Thirty-Year Follow-Up Defines Survival Benefit for Second Internal Mammary Artery in Propensity-Matched Groups

Transcription

1 Thirty-Year Follow-Up Defines Survival Benefit for Second Internal Mammary Artery in Propensity-Matched Groups Paul A. Kurlansky, MD, Ernest A. Traad, MD, Malcolm J. Dorman, MD, David L. Galbut, MD, Melinda Zucker, BSN, and George Ebra, EdD Florida Heart Research Institute, Miami, and Division of Cardiovascular Surgery, JFK Medical Center, Atlantis, Florida Background. The value of the left internal mammary artery (LIMA) graft is well established. However, the incremental value of a second IMA graft is controversial. Despite reports of improved survival with bilateral IMA (BIMA) grafting, the Society of Thoracic Surgeons reports its use in 4% of coronary artery bypass graft operations. We report the influence of BIMA vs SIMA grafting on hospital and late mortality in comparable groups. Methods. Retrospective review was conducted of 4584 consecutive isolated coronary artery bypass graft operations (2369 SIMA and 2215 BIMA) performed from 1972 to The influence of the second IMA was assessed by multivariate analyses of risk factors associated with hospital and late mortality and by propensity score analysis that compares patients with similar baseline characteristics for receiving a second IMA graft. All patients were monitored clinically to assess outcomes. Results. Hospital mortality was 4.5% for SIMA vs 2.6% for BIMA patients (p 0.001). When stratified by propensity score to undergo BIMA grafting, no difference in hospital mortality was found. Multivariate analyses showed SIMA grafting was significantly associated with late but not hospital mortality. Survival curves after 52,572 patient-years of follow-up (mean, 11.5 years; range, 6 weeks to 32 years) demonstrated improved long-term survival for BIMA vs SIMA patients in all quintiles except those with the greatest propensity for SIMA, wherein late survival was comparable between groups. In matched groups, survival favored BIMA patients (p 0.001). Conclusions. BIMA grafting offers a long-term survival advantage over SIMA grafting in propensity-matched groups. (Ann Thorac Surg 2010;90:101 8) 2010 by The Society of Thoracic Surgeons Cardiac surgery is well into its fourth decade of experience with the use of bilateral internal mammary artery (BIMA) grafting for the treatment of ischemic cardiovascular disease. Numerous compelling retrospective studies have documented a clear benefit for BIMA grafting over single IMA (SIMA) grafting in reducing the long-term risk of death [1 3], cardiac death [1, 4] and late cardiac events [1, 4, 5]. Although not all reports have been favorable [6], two independent meta-analyses have corroborated a long-term benefit for BIMA grafting [7, 8]. A multicenter, prospective, randomized trial of SIMA vs BIMA grafting is currently in progress, with results anticipated in 2018 that should help to define the course of IMA grafting in myocardial revascularization [9]. Despite the wealth of knowledge available, the Society of Thoracic Surgeons (STS) National Adult Cardiac Surgery Database reports that only 4% of coronary artery bypass operations (CABG) involve the use of BIMA grafting [10]. Accepted for publication April 2, Presented at the Forty-sixth Annual Meeting of The Society of Thoracic Surgeons, Fort Lauderdale, FL, Jan 25 27, Address correspondence to Dr Kurlansky, Florida Heart Research Institute, 4770 Biscayne Blvd, Ste 500, Miami, FL 33137; drwu18@ aol.com. Reports promoting the benefits of IMA grafting appear to arise from cohorts where nearly all patients had SIMA grafting. We report clinical outcomes in a 30-year experience with IMA grafting in a community hospital in a large cohort where nearly half the patients received BIMA grafting. Patients and Methods This study was presented to the Institutional Review Board, and waiver of requirement of informed patient consent was granted. Patients Between February 1972 and May 1994, 4584 consecutive patients underwent coronary revascularization, of which 2369 received a SIMA and 2215 a BIMA graft. Excluded from the study were 329 patients with concomitant procedures, those with only one distal anastomosis, and those with only saphenous vein grafts (6.7% of the total isolated multigraft coronary revascularization patients during the study period). The coronary and perioperative risk factors, and angiographic findings for the two cohorts are summarized in Table by The Society of Thoracic Surgeons /$36.00 Published by Elsevier Inc doi: /j.athoracsur

