Coronary artery disease (CAD) remains the single largest killer of Americans, accounting for almost half a million deaths per year. It imposes a particular burden on the elderly, with more than 80% of all CAD deaths occurring in those over age 65.¹ The magnitude of this impact takes on great significance because it is expected that the number of Americans older than 65 years of age will more than double over the next four decades.² If you add to this aging population the anticipated increase in the prevalence of important risk factors for CAD such as diabetes mellitus (DM) and obesity, the population at risk for CAD can only be expected to increase.

Myocardial revascularization represents an effective treatment strategy shown to prolong survival. Techniques of revascularization include percutaneous coronary intervention (PCI) and coronary artery bypass graft (CABG) surgery, which can be performed with or without cardiopulmonary bypass. Current techniques for CABG can be carried out with low perioperative morbidity and mortality, with excellent long-term outcomes despite an increasing risk profile.³ CABG surgery with cardiopulmonary bypass remains the standard by which the other techniques (ie, PCI, off-pump CABG, robotic CABG, hybrid revascularization) are measured.^4,5 It is expected that it will continue to be a cornerstone in the management of CAD in the foreseeable future.

The modern era of myocardial revascularization with cardiopulmonary bypass began in 1954 when Dr. John Gibbon reported the development of the cardiopulmonary bypass machine.⁶ An additional seminal advance occurred with the development of coronary angiography by Mason Sones at the Cleveland Clinic in 1957, which opened the door to the elective treatment of coronary atherosclerosis by means of direct revascularization.⁷ Initial reports by Rene Favaloro and Donald B. Effler on their techniques to treat clinical events associated with stenotic lesions of the coronary arteries culminated in the first large series of aorto-to-coronary artery venous grafts reported in 1969.⁸ Simultaneously Dudley Johnson of Milwaukee published a series of 301 patients in 1969.⁹ The success of these techniques was soon demonstrated in larger series initiating the modern era of coronary artery surgery.

The practicing cardiac surgeon is confronted with no clinical question more often than, “Is coronary bypass indicated in this patient?” In brief, the indications established by the American College of Cardiology Foundation and the American Heart Association (ACCF/AHA) consensus panel are based predominantly on the results of trials comparing surgical revascularization with medical therapy for patients with chronic stable angina.¹⁰ Three major historical trials, the Coronary Artery Surgery Study (CASS), the Veterans Administration Coronary Artery Bypass Cooperative Study Group, and the European Coronary Surgery Study (ECSS), demonstrated the greatest survival benefit of revascularization to be among those patients at highest risk of death from the disease itself as defined by the severity of angina and/or ischemia, the number of diseased vessels, and the presence of left ventricular dysfunction.^11-13Even as the technology and methods for all manner of revascularization strategies improve, the principles established by these seminal trials continue to serve as the basis upon which the results of newer trials designed to study specific populations, clinical scenarios, and anatomic patterns of disease are compared and interpreted.

Clinical and Laboratory Assessment of Coronary Artery Disease

A surgeon’s first introduction to a patient with CAD is frequently a conventional coronary angiogram. Indeed most studies of coronary revascularization have stratified risk and grouped patients according to the number and distribution of coronary lesions. For the purposes of this discussion, and in an effort to be consistent with the large body of literature upon which the current ACCF/AHA guidelines are based, “significant” stenosis is defined as ≥70% reduction in luminal diameter of a coronary artery other than the left main coronary artery (LMCA) for which ≥50% reduction is considered significant.¹⁰ A major development in the last decade in assessing the severity of coronary disease has been the introduction of the SYNTAX score,¹⁴ which is a pure angiographic score of severity and complexity that has recently been validated to correlate with 3-year outcomes in patients with triple vessel and/or main stem disease. It uses dedicated computer software to assign points to each lesion based on length, degree of stenosis, and relation to areas of bifurcation/trifurcation, such that the higher the overall score, the more complex the coronary disease. Although the anatomic distribution and severity of hemodynamically significant coronary lesions determines, in large part, the optimal therapy (CABG, PCI, or medical therapy), the clinical presentation and results of noninvasive studies of myocardial perfusion and function are necessary to characterize the pathophysiologic implications of the angiographic disease and its impact on prognosis and, therefore, to make a clinically appropriate recommendation. In the technological era in which we practice, the importance of the clinical history bears emphasis, particularly in an aging population. Because one of the objectives of surgery is to improve symptoms and quality of life, a thorough appreciation of the patient’s functional status is a prerequisite in selecting the optimal therapeutic strategy.

The system proposed by the Canadian Cardiovascular Society for grading the clinical severity of angina pectoris is widely accepted (Table 20-1). Unfortunately, angina is a highly subjective phenomenon for both patient and physician, and prospective evaluation of the assessment of functional classification by the CCS criteria has demonstrated a reproducibility of only 73%.¹⁵ Furthermore, there may be a strikingly poor correlation between the severity of symptoms and the magnitude of ischemia, as is notoriously the case among diabetic patients with asymptomatic “silent ischemia.”

Electrocardiography (ECG), if abnormal, is helpful in assessing ischemic burden. Unfortunately, it demonstrates no pathognomonic signs in half of patients with chronic stable angina. Still the monitoring of an ECG under stress conditions is simple and inexpensive, and is therefore useful as a screening examination. Among patients with anatomically defined disease, stress ECG provides additional information about the severity of ischemia and the prognosis of the disease. The sensitivity of the test increases with age, with the severity of the patient’s disease, and with the magnitude of observed ST-segment shift.¹⁶ If ST-segment depression is greater than 1 mm, stress ECG has a predictive value of 90%, whereas a 2-mm shift with accompanying angina is virtually diagnostic.¹⁷ Early onset of ST-segment depression and prolonged depression after the discontinuation of exercise are strongly associated with significant multivessel disease. Unfortunately, many patients cannot achieve their target heart rates owing to beta blockade or a limitation to their exercise tolerance caused by coexisting disease, decreasing the usefulness of this test in these often high-risk patients. Resting abnormalities in the ECG may also limit the predictive accuracy of the test.

Perfusion imaging with thallium-201 or a technetium-99m–based tracer may be particularly useful in patients with abnormalities on their baseline ECG. Reversible defects demonstrated by comparison of images obtained after injection of the tracer at peak stress with rest images is indicative of ischemia, and hence viability. An irreversible defect indicates a nonviable scar. The results obtained with both tracers are similar, with the average sensitivity around 90% and specificity of approximately 75%.¹⁸ For patients unable to exercise, pharmacologic vasodilators such as adenosine or dipyridamole may be used with similar sensitivity.¹⁹

Echocardiographic imaging during exercise or pharmacologic stress has gained increasing popularity among cardiologists. Comparative studies have demonstrated accuracy similar to that of nuclear studies with sensitivity and specificity both around 85%.¹⁸ Patients unable to exercise may be stressed with high-dose dipyridamole, or more commonly dobutamine at doses from 5 to 40 μg/kg/min. An initial augmentation of contractility followed by loss or “drop out” is diagnostic of ischemia (and accordingly viability), whereas failure to augment contractility at low dose suggests scar. Additionally, information regarding concomitant valvular disease may be obtained during the examination.²⁰

Guidelines for Revascularization

In any discussion regarding the discrete indications for coronary revascularization, it is critical to bear in mind that the overarching goal of revascularization is to either improve symptoms, prolong survival, or, inasmuch as is inextricably tied to both of these goals, to delay and prevent the complications of coronary disease in order to provide improved quality of life over the period of prolonged survival. With this fundamental principle in mind, the guidelines for surgical revascularization established by the ACCF/AHA were significantly revised in terms of organizational structure in 2011 to reflect this understanding (Tables 20-2 and 20-3).¹⁰ These guidelines outline the recommendations regarding whether revascularization is indicated by either CABG or PCI in comparison to medical therapy in specific subsets of patients grouped primarily by the goal of revascularization (either improvement of survival or improvement of symptoms). Indications are then only secondarily stratified by the anatomic pattern and severity of coronary lesions, taking into account the number of involved vessels, involvement of the LMCA, the proximal left anterior descending artery (LAD). Only after these two distinctions are made are the effects of baseline clinical scenario (eg, presence of diabetes, severity of left ventricular dysfunction, presence of unstable angina, non-ST elevation and ST elevation myocardial infarction [MI], and the calculated operative mortality, among others) taken into account to determine the recommendations for specific situations.^21,10 The basis for these guidelines resides in the large body of literature comparing medical therapy with CABG and PCI in patients with stable angina.

Recommendations are categorized into classes whereby Class I denotes that the benefit exceeds the risk and the procedure/treatment should be performed, Class IIa denotes that the benefit likely exceeds the risk and that it is reasonable to administer the treatment, Class IIb may be equivalent to or exceed the risk but additional studies are necessary to confirm, and that the treatment may be considered. Finally Class III represents a recommendation to not perform the procedure or treatment as it is known to have no benefit and in some cases be harmful.¹⁰ Each recommendation is paired with a description of the strength of evidence available to support the recommendation, and ranges from Level A, denoting data derived from multiple randomized controlled trials or meta-analyses, Level B which denotes data from a single randomized trial or nonrandomized studies, and Level C which represents data from just consensus opinions from experts, case studies, and or previously established standard of care.¹⁰

The complexity of the guidelines reflects the overall complexity and heterogeneity of the patient population presenting with CAD. With the major components of the guidelines summarized at face value in Tables 20-2 and 20-3, this section of the chapter aims to simultaneously deepen and simplify the understanding of these guidelines by (1) providing the historical background from seminal randomized controlled trials and registry studies that provided the basis for how to assign patients with stable angina to surgical revascularization versus PCI or medical therapy, and (2) provide an update on the more recent trials that inform the current guidelines in order to provide a framework for how to approach a patient presenting with CAD today.

