Gradients in Aortic Prosthetic Valves In Vitro

Transcription

1 1467 Discrepancies Between Doppler and Catheter Gradients in Aortic Prosthetic Valves In Vitro A Manifestation of Localied Gradients and Pressure Recovery Helut Baugartner, MD, Steven Khan, MD, Michele DeRobertis, RN, Lawrence Cer, MD, and Gerald Maurer, MD Downloaded fro by on Septeber 27, 218 To evaluate possible causes of discrepancy between Doppler and catheter gradients across prosthetic valves, five sies (19-27 ) of St. Jude and Hancock valves were studied in an aortic pulsatile flow odel. Catheter gradients at ultiple sites distal to the valve were copared with siultaneously obtained Doppler gradients. In the St. Jude valve, significant differences between Doppler and catheter gradients easured 3 downstrea fro the valve were found: Doppler gradients exceeded peak catheter gradients of 1 Hg or ore by 81±35% (15±3.6 Hg), and ean catheter gradients by 71±11% (1.3±2.5 Hg). When the catheter was pulled back through the tunnel-like central orifice of the valve, high localied gradients at the valve plane and significant early pressure recovery were found. When the catheter was pulled back through the large side orifices, gradients at the sae level were only 46±6% of the central orifice gradients (ean difference, 7.6±4.5 Hg). Doppler peak and ean gradients showed excellent agreeent with the highest central orifice catheter gradients (ean difference, 1.±3.1 and.9±1.5 Hg, respectively). A significantly better agreeent between Doppler and catheter gradients at 3 was found for the Hancock valve, although Doppler peak and ean gradients were still slightly greater than catheter gradients. Doppler gradients exceeded catheter gradients by 18±1% (3.4± 1.9 Hg) and 13±11% (2.1±.9 Hg), respectively. When the catheter was pulled back through the valve, the highest gradients were found approxiately 2 distal to the valve ring. Doppler peak and ean gradients showed excellent agreeent with these catheter gradients (ean difference, -.6±2.1 and -1.2±.9 Hg, respectively). Therefore, Doppler gradients accurately reflect the highest obtainable catheter gradients, which occur between the two leaflets (but not in the side orifices) of St. Jude valves and 2 distal to the prosthesis in the Hancock valves. However, Doppler gradients ay be significantly higher than catheter gradients easured further downstrea due to localied gradients and pressure recovery. The difference between Doppler and catheter gradients is, therefore, not due to overestiation by Doppler, but to the fact that the two techniques easure gradients in different locations. Doppler easureents reflect the axiu gradient along the interrogation line, whereas catheteriation easures the recovered pressure distal to the valve. The agnitude of this discrepancy is not clinically significant in Hancock valves but ay be substantial in St. Jude valves, particularly in saller valve sies at high flow rates. These findings suggest the iportance of considering valve type, valve sie, flow rate, and catheter position when using continuous wave Doppler to predict clinical catheteriation gradients in prosthetic valves. (Circulation 199;82: ) Fro the Division of Cardiology and Cardiovascular Surgery he siplified Bernoulli equation has been Cedars-Sinai Medical Center and UCLA School of Medicine, Los widely used to deterine pressure gradients Angeles, California. a Supported in part by the Western Cardiological Foundation. contiaoss vavularsteosonnvasivelycwith H.B. was supported by a grant fro the Fonds ur Forderung der continuous wave Doppler ultrasound. The accuracy wissenschaftlichen Forschung, Vienna, Austria. of this technique has been validated in nuerous Address for correspondence: Steven Khan, MD, Division of clinical studies of stenotic heart valvesl-3 and in Cardiology, Cedars-Sinai Medical Center, 87 Beverly Boulevard, experiental studies of siple and coplex Los Angeles, CA 948. Received January 17, 199; revision accepted May 29, 199. orifices.4-6 For prosthetic valves, however, conflicting

2 1468 Circulation Vol 82 No 4, October 199 FIGUR 1. Diagra of the flow odel. Diensions are given in illieters. PF, pressure tap. 1 o 3 Downloaded fro by on Septeber 27, 218 results have been published and the accuracy of the technique reains controversial. Althoughgood agreeent between Doppler and catheter derived gradients has been reported by soe investigators,7-12 significant overestiation of gradients by Doppler in echanical prostheses has been found by others A potential cause of discrepancy between Doppler and catheter gradients is pressure recovery distal to the valve Furtherore, echanical prostheses, such as the St. Jude valve, have ultiple orifices and tunnel-like configurations, resulting in coplex flow and pressure characteristics,18-2 which could lead to the occurrence of localied pressure gradients. This study addresses the hypothesis that discrepancies between Doppler and catheter gradients across prosthetic heart valves are not due