Chapter 26. Coronary Artery Reoperations

Coronary artery reoperations are more complicated than primary operations. Patients undergoing reoperations have distinct, more dangerous pathologies; reoperations are technically more difficult to perform; and the risks are greater.1–12 Vein graft atherosclerosis, present in most reoperative candidates, is a unique and dangerous lesion. Reoperative candidates commonly have severe and diffuse native-vessel distal coronary artery disease (CAD), a problem that has had the time to develop only because these patients did not die from their original proximal coronary artery lesions. Aortic and noncardiac atherosclerosis are also often far advanced in many reoperative candidates. Some technical hazards, including the presence of patent arterial grafts and sternal reentry, are unique to reoperations, and others, such as lack of bypass conduits and difficult coronary artery exposure, are common.

After a primary bypass operation, the likelihood of a patient undergoing a reoperation depends on patient-related variables, primary operation-related variables, adherence to strict medical control of risk factors for disease progression after bypass surgery, the possibility of alternative treatments, physician opinion about the feasibility of reoperation, and time. Studies from our institution noted a cumulative incidence of reoperation of 3% by 5 years, 10% by 10 years, and 25% by 20 postoperative years13 (Fig. 26-1). Factors associated statistically with an increased likelihood of reoperation have been variables predicting a favorable long-term survival (eg, young age, normal left ventricular function [LVF], and single- or double-vessel disease), variables designating an imperfect primary operation (eg, no internal thoracic artery [ITA] graft and incomplete revascularization), and symptom status (eg, class III or IV symptoms at primary operation). Young age at primary operation and incomplete revascularization are also markers of a severe atherogenic diathesis.

More recently, however, the proportion of isolated coronary artery operations that are reoperations has decreased. This decrease is related in part to the more aggressive use of coronary artery interventions for patients with previous bypass surgery and probably to more effective risk factor control. In 1990 about 37% of coronary artery revascularization operations were reoperative interventions, whereas in 2002 this figure decreased to 30%14 (Fig. 26-2). Also, surgery has changed in directions that will decrease the rate of reoperation. Use of the left internal thoracic artery (LITA) to graft the left anterior descending (LAD) coronary artery decreases the risk of reoperation compared with the strategy of using only vein grafts, and the LITA-LAD graft has become a standard part of operations for coronary artery revascularization.15 Furthermore, it now appears that use of bilateral ITA grafts decreases the likelihood of death and reoperation when compared with the single LITA-LAD strategy16 (Fig. 26-3). The use of other arterial conduits such as the radial artery and the gastroepiploic artery in the context of total arterial revascularization may decrease the risk of reoperation further, but as yet the long-term data are insufficient to answer this question.

The patient population of reoperative candidates has evolved. Cleveland Clinic Foundation studies have shown that in the early years of bypass surgery (1967 to 1978), only 28% of patients underwent reoperation solely because of graft failure, and that graft failure often occurred early after the primary operation (mean postoperative interval of 28 months after primary operation). Reoperation because of the progression of atherosclerosis in nongrafted coronary arteries was common in the 1967 to 1978 time period (55% of patients).1,2 Between 1988 and 1991, almost all patients had graft failure as at least part of the indication for reoperation (92%), but that graft failure occurred late after the primary operation at a mean interval of 116 months.3 Today, patients undergoing reoperation usually had a successful primary operation at least 10 years previously for the treatment of multivessel CAD, and the angiographic indications for reoperation are progression of native-vessel distal CAD in combination with late graft failure caused by vein graft atherosclerosis.

An understanding of the pathology and causes of saphenous vein graft failure is important not only for an understanding of the causes of the need for reoperation, but also for understanding the dangers inherent in either the interventional or the conservative treatment of patients with previous bypass surgery. Saphenous vein to coronary artery grafts exhibit different pathologies at different intervals after operation.17–20 Within a few months, they often have diffuse endothelial disruptions with associated mural thrombus. The mural thrombus usually is not obstructing, and when grafts do become occluded early after operation owing to thrombosis, it may not be a result of these intimal changes, but rather may be related to hemodynamic factors. Most saphenous vein grafts examined more than 2 to 3 months after operation have developed a proliferative intimal fibroplasia. This is a concentric cellular process, and it is diffuse, extending the entire length of the graft (Fig. 26-4). It evolves with time to a more fibrous lesion. It is not friable, and although intimal fibroplasia involves most vein grafts, it causes stenoses or occlusions of only a few.

Vein graft atherosclerosis is a distinct pathologic process that often is recognized as early as 3 to 4 years after operation and is characterized by lipid infiltration of areas of intimal fibroplasia (Fig. 26-5). The distribution of vein graft atherosclerosis mimics that of intimal fibroplasia in that it is concentric and diffuse, although as vein graft atherosclerosis progresses, stenotic lesions may become eccentric. In addition, vein graft atherosclerosis is a superficial lesion, it is very friable, and it is often associated with overlying mural thrombus. These characteristics make it different from native-vessel coronary atherosclerosis, a process that is segmental and proximal, eccentric, encapsulated, usually not friable, and usually not associated with overlying mural thrombus. Vein graft atherosclerosis is seen in a majority of grafts explanted more than 10 years after surgery whether or not those grafts are stenotic, and atherosclerotic lesions appear to account for almost all late saphenous vein graft (SVG) stenoses. The extreme friability of vein graft atherosclerosis creates a substantial risk of distal coronary artery embolization during percutaneous interventions to treat stenotic lesions and during reoperations for patients with atherosclerotic vein grafts. It is also probable that spontaneous coronary artery embolization may occur from atherosclerotic grafts. In addition, atherosclerotic stenoses in vein grafts appear to predispose to graft thrombosis. Vein graft atherosclerosis appears to be an "active" event-producing lesion.

