Preparation

After the induction of anesthesia, patients are positioned, prepped, and draped in an identical fashion to an on-pump procedure. Although OPCAB allows for minimally invasive approaches, including small thoracotomy, endoscopic, and robotic-assisted coronary artery bypass, the most common approach is via median sternotomy. A median sternotomy is the most frequently used route for cardiac access when performing multivessel grafting. A median sternotomy may be routinely accomplished via a limited skin incision (10-12 cm), and allows the surgeon to visualize the operative field from an orientation that is familiar and similar to on-pump procedures. This facilitates target-vessel identification as well as harvesting of the internal mammary arteries for use as a conduit. Additionally, should conversion to conventional bypass become necessary, a median sternotomy allows easy access for cannulation for CPB. Isolated grafting of specific individual vessels can be performed using a variation of thoracotomy— anterior for the left anterior descending artery or lateral for access to the marginal vessels.

Conduit Harvesting

During left internal mammary artery harvest, we routinely skeletonize the vessel using the harmonic scalpel (HARMONIC SYNERGY Blade, Ethicon) in order to optimize the length of the vessel while minimizing trauma to the chest wall. Unlike ONCAB, in OPCAB the heart is not decompressed, and the extra artery length is often necessary to avoid tension on the anastomosis during rightward displacement for lateral or inferolateral wall grafting. Dividing or removing the endothoracic fascia, skeletonizing the internal mammary artery during harvest, and dividing the left pericardium vertically toward the left phrenic nerve at the level of the pulmonary artery all provide for extra length and less tension on the LIMA-LAD anastomosis. After dividing the mammary artery, we inject with a soft-tip needle each mammary artery with 10 mL of a solution comprised of 19 mL of the patient’s blood, 10 mg of milrinone (in a concentration of 1 mg/mL) and 1000 U of heparin (1000U/mL solution). The instillation of approximately 5 mL of this solution into the lumen of the mammary artery completely resolves any spasm, thereby creating an ideal conduit for the bypass.

Radial artery and saphenous vein conduits are harvested endoscopically and simultaneously during internal mammary artery harvest. It is our practice to administer 2500 U of heparin before endoscopic vein harvest to minimize thrombus formation within the conduit during the harvest. Concern over graft quality with endoscopic vein harvest has prompted increased vigilance in atraumatic harvest technique to ensure optimal conduits for bypass. We use the same solution described above to store the radial arteries conduits from harvest until the time of grafting.

Pericardiotomy

After single or bilateral internal mammary artery harvest, the heparin dose is administered, and the arterial conduits are divided distally. Once all conduits are obtained and checked, the retractor is positioned inferiorly in the sternal incision. This placement reduces traction on the brachial plexus and generally facilitates mobilization of the heart for positioning. Current retractors used for beating-heart surgery come with attachable devices to aid in positioning the heart as well as stabilizing the target artery.

The pericardium is then incised in an inverted T configuration, and then incised laterally along the diaphragm to facilitate cardiac displacement. It is essential to free the left lateral pericardium from the diaphragm to allow the pericardium to be retracted to displace the heart and effectively expose the lateral wall of the left ventricle. Nonetheless, the phrenic nerves must be identified and preserved during pericardial mobilization. Several pericardial traction sutures are placed to assist with exposure and lateral displacement of the heart; these stitches are positioned away from the opening margin of the pericardium fairly deep in the lateral wall of the pericardium in order to maximize the exposure and take advantage of the rolling ability of the heart within the semicircular circumference of the pericardium. To avoid compression on the right heart during lateral displacement, the right pericardium can be dissected along the diaphragm, or the right pleural space can be opened widely to allow the heart to fall into the right chest during lateral displacement; this trick is particularly useful in the setting of cardiomegaly. Additionally, 1 or 2 rolled towels placed along the inferior aspect of the right side of the retractor helps to elevate the right side of the sternum to allow the heart to be displaced toward or into the right chest.

