Conventional coronary artery bypass grafting (CCAB) is

Transcription

1 ORIGINAL ARTICLES Off-Pump versus Conventional Coronary Artery Bypass Grafting: A Meta-Analysis and Consensus Statement From The 2004 ISMICS Consensus Conference John Puskas, MD, a Davy Cheng, MD, b John Knight, MD, c Gianni Angelini, MD, d Didier DeCannier, MD, e Anno Diegeler, MD, f Mercedes Dullum, MD, g Janet Martin, PharmD, b Masami Ochi, MD, h Nirav Patel, MD, i Eugene Sim, MD, j Naresh Trehan, MD, k and Vipin Zamvar, MD l Background: The purpose of this evidence-based consensus statement is to systematically review and meta-analyze the randomized and nonrandomized evidence comparing off-pump (OPCAB) to conventional coronary artery bypass (CCAB) surgery and to provide consensus on the role of OPCAB in low- and high-risk surgical patients. Methods and Results: This consensus conference was conducted according to the American College of Cardiology (ACC)/American Heart Association (AHA) standards for development of clinical practice guidelines. The Steering Committee collated all published studies of OPCAB versus CCAB through May 2004 and developed six questions central to controversies surrounding OPCAB surgery in mortality, morbidity, and resource utilization. For mixed-risk patient populations, meta-analysis of 37 randomized clinical trials (3,369 patients, Level A) reported across a total of 53 papers, and two meta-analyses of nonrandomized trials (Level B) comparing OPCAB versus CCAB were identified. For high-risk patient populations, we performed a meta-analysis of 3 randomized and 42 nonrandomized trials (26,349 patients, Level B). Conclusion: Meta-analysis of Level A and B evidence provided the basis for the following consensus statements in patients undergoing surgical myocardial revascularization: (1) OPCAB should be considered a safe alternative to CCAB with respect to risk of mortality Institutions: a Division of Cardiothoracic Surgery, Emory University, Atlanta, USA; b Department of Anesthesia & Perioperative Medicine, London Health Sciences Centre, University of Western Ontario, London, Ontario, Canada; c Cardiothoracic Surgical Unit, Flinders Medical Center, Bedford Park, Australia; d Bristol Heart Institute, University of Bristol, Bristol, UK; e Erasme Hospital, Brussels, Belgium; f Herz-Und Gefasse Klinik Bad Neustadt, University of Leipzig, Bad Neustadt, Germany; g Department of Cardiothoracic Surgery, Cleveland Clinic Florida, Weston, Florida, USA; h Department of Cardiovascular Surgery, Nippon Medical School, Tokyo, Japan; i Lenox Hill Hospital, New York, New York, USA; j Department of Cardiovascular Surgery, National University Hospital, Singapore, Singapore; k Escorts Heart Institute and Research Center, New Delhi, India; l Department of Cardiothoracic Surgery, Royal Infirmary of Edinburgh, Edinburgh, UK. Address correspondence to: Dr. John Puskas, Division of Cardiothoracic Surgery, Emory Crawford Long Hospital, 550 Peachtree St., 6 th Floor Medical Office Tower, Atlanta, GA, 30308, USA. Copyright 2005 by Lippincott Williams & Wilkins ISSN: /05/ [Class I, Level A]; (2) With appropriate use of modern stabilizers, heart positioning devices, and adequate surgeon experience, similar completeness of revascularization and graft patency can be achieved [Class IIa, Level A]; (3) OPCAB is recommended to reduce perioperative morbidity [Class I, Level A]; (4) OPCAB may be recommended to minimize midterm cognitive dysfunction [Class IIa, Level A]; (5) OPCAB should be considered as an equivalent alternative to CCAB in regard to quality of life [Class I, Level A]; (6) OPCAB is recommended to reduce the duration of ventilation, ICU and hospital stay, and resource utilization [Class I, Level A]; (7) OPCAB should be considered in high-risk patients to reduce perioperative mortality, morbidity, and resource utilization [Class IIa, Level B]. (Innovations 2005;1: 3 27) Conventional coronary artery bypass grafting (CCAB) is one of the most frequently performed operations and, in specific subgroups, significantly improves survival, symptoms, and quality of life compared with nonsurgical management of coronary artery disease. 1 However, these benefits are tempered by perioperative risks, including mortality (1-3%), stroke (2%), myocardial infarction (1-2%), exposure to allogeneic blood transfusion (30-90%), atrial fibrillation (30%), and cognitive dysfunction (50-75%). 2 5 Portions of these morbidities have been attributed to the use of cardiopulmonary bypass (CPB), cardioplegic cardiac arrest, aortic cannulation, and cross clamping. Consequently, there has been increasing interest in techniques that avoid CPB, such as off-pump coronary artery bypass grafting (OPCAB). 6,7 Numerous studies have been published comparing clinical outcomes after OPCAB versus CCAB. The majority of earlier trials were nonrandomized comparisons of low-risk patients undergoing single- or double-vessel bypass, with the potential risk of unbalanced baseline patient characteristics leading to bias in favor of either OPCAB or CCAB. More recently, nonrandomized comparisons of high-risk patients and randomized comparisons of mixed-risk patients have been published. However, some of these trials have focused on surrogate primary endpoints (such as inflammatory mediators), and have had insufficient statistical power to ade- Innovations Volume 1, Number 1, Fall

