Prosthesis-Patient Mismatch After Aortic Valve Replacement: Impact of Age and Body Size on Late Survival

Transcription

1 ORIGINAL ARTICLES: ADULT CARDIAC SURGERY: To participate in The Annals of Thoracic Surgery CME Program, please visit Prosthesis-Patient Mismatch After Aortic Valve Replacement: Impact of Age and Body Size on Late Survival Marc R. Moon, MD, Michael K. Pasque, MD, Nabil A. Munfakh, MD, Spencer J. Melby, MD, Jennifer S. Lawton, MD, Nader Moazami, MD, John E. Codd, MD, Traves D. Crabtree, MD, Hendrick B. Barner, MD, and Ralph J. Damiano Jr, MD Division of Cardiothoracic Surgery, Washington University School of Medicine, St. Louis, Missouri Background. The purpose of this study was to identify patient subgroups in which prosthesis-patient mismatch most influenced late survival. Methods. Over a 12-year period, 1,400 consecutive patients underwent bioprosthetic (933 patients) or mechanical (467) aortic valve replacement. Prosthesis-patient mismatch was defined as prosthetic effective orifice area/ body surface area less than 0.75 cm 2 /m 2 and was present with 11% mechanical and 51% bioprosthetic valves. Results. With bioprosthetic valves, prosthesis-patient mismatch was associated with impaired survival for patients less than 60 years old (10-year: 68% 7% mismatch versus 75% 7% no mismatch, p< 0.02) but not older patients (p 0.47). Similarly, with mechanical valves, prosthesis-patient mismatch was associated with impaired survival for patients less than 60 years old (10-year: 62% 11% versus 79% 4%, p < 0.005) but not older patients (p 0.26). For small patients (body surface area less than 1.7 m 2 ), prosthesis-patient mismatch did not impact survival with bioprosthetic (p 0.32) or mechanical (p 0.71) valves. For average-size patients (body surface area 1.7 to 2.1 m 2 ), prosthesis-patient mismatch was associated with impaired survival with both bioprosthetic (p < 0.05) and mechanical (p< 0.005) valves. For large patients (body surface area greater than 2.1 m 2 ), prosthesis-patient mismatch was associated with impaired survival with mechanical (p< 0.04) but not bioprosthetic (p 0.40) valves. Conclusions. Prosthesis-patient mismatch had a negative impact on survival for young patients, but its impact on older patients was minimal. In addition, although prosthesis-patient mismatch was not important in small patients, prosthesis-patient mismatch negatively impacted survival for average-size patients and for large patients with mechanical valves. (Ann Thorac Surg 2006;81:481 9) 2006 by The Society of Thoracic Surgeons The impact of prosthesis-patient mismatch (PPM) after aortic valve replacement (AVR) remains controversial. Previous investigators have suggested that PPM may result in higher transvalvular gradients, blunted left ventricular (LV) mass regression, and increased early and late morbidity and mortality [1 7]. Others, however, have suggested that prosthesis size may play a lesser role in determining survival in the late postoperative period [8 11], but the use of internal orifice diameter to quantify effective valve size in some of these studies and a lack of Accepted for publication July 26, Presented at the Forty-first Annual Meeting of The Society of Thoracic Surgeons, Tampa, FL, Jan 24 26, Address correspondence to Dr Moon, Division of Cardiothoracic Surgery, Washington University School of Medicine, 3108 Queeny Tower, 1 Barnes-Jewish Plaza, St. Louis, MO ; moonm@msnotes.wustl.edu. subgroup analysis have led some to question the clinical applicability of these findings [5, 6, 12, 13]. Intuitively, implantation of a small valve in a large patient is not ideal, but it remains unknown in which patients the impact of an undersized valve would be most detrimental. Long-term survival after AVR has also been related to patient age and body size [5, 8, 14 17], but the role of these factors in determining the impact of PPM on late survival remains unclear [3, 5, 6, 8, 11, 18]. Two important clinical questions are as follows: (1) In which subgroups of patients is it necessary to upsize an aortic prosthesis to avoid PPM? (2) In which subgroups of patients does PPM not impact survival, such that a get in and get out approach would be most appropriate? The purpose of the current investigation was to identify patient subgroups in which PPM most influenced late survival; specifically, the impact of patient age and body size was examined by The Society of Thoracic Surgeons /06/$32.00 Published by Elsevier Inc doi: /j.athoracsur

