Differences in Performance of Five Types of Aortic Valve Prostheses

Haemodynamic Assessment by Dobutamine Stress Echocardiography

Jeffrey P Khoo, Joan E Davies, Keng Leong Ang, Manuel Galianes, Derek T Chin Heart. 2013;99(1):41-47. Background In patients being considered for aortic valve replacement, there remains controversy over which design or tissue offers the best performance. We aimed to evaluate in a single study the haemodynamic performances of five different widely used aortic valve prostheses: stentless porcine xenograft (Elan), stentless bovine pericardium (Pericarbon Freedom), stented porcine xenograft (Aspire), stented bovine pericardium (More) and mechanical (Ultracor). We also compared them with normal aortic valves and stenosed valves of variable severity. Methods and results Preoperative echocardiography and dobutamine stress echocardiography at 1 year postoperatively were undertaken in 106 patients (n=1824 from each group). Stentless bioprostheses, whether porcine or bovine, displayed superior haemodynamics across nearly all echocardiographic parameters: lower gradients, larger effective orifice area, higher dimensionless severity index (DSI) and lower resistance, when compared with stented or mechanical prostheses. Comparing both stented designs, bovine tissue performed the worst at rest, but with stress, there was no difference. The stress performances of the stentless bioprostheses were similar to the mildly stenosed native aortic valve, whereas the performances of the stented and mechanical prostheses resembled that of native valves with mild-to-moderate stenoses. Haemodynamic differences, however, did not translate into differences in left ventricular mass reduction at 1 year. Conclusions Stentless bioprostheses displayed haemodynamics superior to stented or mechanical prostheses and had the closest performance to a normal, native aortic valve. Stress DSI data, least reliant on variable annulus/valve sizes and flow rates, provided the best haemodynamic discrimination.

Introduction
There is a wide selection of prostheses available for aortic valve replacement. [1 2] Controversy remains over which design offers the best performance. Stentless valves are designed to preserve interaction between the sinuses of Valsalva and the aortic valve, and in theory are more compliant to high flow rates generated by the exercising heart. Previous studies have indeed demonstrated that stentless bioprostheses have lower gradients and greater effective orifice areas (EOAs) compared with stented bioprostheses.[35] A meta-analysis, however, showed that different haemodynamic parameters may give differing results: although stentless valves had lower peak gradients, they did not display lower mean gradients.[6] Published data on EOAs may also be confounded by inaccuracies in measurements of the left ventricular outflow tract diameter (LVOTd), non-indexation to body surface area (BSA) and differences in prosthesis sizes. Few studies have compared performances during stress, which can be different from rest haemodynamics, and is clinically important in physically active patients. [4 711]

We therefore used dobutamine stress echocardiography (DSE) to compare systematically the performances of five representative aortic valve prostheses: stentless porcine xenograft, stentless bovine pericardium, stented porcine xenograft, stented bovine pericardium and mechanical tilting-disc. DSE is used clinically for assessing native and prosthetic valve function.[1214] Compared with rest echocardiography, DSE could reduce confounding by differences in systolic flow rates. We included assessment of mean pressure drop (MPD) versus a range of flow rates and valve resistance measurements.[1517] We explored the use of rest-stress dimensionless severity indices (DSIs) in assessing performance, as it removes confounding by LVOTd measurement errors and also partially normalises for varying stroke volume. Last, in order to place the performance of the prostheses in the context of native valves, we obtained DSE data from patients with normal trileaflet aortic valves and aortic stenoses of varying severity for comparison. We hypothesised that during stress, stentless valves, designed theoretically to maximise valve orifice areas, would be haemodynamically superior to stented valves; that different tissue compliances of porcine xenograft versus bovine pericardial bioprostheses may affect performance; and finally, that mechanical prostheses, having rigid orifices and defined maximal opening areas, would perform less well.