2 Puskas et al. Innovations Volume 1, Number 1, Fall 2005 quately explore clinically important outcomes, which occur infrequently, such as death, stroke, and myocardial infarction. There remains considerable uncertainty as to the role of OPCAB for patients across the full spectrum of risk groups. The purpose of this evidence-based consensus statement is to systematically review and, where appropriate, meta-analyze the randomized and nonrandomized evidence specifically comparing OPCAB to CCAB and to provide consensus statements that clarify the role of OPCAB and CCAB for coronary revascularization in low- and high-risk surgical patients. METHODS A two-day consensus conference was held in Paris, France, May 21-22, 2004, under the auspices of The International Society for Minimally Invasive Cardiothoracic Surgery (ISMICS). The conference was conducted according to the ACC/AHA standards for the development of clinical practice guidelines [ manual_index.htm accessed April 19, 2004]. A subcommittee of consensus experts performed a literature search for clinical trials in Medline and the Cochrane Library using the keywords off-pump and coronary bypass surgery and collated all published manuscripts on OPCAB versus CCAB, for review prior to the Consensus Conference. The Steering Committee of the 2004 ISMICS Consensus Panel developed six questions central to the controversies presently surrounding the field of OPCAB surgery. These served to focus the systematic review, discussion, and testimony given by the expert panel at the Consensus Conference. Published evidence, such as systematic reviews or meta-analyses of controlled clinical trials, was sought to inform each of these six questions. When systematic reviews or meta-analyses were not found, a subgroup of the consensus panel performed a systematic review and, when required, meta-analysis of all randomized 8 and nonrandomized control trials 9,10 through May 2004 that compared mortality, morbidity, and resource utilization of OPCAB versus CCAB. The patient groups examined included low-risk, high-risk, and mixed-risk populations. Summary data from the systematic review and metaanalyses were presented at the Consensus Conference. The consensus panel provided evidence and/or expert opinion to formulate statements and recommendations regarding OPCAB surgery in various patient populations. These were assessed by classes of support and levels of evidence according to the classifications of the ACC/AHA [ accessed April 19, 2004]. Classes of recommendation were defined as follows: Class I. Conditions for which there is evidence and/or general agreement that a given procedure or therapy is useful and effective; 4 Class II. Conditions for which there is conflicting evidence and/or a divergence of opinion about the usefulness or efficacy of performing the procedure or therapy; Class II a. Weight of evidence or opinion is in favor of usefulness or efficacy; Class II b. Usefulness or efficacy is less well established by evidence or opinion; Class III. Conditions for which there is evidence and/or general agreement that a procedure or therapy is not useful or effective and in some cases may be harmful. These recommendations were based on the following levels of evidence: Level A. The data were derived from multiple randomized clinical trials; Level B. The data were derived from a single randomized study or from nonrandomized studies; Level C. The consensus opinion of experts was the primary source of recommendation. Systematic Review & Meta-Analysis: The methodology and results of the systematic review and meta-analysis of 37 randomized trials of mixed-risk populations undergoing OPCAB or CCAB has been previously described. 8 Since this previous publication did not analyze combined results for graft patency, we conducted a pooled analysis of randomized trials reporting graft patency. Two meta-analyses of nonrandomized trials of mixedrisk populations were identified during searches of the medical literature; 9,10 therefore a separate meta-analysis for this population was not conducted. No systematic reviews or meta-analyses of high-risk populations were identified. Therefore we conducted a full meta-analysis for this patient group. The methodology for the systematic review and meta-analysis of trials of high-risk groups undergoing OPCAB or CCAB is described below. Prior to the consensus conference, a comprehensive search was undertaken, in accordance with the Cochrane Collaboration, 11 to identify all published randomized and nonrandomized trials of OPCAB versus CCAB, in all languages. MEDLINE, Cochrane CENTRAL, EMBASE, Current Contents, DARE, NEED, INAHTA databases were searched from the date of their inception through May The processes of comprehensive literature search and systematic review followed the methodologies and policies of the ACC/ AHA Task Force on Practice Guidelines [ org/clinical/manual/manual_index.htm accessed April 19, 2005 Lippincott Williams & Wilkins

3 Innovations Volume 1, Number 1, Fall 2005 Off-Pump vs. Conventional Coronary Bypass Surgery 2004], while the meta-analysis was conducted in accordance with the Quality of Reporting of Meta-Analyses 12 and Meta-Analysis of Observational Studies of Epidemiology 13 recommendations. Inclusion Criteria All comparative trials of high-risk patients were included, whether randomized or nonrandomized, if they met each of the following criteria: 1 comparison of OPCAB versus CCAB; 2 adult patients undergoing single or multiple vessel bypasses; and 3 reporting at least one pertinent clinical or economic outcome. Studies of robotic surgery and/or combined valve/coronary procedures were excluded. Endpoints All-cause mortality at 30 days and 1 year were the primary endpoints. Secondary outcomes included postoperative incidence of stroke, cognitive dysfunction, acute myocardial infarction, recurrent angina, coronary reintervention, need for inotropes, need for intra-aortic balloon pump (IABP), atrial fibrillation, renal failure, mediastinitis/sternal wound infection, respiratory infection, need for blood transfusion, re-exploration for bleeding, duration of ventilation, ICU length of stay (LOS), hospital LOS, hospital costs, and quality of life (QOL). Postoperative atrial fibrillation, myocardial infarction, respiratory infection, mediastinitis/wound infection, and cognitive dysfunction were defined according to study authors definitions. Need for transfusion was defined as the