An important traction suture is the “deep stitch,” which is placed approximately two-thirds of the way between the inferior vena cava and left pulmonary vein at the point where the pericardium reflects over the posterior left atrium (Figure 1). Care should be taken with placement of this suture to avoid the underlying descending thoracic aorta, esophagus, left lung, and adjacent inferior pulmonary vein. While passing the deep stitch, the right-sided pericardial traction sutures should be relaxed to prevent compression of caval inflow. The deep stitch should be covered with a soft rubber catheter to prevent laceration of the epicardium during retraction. Furthermore, the manual elevation and compression of the heart required to take this stitch may be poorly tolerated in patients with marginal hemodynamics or significant left main coronary artery disease. In that case, grafting and reperfusion of the left anterior descending coronary artery should be accomplished before placing the deep pericardial traction suture.

Figure 1

Placement of the “deep stitch.” Surgeon’s view. The heart is elevated, and the suture is placed approximately two-thirds of the way between the inferior vena cava and left pulmonary vein at the point where the pericardium reflects over the posterior left atrium. The deep stitch should be covered with a soft rubber catheter to prevent laceration of the epicardium during retraction. The purpose of the deep stitch is to elevate the heart up and out of the pericardial well.

Aortic Evaluation

Before starting the distal anastomosis, a final plan of grafting should be made: all the targets should be visualized, and the ideal conduit for each target should be chosen in advance. It is extremely important at this stage to confirm the possibility of safely connecting the proximal anastomosis to the ascending aorta, or opting for a no-touch aortic technique, which requires Y- and T-anastomosis configuration, or both, between the internal mammary arteries and free conduits.

Epiaortic ultrasonography is used in all of our patients undergoing cardiac surgery to evaluate and grade the ascending aorta. It adds only 1 to 2 minutes to the procedure, and provides both the surgeon and the anesthesiologist a simple noninvasive and inexpensive tool for assessing the extent of atheromatous disease in the ascending aorta in preparation for aortic clamping or selection of an alternative clampless technique. Through the operative incision, the ascending aorta, from the aortic root to the origin of the innominate artery, is scanned directly by the surgeon using an ultrasound probe connected to an echocardiography ultrasound scanner. The 8.5 MHz linear array probe is placed inside a sterile sleeve filled with sterile saline to act as a medium between the probe and the surface of the aorta. The information obtained often dictates changes in operative strategy, depending on the grade of atherosclerosis. Similarly, it allows the surgeon to individualize placement of aortic clamps and proximal anastomotic devices to minimize the risk of atheroembolism. Studies using epiaortic ultrasound for aortic screening have shown improvements in clinical outcomes as a result of intraoperative modifications of the surgical plan.[22],[23],[24],[25]

Exposure

Optimal target-vessel exposure and 3-dimensional stabilization are essential for successful off-pump bypass surgery. Cardiac positioners and stabilizers have greatly increased the ability to manipulate the heart with minimal hemodynamic compromise. Two different systems are routinely used in our institution—the Medtronic Octopus Tissue Stabilizer and Starfish or Urchin Heart Positioner (Medtronic, Inc, Minneapolis, MN) and the Maquet ACROBAT stabilizer and XPOSE positioner (Maquet, GMBH & Co, Radstadt, Germany). “Apical” suction devices allow distraction and manipulation of the heart by creating a vacuum-type seal to the epicardial surface. They are generally placed on the apex to expose the anterior wall (LAD territory) and inferior wall (posterior descending territory) of the heart and may be placed on the acute margin to expose the right coronary artery (Figure 2). They are frequently placed off of the apex, especially to the left of the apex, to expose the lateral wall and branches of the left circumflex coronary artery (Figure 3). Because these suction-based cardiac positioning devices pull the heart in the appropriate direction rather than pushing it, the heart is not compressed, functional geometry is maintained, and cardiac positioning is usually well tolerated. The coronary stabilizer devices consist of pods of suction cups within the prongs of the stabilizer that immobilize the target area by creating a vacuum between the epicardial surface and the stabilizer arm. This allows for construction of the anastomoses to take place in a relatively motionless field, approximately recreating the same visual image seen in an arrested heart. The anterior wall vessels often require only the coronary stabilizer for adequate exposure. The stabilizer is positioned along the caudal aspect of the retractor toward the left, with the retractor arm placed out of the way to prevent interference during the anastomosis. The location of these devices on the sternal retractor also requires consideration. For the lateral and inferior wall vessels, the cardiac positioner is usually placed on the surgeon’s side at the most cephalad location of the retractor. The coronary stabilizers can then be placed on either side. A general rule is to place the stabilizer in the assistant’s way instead of the surgeon’s in order to prevent these devices from obstructing the surgeon’s view or interfering with hand positioning during suture placement.

