|Home | About | Journals | Submit | Contact Us | Français|
Medical, percutaneous, and surgical therapies for coronary atherosclerotic disease are developing rapidly, with many recent breakthroughs in metabolic control, improvements in catheter and stent engineering, and advances in surgical technique. Treatment guidelines are still in their infancy and do not take into account several of these recent innovations. Consequently, determining the most appropriate treatment for many patients remains challenging. In this review, we examine the most recent revascularization guidelines, discuss important new data and trials comparing contemporary stent technology and coronary artery bypass surgery, and conclude with updated revascularization recommendations.
Coronary atherosclerotic disease (CAD) is the most common killer of Americans, causing half a million deaths in 2005.1 Although the prevalence of CAD climbed to 16.8 million in 2006, the mortality rate has declined, largely because of improvements in medical, surgical, and percutaneous therapies.1,2 Surgical revascularization by coronary artery bypass (CAB) was proved in the 1970s to be effective therapy for severe CAD. Since then, percutaneous coronary intervention (PCI) with balloon angioplasty augmented by stent implantation has become the primary method to treat severe angina pectoris in patients with less diffuse CAD and to abort myocardial infarction (MI) in patients with unstable symptoms.
Because both surgical and percutaneous therapies are changing rapidly, determining the most appropriate treatment for many patients can be difficult. The use of drug-eluting stents (DESs) instead of bare-metal stents (BMSs) has reduced the rate of restenosis after PCI, and DESs are now used in most of the PCIs in the United States.3,4 New surgical techniques, including off-pump CAB (OPCAB) and the use of multiple arterial conduits, may reduce perioperative morbidity and improve long-term graft patency rates, respectively.5,6 Comparative data on PCI with DESs versus contemporary CAB are just now beginning to accumulate and have not yet been fully incorporated into the latest guidelines. Furthermore, the duration of follow-up in these studies does not allow a realistic comparison of procedural durability and protection from future events. In this review, we look briefly at the recent history of coronary revascularization, examine the most recent revascularization guidelines, discuss important new data and trials comparing PCI with DESs to CAB, and conclude with updated revascularization recommendations.
A brief look back at the development of coronary revascularization is relevant, because support for many of the current guidelines is based on data from the last 3 decades. The Veterans Affairs Cooperative Study,7–10 the Coronary Artery Surgery Study,11–14 and the European Cardiac Surgery Study15 were landmark trials conducted during the 1970s and 1980s that together established the efficacy of CAB. Compared with initial medical therapy with the option of CAB in the future, CAB at the time of randomization provided superior relief of angina and reduced mortality rates in the highest-risk groups: patients with left main coronary artery stenosis greater than 50% or equivalent left main disease, patients with depressed left ventricular (LV) systolic function and either 3-vessel disease or 2-vessel disease involving the proximal left anterior descending coronary artery (LAD), and patients with positive treadmill test results and either 3-vessel disease or 2-vessel disease involving the proximal LAD.7–15 A meta-analysis combining these 3 trials with 4 other randomized trials of similar design showed that CAB provides an absolute risk reduction in mortality rates of 5% at 5 years and 4% at 10 years. The benefit was seen primarily in the high-risk groups described above. In addition, the combined data suggest that the survival benefit extends to patients who have 3-vessel disease with normal LV ejection fractions (LVEFs).16
Percutaneous coronary intervention, first with balloon angioplasty alone and later with BMSs, was developed as a less invasive means of treating obstructive coronary disease. A meta-analysis of 23 randomized controlled trials from the late 1980s to early 2000s summarized the outcomes of 5,019 patients who had single-vessel or multivessel CAD and were treated with PCI: either with balloon angioplasty alone or with BMSs, versus CAB. Survival rates at 10 years were similar for both groups—even in the subgroup of patients with diabetes mellitus. Angina relief at 5 years was more common after CAB (84%) than after PCI (79%; P < 0.001), but procedure-related strokes were slightly more frequent after CAB (1.2%) than after PCI (0.6%; P = 0.002). Repeat revascularization was substantially more common after PCI (absolute risk increase of 33% at 5 years; P < 0.001).17 A Cochrane review of randomized trials of PCI with BMSs, compared with CAB, revealed similar findings.18 The equivalent survival rates with PCI and CAB in these randomized trials is not surprising, because the populations studied were generally lower-risk patients with less to gain from revascularization. Unlike these randomized trials, several large-scale, long-term observational studies of patients with multivessel disease have suggested that CAB may provide a survival advantage over PCI.19–23
Recent practice guidelines from the American College of Cardiology (ACC) and American Heart Association (AHA) regarding revascularization were published before the results of trials comparing PCI with DESs to CAB were reported.24–26 Recommendations from the 2002 practice guidelines for chronic stable angina24 were unchanged in the 2007 update25 and generally match those from the unstable angina/non-ST-elevation MI guidelines from 200726 (Table I).
