Effect of Concomitant Coronary Artery Disease on Procedural and Late Outcomes of Transcatheter Aortic Valve Implantation

Transcription

1 ADULT CARDIAC Effect of Concomitant Coronary Artery Disease on Procedural and Late Outcomes of Transcatheter Aortic Valve Implantation Todd M. Dewey, MD, David L. Brown, MD, Morley A. Herbert, PhD, Dan Culica, MD, PhD, Craig R. Smith, MD, Martin B. Leon, MD, Lars G. Svensson, MD, PhD, Murat Tuzcu, MD, John G. Webb, MD, Alain Cribier, MD, and Michael J. Mack, MD Medical City Dallas Hospital, Dallas, and Cardiopulmonary Research Science and Technology Institute, Dallas, Texas; Department of Surgery, Columbia Presbyterian Hospital, New York, New York; Department of Thoracic and Cardiovascular Surgery, and Department of Cardiology, Cleveland Clinic Foundation, Cleveland, Ohio; St. Paul s Hospital, Department of Cardiology, University of British Columbia, Vancouver, British Columbia, Canada; Department of Cardiology, University Hospital Charles Nicolle, Rouen, France Background. Previous coronary artery bypass grafting increases predicted operative risk for conventional valve replacement, according to the Society of Thoracic Surgeons risk algorithm. Additionally, the presence of coronary artery disease (CAD) has been demonstrated to increase procedural risk with conventional aortic valve replacement. Significant coexisting CAD requires preemptive percutaneous coronary intervention (PCI) in patients under consideration for transcatheter aortic valve implantation (TAVI). This study examined the impact of previous coronary artery bypass grafting or PCI on procedural outcomes and overall survival in patients having TAVI. Methods. Two hundred and one high-risk patients were enrolled in two international feasibility studies from December 2005 to February 2008 for the treatment of aortic stenosis using TAVI. Thirty patients were excluded from analysis due to failure to successfully deploy the valve in the aortic annulus. Data were collected concurrently using an ad hoc database that included operative and long-term survival. Previous cardiovascular intervention prior to TAVI was used to identify the existence of concomitant CAD. Logistic regression along with Kaplan-Meier estimates were employed to establish the association between CAD and survival from TAVI. Results. Overall mortality after TAVI was significantly higher among the CAD group (35.7%) in contrast with the non-cad patients (18.4%), p Logistic regression analysis found that patients who had CAD were 10.1 times more likely to die (95% confidence interval 2.1 to 174.8) within 30 days of the procedure than those who did not. Proportional hazards analysis established that the risk of dying at any point in time was 2.3 times higher among the patients with CAD (95% confidence interval 1.29 to 4.17). Kaplan-Meier survival curves demonstrate improved longterm survival among patients without CAD. Conclusions. Coexisting coronary artery disease negatively impacts procedural outcomes and long-term survival in patients undergoing TAVI, and implies that risk assessment and anticipated outcomes might be inaccurate due to stratification as isolated aortic valve replacement rather than AVR CABG. Comparison of procedural outcomes, based on operative approach without controlling for unequal distribution of CAD in the cohorts, are likely invalid. (Ann Thorac Surg 2010;89:758 67) 2010 by The Society of Thoracic Surgeons Since the first reported clinical implantation of a transcatheter valve in 2002 [1], the rate of assimilation into the clinical arena of this novel technology has been exceedingly rapid. Commercialization of both selfexpanding and balloon expandable platforms in Europe, as an alternative to conventional aortic valve replacement, has led to the implantation of more than 2,000 such valves worldwide. Current patient cohorts considered ideal for these devices primarily center on those at increased or prohibitive risk for conventional valve replacement. Early reports have suggested incremental Accepted for publication Dec 15, Presented at the Forty-fifth Annual Meeting of The Society of Thoracic Surgeons, San Francisco, CA, Jan 26 28, Address correspondence to Dr Dewey, 7777 