2 482 MOON ET AL Ann Thorac Surg PROSTHESIS-PATIENT MISMATCH 2006;81:481 9 Table 1. Valves Implanted and Effective Orifice Area Based on Referenced Normal Values Effective Orifice Area (cm 2 ) Patients 19 mm 21 mm 23 mm 25 mm 27 mm References Bioprosthetic valves Hancock Standard , 20 Hancock Modified Orifice , 20 Hancock II , 22 Carpentier-Edwards Pericardial Carpentier-Edwards Porcine , 20 Medtronic Mosaic , Medtronic Freestyle Stentless , Toronto SPV Stentless , 32, 33 Other bioprosthetic 2 Different values for each valve type Mechanical valves St. Jude Medical , 34, 35 Medtronic-Hall , 37 Carbomedics Other mechanical 9 Different values for each valve type Material and Methods From June 1992 to May 2004, 1,462 consecutive patients underwent AVR at Washington University School of Medicine (Barnes-Jewish Hospital and Christian Hospital Northeast) by 21 different surgeons. Of these, 42 homograft and 20 autograft recipients were excluded, yielding 1,400 patients for statistical analysis. There were 801 men (57%) and 599 women (43%), with a mean age ( 1 SD) of years (range, 20 to 99); 1,007 (72%) were older than 60 years at the time of surgery. Indications for AVR included pure stenosis (44%), pure regurgitation (18%), combined stenosis and regurgitation (29%), and endocarditis (9%). A total of 134 patients (10%) previously underwent coronary artery bypass grafting and 88 (6%) previously underwent AVR. Replacement prostheses included 933 (67%) bioprosthetic and 467 (33%) mechanical. Table 1 enumerates specific valve types, and Figure 1 demonstrates a shift toward bioprosthetic valves since the early 1990s at this institution. Concomitant coronary artery bypass grafting was performed in 615 patients (44%), and concomitant mitral valve repair or replacement was performed in 199 patients (14%). Mean body surface area (BSA) was m 2, and patients were separated into three groups: (1) small patients with BSA less than 1.7 m 2 (275, 20%); (2) averagesize patients with BSA 1.7 m 2 to 2.1 m 2 (788, 56%); and (3) large patients with BSA greater than 2.1 m 2 (337, 24%). Estimates of effective orifice area (EOA) for each valve type and size were obtained from referenced normal valves as summarized in Table 1 [6, 19 40]. Indexed EOA was defined as prosthetic EOA divided by BSA, and PPM was defined as EOA/BSA less than 0.75 cm 2 /m 2 [3, 41]. Survival data were obtained for all 2004 patients during a 2-month closing interval ending August 2004 through interrogation of the Barnes-Jewish Hospital medical records database and the Social Security Death Index, which has been shown to be highly specific (99.5%) and unbiased [42]. Cumulative long-term follow-up totaled 5,194 patient-years. Mean follow-up for all patients was months, and 947 patients (68%) were alive an average of months postoperatively. Data Analysis Operative mortality included any death that occurred during the initial hospitalization or within 30 days of operation for discharged patients. Late survival data included death from all causes. Continuous data are reported as mean 1 SD and were compared between groups using Student s t test or analysis of variance as Fig 1. Numbers of bioprosthetic (solid circles) and mechanical (open circles) aortic valves implanted from 1986 to 2004 (total 2004 projected).

3 Ann Thorac Surg MOON ET AL 2006;81:481 9 PROSTHESIS-PATIENT MISMATCH 483 Table 2. Selected Preoperative and Intraoperative Clinical Characteristics for Patients Undergoing Bioprosthetic or Mechanical Aortic Valve Replacement With and Without Prosthesis-Patient Mismatch (PPM) a Bioprosthetic Valves Mechanical Valves No PPM PPM No PPM PPM Number of patients Age (years) Body surface area (m 2 ) b Female sex 168 (37%) 238 (50%) b 155 (37%) 38 (73%) b Diabetes mellitus 74 (16%) 160 (34%) b 67 (16%) 16 (31%) b Operative mortality 35 (7.7%) 47 (9.9%) 29 (7.0%) 5 (9.6%) a Continuous variables are mean SD. Groups were compared using Student s t test or 2 test as appropriate. b p 0.05 versus No PPM group. appropriate. Clinically important ratios are reported with 95% confidence limits. Actuarial survival estimates were calculated using the Kaplan-Meier method and were compared using the log-rank test. Variability of the actuarial estimates is expressed as 1 SEM. Univariate and multivariate analysis were used to determine the preoperative and intraoperative risk factors that were significant, independent predictors of operative mortality and late death (SigmaStat 2.03; SPSS, Chicago, Illinois). Univariate analysis was performed for categorical variables using the 2 test and for continuous variables using linear regression. Multivariate analysis was performed using stepwise backward regression including only factors identified to be significant during univariate analysis (p 0.05). Odds ratios (OR) are reported with 95% confidence intervals (CI). Twenty-four variables were analyzed: age; year of operation; sex; hypertension; diabetes; pulmonary disease; cerebrovascular disease; peripheral vascular disease; chronic renal insufficiency; history of myocardial infarction; smoking history; family history of coronary disease; congestive heart failure; ejection fraction; status (urgent, elective); ascending aortic aneurysm; endocarditis; New York Heart Association (NYHA) class; previous cardiac operation; concomitant coronary artery bypass grafting; concomitant mitral repair or replacement; valve type (mechanical, bioprosthetic); BSA (small, average, large); and PPM (EOA / BSA 0.75 cm 2 /m 2 ). Results Patient Characteristics Overall, PPM was present in 38% (528 of 1,400) but was more common with bioprosthetic AVR (476 of 933, 51%) than mechanical AVR (52 of 467, 11%; p 0.001). Selected preoperative and intraoperative clinical