Hemodynamic performance of the Medtronic Mosaic and Perimount Magna aortic bioprostheses: five-year results of a prospectively randomized study

European Journal of Cardio-thoracic Surgery 39 (2011) 844 852 www.elsevier.com/locate/ejcts Hemodynamic performance of the Medtronic Mosaic and Perimount Magna aortic bioprostheses: five-year results of a prospectively randomized study María José Dalmau *, José María González-Santos, José Antonio Blázquez, José Alfonso Sastre, Javier López-Rodríguez, María Bueno, Mario Castaño, Antonio Arribas Department of Cardiac Surgery, Salamanca University Hospital, Paseo de San Vicente, N8 58 182, 37007 Salamanca, Spain Received 1 September 2010; received in revised form 30 October 2010; accepted 4 November 2010; Available online 28 December 2010 Abstract Objective: Clinical outcomes of patients undergoing aortic valve replacement may be influenced by the presence of residual gradients and patient prosthesis mismatch. The aim of this study was to compare hemodynamic performance and clinical outcomes at 5 years after prospectively randomized porcine versus bovine aortic valve replacement. We also aimed to determine the effects of valve hemodynamics on left ventricular (LV) mass regression. Methods: A total of 108 patients undergoing aortic valve replacement were randomized to receive either the Medtronic Mosaic (MM) porcine (n = 54) or the Edwards Perimount Magna (EPM) bovine pericardial prosthesis (n = 54). Clinical outcomes, mean gradients, effective orifice area and LV mass regression were evaluated at 1 and 5 years after surgery. Follow-up echocardiograms were performed on 106 (98%) and 87 (92%) patients, respectively. Results: Preoperative characteristics were similar between groups. Mean aortic annulus diameter and mean implant size were comparable in both groups. At 1 and 5 years, mean transprosthetic gradients were lower in the EPM group: EPM 10.3 3.4 mmhg versus MM 16.3 7.6 mmhg ( p < 0.0001) and EPM 9.6 3.5 mmhg versus MM 16.8 8.7 mmhg ( p < 0.0001), respectively. Similarly, indexed effective orifice areas (IEOA) at 1 and 5 years were significantly greater in the EPM group: EPM 1.10 0.22 cm 2 m 2 versus MM 0.96 0.22 cm 2 m 2 ( p < 0.004) and EPM 1.02 0.25 cm 2 m 2 versus MM 0.76 0.19 cm 2 m 2 ( p < 0.0001), respectively. At 5 years, the incidence of patient prosthesis mismatch (IEOA 0.85 cm 2 m 2 ) was significantly lower in the EPM group: EPM 22.9% vs MM 73.9% ( p < 0.0001). Such differences were similar when analysis was stratified by surgically measured annular size and implant valve size. During the first year after surgery, both groups demonstrated similar regression of LV mass index (MM 26.3 43 g m 2 vs EPM 30.1 36 g m 2 ; p = 0.8); however, at 5 years, regression of LV mass index was significantly greater in the EPM group: (EPM 47.4 35 g m 2 vs 4.4 36 g m 2 ; p < 0.0001). Five-year survival was 79.6 4.1% in the MM group and 94.4 2.2% in the EPM group ( p = 0.03). Conclusions: At 5 years, the EPM valve was significantly superior to the MM prosthesis with regard to hemodynamic performance, incidence of patient prosthesis mismatch and regression of LV mass index. The hemodynamic superiority of the EPM prostheses in comparison to MM-prostheses demonstrated at 1 year, increased significantly over time. # 2010 European Association for Cardio-Thoracic Surgery. Published by Elsevier B.V. All rights reserved. Keywords: Aortic valve replacement; Biological prosthesis; Hemodynamic performance; Left ventricular mass index 1. Introduction Contemporary practices in aortic valve replacement (AVR) have seen an increasing use of biological valve substitutes because of the growing number of elderly patients requiring surgery and the anticoagulant-related complications associated with the use of mechanical prostheses. Although aortic biologic prostheses, both porcine and bovine, have proven to be clinically reliable over time, they have undergone modifications in design during the past decades to optimize hemodynamic performance and prolong durability. Presented at the 24th Annual Meeting of the European Association for Cardio-thoracic Surgery, Geneva, Switzerland, September 11 15, 2010. * Corresponding author. Tel.: +34 923291263; fax: +34 923291263. E-mail address: dalmau_mjo@gva.es (M.J. Dalmau). The hemodynamic performance of aortic substitutes has been the focus of many investigations due to the influence of patient prosthesis mismatch (PPM) on left ventricular (LV) mass regression and clinical outcome after AVR [1,2]. Compelling evidence suggests that patients with PPM or high residual transprosthetic gradients have lesser symptomatic improvement and experience poorer LV mass regression at intermediate and long-term follow-up [3 5]. Despite continuous improvement in design and manufacturing of aortic valve biological substitutes, all of them produce a certain degree of LV outflow obstruction. The variability in residual gradients between different aortic bioprostheses may be clinically relevant, whereas the surgical objective sought is to minimize gradients for a given annular size. We conducted a prospective randomized study comparing two biological and last-generation supra-annular aortic valve substitutes, the Edwards Perimount Magna (EPM) pericardial 1010-7940/$ see front matter # 2010 European Association for Cardio-Thoracic Surgery. Published by Elsevier B.V. All rights reserved. doi:10.1016/j.ejcts.2010.11.015

M.J. Dalmau et al. / European Journal of Cardio-thoracic Surgery 39 (2011) 844 852 845 xenograft and the Medtronic Mosaic porcine bioprosthesis (MM). With the aim to provide more objective data, both devices were compared taking the diameter of the patient s aortic annulus as a reference rather than the less reliable manufacturer s labeled valve size [6]. The objective of this study was to compare survival and hemodynamic performance at 1 and 5 years after AVR, and to determine the effects of valve hemodynamics on regression of LV hypertrophy. 2. Patients and methods 2.1. Patients Between February 2004 and February 2006, a total of 116 consecutive patients scheduled to have bioprosthetic valve replacement in the aortic position were randomized to receive either an EPM valve or an MM bioprosthesis. Patients undergoing an isolated AVR and those requiring associated aortocoronary bypass grafting, ascending aortic surgery, or tricuspid annuloplasty were included in the study. Exclusion criteria were the replacement of more than one valve or a pre-existing prosthetic valve in another position. Appropriate institutional research ethics board approval was obtained. Patients were informed of the study and provided written consent. Preoperative, postoperative 12-month and 5-year patient characteristics were extracted from databases constructed during follow-up phases of this randomized trial. The medical records and the echocardiographic outcomes at 1 and 5 years were analyzed. We compared demographics, preoperative clinical data, operative data, hemodynamic profiles, and clinical outcomes at 1 and 5 years. Primary outcomes included transvalvular gradients, effective orifice areas (EOAs), and the regression of LV mass index (LVMI) measured with two-dimensional (2D) echocardiography at 1 and 5 years. Survival and clinical outcomes were secondary end points in this follow-up study. 