In 2009, a group of cardiologists, cardiac surgeons, and other healthcare providers and administrators collaborated to publish appropriateness criteria for revascularization in a variety of clinical situations. Coronary bypass was deemed appropriate for all presented scenarios of severe CAD that required revascularization. Percutaneous coronary intervention was rated appropriate for 2-vessel disease that involved proximal LAD regardless of LV function or diabetes. Those authors concluded that PCI in 3-vessel disease was of uncertain benefit but bordered on appropriate in patients without diabetes or depressed LV function.27 Lesion complexity, which can substantially affect the likelihood of obtaining a high-quality result with PCI, was not included in the scenarios, which is a limitation of these guidelines. The SYNTAX score (discussed later in this article) may be an effective tool for incorporating lesion complexity into decision-making about revascularization approaches.
Because they have drastically reduced the rate of restenosis and target lesion revascularization, DESs are arguably the most important advance in PCI in recent years. Restenosis after angioplasty is driven by 3 main mechanisms: elastic recoil, negative remodeling (vessel contracture), and neointimal hyperplasia.28–30 Bare-metal stents effectively prevent recoil and negative remodeling but may actually exacerbate neointimal hyperplasia.31–33 Drug-eluting stents release antiproliferative medications that inhibit in-stent neointimal hyperplasia, thereby countering all 3 mechanisms of restenosis.34 A meta-analysis found no significant difference in rates of death or MI in trials that involved 5,216 patients who were randomly assigned to receive 1 of 2 types of DES (sirolimus-eluting or paclitaxel-eluting) or a BMS, with a 4-year follow-up. However, rates of target lesion revascularization were greatly reduced by both types of DES (sirolimus-eluting stent vs BMS, 7.8% vs 23.6%; P < 0.01; paclitaxel-eluting stent vs BMS, 10.1% vs 20%, P < 0.01).4 Later, a much larger network meta-analysis of 38 randomized trials and 18,023 patients with follow-up of up to 4 years produced similar findings.3
The initial randomized trials that led to U.S. Food and Drug Administration approval of DESs were limited to single, previously untreated lesions less than 30 mm long, in vessels 2.5 to 3.75 mm in diameter. Subsequently, DESs have been shown in small trials and registries to be better than BMSs in unrestricted real-world use and for off-label indications.35,36 Limited data from subgroup analyses, small trials, and registries suggest that DESs are associated with less repeat revascularization than BMSs in small vessels, long lesions, chronic total occlusions, saphenous vein grafts (SVGs), patients with diabetes, and during primary PCI for acute MI.37–47
Stent thrombosis is a rare but dreaded complication of PCI with either BMSs or DESs and has an associated death rate of 31%.48 Initial studies of DESs versus BMSs used various definitions for stent thrombosis and, therefore, reported variable rates. A meta-analysis of 14 randomized trials involving 6,675 patients examined thrombosis rates as defined and reported by the individual trials; the results suggested that DESs led to a 4- to 5-fold increase in very late (>1 year after PCI) stent thrombosis than did BMSs, with 5 events per 1,000 DES recipients but no events in the BMS recipients.49 However, by excluding events that occurred after a repeat revascularization, the study was biased against DESs, whose recipients were less likely to need repeat revascularization. A more recent meta-analysis of randomized trials, which used patient-level data and the Academic Research Consortium definitions for stent thrombosis, revealed similar rates of definite or probable thromboses during the 4 years after PCI with DESs or BMSs (sirolimus-eluting stents vs BMSs, 1.5% vs 1.7%; P = 0.7; paclitaxel-eluting stents vs BMSs, 1.8% vs 1.4%; P = 0.52).48 Two other meta-analyses of randomized controlled trials with a combined population of 23,284 also found similar rates of in-stent thrombosis within 4 years of PCI with either DESs or BMSs, but most of the thromboses in BMSs occurred during the 1st year, whereas thromboses continued to occur in DESs during the 2nd, 3rd, and 4th years.3,4
