Long-Term Survival After Bovine Pericardial Versus Porcine Stented Bioprosthetic Aortic Valve Replacement: Does Valve Choice Matter?

Transcription

1 Long-Term Survival After Bovine Pericardial Versus Stented Bioprosthetic Aortic Valve Replacement: Does Valve Choice Matter? Asvin M. Ganapathi, MD, Brian R. Englum, MD, Jeffrey E. Keenan, MD, Matthew A. Schechter, MD, Hanghang Wang, MD, Peter K. Smith, MD, Donald D. Glower, MD, and G. Chad Hughes, MD Division of Cardiovascular and Thoracic Surgery, Duke University Medical Center, Durham, North Carolina Background. Bioprosthetic options are increasingly used for surgical aortic valve replacement (AVR). However, current decision making regarding bioprosthetic valve choice is not data-driven given incomplete information on comparative long-term outcomes after stented bovine pericardial (BoAVR) vs porcine (PoAVR) AVR. As such, this study sought to examine the effect of bioprosthetic valve choice on long-term survival and reoperation after AVR. Methods. A retrospective analysis of all stented bioprosthetic AVRs, with or without coronary artery bypass grafting procedures, at a single tertiary referral institution from 1980 to 2013 was conducted using a prospectively maintained database. Procedures were classified as BoAVR or PoAVR. The effect of valve type on long-term survival was assessed with Kaplan-Meier analysis and a Cox proportional hazard model. Subanalyses stratifying by valve size and patient age at the time of AVR assessing mortality and reoperation were also conducted. Results. We identified 2,010 stented bioprosthetic AVR patients, comprising 1,411 BoAVRs (70.2%) and 599 PoAVRs (29.8%), 868 (43.2%) of whom underwent concomitant coronary artery bypass grafting. Kaplan- Meier analysis (p [ 0.12) and Cox models (hazard ratio BoAVR vs PoAVR, 1.08; p [ 0.40) demonstrated no difference in long-term survival or need for reoperation after BoAVR vs PoAVR. Subanalyses of valve size and patient age also failed to show any association between valve choice and long-term survival or need for reoperation. Conclusions. For patients undergoing AVR with a stented bioprosthetic valve, with or without coronary artery bypass grafting, the choice of a porcine vs bovine pericardial bioprosthesis does not appear to affect longterm survival or need for reoperation, regardless of valve size or patient age. As such, stented bioprosthetic valves would appear to be fungible, and therefore, valve choice should be driven by local market factors similar to other commodities. (Ann Thorac Surg 2015;100:550 9) Ó 2015 by The Society of Thoracic Surgeons Bioprosthetic valves, which allow patients to avoid long-term anticoagulation, have come to dominate aortic valve replacement (AVR) operations and now account for almost 80% of valves implanted [1]. Most of these bioprosthetic aortic valves are stented bovine pericardial (BoAVR) or porcine (PoAVR) valves, with recent data from England indicating that nearly two-thirds of all AVRs are with bovine pericardial bioprostheses [2]. The choice between these options, however, is currently guided by poorly quantitated factors, such as surgeon preference or other perceived relative advantages, many of which are driven by valve company marketing departments. Further, current American College of Cardiology/American Heart Association consensus guidelines Accepted for publication Feb 12, Presented at the Sixty-first Annual Meeting of the Southern Thoracic Surgical Association, Tucson, AZ, Nov 5 8, Address correspondence to Dr Hughes, Box 3051 DUMC, Durham, NC 27710; gchad.hughes@duke.edu. do not inform regarding use of bovine pericardial vs porcine xenograft stented bioprostheses [3]. Finally, although numerous studies, including some randomized trials, have examined the differences between BoAVR and PoAVR, many of these studies focused on hemodynamic measurements, such as pressure gradients or effective orifice area at 1 year, or on midterm and long-term survival [4 7]. However, these prior studies were limited by numerous issues, including small patient numbers, lack of data with more modern valves, and inclusion of only selected bovine pericardial and porcine bioprostheses. Consequently, the ability of surgeons to objectively choose between the numerous stented bioprosthetic valve options available for AVR is limited due to a lack of comparative effectiveness data, particularly for long-term survival, which is ultimately the most important clinical end point that should be used to guide valve choice. Given this important knowledge gap, the current study examines the effect of stented bioprosthetic valve choice, amongst multiple valves, on long-term survival and need for reoperation after surgical AVR. Ó 2015 by The Society of Thoracic Surgeons /$36.00 Published by Elsevier