2.2. Sample size Based on earlier studies on LVMI regression after AVR with the Carpentier Edwards Perimount prosthesis [7], an incidence of 41% patients having any residual LV hypertrophy 1 year after surgery was assumed and, therefore, was chosen to determine the sample size of our study. Our objective was to demonstrate that the incidence of successful LV hypertrophy regression with the MM valve would be <20% than that with the EPM prosthesis. A total of 86 patients in two randomized groups was required to declare a significant difference with a b = 0.8 and a = 0.05. 2.3. Randomization The randomization was computer generated and incorporated into sealed envelopes to allow for consecutive intraoperative allocation. Randomization was performed in the operating theater using the sealed envelope technique after patient eligibility was confirmed. Patients were randomized to receive either the EPM valve or the MM valve. After randomization and because of procedural difficulties, two patients randomized to receive a Magna valve received a Mosaic valve, and two randomized to receive a Mosaic valve received a Magna, instead. A total of 108 hospital survivors were finally included in the study and comprised 54 patients with a Mosaic and 54 with a Magna valve. 2.4. Operative technique Operations were performed using standard cardiopulmonary bypass techniques, including mild systemic hypothermia and both antegrade and retrograde cold blood cardioplegia. The native aortic valve was excised and the exact inner aortic annular diameter was assessed based on manual measurements using standardized metric sizers (graduated in millimeters). Thereafter, sizing for both valve types was undertaken in each patient using the appropriate original sizer provided by each manufacturer before the randomization envelope was opened. Surgeons were extremely consistent in selecting similar valve sizes for a given annular diameter, regardless of prosthesis type and avoided any oversizing. As such, each surgeon was required to commit to a specific valve size before valve selection. This protocol was designed to prevent surgeon-specific selection bias. All valves were implanted in the supra-annular position using interrupted, pledgetsupported, non-everting mattress sutures. No patients underwent annular enlargement procedures. For the purpose of the study and further comparisons, patients were stratified for annulus size in three categories: <22 mm, 22 23 mm, and >23 mm. 2.5. Bioprostheses The third-generation MM bioprosthesis (Medtronic, Inc, Minneapolis, MN, USA) is a stented porcine heart valve, which is fixed with glutaraldehyde by using a combination of the zeropressure and root-pressure methods to preserve the natural morphology of the fibers in the leaflets. The Mosaic tissue is treated with alpha-amino-oleic acid to reduce the buildup of calcium. It has been in clinical use since 1994 (Europe) and 2000 (United States), respectively, and its hemodynamic performance and freedom rates from adverse events have been found to be highly satisfactory [8]. Introduced in 2002, the Carpentier Edwards Perimount Magna aortic xenograft (Edwards Lifesciences LLC, Irvine, CA, USA) is a modification of the standard Perimount valve. The EPM prosthesis consists of bovine pericardium mounted on an Elgiloy frame. The cusps are fixed with glutaraldehyde at low pressure and are treated with surfactant combined with thermal treatment (XenoLogiX and ThermaFix processes) to retard calcification. The EPM is characterized by a design specifically intended for supraannular positioning that is claimed to have better hemodynamic and flow characteristics. Although its short-term hemodynamic performance was proved to be superior to that of the Perimount standard model [9], up until now, the longterm hemodynamic results of EPM prostheses are not available. Nevertheless, the Perimount standard bioprosthesis has been in clinical use since 1981 and its long-term clinical and hemodynamic results have previously been reported to be excellent [10].

846 M.J. Dalmau et al. / European Journal of Cardio-thoracic Surgery 39 (2011) 844 852 Comparison of valve performance requires uniform measurements of valve size; and manufacturer s labeled valve size has no standard and can be misleading [11]. Labeled valve size is related to different features of the external diameter of the prostheses; thus, the internal orifice for a given valve size may vary widely among types of prostheses. In general, the inner diameter of the MM prosthesis is smaller across all sizes than the inner diameter of the EPM prosthesis, whereas the external sewing ring diameter of the MM valve is 1 mm larger in sizes 19 and 21; 2 mm larger in size 23; and 3 mm larger in sizes 27 and 29 in comparison to the EPM valve. 2.6. Echocardiographic assessment Patients were followed up by transthoracic Doppler echocardiography at 1 year (median 12 1.5 months) and 5 years postoperatively (median 4.9 0.8 years). The modified Bernoulli equation was used to calculate peak and mean pressure gradients across the prosthetic valve. EOA was calculated by the continuity equation and indexed to body surface area to assess the presence of PPM. According to previous investigations [4,5], PPM was considered as not significant (i.e., mild or no PPM), if the indexed EOA was >0.85 cm 2 m 2 ; significant PPM was defined by indexed EOA 0.85 cm 2 m 2 ; and severe mismatch if indexed EOA was 0.65 cm 2 m 2 or less. LV dimensions were measured according to the recommendations of the American Society of Echocardiography (ASE). LV mass (LVM) was calculated with the corrected ASE formula as follows [12]: LVM = 0.8 1.04 (IVS d + LVID d + PWT d )3 LVID d 3 + 0.6, where IVS d is the enddiastolic interventricular septum thickness, LVID d is the LV end-diastolic internal diameter, and PWT d is the LV enddiastolic posterior wall thickness. Residual LV hypertrophy was defined as an LVM index >131 g m 2 in males and >100 g m 2 in females. LVM regression was calculated by subtracting the mass index at follow-up from the mass index preoperatively. 