APPROACH TO THE LITERATURE

Before reviewing the results of the seminal trials of CABG versus medical therapy performed in the 1970s and those of newer prospectively randomized trials comparing the results of surgery with PCI and medical therapy, some limitations of these trials and of those that followed must be recognized. First, in retrospective or registry studies, it is difficult to ensure comparable patient populations by virtue of the extraordinary anatomical and physiologic complexity of CAD as well as the heterogeneity of the patient substrate. Differences in ventricular function and comorbidities such as age, diabetes, peripheral vascular disease, and pulmonary disease may have a profound impact on outcomes such as survival or quality of life. For example, caution must be exercised in interpreting the results of nonrandomized and registry reports of PCI versus surgery, because the patients subjected to the former more often have one- or two-vessel disease,^22,23 whereas the latter commonly have three-vessel or left main disease.^22,23 Attempts to correct for selection bias with statistical techniques such as propensity matching are only as valid as the parameters entered into the model. Data less tangible than gender or chronologic age, such as socioeconomic status or “physiologic age,” are not easily accounted for in such analyses and yet may be a critical determinant of outcomes. Despite this limitation, retrospective and registry data provide a better glimpse of the real world of CAD. Most prospective randomized trials include only a fraction of the total population undergoing revascularization by virtue of strict entry criteria. For example, the Bypass Angioplasty Revascularization Investigators (BARI) trial entered only 5% of total patients screened.²⁴ Although more recent randomized controlled trials of PCI versus CABG have had better yield (41-71%), these studies have focused on specific subset populations including patients with diabetes, and those with three-vessel and/or left main coronary disease.^4,25,26 Therefore, although the prospectively randomized studies do provide objective data directly applicable to the specific patient subset represented in the study, extrapolation of the results to the more heterogeneous populations seen clinically can only be made if the implicit caveats are clearly understood. We must be mindful of the inescapable tradeoff between selection bias in registry studies and entry bias in randomized studies.

Second, as a consequence of the overrepresentation of patients at lowest risk of death in randomized trials, most are statistically underpowered with respect to survival analysis. For example, given current survival statistics, we would need approximately 2000 patients in each arm of a study to detect a 30% difference in mortality. The problem is compounded by the exclusion of the very patients for whom one would anticipate a survival advantage with adequate revascularization, such as those with depressed ventricular function. Randomized studies frequently employ softer endpoints such as angina or quality of life, or create composites of qualitatively different endpoints, such as death, stroke, and MI. Meaningful analysis is further complicated by relatively short-term follow-up in most studies. Events such as the need for subsequent revascularization and recurrence of angina characteristically occur at different time intervals after these therapies (restenosis after PCI vs graft occlusion after CABG), and an 8- to 10-year follow-up period is needed to adequately compare long-term results. Patients themselves are also generally interested in outcomes measured in years, not months.

Significant improvements in each of the treatment strategies for CAD are occurring constantly. Examples include the increased use of antiplatelet agents, angiotensin-converting enzyme inhibitors, lipid-lowering therapy, internal thoracic aortic (ITA) grafts, and drug-eluting stents (DESs). These advances, along with aggressive secondary prevention after revascularization, have steadily reduced the morbidity and mortality of CAD in all patients, making differences in the hard endpoint of survival difficult but not impossible to demonstrate for any therapy.²⁷ This trend will likely increase as the beneficial effect of secondary prevention becomes more widely appreciated.

Comparative Trials of Revascularization versus Medical Therapy in Stable Angina

MEDICAL THERAPY

In the decades since CABG surgery was popularized and coronary angioplasty introduced, an enormous volume of data on the results of invasive revascularization has been collected. Remarkably, almost from the outset many of these studies have been prospectively randomized. Yet in the current era there is a dearth of data concerning pharmacologic therapies for chronic CAD despite remarkable recent drug development. For example, although nitrates are unquestionably effective in relieving symptoms, the impact of long-acting nitrates on clinical outcomes has never been rigorously tested. Furthermore, there has been only one trial of beta-blocker therapy in the treatment of angina, the Atenolol Silent Ischemia Study, which demonstrated benefit for patients with mild effort induced angina or silent ischemia.²⁸ There remain no randomized studies examining the impact of beta blockers on survival on patients with stable angina. The only evidence for improved survival with beta-blocker use comes from one recent registry study demonstrating a statistically insignificant trend toward improved survival for post-MI patients who received beta blockers.²⁹ A handful of studies of combination therapy with beta blockers and calcium channel blockers have also demonstrated antianginal benefit, but again without any evidence that there is any impact on survival.^30-32

SURGICAL VERSUS MEDICAL THERAPY

Three major randomized studies, the CASS,¹² the Veterans Administration Cooperative Study Group (VA),^13,33 and the ECSS,^34,35 as well as several other smaller randomized trials,^36-38 conducted between 1972 and 1984, provide the historical foundation for comparing the outcomes of medical and surgical therapy. Despite the limitations noted in the preceding, these studies are remarkably consistent in their major findings, and the qualitative conclusions drawn from them continue to be generalizable to current practice.

The central message from all of these studies is that the relative benefits of bypass surgery over medical therapy on survival are greatest in those patients at highest risk as defined by the severity of angina and/or ischemia, the number of diseased vessels, and the presence of left ventricular dysfunction.³⁹ For example, thus far, no study has shown survival benefit for CABG over medical therapy for patients with single-vessel disease not involving the proximal LAD. Accordingly, the current guidelines recommend that CABG be performed on a Class I recommendation for all unprotected left main disease,^39-41 three-vessel disease,^35,39,42-44 and two-vesseldisease with proximal LAD disease,^35,39,42-44 but only with a Class IIa recommendation for two-vessel disease without proximal LAD involvement if extensive ischemia exists,⁴⁵ and one-vessel proximal LAD disease.^10,46,47 It should be emphasized, however, that these trials involved primarily patients with moderate chronic stable angina. These conclusions may, therefore, not necessarily apply to patients with unstable angina or to those with more severe degrees of chronic stable angina.

A landmark meta-analysis by Yusuf et al, of the seven randomized trials cited in the preceding demonstrated a statistically enhanced survival at 5, 7, and 10 years, for surgically treated patients at highest risk (4.8% annual mortality) and moderate risk (2.5% annual mortality), but no evidence of a survival benefit for those patients at lowest risk.³⁹ The overall survival benefit at 12 years for the three large and four smaller randomized studies is shown in Fig. 20-1. Nonrandomized studies have also demonstrated a beneficial effect of surgery on survival of patients with multivessel disease and severe ischemia regardless of left ventricular function.^25-28

Following this, between 1997 and 2004, there were three randomized controlled trials of invasive revascularization by PCI or CABG versus medical therapy. Their results make an even stronger case for revascularization. In the Asymptomatic Cardiac Ischemia Pilot (ACIP) trial, patients with anatomy amenable to CABG were randomized to angina-directed antiischemic therapy, drug therapy guided by noninvasive measures of ischemia, or revascularization by CABG or PCI.⁴⁵ At 2 years, there was a statistically significant difference in mortality of 6.6% in the angina-guided group, 4.4% in the ischemia-guided group, and 1.1% in the revascularization group. The rates of death or MI were also statistically different at 12.1%, 8.8%, and 4.7%, respectively, that is, both symptoms and survival were improved with revascularization. The Medicine, Angioplasty, or Surgery Study (MASS-II) trial randomized patients with multivessel disease among medical therapy, PCI, and CABG.^48,49 Although survival at 1 year was equivalent, freedom from additional intervention was 99.5% for surgical patients and 93.7% for medically treated patients. Reintervention was, incidentally, even higher in the PCI group than the medical group, with 86.7% free of additional intervention. Angina was superior in the CABG group (88%) than in the PCI group (79%) or medical therapy group (46%). Meanwhile the 10-year survival rates were 74.9% with CABG, 75.1% with PCI, and 69% with MT.⁴⁹ In the Trial of Invasive versus Medical Therapy in Elderly Patients with Chronic Symptomatic Coronary-Artery Disease (TIME) study, elderly patients were studied with chronic angina. This failed to demonstrate a difference between optimized medical therapy and an invasive revascularization strategy (PCI or CABG) in terms of symptoms, quality of life, and death or nonfatal MI (20% vs 17%, p = .71). However, medically treated patients were at higher risk because of major clinical events (64% vs 26% for invasive, p < .001), which were mainly attributable to rehospitalization and revascularization.⁵⁰ In this trial of severely symptomatic elderly patients, it was encouraging that the price of an initially conservative strategy, followed by crossover to revascularization in approximately 50% of patients, was not paid for in terms of death or MI.⁵⁰