2 Ann Thorac Surg GANAPATHI ET AL 2015;100:550 9 LONG-TERM SURVIVAL IN BOVINE VS PORCINE AVR 551 Patients and Methods This study was reviewed and approved by the Duke University Institutional Review Board. The need for individual patient consent was waived. Patients and Data Source We identified all patients undergoing elective and nonelective AVR, with or without concomitant coronary artery bypass grafting (CABG), at a single tertiary referral center between 1980 and The analysis excluded patients undergoing a concomitant procedure other than CABG such as multiple valvular operations, aortic operations, or ventricular septal defect closure. Patients who had concurrent atrial septal defect closure or Cox-Maze procedures were included for analysis. The Duke Valve Surgery Database, a prospectively maintained clinical registry of all patients undergoing valvular heart surgery at Duke University Medical Center (Durham, NC), was used to abstract all preoperative, intraoperative, and postoperative variables. We identified 2,010 patients who met the inclusion criteria for analysis. Patients were then classified as having undergone BoAVR or PoAVR (Table 1). The Society of Thoracic Surgeons definitions were used to define patient comorbidities and postoperative outcomes [8]. Follow-up data and long-term mortality status was determined through a combination of record review, national death indices, and Duke Enterprise Data Unified Content Explorer-guided query [9]. Stented Bioprosthetic Valves Analyzed All AVR procedures were conducted at the same institution by 26 surgeons during the study period. Stented bovine pericardial valves included in the analysis were Carpentier-Edwards Perimount models 2700, 3000, and 3300 (Edwards Lifesciences, Irvine, CA), Sorin Mitroflow (Sorin Group Inc, Arvada, CO), and St. Jude Trifecta (St. Jude Medical Inc, St. Paul, MN). Stented porcine xenograft valves included were St. Jude Biocor (St. Jude Medical Inc), Carpentier-Edwards porcine valve (Edwards Life Science), and Medtronic Hancock and Mosaic (Medtronic Inc, Minneapolis, MN). The analysis excluded Ionescu-Shiley bovine pericardial valves (n ¼ 52) due to known premature failure of these firstgeneration valves, which have not been manufactured since 1987 [10]. Outcomes The primary end points of the study were long-term survival and need for reoperation secondary to prosthetic valve dysfunction. Statistical Methods Preoperative, intraoperative, and postoperative variables were compared using the Fisher exact test for categorical variables or the Kruskal-Wallis test for continuous variables. Long-term survival and freedom from prosthetic valve dysfunction requiring reoperation were examined using Kaplan-Meier methods with the log-rank test as Table 1. Stented Bioprosthetic Aortic Valves Included in Study Valves Total (No.) Isolated AVR (No.) AVRþCABG (No.) 1,411 Carpentier-Edwards 1, Perimount a Sorin Mitroflow b St. Jude Trifecta c St. Jude Biocor c Carpentier-Edwards a Medtronic Hancock d Medtronic Mosaic d a Edwards Lifesciences, Irvine, California. b Sorin Group Inc, Arvada, Colorado. c St. Jude Medical Inc, St. Paul, Minnesota. d Medtronic, Minneapolis, Minnesota. AVR ¼ aortic valve replacement; CABG ¼ coronary artery bypass grafting. well as a Cox proportional hazard model to control for covariates other than valve choice. Also conducted were subanalyses with Kaplan-Meier analysis and Cox proportional hazard models, stratifying by small (19 and 21 mm) medium (23 and 25 mm) and large (27 and 29 mm) valve size, as well as younger (18 to 55 years) or older (>55 years) age at the time of AVR, assessing mortality and reoperation. The age cutoff of 55 years was chosen based on clinical relevance and sensitivity analysis to determine the cutoff where an appropriate number of patients for comparisons would exist. For all Cox proportional hazard models, the following variables were adjusted for: age, sex, race, body mass index, hypertension, diabetes, cerebrovascular disease, lung disease, chronic kidney disease, peripheral vascular disease, New York Heart Association Functional Classification, atrial fibrillation, preoperative intraaortic balloon pump, ejection fraction, operative indication, cardiopulmonary Fig 1. Number of porcine (light bars) vs bovine pericardial (dark bars) aortic valve replacements by year.