2.7. Statistical analysis The statistical analysis was performed using the Statistical Package for Social Sciences (SPSS) 17.0 statistical software for Windows (SPSS Inc., Chicago, IL, USA). Continuous variables were expressed as mean values standard deviation (SD) and time variables as median values SD. Comparisons were performed using a t-test in case of normal data distribution and the Mann Whitney U test in case of not normally distributed data. For measurements within groups over time, paired t-test or Wilcoxon test were applied. Categorical variables were presented as frequencies and percentages. Associations among categorical variables were compared by Pearson s x 2 test or Fisher s exact test as appropriate. Statistical analysis of the association of variables was performed with the Pearson (r) or Spearman (r s ) correlation coefficients. After univariate analysis, we include significative variables in a multivariate logistic regression model. A stepwise backward elimination using the likelihood ratio test with elimination defined by a p-value of 0.1 or greater was performed. For each of the explanatory variables, we calculated the coefficient, the odds ratio (OR) and the confidence interval (CI). Survival curves were determined by means of Kaplan Meier method, and comparisons were made using a log-rank test. Statistical significance was defined as a p value of <0.05. 3. Results A total of 112 consecutive patients selected for elective bioprosthetic AVR were prospectively assigned to receive either an EPM valve or an MM bioprosthesis. There were four perioperative deaths, none related to the implanted prosthesis. Thus, a total of 108 hospital survivors (EPM n = 54, MM n = 54) were finally included in the study. Patient preoperative characteristics and operative data were similar for both groups (Table 1). Based on manufacturer s labeled size, there was a difference of borderline statistical significance in mean implanted valve size between groups (EPM 22.7 1.9 mm vs MM 23.4 2.1 mm, p = 0.051). However, actual internal annular diameters were not significantly different in both groups (EPM 23.8 2.1 mm vs MM 23.4 2.3 mm; p = 0.8). Patients were grouped by intra-operatively measured aortic annulus diameter (AAD) as follows: <22 mm (n = 16), 22 23 mm (n = 39) and >23 mm (n = 53). 3.1. Hemodynamic measurements Among 108 patients participating in this trial, a total of 106 patients were echocardiographically evaluated at a median follow-up of 12 months, with a complete follow-up in 98% of the patients. Five-year echocardiograms were performed on 87 of 94 eligible patients (92%). Comparisons of hemodynamic data for both valve types at 1- and 5-year follow-up are listed in Table 2. The EPM prosthesis showed significantly lower mean transvalvular Table 1. Preoperative patient characteristics and surgical data. EPM (n = 54) MM (n = 54) p value Gender (male/female) 34/20 32/22 0.69 Age 75.6 4.5 75.4 4.2 0.77 Body surface area (m 2 ) 1.75 0.16 1.73 0.14 0.32 EuroSCORE (additive) 6.54 1.6 6.54 1.8 1.00 NYHA functional class 0.31 NYHA I II 32 (59%) 34 (62%) NYHA III IV 22 (41%) 20 (37%) Aortic valve lesion 0.09 Stenosis 29 (54%) 26 (48%) Insufficiency 9 (17%) 4 (8%) Mixed 16 (29%) 24 (44%) Etiology 0.15 Rheumatic 5 (9%) 1 (2%) Calcific 49 (91%) 53 (98%) Additional procedures 0.87 CABG 17 (31%) 16 (30%) TA 3 (5%) 2 (4%) AAS 4 (7%) 5 (9%) CPB time (min) 108 33 103 32 0.44 Aortic cross-clamp time (min) Isolated procedures 67 14 62 13 0.14 Combined procedures 96 27 94 20 0.77 EPM: Edwards Perimount Magna; MM: Medtronic Mosaic; CABG: coronary artery bypass graft; TA: tricuspid annuloplasty; AAS: ascending aorta surgery; CPB: cardiopulmonary bypass. Variables are presented as mean SD values or as number of patients (%).

M.J. Dalmau et al. / European Journal of Cardio-thoracic Surgery 39 (2011) 844 852 847 Table 2. Echocardiograms 1 and 5 years postoperatively. 1 year 5 years Differences 1 year vs 5 years p value 1 year vs 5 years Peak gradient (mmhg) Magna 19.7 6.1 20.8 6.3 + 1.3 6.0 0.11 Mosaic 30.4 4.6 36.2 16.6 + 4.8 9.7 0.001 p value <0.0001 <0.0001 0.04 Mean gradient (mmhg) Magna 10.3 3.4 9.6 3.5 0.5 3.6 0.3 Mosaic 16.3 7.6 16.8 8.7 + 0.2 6.1 0.9 p value <0.0001 <0.0001 0.6 EOA (cm 2 ) Magna 1.94 0.44 1.81 0.51 0.13 0.34 0.1 Mosaic 1.66 0.40 1.31 0.32 0.35 0.21 <0.0001 p value 0.001 <0.0001 0.02 Indexed EOA (cm 2 m 2 ) Magna 1.10 0.22 1.02 0.25 0.09 0.21 0.009 Mosaic 0.96 0.22 0.76 0.19 0.19 0.11 <0.0001 p value 0.004 <0.0001 0.02 PPM a Magna 9.2% 22.9% +13.6% 0.007 Mosaic 30.1% 73.9% +43.8% <0.0001 p value 0.006 <0.0001 0.01 EOA: effective orifice area; PPM: patient prosthesis mismatch. a Indexed EOA < 0.85 cm 2 m 2. gradients at 1 and 5 years than did MM valves (EPM 10.3 3.4 mmhg vs MM 16.3 7.6 mmhg, p < 0.0001; EPM 9.6 3.5 mmhg vs MM 16.8 8.7 mmhg, p < 0.0001). Further, average EOAs were significantly larger at these time points ( p = 0.001, p < 0.0001). Similarly, 1 and 5 years echocardiographic studies revealed significantly greater indexed effective orifice areas (IEOAs) in the EPM group: EPM 1.10 0.22 cm 2 m 2 versus MM 0.96 0.22 cm 2 m 2 ( p = 0.004); EPM 1.02 0.25 cm 2 m 2 versus MM 0.76 0.19 cm 2 m 2 ( p < 0.0001). Similar hemodynamic performance of both prosthesis types was observed when patients were grouped according to AAD (Table 3). Further, for each AAD group, mean pressure gradients at 1 and 5 years were slightly lower for the EPM valves. This difference was statistically significant in patients with an AAD of 22 23 mm and >23 mm. Accordingly, significantly larger EOAs were obtained for each AAD group at 1 and 5 years, especially in AAD of 22 23 mm ( p < 0.01). Similarly, in each AAD group, the EPM prosthesis showed slightly higher IEOA, reaching statistical significance in AAD of 22 23 mm and >23 mm. No hemodynamic differences were demonstrated in patients with an AAD of <22 mm, although an obvious trend toward better hemodynamics was also seen in this group. These differences were also apparent between groups when compared by industry-labeled valve size (Table 4). When comparing individual prosthesis sizes, EPM prostheses showed significantly lower mean transvalvular gradients and larger IEOAs at 1 and 5 years, reaching statistical significance when comparing the 21-mm, 23-mm and 25-mm labeled valves. Hemodynamic data for both prosthesis types revealed significant changes over time (Table 2). The mean changes in peak transprosthetic gradients between 1- and 5-year Table 3. Hemodynamic performance at 1 and 5 years postoperatively according to aortic annulus diameter. Size 1 year follow-up (98%) 5 years follow-up (92%) Magna Mosaic p value Magna Mosaic p value AAD <22 (n = 16) Mean gradient (mmhg) 11.4 4.6 18.6 9.9 0.07 14.4 2.7 15.8 4.3 0.4 EOA (cm 2 ) 1.60 0.29 1.44 0.30 0.3 1.27 0.25 1.23 0.28 0.7 IEOA (cm 2 m 2 ) 0.98 0.19 0.90 0.23 0.5 0.79 0.17 0.77 0.20 0.8 PPM 16.6% 30% 0.5 30% 71% 0.5 AAD 22 23 mm (n = 39) Mean gradient (mmhg) 10.1 3.3 17.7 7.9 <0.001 10.58 4.4 18.8 12 <0.02 EOA (cm 2 ) 1.77 0.41 1.50 0.18 <0.01 1.72 0.50 1.22 0.22 <0.01 IEOA (cm 2 m 2 ) 1.03 0.23 0.88 0.08 <0.01 1.00 0.24 0.72 0.13 <0.001 PPM 11.7% 33.3% <0.05 21.4% 84% <0.0001 AAD >23 mm (n = 53) Mean gradient (mmhg) 10.2 3.3 13.8 5.3 <0.003 8.3 2.1 15.1 5.6 <0.0001 EOA (cm 2 ) 2.11 0.42 1.90 0.47 0.1 1.95 0.49 1.46 0.38 <0.001 IEOA (cm 2 m 2 ) 1.14 0.20 1.06 0.27 0.2 1.07 0.25 0.80 0.22 <0.001 PPM 6.4% 27.3% <0.04 17.2% 64.7% <0.002 AAD: aortic annulus diameter; EOA: effective orifice area; IEOA: indexed effective orifice area; PPM: patient prosthesis mismatch.