The exact incidence of late SVG stenoses and occlusions is difficult to determine even with prospective studies because death and reoperation are nonrandom events that remove patients from prospective populations available for late coronary artery angiography. However, it appears that by 10 years after operation, approximately 30% of vein grafts are totally occluded, and 30% of patent grafts exhibit some degree of stenosis or intimal irregularities characteristic of vein graft atherosclerosis.21,22 Although vein graft atherosclerosis is not the only factor related to late SVG occlusion, it is an important one. Native-vessel stenoses distal to the insertion site of vein grafts may decrease SVG graft outflow and contribute to graft failure, but late graft occlusion usually occurs in the presence of vein graft atherosclerosis. Furthermore, when stenotic vein grafts are replaced at reoperation, the late patency rate of the new vein grafts is good.2

Progress has been made toward decreasing the rate of vein graft failure. The early patency rates of SVGs have been improved by the use of perioperative and long-term platelet inhibitors,23–25 but the best data involving patients receiving platelet inhibitors indicate that the 10-year vein graft failure rate is approximately 35%. Some studies now indicate that lipid-lowering regimens decrease late vein graft disease and the risk of late cardiac events.26,27 However, the overall level of improvement has been small.26,27 So far, the only way known to avoid vein graft atherosclerosis is to avoid using vein grafts.

ITA grafts rarely develop late atherosclerosis, and the late attrition rate of patent ITA grafts is extremely low. Left ITA to LAD grafts have a very high late (20 years) patency rate, and for most patients, the LAD is a profoundly important coronary artery.21,28 These factors account for the impact of the LITA-LAD graft not only in decreasing the rate of late death after primary bypass surgery, but also in decreasing the rate of reoperation.15 Multiple ITA grafts provide incremental benefit in decreasing the risk of reoperation.16 It is also important that ITA grafts do not develop graft atherosclerosis, and therefore do not create the risk of coronary artery embolization during reoperation. The presence of patent arterial grafts may create other technical problems during repeat surgery, but embolization is not among them.

The randomized trials of bypass surgery versus medical management that were initiated in the 1970s provided a framework of information concerning the indications for bypass surgery, and subsequent observational studies have added substance to that framework. However, no randomized trials of medical versus surgical management pertain to patients with prior surgery. The coronary pathology of patients with previous bypass surgery is different from that of patients with only native-vessel stenoses, and we cannot assume that the natural history of, for example, triple-vessel disease based on atherosclerotic vein grafts, is equivalent to that of triple-native-vessel disease.

Two nonrandomized, retrospective studies of patients who had angiograms after bypass surgery addressed the issue of late survival.29,30 One study showed that patients with early (fewer than 5 years after operation) stenoses in vein grafts and patients with no stenotic vein grafts had approximately the same outcomes and that these outcomes were relatively good.29However, the presence of late (5 years or more after operation) stenoses in vein grafts predicted poor long-term outcomes, particularly if a stenotic vein graft supplied the LAD coronary artery. When late stenoses in LAD vein grafts were combined with other high-risk characteristics, the late survival rate was particularly dismal. For example, patients with a 50 to 99% stenosis in an LAD vein graft combined with abnormal LVF and triple-vessel or left main stenoses had only a 46% 2-year survival without reoperation. Patients with late stenoses in an LAD vein graft had significantly worse long-term outcomes than did patients with the LAD jeopardized by a native lesion (see Fig. 26-5). This study showed that the difference in the pathology of early (intimal fibroplasia) and late (vein graft atherosclerosis) vein graft stenoses is associated with a difference in clinical outcome and that late stenoses in vein grafts are dangerous lesions.

A second study compared the outcomes of patients with stenotic vein grafts treated with reoperation (REOP group) versus those treated with medical treatment (MED group).30 Again, this was a nonrandomized, retrospective study, and the patients in the REOP group were older and more symptomatic, had worse LVF, and had fewer patent grafts than the patients in the MED group.

The survival of patients with early (fewer than 5 years) SVG stenoses was not different in the two groups. The operative risk for the REOP group was low (no deaths among the 59 patients) and the long-term survival was good, but late survival was just as good for the patients treated medically (Fig. 26-6). It is important to note that the patients in the REOP group were more symptomatic to start with, and at late follow-up, they were less symptomatic than the patients in the MED group. Thus, reoperation for patients with early vein graft stenosis was an effective way of relieving symptoms of angina, but it appears that patients without symptoms can be treated medically with safety, at least over the short term.

However, the overall outcomes were worse for patients with late stenoses in vein grafts, and many subgroups had improved survival rates with reoperation. By multivariate testing (Table 26-1), a stenotic (20 to 99%) LAD vein graft predicted late death, and performing a reoperation increased late survival for these patients. Multivariate testing of smaller subgroups showed that the survival advantage for the REOP group was true even for patients with only class I or class II symptoms, and that reoperation still improved survival for the remaining patients when patients with stenoses in LAD vein grafts were excluded from the analysis.

Univariate comparisons for the REOP and MED subgroups of patients with stenotic LAD grafts are shown in Fig. 26-7, demonstrating the improved survival for the REOP group. When patients with stenotic LAD vein grafts were subgrouped on the basis of severity of the stenotic lesions (Fig. 26-8), the patients with severely stenotic (50 to 99%) vein grafts obviously benefited from surgery, exhibiting a decreased risk of death even early in the follow-up period. For patients with moderate stenoses (20 to 49%) in LAD vein grafts, the survivals of the MED and REOP groups were equivalent for about 2 years, but after that point survival of the patients in the MED group became rapidly worse, so that by 3 to 4 years of follow-up, the survival benefit of reoperation became apparent. Although the patients in these studies did not have consistent functional testing, there is evidence that myocardial perfusion and functional studies can help to identify patients likely to benefit from reoperation. Lauer and colleagues studied 873 symptom-free postoperative patients with symptom-limited exercise thallium-201 studies and found that patients with reversible perfusion defects were more likely to die or experience major cardiac events during a 3-year follow-up.31 Impaired exercise capacity also was strongly predictive of unfavorable outcomes.

Anatomical indications for reoperation to improve survival prognosis include: (1) atherosclerotic (late) stenoses in vein grafts that supply the LAD artery; (2) multiple stenotic vein grafts that supply large areas of myocardium; and (3) multivessel disease with a proximal LAD lesion and/or abnormal LVF based on either native-vessel lesions or stenotic vein grafts or a combination of the two pathologies. Reoperation is also effective in other anatomical situations in which severe symptoms are the indication for invasive treatment, including patients with a patent ITA to LAD graft combined with other ischemia-producing pathology and multiple early vein graft stenoses. The combination of the anatomical characteristics just noted and reversible ischemia and/or worsening LVF during stress constitutes a particularly strong indication for reoperation.