3 552 GANAPATHI ET AL Ann Thorac Surg LONG-TERM SURVIVAL IN BOVINE VS PORCINE AVR 2015;100:550 9 Table 2. Patient Preoperative Characteristics Variables a All Patients (N ¼ 2,010) Bovine Pericardial (n ¼ 1,411) (n ¼ 599) p Value Age, years 73.1 (66.8, 78.7) 73.1 (66.5, 78.7) 73.3 (67.4, 78.8) Male sex 1,267 (63) 888 (62.9) 379 (63.3) Race Caucasian 1,706 (84.9) 1,179 (83.6) 527 (88) African-American 246 (12.2) 183 (13) 63 (10.5) Other 58 (2.9) 49 (3.5) 9 (1.5) Body mass index, kg/m (24.4, 31.8) 27.7 (24.4, 31.8) 28 (24.7, 31.7) 0.51 Hypertension 1,580 (79) 1,129 (80.2) 451 (76.1) Smoker 581 (29.1) 413 (29.4) 168 (28.4) Diabetes 531 (26.5) 381 (27.1) 150 (25.3) Cerebrovascular disease 208 (10.4) 148 (10.5) 60 (10.1) Lung disease b 236 (11.8) 162 (11.5) 74 (12.5) Chronic kidney disease c 317 (15.8) 241 (17.1) 76 (12.8) Peripheral vascular disease 191 (9.5) 137 (9.7) 54 (9.1) New York Heart Association class I 132 (6.6) 99 (7) 33 (5.5) II 741 (36.9) 520 (36.9) 221 (37.1) III 673 (33.5) 489 (34.7) 184 (30.9) IV 460 (22.9) 302 (21.4) 158 (26.5) Pre-op atrial fibrillation 317 (15.9) 213 (15.2) 104 (17.5) Endocarditis in last 6 months 79 (3.9) 53 (3.8) 26 (4.4) Pre-op intraaortic balloon pump 11 (0.6) 8 (0.6) 3 (0.5) Ejection fraction 0.50 (0.45, 0.57) 0.50 (0.45, 0.575) 0.50 (0.41, 0.55) Indication Aortic stenosis 873 (43.6) 619 (44) 254 (42.7) Aortic regurgitation 285 (14.2) 215 (15.3) 70 (11.8) Aortic stenosis and regurgitation 845 (42.2) 574 (40.8) 271 (45.5) Redo sternotomy 182 (9.1) 129 (9.2) 53 (8.9) a Values are expressed as median (interquartile range) or number (percent). baseline creatinine 2.0 mg/dl or being on dialysis, or both. b Defined as mild, moderate, or severe lung disease. c Defined as bypass and cross-clamp times, concurrent CABG, and number of diseased coronary vessels. An a priori decision was made to set the significance level at a ¼ 0.05 for all analyses. Statistical analysis was done using JMP Pro software (SAS Institute, Cary, NC) and R software (The R Foundation for Statistical Computing, Vienna, Austria). Results We identified 2,010 stented bioprosthetic AVRs from 1980 to 2013, of which 1,411 were BoAVRs and 599 were PoAVRs. The most common type of BoAVR implant was the Carpentier-Edwards Perimount valve (models 2700, 3000, and 3300), and the Carpentier-Edwards porcine bioprosthesis was the most commonly implanted porcine valve (Table 1). Use of bioprosthetic AVRs increased over the study period, from a minimum of 11% of all implanted AVRs in 1988 to 86% in Although the early bioprosthetic experience was predominated by PoAVRs, BoAVRs have predominated since the late 1990s, being implanted more than twice as often as PoAVRs at our institution (Fig 1). A comparison of patient preoperative characteristics (Table 2) revealed that the BoAVR and PoAVR groups were similar in age, sex, and comorbidities. Slight differences existed between groups for race and the preoperative incidence of hypertension, chronic kidney disease, New York Heart Association class IV heart failure, and ejection fraction (Table 2). With regards to operative characteristics (Table 3), the groups did not differ significantly in cardiopulmonary bypass or aortic cross-clamp times, in the incidence of concomitant CABG, or in the number of diseased coronary vessels. Distribution of valve sizes implanted was generally similar as well, with 23 mm being the most common valve implant size in both the BoAVR and PoAVR groups (Table 3). Mean length of follow-up was similar between the BoAVR and PoAVR groups (Table 4). Unadjusted survival analysis (Fig 2A) revealed no significant differences between the groups, with a 10-year survival of 47.0% in the BoAVR group and 43.3% in the PoAVR group (Table 4). Similarly, need for redo AVR (Fig 2B) did not differ by valve type. Unadjusted survival analysis stratified by small, medium, or large valve size implanted likewise