characteristics for patients undergoing bioprosthetic or mechanical AVR are listed in Table 2. With mechanical AVR, patient age tended to be slightly higher in the PPM group (p 0.06) but there was no age difference with bioprosthetic AVR (p 0.28). With bioprosthetic AVR, BSA was higher in the PPM group (p 0.001) but not with mechanical AVR (p 0.10). With both bioprosthetic and mechanical AVR, the PPM groups had a higher incidence of female sex (p for both) and diabetes mellitus (p 0.02 for both), but the PPM and no PPM groups were similar with regard to the presence of other comorbidities, including renal insufficiency (p 0.74 biomechanical, p 0.54 mechanical), pulmonary disease (p 0.59, p 0.82), congestive heart failure (p 0.95, p 0.66), left ventricular dysfunction (ejection fraction less than 0.40; p 0.63, p 0.99), peripheral vascular disease (p 0.10, p 0.16), and previous cardiac surgery (p 0.50, p 0.43). With bioprosthetic AVR, PPM occurred in 44% of patients (48 of 109) less than 60 years old and 52% of patients (428 of 821) 60 years old or more. With mechanical AVR, PPM occurred in 9% of patients (26 of 284) less than 60 years old and 14% of patients (26 of 183) 60 years old or more. With bioprosthetic AVR, PPM occurred in 35% of small patients (73 of 211), 47% of average-sizes patients (248 of 525), and 79% of large patients (155 of 197). With mechanical AVR, PPM occurred in 9% of small patients (7 of 77), 9% of average patients (22 of 244), and 16% of large patients (23 of 146). Operative Mortality Operative mortality for patients included in the PPM analysis was 8.3% 1.4% overall (116 of 1,400 patients); 9.8% 2.5% (52 of 528) with PPM and 7.3% 1.7% (64 of 872) without PPM (p 0.12). The difference in operative mortality between PPM and no PPM was not significant with either bioprosthetic (p 0.28) or mechanical (p 0.68) AVR (Table 2). The causes of operative death in 52 patients with PPM were cardiac (n 33), pulmonary (n 5), infection (n 5), and other (n 9). The causes of operative death in 64 patients without PPM were cardiac (n 44), pulmonary (n 7), infection (n 6), and other (n 7). Univariate analysis identified 13 factors associated with operative mortality (Table 3), of which multivariate analysis identified 9 factors to be independent predictors of operative mortality:: advanced age, chronic renal insufficiency, peripheral vascular disease, congestive heart failure, endocarditis, concomitant mitral procedure, previous cardiac surgery, urgent or emergent status, and NYHA class IV (Table 3). Late Survival Of the 1,284 operative survivors, there were 337 late deaths. The causes of late death could not consistently be deter-

4 484 MOON ET AL Ann Thorac Surg PROSTHESIS-PATIENT MISMATCH 2006;81:481 9 Table 3. Risk Factors for Operative Mortality for Patients Undergoing Aortic Valve Replacement (n 1,400 Patients) Number of Patients With Variable (%) p Value Univariate Analysis Odds Ratio a Multivariate Analysis p Value Age 1,400 (100) Earlier operative year 1,400 (100) 0.54 Female sex 599 (43) 0.18 Hypertension 894 (64) 0.27 Diabetes mellitus 317 (23) 0.09 Chronic pulmonary disease 139 (10) ( ) 0.70 Chronic renal disease 160 (11) ( ) Peripheral vascular disease 178 (13) ( ) Cerebrovascular disease 240 (17) 0.09 History of myocardial infarction 324 (23) ( ) 0.88 Smoking history 848 (61) 0.46 Family history of coronary disease 353 (25) 0.20 Congestive heart failure 891 (64) ( ) 0.02 Ejection fraction less than 0.40 b 400 (48) 0.34 NYHA class IV 358 (26) ( ) Previous cardiac surgery 245 (18) ( ) Concomitant coronary procedure 615 (44) ( ) 0.13 Concomitant mitral procedure 199 (14) ( ) Endocarditis 128 (9) ( ) Urgent or emergent status 199 (14) ( ) Ascending aortic aneurysm 124 (9) 0.68 BSA 1,400 (100) EOA/BSA 1,400 (100) 0.14 Bioprosthetic valve implanted 933 (67) 0.39 a Odds ratios are reported with 95% confidence intervals. b Ejection fraction was available in only 830 patients. BSA body surface area; EOA effective orifice area; NYHA New York Heart Association. mined. Univariate analysis identified 17 factors associated with late death (Table 4), of which multivariate analysis identified 11 factors to be independent predictors of late death: advanced age, earlier operative year, chronic renal insufficiency, diabetes mellitus, chronic pulmonary disease, peripheral vascular disease, congestive heart failure, ejection fraction less than 0.40, NYHA class IV, previous cardiac surgery, and lower body surface area (Table 4). Actuarial survival rates (including operative deaths) for all 1,400 patients were 84% 1% at 1 year, 68% 1% at 5 years, and 47% 2% at 10 years (Fig 2). When operative deaths were excluded, late survival was 92% 1% at 1 year, 74% 2% at 5 years, and 51% 3% at 10 years (Fig 2). For patients less than 60 years of age, PPM was associated with impaired survival, diminishing mean survival from months to 67.4 months (p 0.001; Fig 3). In contrast, for patients greater than or equal to 60 years of age, PPM did not have a negative influence on survival (p 0.95; Fig 4). These findings were consistent for both mechanical and bioprosthetic AVR, but