Controversy exists regarding which valve type is best

Treatment of Endocarditis With Valve Replacement: The Question of Tissue Versus Mechanical Prosthesis Marc R. Moon, MD, D. Craig Miller, MD, Kathleen A. Moore, BS, Phillip E. Oyer, MD, PhD, R. Scott Mitchell, MD, Robert C. Robbins, MD, Edward B. Stinson, MD, Norman E. Shumway, MD, and Bruce A. Reitz, MD Department of Cardiovascular and Thoracic Surgery, Stanford University School of Medicine, Stanford, California Background. It remains unknown whether there is any important clinical advantage to the use of either a bioprosthetic or mechanical valve for patients with native or prosthetic valve endocarditis. Methods. Between 1964 and 1995, 306 patients underwent valve replacement for left-sided native (209 patients) or prosthetic (97 patients) valve endocarditis. Mechanical valves were implanted in 65 patients, bioprostheses in 221 patients, and homografts in 20 patients. Results. Operative mortality was 18 2% and was independent of replacement valve type (p > 0.74). Longterm survival was superior for patients with native valve endocarditis (44 5% at 20 years) compared with those with prosthetic valve endocarditis (16 7% at 20 years) (p < 0.003). Survival was independent of valve type (p > 0.27). The long-term freedom from reoperation for patients who received a biologic valve who were younger than 60 years of age was low (51 5% at 10 years, 19 6% at 15 years). For patients older than 60 years, however, freedom from reoperation with a biological valve (84 7% at 15 years) was similar to that for all patients with mechanical valves (74 9% at 15 years) (p > 0.64). Conclusions. Mechanical valves are most suitable for younger patients with native valve endocarditis; however, tissue valves are acceptable for patients greater than 60 years of age with native or prosthetic valve infections and for selected younger patients with prosthetic valve infections because of their limited life expectancy. (Ann Thorac Surg 2001;71:1164 71) 2001 by The Society of Thoracic Surgeons Controversy exists regarding which valve type is best for patients with native (NVE) or prosthetic valve endocarditis (PVE). Studies comparing the use of bioprosthetic and mechanical valves for patients with endocarditis are limited [1 4], and the role that age plays in the selection process has not been critically addressed. The adverse influence of younger age on the durability of pericardial and porcine bioprosthetic valves for noninfectious indications has been reported in a number of recent large series [5 7]; however, the age criteria for implanting bioprosthetic valves may differ in patients with endocarditis, whose life expectancy may be substantially lower. The goals of the current investigation were to determine whether a bioprosthetic or mechanical prosthesis was best for patients with endocarditis who required valve replacement, and to determine the risk factors that portend lower operative survival rate and poor long-term prognosis after the surgical treatment of endocarditis. Accepted for publication Nov 6, 2000. Address reprint requests to Dr Miller, Department of Cardiovascular and Thoracic Surgery, Falk Cardiovascular Research Center, 300 Pasteur Dr, Stanford University School of Medicine, Stanford, CA 94305-5247; e- mail: dcm@leland.stanford.edu. Material and Methods This retrospective review includes 306 consecutive patients who underwent valve replacement for left-sided endocarditis on the full-time faculty service at Stanford University Medical Center from March 1964 through December 1995. There were 221 (72%) men and 85 (28%) women, with a mean age ( 1 SD) of 49 16 years (87 patients were over 60 years of age). There were 209 patients with NVE and 97 patients with PVE. Those with PVE (55 14 years) were older than the subset with NVE (46 16 years) ( p 0.001). All patients were contacted for follow-up by telephone during a 5-month closing interval (March to July 1996). Cumulative long-term follow-up totaled 2,033 patient-years, and was 94% complete. The diagnosis of endocarditis was based on generally accepted clinical criteria, including appropriate combinations of fever, new or