4 Ann Thorac Surg GANAPATHI ET AL 2015;100:550 9 LONG-TERM SURVIVAL IN BOVINE VS PORCINE AVR 553 Table 3. Operative Characteristics Variables a All Patients (N ¼ 2,010) Bovine Pericardial (n ¼ 1,411) (n ¼ 599) p Value Concomitant CABG 868 (43.2) 610 (43.2) 258 (43.1) Isolated AVR CPB time, min 136 (112, 159) 137 (114, 158) 135 (108,161) Aortic cross-clamp time, min 86 (70, 102) 86 (72, 102) 84 (63, 100) AVR þ CABG CPB time, min 178 (149, 207) 177 (148, 207) 180 (151, 208) Aortic cross-clamp time, min 112 (92, 134) 113 (94, 135) 109 (88, 131) Diseased coronary vessels, No (47.6) 671 (48) 277 (46.6) (16.4) 218 (15.6) 109 (18.3) (15.1) 213 (15.2) 88 (14.8) (20.9) 296 (21.2) 121 (20.3) Valve size, mm (9.7) 151 (10.7) 43 (7.2) (30) 438 (31) 164 (27.4) (34.7) 466 (33) 231 (38.6) (19.6) 268 (19) 126 (21) (5.1) 73 (5.2) 30 (5) (1) 15 (1.1) 5 (0.8) a Values are expressed as median (interquartile range) or as number (percent). AVR ¼ aortic valve replacement; CABG ¼ coronary artery bypass grafting; CPB ¼ cardiopulmonary bypass. demonstrated no difference between groups (Fig 3). Furthermore, no difference in unadjusted survival was noted between BoAVR and PoAVR when stratified by patient age at time of AVR (Fig 4). The risk of death, as measured by Cox proportional hazard modeling, revealed a hazard ratio of 1.08 for patients receiving a porcine valve; however, this was nonsignificant (p ¼ 0.40). Because of a perceived difference at approximately 10 years between BoAVR and PoAVR for death, all survival analyses were repeated by only examining survival to 10 years, and no significant differences were noted between the groups in the overall analysis and subanalyses (data not shown). The risk of requiring a repeat AVR was 0.96 for PoAVR compared with BoAVR, but this was also nonsignificant (p ¼ 0.91). Similarly, no significant hazard was associated with porcine valves compared with bovine valves in all subanalyses for survival and need for redo AVR (Table 5). Table 4. Long-Term Outcomes Variables a All Patients (N ¼ 2,010) Bovine Pericardial (n ¼ 1,411) (n ¼ 599) p Value Length of follow-up, years Redo AVR 38 (1.9) 28 (2) 10 (1.7) Survival b 1 year 90.0 ( ) 87.8 ( ) 5 years 72.2 ( ) 68.4 ( ) 10 years 47.0 ( ) 43.3 ( ) Freedom from redo AVR b 1 year 99.4 ( ) 100 ( ) 5 years 98.4 ( ) 98.9 ( ) 10 years 96.1 ( ) 96.5 ( ) a Values expressed as mean standard deviation or number (percent), except as noted. estimated by Kaplan-Meier methods. AVR ¼ aortic valve replacement. b Values are the percentage (95% confidence interval) as

5 554 GANAPATHI ET AL Ann Thorac Surg LONG-TERM SURVIVAL IN BOVINE VS PORCINE AVR 2015;100:550 9 Fig 2. Kaplan-Meier curves for patients with bovine pericardial (solid line) and porcine (dashed line) valves show (A) survival and (B) need for aortic valve reoperation. A Survival (%) Log rank test: p= Years B Survival (%) Log rank test: p= Years Comment AVR remains one of the most common adult cardiac surgical procedures, and in recent years, a marked increase has occurred in the use of bioprosthetic valves, with recent data indicating bioprosthetic valve use in nearly 80% of all surgical AVR procedures [1]. Although consensus guidelines clearly outline populations in whom a bioprosthetic valve should be used [3], no recommendations exist about the specific choice of stented bovine pericardial vs porcine xenografts. This lack of consensus is likely secondary to conflicting data as well as previous studies comparing only one bovine and porcine option, making valve choice driven by surgeon preference, cost, or other unknown factors. Consequently, the current study addressed this knowledge gap directly and demonstrated no appreciable difference between stented bovine pericardial and porcine bioprostheses in longterm survival or the need for aortic valve reoperation. Further, this apparent equivalency persists even when various subgroups are examined by valve size implanted or patient age at time of AVR. Numerous studies have attempted to determine if there are any differences in long-term performance between stented bovine pericardial and porcine bioprostheses. An early manuscript from Gallo and colleagues [5] compared the Ionescu-Shiley bovine pericardial prosthesis to the Carpentier-Edwards and Medtronic Hancock porcine bioprostheses at 9 years and found an increased rate of primary tissue failure in patients with the Ionescu-Shiley

6 Ann Thorac Surg GANAPATHI ET AL 2015;100:550 9 LONG-TERM SURVIVAL IN BOVINE VS PORCINE AVR 555 A B Log rank test: p=0.22 Log rank test: p=0.129 C Log rank test: p=0.412 Fig 3. Overall survival analysis in patients with bovine pericardial (solid line) and porcine (dashed line) valves by subgroups with a (A) small valve (19 and 21 mm), (B) medium valve (23 and 25 mm), or (C) large valve (27 and 29 mm). bovine bioprosthesis. However, their study was limited because the only bovine pericardial valve examined was the Ionescu-Shiley, which has known issues with bioprosthetic valve dysfunction and is no longer marketed [10]. More recent studies comparing bovine pericardial with porcine bioprostheses