Methods
We compared five different prostheses: stentless porcine xenograft (Aortech Elan, Aortech, Glasgow, UK), stentless bovine pericardium (Sorin Pericarbon Freedom, Sorin Group, Gloucester, UK), stented porcine xenograft (Aortech Aspire, Aortech, Glasgow, UK), stented bovine pericardium (Sorin More, Sorin Group, Gloucester, UK) and mechanical tilting-disc (Aortech Ultracor, Aortech, Glasgow, UK). The study was performed in a single institution, and operations on patients with severe aortic stenosis were performed by two surgeons. Operations were done using standard cardiopulmonary bypass techniques, with warm blood cardioplegia, and valve sizing was performed using manufacturers' guidelines. The native aortic annulus diameter was measured by a sizer at the time of surgery. For stented and mechanical prostheses, valve sizes were equivalent to the measured annulus. For the stentless prostheses, valve sizes were chosen to be 2 mm greater than the annulus sizes. Coronary artery bypass surgery was performed concurrently if there was significant coronary disease. Patients with impaired LV systolic function were excluded. Preoperative echocardiography and DSE at 1 year postoperatively were undertaken. A total of 106 patients attended (n=1824 from each group). DSE was also performed on a total of 60 patients with normal trileaflet aortic valve (clinically indicated for assessment of chest pain) or with varying severity of aortic stenosis (13 mild, eight moderate and 18 severe cases). DSE was indicated for assessment of chest pain or for clarification of severity of aortic stenosis.[18] The study was approved by the local research ethics committee.

DSE Technique
Echocardiography was performed with Philips systems (Philips, Guildford, UK). -Blockers and other rate-limiting medications were discontinued for 48 h prior to stress studies. Subjects received low dose stress with intravenous dobutamine (520 g/kg/min in 5 g/kg/min increments every 35 min). Low dose stress was attained when the heart rate rose by 10 beats per minute (bpm) above the resting heart rate, and there was either measurable increase in left ventricular outflow tract velocity time integral (LVOT-VTI) or visible increase in myocardial contractility. At peak stress, 0.31.8 mg of atropine was added to attain a target heart rate of 85% of maximum predicted for age. Parasternal long and short axes views were obtained at rest. Apical 2-, 3- and 4chamber views of the LV were obtained at rest, low dose and peak stress. Aortic valve and LVOT Doppler

envelopes were recorded at rest and every stage of stress. Stress was discontinued if severe symptoms, hypotension, ischaemic contractile change, significant arrhythmia or aortic valve peak velocity of >5 m/s occurred.

LV Measurements
LV ejection fraction was calculated by the modified Simpson's biplane method of discs. BSA was estimated using the formula of Mosteller: BSA (m2) = ((weightkg x heightcm)/3600)0.5. LV mass (LVM) was calculated by the American Society of Echocardiography criteria, using the following equation: LVM = 0.8(1.04((LVIDd + PWTd + SWTd)3-(LVIDd)3)) + 0.6 g, where LVIDd was the left ventricular internal diameter, PWTd the posterior wall thickness and SWTd the inter-ventricular septal wall thickness, all measured in end-diastole. LVM indexed for BSA (LVMI) was calculated by LVM/BSA. LVMI was calculated before and 1 year after valve replacement and the reduction in LVMI was determined.

Aortic Valve Measurements

Peak and mean aortic valve gradients, in mm Hg, were obtained by tracing the continuous wave Doppler envelope, using the modified Bernoulli equation. EOA was calculated with the continuity equation: EOA = (3.14 x (LVOTD/2)2 LVOT-VTI)/(AV-VTI), where LVOTd was the inner to inner edge LVOTd measured in mid-systole immediately below the aortic leaflet hinge points in a zoomed parasternal long axis view, LVOT-VTI measured from pulsed wave Doppler envelope in the LVOT within 10 mm of the AV annulus in the 5-chamber view and AVVTI (aortic valve VTI) measured from continuous wave Doppler envelope across the AV in the 5-chamber view. DSI was calculated by dividing LVOT-VTI by AV-VTI. Severity of aortic stenosis was graded by EOA, according to European and American guidelines: EOA <1.0 cm2 was defined as severe; EOA = 1.01.5 cm2 as moderate; and EOA >1.52.0 cm2 as mild.[1921]

Aortic Valve Resistance Calculations

Aortic valve resistance was calculated as described previously.[17] MPD was calculated by subtracting mean LVOT gradient from mean aortic valve gradient. Stroke volume (SV) was calculated from the formula 3.14 (LVOTD/2)2 LVOT-VTI, and flow (Q) was calculated as SV (1000/ET), where ET is the LV systolic ejection time in milliseconds. For each individual patient, MPD was plotted against flow (Q) at each stage of stress and a linear regression was considered plausible if the Pearson correlation coefficient was >0.5. For individuals with a linear correlation of MPD versus Q, valve resistance (in dynes.s.cm5) was calculated as the slope of the regression line, Hg to multiplied by 1333, as 1333 is the conversion factor from mm dynes.s.cm 5.[15]