848 M.J. Dalmau et al. / European Journal of Cardio-thoracic Surgery 39 (2011) 844 852 Table 4. Hemodynamic performance at 1 and 5 years postoperatively according to labeled valve size. Size 1 year follow-up (98%) 5 years follow-up (92%) Magna Mosaic p value Magna Mosaic p value 19 mm (n =3) MG (mmhg) 15.2 4.3 13.9 0.8 15.5 4.9 16.0 0.9 IEOA (cm 2 m 2 ) 0.87 0.21 0.95 0.8 0.69 0.14 0.69 0.9 21 mm (n = 34) MG (mmhg) 11.2 3.2 22.3 10.7 <0.0001 11.3 3.8 22.9 14 <0.003 IEOA (cm 2 m 2 ) 0.99 0.21 0.83 0.14 <0.01 0.90 0.15 0.67 0.15 <0.001 23 mm (n = 41) MG (mmhg) 0.98 3.4 15.2 5.5 <0.001 8.6 2.5 14.3 3.3 <0.0001 IEOA (cm 2 m 2 ) 1.16 0.21 0.98 0.17 <0.005 1.12 0.28 0.77 0.14 <0.0001 25 mm (n = 17) MG (mmhg) 8.1 2.2 11.1 3.5 <0.05 7.5 2.3 10.1 5.6 <0.05 IEOA (cm 2 m 2 ) 1.22 0.18 1.00 0.24 <0.05 1.18 0.23 0.77 0.20 <0.007 27 mm (n = 13) MG (mmhg) 10.6 3.2 14.5 4.3 0.1 8.6 1.4 17.6 4.8 <0.01 IEOA (cm 2 m 2 ) 1.04 0.01 1.06 0.34 0.9 0.87 0.06 0.85 0.2 0.9 MG: mean gradient; IEOA: indexed effective orifice area. echocardiograms were +1.39 mmhg for EPM prostheses ( p = 0.1) and +4.81 mmhg for MM valves ( p = 0.001). Changes after 5 years in peak transvalvular gradients between EPM and MM prostheses were statistically significant ( p = 0.04). Similarly, changes over time with respect to EOA and IEOA were relevant between both bioprostheses ( p < 0.02), the mean change in IEOA being 0.09 cm 2 m 2 for EPM prostheses ( p = 0.009) and 0.19 cm 2 m 2 for MM valves ( p < 0.0001, Fig. 1). A significant correlation between IEOAs and mean transvalvular gradients were demonstrated in both subgroups at 1 year (EPM group r = 0.366, p < 0.006; MM group, r = 0.365, p < 0.007) and 5 years (EPM group, r = 0.634, p < 0.0001; MM group, r = 0.325, p = 0.02). The prevalence of significant PPM was different according to the type of the implanted bioprosthesis (Table 2). At first year, 30.1% of patients with an MM valve had an IEOA 0.85 cm 2 m 2, whereas this occurred only in 9.2% of those with an EPM valve ( p = 0.006). This difference increased over time and, after 5 years, differences were even more distinctive (EPM 22.9% vs MM 73.9%, p < 0.0001). At 1 [()TD$FIG] Fig. 1. Mean indexed effective orifice areas (95% confidence intervals) of Mosaic and Magna prosthesis at 1 and 5 years after aortic valve replacement. IEOA: indexed effective orifice area. year, severe mismatch occurred in 7.5% of MM valves compared with 1.8% EPM valves ( p < 0.005) and, at 5 years, the prevalence of severe PPM was 32.5% in the MM group and 4% in the EPM group ( p < 0.0001). The presence of PPM was different according to the valve type and the AAD (Table 3). A significant percentage of patients with small AAD (<22 mm) showed a mismatch in both groups at 1 years (MM 30.1% vs EPM 16.6%) and 5 years (MM 71% vs EPM 30%). In the EPM group, the incidence of significant PPM decreased with increasing AAD; however, in the MM group, PPM was present constantly in all AAD. Changes in LVM and LVMI between preoperative echocardiographic measurements and follow-up are shown in Table 5. There was no difference between groups in baseline values of LVM ( p = 0.5) or LVMI ( p = 0.6). During the first year after implantation, LVM and LVMI significantly decreased in both groups. At this time point, there were no significant differences between both valve types regarding absolute LVM regression (EPM 49.2 59.3 vs MM 51.5 77.0) and absolute LVMI reduction (EPM 30.0 36.2 vs MM 26.3 43.8). Neither valve had a significant early sizerelated advantage when patients were stratified by AAD. Mass regression continued up to 5 years in the EPM group, although most of the effects occurred during the first postoperative year. Between the first and fifth year postoperatively, ventricular mass remained relatively stable and did not experience further regression in patients with MM valves while decreasing in patients with EPM prostheses to a significant degree. Therefore, the absolute LVM regression at 5 years follow-up was significantly greater in patients with EPM prosthesis (EMP 82.6 61.1 vs MM 10.4 63.4; p < 0.0001). A similar trend was demonstrated with respect to LVMI, which decreased significantly over time in the EPM group (Fig. 2). At 5 years, differences in LVMI reduction were statistically significant in favor of the EPM prostheses (EPM 47.4 35.1 vs 4.3 36.1; p < 0.0001). By means of simple linear regression analysis, relationships between absolute LVM index regression were correlated to the 5-year IEOA with r = 0.484 ( p < 0.001); to the mean transprosthetic gradient at 5 years with r = 0.310 ( p = 0.002); and to the presence of patient prosthesis mismatch with r s = 0.475 ( p < 0.0001).