Figure 2

Exposure of the inferior wall with aid of an apical suction device. View from the surgeon’s side of table. The cardiac suction device is placed on the apex allowing the heart to be easily displaced to expose the inferior wall vessels. The apical suction device may be attached to the retractor on the surgeon’s side, as shown, or on the assistant’s side.

Figure 3

Exposure of the lateral wall with aid of an apical suction device. View from head of table. The cardiac positioning device is placed slightly off of the apex. A coronary stabilizer may be used additionally to expose lateral wall vessels, such asthe obtuse marginal vessels.

In addition to the positioners and stabilizers, manipulating the traction sutures can greatly enhance exposure. The purpose of the “deep stitch” is to elevate the heart up and out of the pericardial wall. When this suture is retracted toward the patient’s feet, it elevates the heart toward the ceiling and points the apex vertically with remarkably little change in hemodynamics. When retracted toward the patient’s left side, the heart rotates from left to right, exposing the lateral wall vessels. Variable tension on this stitch will enhance exposure to both the anterior and lateral wall. During positioning, the left-sided pericardial sutures should be pulled taut, and the right-sided sutures completely relaxed to avoid compression on the right heart during cardiac displacement. Pericardial sutures on both the right and left sides are never under tension simultaneously when displacing the heart to expose coronary targets, as this will generally lead to diminished venous return to the heart and subsequent hypotension.

Manipulation of the operating table can also facilitate exposure. Placing the patient in steep Trendelenburg position exposes the inferior wall. Turning the table sharply toward the right will aid with exposure of the lateral wall targets. Usually, little manipulation is required for grafting the anterior wall vessels. Occasionally, a warm, moist laparotomy pad can be placed adjacent to the “deep stitch” to assist with elevating and rotating the heart out of the pericardium.

In preparation for each distal anastomosis, a soft silastic retractor tape mounted on a blunt needle (Retractotape, Quest Medical, Inc, Allen, TX) is placed widely around the proximal vessel for transient occlusion. For inferior wall vessels, this suture can be displaced posteriorly and caudally by tying a more posterior pericardial suture loosely around the retractor tape (Figure 4). The pericardial retraction suture serves as a “pulley” that not only enhances coronary exposure and the surgeon’s view, but also keeps this retraction stitch from interfering with the sutures during the anastomosis (Figure 4). Similarly, this maneuver can be done for some lateral wall targets.

Figure 4

Use of retractor tape to create a “pulley” for optimizing inferior wall target vessel exposure. Close-up surgeon’s view. A soft silastic retractor tape is mounted on a blunt needle and placed widely around the target vessel proximally for transient occlusion. For inferior wall vessels, this suture can be displaced posteriorly and caudally by tying a more posterior pericardial suture loosely around the retractor tape. A medium clip secures the suture and retractor tape in place, which can then be retracted to transiently occlude the artery.

Although a well-trained first assistant is necessary for providing an effortless anastomosis, the second assistant, often the scrub nurse, also plays a major role in exposure. The field is kept free of blood with a humidified CO₂ blower (DLP, Medtronic) and Cell Saver (Haemonetics, Braintree, MA), which are managed by the scrub nurse or second assistant. The blower is used to keep the field free of blood, but is also used to open the target vessel and graft during suture placement, and can play a vital role in visualization during the anastomosis. Occasionally an epicardial fat retractor can be used to expose the coronary target in patients with a large amount of epicardial fat. The second assistant usually stands to the right of the surgeon, though better exposure by this assistant standing at the head of the bed, to the surgeon’s left, may be achieved during anastomosis of the inferior wall or lateral wall targets.