2 102 KURLANSKY ET AL Ann Thorac Surg IMA GRAFTING 2010;90:101 8 Table 1. Comparison of Preoperative Clinical Variables and Risk Factors by Patient Group Variables a SIMA BIMA p Value Patients 2369 (100.0) 2215 (100.0)... Gender Male 1761 (74.3) 1886 (85.1) Female 608 (25.7) 329 (14.9) Age, mean SD, y Age groups, y (4.6) 245 (11.0) (14.3) 487 (22.0) (35.3) 859 (38.8) (38.8) 588 (26.6) (7.0) 36 (1.6) Coronary risk factors Family history of CAD 1197 (50.5) 1203 (54.3) Hypertension 873 (36.9) 813 (36.7) Dyslipidemia 350 (14.8) 348 (15.7) Smoking history 1360 (57.4) 1384 (62.5) Diabetes mellitus 646 (27.3) 461 (20.8) Perioperative risk factors Renal dysfunction 106 (4.5) 64 (2.9) Cerebral vascular 143 (6.0) 76 (3.4) disease Peripheral artery 138 (5.8) 94 (4.2) disease Prior myocardial 1369 (57.8) 1172 (52.9) infarction History of CHF 344 (14.5) 189 (8.5) Unstable angina 1636 (69.1) 1314 (59.3) Coronary angiography 3-vessel disease 1951 (82.4) 1850 (83.5) vessel disease 364 (15.4) 340 (15.3)... 1-vessel disease 54 (2.3) 25 (1.1)... Left main disease ( 0.50) 409 (17.3) 431 (19.5) Ejection fraction (64.9) 1510 (68.2) (27.7) 600 (27.1) (6.2) 87 (3.9)... Not done 29 (1.3) 18 (0.8)... a Categoric data are presented as number (%). BIMA bilateral internal mammary artery; CAD coronary artery disease; CHF congestive heart failure; SD standard deviation; SIMA single internal mammary artery. In the SIMA group, 59 patients (2.5%) were in Canadian Cardiovascular System (CCS) class 1, 100 (4.2%) were in class 2, 1045 (44.1%) were in class 3, and 1165 (49.2%) were in class 4. In the BIMA group, 87 patients (3.0%) were in class 1, 144 (6.5%) were in class 2, 1107 (50.0%) were in class 3, and 877 (39.6%) were in class 4. A significantly greater number of SIMA patients had class 3 or 4 symptoms (93.3% vs 89.6%; p 0.001). Clinical Management The operation was performed in the SIMA group electively in 1020 patients (43.1%), urgently in 1132 (47.8%), emergently in 188 (7.9%), and was salvage in 29 (1.2%). In the BIMA group, the operation was performed electively in 1134 (51.2%), urgently in 1013 (45.7%), emergently in 66 (3.0%), and was salvage in 2 (0.1%). A between-group comparison of elective vs nonelective urgency revealed a significantly greater proportion of SIMA operations were not elective (p 0.001). Operations in 179 SIMA patients (7.6%) and 168 BIMA patients (7.6%) were repeat operations. Details of the operative techniques used in the present series, including internal mammary artery (IMA) mobilization, orientation, and reconstruction in BIMA grafting [11], have been previously discussed. The IMA is dissected as an isolated pedicle from the chest wall, free from surrounding muscle and fascia. The vein is initially dissected but subsequently removed to allow maximal length and versatility. All side branches are cauterized carefully or clipped as necessary. Since 1989, combined antegrade and retrograde infusion methods of cardioplegia were implemented to enhance myocardial protection during the operation. Cardiopulmonary bypass was used in all operations. A total of 7432 distal coronary artery grafts were performed (mean, 3.1 per patient; range, 2 to 6) in the SIMA group. In the BIMA group, 7359 coronary artery grafts were performed (mean, 3.3; range, 2 to 6). There were 1248 sequential grafts in the SIMA group and 981 in the BIMA group. The mean number of grafts performed in the BIMA group was significantly higher than in the SIMA group (p 0.001). In situ grafting was performed in 98.9% of SIMA patients and in 97.1% of BIMA patients. The mean cardiopulmonary bypass time was minutes (range, 15 to 523 minutes) for the SIMA group and minutes (range, 25 to 444 minutes) for the BIMA group. The mean duration of aortic crossclamping was minutes (range, 10 to 237 minutes) for the SIMA group and minutes (range, 10 to 200 minutes) for the BIMA group (p 0.001). Data Collection Perioperative data were obtained by prospective review of hospital records, catheterization reports, cine angiograms, and echocardiography. Follow-up information was obtained through comprehensive questionnaires and by telephone interview with surviving patients, family members, or the patient s personal physician. All patients were monitored clinically, and data were recorded in a standardized manner using the guidelines of the Society of Thoracic Surgeons Adult Cardiac Surgery Database. A 98.3% follow-up was obtained in the SIMA group, with 40 patients lost to follow-up after discharge from the hospital. In the BIMA group, a 96.7% follow-up was achieved, with 72 patients lost to follow-up. Statistical Analysis Demographic and clinical data are presented as frequency distributions and simple percentages. Values of continuous variables are expressed as mean standard deviation. Univariate analysis of selected