Percutaneous treatments (PCTs) represent alternative anatomical treatments for postoperative patients and often are useful. The effectiveness of PCTs is related to the vascular pathology to be treated and the clinical implications of treatment failure. Today, native coronary artery stenoses often can be treated with a low restenosis rate as long as those vessels are large enough to allow intracoronary stenting. Unfortunately, many postoperative patients have very diffuse native coronary atherosclerosis that makes PCT difficult or ineffective. Also, PCT has not been as effective in the treatment of diabetic native CAD.

The rate of technologic change in interventional cardiology has been rapid, and multiple percutaneous technologies have been used to treat stenotic vein grafts. Balloon angioplasty, first-generation PCT, was relatively dangerous to perform and produced ineffective long-term revascularization, particularly when used to treat older (atherosclerotic) vein grafts.32 Direct coronary atherectomy (DCA) increased the risk of coronary embolization at the time of the procedure without improving the restenosis rate.33 It has been hoped that the use of intracoronary stents, particularly covered stents and drug-eluting stents (DESs), in stenotic vein grafts might provide better outcomes, and stenting does represent an improvement over balloon angioplasty.32 The Randomized Evaluation of Polytetrafluoroethylene Covered Stent in Saphenous Vein Grafts (RECOVERS) trial, a randomized study designed to compare rates of SVG restenosis between CABG patients treated with covered stents and with bare stents, showed identical restenosis rates at 6 months of follow-up (24.2% versus 24.8%; p = .24).34 In a nonrandomized retrospective study comparing the effects of DESs with those of bare metal stents in treating SVG stenosis, Ge and colleagues reported significant differences in in-stent stenosis between groups at 6 months of follow-up (10 versus 26%; p = .03).35 However, other reports comparing DES with bare metal stents showed that the use of DES lowered the rate of restenosis but increased the risk of death.36

The kinetics of treatment failure after PCT for vein grafts are different from those for native coronary vessels. Restenosis and new stenotic lesions in vein grafts continue to appear with time, and the shoulder on the adverse outcome curve that appears at 6 months to 1 year after PCT for native vessels does not appear for vein grafts. Thus, there is still some uncertainty about the clinical impact of PCTs of stenotic vein grafts. Patients with previous bypass surgery are an extremely heterogeneous group; some subgroups are at low risk without any anatomical treatment at all, and some subgroups are at high risk without effective therapy. To date, the reported studies of PCT of SVG lesions have not included clinical risk stratifications that would allow comparison of patient survival rates.

Despite persistently high restenosis rates after percutaneous interventions, there are still many indications for their use in the treatment of patients with previous bypass surgery. Realistically, the ideal uses of PCTs are in situations in which failure of the anatomical treatment is not likely to be catastrophic as the impact of stenting on survival is unclear. These situations include symptomatic patients with: (1) early vein graft stenoses; (2) native coronary stenoses; or (3) focal late SVG stenoses in vein grafts not supplying the LAD artery. There are many patients with previous surgery who will fall into a middle ground where it is not clear whether percutaneous transluminal coronary angioplasty (PTCA) or reoperation is likely to yield the best outcome, and judgments must be made on the specific advantages and disadvantages of the treatments for those particular patients. Factors making PTCA more attractive than reoperation are listed in Table 26-2.

There are patients with postsurgical repeat ischemic syndromes and very unfavorable coronary anatomy for whom good options for anatomical treatment do not exist. For reoperative coronary surgery to be of benefit there must be bypass conduits available to construct new grafts to graftable coronary arteries that subtend substantial areas of ischemic but viable myocardium. If these conditions do not exist, surgery may not be in the interest of even a symptomatic patient. Studies of diabetic patients undergoing postsurgical repeat revascularization with PCT or surgery have shown unfavorable 10-year survival rates.37 Unless good coronary targets to receive bypass grafts are available, PCT may be the best choice for marginal candidates because of lower initial costs and, in some settings, lower initial mortality.

Reoperations are more complicated than primary operations. The specific technical challenges that surgeons must recognize and solve that are unique to or more common during coronary reoperation are

1. Sternal reentry

2. Stenotic or patent vein or arterial bypass grafts

3. Aortic atherosclerosis

4. Diffuse native-vessel coronary artery disease

5. Coronary arteries located amid old grafts and epicardial scarring

6. Lack of bypass conduits

The overall problem of myocardial protection is more difficult during reoperations, with perioperative myocardial infarction still being the most common cause of in-hospital death.3,6 The metabolic concepts of myocardial protection in use today are valid, but the reasons that myocardial protection sometimes fails during reoperation are related to anatomical causes of myocardial infarction. These anatomical causes of perioperative myocardial infarction include injury to bypass grafts, atherosclerotic embolization from vein grafts or the aorta to distal coronary arteries, myocardial devascularization secondary to graft removal, hypoperfusion through new grafts, failure to deliver cardioplegic solution, early vein graft thrombosis, incomplete revascularization, diffuse air embolization, and technical error.3,38–42 To be consistently successful, coronary reoperations must be designed to avoid these causes of myocardial infarction.

Preoperative Assessment

A complete understanding of the patient's native coronary and bypass graft anatomy is essential. Achieving this goal is sometimes not as easy as it sounds, particularly if the patient has had multiple previous coronary operations. If bypass grafts, venous or arterial, are not demonstrated by a preoperative coronary angiogram, it usually means that they are occluded, but it is also possible that the angiogram simply has failed to demonstrate their location. Examination of old angiograms performed before previous operations and review of previous operative records often help to illustrate the patient's coronary anatomy.

It is also important to know that graftable stenotic coronary arteries supply viable myocardium. Myocardial scar and viability can be differentiated by thallium scanning, positron-emission tomography, and stress (exercise or dobutamine) echocardiography. The intricacies of establishing myocardial viability are beyond this discussion, but it is an important issue. Before embarking on a reoperation, it makes sense to be reasonably sure that there is a matchup between the patient's graftable arteries and some viable myocardium such that grafting those arteries will provide some long-term benefits.

It is also wise to have a preoperative plan for bypass conduit selection and to document that potential bypass conduits are available. ITA angiography often is helpful. Venous Doppler studies can be used to assess the presence of greater and lesser saphenous vein segments, and arterial Doppler studies can assess the radial and inferior epigastric arteries and establish the adequacy of flow to the digits during radial artery occlusion.