Early historical concern over a prohibitive operative mortality among patients with impaired ventricular function has been superseded by the recognition that the survival of these patients on medical therapy was much worse than their survival with revascularization. This, coupled with ever-improving surgical techniques, such as advances in myocardial preservation and perioperative support, has made this specific subgroup the one in which the relative survival benefit of surgical therapy is the greatest. Accordingly, left ventricular dysfunction in patients with documented ischemia is now considered an important indication—rather than contraindication—for surgical revascularization.^{10,12,39,51-53} More recently, the Surgical Treatment for Ischemic Heart failure (STICH) trial was conducted to better evaluate the effect of CABG plus optimal medical therapy to optimal medical therapy alone, an effect that the trial investigators perceived to not have been well-established by the studies referenced above considering (1) the general exclusion of patients with severe left ventricular dysfunction in the three landmark trials from the 1970s and (2) the improvements in both medical and surgical therapy that had developed since then.⁵³ The study of 1212 patients with an ejection fraction (EF) less than 35% and CAD amenable to CABG demonstrated a nonsignificant trend toward decreased mortality from any cause in the CABG group (36%) compared to the medical therapy group (41%). The trial has been criticized for being underpowered, and with just 56 months of mean follow-up and a large percentage of patients in the medical therapy group crossing over into the CABG group (17%) during the study period thereby potentially diminishing the statistical impact of CABG, many have interpreted this to represent evidence that the presence of left ventricular dysfunction remains an important indication to perform CABG for survival benefit. Indeed, the ACCF/AHA guidelines consider left ventricular (LV) dysfunction to represent a Class IIa recommendation for performing CABG if the EF is 35 to 50%, but only Class IIb recommendation if the EF is <35%, in part based on the results of the STICH trial.¹⁰ In addition, recent evidence that ischemic, viable, hypokinetic myocardium (hibernating or stunned) regains stronger contractile function after effective revascularization, has prompted expansion of the indications for surgical revascularization among patients with severe left ventricular dysfunction to include patients who would otherwise be considered candidates for cardiac transplantation.⁵⁴ This subject is discussed in more detail in the following.

In summary, the current guidelines recommend on the basis of a survival advantage that CABG be performed instead of medical therapy alone on a Class I recommendation for all unprotected left main disease,^39-41 three-vessel disease with or without proximal LAD involvement,^35,39,42-44 and two vessel-disease with proximal LAD disease^35,39,42-44; with a Class IIa recommendation for two-vessel disease without proximal LAD involvement but with the presence of extensive ischemia,⁴⁵ one-vessel proximal LAD disease,^10,46,47 and moderate LV dysfunction (EF 35-50%)^39,54-57; and with a Class IIb recommendation for two-vessel disease without proximal LAD disease or extensive ischemia,⁴⁴ and in patients with severe LV dysfunction (EF < 35%)^39,53-57 (Fig. 20-2).

Apart from affording a survival benefit, CABG is also indicated for the relief of angina pectoris and improvement in the quality of life. Between 80 and 90% of patients who are symptomatic on medical therapy become symptom-free after CABG in the early period while approximately 60% remain symptom-free at up to 10 years.⁴⁹ This benefit extends to low-risk patients for whom survival benefit from surgery is not likely.³⁹ Thus, for example, whereas on the basis of a survival advantage alone CABG is only indicated for one-vessel disease if the vessel involved is the left main (Class I) or the proximal LAD (Class IIa), a symptomatic patient with single-vessel disease in any anatomic position represents a Class I, Level A indication to perform either CABG or PCI on the basis of symptom improvement (Fig. 20-2).^{10,49,50,58-60} Consensus agreement by the ACC/AHA recommends on a Class IIa basis CABG is recommended for patients with at least one significant coronary lesions in any position and unacceptable angina (1) for whom guideline-directed medical therapy (GDMT) cannot be implemented because of medication contraindications, adverse effects, or patient preferences; and (2) complex three-vessel CAD (eg, SYNTAX score > 22) with or without involvement of the proximal LAD and a good candidate for CABG (CABG preferred over PCI in this situation).^10,44,61-63 Relief of symptoms appears to relate to both the completeness of revascularization and maintenance of graft patency, with the benefit of CABG diminishing with time. Recurrence of angina following CABG surgery occurs at rates of 3 to 20% per year. Although enhanced survival is reported when an ITA graft is used to the LAD, there is no significant difference in postoperative freedom from angina.^46,47 This may be because of vein graft occlusion or progression of native disease in grafted or ungrafted vessels.¹³

Unfortunately, few patients experience an advantage in work rehabilitation with surgery as compared with medical management. Generally, employment declines in both groups and is determined nearly as much by socioeconomic factors as age, preoperative unemployment, and type of job as by type of therapy or clinical factors such as postoperative angina. Notably, surgical revascularization has not been shown to reduce the incidence of nonfatal events such as MI, although this may be because of perioperative infarctions that offset the lower incidence of infarction in each study follow-up.^64,65

PCI VERSUS MEDICAL THERAPY

PCI continues to evolve as a therapy from its historical roots as a procedure entailing balloon angioplasty alone, followed by the introduction of bare metal stents (BMSs), and subsequently DESs. Despite this evolution and the attendant reduction in overall in-stent restenosis rates in the DES era, and despite a multitude of evidence in both early and more recent trials demonstrating the ability of PCI to effectively reduce the incidence of angina compared to medical therapy in patients with stable ischemic heart disease, there remains no definitive evidence in any study to date that PCI improves survival rates or rates of subsequent MI in this patient population.^43,60,66-71

It is worth describing the specifics of several representative studies on patients with multivessel disease. In the Randomized Intervention Treatment of Angina (RITA)–2, of 1018 patients with stable angina randomized to medicine or PCI, one-third had two-vessel disease and 7% had three-vessel disease.⁵⁹ Perhaps surprisingly, at a median follow-up of 2.7 years, the primary endpoints of death or MI had occurred twice as often in the PCI group (6.3 vs 3.3%, p < .02). Surgical revascularization was required during the follow-up interval in 7.9% of the PCI group and repeat angioplasty was required in 11%. In the medical group, 23% of patients required revascularization. Angina relief and exercise tolerance were improved to a greater degree in the angioplasty group early, but this difference disappeared by 3 years. These results are echoed in the MASS-II trial, in which angina relief was superior with PCI, but rates of intervention/reintervention were actually higher in the PCI group.^48,49 This supports an initial strategy of medical therapy in patients with stable ischemic heart disease.

More recently, the COURAGE trial randomized 2287 patients with stable angina and objective evidence of ischemia to contemporary medical therapy or PCI and medical management (3% of the stents used were DESs). All patients received long-acting metoprolol, amlodipine, and/or isosorbidemononitrate, as well as lisinopril or losartan. Patients had aggressive antilipid therapy and appropriate antiplatelet therapy. The primary finding of this trial was a lack of benefit of PCI over best medical therapy for death, MI, or other major cardiovascular events.⁶⁰ Although this result has caused much controversy in the cardiology community, it is consistent with older studies.

A meta-analysis of percutaneous interventions versus medical management was published in 2005.⁶⁸ In patients with stable CAD, no benefit was found for invasive therapy in terms of death, MI, or need for subsequent revascularization. In 2014, Windecker et al performed a large scale meta-analysis of 100 trials comparing both CABG and PCI to medical therapy found that patients who received new generation everolimus- and zotarolimus-based (Resolute) DESs but not older generation DESs (sirolimus, paclitaxel, and Endeavor zotarolimus), BMSs, or balloon angioplasty alone, had reduced mortality compared to medical therapy.⁷¹ However, given the limitations of this study, including the inclusion of only one trial comparing newer generation DESs to medical therapy and absence of individual patient data, the general consensus remains that unless and until further evidence is collected to the contrary, all patients with stable coronary disease should have a trial of optimized medical therapy before invasive intervention in the form of PCI. Accordingly there are no Class I recommendations for PCI in the setting of stable ischemic heart disease in the 2011 ACCF/AHA guildelines.¹⁰ Based on subgroup analysis from the SYNTAX trial (discussed further in the next section), however, it is recommended on a Class IIa basis for stable ischemic heart disease patients with unprotected left main disease for benefit if they meet the following criteria: (1) anatomic conditions associated with a low risk of PCI complications and have a high likelihood of good long-term outcome; for example, low SYNTAX score ≤22, ostial or trunk left main CAD; and (2) clinical characteristics that impart a high risk of morbidity and mortality with CABG (eg, Society of Thoracic Surgeons (STS)-predicted mortality > 5%).⁷²

PCI VERSUS CABG

Randomized Studies

A number of studies comparing an initial strategy of angioplasty versus early surgery have been carried out, all with similar results. It is important to recognize that, as a rule, these studies are comparisons of treatment strategies and not head-to-head comparisons of revascularization techniques. Accordingly, crossover is permitted and endpoints are selected to determine adverse consequences of the algorithm on an “intention to treat” basis.

Most early trials of CABG versus PCI have in common that PCI was performed first in the form of balloon angioplasty alone, followed by those including BMS deployment with angioplasty. In all, over 20 randomized controlled trials comparing CABG with PCI in the form of angioplasty or BMS implantation have been conducted, and on this basis the following conclusions were made by Bravata et al⁷³ in a recent systematic review and cited in the establishment of the ACC/AHA 2011 guidelines:¹⁰ (1) Survival is similar after CABG and PCI at both 1 year and 5 years, (2) incidence of MI was similar at 5 years after randomization, (3) CABG produced more effective relief from angina than with PCI at 1 and 5 years, and (4) repeat coronary revascularization was less frequent after CABG than after PCI. In other words, at least in the early era of PCI, CABG appears to have a benefit over PCI only insofar as it improves symptoms and reduces the need for repeat coronary revascularization, but without any improvement in overall survival or incidence, which reflects the results of even the earliest studies ever conducted on this topic, beginning with the 1994 Swiss trial of just 134 patients⁷⁴ and both the MASS^75,76 and MASS-II^48,49 trials. The oft-quoted BARI, Emory Angioplasty versus Surgery Trial (EAST), RITA, CABRI, and GABI multivessel PCI versus CABG surgery trials,^77-81 with only a few nonreproducible exceptions, generally support the same conclusions above.

