Coronary artery bypass grafting has evolved into

Transcription

1 J. MAXWELL CHAMBERLAIN MEMORIAL PAPER Does Arterial Revascularization Decrease the Risk of Infarction After Coronary Artery Bypass Grafting? Paul T. Sergeant, MD, PhD, Eugene H. Blackstone, MD, and Bart P. Meyns, MD, PhD Cardiac Surgery Department, Gasthuisberg University Hospital, Leuven, Belgium Background. This study sought to determine whether extensive arterial grafting reduces the prevalence and consequences of infarct after coronary artery bypass grafting. Methods. Post-primary coronary artery bypass grafting infarcts and time-related events thereafter were identified by 99.9% complete follow-up of 9,600 patients (1971 to 1992). The contribution of arterial grafting to freedom from infarct was assessed by multivariable hazard function analysis to adjust for other risk factors. Results. Unadjusted 1-month and 10-year freedom from infarction was 97% and 86%. By multivariable analysis, arterial grafting lowered the prevalence of periprocedural (p 0.005), intermediate term (p and 0.006), and late infarction (arterial grafting to the left anterior descending coronary artery, p ). Unadjusted survival after first infarct after coronary artery bypass grafting was 74% and 52% at 1 and 10 years; arterial grafting improved 10-year survival from 48% to 59% (p 0.002). An additional benefit or cost of extending arterial grafting (n 1,727) beyond a single one could not be identified (p > 0.1). Conclusions. Arterial conduits, particularly to the left anterior descending coronary artery, should be used for coronary artery bypass grafting to reduce early and late myocardial infarction and its consequences. However, use of more than a single arterial graft appears to confer no additional benefit. (Ann Thorac Surg 1998;66:1 11) 1998 by The Society of Thoracic Surgeons Coronary artery bypass grafting has evolved into abundant use of bilateral sequential mammary, radial, and gastroepiploic arteries. The assumption that single arterial grafting to the left anterior descending coronary artery (LAD) [1, 2] improved graft patency and late survival gave rise to the assumption that multiple arterial grafting would contribute further to superior results. However, after adjusting for differences in prevalence of risk factors, we [3] have been unable to identify an additional survival benefit of arterial grafts to other systems. This finding stimulated us to seek possible additional protection of multiple arterial grafting against the morbid event of myocardial infarction after coronary artery bypass grafting (CABG) and its consequences. Material and Methods Patients A consecutive series of 9,600 patients underwent isolated first time CABG at the Gasthuisberg University Hospital of the Katholieke Universiteit Leuven (K.U. Leuven) Belgium, from 1971 to The mean age of the patients increased from years in 1971 to years in Seventeen percent of the patients were women. Single coronary system disease was present in 11%, Presented at the Thirty-fourth Annual Meeting of The Society of Thoracic Surgeons, New Orleans, LA, Jan 26 28, Address reprint requests to Dr Sergeant, Cardiac Surgery Department, Gasthuisberg University Hospital, Herestreet, 3000 Leuven, Belgium. two-system disease in 29%, and three system disease in 60%. Left main coronary artery stenosis of more than 90% stenosis was identified in 355 patients. Mild mitral and aortic valve regurgitation, insufficient to warrant valvular reconstruction or replacement, was present in 643 and 95 patients, respectively. The complete data set has been described and characterized in detail in previous publications [3]. Surgical Technique Intermittent aortic cross-clamping, moderate hypothermia at 28 C, and pretreatment with 1 mg/kg of lidoflazine [4] since 1980, have been the technique of choice for managing the ischemic myocardium during operation. Arterial grafting was started using the internal mammary artery (IMA) in At least one in-situ IMA anastomosis was constructed in 6,074 patients. Bilateral and sequential IMA [5] grafting started in 1973 and 1978, respectively, but the use and the method of use was inconsistent over time. The left IMA was constructed most frequently to the LAD system (n 4,833 patients). Bilateral IMAs were constructed in 207 patients to a single coronary system and in 985 patients to two coronary systems. The distribution of the patients by in-situ IMA distal anastomoses and by age group is presented in Table 1. The number of patients alive, without having suffered an infarct after CABG, within various follow-up intervals, is presented according to the number of in-situ IMA distal anastomoses in Table 2. In addition to the in-situ IMA anastomoses, IMA free grafting and gastro by The Society of Thoracic Surgeons /98/$19.00 Published by Elsevier Science Inc PII S (98)

2 2 CHAMBERLAIN PAPER SERGEANT ET AL Ann Thorac Surg POST-CABG INFARCT AND ARTERIAL GRAFTING 1998;66:1 11 Table 1. Distribution of Patients by Number of In-Situ Internal Mammary Artery Distal Anastomoses and by Age at Operation (excluding the internal mammary artery free and gastroepiploic anastomoses) No. of IMA Distal Anastomoses 60 Years Old 60 to 70 Years Old 70 Years Old 0 2,104 1, ,396 1, IMA internal mammary artery. epiploic artery in-situ grafting were performed in 122 and 43 patients, respectively. In-situ arterial grafting only was achieved in 1,088 patients, of which 442 had multiple system disease. Follow-up The K.U. Leuven Coronary Surgery Database has a core of regular, fully documented, follow-up reports by the referring specialists. These reports document each event, suspicion of event, or regular visit. In addition to these nonsystematic follow-up reports, a formal cross-sectional follow-up was undertaken in 1987 to 1988 and repeated between January 1993 and July The common closing date for outcome information was January 1, All information up to January 1, 1993, was included in the analysis. Follow-up was complete for 99.9% (11 had incomplete follow-up). Median follow-up for the survivors was 6.4 years (range, 30 days to 22 years); average follow-up was years. Event The diagnosis of early perioperative infarct was documented by the intensive care specialist (an independent unit) using repeated surface electrocardiogram (ECG) (Minnesota classification) and routine enzymatic measurements. The cut-off value for positive enzymes was a creatinine kinase MB fraction higher than 8% of total creatinine kinase. The diagnosis of later perioperative infarct was based on repeated routine in-house surface ECG and enzymatic measurements on suspicion of the event. The ECG changes were always compared with earlier ECGs of the patient before confirmation