altered cardiac murmurs, systemic emboli, positive blood cultures, and echocardiographic findings. Characteristic valvular changes were confirmed both at operation and histopathologically. Endocarditis was defined as active (211 patients) versus healed (95 patients) based on whether a planned, standard course of antibiotic therapy had been completed before operation. Although this arbitrary distinction is not a direct index of the activity of the infectious process per se, it correlates with both the operative risk and pathological findings at 2001 by The Society of Thoracic Surgeons 0003-4975/01/$20.00 Published by Elsevier Science Inc PII S0003-4975(00)02665-5

Ann Thorac Surg MOON ET AL 2001;71:1164 71 ENDOCARDITIS, TISSUE VS MECHANICAL VALVES 1165 Table 1. Type of Valve Implanted for AVR and MVR in Patients With NVE and PVE NVE PVE AVR MVR AVR MVR Mechanical Starr-Edwards caged-ball 31 7 0 1 Bileaflet valves 8 10 10 1 Bioprosthetic Hancock porcine 70 46 38 20 Carpentier-Edwards 27 20 13 11 porcine Homograft 12 0 8 0 AVR aortic valve replacement; MVR mitral valve replacement; NVE native valve endocarditis; PVE prosthetic valve endocarditis. operation [8, 9]. The time from symptoms to treatment averaged 35 60 days, while the time from initiation of medical treatment to operation was 51 69 days. Topical and systemic hypothermia alone was used for myocardial protection in 107 (35%) patients, cold crystalloid cardioplegia was added in 158 (52%) patients, and cold blood cardioplegia in 41 (13%) patients. Cardiopulmonary bypass time was 118 71 minutes, and aortic clamp time averaged 74 41 minutes. Isolated aortic valve replacement (AVR) was performed in 190 (62%) patients, isolated mitral valve replacement (MVR) in 89 (29%) patients, and combined AVR and MVR in 27 (9%) patients. Before 1976, Starr-Edwards mechanical caged-ball valves were implanted most often (61% mechanical, 27% bioprosthetic, 12% homograft); between 1976 and 1986, porcine bioprosthetic valves were implanted almost exclusively (98% bioprosthetic, 2% homograft); from 1987 to 1995, valve choice varied (25% mechanical, 65% bioprosthetic, 10% homograft) according to the judgment of the attending surgeon. The use of specific valve types is summarized in Table 1. Concomitant coronary artery bypass grafting was performed in 28 (9%) patients. Thirty-one patients (14% of those who underwent AVR) required aortic root replacement for extensive annular or aortic wall involvement. Seventeen of these patients underwent homograft root replacement, while 2 patients underwent mechanical and 12 patients underwent bioprosthetic composite valve-graft replacement. The responsible organisms for all patients are listed in Table 2. Streptococcal infections were common in both NVE and PVE; however, while Staphylococcus aureus was more common in NVE, S epidermidis predominated in PVE cases. Postoperative antibiotics were continued for 26 23 days. Data Analysis Operative mortality included any death that occurred during the initial hospitalization or within 30 days of operation for discharged patients. Late complications were defined as residual endocarditis (infections with the same organism within 3 months for bacterial or within 2 years for fungal infections), recurrent endocarditis (an Table 2. Microorganisms Responsible for Native (209 Patients) and Prosthetic (97 Patients) Valve Endocarditis Native Prosthetic Staphylococcus aureus 44 (21%) 9 (9%) S epidermidis 4 (2%) 22 (23%) Streptococcus viridans 76 (36%) 17 (18%) Other streptococcal species 22 (11%) 6 (6%) Enterococcus 10 (5%) 9 (9%) Gram-negative bacillus 5 (2%) 3 (3%) Fungus 6 (3%) 4 (4%) Culture negative 13 (9%) 9 (9%) Other 11 (6%) 14 (14%) Unknown 18 (9%) 4 (4%) entirely new episode of endocarditis after documented curative therapy), or development of a bland periprosthetic leak, not necessarily requiring surgical correction. Long-term survival data included death from all causes. Continuous data are reported as mean 1 SD, and clinically important ratios with 70% confidence limits. Actuarial life-table survival estimates were calculated using the Cutler-Ederer method and compared using the Gehan technique (SPSS, Chicago, IL). Variability of the actuarial estimates was expressed as 1 SEM. Freedom from reoperation estimates were also determined using the actual, or cumulative incidence, method of analysis, which takes into account the competing hazard risk of death when calculating the probability of reoperation [10, 11]. Univariate and multivariate regression analysis (Cox proportional hazard model) was used to determine the preoperative and intraoperative risk factors that were significant, independent predictors of operative mortality and the development of a late complication. Thirty variables (see Table 4) were examined, including intraoperative factors (for example, myocardial protection with topical and systemic hypothermia alone vs cold crystalloid or blood cardioplegia) and complications related to visceral organ dysfunction, cardiac dysfunction, underlying comorbid disease, persistent sepsis, and the extent of the infection. Table 3. Postoperative Complications After Valve Replacement for Endocarditis (Excluding 12 Patients Who Died in the Operating Room) Reexploration for bleeding 24/294 (8%) Low output syndrome 100/294 (34%) Myocardial infarction 10/294 (3%) Renal failure 56/294 (19%) Respiratory failure 38/294 (13%) Tracheostomy 27/294 (9%) Cerebrovascular accident 15/294 (5%) Mediastinitis 13/294 (4%) Atrial arrhythmias 97/294 (33%) Ventricular arrhythmias 68/294 (23%) Heart block 28/294 (10%) Pacemaker insertion 15/294 (5%)

1166 MOON ET AL Ann Thorac Surg ENDOCARDITIS, TISSUE VS MECHANICAL VALVES 2001;71:1164 71 Table 4. Univariate and Multivariate Analysis of Preoperative and Intraoperative Variables Operative Mortality by Univariate Multivariate Variable t p F p Gender Male 43/221 (19%) 1.1 0.30 Female 12/85 (14%) Age (years) 40 11/92 (12%) 17.1 0.001 7.1 0.008 40 49 6/72 (4%) 50 59 7/48 (15%) 60 69 21/65 (32%) 70 10/30 (25%) Valve Aortic 32/190 (17%) 0.1 0.70 Mitral 19/89 (21%) Aortic and 4/27 (15%) mitral Diabetes mellitus Yes 12/32 (38%) 9.6 0.002 NS No 43/274 (16%) Hypertension Yes 14/68 (21%) 0.4 0.52 No 41/238 (17%) Hyperlipidemia Yes 5/17 (29%) 0.9 0.35 No 50/288 (17%) COPD Yes 11/34 (32%) 5.6 0.02 NS No 44/272 (16%) Preop renal insufficiency Yes 32/78 (41%) 42.0 0.001 19.3 0.001 No 23/238 (10%) Preop liver dysfunction Yes 10/33 (30%) 3.9 0.05 5.1 0.03 No 45/273 (17%) Preop CHF Yes 46/247 (19%) 0.1 0.82 No 9/59 (15%) NYHA CHF class I 5/33 (6%) 3.8 0.05 NS II 17/101 (17%) III 15/100 (15%) IV 18/71 (25%) Preop angina Yes 9/34 (27%) 2.0 0.16 No 46/272 (17%) Preop MI (acute) Yes 6/12 (50%) 12.1 0.001 NS No 49/294 (17%) Preop MI (remote) Yes 9/34 (27%) 1.3 0.25 No 46/272 (17%) Preop conduction abnormality Yes 21/73 (29%) 6.8 0.009 NS No 34/233 (15%) Preop embolization Yes 17/95 (18%) 0.0 0.98 No 38/211 (18%) Table 4. Continued Operative Mortality by Univariate Multivariate Variable t p F p Indication for operation CHF 35/218 (16%) 4.2 0.04 NS Emboli 3/26 (12%) Sepsis 6/20 (30%) Poor prognosis 1/5 (20%) Combination 11/42 (26%) Organism Staph 27/79 (34%) 21.3 0.001 14.4 0.001 Non-staph 28/227 (12%) Endocarditis classification I NVE 27/209 (13%) 13.1 0.001 8.0 0.005 PVE 28/97 (29%) Endocarditis classification II Active 45/211 (21%) 6.2 0.02 NS Healed 10/95 (11%) Prior history of endocarditis Yes 5/32 (16%) 0.2 0.69 No 50/273 (18%) Emergency operation Yes 17/68 (25%) 3.4 0.07 No 38/238 (16%) Operative year 1964 1975 11/59 (9%) 0.1 0.78 1976 1986 23/131 (18%) 1987 1995 21/116 (18%) Myocardial protection Hypothermia 16/106 (15%) 0.0 0.68 Crystalloid 32/158 (20%) Blood 6/41 (15%) Valve type implanted Mechanical 12/65 (19%) 0.1 0.74 Bioprosthetic 37/221 (17%) Homograft 6/20 (30%) Intraop culture Positive 18/72 (25%) 3.8 0.05 4.0 0.05 Negative 32/223 (14%) Unknown 5/11 (46%) Vegetations Yes 37/179 (21%) 1.8 0.18 No 18/127 (14%) Leaflet destruction Yes 26/188 (14%) 4.3 0.04 NS No 29/118 (25%) Annular abscess Yes 30/118 (25%) 7.9 0.005 NS No 25/188 (13%) Aortic aneurysm Yes 4/14 (29%) 1.4 0.23 No 51/292 (18%) CHF congestive heart failure; COPD chronic obstructive pulmonary disease; MI myocardial infarction; NVE native valve endocarditis; NYHA New York Heart Association; PVE prosthetic valve endocarditis.