Most of the early studies quoted above share the limitation that, in general, only a very small minority of patients undergoing revascularization at any center were entered into these trials.^82,83Accordingly, the populations included in the trials may not be generally reflective of clinical practice. For instance, few patients in these studies had significant LV dysfunction and most randomized patients had only one- or two-vessel disease. In the RITA trial, approximately one-third of patients had single-vessel disease.⁷⁷ Among clinically eligible patients in the BARI^80,84 and EAST⁸¹ trials, approximately two-thirds of patients were excluded on angiographic grounds that included chronic total occlusion, LMCA stenosis, diffuse disease, or other anatomical factors making PCI potentially dangerous. Consequently, these randomized trials contain only a portion of the spectrum of patients with CAD encountered clinically. Entry bias has a significant impact on the likelihood of observing an outcome difference among therapies. Because a high proportion of the randomized patients are in the low-risk group, it is possible that any potential survival benefit of CABG surgery over PCI in high- and moderate-risk groups may be masked.⁸³

A second consideration in evaluating these studies is that the success of revascularization procedures depends not only on the criteria employed to define success, but also on the interpretation of those criteria by both patient and physician. In the 1985 to 1986, National Heart, Lung, and Blood Institute PCI Registry, 99% of patients were discharged alive from hospital, and 92% did not sustain a MI or require CABG surgery.⁸⁵ In the BARI trial, 99% of patients survived hospitalization and 88.6% of PCI-treated patients did not have MI or require repeat revascularization by angioplasty or surgery during the initial hospitalization.⁸⁰ Employing event-free criteria (death, MI, CABG) for the initial hospitalization, PCI can be judged successful. However, if a repeat revascularization procedure within 5 years is regarded as a negative outcome, then far fewer patients are treated successfully. Regardless, the lack of differences in mortality or MI rates permits individuals to select one or the other procedure as initial therapy without the likelihood that they will pay a price with their health.

Considering event-free survival as a more meaningful endpoint than overall survival, several more recent studies of CABG versus early era PCI (balloon angioplasty only) have demonstrated an advantage with surgery. The Argentine Randomized Trial of Coronary Angioplasty versus Bypass Surgery in Multi-vessel Disease (ERACI) trial conducted between 1988 and 1990 demonstrated no difference in death or MI, but superior event-free survival in the CABG group at 1 and 3 years.^86,87 In the French Monocentric Study, 152 patients with multivessel disease underwent PCI or CABG.⁸⁸ Again, superior event-free survival was seen in the surgical group, driven predominantly by a lesser need for subsequent revascularization. Comparing CABG to PCI with BMS implantation, Mercado et al performed a meta-analysis of the ARTS, ERATSII, MASS-II, and SOS trials that demonstrated similar rates of death, MI, or stroke at 1 year and higher repeat revascularization rates with PCI.⁸⁹ A meta-analysis of 5-year data was also recently published, confirming the 1-year results with repeat revascularization significantly more frequent after PCI than CABG (29% vs 7.9%).⁹⁰

In large part fueled by recognition of the relatively high rates of in-stent restenosis seen with bare metal stents (22- 32% at 6 months),^91,92 many investigators had hoped that by lowering the rates of early restenosis,⁶⁷ the introduction of DESs to the PCI armamentarium might improve overall PCI outcomes and expand the indications for PCI over CABG in a wide variety of clinical scenarios. The Synergy between PCI with Taxus and Cardiac Surgery (SYNTAX) trial, assessed the optimal revascularization strategy for patients with three-vessel or left main CAD by randomizing 1800 patients to either CABG or paclitaxel-based DESs.⁴ At 12 months, the rates of MI and death were similar between groups, but stroke was significantly more common in the CABG patients (2.2% vs 0.6% at 1 year). It should be noted that the CABG group had much less aggressive medical management postoperatively, including fewer patients on antiplatelet medications, which may account for some of the increase in stroke risk. Despite the use of DESs, the rates of reintervention were still significantly higher (13.5% vs 5.9%) in the PCI group than the CABG group. Because of the lack of difference in early mortality and MI, some cardiologists have begun to suggest that left main should no longer be an indication for CABG.⁹³ However, others maintain that the evidence for PCI is still inadequate and that the greater freedom from reintervention with CABG suggests that CABG remains the treatment of choice for patients with left main disease.⁹⁴ Meanwhile, the SYNTAX scores themselves, which grade the severity and complexity of coronary disease according to a variety of anatomic features (extent, location, severity), were defined in post hoc analyses as low for scores ≤22, intermediate for scores 23 to 32, and high if ≥33. While the incidence of major adverse cardiac events (MACE) for patients undergoing CABG versus PCI were similar in subgroup analysis of low SYNTAX score patients, those with an intermediate or high SYNTAX score had lower incidence of MACE with CABG.⁴ Furthermore, at 3-year follow-up, those with three-vessel disease had significantly lower mortality in the CABG group compared to the DES group (6.2% vs 2.9%).⁶³ CABG appears to be preferable to PCI for those with diffuse and complex CAD. It is on this basis that CABG is recommended over PCI on a Class IIa basis for patients those with three-vessel disease (regardless of proximal LAD involvement) in patients with SYNTAX scores > 22 who are good candidates for CABG.¹⁰

Notably, the SYNTAX trial also brought to light the concept of the Heart Team approach, described in the SYNTAX trial as a system in which a team consisting of a heart surgeon, an interventional cardiologist, and often the general cardiologist (if available) discuss each case and decide on the basis of mutual agreement the optimal method of revascularization in multidisciplinary fashion. The Heart Team approach, while not validated by any follow-up randomized controlled trial is recommended on a Class I basis for all patients presenting with unprotected left main or complex CAD.¹⁰

Since the SYNTAX trial, there have been five additional randomized controlled trials comparing CABG to PCI. Two of these trials discuss PCI with DESs in comparison to minimally invasive direct coronary artery bypass and therefore are less germane to the present discussion regarding on-pump CABG. The PRECOMBAT trial showed that PCI using DES was noninferior to CABG in UPLM at 2 years for composite endpoint and will be discussed further in the forthcoming section on left main disease.⁹⁵ Meanwhile, the CARDia trial achieved the modest goal of demonstrating that in patients with multivessel CAD and DM, PCI with the use of DES is noninferior to CABG at 1 year with respect to a composite endpoint of all-cause mortality, MI, and stroke.⁹⁶ The FREEDOM trial, which was designed similarly for a similar population but had nearly four times the study population size, demonstrated that for patients with DM and advanced CAD, CABG resulted in lower rates of the same composite endpoint compared to PCI with DES (18.7% vs 26.6%),²⁶ which is the first time a randomized trial of CABG versus PCI with DES has demonstrated this effect. A recent 2014 meta-analysis of six randomized trials of multivessel CAD whose goal was to accurately reflect the more modern era in which DES and arterial grafts are used more frequently in PCI and CABG, respectively, demonstrated that with a weighted average follow-up of 4.1 years, CABG was associated with significantly lower mortality, lower incidence of MI, and repeat revascularization.⁹⁷ These new findings, which now demonstrate a survival advantage and not just freedom from repeat revascularization and symptom improvement with CABG over PCI, were not available at the time of the writing of the 2011 ACCF/AHA guidelines, but have implications for the management of patients in the future if further evidence supporting the same conclusions is gathered.

Nonrandomized Database Comparisons

The information provided by randomized studies is complemented by information gleaned from large, prospectively managed, nonrandomized database studies. Such registries provide insight into the management of the sizeable population of patients who would not have been eligible for randomization. The 1994 Duke Cardiovascular Disease Databank study was one of the first large registry studies that established much of what we know today about the benefit of CABG compared to PCI insofar as it is dependent on the severity of coronary disease.²² From a practical standpoint, in this database study and in the randomized trials, the effect of revascularization on survival depended largely on the extent of the CAD and is an example of the concept of benefit in relationship to a “gradient of risk.” For the least severe (one-vessel) disease, there were no survival advantages of revascularization over medical therapy in up to 5 years of follow-up.²² For intermediate levels of CAD severity (ie, two-vessel disease), there was a higher 5-year survival rate for patients undergoing revascularization than for those treated medically. For patients with the most-severe CAD (ie, three-vessel disease), CABG surgery provided a significant and consistent survival advantage over medical therapy. PCI appeared prognostically equivalent to medical therapy in these patients, but only a small number of patients in this subgroup underwent angioplasty. In comparing PCI with CABG surgery, PCI demonstrated a small survival advantage over CABG surgery for patients with less-severe two-vessel disease, whereas CABG surgery was superior for more severe two-vessel disease (ie, proximal LAD involvement).²²