In chronically occluded vessels that have collateral and/or retrograde flow, bleeding into the field can be controlled with another retractor tape distally, a MyOcclude device (United States Surgical Corp, Norwalk, CT), or an intracoronary shunt.[26]A final preparatory measure is to place temporary atrial or ventricular pacing cables before positioning the heart if it seems likely that intraoperative pacing will be useful. As the heart is rotated toward the right, visualization of the right atrium is more difficult, making placement of temporary clip electrodes on the right atrium challenging. It may be necessary to pace the left atrial appendage in rare circumstances; in this case, it is important to remember how friable that structure can be.

Sequence of Grafting

Careful assessment of the cardiac catheterization is imperative. During on-pump cases, the location and number of vessels requiring bypass usually suffice during evaluation of the catheterization films. However, when planning for OPCAB, particular attention should be paid to the direction of collateral flow between coronary vessels, the presence of intramyocardial vessels, the size of the distal targets, the degree of stenosis, the complexity of coronary disease, and the number of lateral wall vessels requiring grafting. Careful attention must be paid to the sequence of grafting because regional myocardial perfusion is temporarily interrupted in the beating heart (Table 1). As a general rule, the collateralized vessel(s) is grafted first and the collateralizing vessel grafted last. For example, in patients with an occluded right coronary artery with a posterior descending artery supplied by collaterals from the left anterior descending artery, grafting the left anterior descending first would not only leave the anterior wall ischemic, but also disrupt flow to the septum, inferior wall, and right ventricle during the LAD anastomosis. Thus, a more prudent approach would involve grafting the posterior descending artery first, then performing a proximal anastomosis to ensure adequate flow while the proximal left anterior descending is temporarily occluded for construction of the LAD anastomosis. Another scenario that may pose problems is a large, moderately stenotic right coronary artery. Not uncommonly, temporary occlusion of this artery will result in profound bradycardia and hypotension. In these circumstances, the surgeon must be prepared to use an intracoronary shunt or promptly provide temporary epicardial pacing.

If LAD and Diagonal vessels are generally considered easy to demonstrate and graft, an increased level of difficulties and challenges apply for main right coronary artery, posterior descending artery, proximal and lateral obtuse marginal and ramus intermedius. The ramus intermedius artery is challenging to expose and stabilize due to compression on the right ventricular outflow tract and pulmonary artery. The introduction of suction devices to position the heart has decreased the hemodynamic alterations associated with exposing this difficult part of the heart. The apex of the heart is retracted toward the patient’s right hip, and the table is rotated toward the surgeon. This area of the heart tends to be tethered by the nearby pericardial reflection, reducing its mobility and making exposure difficult. Additionally, a large left atrial appendage can present visualization problems when grafting near the atrioventricular groove.

Rapid recovery of regional myocardial function prior to subsequent occlusion of other target arteries is essential for successful multivessel off-pump bypass grafting. Additional options include a “proximals first” approach to allow adequate regional perfusion after completion of each distal anastomosis. Although concern for myocardial protection during OPCAB stems from the brief periods of coronary occlusion necessary to visualize distal target vessels, adequate perfusion can be achieved by maintaining adequate systemic perfusion pressure, selective use of coronary artery shunts, careful use of traction sutures and stabilizers, and proper sequencing of graft anastomoses. Careful placement of intracoronary shunts is important because at least one study demonstrated significant endothelial injury with the use of intracoronary shunts.[27]

Distal Anastomosis

After cardiac positioning and coronary stabilization, the retractor tape can be placed, and the coronary artery dissected. If there are concerns about hemodynamic stability during regional ischemia, the proximal vessel can be test occluded for 2 to 5 minutes. During this time the graft can be prepared. This gives the surgeon some assurance before committing to the anastomosis by creating an arteriotomy. After a brief period of reperfusion of 2 to 3 minutes, the vessel can be reoccluded and the artery prepared for anastomosis. The anastomosis is otherwise performed in a manner identical to on-pump grafting. It is essential to continue communication with the anesthesia team so that adequate steps can be promptly taken if hemodynamic conditions deteriorate. For example, if pulmonary artery pressures begin to rise and mean arterial pressures begin to fall during a lateral wall anastomosis, several steps can be taken to avoid cardiovascular collapse. Gently relaxing on the cardiac positioner or coronary stabilizer can often improve hemodynamics. Optimizing table positioning, inotropes, vasopressors, fluid boluses, or pacing may also help. However, if it appears that hemodynamic conditions are deteriorating despite these interventions, then the safest next step is to place an intracoronary shunt, release both the coronary stabilizer and cardiac positioner, return the heart to the pericardial space, and allow hemodynamics to recover. At this point, a decision must be made to either convert “electively” to an on-pump procedure, including on-pump with the heart beating, or complete the procedure off-pump. With better preparation (eg, fluids, inotropes, vasopressors, pacing, shunt), the anastomosis can usually be completed off-pump. Another option that is frequently used in patients at high risk for complications of CPB is the use of intra-aortic balloon counterpulsation (IABP). An IABP can provide valuable mechanical support during cardiac displacement and positioning to enable safe and controlled completion of a distal anastomosis or entire CABG case that would otherwise require cardiopulmonary bypass.