3 Ann Thorac Surg KURLANSKY ET AL 2010;90:101 8 IMA GRAFTING 103 preoperative and postoperative discrete variables was accomplished by 2, the continuity-adjusted 2 analysis, or a two-tailed Fisher s exact test with the appropriate degrees of freedom to test for the equality of proportions in the case of categoric variables. Twosample t tests (two-tailed) were used to test for the equality of the means of continuous variables. To identify predictors of hospital mortality, a hierarchic logistic regression model was developed using preoperative and intraoperative variables. Propensity score technology was used to calculate the probability of each patient receiving BIMA grafting based on 22 preoperative and intraoperative clinical variables. This process generated a propensity score between 0 and 1, which can be used to summarize a collection of covariates, integrates the likelihood of a patient receiving BIMA grafting, and then stratifies the patients into quintiles [12]. In addition, BIMA patients were matched to SIMA patients using the Rosenbaum optimal matching algorithm [13]. This approach minimizes the overall distance between observations and was conducted using Mahalanobis distance within propensity score calipers (no matches outside the calipers). To identify predictors of late mortality, a Cox proportional hazards regression model was used to discern the influence of multiple clinical variables on late survival. Regression coefficients and odds ratios with 95% confidence intervals were calculated to determine the relative influence of each covariate on the survivor function. Coefficients were computed by the method of maximum likelihood. Lists of all variables used in the multivariate analyses are available from the corresponding author. Actuarial analysis was conducted according to the method of Kaplan and Meier. The equality of survival distribution was tested with the log-rank algorithm. Data collected were analyzed using Number Cruncher Statistical Systems software (NCSS, Kaysville, UT). A significant difference between measurements was defined as p Results Hospital Morbidity The overall incidence of postoperative morbidity for the two groups was low. No complications were experienced in 1880 of the SIMA group (79.4%) or in 1864 of the BIMA group (84.2% p 0.001). A between-group comparison (SIMA vs BIMA) of each of the hospital complications revealed significant differences in renal dysfunction (3.9% vs 2.2%; p 0.001), reoperation for bleeding (3.2% vs 1.8%; p 0.003), pulmonary insufficiency (7.0% vs 4.9%; p 0.003), cerebrovascular accident (2.0% vs 1.1%; p 0.020), and low cardiac output (3.0% vs 2.1%; p 0.040). There was no significant difference in cardiac arrest (3.7% vs 2.8%; p 0.115), perioperative myocardial infarction (5.7% vs 4.8%; p 0.166), gastrointestinal disorder (2.0% vs 2.0%; p 0.999), or deep sternal infection (1.1% vs 1.4%; p 0.289). A between-group comparison revealed that the prevalence of deep sternal wound infection was unaffected by the presence of diabetes mellitus. Among patients with diabetes, the incidence of sternal wound infection was 1.5% for SIMA and 2.8% for BIMA patients (p 0.144). Moreover, a within-group comparison of SIMA patients indicated that diabetes mellitus did not influence the occurrence of deep sternal wound infection (diabetes mellitus, 1.5%; no diabetes, 0.9%; p 0.151). In the BIMA group, the presence of diabetes mellitus did affect the occurrence of deep sternal wound infection (1.0% vs 2.8%; p 0.004). The average postoperative length of stay was days for BIMA patients and days for SIMA patients (p 0.001). Hospital Mortality The hospital mortality rate was 4.6% (108 of 2369) for SIMA patients and 2.6% (58 of 2215) for BIMA patients (p 0.001). The overall hospital mortality rate for the series was 3.6% (166 of 4584). A forward stepwise logistic regression model using 23 preoperative and intraoperative variables identified five independent correlates of hospital mortality (Table 2): age (p 0.001), number of distal grafts (p 0.044), intraaortic balloon pump inserted preoperatively (p 0.001), perfusion time (p 0.001), and surgical urgency (p 0.041). The use or nonuse of BIMA grafting was not a predictor of hospital mortality. Table 3 provides a comparison of the hospital mortality rate for the two patient cohorts after stratification of the total study population into five quintiles using propensity score analysis. No significant difference in the hos- Table 2. Multivariate Analysis of Variables Associated With Hospital Mortality in Patients Undergoing Single Internal Mammary Artery and Bilateral Mammary Artery Coronary Artery Bypass Grafting Predictor Estimate SE 2 OR (95% CI) p Value a Age ( ) Distal grafts ( ) Intraaortic balloon pump ( ) Perfusion time ( ) Surgical urgency ( ) a Only significant variables (p 0.050) are listed. CI confidence interval; OR odds ratio; SE standard error.