Several additional studies in the subsequent PCI with BMS era confirmed these findings. A study of 1999 cases from the New York State Database between 1993 and 1998 showed that a survival benefit was observed with angioplasty at 3 years for those patients with single-vessel disease not involving the LAD, whereas those with LAD or three-vessel disease had superior outcomes with surgery.⁹⁸ In a much larger CABG versus PCI registry study from the BMS era, survival was again higher among the 37,212 patients who underwent CABG than among the 22,102 patients who underwent stent placement after adjustment for known risk factors⁶¹ (Fig. 20-3). This study has limitations of being a nonrandomized study and subject to bias; however, the surprising finding of a survival advantage apparent as early as 3 years postprocedure suggests that improvements in cardiac surgical anesthetic care, myocardial protection, and intensive care management have at least matched if not surpassed advances in percutaneous technology. It is important to note that while PCI targets the culprit lesion, CABG surgery targets both the culprit lesion and potential future culprit lesions by bypassing the diseased vessel. Considering the finding that medium- and long-term clinical outcome after PCI (ie, beyond one year postprocedure) is more dependent on the progression of coronary disease in other culprit lesions than restenosis of the stented lesion,⁹⁹ this in part may explain the apparent mortality benefit derived with CABG that has persisted despite advances in stent technology that have resulted in lower restenosis rates (Fig. 20-3).¹⁰⁰

In the DES era, five of six large registry studies concluded that patients undergoing CABG have lower mortality than those receiving PCI with DES.^62,101-105 In particular, the ASCERT study, upon which the Society of Thoracic Surgeons ASCERT Long-Term Survival Probability Calculator (ascertcalc.sts.org) is based, evaluated outcomes of almost 200,000 patients over the age of 65 years with multivessel disease undergoing nonemergent revascularization, and showed that although there was no difference in adjusted mortality in the CABG and PCI groups at 1 year, CABG patients had significantly lower mortality at 4 years (16.4% vs 20.8%).¹⁰⁴ The long-term benefit of CABG compared to PCI was independent of age, sex, diabetes, renal function, and lung disease and was evident even for patients with propensity scores most consistent with selection for PCI. Another registry study of over 105,000 propensity-matched United States Medicare beneficiaries undergoing either CABG or PCI between 1992 and 2008 showed that at 5 years, patients undergoing CABG had a higher survival rate (74.1% vs 71.9%), an effect which was more pronounced in those with diabetes, tobacco use, heart failure, and peripheral artery disease.¹⁰⁵

In summary, although not void of limitations that randomized controlled trials and associated meta-analyses overcome, these large registry studies provide a more real-world picture of revascularization in actual clinical practice, and support the findings of newer randomized controlled trials that CABG confers survival benefit over PCI, especially in patients with multivessel and complex CAD, even in the era of DES use and the attendant lower rates of restenosis.

Special Circumstances

ACUTE CORONARY SYNDROMES

The acute coronary syndromes (ACS) cover a wide spectrum from ST-segment elevation MI (STEMI) with underlying coronary obstruction to Prinzmetal’s or variant angina in patients with coronary vasospasm in the absence of significant underlying obstruction. The term non-ST-segment elevation (NSTEMI) ACS encompasses the entities of unstable angina, non-Q-wave MI, and postinfarction angina. They denote acute, symptomatic imbalances of the myocardial oxygen supply-demand ratio over a short time span. Prinzmetal angina, or coronary vasospasm, is diagnosed definitively by electrocardiograms obtained during the episode of pain and is treated medically. Unstable angina is not a uniform clinical entity, but comprises the spectrum of myocardial ischemia between chronic stable angina and MI, and is defined as a recent change in the severity, character, or trigger threshold of chronic stable angina or new-onset angina. Approximately 5.6 million Americans have chronic angina, and about 350,000 develop new-onset angina each year. Unstable angina develops in approximately 750,000 Americans each year and is associated with subsequent MI in approximately 10%. Postinfarction angina is defined as the presence of angina or other evidence of myocardial ischemia in a patient with a recent (1- to 2-week) Q-wave or non-Q-wave MI.

CABG in ST Elevation Myocardial Infarction

Prompt myocardial revascularization has been shown to improve outcomes compared to no reperfusion in almost all groups of patients who present after an acute STEMI. Because the degree of improvement in long-term outcomes is dependent on the time from symptom onset to revascularization,^106,107 PCI, which in general can be initiated more quickly that can CABG, is performed as the primary therapy in most cases, relegating CABG largely to a secondary position in this setting (performed in less than 5% of STEMI cases¹⁰⁸). In the majority of PCI cases, the culprit lesion can be determined and successfully stented open, with subsequent improvement in myocardial ischemia. However, in certain cases where coronary disease is diffuse and not amenable to stent placement, or in the setting of complications from PCI such as coronary dissection, perforation with tamponade, stent deployment failure, or hyperacute stent closure, revascularization is either not technically possible or remains incomplete with ongoing ischemia. Emergency CABG in current practice is reserved and, in fact, recommended on a Class I basis primarily for survival benefit in the following circumstances: (1) after unsuccessful or complicated PCI with ongoing ischemia;¹⁰⁹ (2) mechanical complication of acute MI requiring surgery including ventricular septal defect, free wall rupture, or papillary muscle with acute mitral regurgitation; (3) cardiogenic shock, defined as systolic blood pressure < 90 mm Hg for over 30 minutes requiring supportive measures, cardiac index ≤ 2.2 L/min/m², pulmonary capillary wedge pressure ≥ 15 mm Hg, and/or evidence of end-organ malperfusion;^106,107,110 and (4) patients with life-threatening arrhythmias in the presence of significant left main or three-vessel CAD.^10,111 The SHOCK study randomized patients to emergency revascularization versus initial medical stabilization in patients presenting with STEMI and cardiogenic shock and found that among those who were revascularized early, CABG group (36%) of patients had similar survival to those undergoing PCI despite higher rates of diabetes and complex coronary anatomy in the CABG subset.¹¹⁰

The timing of CABG after STEMI remains controversial, but overall the evidence suggests that mortality appears to peak in the 7-to-24-hour period after symptom onset. Several studies have shown that CABG with in-hospital mortality rates of 10.8% if performed within 6 hours of onset, 23.8% at 7 to 24 hours after onset, 6.7% at 1 to 3 days, 4.2% at 4 to 7 days, and 2.4% after 8 days. Another study showed that mortality was 6.1% for CABG performed within 6 hours, 50% at 7 to 23 hours, and 7.1% at 15 days or greater.¹⁰⁹ A reasonable approach appears to be emergency CABG within the 6-hour timeframe if possible to maximize myocardial salvage. Those who present after the 6-hour window would likely benefit from a delay of three to four days to improve survival and decrease bleeding complications.^108,109,112

CABG in Non-ST Elevation Myocardial Infarction and Unstable Angina

The initial approach to the patient with NSTEMI ACS is pharmacologic stabilization followed by risk stratification. The latter is based on multiple clinical, demographic, and ECG variables in addition to the use of serum biomarkers. The TIMI (Thrombosis in Myocardial Infarction) risk score, which takes into account a variety of the above clinical factors, is widely used as a prognostic tool to determine in patients with unstable angina or NSTEMI the 14-day risk of all-cause mortality, new or recurrent MI, or severe recurrent ischemia requiring urgent revascularization,¹¹³ and is often used to risk-stratify patients to a strategy of either immediate angiography, an invasive approach (angiography followed by revascularization by either PCI or CABG), or a conservative approach (initial period of medical therapy intensification). Those with hemodynamic instability, cardiogenic shock, severe LV dysfunction, persistent or recurrent angina at rest despite intensive medical therapy, mechanical complications, sustained ventricular tachycardia, and dynamic ST-T changes, are usually considered high risk irrespective of the TIMI score which does not account for these factors. Patients at intermediate or high risk (the majority) undergo early angiography with a view to revascularization. In many parts of the world, however, angiographic facilities are limited and an alternative approach based upon pharmacologic stabilization followed by mobilization and risk stratification using stress testing is employed.

In general for those undergoing angiography, the choice of PCI versus CABG is largely determined by the anatomy and specific clinical features as in the general population presenting with stable ischemic heart disease, as per the ACCF/AHA guidelines.¹⁰

ASYMPTOMATIC CORONARY ARTERY DISEASE

The role of revascularization either by PCI or CABG versus medical therapy in the setting of asymptomatic CAD has been studied in the ACIP trial.¹¹⁴ Of 558 patients with CAD and medically controlled angina, treatment was randomized to either revascularization or medical therapy directed toward eliminating angina or eliminating ischemia during ambulatory ECG. Revascularization was more effective than medical therapy in relieving ischemia, and CABG was superior to PCI (70% freedom from ischemia vs 46%, p < .002). Mortality at 1 and 2 years was superior for revascularization as compared with angina-directed medical therapy, but not superior to ischemia-directed medical therapy.^45,114 The greatest benefit was among those patients with the most severe disease. Importantly, many of these patients with “silent ischemia” on ambulatory ECG monitoring did have symptomatic angina at other times and thus were not truly asymptomatic.