to intrinsic errors of either easureent technique, but to spatial variability of pressure fields in prosthetic valves. To evaluate this, we designed a well-controlled in vitro odel and tried to quantify the spatial variation of pressure and to isolate areas where Doppler and catheter easureents agree and disagree. To evaluate whether differences in the spatial variation of pressure between different valve types affected the Doppler-catheter gradient relations, we exained both a echanical and a bioprosthetic valve. Methods Model The flow odel consists of a "ventricular" section, "aortic" section, and a copliance chaber constructed of Lucite and PVC tubing (Figure 1). The aortic section was designed to iniie the distance fro the valve to the Doppler transducer and to allow optial alignent of the Doppler bea with flow direction. Ultrasonic gel was used for acoustic coupling. The prosthetic valves were ounted between atched pairs of Lucite plates that had central orifices achined in 1 increents to fit the sewing rings of each valve exactly. The odel was driven by a reciprocating pup (Harvard Apparatus Pulsatile Blood Pup, odel 1423) with a stroke volue adjustable fro 15 to 1 cc, a rate adjustable fro 1 to 9 beats/in, and an ejection tie variable fro 3% to 7% of the cycle. Catheters were inserted through a side branch of the tubing for easureent of pullback pressures. A physiological upstrea pressure of approxiately 12/5 Hg was aintained by varying the outflow resistance and the outflow copliance. The syste was pried with a 7% water-3% glycerol solution (viscosity, 3.5 cp). For Doppler easureents 1 g/l cornstarch was added. In preliinary studies, gradients obtained with and without cornstarch were copared and were siilar (y=.2+ 1.Ox, S= 1.9 Hg, r=.98). Flow was easured with an ultrasonic floweter (Transonic Syste Inc., odel T21). The probe was attached to the noncopliant connection tubing between pup and test odel. The accuracy of the flow easureents was tested by a series of tied collections. The floweter easureents accurately predicted the flow rates calculated fro tied collections (y= x, S=.87 cc/sec, r=.99). Aortic and ventricular pressures were easured with fluid-filled catheters of atched length and electronic pressure transducers (Abbott Critical Care Syste) connected to a 12-channel physiologic recorder syste (lectronics for Medicine, VR 12). The frequency response of the pressure easuring syste was greater than 1 haronics of the fundaental frequency21 as easured with the shock excitation ethod.22 Flow and pressure tracings (Figure 2) were digitied for further calculation using an iage analysis coputer (Microsonics CAD 886). The ean systolic flow rate was obtained by integrating the flow curve over the systolic tie period. The peak systolic gradient was defined as the axial instantaneous gradient between "ventricle" and "aorta" of the flow odel. The ean systolic pressure gradient was obtained by integrating the difference between the siultaneously recorded "aortic" and "ventricular" pressure waves over the systolic tie period, defined as the tie

3 Baugartner et al Doppler and Catheter Gradients in Aortic Prostheses 1469 Downloaded fro by on Septeber 27, 218 u L O \ _j~ ~ < LV ~io., -2 o \ 1 o c 5 - CO FIGUR 2. Siultaneous Doppler, pressure, and flow recordings. Left panel: Siultaneously recorded flow (top) and pressure tracings of ventrticle and aorta (botto) in a Hancock valve (19 ). Right panel: Siultaneously recorded spectral display of continuous wave Doppler signals. LV, "ventricular" pressure; Ao, "aortic" pressure. period during which forward aortic flow occurs.16 ach set of easureents was obtained by averaging two readings of each of three consecutive beats. Doppler chocardiography Continuous wave Doppler easureents were perfored with an Ultraark 4 CAD syste (Advanced Technology Laboratories) using a Duplex probe (3.5 MH-iaging, 2. MH-CW Doppler). The ultrasound bea was carefully aligned to record the highest Doppler velocities, and the transducer was fixed in place with an adjustable clap syste. Peak and ean Doppler gradients were calculated using the on-board quantitation package. Gradients were calculated with the siplified Bernoulli equation: Ap=4 v2. The velocities proxial to the valve were calculated fro the easured peak and ean flow rates and the sie of the inflow. Proxial velocities ranged between.2 and.5 /sec and were, therefore, neglected. Doppler easureents were A) c %- -JTLW---.: WAORTAW *VNTRICL' x flfi AST, CATH ~ PM == l l ST. CATH ~3 t>1>~--- B) LAFLrS ade on three beats and averaged. Doppler velocities (Figure 2) were recorded on paper (Videographic Printer YP 181) and videotape. Test Protocol Five sies (19-27 ) of St. Jude bileaflet echanical prostheses and of Hancock bioprosthetic valves were studied. The sies 19 to 25 of the porcine valves had the odified orifice (odel 25), whereas the 27 had the standard orifice (odel 242). Three experients were perfored. 