of the new event. Surface ECG by the referring cardiologist (an independent unit) at the first visit after CABG confirmed where appropriate the perioperative diagnosis. The diagnosis of follow-up infarcts was documented by the attending cardiologist. If the ECG changes in follow-up did not correlate with clinical symptoms in the same time frame, the date and time of first occurrence of the ECG changes were considered the relevant date and time for the event. Patients not experiencing the event were censored at the common closing date, at the end of follow-up for those with incomplete follow-up, or at death, whichever occurred earliest. Characterization of Overall Freedom From First Infarct Nonparametric estimates of overall nonrisk-adjusted freedom from first infarct were obtained by the method of Kaplan and Meier [6]. A completely parametric method was used to identify the number of hazard phases, identify the form of the equation for each phase, and estimate the parameters that characterize the distribution of times until first infarct [7]. Multivariable Analysis of Freedom From First Infarct The general methods used to identify incremental risk factors for first infarct have been described previously for the event death [3], including the variables (Appendix 1) and their organization for analysis, and the exploratory analyses accompanying the multivariable analyses, which were conducted in the parametric, multiphase hazard function domain. In a directed stepwise entry of variables into the multivariable risk factor model, a p value criterion of 0.05 was used for retention of variables in the final analysis. Regression coefficients are presented plus or minus one standard error. Goodness of fit of the patient procedure experience model to the data was checked in the two ways previously detailed [8, 9]. SEQUENTIAL ANALYSES. To facilitate the analysis and better discover and understand the nature and influence of single and multiple arterial grafting in the face of change in prevalence of other variables, the risk factor analysis Table 2. Number of Patients at Risk by Year After Operation (alive without having suffered a first infarct after coronary artery bypass grafting) and by Number of In-Situ Mammary Artery Distal Anastomoses Years after Operation Internal Mammary Artery Anastomoses At operation 3,526 4,347 1, ,235 4,119 1, ,121 3,670 1, ,996 3, ,755 2, ,576 2, ,392 1, ,108 1, ,789 1, ,474 1, ,

3 Ann Thorac Surg CHAMBERLAIN PAPER SERGEANT ET AL 1998;66:1 11 POST-CABG INFARCT AND ARTERIAL GRAFTING 3 was conducted in a sequential fashion. First, only patient variables were entered, then procedure variables likely to be known or estimated at the time of decision making were added, including the use of arterial grafts, and finally experience variables. Variables from the isolated and preceding analyses were allowed to enter and leave the model. NATURE AND INFLUENCE OF RISK FACTORS. Exploration of the influence of arterial grafting was performed by constructing a series of nomograms representing solutions of the final parametric equation for all variables (patient, procedure, and experience). These were risk-adjusted in the sense that all other variables in the model were set to those of a patient of median risk in the last year of the study. Thus, each figure represents the risk-adjusted prediction for a specific, but typical, hypothetical patient (Table 3). DEPICTIONS. Graphic depictions in this article have been standardized as follows. Each event in the depiction of freedom from first infarct (Fig 1A) is represented by a circle positioned along the horizontal axis at the time of the infarct and on the vertical axis according to Kaplan- Meier lifetable estimates, enclosed at 2-year intervals between confidence limits equivalent to one standard error. The solid line is the parametric estimate of freedom from first infarct enclosed within dashed 70% confidence limits equivalent to one standard error. The numbers in parentheses represent the number of patients without the event and traced beyond that point. The solid line in the hazard function depiction (Fig 1B) is the parametric estimate of the hazard, enclosed within dashed 70% confidence limits equivalent to one standard error. The solid lines in the patient-specific predictions and nomograms represent specific solutions of the parametric equations, enclosed within dashed 70% confidence limits equivalent to one standard error. The solid lines in the nomograms for difference in predicted percentage freedom from first infarct represent the solutions of the parametric equations, enclosed within dashed 90% confidence limits. Results Nonrisk-Adjusted Freedom From First Infarct A first infarct after CABG was identified in 1,030 patients (the remaining 8,570 patients were censored at death, the common closing date or, rarely, at incomplete follow-up). The nonrisk adjusted 1-month, 1-, 10-, and 15-year freedom (Fig 1A) from first infarct was 97.3%, 96.7%, 86%, and 73%, respectively. A three-phase hazard function was identified (Fig 1B). It consists of an early hazard phase, corresponding to the perioperative period, that rapidly declined to disappearance by the fifth day after operation, a constant hazard phase and a late hazard phase rising gradually beyond 2 years after operation for as long as the extent of follow-up. Arterial grafting was associated with a lower prevalence of infarction after CABG (p ) and this reduction was similar in magnitude for Table 3. Patient Profile of the Median Patient Used for the Nomograms and Patient-Specific Predictions (unless specified differently in the appropriate legend) Variable Values Used Patient Demographic Age 58 Blood group A Height of patient (cm) 169 Anginal status Previous infarct within 30 days of operation Stable angina before operation Yes Clinic or ECG result of preop Yes cycloergometric test Unstable ST segment but not acute infarct Left ventricular function Previous inferior infarct Coronary disease distribution Two or three system disease Yes Higher % stenosis of left main Comorbidity (cardiac) Aortic valve incompetence Concomitant planned pacemaker insertion Comorbidity (vascular) History of peripheral vascular disease History of abdominal aortic disease Cerebral, noncarotid, vessel disease Comorbidity (noncardiac, nonvascular) BUN 50 mg/dl Pulmonary vital capacity as % of 95 normal Grade of diabetes Preoperative cholesterol level (mg/dl) 255 Procedure General technical aspects Patch graft Coronary endarterectomy Revascularization Complete Arterial grafting Absence of at least one arterial graft One arterial graft Absence of at least one arterial graft Condition not and single system disease valid Nonarterial graft to the LAD Arterial graft to LAD Institutional Surgeons Lower risk surgeon for first infarct BUN blood urea nitrogen; ECG electrocardiogram; LAD left anterior descending coronary artery. one, two, three, or four IMA distal anastomoses (p ) for the full extent of their follow-up (Fig 2). Risk-Adjusted Influence of Arterial Grafting on Infarction After Coronary Artery Bypass Grafting The reduction of risk of infarction after CABG by arterial grafting across time, from early to late, continued to be