Forest Ln, Ste A323, Dallas, TX 75235; tdewey@csant.com. clinical benefit in selected groups of high-risk patients with critical aortic stenosis [2 6]. Increasing transcatheter experience has allowed standardization in procedural techniques and surmounting of the learning curve in many centers [7]. Additionally, iterative device improvements have enabled safer, faster, and more efficient procedures [8]. Despite these advances, procedural mortality remains high in comparison with some traditional populations having aortic valve replacement, and places a premium on appropriate patient selection. Additionally, conclusively identifying the appropriate high-risk population has been problematic. Dr Cribier discloses a financial relationship with Edwards Lifesciences by The Society of Thoracic Surgeons /10/$36.00 Published by Elsevier Inc doi: /j.athoracsur

2 Ann Thorac Surg DEWEY ET AL 2010;89: EFFECT OF CAD ON OUTCOMES OF TAVI Risk algorithms have been used to assess operative risk, anticipate outcomes, and provide for comparability of patients among diverse centers and countries. Unfortunately, these tools are inherently imprecise and frequently exclude comorbidities encountered in this population [9]. Therefore, the comparison of transcatheter procedural outcomes to anticipated results based upon predictive risk scoring remains somewhat subjective. Discriminative patient selection requires identification of inherent risk factors that unfavorably influence procedural outcomes. By identifying those risk factors, practitioners can analyze procedural outcomes with realistic expectations in mind. The presence of coronary artery disease (CAD) has been clearly demonstrated to increase procedural risk with conventional aortic valve replacement. However, its overall influence on outcomes of transcatheter therapy for aortic stenosis has not been clearly delineated. This is especially true given the fact that transcatheter aortic valve implantation (TAVI) is generally considered a stand-alone procedure, with variable degrees of concomitant coronary artery disease tolerated without intervention. Previous cardiovascular intervention (CVI), either with coronary artery bypass grafting (CABG) or percutaneous coronary intervention (PCI), can be used as a marker for the presence of coexisting CAD. This study examines the impact of CAD on procedural outcomes and long-term survival in patients undergoing TAVI. Patients and Methods This study is a retrospective review of data records of 201 patients enrolled in two studies. It was reviewed and approved by the North Texas Institutional Review Board at Medical City Dallas with a waiver of consent. Between December of 2005 and February of 2008, 201 high-risk patients were enrolled in two feasibility studies utilizing the stainless-steel balloon expandable Sapien valve (Edwards Lifesciences, Irvine, CA) at 12 participating centers in North America and Europe. A total of 40 patients underwent attempted transapical aortic valve implantation (TA-AVI), and 161 patients attempted transfemoral aortic valve implantation (TF-AVI). Inclusion criteria for the TA- AVI patients have been previously reported [10] but included critical aortic stenosis with a valve area of 0.6 cm 2 or less, age 70 years old or greater, a Society of Thoracic Surgeons (STS) predicted risk of mortality greater than 15%, or deemed inoperable. Inclusion criteria for the transfemoral patients included a logistic European system for cardiac operative risk evaluation (EuroSCORE) predicted mortality of greater than 20% and (or) an STS predicted risk of 10% or greater, an aortic valve area of 0.7 cm 2 or less, as well as adequate peripheral access suitable for a retrograde approach to the aortic annulus [11]. Patients were excluded from analysis if the procedure was aborted prior to valve implantation, or if technical complications resulted in embolization of the valve or malplacement such that conversion to an open procedure was required. Baseline demographic data as well as procedural results and long-term outcomes were recorded contemporaneously using an ad hoc database designed for the studies. Hemodynamic data including peak and mean aortic gradients, ejection