4 104 KURLANSKY ET AL Ann Thorac Surg IMA GRAFTING 2010;90:101 8 Table 3. Comparison of Hospital Mortality by Quintiles by Patient Group Quintile SIMA No. (%) BIMA No. (%) p-value Quintile 1 2/285 (0.7) 4/622 (0.6) Quintile 2 7/373 (1.9) 5/535 (0.9) Quintile 3 16/465 (3.4) 15/428 (3.4) Quintile 4 29/536 (5.4) 16/372 (4.3) Quintile 5 53/681 (7.8) 17/225 (7.6) BIMA bilateral internal mammary artery; mammary artery. SIMA single internal pital mortality rate was found comparing the two groups within each quintile. Long-Term Follow-Up The average duration of follow-up for hospital survivors was 11.1 years (range, 6 weeks to 31.6 years) in the SIMA group and 12.7 years (range, 6 weeks to 32.1 years) in the BIMA group. The cumulative follow-up was 25,207 patient-years for the SIMA group and 27,365 patient-years for the BIMA group. Cox regression analysis, which was conducted using 29 preoperative, intraoperative, and postoperative variables, demonstrated the independence of 18 covariates associated with late mortality: 14 preoperative, 2 intraoperative, and 2 postoperative variables (Table 4). SIMA vs BIMA grafting was an independent predictor of late death (p 0.001). The actuarial survival data for SIMA and BIMA patients are shown in Figure 1. At 15 years, survival for SIMA and BIMA patients the standard error of the mean (SEM) was 37.5% 1.1% and 53.5% 1.2%; at 25 years, it was 15.7% 2.0% for SIMA patients and 28.6% 2.2% for BIMA patients (p 0.001). Although true comparisons could not be made due to the dissimilarity in the two groups, these results are indicative of an enhanced survival pattern in BIMA patients. To analyze comparable groups, actuarial survival was completed for the two groups within the five quintiles (Table 5). In all quintiles, BIMA patients experienced a survival advantage. In quintiles 1, 2, and 3 the difference in the two groups was statistically significant, and in quintile 4, the difference approached statistical significance. To further correct for differences in baseline characteristics between groups, a comparison was conducted between 2197 BIMA patients who were optimally matched with 2197 SIMA patients. Survival was 39.0% 1.1% for SIMA patients and 53.5% 1.2% for BIMA patients at 15 years and 16.5% 2.1% for SIMA and 28.5% 2.2% for BIMA patients at 25 years (p 0.001; Fig 2). In these matched groups, median survival for SIMA patients was 11.8 years (95% confidence interval, 11.3 to 12.3) compared with 15.9 years (95% confidence interval, 15.2 to 16.7) for the BIMA patients. Table 4. Variables Influencing Late Mortality by Cox Regression Analysis in Single Internal Mammary Artery and Bilateral Mammary Artery Patients Predictor Regression Coefficient SE HR (95% CI) p Value a Preoperative Age ( ) Angina-stable ( ) Cardiac arrest ( ) Congestive heart failure ( ) Cerebrovascular disease ( ) Diabetes mellitus ( ) Dyslipidemia ( ) Ejection fraction ( ) Female gender ( ) Left main disease ( ) Prior myocardial infarction ( ) Pulmonary insufficiency ( ) Peripheral vascular disease ( ) Renal disease b ( ) Intraoperative BIMA used ( ) Perfusion time ( ) Postoperative Renal insufficiency b ( ) Myocardial infarction ( ) a Only significant variables (p 0.050) are listed. b Noted as creatinine 2.0 mg/dl. BIMA bilateral internal mammary artery; CI confidence interval; HR hazard ratio; SE standard error.

