The concept behind coronary bypass surgery is that establishing grafts to route blood distal to coronary stenoses may relieve angina, improve myocardial function, and decrease the likelihood of death associated with coronary artery disease (CAD). Multiple randomized and observational studies have identified subgroups of patients for whom bypass surgery achieves these goals (1, 2, 3, 4). However, advances in percutaneous therapy (PCT) have occurred, the most recent being drug-coated stents. In addition, the pharmacologic treatment of CAD has progressed. In today’s world, then, which patients may receive incremental benefit from bypass surgery?

In the past, the anatomic treatments for CAD—bypass surgery and PCT—have been both complementary and competitive and a brief review of the development of bypass surgery and PCT may help frame our current questions.

First-generation PCT, balloon angioplasty (BA) or percutaneous transluminal coronary angioplasty (PTCA), involved substantial periprocedural risk, largely because of the occurrence of acute or subacute coronary closure or coronary dissection. Even for the low-risk patients receiving PCT in those early years, the risk of periprocedural death was approximately 1%, and 2% to 10% of patients underwent urgent or emergency operation because of ongoing myocardial ischemia. Even if initial PTCA was successful, the late outcomes were uncertain. Restenosis of the areas of the coronary artery treated with BA was common, occurring in 25% to 50% of treated vessels within the first postoperative year. Restenosis usually was caused by intimal hyperplasia in the area of the endothelial injury.

Post-PTCA symptoms were common, as was reintervention with repeat PCT or bypass surgery, both for patients with single-vessel disease and for those relatively few patients treated with PTCA for multivessel disease. In the randomized Bypass Surgery Angioplasty Revascularization Investigation (BARI) trial of patients with multivessel disease, those patients treated with PCT who had a 1-year postoperative angiogram had angina in 29.6% of cases (compared to 11.9% of surgical patients) and, to achieve even that degree of angina relief, required 48% of PCT patients to undergo either repeat PCT or bypass surgery within the first postoperative year (5). The increased risk of symptoms for the PCT patients correlated with an increased level of myocardial jeopardy that was based both on less complete revascularization at the time of the procedure and an increased risk of treatment failure in the form of restenosis.

A number of randomized trials were undertaken that were designed to see whether patients paid a survival penalty when treated with an initial PTCA strategy instead of having surgery. In general, the patients who were randomized had low-risk characteristics. The majority of patients who underwent angiographic study for a possible inclusion were excluded from entry in this study because of angiographic unsuitability for PTCA, and the majority of patients randomized had two-vessel rather than three-vessel disease. Most of the studies showed equivalent 5- to 7-year survival rates after PTCA and surgery, although patients treated with PTCA had less effective revascularization, more symptoms, took more medication, and had
many more reinterventions. In the largest trial (BARI), diabetic patients did have a decreased long-term survival rate when treated with PCT (6).

The BARI registry was an observational portion of the BARI trial that contained patients who were eligible for randomization but who were not randomized, usually because of physician preference. In that registry, diabetic patients treated with PTCA had less extensive disease than those treated with surgery, and a comparison of these subgroups showed an equivalent survival rate that was better than either of the randomized subgroups. Thus, whereas diabetes as an isolated descriptor did not predict an invariably bad outcome after PTCA, those and other data indicate that there appears to be something different about diabetic patients in regard to the effectiveness of PTCA. This principle has been noted in multiple observational studies of PTCA involving diabetic patients. Finally, the time course of death for the PCT-treated diabetic patients did not correlate with the time course of restenosis.

Second-generation PCT involved the use of bare metal stents (BMSs) combined with effective platelet inhibitors, the glycoprotein IIb-IIIA inhibitors. The most profound effect of these technical and conceptual advances was to decrease the procedure-related risk of PCT. The ability to control localized coronary dissection and/or perforation resulted in a decreased risk of acute and subacute coronary closure and, combined with the increased experience of interventional cardiologists, lowered the procedure-related risks of death, myocardial infarction (MI), and emergency surgery associated with PCT.

The improvement in long-term outcomes based on BMS was not as dramatic as the improvement in procedure-related safety. Although the risk of restenosis dropped roughly in half, from 40% to 50% to 20% to 25%, in randomized trials of patients with multivessel disease, PCT still produced inferior symptom relief and a relatively high risk of reintervention when compared with surgery (7,8). The location of early restenosis was usually within the stent and often extended the length of the stent. Also, patients with diabetes continued to experience substantially inferior outcomes when compared with nondiabetic patients (8). Randomized trials have shown either no survival differences or a slight survival advantage for the surgically treated patients, although, again, the mortality rates for both the surgery and the PCT groups were quite low, based on the favorable risk profile of the randomized patients (7, 8, 9).