Median Sternotomy Incision, Conduit Preparation, and Cannulation

Most coronary reoperations are performed through a median sternotomy. Situations associated with increased risk during a repeat median sternotomy include right ventricular or aortic enlargement, a patent vein graft to the right coronary artery, an in situ right ITA graft patent to a left coronary artery branch, an in situ left ITA graft that curls under the sternum, multiple previous operations, and difficulty reopening the sternum during a previous reoperation. In such situations, vessels for arterial (via the femoral or axillary artery) and venous access for cardiopulmonary bypass are dissected out before sternal reentry. All bypass grafts except for the internal thoracic arteries may be prepared before sternal reentry. Preparation of radial artery and greater and lesser saphenous vein segments can be carried out simultaneously. The most common structure injured during reentry is a bypass graft.4

When reopening a median sternotomy, the incision is made to the level of the sternal wires; the wires are cut anteriorly and bent back but are not removed (Fig. 26-9). An oscillating saw is used to divide the anterior table of the sternum. When the anterior table has been divided, ventilation is stopped, and the assistants elevate each side of the sternum with rake retractors while the posterior table of the sternum is divided in a caudal-cranial direction. The sternal wires that have been left in place posterior to the sternum help to protect underlying structures. Once the posterior table of the sternum has been divided with the saw, the wires are removed, and sharp dissection with scissors is used to separate each side of the sternum from underlying structures. Once the sternum has been divided, it is important that the assistants retract in an upward direction, not laterally. The right ventricle is injured more often by lateral retraction while it is still adherent to the underside of the sternum than it is by a direct saw injury.

In high-risk situations, it can be helpful to perform a small anterolateral right thoracotomy (Fig. 26-10) before the repeat median sternotomy. Underlying structures, such as the aorta, patent bypass grafts, and the right atrium and ventricle, can be dissected away from the sternum via this approach, and thus, with the surgeon's hand placed behind the sternum, reentry is safe. This small additional incision contributes little morbidity.

Another technique for sternal reentry in high-risk patients is to heparinize, cannulate, and initiate cardiopulmonary bypass before median sternotomy. The advantages of this strategy are that the heart can be emptied and allowed to fall away from the sternum, and cardiopulmonary bypass already has been initiated for protection if an injury does occur. The disadvantages of this approach are that extensive mediastinal dissection must be carried out in a heparinized patient, including dissection of the right internal thoracic artery if that is to be used. We rarely employ this approach except in situations in which adherence of an aortic aneurysm to the sternum or a patent right ITA-to-LAD graft creates a specific danger.

Once the sternum has been divided, the pleural cavities are entered. A general principle of dissection during reoperation is that starting at the level of the diaphragm and proceeding in a cranial direction is usually the safest approach. At the level of the diaphragm, few critical structures are injured if the wrong plane is entered. Therefore, at this point in the operation we usually dissect along the level of the diaphragm to the patient's right side until we enter the pleural cavity and then detach the pleural reflection from the chest wall in a cranial direction to the level of the innominate vein. The innominate vein is dissected away from both sides of the sternum with scissors, a maneuver that prevents a "stretch" injury to that vein.

Once the right side of the sternum is separated from the cardiac structures, it is usually possible to prepare a right ITA graft. Because of parietal pleural thickening, it is often more difficult to obtain length on ITA grafts during reoperation than it is during primary procedures, and the right ITA frequently is used as a "free" graft. Once the right ITA dissection is completed to the superior border of the first rib, an incision is made in the parietal pleura to separate the proximal ITA from the area of the phrenic nerve. Thus, if the right ITA needs to be converted to a "free" graft during aortic cross-clamping, it makes division at that point easier because the proximal ITA is clearly identifiable. Although intrapericardial dissection of the left side of the heart is left until later, freeing the left side of the anterior chest wall from the underlying structures (which may include a patent ITA graft) is undertaken now. This is difficult only if there is a patent ITA graft that is densely adherent to the chest wall. Again, it is best to enter the left pleural cavity at the level of the diaphragm and proceed in a cranial direction.

The most difficult point of dissection is usually at the level of the sternal angle, where a patent ITA graft may approach the midline and be adherent to the sternum or the aorta. There are no tricks for dissecting out a patent ITA graft except for being careful. The danger to a patent left ITA graft during sternal reentry and mediastinal dissection is entirely related to the location of the graft at the time of the primary operation. Ideally, the pericardium should be divided at a primary operation, and the left ITA graft should be allowed to run posterior to the lung through the incision in the pericardium and to the LAD or circumflex artery (Fig. 26-11). When this is done, the lung will lay anterior to the left ITA, and that graft will not become adherent to the aorta or to the chest wall.

Once the left side of the chest wall is free, the left IMA is prepared (if it has not been used at a previous operation), the sternal spreader is inserted, and the intrapericardial dissection of the aorta and right atrium is accomplished. Again, in most cases it is safest to find the correct dissection plane at the level of the diaphragm and then to continue around the right atrium to the aorta. The one situation in which this strategy may be dangerous is if an atherosclerotic vein graft to the right coronary artery lies over the right atrium. Manipulation of atherosclerotic vein grafts can cause embolization of atherosclerotic debris into coronary arteries, and it is best to employ a "no touch" technique with such grafts. If a vein graft to the right coronary artery lies in an awkward position over the right atrium, it is best to leave the right atrium alone and use the femoral vein and superior vena cava cannulation to establish venous drainage (Fig. 26-12). Once cardiopulmonary bypass has been established, the aorta has been cross-clamped, and cardioplegia has been given, the atherosclerotic vein graft then can be disconnected.

The goal of dissection of the ascending aorta is to obtain enough length for cannulation and cross-clamping and to avoid the most common error, aortic subadventitial dissection. The correct level of dissection on the aorta usually is found either by following the right atrium to the aorta in a caudal-to-cranial direction or by identifying the innominate vein and leaving all the tissue beneath the innominate vein on the aorta. At the level of the innominate vein, the pericardial reflection on each side of the aorta will be identifiable. Division of the pericardial reflection on the left side in a posterior direction will lead to the plane between the aorta and the pulmonary artery. Once the left side of the aorta is identified, the surgeon then may dissect posteriorly on the medial aspect of the left lung toward the hilum. The segment of tissue between these two dissection planes usually will include a patent left ITA graft, if present, and clamping that tissue will produce occlusion of the ITA graft.