proportion of patients requiring red blood cell transfusion during the intra- and postoperative periods, combined. Renal failure was defined per study authors definitions, typically a new rise in serum creatinine of 50%, a decline in creatinine clearance of 50%, or a requirement for dialysis. Duration of ventilation was measured in hours from end of surgery to time of tracheal extubation. Intensive care LOS and hospital LOS were measured in days, starting from end of surgery to ICU or hospital discharge, respectively. Statistical Analysis Outcomes were analyzed as dichotomous variables, with the exception of duration of ventilation and LOS, which were analyzed as continuous variables when the mean and standard deviation were provided. For dichotomous variables, odds ratios and 95% confidence intervals [OR, 95% CI] were calculated. For continuous variables, the weighted mean difference [WMD, 95% CI] was calculated. was explored using the Q-statistic. Due to the low power of the Q-statistic, 14 a higher threshold of P 0.10 was used to suggest statistically significant heterogeneity across trials. In addition to the Q statistic, the I 2 was calculated to quantify the degree of heterogeneity across trials that could not be attributable to chance alone. 11,15,16 For each outcome, the fixed effect or random effects model was used when the Q-statistic suggested lack or presence of heterogeneity, respectively. Pooled effect estimates and heterogeneity between studies were analyzed by use of Comprehensive MetaAnalysis (v1.0, Biostat, Englewood, NJ, USA 2002) and RevMan (v4.2.2, Cochrane Collaboration, Oxford, England, 2003). Other than for the Q statistic, statistical significance was defined as P All tests of statistical significance were two-sided. Whenever possible, data analysis was by intention-to-treat. Subanalyses for each high-risk group were planned a priori for the following patient risk groups: elderly (age 70 years), obesity (BMI 25 kg/m 2 ), diabetes, renal failure, aortic disease, left main disease, left ventricular (LV) dysfunction, chronic obstructive or other pulmonary disease (COPD), urgent/emergent or re-do bypass, those requiring conversion from OPCAB to CCAB, and those with a high clinical risk score (i.e., usually EUROSCORE 5 or Parsonnet score 15), or mixed populations of patients with or more of the above risk factors. In subgroup analyses, the differences in relative size of effect were tested using a chi-square test for interaction or a chi-square test for trend. RESULTS For mixed-risk patient populations (unselected for high-risk characteristics), the results of a meta-analysis of 37 randomized clinical trials (3369 patients) comparing OPCAB versus CCAB (Level A) has been previously described. 8 Since some randomized trials reported their results over a series of published reports, these 37 unique trials were reported across a total of 53 papers One paper reported two randomized trials. 17 Two meta-analyses of nonrandomized (Level B) OPCAB versus CCAB trials of mixed-risk populations have also been previously reported. 9,10 The latter meta-analysis by Reston et al. combined nonrandomized trials with randomized trials; however the number of randomized trials was small relative to the nonrandomized trials. Therefore, the meta-analysis by Reston et al. will be referred to as Level B evidence. For high-risk patient populations, we identified 45 eligible trials, including 26,349 patients, for our systematic review and meta-analysis (see Fig. 1). Of these, three were randomized controlled trials 27,28,36 and 42 were nonrandomized trials Most of the nonrandomized trials were retrospective cohort analyses, comparing patients undergoing OPCAB with patients undergoing CCAB over the same time period. A number of trials used data from patient registries, some of which may represent overlapping data across trials. All prospectively identified high-risk populations of interest were addressed in these trials. Subgroup analysis of individual risk groups was possible, except for the group of patients undergoing conversion from off- to on-pump or vice versa, since there was insufficient comparative information for this group. Publication bias was not identified for any of the clinical outcomes evaluated in high-risk groups. Upon review of this evidence, the Consensus Panel addressed six prespecified questions, crafted statements and recommendations, and labeled each statement with the highest available level of evidence. Question 1: Is OPCAB surgery associated with similar all-cause mortality at 30 days and > 1 year compared to CCAB surgery? 2005 Lippincott Williams & Wilkins 5

4 Puskas et al. Innovations Volume 1, Number 1, Fall 2005 Early Mortality (up to 30 days): In randomized trials of mixed-risk groups, no differences were found in any trial for mortality at 30 days (29 RCTs, Level A). 17,26-28,30-32,34-37,40-44,47,49,51,52,56-59,62,63,67,68 Pooled analysis of these randomized trials found no difference in 30-day mortality (OR 1.02, 95%CI , P 0.9; n 3,082 patients in 29 RCT studies). 8 (see Table 1) Considering that the pooled risk of mortality was 1.2% for OPCAB and 1.0% in the CCAB, and given that the absolute difference in mortality between the groups was so small (0.2%), it is clear that significant differences in mortality (if they exist at all) are unlikely to be found in patients of mixed- (generally low) risk groups similar to those included in these randomized trials. To find a mortality difference of 0.2%, very large trials (i.e., of 85,000 patients) would be required. 8 While pooled Level A evidence has had inadequate sample size to detect this small potential difference in mortality when low-risk groups are studied, there is a large body of retrospective studies (Level B) of mixed-risk patients showing reduced mortality at 30 days in OPCAB when compared with CCAB. Two meta-analyses of mixed-risk groups that included nonrandomized controlled trials (Level B) 9,10 found reduced risk of mortality at 30 days with OPCAB versus CCAB. Reston et al. found a 36% reduction in risk of early mortality (OR, 0.64, 95%CI , P ; n 39,647 patients in 43 Level A and B studies combined). 