4 4 CHAMBERLAIN PAPER SERGEANT ET AL Ann Thorac Surg POST-CABG INFARCT AND ARTERIAL GRAFTING 1998;66:1 11 and the practice of coronary endarterectomy (Table 4). Later infarcts were related to the aggressiveness and possible progression of atherosclerotic disease (vessel disease and history of peripheral and cerebral vascular disease), grade of diabetes, and higher cholesterol levels. Fig 1. (A) Parametric freedom from first infarct after primary isolated coronary artery bypass grafting (CABG) (n 9,600). (B) Hazard function for first infarct at any time after primary isolated CABG (n 9,600) (instantaneous risk at every point in time after the operation). Events Before and After First Infarct The first infarct was preceded by return of anginal symptoms in 386 patients (37%) of the total of the patients experiencing an infarct after CABG. After the first infarct and within the follow-up interval after CABG, 400 of the 1,030 patients died. The nonriskadjusted parametric survival after first infarct after CABG at 1 month, 5 years, and 10 years was 79%, 65%, and 52%, respectively (Fig 4). Stratified by the presence of an arterial graft, but unadjusted for any residual variability, the 1-month and 10-year survival after first infarct was 77% and 48% without arterial grafting of any kind and 83% and 59% in the presence of an arterial graft (p 0.002) (Fig 5). An additional benefit of multiple arterial grafting over a single arterial graft was not demonstrated (p 0.6). Return of angina after first infarct after CABG was observed in 499 patients. The non risk-adjusted parametric freedom from angina after first infarct after CABG at 1 month and 10 years was 82% and 27% (Fig 6). A non risk-adjusted benefit (p 0.02) in return of angina after CABG was identified in favor of two in-situ IMA anastomoses versus none or only one (Fig 7). This difference was already active early after the infarct (77% versus 66% freedom from angina at 6 months) and increased with the extent of the follow-up (58% versus 40% freedom from angina at 5 years). Reintervention for ischemic disease, cardiologic or cardiosurgical, after first infarct after CABG was observed in 250 patients. The non risk-adjusted freedom from any reintervention after first infarct after CABG at 1 and 10 demonstrated after adjusting for other incremental risk factors (Table 4). Thus, perioperative infarction was reduced by arterial grafting (p 0.005), as was infarction in general (constant hazard phase, p 0.007). Risk was reduced in one-system disease even more than in multisystem disease (p 0.006). The increasing prevalence of infarcts after the second year, accounting for most of the events after CABG, was reduced by use of arterial grafts to the LAD (p ). Importantly, the benefit of arterial grafting persisted into older age in both multiple (Fig 3) and single-system disease. Neither benefit above that of simply using a single arterial graft, particularly to the LAD, nor increased risk, was demonstrated in the use of multiple arterial grafts (p 0.1). However, before adjusting for other risk factors, apparent benefits of multiple grafting were found (Table 5), none of which persisted after risk adjustment. Other Risk Factors for Infarction After Coronary Artery Bypass Grafting Perioperative myocardial infarction was strongly related to an unstable ST-segment, incomplete revascularization, Fig 2. Actuarial freedom from infarct by number of in-situ mammary artery (IMA) distal anastomoses and uncorrected for any patient-related or other procedural variability. The patients without an arterial anastomosis are represented by circles, those with one arterial anastomosis by squares, those with two by triangles, and those with three by diamonds. (CABG coronary artery bypass grafting.)

5 Ann Thorac Surg CHAMBERLAIN PAPER SERGEANT ET AL 1998;66:1 11 POST-CABG INFARCT AND ARTERIAL GRAFTING 5 Table 4. Incremental Risk Factors, Based on Patient, Procedure, and Institutional Variables, for Freedom From First Infarct After Primary Isolated Coronary Artery Bypass Grafting a Risk Variables Hazard Phase (p Value) Early Constant Late Patient Demographic Age Lower Blood group A 0.01 Higher Height of patient (cm) Lower Height of patient (cm) Anginal status Higher Previous infarct within 30 days of operation Lower Previous infarct within 30 days of operation 0.02 Lower Stable angina before operation Lower Clinic or ECG result of preop cycloergometric test 0.01 Unstable ST segment but not acute infarct Left ventricular function Lower Previous inferior infarct Coronary disease distribution Two or three system disease 0.01 Lower Higher percent stenosis of left main 0.03 Comorbidity (cardiac) Aortic valve incompetence Concomitant planned pacemaker insertion Comorbidity (vascular) History of peripheral vascular disease History of abdominal aortic disease Cerebral, noncarotid, vessel disease Comorbidity (noncardiac, nonvascular) BUN 50 mg/dl Lower Pulmonary vital capacity as percent of normal Higher Grade of diabetes Higher Preoperative cholesterol level Procedure General technical aspects Patch graft Coronary endarterectomy Incomplete revascularization Arterial grafting Absence of at least one arterial graft Absence of only arterial grafts and single system disease Nonarterial graft to the LAD Institutional Surgeons Lower risk surgeon for first infarct a See Appendix 2 for the selected variables, their coefficients, their standard error, and their mathematical transformation. BUN blood urea nitrogen; ECG electrocardiogram; LAD left anterior descending coronary artery. years was 79% and 63% (Fig 8). No benefit of either single or multiple (p 0.3) arterial grafting was demonstrated. Comment Limitations of the Studied Event First infarct after CABG is nearly always a clinical event with electrocardiographic and often hemodynamic stigmata. The time relatedness of the event might be misdocumented over a few hours in the very early postoperative events, it might have been misdocumented over a few weeks in the rare silent late infarcts but the number of events will be reliable because the event will be identified at the next physician s visit. Silent infarcts will nearly certainly have been missed in patients with left bundle-branch block.