M.J. Dalmau et al. / European Journal of Cardio-thoracic Surgery 39 (2011) 844 852 849 Table 5. Echocardiograms 1 and 5 years postoperatively. Preoperative 1 year 5 years p value (preop vs 1 y) p value (preop vs 5 y) LV mass (g) Magna 259.5 71.0 222.5 59.5 175.1 38.5 <0.0001 <0.0001 Mosaic 271.4 87.1 220.7 74.8 226.9 82.1 0.008 0.3 p value 0.5 0.8 <0.0001 LVM index (g m 2 ) Magna 149.3 40.9 125.6 30.4 99.7 20.4 <0.0001 <0.0001 Mosaic 152.7 43.7 125.8 42.6 129.6 42.1 0.01 0.5 p value 0.6 0.4 <0.0001 Regr. LVM (g) Magna 49.2 59.3 82.6 61.1 <0.0001 Mosaic 51.5 77.0 10.4 63.4 0.1 p value 0.8 <0.0001 Regr.LVMI (g m 2 ) Magna 30.0 36.2 47.4 35.1 <0.0001 Mosaic 26.3 43.8 4.3 36.1 0.08 p value 0.8 <0.0001 LVM: left ventricular mass; LVMI: left ventricular mass index; Regr.: regression. Although there was an LVMI reduction in both groups, the average LVMI at 5 years for the entire series remained greater than normal in 42.8% of the patients; 60.5% of MM patients and 27.1% of EPM patients had LV hypertrophy 5 years after surgery (x 2, p = 0.002). A multivariate logistic regression analysis revealed the MM valve to be strongly associated with residual LV hypertrophy (OR 10.7, 95% CI 2.6 44.1, p = 0.001). IEOA at 5 years was also identified as a predictor of residual LV hypertrophy (OR 12.6, 95% IC 1.7 21.8, p = 0.005). Neither mean gradients nor the IEOA and PPM at 1 year were statistically significant when using multivariate modeling. At 5-year follow-up, overall survival (freedom from allcause mortality) was 79.6% 4.1% in the MM group (11 patients) and 94.4 2.2% (three patients) in the EPM group ( p = 0.039, Fig. 3). Simple linear regression analysis showed a correlation, between time of survival and the absolute LVMI regression at 5 years (r = 0.23, r 2 = 5%, p = 0.03). During follow-up, three patients in EPM group (5.6 0.1%) and 11 patients in MM-group (20.4 0.2%) died. Death occurred in the first year in 1.8% of MM patients versus none of the EPM patients; and, between the first and fifth year, three patients in the EPM group (5.6 0.1%) versus 10 patients in the MM group (18.5 0.2%) died. All deaths were not related to the valve. Causes of death were malignancies in one EPM [()TD$FIG] and three MM patients. One EPM patient and three MM patients died of infections. Three MM patients died of multiorgan failure after abdominal or urologic surgery. Neurological events were the cause of death in one patient of each group. There was no endocarditis or valve thrombosis. One patient in the MM group underwent re-operation for new onset mitral regurgitation and did not survive due to perioperative complications. Good hemodynamic function was documented in the majority of patients on follow-up echocardiographic measurements. Four patients in the MM group presented significant transvalvular stenosis with reduced prosthetic areas and transvalvular flow velocities >4 ms 1, all four patients being less symptomatic. 4. Discussion The aims of AVR include a reduction of transvalvular gradients to minimal levels, an increase in EOA to allow maximal forward flow and a complete regression of LV hypertrophy. Maximization of EOA area and minimization of [()TD$FIG] Fig. 2. Regression of left ventricular mass index (LVMI) over time in patients with Magna (broken line) and Mosaic (solid line) aortic valve prosthesis. Fig. 3. Actuarial overall survival after stented bovine aortic valve replacement (broken line), and stented porcine aortic valve replacement (solid line).