10 Beattie et al. found a 21% reduction in mortality (OR 0.79, 95%CI ; P ; n 159,845 patients in 7 studies). 9 The apparent discrepancy between Level A and Level B evidence for early mortality may be related to inclusion of higher-risk groups in Level B studies. However, bias in patient selection or management (including surgeonrelated differences between groups) cannot be excluded. Late Mortality (up to 3 years): In randomized trials of mixed-risk groups, no differences were found in any trial for mortality at 1 to 3 years (6 RCTs, n 1,135) (Level A). 17,31,49,52,59 Pooled analysis of these randomized trials found no difference in mortality at 1 to 3 years (OR 0.88, 95%CI , P 0.8; n 1,135 patients in 6 RCT studies) (Level A). 8 (See Table 1) Figure 1. Flowchart of Trial Inclusion for Meta-Analysis of High Risk Patients [Level A/B]. 6 Mortality in Converted OPCAB Patients to CPB: A total of 20 out of 37 identified randomized controlled trials [Level A] 17,27,30,31,38,40,42-45,48,50,52,56-58,63,64,68 reported conversions of patients from off-pump to on-pump surgery. Rates of conversion in these trials ranged from 0% to over 20%, with an overall average rate of 8% of OPCAB patients converted to on-pump surgery. 8 However, no randomized trials reported outcomes separately for patients who were urgently converted. A small body of Level B and C evidence (5 studies, reporting conversion rates of 5 to 13%, for a total of 175 converted patients) reports increased risk of mortality (i.e., absolute risk increase of 6 to 15%) for patients emergently converted from OPCAB to CCAB (Level B and C) Other adverse outcomes, including myocardial infarction, stroke, and multisystem organ failure were increased in urgently converted OPCAB patients. Statement: When compared to CCAB in mixed-risk surgical groups, OPCAB results in similar [Level A] or reduced [Level B] risk of mortality at 30 days. At 1 to 3 years follow-up, mortality is similar between groups [Level A and Level B]. Recommendation: OPCAB should be considered a safe alternative to CCAB with respect to risk of mortality in patients undergoing surgical myocardial revascularization [Class I, Level A] Lippincott Williams & Wilkins

5 Innovations Volume 1, Number 1, Fall 2005 Off-Pump vs. Conventional Coronary Bypass Surgery Table 1. Meta-Analysis of Randomized Trials In OPCAB Surgery [Level A] 8 Clinical Outcomes Outcome n (N) OPCAB % CCAB % OR [95%CI] NNT [95%CI] p-value I 2 P for Overall Effect Death, 30d 3082 (29) [ ] Death, 1-2y 1135 (6) [ ] Atrial Fibrillation, 30d 2425 (17) [ ] 11 [8-17] Transfused Patients, 30d 2412 (17) [ ] 7 [6-10] Respiratory Infections, 30d 896 (7) [ ] 19 [12-52] Inotropes, in hospital 1655 (16) [ ] 12 [8-21] Cognitive Dysfunction, 2-6mos 393 (3) [ ] 10 [6 to 11] Resource Utilization Outcome N WMD [95%CI] p-value I2 P for Overall Effect Hospital LOS, d 1384 (17) 1.0 days [ 1.5 to 0.5] ICU LOS, d 1266 (15) 0.3 days [ 0.6 to 0.1] Ventilation Time, h 1425 (20) 3.4 hrs [ 5.1 to 1.7] NNT number needed to treat; WMD weighted mean difference NNT number needed to treat; Redo/Urgent Repeat, urgent, or emergent coronary bypass surgery; WMD weighted mean difference Question 2: Does OPCAB provide similar completeness of revascularization and graft patency when compared to CCAB? Complete revascularization of all coronary arteries with stenoses 70% has been a hallmark of surgical therapy for coronary artery disease and is important for long-term patient benefits Level B evidence, especially nonrandomized retrospective reviews conducted early in the experience of OPCAB, generally showed fewer grafts performed per patient in the OPCAB groups relative to CCAB Our meta-analysis of randomized trials reporting number of grafts (22/37 RCTs; n 2,062 patients) showed a lower number of grafts per patient in the OPCAB group compared with CCAB [2.6 versus 2.8; P ] (see Fig. 2a). While this difference is statistically significant, the clinical significance of 0.2 fewer grafts per patient is difficult to assess. When the data were analyzed separately for earlier trials (i.e., prior to and including the year 2000) and later trials (after the year 2000), the number of grafts per OPCAB patient increased, but the difference in grafts revascularized remained significantly different: 2.1 OPCAB versus 2.4 CCAB for earlier trials (P 0.004), and 2.7 OPCAB versus 2.9 CCAB for later trials (P ) (see Fig. 2a). Completeness of revascularization was reported in only randomized trials. 27,28,31,40,43,50,53 Carrier et al. 27 and Czerny et al. 31 reported significantly reduced completeness of revascularization in OPCAB group, whereas Covino, 28 Khan, 40 Legare, 43 Nathoe 50 and Puskas 53 reported no difference in completeness of revascularization. Nonrandomized trials suggest that completeness of revascularization may be similar 123,124 or decreased 97,125 with OPCAB [Level B]. Four randomized trials [Level A] of OPCAB versus CCAB evaluated graft patency by angiography at time points ranging from in-hospital to 1 year postoperatively. 40,44,50,53 Of 617 patients randomized in these studies, 451 underwent postoperative catheterization of 1343 grafts. Puskas et al. 53 found no difference in graft patency prior to hospital discharge (622 grafts). Khan et al. 40 found a decrement in graft patency in the OPCAB group at 3 months; in contrast, Lingaas et al. 44 found no significant difference at 3 months. At 1 year, Nathoe et al. 50 and Puskas et al. 53 found no significant difference in graft patency between the OPCAB and CCAB groups. Meta-analysis of graft patency at 1 year confirmed no significant difference between OPCAB and CCAB groups (Fig. 2b). The discrepant results of Khan et al. may result from divergent usage of radial artery conduits and limited surgeon experience in OPCAB techniques, as acknowledged by the authors. 