Ann Thorac Surg MOON ET AL 2001;71:1164 71 ENDOCARDITIS, TISSUE VS MECHANICAL VALVES 1167 Results Operative Morbidity and Mortality Postoperative complications are summarized in Table 3. Atrial and ventricular arrhythmias were common, but heart block occurred in only 28 (10%) patients, 15 (54%) of whom required pacemaker insertion before discharge (6% of all operative survivors). The overall operative mortality rate was 18 2%, including 12 patients who died in the operating room. The most common causes of early death were left ventricular pump failure in 42%, multisystem organ failure in 25%, persistent sepsis in 20%, and cerebrovascular accident in 9%. Interestingly, the operative mortality rate did not change between the three time periods; 19 5% from 1964 to 1975, 18 3% from 1976 to 1986, and 18 4% from 1987 to 1995 ( p 0.78). Operative mortality risk as a function of preoperative and intraoperative variables is summarized in Table 4. With univariate analysis, 15 variables were associated with an increased operative risk; however, multivariate regression analysis identified only six factors were independent predictors of a higher probability of early mortality: (1) increased age ( p 0.008); (2) PVE versus NVE ( p 0.005); (3) Staphylococcal infection ( p 0.001); (4) positive intraoperative culture ( p 0.05); (5) renal dysfunction ( p 0.001); and (6) liver dysfunction ( p 0.03). Early survival was independent of the replacement valve type ( p 0.74). Fig 1. Long-term survival for patients with NVE and PVE. The numbers of patients at risk are indicated. Long-Term Results Of the 251 early survivors, there were 98 late deaths, and 18 patients were lost to follow-up. Median follow-up interval was 15.3 years, with the longest survivor alive 29 years after valve replacement for native mitral endocarditis. The most common causes of late death were myocardial failure in 47% (including myocardial infarction, arrhythmias, and sudden death), recurrent endocarditis in 14%, and various noncardiac causes in 39%. Of 135 patients alive at the time of follow-up, 98 (73%) were in New York Heart Association (NYHA) class I, 34 (25%) were in NYHA class II, and 3 (2%) were in NYHA class III. As expected, long-term survival was substantially lower with increasing age ( p 0.001). Survival at 15 years was 62 6% for patients less than 40 years of age, 45 7% for 40 to 49 years, 45 10% for 50 to 59 years, 23 8% for 50 to 59 years, 23 8% for 60 to 69 years, and 0 0% for those greater than 70 years. Long-term survival was significantly superior for patients with NVE (54 4% at 10 years, 44 5% at 20 years) compared with those with PVE (41 6% at 10 years, 16 7% at 20 years) ( p 0.003) (Fig 1). Excluding operative deaths, long-term survival was slightly higher for patients with NVE (63 4% at 10 years, 51 5% at 20 years) compared with those with PVE (51 5% at 10 years, 22 10% at 20 years) ( p 0.10). The estimate of patients alive without complications also tended to be lower, but statistically insignificantly so, for patients with PVE (10 years: 61 7% PVE vs 75 4% NVE; p 0.17). Multivariate regression analysis identified five factors to be independent predictors of developing a late complication: (1) increased age ( p 0.001); (2) PVE versus NVE ( p 0.03); (3) annular abscess ( p 0.02); (4) acute preoperative myocardial infarction ( p 0.03); and (5) emergency operation ( p 0.02). Late complications were not significantly related to the type of replacement valve ( p 0.90). Bioprosthetic Versus Mechanical Valves There was no significant difference in operative mortality rate according to whether a mechanical (12 of 65, 19 5%), bioprosthetic (37 of 221, 17 3%), or homograft (6 of 20, 30 10%) valve was selected ( p 0.74). Overall long-term survival was also similar at both 10 years (50 8% mechanical, 51 4% bioprosthetic, 40 12% homograft) and at 20 years (38 9%, 34 5%, 40 12%) ( p 0.27) (Fig 2). Similarly, survival was nearly identical in the different valve type groups if operative deaths were excluded at 10 years (62 9% mechanical, 61 4% Fig 2. Long-term survival for patients undergoing valve replacement with mechanical, bioprosthetic, or homograft valves. The numbers of patients at risk are indicated.