1) Pressure gradients were easured between a proxial end-hole catheter (2 c fro the valve) and a side-hole horiontal catheter that was pulled through the orifice (Figure 3). Glycerol solution without cornstarch was used in this experient to allow visual guidance of the catheter. The side hole was initially oved to the level of the end-hole catheter to ensure that the pressure waves were identical for both catheters. Then, the catheter was pulled back through the prosthetic valve; in the St. Jude valve, this was done through both the center (between the two leaflets) as well as through one of the large side orifices (Figure 3). Distance was defined as the plane of the very first structure of the open prosthesis encountered by the flow strea (i.e., sewing ring in Hancock valves and proxial leaflet tips in St. Jude valves). The linear distance fro the level to the pressure tap was 3 (Figure 3). The curvilinear distance that the catheter traversed fro this point to the pressure tap was 35 ; pressure gradients were easured at 1, 3, 5, 7, 1, 15, 2, 25, 3, and 35 in the St. Jude valves and at 1, 5, 1, 15, 2, 25, 3, and 35 in the Hancock valves. The easureent locations were deterined by a pilot study where the spatial pressure distribution was exained for each valve type. The side hole of the catheter was oriented to avoid the underestiation of the lateral pressure gradient that would occur if the pressure port were to face into the flow strea. This experient was perfored for all valve sies at two different ean systolic flow rates ( and cc/sec) with systolic tie periods between 35 and 375 sec. 2) To obtain inforation about the pressure distribution further downstrea, a 2 c long Lucite CATH FIGUR 3. Diagra of experiental setup for pops 1 pressure easureents across the prosthetic valve. Panel A: Pressure gradient is easured between the end-hole catheter in the ventricular chaber CATH and the side hole of the pull-through catheter. Panel B: Cross-section ofa St. Jude valve with the two catheter positions in either the central (cath pos 1) or the side orifice (cath pos 2).

4 147 Circulation Vol 82, No 4, October 199 TABL 1. St. Jude Valves: Peak and Mean Gradients Measured by Doppler and Catheter (at 3 fro valve level) Peak gradient Mean gradient Sie () Doppler ( Hg) Catheter ( Hg) Doppler ( Hg) Catheter ( Hg) ± ± ± ±4. 9.7± ± ± ± ± ± ± ±1.1.7±.7 Values are ean±sd. TABL 2. Hancock Valves: Peak and Mean Gradients Measured by Doppler and Catheter (at 3 fro valve level) Peak gradient Mean gradient Sie Doppler Catheter Doppler Catheter () ( Hg) ( Hg) ( Hg) ( Hg) ± ± ± ± ± ± ± ± ± ± ± ± ±3.1 Values are ean-+sd. Downloaded fro by on Septeber 27, 218 tube (inner diaeter, 32 ) was positioned between the prosthesis and the original aortic chaber. Pressures were recorded in the centerline of the tubing at 2, 3, 4, 5, 6, 75, and 1 fro the level (pull back as described before). The 19 and 23 prostheses of both the St. Jude and the Hancock valve were selected for these easureents. Flow rates of 21±3.2 and ±4.7 cc/sec were used. 3) After deterining the location of the highest pressure gradient ("highest obtainable catheter gradient"), the distal catheter was positioned at this site. For each valve sie and type the flow rate was varied in five steps fro 149.±9.2 to 258.2±2.9 cc/sec, and siultaneous Doppler and catheter gradients were recorded. Cornstarch was added to the glycerol solution in this experient. The distal catheter was then pulled back to the tap 3 downstrea fro the valve and the easureents were repeated. The 3 distance was chosen to eulate typical gradient easureents in clinical catheteriation laboratories. During these experients, the "heart" rate was aintained at 5 beats/in, and the systolic tie periods were aintained between 325 and 37 sec. Statistical Analysis The relation of catheter and Doppler gradients was assessed by linear regression analysis. Pearson correlation coefficients were calculated. A two tailed t test was perfored to test hypotheses about two regression lines. To test the agreeent between the two techniques, the ean differences between Doppler and catheter gradients were calculated.23,24 Mean values and standard deviation were calculated for the percent decrease of pressure gradient with increasing distance fro the valve. The 27- St. Jude valve was excluded fro these calculations because of the very sall gradients. Results Peak and ean gradients easured by Doppler and catheter (at 3 downstrea fro the valve) for St. Jude and Hancock valves are shown in Tables 1 and 2. Pressure Recovery In St. Jude valves, the highest pressure gradient was found in the center of the prosthesis between the two leaflets approxiately 3-5 distal to the defined level (Figure 4A). Pressure gradients steeply decreased within a few illieters (still between the two leaflets) and reached a plateau between 1 and 25 beyond the valve with only slight further decrease. The gradients also decreased by % when the pressure port of the side-hole catheter was oved fro the center of the flow