40 Nonrandomized studies [Level B] report similar patency of arterial conduits for on- and OPCAB Level B and Level C studies have demonstrated excellent arterial graft patency after OPCAB (94% to 100% at 6 months to 6 years follow-up) with no significant differences compared with CCAB in Level B studies While some nonrandomized studies report a decrement in vein graft patency of up to 24%, 128,132 others 129 report similar vein graft patency after OPCAB versus CCAB. [Level B] The ultimate clinical expression of reduced patency is increased need for repeat revascularization over time. Metaanalysis of randomized trials [Level A] has not shown a significantly increased risk of need for repeat revascularization at 1 to 3 years postbypass surgery (OR 1.61, 95%CI 2005 Lippincott Williams & Wilkins 7

6 Puskas et al. Innovations Volume 1, Number 1, Fall 2005 Figure 2. a Meta-Analysis of Number of Grafts Per Patient Before and after Year 2000 in Mixed-Risk Patients [Level A] 2b Meta-Analysis of Graft Patency at One Year in Mixed-Risk Patients [Level A] , P 0.3; n 1,120 in 6 studies). 17,31,49,52,59 Longer follow-up is not yet available. Statement: The number of grafts performed and completeness of revascularization is slightly reduced with OPCAB compared with CCAB [Level A and Level B]. These differences have not translated into measurable increases in need for repeat revascularization or reintervention at 1 to 3 years follow-up [Level A]. Graft patency is similar after OPCAB and CCAB [Level A and Level B]. Completeness of revascularization and patency likely depend on surgeon expertise [Level C]. Recommendation: With appropriate use of modern stabilizers, heart positioning devices and adequate surgeon experience, similar completeness of revascularization and graft patency can be achieved with OPCAB [Class IIa, Level A]. Longitudinal 8 follow-up of randomized trials to explore the long-term impact of OPCAB on recurrent ischemia and need for reintervention should be performed. Question 3: What are the rates of postoperative stroke, acute myocardial infarction, atrial fibrillation, recurrent angina, reintervention for ischemia, renal failure, blood transfusion, re-exploration for bleeding, inotrope dependence, intra-aortic balloon pump (IABP) placement, mediastinitis/wound infection, and respiratory infections after OPCAB versus CCAB surgery? Some but not all randomized trials showed significant reduction in atrial fibrillation, 17,49 red blood cell transfusions, 17,26,35,40,49,52,62,68 inotrope requirements, 17,26,34 and respiratory infections 17 with OPCAB compared to CCAB. In pooled analysis of randomized clinical trials evaluating outcomes in patient populations with mixed-risk factors, OP- CAB reduces the incidence of atrial fibrillation (OR 0.58, 2005 Lippincott Williams & Wilkins

7 Innovations Volume 1, Number 1, Fall 2005 Off-Pump vs. Conventional Coronary Bypass Surgery 95%CI , P ; n 2,425 patients in 17 RCTs), red blood cell transfusions (OR 0.43, 95%CI , P ; n 2,412 patients in 17 RCTs), inotrope requirements (OR 0.48, 95%CI , P ; n 1,655 patients in 16 RCTs), and respiratory infections (OR 0.41, 95%CI , P ; n 896 patients in 7 RCTs) when compared with CCAB (Table 1). 8 [Level A] In patient populations of mixed-risk factors, no randomized trials showed a significant reduction in stroke, myocardial infarction, acute renal failure, IABP requirement, mediastinitis/wound infection, angina recurrence, and need for reintervention within 30 days when compared with CCAB. Similarly, in pooled analysis of these randomized trials, OPCAB had a similar incidence of stroke, myocardial infarction, acute renal failure, IABP requirement, mediastinitis/ wound infection, angina recurrence, and need for reintervention within 30 days when compared with CCAB. 8 [Level A] Furthermore in patient populations of mixed-risk factors, OPCAB, when compared with CCAB at mid-term (1 to 3 years), did not alter the incidence of stroke, myocardial infarction, angina recurrence, and need for reintervention. 8 [Level A] In mixed-risk patient populations, OPCAB was associated with a reduction in markers of subsystem organ dysfunction: troponin T, troponin I, and inflammatory response, compared to CCAB. 19,20,25,26,30,32,35,37,46,47,51,52,55,62,63,66,67 [Level A] Evidence from a large body of retrospective studies [Level B] and two meta-analyses including Level B evidence of mixed-risk surgical groups 9,10 show reduced stroke risk with OPCAB versus CCAB. Reston et al. showed a 45% reduction in the risk of stroke (OR 0.55, 95%CI , P ; n 34,126 patients in 38 studies], 10 and Beattie et al. showed a 40% reduction in the risk of stroke (OR 0.60, 95%CI , P ; n 192,682 in 11 studies). 9 Reston et al. included 53 studies in their meta-analysis (10 RCTs, of which 3 were duplicates; plus 43 observational studies). They also reported a significant reduction for myocardial infarction (OR 0.58, 95%CI , P ; n 24,322 patients in 26 studies), atrial fibrillation (OR 0.69, 95%CI , P ; n 22,092 patients in 28 studies), renal failure (OR 0.62, 95%CI , P ; n 20,845 patients in 17 studies), reoperation for bleeding (OR 0.54, 95%CI , P ; n 33,442 patients in 24 studies), wound infection (OR 0.l55, 95%CI , P 0.004; n 16,039 patients in 17 studies), and reintervention (OR 3.63, P ; n 2,823 patients in 7 studies). 10 [Level B] Meta-analysis of Level A evidence showed no significant difference in the incidence of stroke at 30 days (OR 0.68, 95%CI , P 0.3; n 2,859 in 21 RCTs). 