Nonrandomized but more inclusive observational studies containing higher risk profile patients and a broader spectrum of angiographic disease have demonstrated a survival advantage for surgery. Data from The Northern New England Study Group has shown an excess mortality after PCT for patients with triple-vessel disease and for patients with diabetes (10). More recently, a review of Cleveland Clinic Foundation patients undergoing primary nonemergent PCT and surgery, half of whom had either abnormal left ventricular function or diabetes, showed an excess mortality after PCT for patients in all risk groups (11).

Drug-coated or drug-eluting stents (DES) represent the third generation of PCT, and these appear to represent a significant improvement over previous PCT technologies. Overall risks of restenosis of treated coronary arteries within the first postoperative years has decreased to a 5% to 10% in-segment restenosis rate in the initial trials of DES versus BMS (12, 13, 14). In-stent restenosis still occurred, although at a generally reduced level. In addition, restenosis in the area of the stent (in-segment restenosis) appears to be related to hyperplasia at the end of the stent, a problem that may be at least partially ameliorated by implantation techniques. Although patients with diabetes appear to be at an increased risk of restenosis relative to nondiabetic patients, the use of DES decreases the risk of restenosis for diabetic patients relative to previous PCT strategies. Thus, although drug-coated stents may not be the proverbial “home run,” they are at least a double high-off-the-wall. DES coated with both sirolimus and paclitaxel have been studied, and no substantial difference in outcomes yet is apparent based on the drug.

Early studies of clinical outcomes have noted a decreased risk of target vessel revascularization and possibly MI when DES is compared to BMS. Survival differences have not yet been apparent, because the patients involved in the randomized clinical trials have been low-risk patients with good outcomes after both DES and BMS.

However, some problems with drug-eluting stents are becoming recognized. Restenosis is a problem at branch points, and occurs at roughly the same frequency seen with BMS. Chronic total occlusions are still a problem if long segments of the coronary vessel are involved. Some patients have such diffuse distal disease that coronary vessels too small to accept stents and/or multiple branch points are involved. The most serious issues with DES, however, are subacute thrombosis of stents and the uncertainty of whether diminishing the risk of restenosis improves the long-term survival rate for patients with life-threatening CAD (15).

The success of DES has gone hand-in-hand with treatment using platelet inhibitors, usually clopidogrel. The idea has been that the drug effect that prevents hyperplastic restenosis may also delay endothelial remodeling and decrease the likelihood of stent thrombosis. In-stent thrombosis has not been common as long as patients are treated with clopidogrel (about 1% within 3 postprocedure months), but that risk appears to increase if clopidogrel is discontinued. The hope had been that after 6 months, platelet inhibitors could be discontinued, but as experience accumulates, it has become clear that stent thrombosis may occur at least up to a year postoperatively in the absence of clopidogrel therapy. Whether it is a danger beyond that point is not yet known. This requirement for clopidogrel therapy is a disadvantage if patients need further cardiac or noncardiac surgical procedures.

The second question concerning drug-coated stents is whether decreasing the risk of restenosis of treated areas decreases the risk of death, and if so, by how much? Coronary vascular events may occur in treated areas based on restenosis, but also may occur in nonstented areas. If DES is effective in reducing the rate of restenosis in the treated area, does that affect the life expectancy of high-risk subgroups treated with PCT a lot, a little, or none at all?

Bypass surgery also has evolved and improved. First-generation bypass surgery was usually carried out with reversed segments of saphenous vein grafts (SVGs) and performed with the aid of cardiopulmonary bypass. By the end of the first decade of bypass surgery, however, it became clear that the longevity of the effectiveness of the bypass operation was limited by intrinsic changes that develop in vein grafts—intimal fibroplasias and vein graft atherosclerosis (16). These pathologic changes could lead to the development of stenoses or occlusions of vein grafts that could eventually produce symptoms, MI, and death. Pharmacologic treatment with aspirin and other platelet inhibitors has improved early patency rates of vein grafts, but the most recent long-term data of vein graft patency for patients treated with aspirin show a patency rate of approximately 60% at 10 postoperative years (18, 19, 20). More recently, treatment with statin-type drugs in addition to platelet inhibitors has been shown to decrease the rate of vein graft failure and, in some small recent studies, the patency rates of vein grafts have been as high as 85% to 90% at 5 postoperative years (17,21,22). It is not known whether drugs such as clopidogrel will provide incremental benefits in enhancing patency over aspirin therapy alone. Although encouraging advances have been made in prolonging vein graft patency, late vein graft failure still remains an unsolved problem that compromises the long-term outcomes of bypass surgery.