have not demonstrated any difference in long-term survival [11 13]. These latter studies were limited primarily by small patient numbers, bovine pericardial and porcine prostheses in multiple valve positions, or by including several institutions with different valve choice patterns [11, 13 15]. One very recent study meriting further discussion is from Hickey and colleagues [2]. This study examined the British experience of bovine pericardial and porcine bioprostheses in a modern cohort of nearly 40,000 patients from 2003 to These investigators found results similar to the current study for overall survival and need for reoperation as well as after subanalysis by patient age and valve size. In addition, it is notable that the current study includes 10-year follow up data for more than 300 patients, with many patients having 15 or more years of follow-up, which were not available in the Hickey study. Although these numbers are comparatively small, they confirm that no change in the survival outcomes between the two types of stented bioprostheses develops beyond 10 years. Moreover, the current study provides a comparison of multiple types of bovine pericardial and porcine prostheses performed in a single institution, which may reduce bias introduced by variations in perioperative care and institutional practices. Although studies have suggested equivalent long-term survival of patients after stented bovine pericardial or porcine AVR, other studies have focused on in vivo hemodynamic results obtained from echocardiography. These studies have demonstrated that in both resting and stress (exercise-induced) conditions, patients with bovine pericardial valves tend to have improved hemodynamics, including decreased transvalvular pressure gradients and, in some cases, an increased effective orifice area [4, 6, 16]. Other studies analyzing the effect of valve choice on changes in left ventricular mass have suggested that a similar degree of mass regression is observed after bovine pericardial and porcine AVR [17]. Furthermore, an examination of bovine pericardial and porcine bioprostheses in animal models have suggested that bovine pericardial bioprostheses have an improved ability to maintain collagen and water content, which are important for the structural integrity of the valve, and a decreased rate of calcification in vitro [18, 19]. These human and animal in vivo and in vitro results appear to suggest an overall advantage for bovine pericardial valves in some hemodynamic and structural outcomes. However, the current study, as well as others [2, 7, 12], would suggest that these surrogate end points are unreliable and not predictive of overall survival or the need for reoperation secondary to prosthetic valve

7 556 GANAPATHI ET AL Ann Thorac Surg LONG-TERM SURVIVAL IN BOVINE VS PORCINE AVR 2015;100:550 9 Fig 4. Overall survival analysis in patients with bovine pericardial (solid line) and porcine (dashed line) valves according patient age (A) 18 to 55 years and (B) age older than 55 years at aortic valve replacement. dysfunction, which are more important goals of surgical AVR. Given these findings, the current data suggest that stented bioprosthetic valves appear to be fungible commodities in the sense that the bovine pericardial and porcine valves yield similar results in long-term survival and the need for reoperation. As such, an economist would argue that the choice of stented bioprosthetic valve for surgical AVR should be based solely on cost considerations. Given the skyrocketing cost of health care worldwide, these findings should be increasingly important as health care systems examine ways to reduce costs and maintain the ability to offer life-prolonging therapies, such as AVR, to as many patients as possible. The current study has several limitations. This was a single-institution, retrospective study and has the potential for bias due to its retrospective nature. In addition, data were collected from patients over 30 or more years, and although procedures were conducted similarly throughout this period, there were likely differences in surgical approach and perioperative care secondary to the evolving nature of cardiac surgery. Furthermore, although the long period of follow-up is an important part of the conclusions derived, we acknowledge that most of the patients underwent AVR after the year 2000 and that only 15% of patients had follow-up greater than 10 years. Another limitation is the low number of patients who had a large valve size implanted or who were younger than 55 years at the time of AVR. As a result, the ability to perform multivariable survival analysis in these populations was limited. The low number of