more pronounced with mechanical valves. For mechanical recipients less than 60 years of age, 10-year survival was 79% 4% without PPM versus 62% 11% with PPM (p 0.005). For bioprosthetic recipients less than 60 years of age, 10-year survival was 75% 7% without PPM versus 68% 7% with PPM (p 0.02). For patients greater than or equal to 60 years of age, PPM did not effect survival with either mechanical (p 0.26) or bioprosthetic (p 0.47) AVR. For large patients (BSA greater than 2.1 m 2 ), PPM was associated with impaired survival for the entire group, diminishing mean survival from months to 91.6 months (p 0.02; Fig 5). However, while its impact was significant with mechanical AVR, in which 10-year survival fell from 78% 5% without PPM to 25% 14% with PPM (p 0.04), its impact did not reach statistical significance in large patients undergoing bioprosthetic AVR (p 0.40). For average-size patients (BSA 1.7 m 2 to 2.1 m 2 ), PPM was associated with impaired survival for the entire group, diminishing mean survival from 96.4 months to 72.9 months (p 0.001; Fig 6). These findings were pronounced with both mechanical and bioprosthetic AVR. For mechanical recipients of average size, 10-year survival was 59% 5% without PPM versus 38% 17% with PPM (p 0.005). For bioprosthetic recipients of average size, 10-year survival was 40% 5% without PPM versus 23% 8% with PPM (p 0.05). For small patients (BSA less than 1.7 m 2 ), the impact of PPM on survival was not significant for the entire group (p 0.66; Fig 7) nor specifically for mechanical (p 0.71) or bioprosthetic (p 0.32) AVR. To determine if a different cutoff valve for PPM would yield different results with regard to survival comparisons

5 Ann Thorac Surg MOON ET AL 2006;81:481 9 PROSTHESIS-PATIENT MISMATCH 485 Table 4. Risk Factors for Late Death for Patients Undergoing Aortic Valve Replacement a Univariate Analysis p Value Odds Ratio b Multivariate Analysis p Value Age Earlier operative year Female sex ( ) 0.88 Hypertension 0.12 Diabetes mellitus ( ) 0.04 Chronic pulmonary disease ( ) 0.03 Chronic renal disease ( ) Peripheral vascular disease ( ) 0.04 Cerebrovascular disease 0.18 History of myocardial ( ) 0.80 infarction Smoking history 0.92 Family history of coronary 0.18 disease Congestive heart failure ( ) Ejection fraction less than ( ) NYHA class IV ( ) 0.04 Previous cardiac surgery ( ) Concomitant coronary ( ) 0.43 procedure Concomitant mitral ( ) 0.07 procedure Endocarditis 0.27 Urgent or emergent status 0.07 Ascending aortic aneurysm ( ) 0.75 BSA EOA/BSA 0.07 Bioprosthetic valve implanted ( ) 0.16 a See Table 3 for number of patients with each variable. b Odds ratios are reported with 95% confidence intervals. BSA body surface area; EOA effective orifice area; NYHA New York Heart Association. Fig 2. Late survival estimates after aortic valve replacement for all patients (including operative deaths [solid circles]) and for operative survivors (open circles). The numbers of patients at risk at 1, 5, and 8 years are indicated. group has long had a focus on implantation of relatively large valve sizes when compared with other centers [8]. The incidence of PPM at the 0.75 cm 2 /m 2 level was low (11%) compared with the current report (38% overall, 51% for bioprosthetic valves). At our center, while the operative strategy has been quite diverse owing to the inclusion of data from 21 different surgeons during the 12-year span of this review, implantation of large valves has not been a major focus, similar to the approach reported by the Cleveland Clinic [8, 10, 13]. Rao and coauthors [3] noted improved 12-year survival in patients receiving a 23-mm or larger valve compared with those receiving a 19 to 21-mm valve (50% versus 43%). This type of comparison, however, does not take into account the variability in EOA of the various prosthetic valves [10, 43, 44]. Furthermore, although they noted a decline in late freedom from valve-related mortality between groups, the age and body size analyses were repeated with PPM values of 0.65 cm 2 /m 2, 0.85 cm 2 /m 2, 1.0 cm 2 /m 2, 1.1 cm 2 /m 2, 1.2 cm 2 /m 2, and 1.4 cm 2 /m 2. For patients greater than or equal to 60 years of age, PPM remained insignificant at all levels (p 0.30 for all). For patients less than 60 years of age, PPM continued to impact survival, but was no longer significant above the 1.1 cm 2 /m 2 level (p 0.09). For small patients, PPM remained insignificant at all levels (p 0.33 for all). For average-size patients, PPM significantly impaired survival at all levels except for the highest (p 0.26) and lowest (p 0.16). For large patients, PPM continued to impact survival, but was not significant above the 1.1 cm 2 /m 2 level (p 0.11). Comment Rao and colleagues [3] from Toronto reviewed 2,154 patients who underwent bioprosthetic AVR. The Toronto Fig 3. Impact of prosthesis-patient mismatch (PPM) on late survival after aortic valve replacement in patients less than 60 years of age (p 0.001). The numbers of patients at risk at 1, 5, and 8 years are indicated. (Solid circles PPM 0.75 cm 2 /m 2 or greater; open circles PPM less than 0.75 cm 2 /m 2.)