1168 MOON ET AL Ann Thorac Surg ENDOCARDITIS, TISSUE VS MECHANICAL VALVES 2001;71:1164 71 bioprosthetic, 58 15% homograft; p 0.50) and also after 20 years (46 10%, 41 6%, 58 15%; p 0.5). Recurrent or residual endocarditis occurred in 34 of 251 (or 14%) of operative survivors, and 14 (41%) of them died as a consequence of their ongoing problems with infection. The incidence was similar for patients who had completed a standard medical therapy regimen before undergoing valve replacement (healed: 11 of 95, 12 3%) as it was for those who underwent surgical intervention before completing a full course of antibiotic treatment (active: 25 of 211, 12 2%) ( p 0.90). The linearized rate of recurrent or residual endocarditis during the first 5 years was 2.9 1.0% per patient-year for patients with PVE and 2.1 0.5% for patients with NVE ( p 0.49). After 5 years, the linearized rate of recurrent endocarditis was lower with PVE (2.0 1.0% per patient-year) and NVE (0.9 0.3%), but again were not significantly different between the two groups ( p 0.18). The linearized rate of recurrent or residual endocarditis during the first 5 years was 2.1 1.1% per patient-year in the mechanical valve cohort, 2.3 0.6% among those with bioprostheses, and 3.6 2.5% for the homograft valve recipients ( p 0.88 between groups). After 5 years, the linearized rates were lower (0.5 0.5% per patient-year mechanical, 1.1 0.4% bioprosthetic, and 3.1 2.2% homograft), but differences remained insignificant between groups ( p 0.25). Homograft recipients were excluded from the following analyses due to the small number of patients in each subgroup. For patients less than or equal to 60 years of age, overall long-term survival was similar in those who received a mechanical (61 9% at 10 years, 50 10% at 15 years) or a biologic (58 4% at 10 years, 52 5% at 15 years) valve ( p 0.29) (Fig 3A). For patients greater than 60 years of age, long-term survival tended to be lower in patients who received a mechanical valve (18 12% at 10 years) compared with those who received a biologic valve (31 6% at 10 years, 17 7% at 15 years) valves ( p 0.08) (Fig 3B); however, the number of older patients receiving mechanical valves was small (15 of 87, 17%). Complication-free survival was similar with either mechanical (72 8% at 10 years, 72 8% at 15 years) or bioprosthetic (73 4% at 10 years, 71 4% at 15 years) valve replacement ( p 0.9) (Fig 4). For all patients, the estimate of long-term estimate of freedom from reoperation was relatively high in those with mechanical valves (74 9% at 10 years, 74 9% at 15 years), but started to decline steeply by the 10th year for patients who received bioprosthetic valves (56 5% at 10 years, 22 6% at 15 years) ( p 0.64). In the bioprosthetic group, the indication for reoperation was structural valve degeneration (SVD) in 63% and recurrent/residual endocarditis or a bland periprosthetic leak in 35%. In the younger patients (less than or equal to 60 years of age), the actuarial freedom from reoperation estimate was low after bioprosthetic valve replacement (51 5% at 10 years, 19 6% at 15 years) (Fig 5A). Using the actual (cumulative incidence) method of analysis, the divergence narrowed between the mechanical (81 8% at 10 years, 77 9% at 15 years) and bioprosthetic valve Fig 3. Long-term survival after mechanical or bioprosthetic valve replacement for patients: (A) less than or equal to 60 years of age, or (B) greater than 60 years of age. The numbers of patients at risk are indicated. groups (64 4% at 10 years, 48 4% at 15 years). For patients greater than 60 years of age, the long-term actuarial freedom from reoperation was acceptable with either a mechanical (100 0% at 10 years) or a bioprosthetic (84 7% at 10 years, 84 7% at 15 years) valve (Fig 5B). On the other hand, using the actual method of analysis, the likelihood of being free from reoperation increased slightly in the older patients receiving bioprosthetic valves (91 6% at 10 years, 91 5% at 15 years). Comment Endocarditis remains associated with substantial morbidity and mortality despite improvements in medical and surgical management during the last three decades. Innovative operative techniques have been applied in selected patients, including mitral valve repair when leaflet destruction is not excessive and homograft root replacement when aortic annulus reconstruction is necessary for extensive extravalvular infections [12]. Mitral