8 Ann Thorac Surg GANAPATHI ET AL 2015;100:550 9 LONG-TERM SURVIVAL IN BOVINE VS PORCINE AVR 557 Table 5. Cox Proportional Hazard Models Survival Need for Aortic Valve Reoperation Model HR 95% CI p Value HR 95% CI p Value Overall (Ref: bovine pericardial) Valve size (Ref: bovine pericardial) 19 and 21 mm and 25 mm and 29 mm a a a Age (Ref: bovine pericardial) Age years Age >55 years a Analysis not conducted due to small number of patients and events. CI ¼ confidence interval; HR ¼ hazard ratio. events for prosthetic valve dysfunction also limited the number of variables that could be adjusted for in the Cox proportional hazard model for this outcome. Finally, the analysis did not include short-term or longer-term hemodynamic results or analysis of quality of life or heart failure class with either valve type, which are also clinically important outcomes for AVR patients. In conclusion, for patients undergoing AVR with a stented bioprosthetic valve, with or without CABG, the choice of a porcine vs bovine pericardial bioprosthesis does not appear to affect long-term survival or the need for reoperation, regardless of valve size or patient age. As such, stented bioprosthetic valves would appear to be fungible, and therefore, valve choice should be driven by local market factors similar to other commodities. References 1. Brown JM, O Brien SM, Wu C, Sikora JA, Griffith BP, Gammie JS. Isolated aortic valve replacement in North America comprising 108,687 patients in 10 years: changes in risks, valve types, and outcomes in the society of thoracic surgeons national database. J Thorac Cardiovasc Surg 2009;137: Hickey GL, Grant SW, Bridgewater B, et al. A comparison of outcomes between bovine pericardial and porcine valves in patients in England and Wales over 10 years. Eur J Cardiothorac Surg [published online ahead of print Sept 4, 2014] 3. Nishimura RA, Otto CM, Bonow RO, et al AHA/ACC guideline for the management of patients with valvular heart disease: a report of the American College Of Cardiology/ American Heart Association task force on practice guidelines. Circulation 2014;129:e Eichinger WB, Botzenhardt F, Keithahn A, et al. Exercise hemodynamics of bovine versus porcine bioprostheses: a prospective randomized comparison of the Mosaic and Perimount aortic valves. J Thorac Cardiovasc Surg 2005;129: Gallo I, Nistal F, Arbe E, Artinano E. Comparative study of primary tissue failure between porcine (Hancock and Carpentier-Edwards) and bovine pericardial (Ionescu-Shiley) bioprostheses in the aortic position at five- to nine-year follow-up. Am J Cardiol 1988;61: Ruzicka DJ, Hettich I, Hutter A, et al. The complete supraannular concept: in vivo hemodynamics of bovine and porcine aortic bioprostheses. Circulation 2009;120:S Yap KH, Murphy R, Devbhandari M, Venkateswaran R. Aortic valve replacement: is porcine or bovine valve better? Interact Cardiovasc Thorac Surg 2013;16: The Society of Thoracic Surgeons. Adult cardiac surgery database training manual v2.73; 8/12. Available at Manual Update 8 12.pdf. Accessed June 14, Horvath MM, Winfield S, Evans S, Slopek S, Shang H, Ferranti J. The deduce guided query tool: providing simplified access to clinical data for research and quality improvement. J Biomed Inform 2011;44: Walley VM, Keon CA, Khalili M, Moher D, Campagna M, Keon WJ. Ionescu-Shiley valve failure. I: experience with 125 standard-profile explants. Ann Thorac Surg 1992;54: Reichenspurner H, Weinhold C, Nollert G, et al. Comparison of porcine biological valves with pericardial valves a 12- year clinical experience with 1123 bio-prostheses. Thorac Cardiovasc Surg 1995;43: Gao G, Wu Y, Grunkemeier GL, Furnary AP, Starr A. Durability of pericardial versus porcine aortic valves. J Am Coll Cardiol 2004;44: Chaudhry MA, Raco L, Muriithi EW, Bernacca GM, Tolland MM, Wheatley DJ. versus pericardial bioprostheses: eleven-year follow up of a prospective randomized trial. J Heart Valve Dis 2000;9:429 37; discussion Pelletier LC, Carrier M, Leclerc Y, Lepage G, deguise P, Dyrda I. versus pericardial bioprostheses: a comparison of late results in 1,593 patients. Ann Thorac Surg 1989;47: Jamieson WR, Germann E, Aupart MR, Neville PH, Marchand MA, Fradet GJ. 15-year comparison of supraannular porcine and Perimount aortic bioprostheses. Asian Cardiovasc Thorac Ann 2006;14: Chambers JB, Rajani R, Parkin D, et al. 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:

9 558 GANAPATHI ET AL Ann Thorac Surg LONG-TERM SURVIVAL IN BOVINE VS PORCINE AVR 2015;100: Suri RM, Zehr KJ, Sundt TM 3rd, et al. Left ventricular mass regression after porcine versus bovine aortic valve replacement: a randomized comparison. Ann Thorac Surg 2009;88: Liao K, Seifter E, Hoffman D, Yellin EL, Frater RW. Bovine pericardium versus porcine aortic valve: comparison of DISCUSSION DR WILLIAM T. RYAN (Dallas, TX): I would like to thank the Society for the opportunity to begin discussion of this timely and well-conceived and really very well presented paper. Dr Ganapathi, you have created a lot of good fellowship in this room. It is nice to know that those of us who have known with certainty which stented valve was best have at last been proved correct. Now, if you were arguing for the bovine valve, after looking at your paper, I would say that chronic kidney disease was higher in my patients as was multivessel coronary artery disease although the incidence was no different although this really did not reach statistical significance. I would also say that 1-, 5-, and 10-year survivals were better in the bovine group, even though this did not reach statistical significance. If I were arguing for the porcine valve, I would say that my patients were followed up longer with less aortic valve replacement (AVR) reoperations, so survival must be due to patient factors and not the valve. In addition, most of my patients were done early in the study with less newer technology. So these examples, of course, which you just heard, are typical marketing ploys and illustrate the classic cardiac surgeon rationale of when you need a stented valve excuse, any excuse will do. So your paper, I believe as one of many, by removing or correcting for many of the variables, debunks the myth and marketing of stented valves, and we hope for a long time. I have two questions. With the publication of your paper and others recently like it, can we finally treat stented valves, like you say, as a commodity, and ask our hospital administrators to send out request for proposals, or are there other studies that you really feel should be done, particularly with particular valve models and sizes before this is done? DR GANAPATHI: Thank you, Dr Ryan, for being a discussant on this paper and your questions and comments. In terms of treating the valve as a commodity, there has been quite a bit of data now our study, and then there was a recent study published out of the United Kingdom looking at almost 38,000 patients, although their follow-up was shorter than what we have and it only went to 10 years, which suggests that if you look at all these different types of stented bioprostheses, the long-term survival, which probably is the most clinically meaningful outcome, is essentially equivalent. In terms of other studies, if you look at the data regarding valvular gradients, effective orifice area, really that data is only concentrated on 1 year, and that is the longest that they go out. The only other study that I think would be useful going forward is one in which the data would show that preferential hemodynamics for the bovine valves at 1 year continues long term and has some bearing on patient quality of life and heart failure class going forward. DR RYAN: Thank you. The second question is probably one of the most important questions that we all in this room have thought about or should think about as surgeons. If you look at your paper, realizing your survival curves for AVR are virtually identical to the survival curves of 20 years ago, with 40% of the tissue biological properties as prosthetic valves. Artif Organs 1992;16: Mavrilas D, Kapolos J, Koutsoukos PG, Dougenis D. Screening biomaterials with a new in vitro method for potential calcification: porcine aortic valves and bovine pericardium. J Mater Sci Mater Med 2004;15: patients who are 18 to 55 dead at 10 years, are biologic stented valves really the question at all, and are we putting something in that is not actively or accurately treating the disease process? And at that point, I have a question for you. What national randomized trial would you design that might get at the problem that we have as surgeons about a pathetic 10-year survival after AVR? DR GANAPATHI: Thank you for your question. I think if you were designing a national randomized trial for this, ideally, as we have