Statistics
Normally distributed continuous data were expressed as mean SE and analysed using one-way analysis of variance. Non-normally distributed or nominal data were expressed as median IQR and analysed using the KruskalWallis test. Categorical variables were analysed using 2 or Fisher's exact test as appropriate. A p value of 0.05 was taken to indicate significance.

Results
Demographic and operative data are shown in Table 1. As expected of current practice, patients receiving mechanical prostheses were younger and had lower Euroscores. Total bypass and cross-clamp times were on average 30 min longer in the stentless groups. The stentless porcine group had a lower BSA compared with the

stented porcine and mechanical groups (p=0.01 vs stented porcine and p=0.001 vs mechanical). Both stentless groups had smaller native mean annulus sizes compared with the stented and mechanical groups, and the stentless porcine group had a slightly smaller annulus compared with the stentless bovine (p<0.05). Prosthetic sizes are also shown in Table 1 and given as median IQR, as these are categorical variables. The median stented bovine prosthesis size was smaller than the stentless bovine prosthesis and the mechanical prosthesis (21 mm vs 25 mm, p=0.001 and p=0.02, respectively). During stress, heart rates rose on average from 74 bpm at rest to 90 bpm at low dose stress, and then to 133 bpm at peak stress; blood pressure was on average 153/82 at rest, 149/75 at low dose stress and 147/80 at peak stress (no differences between groups). DSE results are shown in Table 2. At rest, the stentless prostheses had the lowest gradients, with no difference between them. These gradients were significantly lower than those from the stented and mechanical prostheses (p<0.05). In keeping with this, the stentless bovine valve rest EOA was larger than the stented bovine and mechanical prosthetic rest EOAs (p<0.05). Conversely, the stentless porcine valve had similar rest EOAs to the stented porcine and mechanical prostheses. The stented bovine prosthesis had higher rest gradients than the stented porcine and the mechanical prostheses (p<0.01); it had the smallest rest EOA in comparison with the other four groups (p<0.01), as well as the smallest prosthesis size. At low dose and peak stress, there was no difference in gradients between the stentless prostheses; both stentless prostheses had lower gradients than stented and mechanical prostheses (p<0.01). The stentless bovine prosthesis had larger stress EOAs compared with the stentless porcine, stented bovine and mechanical prostheses (p<0.05). With stress, the stented bovine EOA increased to 1.4 cm2, against a static stented porcine EOA of 1.5 cm2. As a result, the difference in gradients seen at rest was no longer significant during stress. In contrast to EOAs, there were no differences in DSIs or EOAIs between the two stentless prostheses. The stentless prostheses had higher DSIs compared with the stented prostheses at rest (p<0.001 comparing bovine prostheses and p=0.07 comparing porcine prostheses). These differences became highly significant during peak stress (p<0.001 comparing bovine prostheses, and p<0.05 and p<0.001 comparing porcine prostheses at low dose and peak stress, respectively). The stentless valves also had higher DSIs compared with the mechanical valves (p<0.01 at rest and low dose stress, and p<0.001 at peak stress). When comparing the stented prostheses, we found a significant difference in DSI at rest, but not with stress. This was due to the stented bovine DSI, but not the stented porcine DSI, increasing with stress. Averaged linear regressions of MPD versus flow are plotted on figure 1 and individual regressions in figure 2. Most valves had a linear relationship with high Pearson correlation coefficients. The stentless prostheses again had the best performance, with the lowest resistance. Comparing porcine prostheses, the stentless design had lower resistance than the stented design but this did not reach statistical significance (p=0.18). Comparing bovine prostheses, the stentless design also had lower resistance which was short of statistical significance (p=0.08). Combining resistance data from the two stentless groups and comparing this with combined resistance from the two stented groups, there was a significantly lower resistance in the combined stentless group (83 vs 163 dynes.s.cm5, p=0.03).