When the aorta has been dissected out, heparin is given, and cannulation is undertaken. Cannulation of an atherosclerotic ascending aorta may cause atherosclerotic embolization leading to stroke, myocardial infarction, or multiorgan failure, so the ascending aorta should be studied with palpation and echocardiography to detect atherosclerosis before cannulation. Although the most widely used alternative arterial cannulation site is the femoral artery, arteriopathic patients often have severe femoral artery atherosclerosis. The axillary artery is an alternative arterial cannulation site that we have used with increasing frequency because atherosclerotic disease is usually not present in that vessel, and its cannulation allows antegrade perfusion44 (Fig. 26-13). If atherosclerotic disease or calcification of the aorta makes any aortic occlusion hazardous, the options are off-pump bypass surgery (see Other Options) or replacement of the aorta with axillary artery cannulation, hypothermia, and circulatory arrest. Venous cannulation usually is accomplished with a single two-stage right atrial cannula. A transatrial coronary sinus cardioplegia cannula is inserted via a right atrial purse string with the aid of a stylet, and a needle is placed in the ascending aorta for delivery of antegrade cardioplegia and for use as a vent (Fig. 26-14).

Myocardial Protection

The myocardial protection strategy used by us during most coronary artery reoperations is a combination of antegrade and retrograde delivery of intermittent cold blood cardioplegia combined with a dose of warm reperfusion cardioplegia ("hot shot") given before aortic unclamping, principles developed by Buckberg and colleagues.45 Multiple types of cardioplegic solutions have been described, and most appear to provide a metabolic environment that effectively protects the myocardium. Because of the potential anatomical challenges to cardioplegic myocardial protection during reoperations, the details of how the cardioplegic solution is delivered are very important. In most primary bypass operations, antegrade cardioplegia works well by itself. During reoperations, however, antegrade cardioplegia may not be effective for areas of myocardium that are supplied by patent in situ arterial grafts and may be dangerous because of the risk of embolization of atherosclerotic debris into the coronary arteries from old vein grafts. The delivery of cardioplegia through the coronary sinus and through the cardiac venous system to the myocardium (retrograde cardioplegia) has been a step forward in myocardial protection during reoperations.46,47 Retrograde cardioplegia delivery avoids atheroembolism from vein grafts, can be helpful in removing atherosclerotic debris and air from the coronary artery system, and can deliver cardioplegia to areas supplied by in situ arterial grafts. The biggest disadvantage of retrograde cardioplegia is that it is not always possible to place a catheter in the coronary sinuses that will deliver cardioplegia consistently. It is important to monitor the adequacy of cardioplegia delivery by measuring the pressure in the coronary sinus, noting the distention of cardiac veins with arterial blood, the cooling of the myocardium, and the return of desaturated blood from open coronary arteries.

Cardiopulmonary bypass is begun, the perfusionist empties the heart and produces mild systemic hypothermia (34°C), and the aorta is cross-clamped. We usually initiate cardioplegia induction with aortic root cardioplegia. To induce and maintain cardioplegic protection, it is helpful to be able to occlude patent arterial grafts. If it has not yet been possible to dissect out a patent arterial graft so that it can be clamped, the systemic perfusion temperature is decreased to 25°C until control of the graft is achieved. After antegrade cardioplegia has been given for 2 to 3 minutes, we shift to retrograde induction for another 2 to 3 minutes. Giving any antegrade cardioplegia does risk embolization from atherosclerotic vein grafts, but if these grafts have not yet been manipulated, that danger is relatively small. Once the adequacy of retrograde cardioplegia delivery has been established, it is often possible to use that route predominantly for maintenance doses.

Intrapericardial Dissection

When the heart has been arrested completely, intrapericardial dissection of the left ventricle is undertaken, starting at the diaphragm and extending out to the left of the apex of the heart. After the apex is identified, the surgeon divides the pericardium in a cranial direction on the left side of the LAD artery (Fig. 26-15). A patent LITA-to-LAD graft will be contained within the strip of pericardium that lies over the LAD artery. Dissection of this pedicle from the anterior aspect of the pulmonary artery will allow an atraumatic clamp to be placed across the patent ITA graft and also will allow the passage of new bypass grafts from the aorta underneath the patent ITA graft to left-sided coronary arteries. The advantages of waiting until after aortic clamping and arrest to dissect out the left ventricle are that dissection is more accurate, there is less damage to the epicardium and less bleeding, manipulation of atherosclerotic vein grafts is less likely to cause coronary embolization, and the dissection of patent ITA grafts is safer.

After the heart is dissected out completely, the coronary arteries to be grafted can be identified, the lengths that bypass conduits need to reach those vessels may be assessed, and the final operative plan can be established. The old grafts and epicardial scarring that are present during reoperations make the preoperative prediction of the lengths of conduits needed for bypass grafts quite difficult, particularly the lengths of arterial grafts, and it is wise to have some flexibility in the operative plan. Before the construction of the anastomoses, those patent but atherosclerotic vein grafts that are going to be disconnected are identified and are disconnected with a scalpel. The order of anastomosis construction that is used by the authors is: (1) distal vein graft anastomoses; (2) distal free arterial graft anastomoses; (3) distal in situ arterial graft anastomoses; and (4) proximal (aortic) anastomoses.

Stenotic Vein Grafts

When should patent or stenotic vein grafts be replaced, and with what should they be replaced? Atherosclerosis in vein grafts is common if those grafts are more than 5 years old, and leaving them in place risks embolization of atherosclerotic debris at the time of reoperation and subsequent development of premature graft stenoses or occlusions after reoperation. On the other hand, replacement of all vein grafts extends the operation and may use up available bypass conduits.

In the past, our general rule has been to replace all vein grafts that are more than 5 years old at the time of reoperation, even if those grafts are not diseased angiographically. However, this strategy assumes that conduits are available that can replace these old grafts. Today, many patients have very limited conduits at reoperation because of the large numbers of vein grafts used at primary surgery or because of multiple previous operations. Thus, graft replacement must be individualized. Inspection of vein grafts at reoperation occasionally will identify a graft that looks normal angiographically and does not appear to have any thickening or atherosclerosis on visual inspection. Often such vein grafts will be left alone.