5 Ann Thorac Surg KURLANSKY ET AL 2010;90:101 8 IMA GRAFTING 105 Fig 1. Actuarial survival of patients who underwent coronary artery bypass grafting with single internal mammary artery grafting ( SIMA) and bilateral internal mammary artery grafting (ΠBIMA). Number of patients at risk is in parentheses; results are mean standard error of the mean. Fig 2. Actuarial survival of optimally matched patients who underwent coronary artery bypass grafting with single internal mammary artery grafting ( SIMA) and bilateral internal mammary artery grafting (ΠBIMA). Number of patients at risk is in parentheses; results are mean standard error of the mean. There were 758 SIMA patients and 1110 BIMA patients alive at follow-up, of which 99.3% in the SIMA group and 97.8% in the BIMA group were in CCS class I or II (p 0.011). The occurrence rate of major adverse late cardiac events in SIMA vs BIMA survivors was nonfatal myocardial infarction, 3.0% vs 2.7%; reoperation, 2.7% vs 1.8%; percutaneous coronary intervention, 12.1% vs 10.3%; permanent stroke, 2.7% vs 3.0%, and composite freedom from major adverse late cardiac events was 84.0% vs 85.6%. None of these comparisons achieved statistical significance. Because the quintile most closely associated with the decision to perform SIMA vs BIMA grafting demonstrated the least survival benefit from BIMA grafting, it was thought that those variables most closely associated with that decision would be least likely to demonstrate a benefit for BIMA grafting. Actuarial survival was examined comparing SIMA with BIMA patients among each of the groups which defined a greater propensity for SIMA grafting. Equality of survival distribution favored BIMA grafting among women (p 0.001), the elderly (age 75 years; p 0.015), patients with diabetes (p 0.001), patients with impaired ventricular function (ejection fraction 0.50; p 0.001), nonelective surgical urgency (p 0.001), patients with intraaortic balloon pump inserted preoperatively (p 0.003), patients with unstable angina (p 0.001), patients with a history of congestive heart failure (p 0.010), and patients with preoperative renal dysfunction (p 0.003). Among these variables, we were unable to identify any that independently failed to denote a significant long-term survival benefit for BIMA grafting. Comment Controversy persists regarding the long-term benefits of BIMA grafting. Despite a number of studies, there is a wide discrepancy between findings in the literature and Table 5. Actuarial Survival by Quintiles by Patient Group Quintile/Group 10 Years Patients At Risk 20 Years Patients At Risk Mean SEM, % No. Mean SEM, % No. p Value Quintile 1 SIMA BIMA Quintile 2 SIMA BIMA Quintile 3 SIMA BIMA Quintile 4 SIMA BIMA Quintile 5 SIMA BIMA BIMA bilateral internal mammary artery; SEM standard error of the mean; SIMA single internal mammary artery.