channel to the wall at 3. The ean decrease between 5 (highest obtainable gradient) and 3 at the wall was %. When the longer outflow channel was used, a decrease in gradient of % was found between 2 and 1 downstrea fro the valve. Most of this decrease occurred between 2 and 5, whereas the decrease between 5 and 1 was insignificant. The high gradients between the leaflets were not found when either of the large side orifices was crossed by the catheter (Figure 3B). At the level where the highest gradients were found across the central orifice, the gradients across the side orifice were only 46±6% of the highest central orifice gradients (ean difference, 7.6±4.5 Hg, axiu 14.6 Hg; flow rate 245 cc/sec). Side orifice gradients increased continuously to a axiu at 15-2 beyond the valve (Figure 4B). Mean decrease to the 3 wall easureent was 22.9±11.2%. A different pressure distribution was found distal to the Hancock valves (Figure 4C). The highest gradients were found 2 distal fro the valve ring. Gradients decrea ed % to the 3 wall easureent. When the longer outflow channel was used, gradients decreased 16.±4.4% between 2 and 1. Correlation Between Doppler and Catheter Gradients St. Jude valve (Figures 5 and 6): The peak and ean Doppler gradients substantially exceeded the peak and ean catheter gradients easured 3 beyond the valve at the wall, although there was a strong linear relation between the two easureents (r= ). Peak Doppler gradients exceeded peak catheter gradients of 1 Hg and ore by 81+35% ( Hg), and ean Doppler gradients exceeded ean catheter gradients greater than or equal to 1 Hg by 71+11% ( Hg) on average. The greatest discrepancy was seen for the sallest valves.

5 Baugartner et al Doppler and Catheter Gradients in Aortic Prostheses 1471 A 3 ST. JUD ) I- WJ a er C, w W- Wu!- 'C C, Lu WC FIGUR 4. Plots of ean catheter gradient and distance fro valve. Panel A: Gradients obtained when the catheter was pulled back through the central onfice of the St. Jude valve (sie, 19-27; flow rate, cclsec). Panel B: Gradients obtained for two sies of St. Jude valves when the catheter was pulled back through either the central or the side onfice. Panel C: Gradients as a function of distance fro the valve ring for Hancock valves (sie, ; flow rate, 245.5±12 cc/sec). Downloaded fro by on Septeber 27, DISTANC FROM VALV () 8 ~ 6 a cc c4 ~ cc O 2 a a. 2 y= x r=.98 S=3.5 y= x r=o.99 S=2.7 CATHTR GRADINT O Highest At 3O 4 CD S 4 3 co 2 a. a. 1 an PAK CATHTR GRADINT (H9) FIGUR 5. Plot ofcorrelation of the highest obtainable peak catheter gradients (at valve level) and the peak catheter gradients at 3 fro the valve with the peak Doppler gradient for the St. Jude valve (19-27 ). Dashed line represents the line of identity. Difference between the slopes of the two regression lines is significant (p<.1) MAN CATHTR GRADINT (Hg) FIGUR 6. Plot ofcorrelation of the highest obtainable ean catheter gradients (at valve level) and the ean catheter gradients at 3 with the ean Doppler gradient for the St. Jude valve (19-27 ). Dashed line represents the line of identity. Difference between the slopes of the two regression lines is significant ( p<.1).

6 1472 Circulation Vol 82, No 4, October 199 Downloaded fro by on Septeber 27, 218 I I- Wa 6 4 y= x r=.99 S= 1.9 y= x r=.99 S =.9 a er. W CATHTR GRADINT 2 O Highest * At 3 w a PAK CATHTR GRADINT (Hg) FIGUR 7. Plot of correlation of the highest obtainable peak catheter gradients (2 distal to the valve ring) and the peak catheter gradient at 3 with the peak Doppler gradient for the Hancock valve (19-27 ). Dashed line represents the line of identity. Difference between the slopes of the two regression lines is statistically significant (p<.5). Doppler gradients, however, showed excellent agreeent with the highest obtainable catheter gradients that were found between the two leaflets of the valve. At the site of the highest obtainable gradient, the peak Doppler and catheter gradients differed by only 1.±3.1 Hg. The ean gradient at this site differed by only.9±1.5 Hg. Hancock valve (Figures 7 and 8): A significantly better agreeent was found between Doppler gradients and catheter gradients at 3 for the Hancock valve, although Doppler gradients were still slightly greater than catheter easureents. Peak catheter gradients of 1 Hg and ore were exceeded by 18±1% (3.4±1.9 Hg) and ean gradients were exceeded by 13±11% (2.1±.9 Hg) on average. At the site of the highest obtainable gradient, excellent agreeent was found between Doppler and catheter gradients. Mean differences between Doppler and catheter gradients were -.6±2.1 Hg for the peak gradient and - 1.2±.9 Hg for the ean gradient. Discussion The accuracy of gradient estiation by continuous wave Doppler using the siplified Bernoulli equation has been validated in clinical studies of stenotic heart valves' 3 and in vitro studies of siple and coplex orifices.4-6 Siilar results have been reported for bioprosthetic valves.8-11 For echanical prostheses, however, soe investigators