6 486 MOON ET AL Ann Thorac Surg PROSTHESIS-PATIENT MISMATCH 2006;81:481 9 Fig 4. Impact of prosthesis-patient mismatch (PPM) on late survival after aortic valve replacement in patients 60 years of age or more (p 0.95). The numbers of patients at risk at 1, 5, and 8 years are indicated. (Solid circles PPM 0.75 cm 2 /m 2 or greater; open circles PPM less than 0.75 cm 2 /m 2.) Fig 6. Impact of prosthesis-patient mismatch (PPM) on late survival after aortic valve replacement in average-size patients (body surface area 1.7 m 2 to 2.1 m 2 ;p 0.001). The numbers of patients at risk at 1, 5, and 8 years are indicated. (Solid circles PPM 0.75 cm 2 /m 2 or greater; open circles PPM less than 0.75 cm 2 /m 2.) with PPM (84% versus 75%), overall survival at 12 years was similar (49% versus 50%) [3]. In the current report, subgroup analysis based on patient age, body size, and mechanical versus bioprosthetic implantation elucidated important differences. The current findings demonstrate that PPM is a sizeand age-dependent phenomenon. In patients receiving mechanical valves, PPM should be avoided in averagesize or large patients, especially if they are young. Patients with BSA greater than 2.1 m 2 had a dramatic fall in survival from 78% to 25% with PPM, whereas patients with BSA less than 1.7 m 2 did not experience the same response with PPM. In patients receiving bioprosthetic valves, the data suggest that PPM should be avoided in average-size patients as 10-year survival diminished from 40% to 23%, but smaller and larger patients did not demonstrate a similar decline in survival. With both mechanical and bioprosthetic AVR, young patients were more prone to experience the negative impact of PPM on late survival than were older patients. In an elegant, large, multicenter retrospective study, Blackstone and coinvestigators [11] found increased operative mortality, but no change in late survival with diminished valve-to-patient sizing; however, internal orifice diameter was used instead of EOA to quantify prosthesis size. This approach has limitations when comparing various mechanical and bioprosthetic valves that may have the same internal orifice diameter but widely varying flow patterns and opening characteristics [5, 6, Fig 5. Impact of prosthesis-patient mismatch (PPM) on late survival after aortic valve replacement in large patients (body surface area greater than 2.1 m 2 ;p 0.02). The numbers of patients at risk at 1, 5, and 8 years are indicated. (Solid circles PPM 0.75 cm 2 /m 2 or greater; open circles PPM less than 0.75 cm 2 /m 2.) Fig 7. Impact of prosthesis-patient mismatch (PPM) on late survival after aortic valve replacement in small patients (body surface area less than 1.7 m 2 ;p 0.66). For consistency, the numbers of patients at risk at 1, 5, and 8 years are indicated below the lines for PPM and above the lines for no PPM. (Solid circles PPM 0.75 cm 2 /m 2 or greater; open circles PPM less than 0.75 cm 2 /m 2.)

7 Ann Thorac Surg MOON ET AL 2006;81:481 9 PROSTHESIS-PATIENT MISMATCH , 13]. Similar to the Blackstone study, the current report focused on survival alone, not taking into account left ventricular mass regression or late functional status [2, 5, 7, 33, 43]. It is likely that PPM also impacts these two important endpoints. Reul and colleagues [7] found that PPM was independently associated with persistent or recurrent congestive heart failure in the late postoperative period. They did not identify an impact of PPM on late mortality, but subgroup analysis based on age and body size was not performed. Blais and associates [6] examined the impact of PPM on operative mortality in 1,266 patients collected over 10 years. They quantified PPM as absent (greater than 0.85 cm 2 /m 2 ), moderate (0.65 to 0.85 cm 2 /m 2 ), or severe (less than 0.65 cm 2 /m 2 ). Moderate or severe PPM was present in 38% of their patients, similar to the current series. Operative mortality for the entire series was 4.6% but increased significantly from 3% without PPM to 6% with moderate PPM and 26% with severe PPM. In the current report, when the analyses were performed using other EOA/BSA values as the PPM cutoff point, the impact on late survival was similar. The impact of PPM is likely linear, rather than binary. Below a critical level of PPM, which likely varies for each subgroup identified to be at increased risk with PPM, the impact of PPM increases as the mismatch becomes more severe. However, whether one chooses a cutoff value for PPM of 0.75 cm 2 /m 2, 0.85 cm 2 /m 2, or 1.0 cm 2 /m 2, the impact of age and body size is similar. When a cutoff value of 1.2 cm 2 /m 2 was employed, the impact of PPM on late survival was no longer significant in most subgroups. Study Limitations The current study was subject to all the limitations inherent to a retrospective, nonrandomized comparison of surgical results, including selection bias as to which patients received which prostheses during valve replacement. Multivariate analysis was used to help account for selection bias and other confounding risk factors, but the possible impact of undersizing valves in sicker patients to get in and get out may have influenced the results. We were not surprised that PPM was not an independent predictor of operative mortality (p 0.14) or late death (p 0.07) in the initial univariate and multivariate