fraction, aortic valve area, and end diastolic volumes were entered into the database. Data were analyzed using SAS 9.2 (SAS Institute, Cary, NC). Categoric variables were analyzed using 2 statistics, or the Fisher exact test when small cell counts were Table 1. Descriptive Characteristics of Patients Who Had Transapical Approach With or Without CAD Categoric Variables CAD (n 26) No CAD (n 9) p Value Female 50.0% (13) 44.4% (4) 1.00 Past smoker 46.2% (12) 55.6% (5) Current smoker 0 3.9% (1) 0.84 Diabetes 34.6% (9) 44.4% (4) 0.69 Hyperlipidemia 92.3% (24) 44.4% (4) Chronic renal disease 23.1% (6) 11.1% (1) 0.64 Hypertension 88.5% (23) 66.7% (6) 0.16 History of COPD 23.1% (6) 55.6% (5) 0.10 Peripheral vascular disease 65.4% (17) 22.2% (2) 0.05 Cerebral vascular disease 26.9% (7) 33.3% (3) 0.69 Myocardial infarction 30.8% (8) 11.1% (1) 0.39 Congestive heart failure 88.5% (23) 88.9% (8) 1.00 Angina 69.2% (18) 22.2% (2) 0.03 Previous aortic valvuloplasty 42.3% (11) 44.4% (4) 1.00 Aortic regurgitation: 0.17 None 19.2% (5) 33.3% (3) % (10) 11.1% (1) % (10) 33.3% (3) 3 3.9% (1) 22.2% (2) Baseline NYHA functional 0.45 class: I 0 0 II 0 0 III 61.5% (16) 77.8% (7) IV 38.5% (10) 22.2% (2) Mitral regurgitation: % (7/25) 37.5% (3/8) % (8/25) 37.5% (3/8) % (8/25) 25.0% (2/8) 4 8.0% (2/25) 0 Implanted valve size (mm): % (10) 22.2% (2) % (16) 77.8% (7) Continuous Variables Mean SD Mean SD 759 Age Logistic EuroSCORE STS Predicted risk of mortality Body surface area (Dubois) (m 2 ) Ejection fraction Aortic valve area (cm 2 ) Mean gradient (mm Hg) Peak aortic gradient (mm Hg) BMI (kg/m 2 ) BMI body mass index; CAD coronary artery disease; COPD chronic obstructive pulmonary disease; EuroSCORE European system for cardiac operative risk evaluation; NYHA New York Heart Association; STS Society of Thoracic Surgeons. ADULT CARDIAC

3 ADULT CARDIAC 760 DEWEY ET AL Ann Thorac Surg EFFECT OF CAD ON OUTCOMES OF TAVI 2010;89: observed. The t tests were used to compare continuous variables. Univariate and multivariate logistic regression was performed as noted, and survival curves were generated with data from Kaplan-Meier statistics. Hazard ratios were calculated using Cox proportional hazards. Log-rank statistics were used for comparing curves. Results A total of 201 patients were identified as having undergone attempted TA-AVI (n 40) or TF-AVI (n 161) from 12 centers in North America and Europe between December 2005 and February 2008 as part of transapical Table 2. Descriptive Characteristics of Patients Who Had Transfemoral Approach With or Without CAD Categoric Variables CAD (n 58) No CAD (n 78) p Value Female 34.5% (20) 64.1% (50) Past smoker 48.3% (28) 29.5% (23) Current smoker 1.7% (1) 1.3% (1) 0.09 Diabetes 27.6% (16) 24.4% (19) 0.67 Hyperlipidemia 62.1% (36) 37.2% (29) Chronic renal disease 41.4% (24) 7.7% (6) Hypertension 82.8% (48) 68.0% (53) 0.05 Cerebrovascular accident 17.2% (10) 15.4% (12) 0.77 History of COPD 19.0% (11) 35.9% (28) 0.05 Peripheral vascular disease 27.6% (16) 16.7% (13) 0.19 Cerebral vascular disease 20.7% (12) 11.5% (9) 0.03 Myocardial infarction 31.0% (18) 5.1% (4) Congestive heart failure 87.9% (51) 79.6% (60) 0.15 Angina 65.5% (38) 20.5% (16) Previous aortic valvuloplasty 36.2% (21) 34.6% (27) 0.85 Aortic regurgitation: 0.40 None 4.4% (2/46) 14.3% (8/56) % (11/46) 23.2% (13/56) % (24/46) 44.6% (25/56) % (8/46) 17.9% (10/56) 4 2.2% (1/46) 0 Baseline NYHA functional class: 0.06 I 0 1.4% (1/73) II 12.7% (7/55) 6.9% (5/73) III 45.5% (25/55) 65.8% (48/73) IV 41.8% (23/55) 26.0% (19/73) Mitral regurgitation: 0.22 None 2.2% (1/46) 5.7% (3/53) % (6/46) 28.3% (15/53) % (20/46) 41.5% (22/53) % (17/46) 22.6% (12/53) 4 4.4% (2/46) 1.9% (1/53) Implanted valve size (mm): % (13) 48.7% (38) % (45) 51.3% (40) Continuous Variables Age Logistic EuroSCORE STS Predicted risk of mortality (27/58) (18/78) 0.01 Body surface area (Dubois) (m 2 ) Ejection fraction Aortic valve area (cm 2 ) Mean gradient (mm Hg) Peak aortic gradient (mm Hg) BMI (kg/m 2 ) BMI body mass index; CAD coronary artery disease; COPD chronic obstructive pulmonary disease; EuroSCORE European system for cardiac operative risk evaluation; NYHA New York Heart Association; STS Society of Thoracic Surgeons.