have reported good agreeent between Doppler and catheter gradients,9-11 whereas others have reported that Doppler substantially overestiates catheter gradients The rea- ) I- W lii C) cc cc -J W y=o+ 1.Ix r=.98 S= 1.4 y=-o.6+.9ft r=.99 S=.9 CATHTR GRADINT O Highest * At MAN CATHTR GRADINT (Hg) FIGUR 8. Plot ofcorrelation of the highest obtainable ean catheter gradient (2 distal to the valve ring) and the ean catheter gradient at 3 with the ean Doppler gradient for the Hancock valve (19-27 ). Dashed line represents the line of identity. Difference between the slopes of the two regression lines is significant (p<.2). son for these conflicting reports is not known. Several investigators have deonstrated the adequacy of the assuptions that the siplified Bernoulli equation is based on',6,'7,25; however, several potential sources of error reain. The gradient ay be underestiated if the Doppler bea and poststenotic jet are iproperly aligned. Gradients ay be overestiated if the velocity proxial to the lesion is high and is neglected.1,2 Both errors can be either eliinated or their agnitude estiated in controlled in vitro easureents. Pressure Recovery One possible cause of discrepancy between Doppler and catheter easureents is pressure recovery. The pressure in a stenotic lesion will be lowest where the velocity is highest, that is, at the vena contracta, and will increase again distal to the stenosis. The actual aount of pressure recovery will depend on how uch energy has been lost at the stenosis due to viscous and turbulent energy losses. Therefore, Doppler gradients that correspond to the gradients at the vena contracta will be higher than catheter gradients that are usually easured further downstrea at varying distances fro the valve. Significant pressure recovery has been shown in vitro in odels of aortic stenosis26,27 and in several valve prostheses,16,28,29 although the available data have liitations. For exaple, Tindale et a128 and Schra et a129 used only steady flow odels. Although Yoganathan et al'6 reported pulsatile flow experients, the closest pressure easureents were taken 3-4 distal to the valve. Furtherore, neither of these studies used Doppler ultrasonography. The present study used pulsatile flow, and the

7 Baugartner et al Doppler and Catheter Gradients in Aortic Prostheses 1473 Downloaded fro by on Septeber 27, 218 valve was crossed with a catheter to deterine where the highest gradients occur. When Hancock valves or the large side orifice of St. Jude valves were crossed with the catheter, the highest gradients were found approxiately 2 distal fro the first valve structure. The sae distance has been reported by Tindale et a128 for bioprostheses in a steady flow odel and by Clark26'27 in a odel iitating aortic stenosis. Pressure recovery has been described up to fro the prosthetic valve.16,28 However, in vivo, pressure waves change along the aorta. Tie delay and reflections fro the periphery change the wave shape and the axiu systolic pressure ay increase.3 We found siilar changes of the aortic pressure waves in our odel, and we excluded fro further analysis recordings that were obtained ore than 1 downstrea fro the valve. The ean decrease of the pressure gradient between 2 and 1 was approxiately 15% and ost of the pressure recovery occurred within the first 5. A siilar change in pressure gradient was found between 2 (at the center) and 3 (at the wall), indicating a significantly higher flow velocity in the center of the aorta. Differences Between St. Jude and Hancock Valves In the St. Jude valve, significantly higher gradients were found within the valve's central orifice (between the two leaflets). These gradients decreased rapidly when the catheter was advanced a few illieters downstrea (still between the two leaflets). This observation suggests that the highest localied gradient in this echanical prosthesis occurs at the valve plane. The gradients easured at the side orifices were lower, deonstrating that localied gradients also occur across the cross-section of the flow channel. These local gradients cannot be explained by the ultiple orifices in this valve. Teirstein et a16 deonstrated that Doppler velocities for pairs of large and sall orifices were identical and that the calculated gradients correlated well with those easured by anoeter. If pressure gradients across two orifices are the sae, the flow velocity generated by those pressure gradients ust also be the sae, according to the Bernoulli equation. However, the unique geoetry of echanical valves is ore coplex than the siple plates studied by Teirstein. Flow velocities across disk prostheses, in particular, are not unifor18-2 and deonstrate ore spatial variation than in Hancock valves.31 Current ethods are liited in their ability to visualie flow within the echanical valve itself. Therefore, no data are available on flow velocities or gradients between the two leaflets of a St. Jude valve, although velocities further downstrea have been reported.18 