The documentation of ischemia, however, is critical. Several studies emphasize the flaws in the assumption that one can identify future culprit lesions in the absence of symptoms or documentation of ischemia. Among patients undergoing serial coronary arteriography who subsequently developed acute MI or unstable angina, the severity of stenosis at the time of initial angiogram is poorly predictive of the culprit lesion causing the acute ischemic syndrome.^115,116 In most cases the severity of the lesion responsible for subsequent ischemia was less than 50%, and in many patients it was not present at all on the initial angiogram, raising concern regarding the number of asymptomatic or minimally symptomatic patients undergoing angioplasty without stress testing.¹¹⁷ Bech and associates have demonstrated that fractional flow reserve exceeded 0.75 in 91 of 325 patients planned for PCI without noninvasive evaluation of ischemia, and that among those patients, angioplasty had no impact on event-free survival or angina.¹¹⁸

DRUG-ELUTING STENTS

Recent progress in stent technology, particularly the introduction of DESs, has further reduced the incidence of restenosis. The use of stents has reduced adverse remodeling, and stents-eluting medications such as sirolimus, paclitaxel, or everolimus, are having a profound effect on the patterns of interventions for CAD. When compared with bare-metal stents, DES have not been shown to convey any advantage in terms of MI or mortality, but do demonstrate decreased rates of angiographic restenosis in a meta-analysis of 11 randomized trials.⁶⁷ The results of both randomized controlled trials and observational studies comparing CABG to PCI using DES are described in the preceding section, and suggest that the addition of DES, while successful in decreasing the restenosis rate inherent in BMS, the indications for PCI and CABG have not changed.¹⁰¹ Some of the most recent studies comparing these two strategies (FREEDOM trial, a meta-analysis of six randomized trials by Sipahi et al, and the ASCERT study)^26,97,104 have actually demonstrated a long-term survival benefit of CABG over PCI despite the use of DES in the PCI groups. It is important to note that because DES is associated with higher rates of in-stent thrombosis (as distinct from restenosis), patients receiving DES in general require 6 to 12 months of dual antiplatelet therapy (usually clopidogrel and aspirin) to minimize this risk.¹¹⁹ Consideration for CABG or BMS should be given over DES on an individualized basis for patients in whom comorbid conditions will require withholding clopidogrel in order to perform subsequent surgery.

LEFT MAIN DISEASE

Left main CAD is present in about 5% of patients undergoing coronary angiography,¹²⁰ and when present is associated with multivessel CAD around in 70% of patients.^4,94 Significant unprotected left main disease (ie, left main disease without any preexisting distal CABGs) has been considered an indication for CABG rather than PCI for many years, but recently studies have begun to question this. A handful of registry studies²⁵ as well as subgroup analysis of the SYNTAX trial,⁷² and finally several randomized trials of CABG versus PCI with DES^95,121 have suggested that PCI may be an acceptable method of revascularization in patients with left main disease. In a matched cohort of more than 1100 patients with unprotected left main disease, there was no difference in the rates of death or the composite outcome of death, Q-wave MI, or stroke between patients undergoing CABG or PCI despite a greater rate of target-vessel revascularization in the PCI group, including the patients who received DES.²⁵ The PRECOMBAT trial compared PCI using sirolimus-eluting stents to CABG in patients with left main disease and found that the risk of major adverse cardiac or cerebrovascular events (MACCE) was not significantly higher after PCI compared to CABG at two years, but did have about double the incidence of repeat revascularizations.⁹⁵ A subgroup analysis of patients in the SYNTAX trial presenting with left main disease supported these findings by showing that PCI with a paclitaxel-eluting DES did not increase the 12-month rate of MACCE in comparison to CABG, but did have a higher rate of repeat revascularization.⁷² Five-year outcomes also demonstrate the same findings qualitatively.¹²² However, when grouped according to SYNTAX score, while those with low or intermediate scores had similar outcomes regardless of the method of revascularization, those with high-risk scores (>32) undergoing PCI had higher incidence of MACCE than CABG patients.^72,122,123 Based primarily on this evidence, the ACCF/AHA guidelines regard PCI as a Class IIa or IIb recommendation in patients with unprotected left main disease and stable ischemic heart disease depending on how favorable the anatomy and how high the overall surgical risk (Class IIa if SYNTAX score is ≤22 and STS-predicted mortality is ≥5%, Class IIb if SYNTAX score < 33 and STS-predicted mortality > 2%).¹⁰ PCI may be performed on a Class IIa recommendation if the patient is having unstable angina/NSTEMI and is not a CABG candidate due to high risk of operative morbidity and mortality, or in the setting of a STEMI when distal coronary flow is grade 3 and PCI can be performed more rapidly than CABG.¹⁰ For all other instances (ie, the majority of patients), until additional data are gathered to the contrary, CABG remains the definitive Class I-recommended treatment for most patients with unprotected left main disease.¹⁰ The EXCEL trial which is currently ongoing seeks to compare CABG to PCI using newer generation everolimus-eluting stents in patients with unprotected left main disease.

SEVERE LEFT VENTRICULAR DYSFUNCTION

Left ventricular dysfunction is a predictor of increased operative risk for most cardiac surgical procedures, and in the early days of coronary revascularization these patients were not offered CABG. Like age, however, it is also a strong predictor of poor outcome with medical therapy. Accordingly, more recently significant LV dysfunction has been considered an indication rather than contraindication to surgical revascularization. In some patients, LV function has been shown to improve—sometimes dramatically so—after revascularization, leading to the concept of “hibernating” or “stunned” myocardium.^124,125 The identification of viable myocardium, which is potentially recoverable, depends on the identification of preserved metabolic activity by positron emission tomography, cell membrane integrity by thallium-201 or technetium-99m single-photon emission computed tomography (SPECT), or dobutamine stress echocardiography.¹²⁶

Five studies from the early era of CABG as well as several meta-analyses of these studies and others that followed showed that patients with LV systolic dysfunction (mainly mild to moderate) showed that CABG conferred improved survival in comparison to medical therapy.^39,54-57,127 In regards to severe left ventricular dysfunction, however, as mentioned above the STICH trial, showed no reduction in all-cause mortality for CABG in comparison to GDMT in patients with CAD amenable to CABG with an EF of <35%.⁵³ CABG did, however, demonstrate a benefit compared to medical therapy with respect to the composite endpoint of death from any cause or (1) hospitalization for heart failure, (2) hospitalization for cardiovascular access, (3) hospitalization for any cause, and (4) revascularization with PCI or CABG. In other words, event-free survival was better with CABG in this population, suggesting that with longer follow-up, an overall survival benefit might resolve. The study is being continued out to 10 years of follow-up.

The relative roles of CABG and PCI in this population are not clearly defined despite the number of randomized trials of PCI versus surgery, largely because these patients were excluded from trial entry. In an early multicenter study of patients with left ventricular dysfunction (EF < 40%), slightly more than one-fourth of the patients died in the 2 years following multivessel PCI.¹²⁸ The majority of studies comparing CABG to PCI in this population have demonstrated similar survival,^129-133 while some showed improved outcomes with CABG.⁶² Overall, aside from the STICH trial, there remains a paucity of meaningful randomized trial data informing the choice of CABG versus PCI in the setting of left ventricular dysfunction. For patients with LV dysfunction and CAD amenable to revascularization, the recommendation is to perform CABG on a Class IIa basis if the EF is slightly depressed (35-50%), and on a Class IIb basis if the EF is very low (<35%) as long as the LMCA is not involved, in which case CABG is more clearly indicated.¹⁰ The ACCF/AHA guidelines recommend that patients in this category should have the method of revascularization tailored to clinical variables such as the presence of renal disease, diabetes, and the complexity of the coronary anatomy.¹⁰

CHRONIC TOTAL OCCLUSION

Chronic, totally occlusion of a coronary artery is found on up to 20% of all coronary angiograms, and of these, three-fourths have multivessel disease.^134,135 Revascularization is not possible nor attempted in 65% of patients with chronic total occlusion.¹³⁴ This population remains poorly studied in the randomized trials of multivessel PCI versus CABG as 35 to 37% of the patients excluded from the overall study were disqualified because of the presence of a chronic total occlusion of a coronary artery serving viable myocardium. Notably, the SYNTAX trial did not exclude patients with chronic total occlusion (a condition which alone contributes 10 to 15 points to the SYNTAX score), and found that complete revascularization, which has a known association with reduction in long-term mortality, MI, and repeat coronary interventions,¹³⁶ was significantly higher in patients treated with CABG compared to PCI (69% vs 49%).⁴ Considering that the presence of chronic total occlusion was the most common reason for the inability to achieve complete revascularization in the PCI group, this may account for the lower rates of major adverse cardiovascular events including death in the CABG group compared to PCI.¹³⁵ However, success rates for achieving complete revascularization by PCI continue to improve as the interventional cardiology community increases its aggressiveness in using PCI to treat these lesions, as evidenced by the recent launching of at least three trials (EXPLORE, DECISION-CTO, and EURO-CTO), all designed to determine the benefit of PCI in this setting compared to medical therapy.¹³⁴ In summary, CABG surgery improves survival in patients with multivessel and complex CAD compared with PCI, in large part due to the ability of CABG to provide higher rates of complete revascularization.