kind of touched on in this paper, it would look at long-term survival. So obviously you would have to look at survival not just to 5 years but 10, 15 years, possibly longer, particularly as patients who receive bioprosthetic valves are becoming younger but also as older patients tend to live longer. The second thing is we would also want to analyze things like I just mentioned before, quality of life data, New York Heart Association class, and, if possible, hemodynamic measurements, because that might help to basically define why these patients are dying towards the 5-, 10-year marks, if you see any differences there. The only problem is it would be difficult to do a large randomized trial with all of these valves. As demonstrated by the United Kingdom paper, which looked at a national registry, one of the ideas that we could do within the United States would be to use some of these large databases to help recapitulate a randomized trial such as The Society of Thoracic Surgeons database and those cooperatives out of Virginia and Michigan, and that might be useful as well to increase the numbers, particularly with long-term data, to see how these valves actually behave over long periods of time, particularly with median follow-up times of 10, 15 years. DR RYAN: Great. Thank you. I know this paper could be discussed probably for another hour, and hopefully I have stimulated some thoughts. DR VINOD H. THOURANI (Atlanta, GA): I will keep this brief. Just to add a little bit to what Will said, to me it makes a difference if you are talking about a 19 valve and you are talking about a 27 valve, and maybe I missed it in the presentation. Did you look at the 19 and 21 pericardial and compare it with Kaplan- Meier curves for a 19 and 21 bovine, because this could have profound differences? If you are looking at all valves together, I am not sure it really matters if you put a 25 pericardial or porcine. I think it may matter in the 19 and 21 range of which I use the pericardial valves. DR GANAPATHI: Thank you for your question. We actually did do that, and for time constraints, I did not show all of that data. But when you look at the survival curves, we stratified by 19 and 21 mm, 23 and 25 mm, and 27 and 29 mm, in three separates analyses. The 19- and 21-mm survival curves were actually similar between the bovine and the porcine. Additionally, we then carried that analysis forward using the Cox proportional hazard model, and, again, there was no

10 Ann Thorac Surg GANAPATHI ET AL 2015;100:550 9 LONG-TERM SURVIVAL IN BOVINE VS PORCINE AVR 559 significant difference in hazard mortality for long-term survival. DR WILLIAM F. NORTHRUP, III (Kennesaw, GA): Do you think it is important in these comparisons between valve types to include expected survival curves? The reason I ask is because there are a number of papers that have shown excess mortality in younger patients. I think Will was kind of alluding to this. Just within the last 2 months we have seen a series from Montreal, over 20 years, with excess mortality in the range of 1% to 2% per year in younger patients with stented bioprostheses; the Cleveland Clinic; Vancouver. I mean, there are a lot of data out there, thousands of younger patients, demonstrating excess mortality with this valve, which is a popular valve. But to me, a paper that does not have a comparison of observed mortality with expected mortality leaves out an important data set that I personally think should be in every valve paper. DR GANAPATHI: Thank you for your question. I do agree. I think expected mortality vs observed mortality would be a nice comparison to have. Unfortunately, I think that, at least for our study, one of the difficulties that we would have with expected mortality is that current risk models for expected mortality are developed with more modern data from The Society of Thoracic Surgeons, it would be difficult to extrapolate that data back to some of our patients, particularly from the 1980s and early 1990s, and say that that mortality data are appropriate to this. That said, I think, to the earlier question of looking at a randomized trial, that is a very interesting question, and it is particularly interesting in the young patients, which you alluded to, where it is going to be difficult to get the numbers. We touched on the limitations, but we only had 150 patients who were under the age of 55, which makes all of these analyses a little difficult, particularly when adjusted for other covariates.