850 M.J. Dalmau et al. / European Journal of Cardio-thoracic Surgery 39 (2011) 844 852 transprosthetic gradient are seen as the hemodynamic objectives for an aortic prosthesis. Compelling evidence suggests that persistently elevated transvalvular gradients negatively affect the optimal regression of LV hypertrophy [1,13]. Stented biologic aortic substitutes, both porcine and bovine, are prone to generate some postoperatively transvalvular gradients due to suboptimal leaflet opening and obstruction by sewing rings and stents. The ratio of flow orifice diameter to external valve mounting dimensions is one of the most important determinants of a heart valve s hemodynamic potential; therefore, changes in design of new aortic prostheses seek to maximize the ratio between EOA and the tissue annulus, to minimize pressure gradients. Significant variability between the manufacturer s provided actual dimensions of the prosthesis lead to question the scientific value of comparisons based only on the industry labeled valve size [11]. However, comparisons performed in relation to the actual dimensions of the native aortic annulus, an independent parameter of valve size, appear to offer more objective data [6]. The hemodynamic results of the current study are depicted according to the AAD, which was measured intra-operatively with a standardized metric sizer, as well as referring to the industry labeled valve size. When labeled valve sizes were compared, 1 and 5 years data clearly showed the hemodynamic advantage of the EPM prosthesis, especially in valve sizes 21, 23, and 25 mm. This can be explained because the size-matched internal diameter of the EPM prosthesis is between 1.5 and 2.0 mm larger than the respective manufacturer-reported internal diameter of the MM valve [14]. Upon implanting valves with larger internal diameters, the current study demonstrated the resultant hemodynamic advantages of the EPM prosthesis: lower mean pressure gradients and larger EOA. These findings correlated closely with previously published studies comparing the Mosaic and Perimount standard valves [6,15,16] and confirm the observed hemodynamic superiority of the new Magna prosthesis compared with Mosaic valves observed in short-term follow-up studies [17,18] and differences being more distinctive under stress conditions [19]. Similarly, the results of our study indicated that the overall hemodynamic performance of the Magna valve was superior to the Mosaic prosthesis even when their performance was related to the inner diameter of the aortic annulus. In patients with an AAD < 22 mm, the implanted valve did not influence the hemodynamic outcome after AVR, although the number of patients in this group (EPM n = 7 and MM n = 9) was too small to permit meaningful analyses. By contrast, in AAD of 22 23 mm and >23 mm, the EPM prosthesis was significantly superior regarding mean pressure gradient, EOA and IEOA at 1 and 5 years. These differences definitively demonstrate the hemodynamic advantage of EPM valves: upon maximizing the ratio EOA/tissue annulus, the third-generation Magna prosthesis achieved a reduction of transvalvular pressure gradients and increased EOAs in comparison to MM valves. In the current study, echocardiographic quantification of IEOA, the only valid parameter that identifies PPM [20], has been employed to define PPM. The hemodynamic consequence of mismatch is to generate high residual transvalvular gradients, which are responsible for an incomplete LVM regression [7], a phenomenon associated with a negative effect on intermediate and long-term survival [4]. The incidence of PPM was statistically different between groups. In the first year, PPM was present in 30.1% of patients with an MM valve and in 9.2% of those with an EPM valve. This difference increased over time and, after 5 years, the differences were more distinctive (EPM 22.9% vs MM 73.9%). At 1 and 5 years, the prevalence of severe mismatch was significantly higher in the MM group. Our data confirm the outcomes reported in other studies [17 19,21] and showed that the use of an EPM valve may contribute to reduce the incidence of PPM, even in patients with a small AAD. When analyzing the effect of PPM on the hemodynamic results, the transprosthetic pressure gradient is expected to decrease with increasing IEOA, a correlation that could be demonstrated in both groups at 1 and 5 years. Nevertheless, at 1 year, the effect of PPM magnitude on LVM regression was less evident. At this time point, our patients showed a significant regression in LVM and LVMI, irrespective of prosthesis type or AAD, the LVMI reduction being similar for both groups. The absence of differences in early LVM regression seen in our series confirm findings of other studies showing equivalent LVM regression after 1 year with Mosaic and pericardial Edwards Perimount valves [15,16]. Nevertheless, it is assumed that mass regression is a continuing process and further reductions in LVM may occur up to 5 years postoperatively [22]. Accordingly, a longer follow-up of our patients has been necessary to determine whether the difference between these prostheses increases over years. Our study demonstrated that small differences at 1 year increased over time and became statistically significant after 5 years. Between the first and fifth year postoperatively, LVMI remained relatively stable in patients with MM valves, while it decreased in patients with EPM prostheses to a significant degree (EPM 47.4 35.1 vs 4.3 36.1; p < 0.0001) (Fig. 2). The reduction of LVMIs observed in patients with Magna prostheses are of special interest, with possible clinical implications, which could only be elucidated with future studies designed and powered to detect differences in clinical event rates. A large amount of literature is available on the effect of AVR on LV hypertrophy regression. However, there are very few studies directly addressing this issue in relation to PPM. In a study including 1103 patients with a porcine bioprosthetic valve, Del Rizzo and co-workers [13] found a strong and independent relationship between IEOA and the extent of LVM regression following AVR. There was a mean decrease in LVM of 23% in patients with an IEOA > 0.8 cm 2 m 2 compared with only 4.5% in those with an IEOA 0.8 cm 2 m 2 ( p = 0.0001). In contrast to these results, Hanayama and co-workers [23] found no significant relationship between PPM and regression of LV hypertrophy in a retrospective study. The major finding of our study was that IEOA (i.e., PPM) was associated with lesser regression of LVM after AVR. This finding was consistent with the pressure gradient IEOA relation, whereby the pressure gradient and, thus, the LV workload increase markedly when the IEOA falls below 0.8 0.9 cm 2 m 2 [20,24]. In the current study, absolute LVMI regression was correlated with the 5-year IEOA, to the mean transprosthetic gradient at 5 years and to the presence of PPM at 5 years. Our study demonstrated that patients with