THE ELDERLY

Age is a predictor of operative risk in most models, but is also a predictor of poor outcome with medical therapy in the presence of CAD. Mortality is estimated to be approximately 8 to 11% in octogenarians after CABG.¹³⁷ The Swiss Multicenter Trial of invasive versus medical therapy in the elderly (TIME) trial examined 301 patients over the age of 75 years with chronic angina and randomized them to medical therapy with or without invasive evaluation.⁵⁰ Of those undergoing angiography, two-thirds had revascularization. At the 1-year follow-up interval, there was no statistical difference in death or nonfatal MI rates. Symptoms and quality of life were also similar. However, there was a substantial increased risk of rehospitalization with revascularization in those who had medical treatment. Early interest in using off-pump CABG as a method of reducing risk in the elderly has not demonstrated any benefit in comparison to traditional on-pump CABG in two randomized controlled trials focusing specifically on patients aged 75 years and older.^138,139 A recent propensity-matched analysis of 1932 patients in this same age group showed that CABG and PCI with DES outcomes were similar at a mean follow-up interval of 1.5 years, with a higher incidence of repeat revascularization in the PCI group, similar to the conclusions drawn from studies in younger populations.¹⁴⁰ These data suggest that invasive evaluation should be offered to elderly people who are symptomatic only after adequate medical therapy. The choice of mode of revascularization will be particularly impacted in this group of patients by the presence of comorbidities that may increase the risk of surgical intervention, such as cerebrovascular disease, renal dysfunction, and pulmonary disease.

DIABETES MELLITUS

It has been recognized for many years that patients with DM are at higher risk following percutaneous¹⁴¹ or surgical¹⁴² revascularization. The BARI trial was the first trial large enough to identify significant differences in outcome between diabetic and nondiabetic patients. In this study, which included 353 diabetic patients, a survival benefit was observed among insulin-dependent patients undergoing CABG with an ITA, as compared with those undergoing PCI. The explanation for this is not entirely clear, although an intriguing observation is that while the incidence of subsequent MI is similar between groups, survival after MI is superior among those who have undergone surgical revascularization.¹⁴³ In fact, unlike nondiabetics, diabetics suffering spontaneous Q-wave MI were more than 10 times as likely to die with their infarction if they had been treated with PCI as compared with CABG. This survival difference was even more pronounced at 7 years with 76.4% of diabetics in the surgical arm alive as compared with 55.7% in the angioplasty arm.¹⁴⁴

The physiologic basis for this difference remains a matter of speculation, although the completeness of revascularization may be a factor. Because of the significant incidence of restenosis after PCI in diabetics, Van Belle and colleagues¹⁴⁵ analyzed EF at 6 months and long-term cardiac mortality and morbidity among 513 diabetic patients stratified according to the presence of occlusive restenosis (n = 94), nonocclusive restenosis (n = 257), and no restenosis (n = 162). The mortality risk rose with restenosis (24% without restenosis, 35% with nonocclusive restenosis, and 59% with occlusion), and EF fell with occlusion (decrease of 4.8 ± 12.6%).

The results of the BARI trial prompted retrospective post hoc analyses of several earlier trials. The results have been variable. There was a nonsignificant trend for better survival among diabetic patients treated surgically in the EAST trial.¹⁴⁶ A meta-analysis of pooled data pertaining to diabetic patients from CABRI, EAST, and RITA, however, found similar 5-year mortality rates following CABG or PCI.¹⁴⁷ Meanwhile, a larger meta-analysis of 10 randomized trials showed that long-term survival rates in patients with diabetes were lower in patients receiving PCI with balloon angioplasty or BMS compared to CABG.¹³³

The follow-up study to BARI, BARI-2D,¹⁴⁸ enrolled 2368 patients with DM and stable CAD and randomized them to receive either intensive medical therapy or intensive medical therapy and prompt revascularization (CABG or PCI left to clinical judgment). At 5 years there was no difference in survival or freedom from major cardiovascular events (MI or stroke) between medical therapy and revascularization groups. This trial was not designed to compare CABG and PCI in diabetes, but rather to examine a strategy of intensive medical therapy compared with revascularization. However, in the stratum of patients who received CABG, there were fewer major cardiovascular events when compared with medical therapy (22.4 vs 30.5%, p = .002). In the stratum of patients who received PCI, there was no difference in major cardiovascular events when compared with medical therapy.

In terms of more recent randomized trials, the SYNTAX trial showed that those with higher SYNTAX scores had higher rates of repeat revascularization after PCI than after CABG.¹⁴⁹ More importantly, the FREEDOM trial demonstrated the most definitive and current evidence that even in the DES era, CABG confers survival benefit over PCI in diabetics with multivessel disease. It compared CABG to PCI with sirolimus- and paclitaxel-eluting stents in 1900 patients with multivessel disease and diabetes and found that 5-year rates of a composite endpoint of death from any cause, nonfatal MI, or nonfatal stroke were significantly lower in the CABG group compared to PCI (18.7 vs 26.5%)²⁶, supporting the results of the BARI and BARI-2D trials, but with the added caveat that the difference in outcomes was driven by a lower rate of MI and death from any cause in the CABG group.

Diabetes is a condition characterized biologically by an inflammatory, proliferative, and prothrombotic state. This may account in part for the increased risk of restenosis and occlusion. Because diabetics tend to have more diffuse disease, the importance of complete revascularization, which is more often achieved surgically that percutaneously, may be enhanced. Another explanation may have more to do with patient selection than vascular biology. It has long been recognized from the previously cited studies that the survival advantage of CABG over medical therapy is greater the more extensive the coronary disease; and more recent studies of PCI versus CABG have demonstrated similar trends. Diabetic patients tend to have more extensive disease, and in BARI diabetic patients had a higher frequency of three-vessel disease, diffuse disease, proximal LAD disease, and left ventricular dysfunction.

From a clinical standpoint with regard to patient selection for coronary revascularization and the method of revascularization, the assessment of diabetics should be made on standard principles, namely the severity and extent of coronary disease, the potential for complete revascularization, the presence or absence of left ventricular dysfunction, and the technical suitability of the lesions for PCI. The results of the aforementioned studies suggest that a preference for surgical over percutaneous revascularization remains appropriate among diabetics, especially those with complex disease and/or ventricular dysfunction. As evident from the results of the FREEDOM and SYNTAX trials, this recommendation has not changed since the introduction of DESs.

END-STAGE RENAL DISEASE

End-stage renal disease (ESRD) is a growing problem. Cardiovascular disease is the most common cause of death among those with ESRD; therefore, there will likely be high rates of revascularization required in these patients in the future. Comorbidities complicating surgical or percutaneous revascularization, such as diabetes, hypertension, and calcified vessels, are more common in patients with ESRD, increasing the risk of intervention. A study conducted by the Northern New England Consortium found dialysis-dependent patients with renal failure to be 3.1 times more likely to die after CABG after adjusting for known risk factors (OR 3.1, [2.1 to 3.7], p < .001).¹⁵⁰ They also found significantly increased rates of mediastinitis (3.6% vs 1.2%) as well as postoperative stroke (4.3% vs 1.7%). The long-term survival was also decreased with renal failure, which was found to be a highly significant predictor of mortality after adjustment.¹⁵¹ Despite these risks, the prognosis without surgical correction of CAD is poor. Revascularization in patients with ESRD is associated with improved survival compared with medical management,^152-154 with some studies showing that CABG confers improved survival compared to PCI.^{152,153,155-159}

Concomitant Carotid Disease

Perioperative stroke remains one of the most dreaded complications associated with CABG, and aside from the debilitating features of the stroke itself, it is associated with 21 to 23% hospital mortality.^160,161 Concomitant carotid artery disease is present in approximately 8% of patients undergoing CABG (range 2-22%, depending on the definition of stenosis, method of diagnosis, and frequency of screening).^162-165 Its presence increases the risk of perioperative stroke in some^166-169 but not all studies,^170-173 and as such, the question of whether to intervene on the carotid artery, with the goal of decreasing the perioperative as well as long-term stroke rate, has garnered significant attention in the literature. Controversy continues regarding the utility and indications for carotid endarterectomy (CEA) either before (staged), during (concomitant), or after (reverse-staged) CABG, especially in patients with asymptomatic carotid artery stenosis for whom data regarding increased risk of perioperative stroke are less wellestablished. This section will describe the relationship between extracranial carotid artery stenosis and neurologic outcomes after CABG and discuss the approach to determining whether and in what manner intervention on the carotid artery is indicated.

PERIOPERATIVE STROKE

Incidence and Causes of Perioperative Stroke

The overall incidence of perioperative stroke or TIA after CABG has been decreasing in the last two decades. Several large registry studies from prior to 2002 had estimated the incidence to be around 3%,^160,174 but recently a large registry study of almost 1.5 million patients undergoing isolated CABG between 2000 and 2009 in all STS-participating institutions has put that risk at about 1.2%.¹⁷⁵ The most common cause of strokes after CABG is not embolism from carotid plaques, but rather athero- or thromboembolism from complex plaques liberated from the ascending aorta as a result of direct physical manipulation inherent in the conduct of the CABG operation including cross-clamping, aortic cannulation, and the creation of proximal graft anastomoses.^176,177 Other causes include intracardiac emboli which can arise in the setting of valvular disease, atrial fibrillation, prosthetic valve implantation, suture lines, left-sided heart catheters, mural thrombus after MI, and less commonly, entrapped air. Finally, miscellaneous causes of perioperative stroke include small-vessel occlusive disease, cerebral hypoperfusion from low perfusion pressure during cardiopulmonary bypass, perioperative MI, acute arterial dissection from cannulation, and cerebral hemorrhage. Overall cardioembolic sources including those from the aorta and the aortic arch account for about 75% of perioperative strokes after CABG, while large artery stenosis, which includes carotid artery disease, accounts for 5%.¹⁷² Although carotid disease is implicated in only a small minority of perioperative strokes, it is the one situation in which the surgeon can take definitive action to remove the pathology and in so doing, potentially also decrease the risk associated therewith.