Replacing old vein grafts with new vein grafts may often be accomplished by creating the new vein-to-coronary-artery anastomosis at the site of the previous distal anastomosis, leaving only 1 mm or so of the old vein in place (Fig. 26-16). If significant native-vessel stenoses have developed distal to the old vein graft, it is often best to place a new graft to the distal vessel in addition to replacing the vein graft. Many reoperative candidates have proximal occlusions of the native coronary artery system and multiple stenoses throughout the vessel, and if only new distal grafts are constructed, the proximal segments of coronary arteries and their branches that are supplied by atherosclerotic vein grafts may be jeopardized. More than one graft to a major coronary artery may be desirable during reoperation (Fig. 26-17).

Sequential vein grafts often are very helpful during reoperation because they allow more distal anastomoses and fewer proximal anastomoses. Sites for proximal anastomoses are often at a premium in the scarred reoperative aorta.

Artery-to-coronary-artery bypass grafts have many advantages during reoperations. First, they are often available. Second, the tendency of arteries to remain patent even when used as grafts to diffusely diseased coronary arteries makes them particularly applicable to reoperative candidates. Third, in situ arterial grafts do not require a proximal anastomosis. If the left ITA has not been used as a graft at a previous operation, a strong attempt should be made to use it as an in situ graft to the LAD artery. During primary operations, the right ITA usually can be crossed over as an in situ graft to left-sided vessels, but such a plan is more difficult during repeat surgery, so the right ITA is often used as a free graft.

Arterial graft proximal anastomoses are a problem at reoperation because the scarring and thickening of the reoperative aorta often make direct anastomoses of arterial grafts to the aorta unsatisfactory. However, when old vein grafts become occluded, there is usually a "bubble" of the hood of the old vein graft that is not atherosclerotic and that often is a good spot for construction of a free (aorta-to-coronary-artery) arterial graft anastomosis (Fig. 26-18). In addition, if new vein grafts are performed, the hood of that new vein graft represents a favorable location for an arterial graft anastomosis. Late angiographic data regarding this strategy are not available, but the relative freedom of the hood of vein grafts from the development of atherosclerosis means these grafts are likely to be successful.

Another effective strategy is to use either an old arterial graft or a newly constructed arterial graft for the proximal anastomosis of a free arterial graft (Fig. 26-19). Composite arterial grafts, usually using a new in situ left ITA graft at the proximal anastomotic site for a free right ITA graft, have been employed with increasing frequency, and early outcomes have been favorable.48,49 This method is particularly useful during reoperations because it may avoid an aortic anastomosis, and less right ITA graft length is needed to reach distal circumflex arteries. Other advantages of using a previously performed patent ITA graft for the proximal anastomosis of a new arterial graft are that the old left ITA graft often has increased in size, and the preoperative angiogram has demonstrated its integrity. In situations in which the effectiveness of an LITA-to-LAD graft has been jeopardized by a distal LAD lesion, a short segment of a new arterial graft can be used to bridge that stenosis from the old arterial graft to the distal LAD artery (see Fig. 26-18).

Can an ITA graft be used to replace a vein graft during reoperation? When faced with replacing a stenotic or patent vein graft during reoperation, the surgeon has a number of options, all of which have some potential disadvantages:

1. The surgeon may leave the old vein graft in place and add an arterial graft to the same coronary vessel. The dangers of this approach are that atherosclerotic embolization from the old vein may occur during the reoperation, and competitive flow between the vein graft and the arterial graft may jeopardize the ITA graft after reoperation.

2. The surgeon may remove the old vein graft and replace it with an ITA graft. This decreases the likelihood of atherosclerotic embolization and competitive flow but risks hypoperfusion during reoperation if the arterial graft cannot supply all the flow that had been generated previously by the vein graft.

3. The surgeon may replace the old vein graft with a new vein graft. The disadvantage of this approach is a long-term one: The coronary vessel is left dependent on a vein graft.

When we examined these choices in a retrospective study of operations for patients with atherosclerotic vein grafts supplying the LAD artery, we found that the worst outcomes resulted from removing a patent (although stenotic) vein graft and replacing it with only an ITA graft.39 This strategy was associated with a significant incidence of hypoperfusion and severe hemodynamic difficulties during reoperation that were treated effectively only by adding a vein graft to the same coronary artery. The incidence of myocardial infarction associated with leaving a stenotic vein graft in place was low. Thus, atherosclerotic embolization from an atherosclerotic vein graft is a danger, but it appears that with the use of retrograde cardioplegia, it is not commonly a major catastrophe.

Another potential disadvantage of the strategy of adding an ITA graft to a stenotic vein graft is that competition in flow from the stenotic vein graft may lead to failure of the new ITA graft. However, this is unlikely to occur as long as the stenosis in the SVG is severe.50 Our usual approach, therefore, is to remove atherosclerotic vein grafts when replacing them with a new vein graft but leave stenotic vein grafts in place when grafting the same vessel with an arterial graft (Fig. 26-20).

Alternative arterial grafts often are very useful during reoperation. The radial artery has particular advantages during repeat surgery because it is larger and longer than other free arterial grafts. These qualities increase the range of coronary arteries that can be grafted. Early studies of radial artery grafts have shown favorable patency rates, but few long-term data currently exist. If the high patency rates that have been documented by early studies are confirmed by the tests of time, the radial artery will be used extensively during reoperations. The inferior epigastric artery often is too short to function as a separate aorta-to-coronary-artery graft during reoperation but can be extremely useful as a short composite arterial graft, as illustrated in Fig. 26-19.

The right gastroepiploic artery (RGEA) has established a good midterm graft patency rate record and often is useful during reoperation because it is an in situ graft.51 Furthermore, it can be prepared before the median sternotomy. It is effective most often as an in situ graft to the posterior descending branch of the right coronary artery or the distal LAD artery (Fig. 26-21).

The aortic anastomoses of the vein and arterial grafts are performed last during the single period of aortic cross-clamping. Sites for aortic anastomoses are often at a premium owing to previous scarring, atherosclerotic disease, or the use of Teflon felt during the primary operation, and often the locations of the previous vein graft proximal anastomoses are the best locations for the new ones. The advantages of constructing aortic anastomoses during a single period of aortic cross-clamping are that it minimizes aortic trauma and allows excellent visualization of the proximal anastomoses. In addition, if patent or stenotic vein grafts have been removed and replaced, reperfusion is not accomplished by aortic declamping until the aortic anastomoses have been completed.