(Enlarge Image) Figure 1. Averaged linear regression plots of mean pressure drop versus flow for valves demonstrating linear correlations. For each prosthesis or valve group, the average of the mean pressure drop was plotted against average flow, at rest and at each stage of stress, and a linear regression line plotted.

(Enlarge Image) Figure 2. Individual linear regression plots of mean pressure drop versus flow for each prosthesis with linear correlation. The averaged regression lines for each group are shown in bold. Comparing performances with normal or diseased aortic valves, the mean gradients of the stentless valves were similar to the gradients of mild native aortic stenosis (mean gradients 91, 164, 194 mm Hg at rest, low dose and peak stress; peak gradients 161, 316, 307 mm Hg). The stented and mechanical valves had gradients similar to moderate native aortic stenosis (mean gradients 132, 256, 254 mm Hg; peak gradients 243, 469, 447 mm Hg). EOA of the stentless bovine valve was closest to mild native aortic stenosis (1.760.04, 2.040.08 and 2.100.07 cm2 at rest, low dose and peak stress), whereas the other valves resembled mild to moderate stenoses (moderately stenosed valve EOAs were 1.190.06, 1.210.09, 1.290.11 cm2). The stentless prostheses had similar resistances to the mildly stenosed native valve (338 vs moderate 20925, and severe 31445 dynes.s.cm5). Finally, DSI changes in stentless valves were also closest to those of mild native aortic stenosis (0.670.04, 0.760.04 and 0.800.04 cm2 at rest, low dose and peak stress). DSI changes in the stented and mechanical valves resembled those of mild to moderate native aortic stenosis (native moderate aortic stenosis DSIs were 0.330.02, 0.340.02 and 0.35 0.02 at rest, low dose and peak stress). Despite the haemodynamic differences observed, there were no differences in LVM regression at 1 year (Table 1).

Discussion
This study used DSE to compare the haemodynamics of five representative aortic valve prostheses in common use against normal and stenotic native aortic valves. Previous studies have directly examined two types of prostheses, usually only with rest echocardiography, and with conflicting results. The Sorin Freedom stentless

bovine prosthesis has previously been demonstrated to have a lower resting aortic gradient and a larger orifice area, compared with the Sorin More stented bovine prosthesis.[3 5] In a second trial comparing the CarpentierEdwards Prima Plus bovine pericardium stentless prosthesis with the Carpentier-Edwards Perimount stented bovine pericardium prosthesis at rest, the stentless prosthesisnotably in smaller sizeshad a trend towards greater EOAs and lower gradients that did not reach significance (p=0.09).[22] A third trial (ASSERT) compared Medtronic Mosaic porcine xenograft stentless versus Freestyle porcine xenograft stented prostheses and found that the stentless group had lower peak gradients and larger EOAs.[23] A further meta-analysis showed that although stentless valves had lower peak gradients, they did not have lower mean gradients. [6] Cohen et al recently published long-term (10-year) follow-up data comparing the Carpentier-Edwards bovine pericardium stented and the Toronto porcine xenograft stentless prostheses and found superior haemodynamics (larger EOAs and lower gradients) in the stentless group.[4] They also undertook DSE, which showed greater haemodynamic differences with stress in favour of the stentless prosthesis. We found that stentless bioprostheses have superior haemodynamics to stented and mechanical valves at rest, which became more significant during stress. The only exceptional parameter was similar EOAs of the porcine stentless and stented prostheses, likely to be due to the differences in annular size and BSA. This was partly corrected for by using EOAI data. Data in Table 2 also demonstrate that the stentless porcine design allowed EOA, EOAI and DSI to increase in stress, whereas interestingly, the stented porcine EOA, EOAI and DSI did not do so. Superior haemodynamics of the stentless prostheses were demonstrated despite being implanted in patients with smaller annulus sizes. Relatively larger stentless prostheses can be implanted in the supra-annular position, whereas the intra-annular stented and mechanical valve sizes are governed by the native annulus sizes. In comparing the stented prostheses, the stented porcine xenograft had better rest haemodynamics (lower mean and peak gradients, higher EOA and DSI) than the stented bovine pericardium prosthesis. However, the stented porcine xenograft was unable to increase its EOA or DSI with stress. In contrast, the stented bovine pericardium prosthesis was able to increase both EOA and DSI at higher flow rates. The haemodynamic advantage of the stented porcine xenograft at rest therefore disappeared during stress, and the parameters of the two groups at both stages of stress became similar. Eichinger et al used exercise echocardiography to compare the Mosaic stented porcine xenograft valve with the Carpentier-Edwards Perimount stented bovine pericardium valve and found that mean pressure gradients were superior for the bovine prosthesis, albeit at small sizes of 21 and 23 mm.[24] In comparison, there were no differences in the echocardiographic parameters of the stentless porcine xenograft and bovine pericardium prostheses (equivalent gradients, DSIs and resistances) except that the stentless bovine EOAs were larger than the porcine stentless EOAs; this is likely due to differences in BSA, annular and prosthetic sizes, partly corrected for by using EAOI and DSI data. Nevertheless, our data suggest that tissue type affects performance in the stented but not the stentless design, and then only at rest. We did not have sufficient numbers to assess the impact of tissue type and valve design in the smaller prostheses. We also showed that the mechanical prostheses had the highest resistance. Resistance was significantly higher than in stentless prosthetic valves. Although the average resistance was also higher than in the stented prostheses, these comparisons did not reach statistical significance. With the other parameters, the mechanical prostheses had largely similar haemodynamics to stented bioprostheses. These findings are comparable with a previous study which showed that during stress, the Ultracor prosthesis had higher resistance than the Carpentier-Edwards porcine xenograft stented prosthesis.[7] A second smaller study found higher stress gradients in a bileaflet mechanical prosthesis compared with a stentless porcine xenograft. [25]

