "pressure recovery" and, in the case of the bileaflet mechanical valves, the recording of a high central velocity (1,3,27,29). The former is a result of changes in blood flow velocity beyond the aortic valve caused by the shape and size of the immediate ascending aorta. The effects are thought to be more significant with moderately stenotic valves and relatively small aortas (27). This has only been described in native aortic valves (27,29). These two errors would occur with both simultaneously and nonsimultaneously measured V1 and V2. One source of error particular to the DE technique would result in the underestimation of V1. Although we directed the ultrasound beam to record the maximum V2, it is possible that the ultrasound beam was not in parallel with LVOT flow. All of these errors would cause underestimation of DVI and subsequently lower prosthetic valve areas. Our DVI results showed excellent agreement with reported data. This would suggest that LVOT or subvalvular diameter measurement contributed more significantly to discrepancies and variabilities seen with data in this study.
Regression analysis between prosthetic AVA and DVI did not show a strong correlation. Although a correlation between the two may have supported that valve assessment could be simplified by measuring the DVI, the lack of correlation does not necessarily reduce its value. Several manuscripts have suggested that the DVI (or VR) is a useful "performance index of a prosthetic valve," and is consistent over time (1,3,5,8–10). The measurement of DVI has an additional advantage in that it does not require calculation of the subvalvular cross-sectional area.
Unfortunately, the value of DVI that constitutes normal or abnormal flow across the prosthetic AoV has not been defined. Furthermore, this value may differ between mechanical and bioprosthetic valves (1,10). Rothbart et al. (10) compared TEE data from 38 asymptomatic patients, within 24 months of aortic valve replacement using bioprosthetic valves, to 12 symptomatic patients who, by cardiac catheterization, had bioprosthetic valve stenosis. In this evaluation, a DVI > 0.35 was found in patients with normal valve function, whereas a DVI <0.35 was found in stenotic valves. Of the 12 patients with prosthetic valve stenosis, only 4 had evidence of obstruction by two-dimensional TEE analysis (leaflet and annular thickening/calcification, decreased valve leaflet mobility) (10). In an evaluation of patients with a bileaflet mechanical St. Jude Medical valve, three patients with mechanical valve obstruction were reported to have DVIs of 0.22, 0.16, and 0.18 (1). These authors (1) suggested that a DVI <0.23 "should cause suspicion of an obstruction." This value was suggested because it was two SDs below the mean DVI found in that study (1). This ratio has been associated with a native AVA less than 1.0 cm2 (30,31). We did not show significant differences in DVI between the valve types with the exceptions of the larger valve sizes, of which there were only a small number of patients, making it difficult to draw statistically significant conclusions. The mean DVI for mechanical valves was 0.44 with a SD of 0.06 (2 SD 0.12). The mean DVI for bioprostheses was 0.47 with a SD of 0.07 (2 SD 0.14). All patients had a DVI greater than 0.35 with the exception of two patients who had received 21-mm CM bileaflet valve prostheses (DVIs; 0.32 and 0.31: AVA 0.96 and 0.92, respectively). One patient (DVI 0.31; AVA 0.92 cm2) had repeated episodes of congestive heart failure on follow-up. An increased transvalvular pressure gradient and moderate prosthetic stenosis had been demonstrated on subsequent TTE examination. At the time of surgery, qualitative assessment, using two-dimensional and color Doppler echocardiography, showed all valves to be functioning normally.
A conservative approach would be to consider 0.35 as a cutoff for normal flow for both types of prosthetic aortic valves, which represents approximately two SDs below the means of both types of valves in the present study. In these patients, the AVA should be calculated and evidence of valve dysfunction should be sought by using two-dimensional TEE. These patients may benefit from regular postoperative examinations to assess valve function. Measurement of bioprosthetic valve area by using the reported prosthetic valve annular or internal orifice diameter to calculate subvalvular cross-sectional area resulted in decreased agreement. Therefore, we recommend using the measured LVOT diameter, as suggested by Pibarot et al. (2). The prosthetic valve annular diameter has been reported to improve the accuracy of AVA measurements of bileaflet mechanical valves; however, we did not demonstrate any added benefit (1).
Quantitative assessment of prosthetic aortic valves may be useful in predicting adverse events (1,6,7,10,14–16,26,32). Combined with clinical data, TEE data may be prognostically useful (6,7,14–16,26,32). Assessment of prosthetic AVA index (effective orifice area index), or the ratio of the prosthetic valve area to the body surface area, may predict subsequent adverse cardiac events (6,7,14–16,26). Studies have suggested that the effective orifice area index of > 0.7 is associated with lower transvalvular gradients and fewer cardiac events (6,7,14–16,26). Selection of prosthetic valve size may, therefore, be based on the size of the native aortic annulus, reported prosthetic valve areas, and the body surface area (6,7,14–16,26).
One limitation of the study is the lack of a gold standard of in vivo prosthetic valve area for data comparison. For example, published prosthetic valve areas for 21-mm CM valves have ranged from 1.23 to 1.54 cm2 (3,6,7,26). Furthermore, within the individual studies, there is significant variability of the calculated valve areas (1,3). This is also true for DVIs, which have been reported in relatively few studies (1,10).
We were unable to find DVI data for CM valves and were able to find data only for the #21 CE valve. However, our data compared favorably with the available data. Although it can be argued that an accurate comparison of our data is not possible, the results of this study can be used for future comparison.
The study sample size is relatively small. Only one to two patients represent several valve sizes. This, statistically, makes the results more speculative. Although we demonstrated good agreement with available data for prosthetic valve areas, the reported confidence intervals were relatively large.
Postoperative TTE assessment of prosthetic aortic valves would have added to the overall assessment; however, TTE is not performed routinely in the absence of symptoms. In addition, available acoustic windows often limit TTE performed in the early postoperative period (within one week).
This study demonstrates that quantitative Doppler assessment during intraoperative TEE examination of the prosthetic aortic valve can be performed and agrees with available valve area data. Because measurement of AVA using the continuity equation is dependent on the instantaneous ratio of the LVOT velocity to the AoV velocity, the use of the DE technique may simplify assessment and minimize error introduced by beat-to-beat variability. Further study is needed to assess the role of DVI; however, this index of valve function may further simplify assessment and eliminate error caused by calculation of subvalvular cross-sectional area. As noted by several authors, the quantitative Doppler assessment should be used as a part of the assessment of prosthetic valve function along with two-dimensional and color Doppler TEE. Prediction of future cardiac events based on intraoperative assessment remains undetermined and should be evaluated by a larger sample size with longer follow-up.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