Determining the hemodynamic significance of a coronary stenosis by measuring the fractional flow reserve (FFR) of the stenotic artery is a valuable new technique that can improve interventional decisions and patient outcomes.50,51 Fractional flow reserve is measured with a coronary pressure wire after adenosine administration to achieve maximal coronary flow. The ratio of the pressure distal to the lesion to the aortic pressure equals the ratio of maximal blood flow in the stenotic artery to normal maximal flow and defines the FFR.52 The FFR in a normal coronary artery is 1; lower values indicate more functionally significant stenosis. In the DEFER study, patients with a single coronary stenosis of more than 50% but an FFR of 0.75 or greater had similar 5-year rates of composite death and acute MI whether PCI with BMS was performed at that time or was deferred (7.9% vs 3.3%; P = 0.21). The percentages of patients free of chest pain and needing target-vessel revascularization were also similar.50 In the FAME study, patients with multivessel disease and an indication for PCI were treated with DESs after randomization into 2 groups: 1 in which all angiographically significant lesions were stented, and 1 in which FFR was measured on all such lesions and only the ones with an FFR of 0.8 or less were stented. Patients in the FFR-guided arm of the study required fewer stents and at 1 year had less composite death, MI, and repeat revascularization than did patients in the angiography-guided arm (13.2% vs 18.3%; P = 0.02). These studies indicate, as would be expected, that the benefits of PCI are limited to vessels with hemodynamically significant stenoses. In addition, the results illustrate that visual assessment of angiographic stenosis severity can be improved with the addition of FFR measurement.
Although CAB is performed on older patients with more comorbidities and more complex CAD than ever before, CAB-related perioperative mortality rates continue to decline and are now reported to be 1% to 2%.6,53,54 Refinements in cardiopulmonary bypass (CPB), intraoperative myocardial preservation, and perioperative care, including intensive glucose control, all improve outcomes.5 Two important trends during the last decade have been an increase in the use of both OPCAB and multiple arterial bypass conduits.
During CPB, blood is exposed to artificial materials that can trigger a systemic inflammatory response, resulting in dysfunction in multiple organs.55 In an attempt to avoid this proinflammatory state, OPCAB is performed on a beating heart through a traditional median sternotomy without CPB. This can be done without cross-clamping the aorta, which avoids 1 possible contributing factor for perioperative stroke. Despite evidence associating OPCAB with reduced markers of inflammation,56,57 rates of the most important clinical endpoints, such as death, MI, and stroke, appear to be similar in randomized trials of OPCAB versus traditional CAB with CPB in patients from mixed populations.53,58 A meta-analysis of 37 such randomized controlled trials revealed no significant difference in 30-day mortality, MI, stroke, renal dysfunction, wound infection, or reintervention rates. However, OPCAB did reduce rates of atrial fibrillation, inotropic support, and respiratory infection, as well as ventilation time and length of intensive care unit and hospital stays.53 In contrast with these studies, a meta-analysis that included nonrandomized trials associated OPCAB with reductions in perioperative death (odds ratio [OR], 0.64), MI (OR, 0.58), and stroke (OR, 0.55) (P < 0.001 for all 3 outcomes).59 A scientific statement issued by the AHA in 2005 stated that definitive conclusions could not be drawn, but that trends in most studies suggest that OPCAB leads to less blood loss, less perioperative myocardial enzyme release, less early neurocognitive dysfunction, and less renal insufficiency than does traditional CAB.60 The latest meta-analysis (from 2009) of 10 randomized controlled trials found no significant difference in the rates of death, MI, stroke, and repeat revascularization associated with OPCAB versus CAB with CPB.58 Notwithstanding these mixed results, it seems likely that OPCAB provides some degree of protection for patients at highest risk for stroke, such as patients with atheromatous or calcified ascending aortas, patients with chronic obstructive pulmonary disease, and those with renal insufficiency. Conventional CAB is still preferred for most patients without these features, patients with emergent or unstable conditions, and patients with complex CAD for which better exposure and cardioplegia are beneficial.5,6