6 6 CHAMBERLAIN PAPER SERGEANT ET AL Ann Thorac Surg POST-CABG INFARCT AND ARTERIAL GRAFTING 1998;66:1 11 Fig 3. Nomogram for a median patient (see Table 4 for the patientprofile used) validating the difference in predicted percent freedom from first infarct after coronary artery bypass grafting (CABG) between one and no arterial anastomosis (preferably to the left anterior descending coronary artery) at 5, 10, and 15 years after CABG and for the whole spectrum of age at operation. Limitations of the Data Set All statements related to arterial grafting are limited to the techniques of arterial grafting used during the time frame of the study reported. Sequential and bilateral in-situ IMA anastomoses were considered normal practice. A free IMA graft was only constructed when after planning in-situ grafting, the IMA was found to be too short or was damaged proximally. Only a limited number of patients with in-situ gastroepiploic grafting and none with radial artery grafting were performed during this time. Inferences about multiple arterial grafting are also limited by this technique being used more recently. However, even at 9 years of follow-up, more than 1,000 patients with multiple arterial grafts are alive without infarct, and at 10 years, more than 800. Fig 4. The actuarial survival and superimposed parametric function of survival after first infarct after coronary artery bypass grafting (CABG). Clinical Relevance of the Event for the Patient The clinical relevance of infarct after CABG is increased by the fact that few of the patients received anginal warnings of the impending infarct. The hazard function indicates that monitoring for early infarcts is advisable, with diminishing returns, for about 5 days after operation, and the risk of infarction must be weighted against the benefits of early discharge after CABG. A higher prevalence, early and late, of infarct after CABG was reported in the Bypass Angioplasty Revascularization Investigation trial [10], 80.4% at 5 years. This higher prevalence could have been induced by the absence of single vessel disease, the smaller sample size (n 914), the higher periprocedural infarct rate of 4.6% and the multicenter aspect of the trial. The results of our study are similar to the 10- (89% 2%) and 15-year (77% 3%) freedom given for a smaller group (n 428) of patients but also with very extended and complete follow-up [11], using the anniversary method. This study gives a first insight in several events after infarct. The first infarct after CABG has a 1-month lethal cost of close to 21% (Fig 2) (uncorrected for patient and procedure variability). This mortality is comparable to mortality after infarct without previous CABG, because it includes patients in all grades of cardiac failure at the moment of infarct. The 30-day mortality in the Gruppo Italiano per lo Studio della Streptochinase nel Infarto miocardico 1 trial [12] increased from 7% for Killip class Table 5. Univariate Non Time-Related Comparison of the Number of Events Within the Follow-up Interval by Number of In-Situ Mammary Artery Distal Anastomoses a Internal Mammary Artery Distal Anastomoses No. of Events Total 3,526 4,347 1, Infarct Percent of total a p by 2 analysis. 17% 8% 6% 3% 0% Fig 5. The actuarial survival after first infarct after coronary artery bypass grafting (CABG), stratified by the presence of an arterial graft at the primary procedure.

7 Ann Thorac Surg CHAMBERLAIN PAPER SERGEANT ET AL 1998;66:1 11 POST-CABG INFARCT AND ARTERIAL GRAFTING 7 Fig 6. The actuarial freedom from angina and superimposed parametric function of freedom from angina after first infarct after coronary artery bypass grafting (CABG). I to 81% for Killip class IV. Stratified actuarials explored the relation between some of the patient variability and survival after first infarct after CABG: age more than 65 years of age (p ), insulin-treated diabetes (p ), and ventricular function (p ). Even if the patient survives, quality of further life will be reduced by the loss of ventricular mass and function, concomitant with myocardial cell necrosis. Return of angina was noted fairly rapidly in most patients after their first infarct after CABG. This high rate could have been induced by the use of thrombolysis, often reducing the impact of the infarct but not taking away the cause. Impact of Arterial Grafting on Freedom From First Infarct From the first reports by Kay [13] and Barner [14] and their colleagues emphasizing the good patency rates after direct anastomosis between an internal mammary artery graft and a coronary artery, surgeons have been increasing their expectation in this graft. This was further Fig 7. The actuarial freedom from angina after first infarct after coronary artery bypass grafting (CABG) stratified for 0 or 1 internal mammary artery (IMA) anastomosis versus more than 1 IMA anastomosis. Fig 8. The actuarial freedom from any cardiologic or cardiosurgical reintervention for ischemic disease after first infarct after coronary artery bypass grafting (CABG). increased after Siegel and Loop [15] reported higher patency rates compared to saphenous vein grafts, certainly when this was correlated with improved survival [16]. Tector and coworkers [17] insisted in 1983 to use this arterial conduit for grafting the anterior descending coronary artery. Several investigators reported reduced early postoperative infarction rates in the presence of arterial or more extensive arterial grafting, sometimes without [18], sometimes after correction [19, 20] for patient variability. These findings were confirmed in this study with the first-month reduction of freedom from infarct for the median patient from 97% to 98% by having at least one arterial graft, preferably to the LAD. The additional early benefit of increasing the number of arterial grafts above a single one was lost after correction for patient variability. Loop and colleagues [1] demonstrated a better late cardiac event-free survival after a left IMA anastomosis, versus a saphenous vein, to the LAD. These findings are confirmed in this study for the event infarct, after correction for patient variability, and are quantified for a median patient (see Table 3) in Figure 3. This benefit is increasing over time and as far as the extent of the follow-up. An additional benefit is identified in the interaction between single vessel disease and the presence of only arterial grafts. This benefit is active over the entire follow-up interval but most visible between a few days and 2 years after operation, because this is the interval when the constant phase is proportionally the most important. The positive findings for a single mammary artery graft prompted surgeons to use both arteries. In a casematched study with a follow-up of 15 years, Fiore and colleagues [21] observed, a significant (p 0.02) reduction (59% to 75% freedom) of late infarcts in increasing the number of mammary artery grafts to two. But the curves for both groups crossed one another at 10 years after operation. These findings were not reconfirmed in this study after correction for patient variability. A median patient (Table 3) improves freedom from infarct at 10 years from 86% to 91% by having one arterial graft,