Ann Thorac Surg MOON ET AL 2001;71:1164 71 ENDOCARDITIS, TISSUE VS MECHANICAL VALVES 1169 Fig 4. Complication-free survival for patients undergoing valve replacement with mechanical, bioprosthetic, or homograft valves. valve repair is a very attractive option for patients with NVE, as long as adequate resection of all infected tissue does not compromise the durability of valve reconstruction. For aortic valve endocarditis, if greater than 50% of the annulus has been destroyed or there is extensive ventricular-aortic discontinuity, homograft root replacement may be preferable. For most patients with endocarditis, however, simple valve replacement remains the mainstay of surgical therapy. Aortic or mitral valve replacement is expedient, and can yield satisfactory longterm results for patients who otherwise have a dismal prognosis. In the current series, operative mortality rates were soberingly high (13 2% for NVE and 29 5% for PVE) but are consistent with those reported previously [8, 9] and with a recent statistical metaanalysis of 30 surgical series (12% NVE, 25% PVE) [12]. Historically, it has been suggested that bioprosthetic valves may be less susceptible to early recurrent endocarditis than mechanical valves, but that they may be more susceptible to late endocarditis due to infection of the tissue leaflets [13, 14]. An early review of 2,184 patients reported that resistance to infection, pathologic behavior once affected by PVE, and the survival rate after PVE were similar between bioprosthetic and mechanical valves [15]. Recent reviews have substantiated these findings with current reported linearized PVE rates of less than 1.0% per patient-year for bioprosthetic and mechanical valves [5 7, 14, 16]. In the current series, the linearized rate of recurrent or residual endocarditis was 1.5 0.3% per patient-year for NVE and 2.5 0.7% for PVE, and did not differ significantly between mechanical (1.2 0.5%) and bioprosthetic (1.8 0.3%) valves. Operative mortality and complication-free survival were also similar with mechanical and bioprosthetic valves. Infected bioprosthetic valves are, in general, more easily sterilized than mechanical valves if treatment is initiated before extension of the infection into the annulus occurs; for example, streptococcal bioprosthetic infections can be cured in 80% of patients with antibiotics as long as valvular dysfunction is not present [15, 17]. Infection of a bioprosthetic valve may, however, hasten structural valve degeneration due to injury to the bioprosthetic leaflets. This can also occur with homografts, where leaflet destruction and valvular dysfunction can occur early in the infective process, prompting surgical intervention despite the absence of annular extension [18]. Mechanical or bioprosthetic infections involving the sewing ring or annulus invariably necessitate valve rereplacement. In the current report, univariate analysis demonstrated that operative mortality was higher for patients with annular abscesses (25 4% vs 13 2%); while extravalvular extension was not an independent predictor of operative mortality, annular abscesses were associated with more late complications. Complete excision of all infected or necrotic tissue along with generous margins of clearly healthy tissue is the cornerstone of surgical treatment of patients with extravalvular involvement. Extensive debridement may necessitate novel, complex annular reconstructive techniques with or without homograft aortic root replacement in some cases [12, 19, 20]. Haydock and associates compared the use of freehand aortic homografts (78 patients) with mechanical or bioprosthetic valves (30 patients) in the surgical treatment of patients with active endocarditis [2]. They noted that with homografts, the hazard function for recurrent endocarditis had only a low constant phase, while with prosthetic valves, there was a high early hazard phase in addition to the low-hazard late constant phase. In the current report, only 20 homografts were implanted in highly selected patients with extensive destruction of surrounding cardiac structures, but 2 of 14 (14 9%) operative survivors developed recurrent endocarditis within 5 years. In this admittedly small subset of homograft patients, there was no apparent advantage to using this technique. Looking specifically at patients with prosthetic aortic infections, Lytle and coauthors found no difference in survival or freedom from reoperation if