8 Whether the discrepancy between Level A and Level B evidence is related to insufficient statistical power in Level A trials or to a greater apparent benefit of OPCAB in higher-risk groups included in Level B trials remains uncertain. Statement: In mixed-risk patient populations, OPCAB is associated with reduced risk of perioperative atrial fibrillation, red blood cell transfusion, inotrope requirements, and respiratory infections [Level A]. OPCAB is also associated with reduction in serum levels of myocardial enzymes and inflammatory mediators [Level A]. Level B evidence suggests potential reduction in additional perioperative morbidity (stroke, myocardial infarction, renal failure, reoperation for bleeding, wound infection, and reintervention). At up to 3 years follow-up, the risk of stroke, myocardial infarction, angina recurrence, and need for reintervention similar between OPCAB and CCAB [Level A]. Recommendation: OPCAB is recommended in patients undergoing surgical myocardial revascularization to reduce perioperative morbidity [Class I, Level A]. Question 4: Are there differences in cognitive function and quality of life outcomes between OPCAB and CCAB? Cognitive Outcome: Early: Significant reduction in selected measures of cognitive decline was found up to 2 weeks postoperatively in two randomized studies. 42,68 One randomized study did not show any difference in cognitive outcomes between OPCAB and CCAB at month. 60 Pooled analysis of these randomized trials showed a reduction in patients experiencing cognitive decline (OR 0.57, 95%CI 0.21 to 1.54, P 0.04; n 335 in 3 studies), 8 but the results were not statistically significant when heterogeneity between results was accounted for statistically (P 0.04 with fixed effects model; P 0.3 with random effects model). Midterm. Of four randomized trials 32,60,68,69 reporting cognitive outcomes at 2 to 6 months, 3 showed significant difference 32,68,69 and showed no significant difference 60 in selected measures of cognitive decline. Pooled analysis of 3 of these randomized trials 60,68,69 (the fourth trial 32 did not meet criteria for meta-analysis) showed a 46% reduction in the number of patients with cognitive dysfunction in the off-pump group at 2 to 6 months (OR 0.56, 95%CI , P 0.01; n 393 in 3 studies). (Table 1) 8 [Level A] Late. Four randomized trials compared cognitive function at year and did not find a difference in number of patients with cognitive decline in OPCAB versus CCAB groups. 42,46,60,69 [Level A] Pooled analysis of these randomized trials showed no significant difference in number of patients with cognitive decline at 1 year (OR 0.91, 95%CI , P 0.7; n 334 in 2 trials). 8 [Level A] No Level A or Level B studies of cognitive outcomes beyond 1 year were found. Comparative cognitive outcome after OPCAB and CCAB depends to a great extent upon the tests used, the time period of assessment, the definition of significant cognitive dysfunction, and the statistical methodology. 133,134 There is additional rationale for improved results in OPCAB from studies of surrogate markers of brain dysfunc Lippincott Williams & Wilkins 9

8 Puskas et al. Innovations Volume 1, Number 1, Fall 2005 Table 2. Direct In-Hospital Cost Comparison Between OPCAB and CCAB [Level A]: Study Change in Mean Direct Costs OPCAB vs CCAB p-value Ascione 1999 [BHACAS 1] -30% p Lee % p Nathoe 2003 [Octopump] -14% p 0.01 Puskas 2003 [SMART] -11% p Straka 2004 [Prague-4] -79% p 0.01 tion, such as measurement of differences in brain water [Level B], 135,136 brain perfusion [Level A], 48,63 and cerebral microemboli [Level A] 69 [Level B]. 137 However, the clinical impact of these results is unclear. Quality of Life: Early: Two randomized controlled trials reported no significant differences between OPCAB and CCAB in QOL scores as measured by the EuroQOL and SF-36 at 4 to 6 weeks. 53,60 [Level A] Pooled analysis of QOL results was not attempted, due to heterogeneity in definitions and endpoints. Late. Three randomized studies 24,50,53 reported similar improvements in measures of quality of life after 1 year of follow-up, with no significant differences between OPCAB and CCAB groups [Level A]. Of nonrandomized trials reporting QOL, showed no difference 73,125 and showed significantly improved QOL 138 with OPCAB versus CCAB. [Level B] Table 3a. 30-day Death in High-Risk Patients [Level B/A]* Risk Groups n (N) OPCAB % CCAB % OR [95%CI] NNT p-value [95%CI] I 2 All Hi-Risk Pts 24,989 (51) [ ] < LVD 3,223 (9) [ ] Elderly 1,672 (8) [ ] Left Main 1,988 (2) [ ] Diabetics 2,478 (2) [ ] Renal Dysfunction 643 (4) [ ] Euroscor5 2,456 (5) [ ] < Multirisks 10,374 (9) [ ] < Aorta, atheromatous 762 (3) [ ] Redo/Urgent 1,213 (6) [ ] COPD 171 (3) [ ] *Meta-analysis of 45 trials including 26,349 highrisk patients (3 randomized and 42 nonrandomized trials) COPD chronic obstructive pulmonary disease; Euroscore5 High Euroscore/Parsonnet; LOS Length of stay; LVD left ventricular dysfunction; NNT number needed to treat; Redo/Urgent Repeat, urgent, or emergent coronary bypass surgery; WMD weighted mean differencetable 3b: 30-day Stroke in High-Risk Patients [Level B/A]* Table 3b. Risk Groups n (N) OPCAB % CCAB % OR [95%CI] NNT p-value [95%CI] I 2 All Hi-Risk Pts 24,353 (37) [ ] < LVD 1,890 (5) [ ] Elderly 845 (5) [ ] Left Main 1,165 (1) [ ] Diabetics 5,369 (3) [ ] Renal Dysf 617 (3) [ ] Euroscore5 2,465 (5) [ ] Multirisks 10,161 (8) [ ] <0.001 Aorta, atheromatous 701 (2) [ ] Redo/Urgent 1064 (4) [ ] COPD 76 (1) [ ] *Meta-analysis of 45 trials including 26,349 highrisk patients (3 randomized and 42 nonrandomized trials) COPD chronic obstructive pulmonary; Euroscore5 High Euroscore/Parsonnet disease; LOS Length of stay; LVD left ventricular dysfunction; NNT number needed to treat; Redo/Urgent Repeat, urgent, or emergent coronary bypass surgery; WMD weighted mean differencetable 3c: 30-day Myocardial Infarction in High-Risk Patients [Level B/A]* Lippincott Williams & Wilkins