analyses. That is consistent with our impression that PPM is not important in all patients. It was our initial impression that analysis of the entire group was not what was going to be of interest, and that is why we additionally focused on specific subgroups of patients that we thought were of most clinical interest (body surface area, age, and mechanical versus bioprosthetic recipients). This series included 1,400 patients who underwent AVR by 21 different surgeons over a 12-year period. Some may consider such diversity a limitation of the study because of a lack of a consistent approach to the patient with a small aortic root, but we believe that our data may better predict the expected outcome for a given patient and a given surgeon in practice than does a similar series from a single surgeon or from a center with a standard surgical approach. In the current report, late functional status and quality of life were not assessed and left ventricular mass regression was not evaluated with postoperative echocardiographic data. The goal of the current study was to identify in which patient subgroups late survival was most affected by PPM and in whom consideration of advanced surgical techniques to increase valve size during AVR may be warranted on a survival basis. We made no attempt to correlate PPM findings with functional status or symptomatic improvements. Previous reports have suggested that PPM may impact each of these parameters and that larger valves are associated with improved LV mass regression and improved functional status [2, 5, 7]. Further analysis would be required to assess these findings in our series in which postoperative echocardiography data was inconsistent. In summary, the current findings demonstrated that PPM is a real phenomenon and can impact late survival in selected patients. With an EOA/BSA ratio of 1.2 cm 2 /m 2, PPM was not present in any subgroup. However, with an EOA/BSA ratio of 0.75 cm 2 /m 2, PPM had a significant negative impact on survival in average-size and large patients, especially if they were young. Thus, it is in these patients that we should focus our attempts to improve prosthesis-patient matching [31 33, 45], while adopting a less aggressive approach in small, elderly patients [8, 13, 18]. Castro and associates [45] demonstrated that an aggressive approach to aortic root enlargement can decrease the incidence of PPM from 17% to 3% without a significant rise in operative mortality. However, while the increased perioperative mortality rate that surely accompanies more complex procedures in most hands may be warranted in young patients of average size or greater, small patients, especially when elderly, may be better served with a get in and get out approach, as PPM does not appear to have a significant impact on late survival. The authors gratefully acknowledge the clinical contributions of Charles B. Huddleston, MD, William A. Gay, Jr, MD, James L. Cox, MD, Thoralf M. Sundt III, MD, Michael Rosenbloom, MD, Thomas L. Spray, MD, T. Bruce Ferguson, Jr, MD, Scott H. Johnson, MD, Eric N. Mendeloff, MD, Alfredo Rego, MD, Richard Shaw, MD, Lawrence Creswell, MD, Nicholas T. Kouchoukos, MD, and Thomas B. Ferguson, Sr, MD. References 1. 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 bioprosthetic heart valve. J Heart Valve Dis 1998;7: Del Rizzo DF, Abdoh A, Cartier P, Doty D, Westaby S. Factors affecting left ventricular mass regression after aortic valve replacement with stentless valve. Semin Thorac Cardiovasc Surg 1999;11: Rao V, Jamieson WRE, Ivanov J, Armstrong S, David TE. Prosthesis-patient mismatch affects survival after aortic valve replacement. Circulation 2000;102(Suppl 3): Milano AD, DeCarlo M, Mecozzi G, et al. Clinical outcome in patients with 19-mm and 21-mm St. Jude aortic prostheses. Comparison at long-term follow-up. Ann Thorac Surg 2002; 73:37 43.

8 488 MOON ET AL Ann Thorac Surg PROSTHESIS-PATIENT MISMATCH 2006;81: Rahimtoola SH. Choice of prosthetic heart valve for adult patients. J Am Coll Cardiol 2003;41: 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: Ruel M, Rubens FD, Masters RG, et al. Late incidence and predictors of persistent or recurrent heart failure in patients with aortic prosthetic valves. J Thorac Cardiovasc Surg 2004;127: Medalion B, Blackstone EH, Lytle BW, White J, Arnold JH, Cosgrove DM. Aortic valve replacement. Is valve size important? J Thorac Cardiovasc Surg 2000;119: Hanayama N, Christakis GT, Mallidi HR, et al. Patient prosthesis mismatch is rare after aortic valve replacement. Valve size may be irrelevant. Ann Thorac Surg 2002;73: Gillinov AM, Blackstone EH, Rodriquez LL. Prosthesispatient size. Measurement and clinical implications. J Thorac Cardiovasc Surg 2003;126: Blackstone EH, Cosgrove DM, Jamieson WRE, et al. Prosthesis size and long-term survival after aortic valve replacement. J Thorac Cardiovasc Surg 2003;126: Seitelberger R, Bialy J, Gottardi R, et al. Relation between size of prosthesis and valve gradient. Comparison of two aortic bioprosthesis. Eur J Cardiothorac Surg 2004;25: Dumesnil JG, Pibarot P. Prosthesis size and prosthesispatient size are unrelated to prosthesis-patient mismatch. J Thorac Cardiovasc Surg 2004;127:1852 [author reply ]. 