bioprosthetic, mechanical, or homografts where used for valve re-replacement ( p 0.64) [4]. Sweeney and colleagues from the Texas Heart Institute reviewed 185 patients who underwent mechanical (97 patients) or bioprosthetic (88 patients) valve replacement for endocarditis; mean follow-up time was 20 months [1]. They noted no difference in operative mortality between the groups, but found that reoperation for recurrent endocarditis or perivalvular leak was necessary in 15 (20% of operative survivors) of patients with a bioprosthetic valve but in only 5 (6% of operative survivors) mechanical valve recipients within 3 years. In contrast, freedom from reoperation in the current series was similar between mechanical and bioprosthetic valves at 3 years (86 5%, 87 3%) and at 5 years (86 5%, 84 3%). The Texas Heart group also suggested a small, but significant, survival advantage at 4 years (excluding operative deaths) with mechanical rather than bioprosthetic valves (87% vs 79%, p 0.05). In the current series, however, there was no difference in 4-year survival with mechanical or bioprosthetic valves (82 6%, 79 3%) or, more importantly, in late survival at 10 years (62 9%,

1170 MOON ET AL Ann Thorac Surg ENDOCARDITIS, TISSUE VS MECHANICAL VALVES 2001;71:1164 71 Fig 5. Actuarial freedom from reoperation with mechanical or bioprosthetic valve replacement for patients: (A) less than or equal to 60 years of age, or (B) greater than 60 years of age. The numbers of patients at risk are indicated. The inset depicts freedom from reoperation using the actual or cumulative incidence method of analysis. 61 4%), 15 years (46 10%, 52 5%), or even 20 years (46 10%, 41 6%) ( p 0.50). In this context it is important to note that most of the bioprosthetic valves implanted in the Texas Heart series were Ionescu-Shiley pericardial valves; most of these were an early version with a Teflon (polytetrafluorethylene) sewing ring, which does not promote rapid tissue ingrowth as does the Dacron velour sewing rings used in later Ionescu-Shiley valves and many current valves. Furthermore, this particular type of early pericardial valve was prone to early SVD, and is no longer on the market. These factors may have played a role in the high early reinfection rate reported in their bioprosthetic recipients. In the current series, no Ionescu-Shiley valves were implanted; only stented Hancock or Carpentier-Edwards porcine valves were used. In younger patients, the long-term reoperation rate was higher with bioprosthetic valves than with mechanical valves due to SVD, but, as patient age increased, the freedom from reoperation rates converged. Accepting that there are many different opinions among surgeons and a myriad of exceptions to the standard rule, a reasonable initial age criteria for implanting biologic prostheses for noninfectious valvular disease is 70 years of age for MVR and 65 to 70 years of age for AVR, if no comorbidities exist that would otherwise limit life expectancy. Fann and associates reported that the actuarial estimate of freedom from reoperation was very high in patients greater than 70 years of age after bioprosthetic AVR (93 2% at 10 years) and MVR (84 6% at 10 years), but as patient age fell below 70 years, the reoperation rates progressively rose [5]. In the current series of

Ann Thorac Surg MOON ET AL 2001;71:1164 71 ENDOCARDITIS, TISSUE VS MECHANICAL VALVES 1171 patients with endocarditis, freedom from reoperation appeared higher with mechanical valves in younger patients, but this risk was similar with either mechanical (100 0% at 10 years) or bioprosthetic (actuarial: 84 7% at 10 years, actual: 91 6% at 10 years) valves for patients greater than 60 years of age. It appears that for patients with infectious valvular disease, the age threshold for selecting a bioprosthetic valve should be lower than that for patients with noninfectious valvular disease. Furthermore, long-term survival was low in all patients with PVE (31 7% at 15 years, 16 7% at 20 years), suggesting that bioprosthetic valves may be appropriate for selected younger patients in this subset. The competing risk of death for patients with endocarditis markedly limits life expectancy, which means that fewer patients will actually live long enough to experience SVD of a tissue valve. This high competing hazard (eg, death) exemplifies the importance of using actual methods instead of actuarial techniques to examine what the true risk of a nonfatal