Tindale et a128 reported higher and earlier occurring gradients in the inor orifice than in the ajor orifice of a Bjork-Shiley valve. An explanation for these conflicting results ight be that echanical prostheses are not siple flat plates with ultiple orifices, but coplex, three-diensional structures with tunnel-like configurations. This is particularly obvious in the St. Jude valve, which has a central "tunnel" between the two leaflets. The role of pressure recovery in causing apparent overestiation of pressure gradients by Doppler in tunnel-like orifices has recently been reported.17'32 Fro the results of the present study, one could hypothesie that high velocities and low pressures occur in the tunnel between the two leaflets of a St. Jude valve. The reversed conical shape of the tunnel creates a strealined stenosis beyond which flow expands gradually back into the larger luen. Therefore, energy loss across the central orifice is reduced and pressure recovery increased, causing a steep decay of the pressure gradient (Figure 4A). In the present study, the observed decrease of pressure gradient between 5 and 3 (wall) distance was considerable (ean, %). Correlation Between Doppler and Catheter Gradients In the present study, Doppler gradients significantly exceeded catheter gradients easured 3 distal fro the valve, although a strong linear relation was found. The differences were less in bioprosthetic valves but substantial in echanical valves. Doppler gradients, however, showed excellent agreeent with the highest obtainable catheter gradients deterined by catheter pull back through the valve. Therefore, Doppler ultrasound can accurately easure the highest spatial gradients across these valves that occur at approxiately 2 beyond the valve ring in the Hancock valve and within the prosthesis itself in the St. Jude valve. When catheter gradients are easured further downstrea, as is usually done in clinical catheteriation studies, discrepancies between the Doppler and catheter gradients ay occur. These discrepancies are not due to an intrinsic liitation of either technique, but to the fact that pressure fields have spatial variability and gradients are easured in different locations by the two techniques. Continuous wave Doppler gives us the axiu velocity along the line of interrogation and, therefore, reflects the highest gradient along the path of the bea. The catheter technique, on the other hand, easures pressure only at the site of the sapling port. Gradient easureents are, therefore, highly dependent on the sapling sites and, when perfored at a distance fro the valve, reflect recovered pressures. One ight argue that the Doppler gradient is of greater clinical ipact because it reflects the axiu gradient experienced by the working left ventricle. The work done by the ventricle in systole, however, is given by the integral of the instantaneous pressure-flow product. Therefore, the absolute pressure generated in the ventricle and the volue of blood ejected fro the ventricle deterine ventricular work. Attepts to estiate this absolute ventricular pressure by adding Doppler gradients to the systolic aortic pressure ay lead to erroneous results due to pressure recovery and inhoogeneous velocity and pressure distribution in echanical valves. The high-

8 Downloaded fro by on Septeber 27, Circulation Vol 82, No 4, October 199 est obtainable gradients in St. Jude valves are localied phenoena occurring only in the center of the valve but not across the larger side orifices. Therefore, the gradients reflected by Doppler ay differ substantially fro the average gradient present across the entire valve orifice area at any point in tie. The iportant finding of this study is that spatial variation of pressure distal to prosthetic valves ay cause Doppler and catheter to easure gradients in different locations. This spatial variation has to be considered when interpreting Doppler and catheter gradients and, particularly, when coparing these gradients. We have deonstrated how the two techniques differ, not the inherent superiority of either technique. Coparison to Previous Studies Although the difference between Doppler gradients and catheter gradients easured 3 downstrea fro the valve was significant for the Hancock valve, its agnitude is probably of no clinical relevance. It is not surprising, therefore, that previous in vitro8 and clinical studies9-11 reported good agreeent between Doppler and catheter gradients in bioprosthetic valves-even though the latter were easured at various distances fro the valve. The observation of substantial discrepancy between Doppler and catheter gradients in St. Jude valves agrees with siilar results that have been reported previously for echanical valves in published13-15 and unpublished sources.33 In absolute ters, the differences are less significant for the 25- and 27- prosthesis where gradients below 1 Hg can be expected. However, in the saller 19- and 21- sies, the difference between Doppler and catheter