4 Ann Thorac Surg DEWEY ET AL 2010;89: EFFECT OF CAD ON OUTCOMES OF TAVI and transfemoral feasibility trials. In the 40 patient TA- AVI trial, five patients were excluded from analysis due to technical complications that resulted in embolization of the valve or malplacement such that conversion to an 761 open procedure was required. Table 1 characterizes the 35 remaining patients; 26 (74%) had a history of CAD, as indicated by previous PCI or CABG, and 9 (26%) without CAD. Significant differences were seen within the TA- ADULT CARDIAC Table 3. Descriptive Characteristics of All Transcatheter Patients With or Without CAD Categoric Variables CAD (n 84) No CAD (n 87) p Value Female 39.3% (33) 62.1% (54) Current smoker 2.4% (2) 1.2% (1) 0.15 Diabetes 29.8% (25) 26.4% (23) 0.62 Hyperlipidemia 71.4% (60) 37.9% (33) Chronic renal disease 35.7% (30) 8.1% (7) Hypertension 84.5% (71) 67.8% (59) 0.01 Cerebrovascular accident 11.9% (10) 13.8% (12) 0.71 History of COPD 20.2% (17) 37.9% (33) 0.04 Peripheral vascular disease 39.3% (33) 17.2% (15) Cerebral vascular disease 22.6% (19) 13.8% (12) 0.29 Myocardial infarction 31.0% (26) 5.8% (5) Congestive heart failure 88.1% (74) 78.2% (68) 0.20 Angina 66.7% (56) 20.7% (18) Previous aortic valvuloplasty 38.1% (32) 35.6% (31) 0.73 Aortic regurgitation: 0.41 None 9.7% (7/72) 16.9% (11/65) % (21/72) 21.5% (14/65) % (34/72) 43.1% (28/65) % (9/72) 18.5% (12/65) 4 1.4% (1/72) 0 Baseline NYHA functional class: I 0 1.2% (1/82) 0.08 II 8.6% (7/81) 6.1% (5/82) III 50.6% (41/81) 67.1% (55/82) IV 40.7% (33/81) 25.6% (21/82) Mitral regurgitation: 0.20 None 1.4% (1/71) 4.9% (3/61) % (13/71) 29.5% (18/61) % (28/71) 41.0% (25/61) % (25/71) 23.0% (14/61) 4 5.6% (4/71) 1.6% (1/61) Implant approach: Transapical 31.0% (26) 10.3% (9) Transfemoral 69.1% (58) 89.7% (78) Implanted valve size (mm): % (23) 46.0% (40) % (61) 54.0% (47) Continuous Variables Age Logistic EuroSCORE STS Predicted risk of mortality Body surface area (Dubois) (m 2 ) Ejection fraction Aortic valve area (cm 2 ) Mean gradient (mm Hg) Peak aortic gradient (mm Hg) Left ventricular systolic volume (ml 3 ) Left ventricular diastolic volume (ml 3 ) CAD coronary artery disease; COPD chronic obstructive pulmonary disease; EuroSCORE European system for cardiac operative risk evaluation; NYHA New York Heart Association; STS Society of Thoracic Surgeons.

5 ADULT CARDIAC 762 DEWEY ET AL Ann Thorac Surg EFFECT OF CAD ON OUTCOMES OF TAVI 2010;89: Table 4. Complications and Adverse Events Factor CAD No CAD p Value Mortality: Pooled patients 35.7% (30/84) 18.4% (16/87) 0.01 Transapical 42.1% (11/26) 22.2% (2/9) 0.43 Transfemoral 32.8% (19/58) 18.0% (14/78) Day Mortality: Pooled patients 13.1% (11/84) 1.2% (1/87) Transapical 19.2% (5/26) Transfemoral 10.3% (6/58) 1.3% (1/78) 0.04 Number of Adverse Events: % (39) 65.5% (57) % (31) 29.9% (26) % (14) 4.6% (4) CAD coronary artery disease AVI cohorts in terms of the distribution of patients with a history of hyperlipidemia, peripheral vascular disease, and angina. Additionally, patients undergoing TA-AVI with CAD had significantly higher logistic EuroSCOREs ( vs , p 0.03). There were no differences between the groups in terms of average age, ejection fraction, aortic valve area, and peak and mean gradients across the valve. Additionally, patients with CAD had similar degrees of mitral regurgitation and baseline New York Heart Association functional class as those without CAD. Twenty-five of the 161 TF-AVI patients were excluded from analysis for technical complications resulting in embolization of the valve or malplacement, as well as aborting the procedure prior to deploying the valve. Table 2 characterizes the remaining 136 patients, 58 (43%) had CAD, and 78 (57%) had no CAD. There were a number of significant differences between the TF-AVI patients who had CAD compared with those who did not, including a history of hyperlipidemia, chronic renal disease, hypertension, chronic obstructive pulmonary disease, previous myocardial infarction, and angina. Additionally, patients undergoing TF-AVI with CAD had significantly higher logistic EuroSCOREs ( vs , p 0.001), decreased average ejection fractions ( vs , p 0.04), and larger body surface areas (BSA) (1.8 m vs 1.7 m 2 0.2, p 0.03). There were no differences between the groups in terms of average age, aortic valve area, and peak and mean gradients across the valve. Additionally, patients with CAD had similar baseline New York Heart Association functional class as those without CAD, but a significantly greater incidence of grade III or grade IV mitral regurgitation. The patients having had successful TA and TF valve implantations were then combined to create the pooled cohort used for analysis. Table 3 characterizes this pooled group of 171 patients: 84 (49%) patients had CAD as indicated by previous PCI or CABG, and 87 (51%) patients no CAD. There were significantly more females in the no CAD group. Patients with CAD had significantly greater rates of hyperlipidemia, chronic renal disease, hypertension, chronic obstructive pulmonary disease, peripheral vascular disease, previous myocardial infarctions and symptoms of angina. Additionally, the pooled patients with CAD had significantly higher logistic EuroSCOREs ( vs , p 0.001), decreased average ejection fractions ( vs , p 0.05), and larger BSA (1.8 m vs 1.7 