We found that DSI, particularly at peak stress, was the most useful index for comparing prostheses of varying designs and sizes. Unlike EOA or EOAI, DSI does not rely on accurate LVOT measurements, thereby reducing variability when prostheses of different sizes are compared. Dunning et al previously used rest echocardiography to compare the Sorin Freedom stentless bovine pericardium versus the Sorin More stented bovine pericardium prosthetic valves and concluded that the stentless valve resulted in lower gradients and greater EOA. [3] However, the average size of their stentless prostheses was significantly larger than the stented prostheses. Therefore, their observed differences may be due to larger prosthetic size rather than design. They did not compare DSI. The use of DSI in assessing aortic prostheses is supported by the American Society of Echocardiography. [26 27] Valve resistance data provided further discrimination, but may still be partially confounded by the different annular and prosthetic sizes. Moreover, quantification is time-consuming and linear correlations cannot always be achieved. Our resistances obtained for native mild, moderate and severe aortic stenosis were comparable with previous published data.[17] Overall performances of the stentless bioprostheses were very similar to the mildly stenosed native aortic valve, whereas performances of the stented and mechanical prostheses were closer to native mild-to-moderate stenoses. However, there were no differences in LVM reduction between all groups at 1 year, as noted in previous studies.[36 2223 28] This may suggest that the haemodynamic advantage of stentless valves, which took longer to implant, did not translate into clinical advantage. However, the finding may also be explained by the fact that none of the prostheses performed worse than native valves with mild-to-moderate aortic stenosis. Other confounding factors, such as age, activity or medication, may coexist across the groups in this non-randomised study. Two previous studies did find that LV reverse remodelling occurred earlier in stentless compared with stented valves at 6 months.[3 5] In our study, we did not measure LVM at this timepoint. In terms of study limitations, our sample sizes were relatively small. However, the number of patients was similar to previous studies,[7] and in most parameters, we achieved consistently meaningful and significant comparisons. Second, having valves of various sizes in each group may have affected some comparisons, particularly with EOAs. However, we were able to partly overcome this by using DSI and EOAI. Third, having more stress stages would have improved the linear regression plots and resistance measurements. Fourth, valve replacements vary widely by design, and our findings may not be generally applicable. For example, there were no haemodynamic differences between the Toronto stentless prosthesis and the Perimount stented prosthesis. [8] Even within stentless designs, the Cryolife O'Brien single suture stentless valve performed better than the Toronto stentless.[29] In summary, we have compared five different commonly used aortic valve prostheses using DSE and demonstrated that stentless bioprostheses, regardless of tissue type, had superior haemodynamics, compared with stented or mechanical prostheses. DSE provides a method for comparing different aortic prostheses in an era of new transcatheter and sutureless designs.