gradients ay be ore than 2 Hg, particularly at high flow rates. Differences of this agnitude can lead to erroneous conclusions about valve function in clinical situations. Previous studies where good agreeent between Doppler and catheter gradients was shown9-11 ay not have included this subgroup of sall valve sies with easureents at high flow rates. Thus, Wilkins et a19 only studied itral prostheses, where valve sies are larger and flow velocities lower. In the study by Sagar et al1 only three patients with echanical valves are reported, and in each case, Doppler overestiated the nonsiultaneous catheter gradient, in one instance by 3 Hg. Burstow et al"l reported good agreeent for siultaneous Doppler and catheter gradients in a variety of echanical aortic prostheses including two 21- St. Jude valves. Cardiac index was only given in one of these two patients and was consistent with a low output state. Liitations In applying these results to clinical situations, it ust be reebered that in vitro odels like the one used in this study cannot precisely duplicate the coplex flow dynaics in a patient with an iplanted aortic prosthesis. For exaple, pressure recovery is partly deterined by the sie of the receiving chaber.28 The diaeter of the aortic chaber used in this study is larger (32 versus 25.4 ) than that reported by Yoganathan et al.16 The larger sie of the aorta ay have caused a saller aount of pressure recovery Thus, using a saller aorta would have resulted in an even greater discrepancy between Doppler and catheter gradients easured downstrea fro the valve. In addition, the axiu flow rates used in this study were liited by the pup characteristics. It is possible that under clinical conditions of high flow rates discrepancies between Doppler and catheter easureents ay be even greater. There are also technical liitations to our study. Placing the catheter across the prosthesis causes an unphysiological situation that ay be ore iportant in a echanical valve because coplete closure during diastole is prevented. It is unlikely, however, that having the valve partially open at the beginning of systole would significantly affect the correlation of Doppler and catheter gradients. In addition, the accuracy of the fluid filled catheter syste used in this study is liited, particularly when easuring sall gradients. Transducer tip catheters would have been ore accurate but do not allow the special set up (Figure 3) that was necessary for precise placeent of the distal pressure port. Clinical Iplications Doppler ultrasound accurately reflects the highest obtainable gradients across prosthetic valves, which occur at approxiately 2 beyond the valve ring in the Hancock valve, and within the prosthesis itself in the St. Jude valve. However, discrepancy between Doppler and catheter gradients ay occur when catheter easureents are obtained further downstrea as is usually the case in clinical catheteriation studies. This discrepancy is probably of no clinical relevance in Hancock porcine valves but ay be substantial in the echanical St. Jude prostheses, particularly with saller valve sies studied at high flow rates. The discrepancies between Doppler and catheter gradients are not due to erroneous easureents by either technique, but to spatial variation of pressure within and distal to the valve prostheses. This variation is ore pronounced in St. Jude valves with the occurrence of high localied gradients between the two leaflets of the valve. Agreeent or disagreeent between Doppler and catheter easureents depends on the site where catheter easureents are obtained. Furtherore, one should exercise caution when trying to deterine echanical valve stenosis fro Doppler gradient easureents alone. Calculation of valve area using the continuity equation ay also be inappropriate in echanical prostheses because velocity distribution across the valve orifice ay be inhoogeneous. This study deonstrates the iportance of considering valve type, valve sie, flow rate, and catheter position when using continuous wave Doppler to predict clinical catheteriation gradients in prosthetic valves.

9 Baugartner et al Doppler and Catheter Gradients in Aortic Prostheses 1475 Downloaded fro by on Septeber 27, 218 Acknowledgent We thank Ajit Yoganathan, PhD, for helpful discussions during this work. References 1. Hatle L, Angelsen BA, Tosdal A: Non-invasive assessent of aortic stenosis by Doppler ultrasound. Br Heart J 198; 43: Hegrenaes L, Hatle L: Aortic stenosis in adults - Noninvasive estiation of pressure differences by continuous wave Doppler echocardiography. Br Heart J 1985;54: Currie P, Seward JB, Reeder GS, Vliestra R, Bresnahan DR, Bresnahan JF, Sith HC, Hagler GJ, Tajik AJ: Continuouswave Doppler echocardiographic assessent of severity of calcific aortic stenosis: a siultaneous Doppler-catheter correlative study in 1 adult patients. Circulation 1985; 71: Requarth JA, Goldberg SJ, Dietrich Vasko S, Allen HD: In vitro verification of Doppler prediction of transvalve