m 2 0.2, p 0.04). There were no differences between the groups in terms of average age, aortic valve area, and peak and mean gradients across the valve. Within the pooled group, patients with CAD had similar baseline New York Heart Association functional class as those without CAD, but a significantly greater incidence of grade III or grade IV mitral regurgitation (p 0.03), and larger left ventricular end-diastolic volumes (107.1 ml 42.0 vs 92.3 ml 36.2, p 0.05). There were also a significantly greater number of patients having TA-AVI in the pooled CAD cohort (26 of 84, 31%) versus the cohort that had no CAD (9 of 87, 10.3%), p Within the pooled group, 30-day mortality was 13.1% (11 of 84) in those patients having CAD as compared with those that did not (1.2% [1 of 87]), p (Table 4). Thirty-day mortality among those patients just having TF-AVI was also significantly higher with associated CAD; 10.3% (6 of 58) versus 1.3% (1 of 78), p Increased 30-day mortality was also observed in the TA-AVI patients with CAD, 19.2% (5 of 26) compared with those without CAD 0% (0/9), although statistical significance was not reached due to small numbers. Overall mortality (30-day plus follow-up deaths) was also significantly greater in the pooled and TF-AVI groups with CAD (35.7% [30 of 84] and 32.8% [19 of 58], respectively), compared with the pooled and TF cohorts without CAD (18.4% [16 of 87], p 0.01 and 18.0% [14 of 78], p 0.04) (Table 4). There were significantly more patients having two or more adverse events in the pooled CAD cohort (16.7% [14 of 84]) versus those not having CAD (4.6% [4 of 87]), p These adverse events were described as bleeding, emergent cardiac surgery, death, myocardial infarction, stroke, vascular injury, or valvular thrombosis (Table 5). Furthermore, when analyzing two adverse events that were noted to lead to death (ie, stroke or myocardial infarction), we see that the majority of the events still occurred in the CAD populations irrespective of surgical approach (Table 6). Table 5. Reported Adverse Events in Study Groups All Postoperative Adverse Events a CAD (n 84) No CAD (n 87) Patients with at least 1 adverse event 53.6% (45) 34.5% (30) Bleeding 8.3% (7) 3.4% (3) Emergent cardiac surgery 0 2.3% (2) Death 35.7% (30) 18.4% (16) Myocardial infarction 14.3% (12) 10.3% (9) Stroke 13.1% (11) 1.1% (1) Valvular thrombosis 0 1.1% (1) Vascular injury 4.8% (4) 2.3% (2) a Event rates are calculated as percent of all patients. CAD coronary artery disease.

6 Ann Thorac Surg DEWEY ET AL ;89: EFFECT OF CAD ON OUTCOMES OF TAVI Table 6. Patients Dying After Having an Adverse Event of Stroke or MI Adverse Events Stroke Then Death MI Then Death Approach Subgroup No. patients p value No. patients p value ADULT CARDIAC Transapical With CAD 7.7% (2/26) % (0/26) N/A No CAD 0% (0/9) 0% (0/9) Total 5.7% (2/35) 0% (0/35) Transfemoral With CAD 8.6% (5/58) % (2/58) 0.58 No CAD 1.3% (1/78) 1.3% (1/78) Total 4.4% (6/136) 2.2% (3/136) Transapical vs Transfemoral p 0.67 (stroke then death) p 1.00 (MI then death) CAD coronary artery disease; MI myocardial infarction. Logistic regression was used to test factors for their effect on patient mortality over the study period. In the pooled group, the presence of CAD was a significant predictor of over all mortality (30-day plus follow-up deaths) with an odds ratio of 20.3 (95% confidence interval [CI] 2.4 to 172.3; p 0.006) (Table 7). Proportional hazards analysis established that the risk of dying at any point in time was 2.3 times higher among the patients with CAD (95% CI 1.29 to 4.17). Coronary artery disease was also a univariate predictor for the TF-AVI patient cohort with an odds ratio of 2.3 (95% CI 1.00 to 5.09; p 0.050). Coronary artery disease was not statistically significant for the TA subgroup, likely due to the fact that the majority of the patients had CAD, and this group was comprised of a smaller number of patients with fewer deaths. Controlling for other covariates (age, gender, mitral regurgitation, ejection fraction, and BSA) did not change the effect significantly. The BSA alone was a significant factor as a univariate predictor in the TA patients where increasing size reduced operative risk with an odds ratio of 0.32 for every increase in BSA of 0.1 m 2 (p 0.003) (Table 7). Logistic regression was also used to test for factors that increased the 30-day mortality rate (Table 8). In the pooled patient group, the presence of CAD was found to be the most significant factor associated with perioperative mortality with an odds ratio of 10.1 (95% CI 2.1 to 174.8; p 0.009). The second most significant risk factor was male gender, and again increasing BSA had a protective effect. Controlling for other covariates, the procedural approach Table 7. Logistic Regression for Overall Mortality Pooled patients Variable Odds Ratio 95% CI p Value CAD (yes vs no) BSA (0.1 m 2 increase) TA BSA (0.1 m 2 increase) TF CAD (yes vs no) BSA body surface area; CAD coronary artery disease; CI confidence interval; TA transapical; TF transfemoral. (TA or TF) was not a significant factor in 30-day mortality. Looking at just the TF group, presence of CAD, gender, and increasing BSA were all significant risk factors. The Kaplan-Meier survival curve for the pooled cohort also demonstrates significantly reduced long-term survival for those patients