The superiority of the left internal mammary artery (LIMA) over the SVG as a conduit to the LAD has been clear since the 1980s, when the 10-year patency rates of LIMA grafts were shown to be 80% to 95%.61–66 Since then, surgeons have used a variety of arterial conduits with the hope of producing similar patency rates. Unfortunately, other arterial conduits appear to be less effective than the LIMA. This may be explained partly by factors intrinsic to the arteries themselves, the different flow requirements of the non-LAD bypass territories to which some of these conduits are frequently grafted, and the importance of higher flows for maintaining arterial bypass patency. Right internal mammary artery (RIMA) grafts have shown patency rates between those of SVGs and LIMA grafts.67,68 Using both IMAs in multivessel disease has been evaluated in multiple observational studies,69,70 and a meta-analysis of these observational data suggests that dual IMA grafts lead to better survival than do single IMA grafts and SVGs.71 However, not all studies have found this survival benefit,72 and data from randomized trials are lacking. Furthermore, enthusiasm for the RIMA has been diminished by reports of longer harvest times and a 2.5- to 5-fold higher risk of mediastinal infections.73 The radial artery appears more prone to spasm during harvesting and to thrombosis when faced with competitive flow after being grafted to less severely stenosed native vessels.5,6 Several observational studies and at least 1 randomized control trial have produced mixed results, providing no convincing evidence that radial artery grafts produce better short- to midterm results than do SVGs.74 In the future, studies with follow-ups extending into and beyond the period when SVGs begin to fail will help to clarify the merits of the radial artery conduit. One retrospective study found that 12-year survival was better after CAB with 2 or more arterial grafts (IMA or radial artery) than after CAB with a single IMA graft and SVGs.75 Total arterial OPCAB is being performed with excellent short- and midterm results as reported in observational studies (Table II), but its efficacy, compared with conventional CAB, has not yet been proved in randomized studies.76–81
Comparative data are beginning to accumulate concerning the outcomes of PCI with DESs and of CAB with use of current techniques in patients with multivessel disease. Of the several observational studies that have been reported (Table III),82–90 most show that patients with multivessel CAD who are treated with DESs or CAB have similar rates of death, MI, and stroke, but repeat revascularization is more frequent after DES. Two studies showed lower mortality rates after CAB: one was the New York registry study,83 but the benifit was shown only after risk adjustments. The other was the overall population reported by Javaid and colleagues84; however, after adjustments, only the CAB patients with diabetes mellitus received the survival benefit. Most of these studies show that diabetic patients have worse outcomes regardless of treatment approach, and some but not all of these studies suggest that diabetic patients receive more benefit from CAB than from PCI with DESs.82,84–90 Two studies that analyzed subgroups with low LVEFs produced contradictory results: Hannan and associates83 found a survival benefit with CAB over DES placement, whereas Park and coworkers86 found no difference. The only study that looked at costs and time in the hospital showed that treatment with DESs was less expensive and required less time in the hospital than did CAB.89 All of these studies are limited by the potential for bias that is inherent in observational studies. In addition, most are single-center studies of relatively small size and short follow-up.