pressure gradient and orifice area in stenosis. A J Cardiol 1984; 53: Callahan MJ, Tajik AJ, Su-Fan Q, Bove AA: Validation of instantaneous pressure gradients easured by continuouswave Doppler in experientally induced aortic stenosis. A J Cardiol 1985;56: Teirstein PS, Yock PG, Popp RL: The accuracy of Doppler ultrasound easureent of pressure gradients across irregular, dual, and tunnellike obstructions to blood flow. Circulation 1985;72: Willias GA, Labovit AJ: Doppler heodynaic evaluation of prosthetic (Starr-dwards and Bjork-Shiley) and bioprosthetic (Hancock and Carpentier-dwards) cardiac valves. A J Cardiol 1985;56: Sipson IA, Fisher J, Reece IJ, Houston AB, Hutton I, Wheatley DJ: Coparison of Doppler ultrasound velocity easureents with pressure differences across bioprosthetic valves in a pulsatile flow odel. Cardiovasc Res 1986; 2: Wilkins GT, Gilla LD, Kriter GL, Levine RA, Palacios IF, Weyan A: Validation of continuous-wave Doppler echocardiographic easureents of itral and tricuspid prosthetic valve gradients: a siultaneous Doppler-catheter study. Circulation 1986;74: Sagar KB, Wann S, Paulsen WHJ, Rohilt DW: Doppler echocardiographic evaluation of Hancock and Bjork-Shiley prosthetic valves. JA Coil Cardiol 1986;7: Burstow DJ, Nishiura RA, Bailey KR, Reeder GS, Holes DR, Seward JB, Tajik AJ: Continuous-wave Doppler echocardiographic easureents of prosthetic valve gradients a - siultaneous Doppler-catheter correlative study. Circulation 1989;8: Holen J, Sionsen S, Froysaker T: An ultrasound Doppler technique for the noninvasive deterination of the pressure gradient in the Bjork-Shiley itral valve. Circulation 1979; 59: Rothbart RM, Sucker ML, Gibson RS: Overestiation by Doppler echocardiography of pressure gradients across Starr- dwards prosthetic valves in the aortic position. A J Cardiol 1988;61: Arabia FA, Talbot TL, Stewart SF, Nast P, Clark R: A coputeried physiologic pulse duplicator for in-vitro hydrodynaic and ultrasonic studies of prosthetic heart valves. Bioed Instru Technol 1989;23: Yoganathan AP, Jones M, Sahn DJ, Ridgeway A, Jioh A, Taura T: Bernoulli gradient calculations for echanical prosthetic aortic valves: in vitro Doppler Study (abstract). Circulation 1986;74(suppl II):II Yoganathan AJ, Corcoran WH: Pressure drops across prosthetic aortic heart valves under steady and pulsatile flow-in vitro easureents. J Bioechanics 1979;12: Levine RA, Jioh A, Cape G, McMillan S, Yoganathan AP, Weyan A: Pressure recovery distal to a stenosis: potential cause of gradient 'overestiation' by Doppler echocardiography. J A Coll Cardiol 1989;13: Yoganathan AJ, Chaux A, Gray RJ, Yi-Ren Woo, DeRobertis M, Willias FP, Matloff JM: Bileaflet, tilting disc and porcine aortic valve substitutes: in vitro hydrodynaic characteristics. JA Coll Cardiol 1984;3: Woo Y-R, Yoganathan AP: In vitro pulsatile flow velocity and shear stress easureents in the vicinity of echanical itral heart valve prostheses. J Bioechanics 1986;19: Chandran KB: Pulsatile flow past St. Jude Medical bileaflet valve - an in vitro study. J Thorac Cardiovasc Surg 1985; 89: Yang S, Bentivoglio LG, Maranhao V, Goldberg H: Fro Cardiac Catheteriation Data to Heodynaic Paraeters, ed 3. Philadelphia, F.A. Davis Co, 1988, p Grossan W: Cardiac Catheteriation and Angiography, ed 3. Philadelphia, Lea & Febiger, 1986, pp Wevans SJ, Mills P, Dawson J: The end of the p value? Br Heart J 1988;6: Bland JM, Altan DG: Statistical ethods for assessing agreeent between two ethods of clinical easureent. Lancet 1986;1: Hatle L, Angelsen B: Doppler Ultrasound in Cardiology: Physical Principles and Clinical Applications, ed 2. Philadelphia, Lea & Febiger, 1986, pp Clark C: The fluid echanics of aortic stenosis-i. theory and steady flow experients. J Bioechanics 1976;9: Clark C: The fluid echanics of aortic stenosis-ii. unsteady flow experients. J Bioechanics 1976;9: Tindale WB, Black MM, Martin TRP: In vitro evaluation of prosthetic heart valves: anoalies and liitations. Clin Phys Physiol Meas 1982;3: Schra D, Baldauf W, Meisner H: Flow pattern and velocity field distal to huan aortic and artificial heart valves as easured siultaneously by ultraicroscope aneoetry in cylindrical glass tubes. Thorac Cardiovasc Surg 198; 28: Hatle L, Angelsen B: Doppler Ultrasound in Cardiology: Physical Principles and Clinical Applications, ed 2. Philadelphia, Lea & Febiger, 1986, p Yoganathan AP, Woo Y-R, Sung H-W, Willias FP, Franch RH, Jones M: In vitro heodynaic characteristics of tissue bioprostheses in the aortic position. J Thorac Cardiovasc Surg 1986;92: Yoganathan AP, Valdes-Cru LM, Schidt-Dohna J, Jioh A, Berry C, Taura T, Sahn DJ: Continuous-wave Doppler velocities and gradients across fixed tunnel obstructions: studies in vitro and in vivo. Circulation 1987;76: Ridgway AJ: Ultrasound Doppler valuation of Mechanical Aortic Heart Valves. Thesis. Georgia Institute of Technology, Atlanta, Georgia, 1986 KY WORDS * prosthetic valve * echocardiography, Doppler e localied gradients * pressure recovery