with CAD (p 0.009) as compared with those without CAD (Fig 1). Survival curves for the TA-AVI patients did not demonstrate reduced long-term survival (p 0.32), likely secondary to low numbers and that the majority of the patients had CAD (Fig 2). Survival curves for the TF-AVI strongly trended toward reduced survival in groups having CAD as well (p 0.064) (Fig 3). The hemodynamic results of transcatheter valve implantation, either TA or TF, were excellent and durable in both the pooled group with CAD and in the group without CAD. Significant increases in aortic valve area and average ejection fraction, as well as significant decreases in peak gradients were demonstrated on follow-up (Figs 4; 5; 6). Comment It is axiomatic that aortic valve replacement is the treatment of choice for patients with critical aortic stenosis. However, certain patient populations have been excluded from conventional surgical therapy due to perceived elevated operative risk [12, 13]. Transcatheter Table 8. Logistic Regression for 30-Day Mortality Pooled patients TA Parameter Odds Ratio 95% CI p Value CAD (yes vs no) BSA (0.1 m 2 increase) Sex (male vs female) No significant variables TF CAD (Yes vs. No) BSA (0.1 m 2 increase) Sex (male vs female) BSA body surface area; CAD coronary artery disease; CI confidence interval; TA transapical; TF transfemoral.

7 ADULT CARDIAC 764 DEWEY ET AL Ann Thorac Surg EFFECT OF CAD ON OUTCOMES OF TAVI 2010;89: Fig 1. Kaplan-Meier survival curve for pooled patient data, stratified by the presence or absence of CAD. (CAD coronary artery disease; censored; log-rank, p ) Fig 2. Kaplan-Meier survival curve for TA patients, stratified by the presence or absence of CAD. (CAD coronary artery disease; TA transapical; censored; log-rank, p ) Fig 3. Kaplan-Meier survival curve for TF patients, stratified by the presence or absence of CAD. (CAD coronary artery disease; TF transfemoral; censored; log-rank, p )

8 Ann Thorac Surg DEWEY ET AL 2010;89: EFFECT OF CAD ON OUTCOMES OF TAVI Fig 4. Change in aortic valve area (AVA) after implantation. (CAD coronary artery disease; Πno; yes; h hours; d days; m months.) 765 ADULT CARDIAC Fig 5. Change in ejection fraction (EF) after implantation. (CAD coronary artery disease; Πno; yes; h hours; d days; m months.) Fig 6. Change in peak gradient after implantation. (CAD coronary artery disease; Πno; yes; h hours; d days; m months.)

9 ADULT CARDIAC 766 DEWEY ET AL Ann Thorac Surg EFFECT OF CAD ON OUTCOMES OF TAVI 2010;89: aortic valve implantation, either retrograde through a transfemoral approach or antegrade through a transapical approach, has become a clinical reality in the treatment of critical aortic stenosis in high-risk patients. Early clinical results are positive, and have generated intense interest in this technology, along with intense scrutiny. Experience has shown that proper patient selection is the leitmotif with any new technology. Previous studies have demonstrated that CAD and (or) previous CABG are independent markers for increased mortality in patients undergoing aortic valve replacement [14 16]. However, delineation of risk and patient selection for transcatheter therapy is still a work in progress. The technical complications identified in this study were primarily related to the inherent learning curve of these procedures. In no cases were device-related complications or malfunctions identified. Multiple strategies designed to mitigate against procedural failure were implemented, including having surgeons and cardiologists work in tandem on both the transfemoral and transapical procedures, performing the procedures in either catheterization labs or hybrid operating rooms when possible, and using preprocedure checklists to ensure appropriate disaster planning. Interestingly, some adjunctive imaging techniques such as intracardiac ultrasound and three-dimensional transesophageal imaging were not found to be particularly helpful in performing the procedures or preventing complications. And while technical misadventures in sporadically performed procedures can overshadow the effect of risk factors on outcomes, eventually learning curves are surmounted and choosing the appropriate patients becomes paramount. This study demonstrates that CAD, as defined by previous CABG or PCI, is a significant risk factor for mortality in patients having transcatheter aortic valve implantation. Patients with procedural mishaps were excluded from analysis to control for learning curve. This in effect deemphasized the technical aspects of the procedure, and highlighted the risk variables associated with mortality. The patients with CAD clearly constitute a higher-risk cohort as demonstrated by the higher logistic EuroSCOREs, lower ejection fractions, larger left ventricular enddiastolic volumes, and more significant mitral regurgitation (Table 3). It is compelling to note that every 30-day death except one out of 171 patients was in those patients who had CAD. In fact, in the TA group there were no 30-day deaths in patients who did not have CAD. This substantiates the significance of CAD in this population in regard to procedural outcomes with either the TA or TF approach, and illustrates the peril of attempting to compare outcomes between groups with different numbers of patients with this risk factor. As a case in point, the 30-day