9 Innovations Volume 1, Number 1, Fall 2005 Off-Pump vs. Conventional Coronary Bypass Surgery Table 3c. Risk Groups n (N) OPCAB % CCAB % OR [95%CI] NNT p-value [95%CI] I 2 All Hi-Risk Pts 2,166 (32) [ ] LVD 2,207 (6) [ ] Elderly 1,004 (3) [ ] Left Main 1,988 (2) [ ] Diabetics 4,188 (2) [ ] Renal Dysfunction 570 (3) [ ] Euroscor5 2,465 (5) [ ] Multirisks 8,008 (5) [ ] Aorta, atheromatous 211 (1) [ ] Redo/Urgent 1032 (4) [ ] COPD 58 (1) [ ] *Meta-analysis of 45 trials including 26,349 highrisk patients (3 randomized and 42 nonrandomized trials) COPD chronic obstructive pulmonary disease; Euroscore5 High Euroscore/Parsonnet; LOS Length of stay; LVD left ventricular dysfunction; NNT number needed to treat; Redo/Urgent Repeat, urgent, or emergent coronary bypass surgery; WMD weighted mean difference Table 3d: 30-day Atrial Fibrillation in High-Risk Patients [Level B/A]* Table 3d. Risk Groups n (N) OPCAB % CCAB % OR [95%CI] NNT p-value [95%CI] I 2 All Hi-Risk Pts 18,02 (27) [ ] < LVD 1,851 (4) [ ] Elderly 1,053 (4) [ ] Left Main 1,165 (1) [ ] Diabetics 2,891 (1) [ ] Renal Dysfunction 570 (3) [ ] Euroscor5 2,119 (4) [ ] Multirisks 7,353 (6) [ ] < Aorta, atheromatous Redo/Urgent 960 (3) [ ] COPD 58 (1) [ ] *Meta-analysis of 45 trials including 26,349 highrisk patients (3 randomized and 42 nonrandomized trials) COPD chronic obstructive pulmonary disease; Euroscore5 High Euroscore/Parsonnet; LOS Length of stay; LVD left ventricular dysfunction; NNT number needed to treat; Redo/Urgent Repeat, urgent, or emergent coronary bypass surgery; WMD weighted mean differencetable 3e: RBC Transfusion in High-Risk Patients [Level B/A]* Statement: OPCAB surgery may have a positive impact in midterm (2 to 6 months) cognitive function [Level A]. For early (up to 30 days) and late cognitive function (1 year), no difference has been shown between OPCAB and CCAB [Level A]. Limited data suggest no difference in QOL outcomes at 1 month to 4 years [Level A]. Recommendation: OPCAB may be recommended to minimize midterm cognitive dysfunction in patients undergoing surgical coronary revascularization. (Class IIa, Level A) OPCAB should be considered as an equivalent alternative to CCAB in regard to QOL for patients undergoing surgical myocardial revascularization. (Class I, Level A) Question 5: Are there differences between OPCAB and CCAB surgery in resource utilization, including duration of ventilation, ICU length of stay (LOS), hospital LOS, and hospital costs? A significant reduction in duration of ventilation, 17,30,31,35,36,39,47,49,51 ICU stay, 17,26,28,30,36 hospital stay, 17,26,35,47,49 and in-hospital costs 22,42,50,52,56 in OPCAB compared to CCAB has been confirmed by a large body of randomized trials [Level A]. Pooled analysis of randomized trials showed significant reductions in hospital stay (WMD 1.0 day, 95%CI 1.5 to 0.5 days, P ; n 1,384 patients in 17 RCTs), ICU stay (WMD 0.3 days, 0.6 to 0.1 day, P 0.003; n 1,266 patients in 15 RCTs), and duration of ventilation (WMD -3.4 hours, 95%CI 5.1 to 1.7 hours, P ; n 1,425 patients in 20 RCTs) (Table 1). 8 [Level A] The reduction in direct hospital cost (Table 2) is secondary to a reduction in ICU stay, hospital stay, blood transfusion, and a lower incidence of postoperative complications Lippincott Williams & Wilkins 11

10 Puskas et al. Innovations Volume 1, Number 1, Fall 2005 Table 3e. Risk Groups n (N) OPCAB % CCAB % OR [95%CI] NNT p-value [95%CI] I 2 All Hi-Risk Pts 18,033 (27) [ ] 5 < < LVD 2,580 (5) [ ] Elderly 1,028 (4) [ ] < Left Main 2,117 (3) [ ] Diabetics 4,142 (2) [ ] Renal Dysfunction 97 (2) [ ] Euroscor5 458 (1) [ ] Multirisks 7,429 (8) [ ] < Aorta, atheromatous Redo/Urgent 104 (1) [ ] COPD 76 (1) [ ] *Meta-analysis of 45 trials including 26,349 highrisk patients (3 randomized and 42 nonrandomized trials) COPD chronic obstructive pulmonary disease; Euroscore5 High Euroscore/Parsonnet; LOS Length of stay; LVD left ventricular dysfunction; NNT number needed to treat; Redo/Urgent Repeat, urgent, or emergent coronary bypass surgery; WMD weighted mean differencetable 3f: 30-day Renal Dysfunction in High-Risk Patients [Level B/A]* Table 3f. Risk Groups n (N) OPCAB % CCAB % OR [95%CI] NNT p-value [95%CI] I 2 All Hi-Risk Pts 24,111 (33) [ ] < LVD 1,829 (3) [ ] Elderly 583 (3) [ ] Left Main 1,204 (2) [ ] Diabetics 5,369 (3) [ ] Renal Dysfunction 326 (2) [ ] < Euroscor5 2,465 (5) [ ] Multirisks 7,088 (6) [ ] Aorta, atheromatous 701 (2) [ ] Redo/Urgent 932 (2) [ ] COPD 76 (1) [ ] *Meta-analysis of 45 trials including 26,349 risk patients (3 randomized and 42 nonrandomized trials) COPD chronic obstructive pulmonary disease; Euroscore5 High Euroscore/Parsonnet; LOS Length of stay; LVD left ventricular dysfunction; NNT number needed to treat; Redo/Urgent Repeat, urgent, or emergent coronary bypass surgery; WMD weighted mean differencetable 3g: Inotrope Requirements in High-Risk Patients [Level B/A]* Statement: Substantial evidence exists for a reduction in resource utilization, including duration of ventilation, ICU and hospital LOS, in OPCAB versus CCAB surgery [Level A] Recommendation: OPCAB is recommended in patients undergoing surgical myocardial revascularization to reduce the duration of ventilation, ICU and hospital stays, and resource utilization. [Class I, Level A] Question 6: Are there differences in a) mortality, b) morbidity and c) QOL and resource utilization in high-risk patients having OPCAB versus CCAB surgery? a) Mortality in High-Risk Patients Only randomized trials of high-risk patients were identified: in patients with COPD 28,36 and in patients with at least high-risk factor. 27 None of these trials found a significant 12 difference in mortality; however their combined sample size was low (n 160) [Level A]. Forty-two nonrandomized trials of high-risk patients were identified, a number of which demonstrated significant reduction in mortality after OPCAB versus CCAB in various high-risk patient subsets [Level B]. Our pooled analysis of Level B evidence (Table 3a) showed a significantly improved survival when all high-risk patient groups were combined (OR 0.58, 95%CI ; P , n 24,989) [Level B]. In addition, pooled analysis showed that mortality was reduced in patient subgroups with the following high-risk factors (Table 3a, and Fig. 3a) [Level B]: Euroscore 5 98,99,102,109,139 LV dysfunction 70,73,75,76,78,82,92 94 Atheromatous Aorta 75,140,141 Presence of at least high-risk factor 27,71,73,75,84 86,97,105, Lippincott Williams & Wilkins