The SYNTAX trial, whose results were recently published, was the 1st large randomized controlled trial of PCI with DESs versus CAB in patients with severe CAD.91 This study included patients from 17 countries. Each patient had 3-vessel or left main coronary disease with either chest pain or documented ischemia. Exclusion criteria were limited to prior PCI or CAB, acute MI, and other indications for cardiac surgery. All patients' angiograms were reviewed by an interventional cardiologist and a cardiac surgeon. Patients who could be revascularized equally well with PCI or CAB were randomized 1:1 to PCI with TAXUS® Express® paclitaxel-eluting stents (Boston Scientific Corporation; Natick, Mass) or CAB. Patients deemed better treated by 1 approach than the other were entered into nested registries. The baseline characteristics of the 1,800 randomized patients were similar in both arms of the study and included diabetes mellitus in 25% of patients and congestive heart failure or LVEF of less than 0.30 in 2% to 5%. Patients in both groups had higher risk scores for PCI and CAB than did patients in most previous PCI-versus-CAB trials. The goal of revascularizing all vessels that were 1.5 mm in diameter or larger with 50% or greater stenosis led to a mean of more than 4 stents placed per patient in the PCI group, and 63% involved at least 1 bifurcation or trifurcation. The CAB group averaged 3 conduits per patient; 97% of cases involved at least 1 arterial conduit, and 15% were performed off-pump. Medical therapy was not controlled, and it differed between the 2 groups: the PCI patients received more aspirin, clopidogrel, statins, β-blockers, and angiotensin-converting enzyme inhibitors, whereas the CAB patients more often received amiodarone and warfarin. At 12 months, the primary outcome of major adverse cardiac or cerebrovascular events (MACCE) (death, MI, stroke, or repeat revascularization) occurred more frequently after PCI than after CAB (17.8% vs 12.4%; P = 0.002). However, this difference was driven by higher rates of repeat revascularization after PCI (13.5% vs 5.9%; P < 0.001); the rates of death and MI were similar between the 2 groups. Stroke occurred more often after CAB (2.2% vs 0.6%, P = 0.003). The rates of symptomatic graft occlusion after CAB and stent thrombosis after PCI were both approximately 3%.
A predictive tool that could help objectively identify the superior revascularization method, the SYNTAX score, was developed as part of the SYNTAX study. The score characterizes a patient's CAD with respect to the number of lesions and their functional impact, location, and complexity.92 In patients with SYNTAX scores of 33 or higher (indicating diffuse or complex CAD that is hard to treat percutaneously), the rate of MACCE was more than twice as high after PCI (23.4%) as after CAB (10.9%; P < 0.001). Patients with low or intermediate SYNTAX scores had similar rates of MACCE with either revascularization approach.
The limitations of this study included the variation in medical therapy between the 2 groups. More patients in the PCI group received medications that have been proved to reduce cardiovascular events, which may have biased the results in favor of PCI. In addition, relatively few women and patients with diabetes mellitus or heart failure were included, so the SYNTAX data may or may not be applicable to these important populations. Furthermore, 1 year of follow-up may not have been enough to reveal the full benefits of CAB; other studies, such as ERACI III,87 showed better outcomes with DES during the 1st year but relatively more benefit from CAB over a longer period. Finally, the goal of achieving total revascularization may have been too ambitious in light of the FFR data presented above, which suggest that treating intermediate stenoses medically produces similar or better outcomes than PCI with DESs.
Coronary artery bypass surgery offers a clear survival advantage over medical therapy in patients with >50% stenosis in the left main coronary artery.16 Despite the lack of clear evidence for their long-term safety and efficacy, DESs are being used in many of these patients. A 2004 survey found that 21% of patients in North America and 26% of patients in Europe with left main disease were treated with PCI.93 Taggart and colleagues,94 who examined data from 7 different groups that used DESs for left main disease in 599 patients, summarized the outcomes as follows: in-hospital death averaged 2.4% (range, 0–11%), the immediate repeat revascularization rate was 2% (0–6%), and biochemical evidence of periprocedural MI was found in 6% of patients (0–9%). At a mean follow-up of 11 months, the mortality rate was 7% (range, 0–14%), and the repeat revascularization rate was 13% (2%–38%). The wide ranges of outcomes were probably due to the variety in patient selection, interventional techniques, and location and complexity of the disease. Outcomes appear to be much better when the disease is proximal and does not involve the distal bifurcation or trifurcation.95,96 In 2 registry studies that compared the outcomes of CAB versus PCI with DESs in patients with left main disease, both treatments produced similar 1-year mortality rates, but repeat revascularization was more frequent after PCI.97,98 To date, there has not been a large randomized controlled trial comparing CAB to PCI with DESs in left main CAD. The closest is the SYNTAX trial, in which subgroup analysis of patients with left main disease revealed similar 1-year rates of MACCE for PCI with DESs versus CAB (16% vs 14%; P = 0.44). More frequent repeat revascularizations were required after DES (12% vs 7%; P = 0.02), but more frequent strokes occurred after CAB (0.3% vs 2.7%; P = 0.01).91
Given the current data, CAB remains the standard of care for patients with left main stenosis. This treatment will likely continue to be the best option for most of these patients, because most left main disease is accompanied by other features that make PCI challenging (such as bifurcating or multivessel disease and heavy calcification). However, it seems that PCI with DESs can achieve reasonable results and is still an option for patients who are hemodynamically unstable or ineligible for CAB—especially those without distal bifurcating disease. Further randomized studies are warranted.