mortality in the TA group was 14.3% (5 of 35) compared with 5.1% (6 of 136) in the TF cohort, but 74% of the TA patients had CAD versus only 43% of the TF patients. Furthermore, the logistic regression analysis demonstrated that surgical approach, either TA or TF, was not a risk factor for either 30-day or overall mortality. Patients with CAD also had increased risk of stroke, particularly with the transfemoral procedure. Possible explanations for increased stroke in patients with CAD include an increased atheromatous burden which becomes at risk with the passage of the transfemoral catheter around the aortic arch, or manipulation of guide wires in the ascending aorta or arch with both approaches. The need for embolic protection devices has not been convincingly demonstrated at this point but remains an area of research. The predominance of CAD, the most significant risk factor for mortality, clearly distinguishes the TA patients as a higher risk group and discourages outcome comparisons between these two approaches. There was no indication in the analysis that either CABG or PCI, individually, demonstrated more significance than the other in regard to procedural mortality. In fact, the Kaplan-Meier survival curve for the pooled patients was only significant when CABG and PCI were compared together, and not individually, to the group without CAD. Furthermore, the pooled patients with CAD were evenly distributed in regard to previous coronary interventions with 48 having undergone previous CABG, 49 previous PCI, and 13 of these two groups both interventions. Patients with CAD remain good candidates for transcatheter therapy, but anticipated results should be viewed with recognition of the influence of this risk factor on procedural outcomes. Given that transcatheter AVR is a stand-alone procedure, it is unknown how many of these patients would have received concomitant bypass grafting if undergoing conventional valve replacement. However, it cannot be overemphasized that transcatheter aortic valve implantation is not indicated for those patients with significant ischemic myocardium without prior revascularization. Outcomes in patients with concomitant CAD also raises the question of whether predicted mortality by the Society of Thoracic Surgeons risk algorithm for patients with CAD should be calculated using the formula for AVR plus CABG rather than one for isolated AVR as is currently done. Using the AVR plus CABG algorithm might provide a more realistic expectation of anticipated mortality in this higher risk population given the differences between the calculated coefficients between the two formulas. Hemodynamic improvement, in regard to increased aortic valve area, improved ejection fraction, and decreased peak gradients, was significantly better than baseline in all patients having transcatheter aortic valve implantation. Whether this translates into increased life expectancy likely will require further studies in lower risk populations. In summary, patients with CAD as indicated by previous CABG or PCI had significantly higher 30-day and overall mortality with transcatheter valve implantation. Choice of operative approach, either TA or TF, was not found to be a risk factor for mortality. Comparison of procedural outcomes, based on operative approach, without controlling for unequal distribution of CAD in the cohorts are likely invalid. Excellent hemodynamic outcomes that appear to be durable can be achieved with this technology. References 1. Cribier A, Eltchaninoff H, Bash A, et al. Percutaneous transcatheter implantation of an aortic valve prosthesis for calcific aortic stenosis: first human case description. Circulation 2002;106:

10 Ann Thorac Surg DEWEY ET AL 2010;89: EFFECT OF CAD ON OUTCOMES OF TAVI 2. Lichtenstein SV, Cheung A, Ye J, et al. Transapical transcatheter aortic valve implantation in humans: Initial clinical experience. Circulation 2006;114: Webb JG, Pasupati S, Humphries K, et al. Percutaneous transarterial aortic valve replacement in selected high-risk patients with aortic stenosis. Circulation 2007;116: Walter T, Simon P, Dewey TM, et al. Transapical minimally invasive aortic valve implantation: multicenter experience. Circulation 2007;116(11 suppl): Webb JG, Chandavimol M, Thompson CR, et al. Percuaneous aortic valve implantation retrograde from the femoral artery. Circulation 2006;113: Walther T, Falk V, Kempfert J, et al. Transapical minimally invasive aortic valve implantation: The initial 50 patients. Eur J Cardiothorac Surg 2008;33: Walther T, Dewey T, Borger M, et al. Transapical aortic valve implantation: step by step. Ann Thorac Surg 2009;87: Grube E, Schuler G, Buellesfeld L, et al. Percutaneous aortic valve replacement for severe aortic stenosis in high-risk patients using second and current third generation selfexpanding CoreValve prosthesis. J Am Coll Cardiol 2007;50: Dewey TM, Brown D, Ryan WH, et al. Reliability of risk algorithms in predicting early and late operative outcomes in high-risk patients undergoing aortic valve replacement. J Thorac Cardiovasc Surg 2008;135: Svensson LG, Dewey T, Kapadia S, et al. United States feasibility study of transcatheter insertion of a stented aortic valve by the left ventricular apex. Ann Thorac Surg 2008;86: Leon MB, Kodali S, Williams M, et al. Transcatheter aortic valve replacement