6 106 KURLANSKY ET AL Ann Thorac Surg IMA GRAFTING 2010;90:101 8 actual clinical behavior. The current study provides valuable information concerning outcomes in broadly applied BIMA grafting that can be used in informed clinical decision making. One of the leading impediments to the widespread use of both IMAs has been concern about the risk of perioperative complications. In the present study, patients with two IMAs experienced an overall reduced rate of postoperative morbidity than SIMA patients (p 0.001). The most common concern was sternal wound complications, especially in patients with diabetes [14, 15]. Bleeding and pulmonary morbidity have also been reported [16, 17]. BIMA patients in the present study demonstrated no higher incidence of sternal wound infection than SIMA patients, regardless of the presence of diabetes. Neither were there increased bleeding nor pulmonary complications. The skeletonized approach to IMA harvesting, with its preservation of sternal blood supply and lymphatic draining [18], may account for the amelioration of sternal and other complications reported here and elsewhere [19 21]. Although complications were not assessed in matched groups, BIMA grafting by itself appears to have no increased risk of serious complications. The hospital mortality rate was greater in SIMA than BIMA patients but similar in propensity-matched groups within quintiles. Moreover, BIMA grafting was not identified as an independent predictor of hospital death in multivariate analysis. This finding corroborates that of others [2] and assists in addressing concerns regarding the adequacy of IMA flow in the acute postoperative period [22]. The maintenance of adequate hemodynamics during the postoperative period is critical in patients relying on two IMA grafts. The present study, however, found no evidence of increased perioperative hemodynamic instability. Although patient selection favored more stable patients in the BIMA group, propensitymatched patients performed equally well in the two cohorts. The most striking finding in this study is the long-term survival benefit, which was widespread despite the broad spectrum of patients who underwent BIMA grafting. Because there were more grafts per patient in BIMA than SIMA patients, it could be hypothesized that the reason for improved survival was more complete revascularization rather than the use of a second IMA. Even without an adequate metric to assess the completeness of revascularization, BIMA grafting, and not the number of grafts per patient, was an inverse predictor of late death. Among propensity-matched groups, the survival benefit of BIMA patients was apparent not only in the two quintiles mostly likely to receive two IMA grafts but also in the cohort with equivalent probability of receiving either graft. Even in patients with an intermediate probability of receiving a SIMA graft (quintile 4), the survival benefit of BIMA grafting approached statistical significance (p 0.059). When the two cohorts were compared in optimally matched groups, a clear survival benefit was seen for patients with two IMA grafts, which resulted in a 34% prolongation of median survival. It might be anticipated that those risk factors most closely associated with the decision to perform one rather than two IMA grafts would assist in defining the groups least likely to receive long-term survival benefits from BIMA grafting. However, actuarial survival was enhanced with BIMA grafting in women, the elderly ( 75 years), patients with diabetes, prior myocardial infarction, nonelective surgical urgency, impaired ventricular function (ejection fraction 0.50), unstable angina, history of congestive heart failure, preoperative renal dysfunction, and intraaortic balloon pump inserted preoperatively. Propensity score analysis has inherent limitations that do not allow it to factor out all of the complex interrelated variables that enter into clinical decision making. Further analysis of these subgroups in matched pairs (if not prospective randomized studies) would be required before concluding that BIMA grafting is preferable for each cohort. What emerges from these preliminary findings is that there appears to be no single contraindication to the use of two IMA grafts. Almost all patients in whom BIMA grafting is technically feasible and can be safely performed are likely to derive a significant long-term survival benefit. The early and continued divergence of the actuarial curves suggests that the survival benefit generally becomes apparent fairly early during the follow-up period and increases over time. Study Limitations This study has some inherent limitations that must be considered in evaluating the results. As a single-center nonrandomized experience, the findings may represent technical subtleties that might preclude equivalent success in the widespread application of BIMA grafting. Second, most of the patients in this study were operated on during the 1980s and 1990s. Although this allowed for a longer follow-up, the introduction and advancement of percutaneous techniques as well as improvements in anesthesia management, surgical, and intensive care have resulted in more high-risk patients being referred for surgical revascularization. The data suggest that it is reasonable to hypothesize that