14. He GW, Grunkemeier GL, Gately HL, Furnary AP, Starr A. Up to thirty-year survival after aortic valve replacement in the small aortic root. Ann Thorac Surg 1995;59: Fann JI, Miller DC, Moore KA, et al. Twenty-year clinical experience with porcine bioprosthesis. Ann Thorac Surg 1996;62: Banbury MK, Cosgrove DM, White JA, Blackstone EH, Frater RWM, Okies JE. Age and valve size effect on the long-term durability of the Carpentier-Edwards aortic pericardial bioprosthesis. Ann Thorac Surg 2001;72: Ikonomidis JS, Kratz JM, Crumbley AJ, et al. Twenty-year experience with the St. Jude Medical mechanical valve prosthesis. J Thorac Cardiovasc Surg 2003;126: Freed DH, Tam JW, Moon MC, Harding GE, Ahmad E, Pascoe EA. Nineteen-millimeter prosthetic aortic valves allow normalization of left ventricular mass in elderly women. Ann Thorac Surg 2002;74: Rashtian MY, Stevenson DM, Allen DT, et al. Flow characteristics of bioprosthetic heart valves. Chest 1990;98: Desai ND, Christakis GT. Stented mechanical/bioprosthetic aortic valve replacement. In: Cohn LH, Edmunds LH Jr, eds. Cardiac surgery in the adult. 2nd ed. Chicago: McGraw-Hill, 2003: David TE, Pollick C, Bos J. Aortic valve replacement with stentless procine aortic bioprosthesis. J Thorac Cardiovasc Surg 1990;99: David TE, Armstrong S, Sun Z. Clinical and hemodynamic assessment of the Hancock II bioprosthesis. Ann Thorac Surg 1992;54: Salomon NW, Okies JE, Krause AH, Page US, Bigelow JC, Colburn LQ. Serial follow-up of an experimental bovine pericardial aortic bioprosthesis: usefulness of pulsed Doppler echocardiography. Circulation 1991;84(Suppl 3): Banbury MK, Cosgrove DM, Thomas JD, et al. Hemodynamic stability during 17 years of the Carpentier-Edwards aortic pericardial bioprosthesis. Ann Thorac Surg 2002;73: Dellgren G, David TE, Raanani E, Armstrong S, Ivanov J, Rakowski H. Late hemodynamic and clinical outcomes of aortic valve replacement with the Carpentier-Edwards perimount pericardial bioprosthesis. J Thorac Cardiovasc Surg 2002;124: Seitelberger R, Bialy J, Gottardi R, et al. Relation between size of prosthesis and valve gradient. Comparison of two aortic bioprosthesis. Eur J Cardiothorac Surg 2004;25: Thomson DJ, Jamieson WR, Dumesnil JG, et al. Medtronic mosaic porcine bioprosthesis. Satisfactory early clinical performance. Ann Thorac Surg 1998;66(Suppl): Botzenhardt R, Gansera B, Kemkes BM. Mid-term hemodynamic and clinical results of the stented porcine Medtronic mosaic valve in aortic position. Thorac Cardiovasc Surg 2004;52: Dumesnil JS, LeBlanc MH, Cartier PC, et al. Hemodynamic features of the freestyle aortic bioprosthesis compared with stented bioprosthesis. Ann Thorac Surg 1998;66(Suppl): Yun KL, Sintek CF, Fletcher AD, et al. Aortic valve replacement with the Freestyle stentless bioprosthesis. Five-year experience. Circulation 1999;100(Suppl 2): Yun KL, Jamieson WR, Khonsari S, Burr LH, Munro AI, Sintek CF. Prosthesis-patient mismatch. Hemodynamic comparison of stented and stentless aortic valves. Semin Thorac Cardiovasc Surg 1999;11: Del Rizzo DF, Goldman BS, Christakis GT, David TE. Hemodynamic benefits of the Toronto stentless valve. J Thorac Cardiovasc Surg 1996;112: Dellgren G, David TE, Raanani E, Bos J, Ivanov J, Rakowski H. The Toronto SPV. Hemodynamic data at 1 and 5 years postimplantation. Semin Thorac Cardiovasc Surg 1999;11: Chafizadeh ER, Zoghbi WA. Doppler echocardiographic assessment of the St. Jude Medical prosthetic valve in the aortic position using the continuity equation. Circulation 1991;83: Carrel T, Zingg U, Jenni R, Aeschbacher B, Turina MI. Early in vivo experience with the hemodynamic plus St. Jude Medical heart valves in patients with narrowed aortic annulus. Ann Thorac Surg 1996;61: Wiseth R, Levang OW, Sande E, Tangen G, Skjaerpe T, Hatle L. Hemodynamic evaluation by Doppler echocardiography of small (less than or equal to 21 mm) prostheses and bioprostheses in the aortic valve position. Am J Cardiol 1992;70: Gerdts E, Stangeland L, Engedal H, Nordrehaug JE. Postoperative Doppler echocardiographic evaluation in different sizes of Metronic-Hall, Bicor and Carpentier-Edwards S.A.V. prosthetic aortic valves. Scand J Thorac Cardiovasc Surg 1994;28: Ihlen H, Molstad P, Simonsen S, et al. Hemodynamic evaluation of the Carbomedics prosthetic heart valve in the aortic position. Comparison of noninvasive and invasive techniques. Am Heart J 1992;123: Chambers J, Cross J, Deverall P, Sowton E. Echocardiographic description of the Carbomedics bileaflet prosthetic heart valve. J Am Coll Cardiol 1993;21: Bernal JM, Martin-Duran R, Rabasa JM, Revuelta JM. The Carbomedics Top-Hat supraannular prosthesis. Ann Thorac Surg 1999;67: McDonald ML, Daly RC, Schaff HV, et al. Hemodynamic performance of small aortic valve bioprosthesis. Is there a difference? Ann Thorac Surg 1997;63: Newman TB, Brown AN. Use of commercial record linkage software and vital statistics to identify patient deaths. J Am Inform Assoc 1997;4: Pibarot P, Dumesnil JG. Hemodynamics and clinical impact of prosthesis-patient mismatch in the aortic valve position and its prevention. J Am Coll Cardiol 2000;36: Vannan MA, Sarkar K. Assessment of mechanical aortic valve prosthesis by means of Doppler echocardiography. What to measure and why? J Thorac Cardiovasc Surg 2003;126: Castro LJ, Arcidi JM, Fisher AL, Gaudiani VA. Routine enlargement of the small aortic root. A preventive strategy to minimize mismatch. Ann Thorac Surg 2002;74:31 6.