in patients with critical aortic stenosis: rationale, device description, early clinical experiences, and perspectives. Semin Thorac Cardiovasc Surg 2006;18: Varadarajan P, Kapoor N, Bansal RC, Pai RG. Clinical profile and natural history of 435 nonsurgically managed patients with severe aortic stenosis. Ann Thorac Surg 2006;82: Lung B, Baron G, Butchart EG, et al. A prospective survey of patients with valvular heart disease in Europe: The Euro Heart Survey on Valvular Heart Disease. Eur Heart J 2003; 24: Nowicki ER, Birkmeyer NJ, Weintraub RW, et al. Multivariate prediction of in-hospital mortality associated with aortic and mitral valve surgery in northern New England. Ann Thorac Surg 2004;77: Tjang YS, Hees YD, Körfer R, Grobbee DE, van der Heijden GJ. Predictors of mortality after aortic valve replacement. Eur J Cardiothorac Surg 2007;32: Florath I, Rosendahl UP, Mortasawi A, et al. Current determinants of operative mortality in 1400 patients requiring aortic valve replacement. Ann Thorac Surg 2003;76: ADULT CARDIAC DISCUSSION DR VALAVANUR SUBRAMANIAN (New York, NY): I am rather perplexed by the results. The concept of transapical catheter-based AVR [aortic valve replacement] being advantageous in a high-risk aortic valve patient is based on the fact that it can be done off-pump with no aortic cross-clamping and global myocardial ischemia. So tell me why you have an adverse event in the presence of a CAD [coronary artery disease] in this group, especially when you do not have global myocardial ischemia, no cardiopulmonary bypass, no cross-clamping. Is something else happening in this group of patients? Now, I asked this question many times prior to this meeting. Something seems to trigger these patients to have a rapid downfall after the transapical AVR. You can argue that these patients have a high-risk EuroSCORE, and they are very sick patients. But if you take sick patients with no aortic stenosis and the natural history of those patients with comorbidities like renal failure, COPD [chronic obstructive pulmonary disease], hypertension, they don t die within six months after Dx [diagnosis]. They have a slow progressive mortality rate over many years. So, tell me why they die because of their concomitant CAD soon after the procedure. DR DEWEY: Well, the easy answer to that is they are sicker. If you look at the logistic EuroSCOREs, they are significantly higher than those without CAD. And I think you have to think about the fact that CAD in these patients is just a marker for more diffuse atherosclerotic disease, peripheral vascular disease, renal disease, cerebrovascular disease. And so these patients are at the farthest end of this spectrum and have even less reserve. Short of that answer, there is almost no way to know why you can do a perfect 30-minute transapical implantation and have the patient still not survive. That is probably the best answer I can give you at this point. DR VINOD H. THOURANI (Atlanta, GA): In your patient population you have included both patients with previous coronary bypass and PCI [percutaneous coronary intervention]. Can you differentiate roughly what percentage had a previous coronary versus a PCI, and, consequently, for those patients who had a transapical, that would become a redo operation for those patients, and do you think that makes a difference? DR DEWEY: It is interesting you should ask me that, Vinod, because the patients were evenly distributed between previous bypass surgery, previous PCI, there was 46 and 47%, and 13% had had both. And we looked at actual survival curves just based on previous PCI or previous CABG [coronary artery bypass surgery] in and of themselves, and each one of those in isolation was not significant. It was the combination of the two. Now, it may just be because the numbers are still rather small and with a larger population each one of those will become significant. But it was the combination of the two that became significant. Given the fact that many more of the transapicals had coronary disease, they were more likely to have had previous bypass grafting, and so they are redo operations. The other thing to keep in mind is transapical is a stand-alone procedure. And so various degrees of coexisting coronary disease tend to be tolerated in these procedures that you might otherwise be obliged to bypass if you were going to do an open procedure. DR THOURANI: And did you see an increased perioperative myocardial infarction rate between the different groups? DR DEWEY: We actually did not. DR MICHAEL A. ACKER (Philadelphia, PA): Todd, when you looked at those patients using the STS [Society of Thoracic Surgeons] preoperative risk score, were they still sicker? Was that as good to differentiate the two? DR DEWEY: In this population with the bulk of the patients actually being done with Europe, we didn t have access to the STS in those patients. But I would imagine that all the other studies that have been performed, the patients with the higher STS scores and the higher logistic EuroSCOREs all do worse. Their survival is the worst of the bunch. DR ACKER: Does the STS, though, take into consideration coronary artery disease as part of their risk stratification? DR DEWEY: It takes into consideration left main disease and number of diseased vessels. So it is a calculated covariate in the risk model.