similar results could be obtained in the current surgical environment. That theory remains to be tested. Moreover, the introduction of statins, angiotensin-converting enzyme inhibitors, and other evolving medical therapies may ultimately make the long-term benefit of BIMA grafting less apparent. Third, statistical efforts notwithstanding, this is a retrospective nonrandomized study. Although the decisionmaking pattern may have evolved over time, the year of operation did not emerge as a significant determinant of surgical approach during the course of the study. In a study spanning 23 operative years, there may have been a series of subtle alterations that cannot be discerned in a retrospective study. Conclusions Despite its inherent limitations, this study provides convincing evidence that BIMA grafting, when broadly ap-

7 Ann Thorac Surg KURLANSKY ET AL 2010;90:101 8 IMA GRAFTING 107 plied, furnishes a long-term survival advantage, over SIMA grafting in comparable patients. We thank Dr Debra D. Guest for technical assistance in the preparation of this report. References 1. Pick AW, Orszulak TA, Anderson BJ, Schaff HV. Single versus bilateral internal mammary artery grafts: 10-year outcome analysis. Ann Thorac Surg 1997;64: Stevens LM, Carrier M, Perrault LP, et al. Single versus bilateral internal thoracic artery grafts with concomitant saphenous vein grafts for multivessel coronary artery bypass grafting: effects on mortality and event-free survival. J Thorac Cardiovasc Surg 2004;127: Lytle BW, Blackstone EH, Sabik JF, Houghtaling P, Loop FD, Cosgrove DM. The effect of bilateral internal thoracic artery grafting on survival during 20 postoperative years. Ann Thorac Surg 2004;78: Calafiore AM, DiGiammarco G, Teodori G, et al. Late results of first myocardial revascularization in multiple vessel disease: single versus bilateral internal mammary artery with or without saphenous vein grafts. Eur J Thorac Cardiovasc Surg 2004;26: Berreklouw E, Raclemakers PP, Koster JM, van Leur L, van der Wielen BJ, Westers P. Better ischemic event-free survival after two internal thoracic artery grafts:13 years of follow up. Ann Thorac Surg 2001;72: Dewar LR, Jamieson WR, Janusz MT, et al. Unilateral versus bilateral internal mammary revascularization. Survival and event-free performance. Circulation 1995;92:IV Taggart DP, D Amico R, Altman DG. Effect of arterial revascularization on survival: a systematic review of studies comparing bilateral and single internal mammary arteries. Lancet 2001;356: Rizzoli G, Shiavon L, Bellini P. Does the use of bilateral internal mammary artery (IMA) grafts provide incremental benefit relative to the use of single IMA graft? Eur J Cariothorac Surg 2002;22: Taggart DP, Lees B, Gray A, et al. Protocol for the Arterial Revascularization Trial (ART). A randomized trial to compare survival following bilateral versus single internal mammary grafting in coronary revascularization. Trials 2006; 7: Tabata M, Grab JD, Khalpey Z, et al. Prevalence and variability of internal mammary artery graft use in contemporary multivessel coronary artery bypass surgery. Circulation 2009;120: Galbut DL, Traad EA, Dorman MJ, et al. Seventeen-year experience with bilateral internal mammary artery grafts. Ann Thorac Surg 1990;49: Rosenbaum PR, Rubin DR. Reducing bias in observational studies using subclassification on the propensity score. J Am Stat Assoc 1984;79: Rosenbaum PR. Optimal matching for observational studies. J Am Stat Assoc 1989;84: The Parisian Mediastinitis Study Group. Risk factors for deep sternal wound infection after sternotomy: a prospective multicenter study. J Thorac Cardiovasc Surg 1996;111: Savage EB, Grab JD, O Brien SM, et al. Use of both internal thoracic arteries in diabetic patients increases deep sternal wound infection. Ann Thorac Surg 2007;83: Gansera B, Schmidtler F, Gilrath G, et al. Does bilateral ITA grafting increase perioperative complications: Outcome of 4462 patients with bilateral versus 4204 patients with single ITA bypass. Eur J Cardiothorac Surg 2006;30: Goyal V, Pinto RJ, Mukherjee K, Trivedi A, Sharma S, Bhattacharya S. Alteration in pulmonary mechanics after coronary artery bypass surgery: comparison using internal mammary artery and saphenous vein grafts. Indian Heart J 1994;46: Kamiya H, Akhyari, Martens A, Karck M, Haverisch A, Lichtenberg A. Sternal microcirculation after skeletonized versus pedicled harvesting of the internal thoracic artery: a randomized study. J Thorac Cardiovasc Surg 2006;135: De Paulis R, de Notaris S, Scaffa R, et al. The effect of bilateral internal thoracic artery harvesting on superficial and deep sternal infection: the role of skeletonization. J Thorac Cardiovasc Surg 2005;129: Toumpoulis IK, Theakos N, Dunning J. Does bilateral internal thoracic artery harvest increase the risk of Mediastinitis? Interact Cardiovasc Thorac Surg 2007;6: Bonacchi M, Prifti E, Giunti G, Salica A, Frati G, Sani G. Respiratory dysfunction after coronary artery bypass grafting employing bilateral internal mammary arteries: the influence of intact pleura. Eur J Cardiothorac Surg 2001;19: Spence PA, Montgomery BS, Santamore WP. High flow demand on small arterial coronary bypass conduits promotes graft spasm. J Thorac Cardiovasc Surg 1995; DISCUSSION DR JOSEPH F. SABIK (Cleveland, OH): Dr