8 8 CHAMBERLAIN PAPER SERGEANT ET AL Ann Thorac Surg POST-CABG INFARCT AND ARTERIAL GRAFTING 1998;66:1 11 preferably to the LAD. A similar additional increase with a second arterial graft is very unlikely because it would annihilate all other influences. The absence of additional advantage of a second or third IMA anastomosis could have been foreseen as the first IMA anastomosis was most frequently selected for the supply of the left anterior descending artery, the dominant artery responsible for the preservation of the ventricular function and capable of delivery of collateral flow. Impact of Arterial Grafting on Events After First Infarct After Coronary Artery Bypass Grafting In addition to a first insight in the prevalence and time structure of events after-infarct, this study has explored the influence of arterial grafting on these events. An arterial graft protected the patient s survival after the first infarct after CABG (Fig 5). No additional benefit of single arterial grafting was documented in return of postinfarct angina or reintervention. Double mammary artery reconstruction reduced return of angina after CABG (Fig 7), certainly versus a single or no arterial graft. Further research is warranted to determine whether these benefits remain active after adequate correction for some of the identified patient variability. Clinical Inferences The first infarct after CABG is a rare event with major clinical relevance for the patient. Patients in whom the demographic, cardiac or noncardiac comorbidity reduces the 10-year survival are unlikely to suffer the event. Only one-third of the patients is warned by preinfarct angina. The occurrence of this first infarct after CABG should be avoided by all therapeutic and procedural means, as well during the primary CABG procedure as in follow-up. Arterial grafting, certainly to the LAD, seems one of these therapeutic possibilities. No increased risk or benefit was identified with more extensive or complete arterial grafting in any hazard phase. The age of the median coronary bypass patient in 1998 has reached or exceeded 70 years in most surgical programs. The comorbidity has risen in parallel. Therefore, it is unlikely that more extensive arterial revascularization will ever reduce the prevalence of first infarct after CABG due to the rarity of the event, the limited life expectancy of the median patient, and the not yet identifiable benefit of this extensive arterial grafting. We thank Robert Brown (UAB) for his unsurpassed expertise in handling the most complex database, analytic, and graphic requests. References 1. Loop FD, Lytle BW, Cosgrove DM, et al. Influence of the internal mammary artery graft on 10-year survival and other cardiac events. N Engl J Med 1986;314: Sergeant PT, Lesaffre E, Flameng W, Suy R. Internal mammary artery: methods of use and their effect on survival after coronary bypass surgery. Eur J Cardiothorac Surg 1990;4: Sergeant PT, Blackstone EH, Meyns B, K.U. Leuven Coronary Surgery Program. Validation and interdependence with patient-variables of the influence of procedural variables on early and late survival after CABG. Eur J Cardiothorac Surg 1997;12: Flameng W, Borgers M, Van der Vusse G, et al. Cardioprotective effects of lidoflazine in extensive aorta coronary bypass grafting. J Thorac Cardiovasc Surg 1983;92: Sergeant P, Flameng W, Suy R. Internal mammary artery jumpgraft. J Cardiovasc Surg 1988;29: Kaplan E, Meier P. Non-parametric estimation from incomplete observations. J Am Stat Assoc 1958;53: Blackstone EH, Naftel D, Turner M. The decomposition of time-varying hazards into phases, each incorporating a separate stream of concomitant information. J Am Stat Assoc 1986;81: Blackstone EH, Kirklin JW. Recommendations for prophylactic removal of heart valve prostheses. J Heart Valve Dis 1992;1: Ferrazi P, McGiffin DC, Kirklin JW, et al. Have the results of mitral valve replacement improved? J Thorac Cardiovasc Surg 1986;92: The Bypass Angioplasty Revascularisation Investigation (BARI) Investigators. Comparison of coronary bypass surgery with angioplasty in patients with multivessel disease. N Engl J Med 1996;335: Van Brussel BL, Plokker HW, Voors AA, et al. Multivariate risk factor analysis of clinical outcome 15 years after venous coronary artery bypass graft surgery. Eur Heart J 1995;16: Gruppo Italiano Per Lo Studio Della Streptochinasi Nell Infarto Miocardico (GISSI). Effectiveness of intravenous thrombolytic treatment in acute myocardial infarction. Eur Heart J 1990;11(Suppl B): Kay EB, Naraghipour H, Beg RA, et al. Internal mammary artery bypass graft long-term patency rate and follow-up. Ann Thorac Surg 1974;18: Barner HB, Mudd JG, Mark AL, et al. Patency of internal mammary coronary grafts. Circulation 1976;54(Suppl): Siegel W, Loop FD. Comparison of internal mammary artery and saphenous vein bypass grafts for myocardial revascularization: exercise test and angiographic correlation. Circulation 1976;54(Suppl 3): Loop FD, Irarrazaval MJ, Bredee JJ, et al. Internal mammary artery graft for ischemic heart disease. Effect of revascularization on clinical status and survival. Am J Cardiol 1977;39: Tector AJ, Schmahl TM, Canino VR. The internal mammary artery graft: the best choice for bypass of the diseased left anterior descending coronary artery. Circulation 1983;68 (Suppl 2): Tector AJ, Kress DC, Schmahl TM, et al. T-graft: a new method of coronary arterial revascularization. J Cardiovasc Surg (Torino) 1994;35(6 Suppl 1): Grover FL, Johnson RR, Marshall G, et al. Impact of mammary grafts on coronary bypass operative mortality and morbidity. Ann Thorac Surg 1994;57: Del Rizzo DF, Fremes SE, Christakis GT, et al. Coronary bypass with arterial conduits. Cardiovasc Surg 1998;6: Fiore AC, Naunheim KS, Dean P, et al. Results of internal thoracic artery grafting over 15 years: single versus double grafts. Ann Thorac Surg 1990;49: Appendix 1. Variables Considered in the Multivariable Analysis of Infarct After Coronary Artery Bypass Grafting Patient Variables demographics. Sex; age (years) at operation; weight; height; body surface area; body mass index; weight height ratio, difference and ratio of weight to ideal body weight where ideal weight