The disadvantage of this approach is that it prolongs the period of aortic cross-clamping. However, our strategies for reoperation are not based on trying to minimize myocardial ischemic time. If cardioplegia can be delivered effectively, its metabolic concepts are valid, and myocardial protection is secure. Failure of myocardial protection usually is caused by anatomical events, not by metabolic failure. Once the proximal anastomosis has been constructed, a "hot shot" of substrate-enhanced blood cardioplegia is given, and the aortic cross-clamp is removed.

Other Options

Although most reoperations are performed through a median sternotomy with the use of cardiopulmonary bypass, the strategies of small-incision surgery and off-pump surgery that have been gaining increasing use for primary coronary artery operations also can be helpful during reoperations. Reoperations in situations in which a limited area of myocardium needs revascularization often can be accomplished through a limited incision and without the use of cardiopulmonary bypass (known as the minimally invasive direct coronary artery bypass [MIDCAB] operation). The distal LAD artery may be exposed with a small anterior thoracotomy, and the LAD or diagonal artery may be grafted with a left ITA graft. A stabilizing device usually is employed for anastomotic construction, although the intrapericardial adhesions provide some stability during reoperations. If the left ITA is not available, a segment of saphenous vein can be anastomosed to the subclavian artery and routed in a transthoracic path to the LAD artery. If the right ITA is to be used as an in situ graft to the LAD artery, a median sternotomy is indicated, but if this is the only graft, off-pump surgery usually is possible.

The lateral wall of the heart can be exposed through a left lateral thoracotomy (Fig. 26-22), and the circumflex and distal right coronary artery branches can be grafted with this approach. Often the LITA already has been used for a graft, but the descending thoracic aorta may be used as a site for the proximal anastomosis of a vein graft or a radial artery graft using a partial occluding clamp. The disadvantages of this approach are that the right ITA is difficult to use as an in situ graft, and if the circumflex vessels are deeply intramyocardial, they may be difficult to expose and isolate with the off-pump strategy.

In addition to avoiding potential complications of cardiopulmonary bypass, the "limited-area, off-pump" approach also avoids extensive dissection of the heart and possible manipulation of atherosclerotic vein grafts. The disadvantage of this approach is that most patients who are candidates for reoperation need grafts to multiple vessels in multiple myocardial areas.

Use of a median sternotomy and the off-pump strategy to graft multiple myocardial areas is now a standard approach to primary coronary revascularization and also can be used during reoperation. However, because of the need to access all areas, extensive dissection sometimes is necessary for lysis of adhesions to be able to mobilize the heart. If patients have atherosclerotic vein grafts, dissection and manipulation create the dangers of embolization of atherosclerotic debris and myocardial infarction. This problem was encountered during the early years of bypass surgery when the risks of atherosclerotic embolization were less recognized. Another disadvantage of off-pump reoperative strategies is that reoperative candidates often have very distal and diffuse CAD, which leaves intramyocardial segments as the best areas for grafting. These characteristics stress off-pump isolation and immobilization techniques. In addition, the aortic anastomoses of vein or free arterial grafts may be difficult because of aortic atherosclerosis, adhesions, or previous aortic anastomoses that may limit the application of a partial occluding clamp. On the other hand, the use of off-pump techniques may minimize aortic trauma, particularly if in situ arterial grafts can be employed to provide inflow to new grafts.

In an individual case, the disadvantages of off-pump surgery may be important or irrelevant. Surgeons who perform reoperative coronary artery surgery in a wide spectrum of situations will find both on- and off-pump strategies helpful.

Coronary artery reoperations are riskier than primary operations. A study from the Society of Thoracic Surgeons (STS) database reported an in-hospital mortality rate of 6.95% associated with reoperations for the years 1991 to 1993, and in a multivariate analysis of all isolated coronary artery bypass surgery, "previous operation" was identified as a factor that increased the mortality rate.12 At the Cleveland Clinic Foundation, the in-hospital mortality rate of a first reoperation ranged between 3 and 4% from 1967 through 1991, and the rate was 3.7% for 1663 patients having repeat surgery from 1988 through 1991.1–3 Progress during the last decade has continued to lower this risk. In a recent report by Sabik and colleagues, the hospital mortality rate for patients undergoing reoperative CABG was reduced to 2.5% in 2002, and risk adjustment identified the comorbidity burden carried by reoperative patients as a factor that increased risk, not reoperative status itself.14

Recent mortality rates from other large series range from 4.2 to 11.4%, most being around 7%.4–9,52 All these figures are two to five times higher than the rates we would expect for the risk of primary CABG.

Coronary artery reoperations have been associated with a higher in-hospital mortality mostly because of an increased risk of perioperative myocardial infarction. In the Cleveland Clinic Foundation series, the cause of perioperative death was cardiovascular in 85% of cases in the most recent cohort of patients undergoing reoperation, a figure that contrasts with recent studies of primary operations, in which noncardiac causes of death have been increasingly important.3,15 Furthermore, in the reoperative series, in-hospital mortality was associated with new perioperative myocardial infarction in 67% of cases. Multiple causes of myocardial infarction have been identified, including incomplete revascularization owing to distal CAD, vein graft thrombosis, ITA graft failure, atherosclerotic embolization from vein grafts, injury to bypass grafts, hypoperfusion from arterial grafts, preoperative myocardial infarction, and complications of PTCA.

Sternal reentry still creates risk. In a study of 1847 patients undergoing reoperation, 7% were associated with an adverse event and only previous radiation and the number of previous operations could be identified as predictors of injury. Of the 127 patients sustaining an injury, 24 (19%) experienced a major adverse outcome (stroke, myocardial infarction, or death) compared with a risk of 6.2% for those without an injury.43

Multiple studies of patients undergoing reoperation have identified increased age, female gender, and emergency operation as clinical variables that have a high association with in-hospital mortality. Emergency operation is a particularly strong factor. Although there is not a standard definition of emergency, mortality rates after emergency reoperations that have been reported range from 13 to 40%.3,5–8 Data from the STS for the year 1997 documented a risk of 5.2% for elective reoperations, 7.4% for urgent reoperations, 13.5% for emergency reoperations, and 40.7% for "salvage" reoperations. There is clearly a major increment in risk associated with emergency reoperations, a larger increment than has existed for patients undergoing primary surgery.