M.J. Dalmau et al. / European Journal of Cardio-thoracic Surgery 39 (2011) 844 852 851 MM prostheses showed over time a progressive rise of transprosthetic gradients (peak gradient + 4.8 mmhg, p = 0.001), while decreasing IEOAs ( 0.19 cm 2 m 2 ; p < 0.0001). Consequently, an increase in prevalence of PPM was observed, the presence of mismatch being greater than 70% at 5 years. These factors may have been responsible for the poorer extent of LVMI regression seen in this group of patients. Although the reduction in LVM was evident in our patients, the average postoperative LVMI for the entire series remained greater than normal in 42.8% of the patients, and 60.5% MM versus 27.1% EPM patients had LV hypertrophy 5 years after surgery ( p = 0.002). Furthermore, when multivariate modeling was used, the MM valve and the IEOA at 5 years have been identified as independent predictors of residual LV hypertrophy. The reasons for incomplete hypertrophy regression are mainly due to residual aortic gradients and PPM; however, other hemodynamic and non-hemodynamic factors such suboptimal hypertension treatment, physical activity, genotype (angiotensin phenotype expression), and the environment can also affect the degree of mass regression [25]. Previous studies have reported that PPM after AVR is associated with inferior hemodynamics, incomplete LVM regression, more cardiac events and higher mortality rates, all these factors affecting negatively intermediate and longterm survival [1,2,4,5]. The results of the present study suggest that the persistence of LV hypertrophy associated with PPM may be one of the factors contributing to worse clinical outcomes. Accordingly, our study showed that there was a significant difference in survival between groups (79% MM vs 94% EPM) (Fig. 3), and simple linear regression analysis demonstrated a correlation between the LVMI regression at 5 years and time survival. Although, the overall mortality was acceptable when considering patient age, during the follow-up period of 5 years, 20.4% of the patients after MM and 5.4% of the patients after EPM died of different, mostly non-valve-related, causes. Nonetheless, in the majority of patients surviving follow-up, a good hemodynamic function was documented on echocardiographic measurements and both the porcine and pericardial aortic valve types provided good clinical outcomes with acceptable survival at medium-term follow-up. Additional advantages of the Magna prostheses, if any, will only be determined through long-term follow-up to assess late patient outcome and valvular durability. The study has a number of limitations. First, although our original study was randomized in nature, patient numbers were too small to enable any definitive conclusions regarding group-related differences in midterm mortality or clinical outcome. The study was primarily concerned with hemodynamic function and was not powered to detect small differences in clinical event rates. Second, operations were performed in a group practice with multiple surgeons; although all of them used a similar surgical implantation technique, we cannot exclude the fact that small differences in sizing tendencies exist, leading to this issue becoming a possible confounding factor. Third, although 5-year echocardiograms were performed on 87 of 94 eligible patients (92%), seven surviving subjects could not have echocardiography at 5 years due to co-morbidities. Fourth, this study reports information up to only 5 years, and it is possible that late event rates or durability may differ between groups. Finally, the regression of LVM is indeed a complex phenomenon that is influenced by several patient-related and prosthesis-related factors. Furthermore, non-hemodynamic factors may also be involved in the process of LVM regression. These factors were not measured in this study. In conclusion, both porcine and pericardial aortic valves were found to be suitable options for AVR. Our study clearly demonstrates a favorable hemodynamic function of the bovine pericardial Edwards Perimount Magna compared with the porcine Medtronic Mosaic aortic valve prosthesis up to 5 years after implantation, thus achieving lower gradients and larger IEOA. Although short-term follow-up did not show any differences in LVM regression between both prostheses, with longer-term follow-up, Magna valves were found to hemodynamically outperform the Mosaic valves; and such improvements positively affected LV hypertrophy regression. In view of the above findings, we believe that patients clearly benefit from the implantation of the EPM prosthesis, thus resulting in a significantly superior hemodynamic performance, a minimized risk of PPM, and a greater regression of LV mass. References [1] Pibarot P, Dumesnil JG. Hemodynamic and clinical impact of prosthesis patient mismatch in the aortic valve position and its prevention. J Am Coll Cardiol 2000;36:1131 41. [2] Rao V, Jamieson WR, Ivanov J, Armstrong S, David TE. Prosthesis patient mismatch affects survival after aortic valve replacement. Circulation 2000;102(19 Suppl. 13):III-5 9. [3] Pibarot P, Dumesnil JG, Lemieux M, Cartier P, Metras J, Durand LG. Impact of prosthesis patient mismatch on hemodynamic and symptomatic status, morbidity and mortality after aortic valve replacement with a bioprosthetic heart valve. J Heart Valve Dis 1998;7:211 8. [4] Blais C, Dumesnil JG, Baillot R, Simard S, Doyle D, Pibarot P. Impact of valve prosthesis patient mismatch on short-term mortality after aortic valve replacement. Circulation 2003;108:983 8. [5] Fuster RG, Montero Argudo JA, Albarova OG, Sos FH, Lopez SC, Codoner MB, Buendia Minano JA, Albarran IR. Patient prosthesis mismatch in aortic valve replacement: really tolerable? Eur J Cardiothorac Surg 2005;27:441 9. [6] Seitelberger R, Bialy J, Gottardi R, Seebacher G, Moidl R, Mittelbock M, Simon P, Wolner E. Relation between size of prosthesis and valve gradient: comparison of two aortic bioprosthesis. Eur J Cardiothorac Surg 2004;25:358 63. [7] Tasca G, Brunelli F, Cirillo M, DallaTomba M, Mhagna Z, Troise G, Quaini E. Impact of valve prosthesis patient mismatch on left ventricular mass regression following aortic valve replacement. Ann Thorac Surg 2005;79:505 10. [8] Riess FC, Cramer E, Hansen L, Schiffelers S, Wahl G, Wallrath J, Winkel S, Kremer P. Clinical results of the Medtronic Mosaic porcine bioprosthesis up to 13 years. Eur J Cardiothorac Surg 2010;37(1):145 53. [9] Dalmau MJ, Maríagonzález-Santos J, López-Rodríguez J, Bueno M, Arribas A. The Carpentier Edwards Perimount Magna aortic xenograft: a new design with an improved hemodynamic performance. Interact Cardiovasc Thorac Surg 2006;5(3):263 7. [10] Banbury MK, Cosgrove DM, Thomas JD, Blackstone EH, Rajeswaran J, Okies E, Frater RM. Hemodynamic stability during 17 years of the Carpentier Edwards aortic pericardial bioprosthesis. Ann Thorac Surg 2002;73:1460 5. [11] Christakis GT, Buth KJ, Goldman BS, Fremes SE, Rao V, Cohen G, Borger MA, Weisel RD. Inaccurate and misleading valve sizing: a proposed standard for valve size nomenclature. Ann Thorac Surg 1998;66(4):1198 203. [12] Lang RM, Bierig M, Devereux RB, Flachskampf FA, Foster E, Pellikka PA, Picard MH, Roman MJ, Seward J, Shanewise JS, Solomon SD, Spencer KT, Sutton MS, Stewart WJ. Recommendations for chamber quantification: a report from the American Society of Echocardiography s Guidelines and

852 M.J. Dalmau et al. / European Journal of Cardio-thoracic Surgery 39 (2011) 844 852 Standards Committee and the Chamber Quantification Writing Group, developed in conjunction with the European Association of Echocardiography, a branch of the European Society of Cardiology. J Am Soc Echocardiogr 2005;18:1440 63. [13] Del Rizzo DF, Abdoh A, Cartier P, Doty DB, Westaby S. Factors affecting left ventricular mass regression after aortic valve replacement with stentless valves. Semin Thorac Cardiovasc Surg 1999;11:114 20. [14] Jamieson WER, Janusz MT, MacNab J, Henderson C. Hemodynamic comparison of second- and third-generation stented bioprostheses in aortic valve replacement. Ann Thorac Surg 2001;71:S282 4. [15] Suri RM, Zehr KJ, Sundt 3rd TM, Dearani JA, Daly RC, Oh JK, Schaff HV. Left ventricular mass regression after porcine versus bovine aortic valve replacement: a randomized comparison. Ann Thorac Surg 2009;88(4):1232 7. [16] Chambers JB, Rajani R, Parkin D, Rimington HM, Blauth CI, Venn GE, Young CP, Roxburgh JC. Bovine pericardial versus porcine stented replacement aortic valves: early results of a randomized comparison of the Perimount and the Mosaic valves. J Thorac