9 Ann Thorac Surg CHAMBERLAIN PAPER SERGEANT ET AL 1998;66:1 11 POST-CABG INFARCT AND ARTERIAL GRAFTING 9 is (height in cm minus 100) for men and (height in cm minus 110) for women; blood group; rhesus factor. preoperative rhythm disturbances. Atrial fibrillation; right hemiblock, left partial or total hemiblock (and either); recent ventricular tachycardia or ventricular fibrillation or intractable ventricular tachycardia or repeated ventricular fibrillation; permanent pacemaker implanted before or at the coronary artery bypass grafting. previous procedures. Percutaneous transluminal coronary angioplasty (PTCA) unsuccessful PTCA, successful PTCA, number of previous PTCAs, interval since last PTCA, duration of freedom from angina after last PTCA; previous noncarotid vascular operation, previous carotid endarterectomy, or either. acute myocardial infarction. Interval between infarct and operation; location of infarct (coronary distribution); use of preoperative thrombolytic therapy; interval since last thrombolytic therapy. hemodynamic status. Cardiogenic shock without or with cardiopulmonary resuscitation; hemodynamic instability (0 stable, 1 unstable on medication but not in shock, 2 shock, 3 shock with cardiopulmonary resuscitation; unstable (yes/no), and graded with states 2 and 3 combined); New York Heart Association class (limitation by either angina or heart failure); increment of limitation by heart failure above that of angina (0 none, 1 mild, 2 moderate, 3 severe). symptoms of reversible ischemia. K.U. Leuven Angina Class (0 mild, 1 mild symptoms, 2 symptoms with normal activities, 3 severe with symptoms even at rest, 4 unstable angina); Canadian Angina Class (0 4, and augmented to grade 5 by unstable angina); duration of anginal symptoms; angina at rest; unstable angina; treatment requirement for unstable angina (0 no unstable angina, 1 unstable angina controlled by intravenous medication, 2 changing ST segment in the hours before operation despite maximal intravenous medication. exercise testing. Positive exercise test by electrocardiogram or on clinical grounds, clinically positive test; electrocardiogram positive test. distribution of coronary artery disease (limited to those predictable at the time of decision making). Number of coronary systems diseased (70% diameter reduction or more), one, two, or three system disease, left main stenosis (percent, 50%, 70%, and 90%). left ventricular function. Ejection fraction; left ventricular end-diastolic pressure; number of previous myocardial infarcts; left ventricular hypertrophy on electrocardiogram; graded (subjective) ventricular dysfunction (0 none, 1 mild, 2 moderate, 3 severe). coexisting conditions (cardiac). Ischemic mitral incompetence (insufficient to require surgical intervention); aortic valve stenosis; aortic valve insufficiency. coexisting vascular disease. Abdominal aortic disease; peripheral vascular disease; cerebral vascular disease; carotid artery disease; internal carotid artery occlusion, internal carotid artery percent stenosis; 80% to 99% uni- or bilateral internal carotid artery stenosis, previous history of vascular operation, previous history of carotid operation, history of stroke; history of transient ischemic attack, calcification of ascending aorta. hyperlipidemia. Cholesterol level; HDL level; low-density lipoprotein level; triglyceride level. coexisting conditions (noncardiac). Diabetes (graded as 0 none, 1 oral treatment, 2 insulin treatment, and each grade separately); overweight ( 10 kg above calculated ideal weight, based on sex and height as defined under demographic variables above); difference from ideal weight; ratio of actual weight versus ideal weight; hypertensive (systolic blood pressure 160 mm Hg or diastolic pressure 100 mm Hg, or on antihypertensive medication); current or history of malignancy; on dialysis; history of severe renal failure; history of renal transplantation; history of nephrectomy; history of hepatic disease (hepatitis or clinical hepatic dysfunction); history of smoking; family history of ischemic heart disease; pulmonary disease (legally confirmed incapacitating anthracosilicosis, important reduction in diffusing capacity, chronic obstructive pulmonary disease, asthmatic bronchitis), chronic obstructive pulmonary disease, pulmonary function (forced vital capacity volumes and 1-second expiratory volumes, these normalized to gender height and age); psychiatric history. Procedure Variables (limited to those predictable at the time of decision making) bypassing conduit. Use of saphenous vein conduits (solely or with other conduits), number of saphenous vein conduits used, number of vein graft distal anastomoses, location of coronary systems to which vein grafts are anastomosed; use of internal thoracic artery (solely or with other conduits); number of internal mammary artery arteries used; number of internal mammary artery distals; location and number of coronary systems receiving internal mammary artery conduits; gastroepiploic artery as the conduit; number of distals; and location of systems receiving this conduit; total arterial conduits, distals and location of anastomoses; end-to-end grafts used; prosthetic conduits used; free internal mammary artery used; ratio of total conduits to total distals, and by extent of coronary disease; ratio of arterial conduits to total arterial distals, and by extent of coronary disease; ratio of nonarterial conduits to total nonarterial distals, and by extent of coronary disease; the largest number of distals on an arterial graft, largest number of distals on a