Advanced age, by itself, does not increase the risk of reoperation substantially but does so when combined with other variables. In a review of 739 patients aged 70 years or older undergoing reoperation, we noted an overall in-hospital mortality rate of 7.6% and identified emergency operation, female gender, left ventricular (LV) dysfunction, creatinine concentration greater than 1.6 μg/dL, and left main coronary artery stenosis as specific factors increasing risk. For patients with none of these characteristics, the in-hospital mortality rate was only 1.5%.53

Specific anatomical situations, in particular, the presence of patent ITA grafts and atherosclerotic vein grafts, can increase the risk of reoperation, but with experience, these technical factors largely have been neutralized. We have never documented an increased mortality rate for patients with patent ITA grafts but have noted that the risk of ITA damage has dropped from 8% in our early experience to 3.7% more recently, an improvement almost entirely related to increased surgical experience. With proper positioning of an ITA graft at primary operation, a patent LITA-to-LAD-artery or LITA-to-circumflex-artery graft should not represent an impediment to reoperation. Situations in which a patent in situ right ITA graft crosses the midline to supply the LAD or circumflex system are more difficult and require extreme care in reoperating using a median sternotomy incision. Although these situations are uncommon and provide difficult technical challenges, the risks for these patients have not been increased.

Studies from the past noted that the presence of atherosclerotic vein grafts did increase perioperative risk. Perrault and colleagues documented mortality rates of 7, 17, and 29% for patients with one, two, or three stenotic vein grafts, respectively, and in a previous study of patients with atherosclerotic vein grafts, we noted that the presence of an atherosclerotic vein graft to the LAD artery increased in-hospital risk.30,36 However, in our more recent study we found that atherosclerotic vein grafts did not increase mortality, although there was a nonsignificant trend toward increased risk for patients with multiple stenotic grafts.3 The favorable results for these patients have been based on a combination of improved technology, the use of retrograde cardioplegia delivery, and increased surgeon experience.

Although arterial grafts may offer advantages at reoperation, their use may prolong an already complex operation, and the influence of arterial grafting on perioperative risk has been a concern. However, we have specifically studied this issue and found that the use of single or double ITA grafts at reoperation does not increase perioperative risk, and in fact, not having an ITA graft at either the first or second operation appeared to be a factor associated with increased in-hospital mortality.3 Graft selection in that study was not randomized, and it is certainly possible that the increased risk for patients receiving only vein grafts was related to patient-related variables rather than surgical strategy. It does appear, however, that the use of arterial grafts does not increase risk. Except for an increased incidence of perioperative myocardial infarction, in-hospital morbidity does not seem to be increased for patients undergoing reoperation. One important observation relates to wound complications. Multiple groups, including ours, have noted an increased risk of wound complications when diabetic patients have received bilateral (simultaneous) ITA grafts. However, there does not appear to be an increased risk of wound complications for diabetic patients who receive staged ITA grafts, one at the first and another at a second operation.

It is important to note that only the variables that can be identified and quantified are included in studies consistent enough to be identified as risk factors. For example, experience and logic dictate that severe atherosclerosis of the ascending aorta is a major risk factor, but this is rarely identified in large studies because patients do not routinely undergo echocardiography to identify the presence of aortic atherosclerosis.

Late Results

Patients who are undergoing reoperation are at a later stage in the progression of their native coronary atherosclerosis compared with the point when they underwent primary surgery, and the anatomical corrections achieved at reoperation are less perfect. Although the definition of complete revascularization varies widely, few reoperative candidates undergo an operation in which all diseased segments of all arteries receive bypass grafts. It is not surprising that the long-term results of reoperation have not been as favorable as the long-term results of primary operations.

The likelihood of recurrent angina after any bypass operation is related to time, but angina symptoms are more common after repeat surgery than they are after primary operation. Follow-up of our reoperative patients at a mean interval of 72 months after reoperation showed that 64% of patients were in New York Heart Association (NYHA) functional class I, although only 10% of patients had class III or class IV symptoms.2 Weintraub and colleagues also noted at a 4-year follow-up that 41% of reoperative patients had experienced some angina.6

Late survival rates after reoperation are also inferior to those after primary surgery. Weintraub and colleagues noted 76% 5-year and 55% 10-year survival rates, and our most recent follow-up study found a 10-year survival rate of 69% for in-hospital survivors2,6 (Fig. 26-23). The predictors of late survival have varied among studies, but LV dysfunction, advanced age, and diabetes consistently have been associated with a decreased late survival rate. The variables identified by multivariate testing as decreasing the late survival for 2429 hospital survivors of a first reoperation are listed in Table 26-3. The influence of ITA grafts on late survival has been difficult to determine for reoperations. We found a positive influence of a single ITA graft on late survival, as have others,54 but the effect was not as dramatic as has been noted after primary operations. Weintraub and colleagues did not document an improved survival associated with ITA grafting.6

Multiple Coronary Artery Reoperations

Patients who have had more than one previous coronary artery operation are like patients undergoing first reoperations, only more so. Many patients undergoing multiple reoperations had their first procedure more than 15 years ago, and severe native-vessel disease and lack of bypass conduits are a common combination of problems. Selection criteria vary widely among institutions, but in-hospital mortality rates are increased relative to first reoperations.10,11 Through 1993, we reoperated on 392 patients who had more than one previous bypass operation, with an in-hospital mortality rate of 8%. Over the next 10 years, this mortality rate has decreased to 5.8%.14 Follow-up of the in-hospital survivors in the former group found late survival rates of 84% at 5 and 66% at 10 postoperative years. Thus, although the in-hospital risks were increased for these patients, the long-term outcome has been relatively favorable. Age was a major determinant of outcome. Recently, in-hospital mortality for patients younger than 70 years of age has decreased to 1 to 2%, but for patients greater than age 70, it has remained higher than 10%. Furthermore, patients greater than age 70 who did survive operation in our series had only a 50% 5-year late survival.

Coronary artery reoperations continue to present adult cardiac surgeons with their most difficult challenges, in part because of the many technical pitfalls that exist, but also because significant comorbidity is common in reoperation candidates. The overall number of reoperations appears to be declining for multiple reasons, one important one being the increased use of arterial grafts at primary operations. However, when recurrent ischemia syndromes do occur reoperation will often be the best choice of therapy.

1. Make certain the end justifies the means before performing a coronary reoperation.

2. Avoid cardiac injury.

3. Do not replace a patent or stenotic vein graft with an ITA graft.

4. Dissect out the left side of the heart with the heart arrested.

5. Use the axillary artery as an alternative cannulation site.