Murray, Dr Wood, members and guests. I would like to congratulate Dr Kurlansky and colleagues for a fine study and presentation and thank them for providing me with a copy of both their manuscript and presentation in advance of this meeting. Dr Kurlansky, your study joins many other observational studies that demonstrate the benefit of bilateral ITA (internal thoracic artery) over single ITA grafting in prolonging survival after coronary surgery. Your findings make sense. The results of coronary surgery are directly related to graft patency, and since ITA grafts have better patency than vein grafts, a surgical strategy that includes more ITA grafts and fewer saphenous vein grafts should result in better long-term outcomes. However, despite work such as yours, there remains a discrepancy between what has been published and what continues to be practiced. As you stated, data from our database suggests that only 4% of all coronary operations are done with bilateral ITAs. An explanation for this inconsistency between clinical practice and literature is that many do not agree with conclusions drawn from studies such as yours. They believe that the bias in selecting patients for bilateral ITA grafting is so strong that no statistical method can adjust for the preoperative differences observed between these two groups and that the better outcomes observed in bilateral ITA patients are due to the fact that the lower-risk patients receive bilateral ITAs. Your study supports this bias, with the bilateral ITA patients being younger, less likely diabetic, and less likely to have renal dysfunction, vascular disease, congestive heart failure, and left ventricular dysfunction. My first question has to do with this. How confident are you that your statistical methods are able to adjust for this selection bias, and how do you convince the nonbelievers in the audience that bilateral ITA grafting is truly beneficial? My second question has to do with a lack of consistency in survival and freedom from cardiac events in your paper. Although survival is consistently better in your bilateral ITA patients, for the first 24 years of follow-up, the freedom from cardiac events is similar in both groups. Since we believe that bilateral ITA grafting

8 108 KURLANSKY ET AL Ann Thorac Surg IMA GRAFTING 2010;90:101 8 improves outcomes by decreasing cardiac events, how do you explain the lack of consistency between freedom from cardiac events in the bilateral ITA patients? Thank you. DR KURLANSKY: Thank you, Dr Sabik. It is a privilege for us to have commentary from the Cleveland Clinic. I think we all owe a tremendous debt of gratitude to Cleveland for introducing and teaching the profession not only about the internal mammary experience but also about how to understand it. I would like to parenthetically comment that as we find out more about the internal mammary graft as to why it stays patent, one thing we have learned is that it produces a lot of nitric oxide. And so it is not merely a function of the graft staying open or patent for a longer period of time, but essentially you are putting a nitric oxide pump or a nitroglycerin pump into the patient s heart, and I think this may also explain some of the long-term benefits. With regards your specific questions, I agree with you that even the finest statistical analyses cannot account for all factors that go into bias. I think one of the more compelling parts about this study is that 48% of the patients, essentially half the patients, had bilateral internal mammary grafting, and as well as we could do with matched groups, 60% of the patients had a statistically significant survival benefit. So that means that even though there are selection criteria, perhaps the selection criteria need to be considerably broader than what is currently practiced in the country. As for the end point of event-free survival, this is a weakness in our data, which is why we have emphasized the hard end point of mortality much more so in our study, because we do not have a prospective clinical ongoing follow-up. What we had was a crosssectional follow-up of survivors. Therefore, survival becomes a competing variable with all of the other parameters being measured, and if the patient had an event prior to his demise, we would not have necessarily known about it. So therefore, although it is encouraging, I think it is probably not as strong as it could be due to the nature in which we collected the data. The Society of Thoracic Surgeons Policy Action Center The Society of Thoracic Surgeons (STS) is pleased to announce a new member benefit the STS Policy Action Center, a website that allows STS members to participate in change in Washington, DC. This easy, interactive, hassle-free site allows members to: Personally contact legislators with one s input on key issues relevant to cardiothoracic surgery Write and send an editorial opinion to one s local media senators and representatives about upcoming medical liability reform legislation Track congressional campaigns in one s district and become involved Research the proposed policies that help or hurt one s practice Take action on behalf of cardiothoracic surgery This website is now available at by The Society of Thoracic Surgeons Ann Thorac Surg 2010;90: /$36.00 Published by Elsevier Inc