nonarterial graft; patch grafts used; completeness of revascularization. quality of distal vessels. Proportion of distal anastomoses to small coronary arteries (1 mm or less in size). concomitant procedures. Repair of abdominal aortic aneurysm; carotid endarterectomy; plication or resection of left ventricular aneurysm (limited anterior or apical). Institutional Experience Variables Each surgeon; inexperienced surgeons; surgeon s experience (number of isolated coronary cases operated on previous to that patient); surgeon s routine (number of isolated coronary cases operated on in the previous calendar year); date of operation (number of years since 1971); month of operation. Appendix 2. Parameter Estimates, Selected Variables, Their Mathematical Transformation, Their Coefficients, and Their Standard Error for the Final Model Early Phase Intercept ; Delta 0; Rho ; Nu ; M 0 Height of the patient (inverse transformation, 170 divided by length in cm) ; Unstable ST-segment but not acute infarct ; Infarct within 30 days before

10 10 CHAMBERLAIN PAPER SERGEANT ET AL Ann Thorac Surg POST-CABG INFARCT AND ARTERIAL GRAFTING 1998;66:1 11 operation ; Preoperative inferior infarct ; History of peripheral vascular disease ; Preoperative pulmonary vital capacity as a percent of normal ; Presence of an arterial graft ; Coronary endarterectomy performed ; Planned concomitant pacemaker insertion ; Incomplete revascularization ; Low-risk-for-infarct-return surgeon Constant Phase Intercept E-06 Height of the patient (inverse transformation, 170 divided by length in cm) ; Presence of an arterial graft ; Incomplete revascularization ; Age at operation (squared 50 divided by age at operation) ; Stable angina before operation ; Clinical or electrocardiographical positive stress test before operation ; Percent left main coronary artery stenosis ; Preoperative aortic valve incompetence ; Preoperative history of abdominal aortic disease ; Preoperative blood urea nitrogen higher than 50 mg/dl ; Presence of a patch graft ; One system disease and no arterial grafts constructed Late Phase Intercept E-07; Tau 1; Gamma ; Alpha 1; Eta 1 Infarct within 30 days before operation ; History of peripheral vascular disease ; Preoperative pulmonary vital capacity as a percent of normal ; Height of the patient ; Blood group A ; Two or three coronary system disease ; Cerebral (noncarotid) vessel disease ; Preoperative cholesterol value (inverse transformation) ; Higher grade of diabetes (grade 0 no diabetes; grade 1 abnormal glucose tolerance test with dietary treatment; grade 2 diabetic, receiving oral hypoglycemic treatment; grade 3 insulin-treated diabetes) (the squared transformation is used) ; Nonarterial graft to the body of the left anterior descending coronary artery DISCUSSION DR ALFRED J. TECTOR (Milwaukee, WI): My compliments to Dr Sergeant and his colleagues on their wonderful study and their 99% follow-up of 9,600 patients who had coronary bypass grafting over a 21-year period. Their completeness of follow-up and their meticulous and extensive statistical analysis certainly merits the J. Maxwell Chamberlain award, and I congratulate them for their achievement. I am sure many of you in the audience share my envy of their K.U. Leuven database. I would also like to thank them for sending me a copy of their manuscript and their later revisions. The main goals of the ideal coronary bypass graft operation are to prolong survival to its fullest extent, minimize the recurrence of coronary events, and reduce or abolish the need for further intervention such as angioplasty, stenting, and repeat coronary artery bypass grafting. Ultimately this should be the only procedure the patient will need for the remainder of his or her life. One point of interest is the devastating effect of perioperative infarct on the outcome of patients early and later after their operation. We, as cardiovascular surgeons, should strive to eliminate perioperative and myocardial infarctions. Early graft failure from spasm, technical errors, and intimal hyperplasia is a major cause of perioperative myocardial infarction. The properly prepared and anastomosed internal thoracic artery is less likely to fail early than the saphenous vein. We can assume failure of one of the bypass grafts was responsible for the perioperative myocardial infarction in many but not all of their patients. Likewise, the later myocardial infarctions occur most frequently from obstruction of bypass grafts, and 15% to 20% occur from new disease in one of the coronary arteries. In the group of patients who had one or more arterial grafts as well as one or more saphenous vein grafts, there is no mention in the article of any attempt to identify the site of infarction or the presence of a culprit graft. Other procedures such as angioplasty, stents, repeat coronary artery bypass grafting are more frequently necessary for failing saphenous vein grafts than for failing arterial conduits 5 or more years after the operation. These procedures will reduce the number of myocardial infarctions from failing bypass grafts and could bias the benefits the patients receive from multiple arterial grafting. I have three questions for the investigators. Have they tabulated the number of patients who did not have a myocardial infarction in their constant or late phase after their operation but did have another invasive procedure like angioplasty, stenting, or repeat coronary artery bypass grafting? The second question is, during the follow-up period has there been any attempt to identify the culprit graft by postoperative angiograms or by localizing the anatomic site of the infarction and its supplying coronary artery from the electrocardiogram? And my last question is, do they believe total bypass grafting with only arterial conduits could alter the incidence of myocardial infarction at 1 month, 1, 5, 10, and 15 years postoperatively? Again, I compliment Dr Sergeant and associates on their marvelous report. DR HENDRICK B. BARNER (St. Louis, MO): Benefit from arterial conduits is based on their distribution and continued patency. We now know that 10-year survival advantage is realized when the second internal thoracic artery (ITA) is placed to the circumflex system rather than to the right coronary as reported in November by James Jones of Baylor. Our experience from 