Because revascularization can be achieved equally well with PCI or CAB in many patients, a reliable method of estimating a given patient's risk with each procedure would be helpful. Several risk prediction scores have been validated for CAB and PCI individually,99,100 but these have not been applied to both approaches concurrently. These scores frequently do not help one determine the risk-to-benefit ratio of a given procedure, because many of the same features that suggest high periprocedural risk also predict the most potential benefit (for example, LV dysfunction). In addition, risk scores rarely help determine the merits of 1 revascularization approach over the other, because many of the risk factors are shared (such as age, LV dysfunction, and urgency). However, risk scores based on complexity of CAD from a PCI standpoint, such as the ACC/AHA lesion classification system and the SYNTAX score, may not be subject to this limitation. The SYNTAX study described above showed that the SYNTAX score could define a group of patients whose MACCE rates were significantly higher after PCI than CAB,91 so this score could be a useful tool in guiding decisions about revascularization approach.
Do patients with certain comorbidities do better with 1 procedure or another? Answering this question with reasonable certainty would require large, prospective, randomized trials designed specifically to isolate the comorbidity in question; such studies generally have not been done. Most attempts to compare PCI and CAB outcomes in patients with particular risk factors or comorbidities have been retrospective, nonrandomized, subgroup analyses, which are limited by potential selection bias and by variations in baseline characteristics; these limitations make the validity of any conclusions questionable.
With these limitations in mind, the preponderance of evidence appears to provide weak support for the following statements: First, patients with severe pulmonary disease have particularly high perioperative mortality rates after CAB.101–103 Second, patients with diabetes mellitus have particularly high rates of repeat revascularization after PCI and may survive longer after surgery—particularly if the surgery involves a LIMA-to-LAD bypass.104–106 Third, patients with reduced LV systolic function may live longer after CAB than after PCI.15,107,108 Fourth, patients with dementia have a decreased rate of further cognitive decline after PCI than after CAB.109,110 The literature is mixed and indeterminate for patients with chronic kidney disease.111–114 Several other clinical situations and factors that may affect the choice of 1 approach over another are listed in Table IV.
In our current approach (Fig. 1), we recommend CAB for most patients with the highest-risk CAD (that is, the most myocardium at risk and the least reserve), regardless of symptoms, because of CAB's proven survival benefit in this population. For patients with less severe CAD, symptom severity and noninvasive test results are used to stratify patients into 1 of 2 groups: those who will benefit from immediate revascularization and those in whom an initial trial of aggressive medical therapy alone may be safely attempted. In patients without the highest-risk CAD who require revascularization because of unacceptable symptoms, noninvasive test results indicating high risk, or failure of medical therapy, PCI with DESs and CAB appear to result in similar rates of death and MI. Therefore, the choice depends largely upon how effectively the lesions can be treated with PCI, and upon the patient's feelings about the temporary disability and the slightly increased stroke risk associated with CAB versus the increased risk of repeat revascularization with PCI. A patient with 3-vessel disease due to 3 focal stenoses, who is able to take clopidogrel and is willing to accept the risk of repeat revascularization, will probably have excellent results with PCI—especially if FFR measurement reveals that 1 lesion is not hemodynamically significant and can therefore be left alone. On the other hand, a patient with 1- or 2-vessel disease involving chronic total occlusions, bifurcations, or highly calcified or angulated lesions may have better results with CAB. Regardless of the revascularization approach taken, the underlying atherosclerotic disease process continues, and aggressive medical management of risk factors is imperative.
We thank Stephen N. Palmer, PhD, ELS, who assisted with the editing of this report.
Address for reprints: James M. Wilson, MD, St. Luke's Episcopal Hospital (MC 1-133), 6720 Bertner Ave., Room P-322, Houston TX 77030 E-mail: moc.hels@nosliwj
Disclosure: Dr. Wilson has received research funding from the Cordis Corporation to record and examine the outcomes of patients after surgical and percutaneous therapy.