9 Ann Thorac Surg MOON ET AL 2006;81:481 9 PROSTHESIS-PATIENT MISMATCH 489 DISCUSSION DR STEVEN F. BOLLING (Ann Arbor, MI): Obviously these two papers were put together by the Program Committee, and perhaps you could make a comment on the difference between this paper and the last one. They used 0.65 as their cutoff for severe PPM and you used 0.75 for your overall cutoff of PPM; they found no impact of even more severe PPM, and yet you found an impact upon survival with PPM; assuming that four fifths of your valves were St. Jude valves, so you had the same substrate, the same limitations of your study, and yet very different findings. DR MOON: Well, St. Jude or Carbomedics. I did the analysis with 0.65, 0.75, 0.85, 1, 1.1, 1.2, 1.4. The results were similar when we used 0.75, 0.85, and 1. When we went to the extremes, though, the numbers in the groups became so small, especially with mechanical valves, we had very few mechanical patients that were less than So that is why we picked 0.75 as our cutoff point. As I mentioned, prosthesis-patient mismatch is probably not a binary effect but more of a linear effect and the results may be different with various levels. DR BOLLING: But even given that, please explain how you would have a different conclusion from the previous paper. DR MOON: Overall in our mechanical patients when we had patient-prosthesis mismatch less than 0.75, we had diminished survival. At 0.65, when I redid the analysis, in our series anyway, our numbers of patients were quite small and none of our differences came out statistically significant. So I think with more numbers and more power maybe these small differences can come out. We put in a lot of bioprosthetic valves, especially in older patients, and maybe in their series it was a little bit older. It is hard to say. DR ROBERT W. M. FRATER (Bronxville, NY): The value for your effective orifice area is, I presume, a derived EOA from echocardiography, and I wonder whether the ventricle has any idea of what an EOA is. What the ventricle notices is the pressure generated in order to open the valve and eject. Do you have any data on left ventricular pressure in the patients who were supposed to have patient-prosthesis mismatch? DR MOON: The short answer is no, I don t have any specific data. I have looked at some small subsets of recent patients in whom we do have data, and I have found substantial differences in the gradients, but we don t have any consistent data on that, especially for the valves, and it makes it difficult to compare all the different types of valves. They have all been studied in different ways, in vitro, in vivo. I tried to pick studies to use for our reference values that appeared to be unbiased, and any studies that had values that were way out of whack, I essentially didn t use. DR W. R. ERIC JAMIESON (Vancouver, British Columbia): A question with regard to your biological population. In the United States, it has only been in the last few years that you have had the advanced bioprosthetic valves, the SupraAnnular type valves to implant. What percentage of your population of bioprostheses were old standard Hancocks and standard Edwards? There has been a tremendous advance in the use of bioprostheses in the United States in the last year or so, up to 77%. We have to take these data with some consideration that your old valves are contributing to your biological results. DR MOON: One nice thing about using the effective orifice area and having a series that is so old is that we have got patients and valves all over the map. Any given valve can have any given effective orifice area, and if we take a new SupraAnnular 17 valve, it is probably no better than a 21 porcine Hancock. We don t have to know the specifics. All we have to know is the effective orifice area in order to compare the effects of the mismatch. Obviously, we will have mismatch less often now with the new SupraAnnulars that have better flow characteristics than we did in the past. DR ROBERT W. EMERY (St. Paul, MN): That was a very nice presentation. The differences in our two series are interesting. In thinking about this, three things seem to differentiate what we both reported and may offer some answers. One is duration of follow-up. In a 10-year follow-up, there has not been enough time for biologic valves to degenerate. If you look at Steve Kahn s 20-year follow-up between mechanical and tissue valves, the incidence curves cross at 10 years and the incidence of valve-related complications are higher in the latter 10 years in tissue valves, predominantly due to valve failure. That may be an answer to one of the differences in what we are seeing things. The second is our surgical technique was very consistent, because basically all of us were taught by Dr Nicoloff. So for 25 years, the members of our group were using standard interrupted mattress sutures in the aortic annulus and just changed a little bit the number of sutures to make the valves fit with the diameter-enhanced model. So we had a very consistent surgical approach, and in your series there were 26 different surgeons and a multiple admixture of both mechanical and biologic valves; it may be that one type of valve or one surgical technique contributed to the different in our results. The third issues is that we followed only one valve type whereas your report contains multiple valve models. DR MOON: I agree with you one hundred percent. One thing we found amazingly when we looked at the data was how low our 10-year survival was for patients all across the board: patients just don t live as long as we think they are going to. And so to make a 30-year decision in an older patient is probably not warranted. And I agree, we did have quite a diverse group, and all of my partners and previous partners know, none of us did anything the same.