11 Innovations Volume 1, Number 1, Fall 2005 Off-Pump vs. Conventional Coronary Bypass Surgery Table 3g. Risk Groups n (N) OPCAB % CCAB % OR [95%CI] NNT p-value [95%CI] I 2 All Hi-Risk Pts 10,325 (12) [ ] 14 < < LVD 250 (1) [ ] Elderly 595 (1) [ ] (significantly increased use of inotropes) Left Main 2117 (3) [ ] 4 < < Diabetics Renal Dysfunction 253 (1) [ ] Euroscor (1) [ ] (significantly increased use of inotropes) Multirisks 780 (2) [ ] <0.001 Aorta, atheromatous Redo/Urgent COPD 95 (2) [ ] *Meta-analysis of 45 trials including 26,349 highrisk patients (3 randomized and 42 nonrandomized trials) COPD chronic obstructive pulmonary disease; Euroscore5 High Euroscore/Parsonnet; LOS Length of stay; LVD left ventricular dysfunction; NNT number needed to treat; Redo/Urgent Repeat, urgent, or emergent coronary bypass surgery; WMD weighted mean differencetable 3h: Intra-Aortic Balloon Pump in High-Risk Patients [Level B/A]* Table 3h. Risk Groups n (N) OPCAB % CCAB % OR [95%CI] NNT p-value [95%CI] I 2 All Hi-Risk Pts 15,311 (18) [ ] LVD 3039 (4) [ ] Elderly 735 (2) [ ] Left Main 2117 (3) [ ] Diabetics Renal Dysfunction 253 (1) [ ] Euroscore5 0 Multirisks 6,149 (5) [ ] < Aorta, atheromatous Redo/Urgent 905 (2) [ ] COPD 58 (1) [ ] *Meta-analysis of 45 trials including 26,349 highrisk patients (3 randomized and 42 nonrandomized trials) COPD chronic obstructive pulmonary disease; Euroscore5 High Euroscore/Parsonnet; LOS Length of stay; LVD left ventricular dysfunction; NNT number needed to treat; Redo/Urgent Repeat, urgent, or emergent coronary bypass surgery; WMD weighted mean differencetable 3i: Reoperation for Bleeding in High-Risk Patients [Level B/A]* In contrast, no effect on early mortality was found for OPCAB versus CCAB in the following high-risk patient subgroups [Level B] (Fig. 3a): Increased age 73,75,77,80,90,96,100,101 Left Main Disease 74,79 Diabetes 95,103 Renal Dysfunction 72,75,83,108 COPD 28,36,75 Statement: OPCAB is associated with reduced mortality in pooled analysis of high-risk patients and in specific high-risk subgroups of patients identified above [Level B]. b) Morbidity in High-Risk Patients Figures 3b-i and Tables 3b-k outline the meta-analysis results for morbidity in high-risk patients and their subgroups. Overall, our pooled analysis of all high-risk groups combined showed that OPCAB significantly reduced risk overall for stroke, myocardial infarction, atrial fibrillation, transfusions, renal dysfunction, inotrope requirement, IABP, and reoperation for bleeding compared with CCAB. No significant difference between OPCAB and CCAB was found for mediastinitis/wound infection or pulmonary complications (including respiratory infections) when all high-risk groups were combined Lippincott Williams & Wilkins 13

12 Puskas et al. Innovations Volume 1, Number 1, Fall 2005 Table 3i. Risk Groups n (N) OPCAB % CCAB % OR [95%CI] NNT p-value [95%CI] I 2 All Hi-Risk Pts 21,667 (27) [ ] LVD 3047 (6) [ ] Elderly 963 (3) [ ] Left Main 1165 (1) [ ] Diabetics 4118 (2) [ ] Renal Dysfunction 253 (2) [ ] Euroscor (4) [ ] Multirisks (4) [ ] Aorta, atheromatous 701 (2) [ ] Redo/Urgent 960 (3) [ ] COPD 0 *Meta-analysis of 45 trials including 26,349 highrisk patients (3 randomized and 42 nonrandomized trials) COPD chronic obstructive pulmonary disease; Euroscore5 High Euroscore/Parsonnet; LOS Length of stay; LVD left ventricular dysfunction; NNT number needed to treat; Redo/Urgent Repeat, urgent, or emergent coronary bypass surgery; WMD weighted mean differencetable 3j: Pulmonary/Respiratory Complications in High-Risk Patients [Level B/A]* Table 3j. Risk Groups n (N) OPCAB % CCAB % OR [95%CI] NNT p-value [95%CI] I 2 All Hi-Risk Pts 16,940 (21) [ ] LVD 632 (3) [ ] Elderly 715 (2) [ Left Main 1165 (1) [ ] Diabetics 5362 (3) [ ] Renal Dysfunction 253 (1) [ ] Euroscor (3) [ ] Multirisks (3) [ ] Aorta, atheromatous Redo/Urgent 72 (1) [ COPD 95 (2) [ ] *Meta-analysis of 45 trials including 26,349 highrisk patients (3 randomized and 42 nonrandomized trials) COPD chronic obstructive pulmonary disease; Euroscore5 High Euroscore/Parsonnet; LOS Length of stay; LVD left ventricular dysfunction; NNT number needed to treat; Redo/Urgent Repeat, urgent, or emergent coronary bypass surgery; WMD weighted mean differencetable 3k: 30-day Mediastinitis/Wound Infection in High-Risk Patients [Level B/A]* In our pooled analysis of individual patient risk groups, OPCAB was associated with lower morbidity when compared to CCAB in patients with the following identifiable risk factors (Tables 3b-k) [Level B]: Euroscore 5: AMI 98,99,102,109,139 Age 75: Stroke, 75,80,90,100,101 AF, 90,96,101,139 Transfusions, 73,75,90,96 IABP 77,96 Diabetes: Stroke, 88,88,95,103 AF 88 Renal Failure: Stroke, 72,75,108 Transfusions, 75,83 Dialysis, 72,75 Inotropes 72 LV dysfunction: Transfusions, 70,73,75,76,78 Renal Dysfunction, 70,93,139 Inotropes 76 Left Main Disease: AF, 79 Transfusions, 78,79,104 Inotropes 78,79,104 Redo or Urgent/Emergent CABG: Renal dysfunction 75, COPD: Transfusions 75 Presence of at least high-risk factor: Stroke, 27,71,73,75,84 86,105,107 AF, 71,84 86,97,107 Transfusions, 27,71,73,75,84 86,105,107 Inotropes, 71,86,105 Reoperation for bleeding 73,84-86,105 ; IABP 71,85,86,105,107 In general, perioperative risks were decreased in OP- CAB versus CCAB, with the possible exception of increased inotrope requirement in the elderly 96 and in those with EU- ROSCORE Statement: OPCAB is associated with reduced morbidity in pooled analysis of high-risk patients and in specific highrisk subgroups of patients identified above [Level B] Lippincott Williams & Wilkins