1972 to 1974 indicates that the right ITA grafts to the right coronary must be placed to the posterior descending artery to potentially provide long-term benefit because of distal disease developing in the right coronary. Did the investigators only use the right ITA as an in-situ graft, which will seldom reach the posterior descending artery? The Mayo Clinic data, and ours, demonstrate a reduction in the occurrence of myocardial infarction during 10 or 15 years of follow-up when two rather than one ITAs are used. I have long admired the statistical prowess of Paul Sergeant and Gene Blackstone, but I have concerns about their use of arterial conduits. We know from their 1990 publication that their experience with bilateral ITAs was small, at 276, and that in a large number of these the second ITA was grafted to the diagonal artery when the first was grafted to the left anterior descending (a configuration I have never used). Even if their subsequent use of the second ITA has been appropriate, this would not provide 10- and 15-year data, which are the essence of

11 Ann Thorac Surg CHAMBERLAIN PAPER SERGEANT ET AL 1998;66:1 11 POST-CABG INFARCT AND ARTERIAL GRAFTING 11 this report. I would submit that use of both ITAs for the left anterior descending coronary artery system is inappropriate. Patients having two ITAs grafted to one coronary system should be considered as having only one ITA graft in this analysis. Because benefit has been realized by grafting one ITA to the left anterior descending (arguably the most important coronary), it is clear that the second ITA must be placed to the next most important coronary with regard to the mass of viable myocardium subtended by that artery. Doctor Sergeant, do you believe you have achieved this in your database? If not, your conclusions must be viewed with reservation. DR CONSTANTINE E. ANAGNOSTOPOULOS (New York, NY): I also rise to compliment the investigators, whose original paper I heard at the meeting of The American Association for Thoracic Surgery. The questions that I have echo Dr Barner s comments. Namely, does the patency of the second ITA depend on precisely how and where it is used? And how many pedicle versus free right ITAs does the study contain? We, at St. Luke s Hospital in New York, follow the lead of Dr George Green and Dr Daniel Swistel by performing free right ITA bypass to the circumflex system. The series now exceeds 3,000 patients operated on by all three teams. DR SERGEANT: I thank the three discussants, to whom I have been indebted for part of my training. I had the pleasure 20 years ago to see Mr Tector use the mammary arteries, and this has impressed me for the rest of my life. He has posed to me three questions. The first question is, have we tabulated the relationship with the reinterventions after operation? We have made an analysis similar to this one on the prevalence of reintervention, whether it be a cardiologic or a cardiosurgical reintervention, and we have presented that at the last European meeting in Copenhagen. The article is in the review process and will be out very shortly. Have we identified the culprit graft in the follow-up angiograms? The database is much more complete than we can show in this presentation, and every postoperative angiogram any of these 9,600 patients had, wherever in the world, is stored in the database by patency of the graft. Such a complex analysis has not been looked at yet, but we do have the information. And so to answer your question directly, we have not as of yet looked at that particular problem. Do we believe something like total arterial grafting would reduce the incidence of return of infarct? There is a series of variables expressing the use of arterial grafting. One of the variables is total arterial grafting in the presence of single-vessel disease, two-vessel disease, and three-vessel disease. So a series of variables is proposed for the multivariable analysis. Up to this time frame, we have not been able to identify this variable as being related to a reduction of the infarct rate. Therefore, I can only express my comments in this time frame. And for sure, the comments of the discussants are well taken, we can only make conclusions on the method of use of the arterial grafting as done in this article. They are described much more in detail in the article itself. To respond to Henry Barner, who is, I believe, a friend of mine, and we are very grateful of his comments. He is correct. In the beginning we preferred in-situ mammary artery grafting versus free mammary artery grafting; therefore, any free mammary artery grafting has been done when double mammary artery grafting was not feasible because of the length of the mammary artery pedicle. In this data set, at the time of operation, 207 patients had double mammary artery grafts to one single area, meaning diagonal or left anterior descending coronary artery. But 987 patients at the start of the follow-up had a double mammary artery graft to two different areas, whether it be anterior and right, anterior and lateral, or any other combination. After 1987, when we found in our first analysis that the double mammary artery graft did not increase the benefit, we expanded this even more and made it nearly a standard rule not to use it anymore to the same area. But we have indeed changed our practice, and this is why we have presented this table of the change in use over time. The time relationship aspect is an important issue. At 10 years after operation we still have, with two mammary artery anastomoses, 109 patients alive who have not suffered the event. Therefore, this is 8 years into the late phase. And normally these strengths should have shown up already if they have any considerable value. To comment on Mr Anagnostopoulos remarks, we have reduced the practice of free mammary artery grafting and only used it when it could not be avoided. But in this data set there are 102 patients with additional free arterial grafting and 43 patients with gastroepiploic artery grafting. You must understand that this data set was closed in The common closing date was in It took 2 years to complete the follow-up, as you might expect, and certainly with the secondary and tertiary events, and it has taken close to 2 to 3 years of analysis, so this is the reason why these numbers still look so small.