Risk Factors for Perioperative Stroke

Age is one of the most well-established risk factors for the development of perioperative stroke. In a study from 1986, Gardner et al, showed that stroke rates increased with age, such that those younger than 45 years had stroke rates of 0.2%, rising to 3.5% for those in their 60s and 8.0% for those older than 75 years.¹⁷⁸ Another study in 1992 showed that stroke rates were 0.9% for those younger than 65 years but 8.9% for those older than 75 years. Other risk factors from the cardiac surgery literature include aortic calcification, renal failure, prior stroke, tobacco use, age, peripheral vascular disease, diabetes, and carotid artery disease.^161,179

Relationship of Carotid Stenosis to Perioperative Stroke

There are several major studies dating back to the 1980s that show an increased risk of perioperative stroke after cardiac surgery in patients with significant carotid artery stenosis. In 1987, Brener and colleagues showed that among 4047 patients undergoing cardiac surgery, stroke or transient ischemic attacks (TIAs) occurred in 9.2% of patients with asymptomatic carotid artery stenosis (defined as greater than 50% luminal narrowing on carotid angiography), while those without had a stroke rate of 1.9%.¹⁶⁶ Faggioli and colleagues investigated this in 1990 and found similarly that those with >75% carotid stenosis and age over 60 years, the rate of stroke was 15% versus 0.6% in patients in the same age group without carotid stenosis.¹⁶⁷ CEA appeared to have a protective effect; none of the patients who underwent this procedure concomitantly (19 patients) had strokes, while 14.3% (4 of 28 patients) who had carotid stenosis but did not undergo CEA developed strokes.¹⁶⁷ For patients 65 years or older, who are at higher risk of stroke from any cause to begin with, the degree of carotid artery stenosis was shown to affect the perioperative stroke rate such that the total neurologic event rate (stroke or TIA) was 2.5% for <50% stenosis, 7.6% for ≥50% stenosis, 109% for ≥80% stenosis, and 10.9% for unilateral occlusion.¹⁸⁰

It remains unclear, however, to what extent the presence of carotid stenosis in this patient population is the direct causative element in the development of perioperative stroke, or whether its main significance is as a marker of advanced overall cardiovascular disease. Determining which of the strokes occurring in those with significant carotid stenosis actually occur ipsilaterally has given further insight into what the actual impact of performing CEA would be, especially in asymptomatic patients. Several small retrospective studies have shown that the perioperative stroke rate directly attributable to ipsilateral carotid stenosis is small in asymptomatic patients.^172,173 Li et al showed that among 18 patients with ≥50% carotid stenosis who developed perioperative stroke after CABG, only four occurred on the same side as the disease itself, and that of these four, the carotid was totally occluded in three patients, that is, not amenable to intervention. Therefore only one of 18 strokes was potentially preventable by CEA. Since 2005, there have been four studies demonstrating a 0% rate of stroke in patients with asymptomatic carotid stenosis of ≥50% (n = 156) and ≥70% (n = 42).^{170,171,181,182} These data suggest, though not definitively, that at least for asymptomatic patients, the risk of perioperative stroke directly attributable to the carotid stenosis itself may be minimal, lending credence to the idea that carotid disease is perhaps more of a surrogate and that prophylactic CEA may have little benefit.

Relationship of Uncorrected Carotid Stenosis to Late Stroke

The potential benefit of CEA in patients who have asymptomatic carotid artery disease has been shown to extend beyond the immediate perioperative period. Barnes et al showed that at 22 months following CABG in 40 patients with untreated asymptomatic carotid stenosis, the mortality rate was 10% while 17.5% had suffered a stroke.¹⁸³ One half of the patients had progression of the severity of carotid disease by noninvasive testing. Another study found that at 48 months, the risk of stroke after CABG in the setting of uncorrected significant carotid disease was 10%, that is, tenfold higher than patients who underwent combined CABG and CEA.¹⁸⁴ Contemporary randomized trials of surgery versus medical therapy for significant carotid stenosis have defined the late risk of carotid-related stroke in medically treated patients. In the Asymptomatic Carotid Surgery trial, actuarial risk of stroke at 5 years was 12% in medically treated patients with asymptomatic high-grade carotid stenosis.¹⁸⁵

INDICATIONS FOR NONINVASIVE CAROTID TESTING

While some centers have practiced a policy of screening all patients undergoing CABG for carotid disease, studies have shown that identifying patients who are at high risk of having significant carotid disease to begin with can be screened selectively with negligible impact on the overall sensitivity. A retrospective analysis of 1421 patients undergoing CABG limited routine screening by ultrasound to only those that met the following criteria: age over 65 years, history of a stroke or TIA, and presence of a carotid bruit.¹⁸⁶ The investigators determined that this reduced the need for preoperative testing by 40% overall while missing only 2% of all candidates with ≥70% carotid stenosis. The ACCF/AHA issued a Class IIa recommendation that carotid artery duplex scanning should be performed for patients who have clinical features associated with a high risk of concurrent carotid artery stenosis, including age over 65 years, left main coronary stenosis, peripheral vascular disease, history of stroke or TIA, hypertension, smoking, and DM.¹⁰

MYOCARDIAL ISCHEMIC EVENTS IN PATIENTS AFTER CAROTID ENDARTERECTOMY

The incidence of CAD in the general population of patients undergoing CEA is high. A study from the Cleveland Clinic found that in patients studied with routine preoperative coronary angiography prior to planned CEA, only 7% had normal coronary arteries, while 28% had mild-to-moderate CAD, 30% had advanced but compensated disease, 28% had severe but correctable disease, and 7% had severe, inoperable CAD.¹⁸⁷ Early studies demonstrated that the risk of perioperative MI in the setting of CAD as significantly higher (4.3% compared to 0.5%).¹⁸⁸ Hertzer and colleagues showed that for patients undergoing CEA, MI caused more late deaths (37%) than did stroke (15%), with improved 10-year survival in the cohort of patients who had undergone incidental CABG compared to those with suspected but undocumented coronary disease (55% vs 32%).¹⁸⁹

COMBINED CORONARY AND CAROTID REVASCULARIZATION

Guidelines for Combined Coronary and Carotid Revascularization

To date, there exist no randomized trials comparing a combined or staged approach to carotid disease in asymptomatic patients to CABG alone. The CABACS trial which began in 2010 is currently ongoing and will report on the relative incidence of strokes and death from any cause by randomly assigning patients to concomitant (ie, synchronous) CEA and CABG versus CABG alone in patients with high grade carotid stenosis undergoing CABG.¹⁹⁰ Until then, the current guidelines from the ACCF/AHA are based on a large body of retrospective data that have not conclusively established the optimal approach to treating this patient population. Indeed, the formal Class I recommendation is that for patients with clinically significant carotid artery disease for whom CABG is planned, a multidisciplinary team consisting of a cardiac surgery, cardiologist, neurologist, and vascular surgeon should convene and determine an individualized plan regarding whether and when to perform CEA.¹⁰ Meanwhile, for patients with a previous TIA or stroke and a significant carotid artery lesion (≥50% stenosis), combined CABG and carotid revascularization is recommended on a Class IIa basis. The sequence and timing of intervention is determined by the relative magnitudes of cerebral and myocardial dysfunction.¹⁰ Finally, in patients scheduled for CABG who have no history of TIA or stroke but have bilateral severe (70-99%) carotid stenosis or unilateral severe carotid stenosis with a contralateral occlusion, combined carotid revascularization is recommended on a Class IIb basis.¹⁰

These guideline recommendations notwithstanding, Cambria and colleagues at our institution have argued that if one accepts that (1) uncorrected carotid stenosis is associated with an increase in stroke risk for patients with severe carotid and CAD who have only isolated CABG; (2) CEA is the indicated treatment for severe symptomatic and asymptomatic carotid stenosis; (3) CAD increases the early and late risk of death for CEA patients; and (4) CABG is an indicated treatment for CAD, then the important question becomes not the indication for but the timing of the two operative procedures. It is on these grounds that some surgeons in our institution have, since the 1970s, taken an aggressive approach to patients with both CAD and carotid stenosis, with the concomitant operation used as the standard approach.

Staged versus Concomitant Carotid and Coronary Artery Operations

The technique of performing CEA before coronary bypass grafting is referred to, by convention, as a staged procedure, while performing the CABG first and the CEA in delayed fashion is termed a reverse staged procedure. Performing both procedures under the same general anesthetic administration is a concomitant procedure.

On the one hand, performing CEA first might increase the risk of perioperative MI, while performing CABG first might increase the risk of perioperative stroke. If the patient does not have signs of active ischemia or hemodynamic instability, especially if the CEA can be performed safely with regional anesthesia, most agree that performing CEA first is a reasonable strategy. This argument is strengthened further in elderly patients with a history of prior stroke or TIA in whom the incidence of perioperative MI might be expected to be significantly higher. Meanwhile, for patients with active ischemia and/or hemodynamic instability secondary to coronary disease, the reversed staged procedure is sometimes favored, especially for those with asymptomatic carotid stenosis and no history of stroke or TIA. These general principles are considered on a case-by-case basis without the benefit of any randomized controlled trial data to definitively support one strategy versus the other.