Cardiovasc Surg 2008;136(5):1142 8. [17] Ruzicka DJ, Hettich I, Hutter A, Bleiziffer S, Badiu CC, Bauernschmitt R, Lange R, Eichinger WB. The complete supraannular concept: in vivo hemodynamics of bovine and porcine aortic bioprostheses. Circulation 2009;120(September (11 Suppl.)):S139 45. [18] Dalmau MJ, María González-Santos J, López-Rodríguez J, Bueno M, Arribas A, Nieto F. One year hemodynamic performance of the Perimount Magna pericardial xenograft and the Medtronic Mosaic bioprosthesis in the aortic position: a prospective randomized study. Interact Cardiovasc Thorac Surg 2007;6(3):345 9. [19] Wagner IM, Eichinger WB, Bleiziffer S, Botzenhardt F, Gebauer I, Guenzinger R, Bauernschmitt R, Lange R. Influence of completely supraannular placement of bioprostheses on exercise hemodynamics in patients with a small aortic annulus. J Thorac Cardiovasc Surg 2007;133(5):1234 41. [20] Pibarot P, Dumesnil JG, Cartier PC, Métras J, Lemieux MD. Patient prosthesis mismatch can be predicted at the time of operation. Ann Thorac Surg 2001;71:S265 8. [21] Flameng W, Meuris B, Herijgers P, Herregods MC. Prosthesis patient mismatch is not clinically relevant in aortic valve replacement using the Carpertier Edwards Perimount valve. Ann Thorac Surg 2006;82:530 6. [22] Krayenbuehl H, Hess OM, Monrad S, Schneider J, Mall G, Turina M. Left ventricular myocardial structure in aortic valve disease before, intermediate, and later after aortic valve replacement. Circulation 1989;79:744 55. [23] Hanayama N, Christakis GT, Mallidi HR, Joyner CD, Fremes SE, Morgan CD, Mitoff PR, Goldman BS. Patient prosthesis mismatch is rare after aortic valve replacement: valve size may be irrelevant. Ann Thorac Surg 2002;73:1822 9. [24] Dumesnil JG, Yoganathan AP. Valve prosthesis hemodynamics and the problem of high transprosthetic pressure gradients. Eur J Cardiothorac Surg 1992;6:S34 8. [25] Dellgren G, Eriksson MJ, Blange I, Brodin LA, Radegran K, Sylven C. Aongiotensin-converting enzyme gene polymorphism influences degree of left ventricular hypertrophy and its regression in patient undergoing operation for aortic stenosis. Am J Cardiol 1999;84:909 13. Appendix A. Conference discussion Dr R. Dion (Genk, Belgium): This is a very detailed analysis from the Salamanca group of the function of two bioprostheses in the aortic position. The authors present overwhelming evidence of the superior hemodynamic characteristics of the Edwards Perimount Magna. This leads to significantly greater left ventricular mass regression, and even survival, at five years. However, although the multivariate analysis identified the indexed orifice area at one and five years as independent predictors of death, the causes of death in this report are not all cardiac-related and therefore the statement must be taken with caution. I also regret the absence of stress tests in the comparison of the prostheses. It might be that, in view of the presence of a muscular band in one of the leaflets of the Mosaic prosthesis, the gradient during the stress test would rise proportionately less than that of the pericardial prosthesis, because the increased flow would force this leaflet open. In the Mosaic group, the fact that the PPM, the mean gradient, and the indexed effective orifice area are worst in the 22 23 aortic annulus diameter, is certainly a matter of concern. It is not only in the small diameters but also in the middle cohort of patients. Even in the greater than 23 aortic annulus diameter group, the Mosaic yields a PPM at five years in 65% of the patients. However, recently Jamieson has questioned the influence of a moderate PPM on the postoperative evolution and underlined that only severe PPM, <0.65 cm 2 m 2, is a problem. This leads to my first question. Why did the authors choose not to follow only what happens in the patient with a severe PPM? In the manuscript the authors only mention in the discussion that severe PPM was present at one year in 7.5% of the Mosaic patients versus 1.8% in the Perimount patients, and at five years, 32.5% versus 4%. My second question would be, how do the authors explain the less evident effect of PPM magnitude on the left ventricular mass regression at one year compared to five years? Dr Dalmau: In the manuscript we described the incidence of severe PPM in our patients. Unfortunately an analysis of the effect of severe PPM on clinical or hemodynamic outcomes has not been performed. The second question, in the Magna group the prevalence of mismatch increased over time, but we observed a greater increase in the Mosaic group. This is explained because the achieved indexed effective orifice area in the Mosaic group decreased significantly over time. As PPM is reflected, or is defined by the indexed effective orifice area, when orifice areas decreased over time, the prevalence of mismatching increased. Dr Dion: So you believe you explain the difference in decrease of left ventricular mass by the fact that the indexed orifice area is constantly decreasing with time? Dr Dalmau: Yes. Dr F.C. Riess (Hamburg, Germany): I am very astonished, because our results with long-term follow-up are in contrast to your findings. We had a chance to take part in the FDA trial, and for the 300 cases we operated in our center, we now have 15 years follow-up available. We found that the gradients are higher compared to other valves described in the literature, which is in contrast to your results. By the way, we looked at each patient each year with echocardiography, and we found very stable gradients, a very slight increase, and a very small reduction of orifice area. So this is in contrast. My question to you is, how is the measurement of the aortic root performed by your surgeons? Do you use the original sizers or do you use metal devices? Dr Dalmau: Before randomization, the aortic valve was excised, and the aortic annulus diameter was assessed using standardized metric sizers. We routinely don t perform any oversizing and all valves were implanted in the supra-annular position. Surgeons used the same surgical implantation technique. Dr Riess: My second question is, we found that for porcine valves, they have very good closure compared with the pericardial valves. So my question to you is, did you investigate with echocardiography and do you have some details about how high the degree of regurgitation was after five years? Dr Dalmau: Can you repeat the question. Dr Riess: Concerning regurgitation: pericardial valves always have a small amount of regurgitation because their closure is not so rapid as compared with porcine valves. Did you look at the degree of regurgitation? How many patients had regurgitation at the follow-up? Dr Dalmau: All surviving patients were echocardiographically followed up at five years, and we found no difference with respect to aortic valve regurgitation between both prosthesis types. Dr P. Myken (Gothenburg, Sweden): I agree with Professor Dion that there might be other reasons that the left ventricular mass decrease differs after five years. You have just 87 patients to follow-up at five years, and we don t even know how many in each group. Did you look at hypertension, which is more relevant than the valves? It might even be that the valves are not that important. Dr Dalmau: Left ventricular mass regression is a complex phenomenon in which several patient-related and prosthesis-related factors are involved. Also, non-hemodynamic factors could affect LV mass regression. In our study, the most important factor was the existence of residual transvalvular gradient and the presence of patient prosthesis mismatch. However, other factors, such as hypertension, physical activity and genetic factors, were not measured in this study, but they can also affect left ventricular mass regression. I agree with you.