valve-related complication is for individual patients; this is because the actuarial freedom curves overestimate the true incidence of such an event occurring. In summary, when valve replacement is necessary, we recommend mechanical prostheses for most patients with NVE who are less than 60 years of age, have no contraindication to long-term anticoagulation, and have a life expectancy that is otherwise not limited by other major medical problems. Bioprosthetic valves, on the other hand, are used for patients greater than 60 years of age with either NVE or PVE. Bioprosthetic valves are also acceptable for selected younger patients with PVE who have limited life expectancy, eg, coronary artery disease, left ventricular dysfunction, or end-stage renal failure. These data continue to argue strongly for early valve replacement (or repair) as the initial form of treatment for patients with congestive heart failure, in lieu of prolonged stabilization with medical therapy. Earlier surgical intervention, before the adverse multiorgan consequences of persistent infection and prolonged hemodynamic instability become manifest, may be the only way to improve the otherwise poor expected outcome in these often critically ill patients. References 1. Sweeney MS, Reul GJ, Cooley DA, et al. Comparison of bioprosthetic and mechanical valve replacement for active endocarditis. J Thorac Cardiovasc Surg 1985;90:676 80. 2. Haydock D, Barratt-Boyes B, Macedo T, Kirklin JW, Blackstone E. Aortic valve replacement for active infectious endocarditis in 108 patients. A comparison of freehand allograft valves with mechanical prostheses and bioprostheses. J Thorac Cardiovasc Surg 1992;103:130 9. 3. Wos S, Janinski M, Bachowski R, et al. Results of mechanical prosthetic valve replacement in active valvular endocarditis. J Cardiovasc Surg 1996;37(6 Suppl 1):29 32. 4. Lytle BW, Priest BP, Taylor PC, et al. Surgical treatment of prosthetic valve endocarditis. J Thorac Cardiovasc Surg 1996;111:198 210. 5. Fann JI, Miller DC, Moore KA, et al. Twenty-year clinical experience with porcine bioprostheses. Ann Thorac Surg 1996;62:1301 12. 6. Jamieson WRE, Burr LH, Munro AI, Miyagishima RT. Carpentier-Edwards standard porcine bioprosthesis: a 21-year experience. Ann Thorac Surg 1998;66(Suppl):S40 3. 7. Poirer NC, Pelletier LC, Pellerin M, Carrier M. 15-year experience with the Carpentier-Edwards pericardial bioprosthesis. Ann Thorac Surg 1998;66:S57 61. 8. Baumgartner WA, Miller DC, Reitz BA, et al. Surgical treatment of prosthetic valve endocarditis. Ann Thorac Surg 1983;35:87 104. 9. D Agostino RS, Miller DC, Stinson EB, et al. Valve replacement in patients with native valve endocarditis. What really determines operative outcome? Ann Thorac Surg 1985;40: 429 38. 10. Grunkemeier GL, Jamieson WRE, Miller DC, Starr A: Actuarial versus actual risk of porcine structural valve deterioration. J Thorac Cardiovasc Surg 1994;108:709 18. 11. Grunkemeier GL, Anderson RP, Miller DC, Starr A: Timerelated analysis of nonfatal heart valve complications: Cumulative incidence (actual) versus Kaplan-Meier (actuarial). Circulation 1997;96(Suppl II):II70 5. 12. Moon MR, Stinson EB, Miller DC. Surgical treatment of endocarditis. Prog Cardiovasc Dis 1997;40:239 64. 13. Cohn LH. Valve replacement for infective endocarditis. An overview. J Cardiac Surg 1989;4:321 3. 14. Agnihotri AK, McGiffin DC, Galbraith AJ, O Brien MF. The prevalence of infective endocarditis after aortic valve replacement. J Thorac Cardiovasc Surg 1995;110:1708 24. 15. Rossiter SJ, Stinson EB, Oyer PE, et al. Prosthetic valve endocarditis. Comparison of heterograft tissue valves and mechanical valves. J Thorac Cardiovasc Surg 1978;76:795 803. 16. Akins CW. Results with mechanical cardiac valvular prostheses. Ann Thorac Surg 1995;60:1836 44. 17. Durack DT. Infective and noninfective endocarditis. In: Schlant RC, Alexander RW, ed. Hurst s The Heart, 8th ed. San Francisco: McGraw-Hill, Inc., 1994:1681 709. 18. Clarkson PM, Barratt-Boyes BG. Bacterial endocarditis following homograft replacement of the aortic valve. Circulation 1970;42:987 91. 19. David TE, Feindel CM, Armstrong S, Sun Z. Reconstruction of the mitral anulus. A ten-year experience. J Thorac Cardiovasc Surg 1995;110:1323 32. 20. Ergin MA, Raissi S, Follis F, Lansman SL, Griepp RB. Annular destruction in acute bacterial endocarditis. Surgical techniques to meet the challenge. J Thorac Cardiovasc Surg 1989;97:755 63.