Proximal Anastomosis

Traditionally, proximal anastomoses during OPCAB have been performed with the use of an aortic partial-occluding clamp. In preparation for an aortic clamp, the systolic blood pressure is lowered to less than 95 mmHg. Once the clamp is applied, aortotomies can be made with a standard aortic punch device. Proximal anastomoses are then performed using 6/0 polypropylene suture, and the author’s choice is for an RB2 needle or a custom-bent SH1 needle. Before tying down the most anterior proximal, the clamp is released and the aorta deaired through the proximal anastomosis. After the suture is tied down, the vein grafts can be deaired with a 25-gauge needle before removing their bulldog occlusion clamps. Arterial grafts are not punctured but are allowed to bleed backward before clamp removal.

Unlike on-pump coronary artery bypass, OPCAB provides the opportunity to minimize or completely avoid manipulation of the aorta. Avoiding partial clamping during proximal anastomoses can be achieved by performing proximal anastomoses to in situ arterial grafts, or using proximal automated anastomotic connectors or facilitating devices.[28] This may be particularly relevant in patients with advanced aortic atheromatous disease detected by epiaortic ultrasound. Commercially available devices for clampless proximal anastomoses include the Heartstring III (Maquet Cardiovascular, San Jose, CA) or PAS-Port Proximal Anastomosis System (Cardica, Redwood City, CA). The Heartstring device creates a hemostatic seal with the inner surface of the ascending aorta that allows the creation of a handsewn anastomosis with a relatively bloodless field. After completion of the anastomosis, the device is removed by unwinding the sealing cup from the aorta before tying down the suture; there is no foreign material other than suture material left in the anastomosis (Figure 5). However, this device still requires a handsewn anastomosis to be performed between the graft and the aorta, and can be associated with some blood loss.[29] The PAS-Port Proximal Anastomosis System was specifically designed to create a consistent anastomosis between a saphenous vein graft and the aorta during either on-or off-pump coronary bypass surgery. It is a fully integrated, automated system that cuts the aortotomy and attaches the vein graft to the aorta in seconds, producing consistent, reproducible anastomoses independent of surgical technique and skill. Compared with earlier devices, the PAS-Port system allows the endothelium of the vein graft to be untouched during the loading and deployment process;[30] however, there is a small amount of metallic foreign material left within the graft lumen.

Figure 5

Use of a Heartstring device to perform a proximal anastomosis without aortic crossclamping. Close-up surgeon’s view. After use of a standard aortic puncture device, the Heartstring device (A) creates a hemostatic seal with the inner surfaceof the ascending aorta that allows the creation of a handsewn anastomosis (B) within a relatively bloodless field. After completion of the anastomosis, the device is removed by unwinding the sealing cup from the aorta before tying down thesuture; there isno foreign material other than suture material left in the anastomosis.

A number of large retrospective studies have shown OPCAB to be associated with a decreased incidence of perioperative stroke compared with conventional on-pump CABG.[31] Conversely, randomized studies comparing the 2 revascularization strategies have often failed to show a significant advantage with OPCAB regarding postoperative stroke. This apparent contradiction may be explained by the fact that stroke is a relatively rare occurrence after CABG surgery, and even the largest randomized studies are underpowered to find a difference in the incidence of stroke.[32] A single-center institution reviewed more than 12000 patients who underwent primary isolated CABG, and compared the incidence of stroke in patients with a complete no-touch aortic manipulation technique (inflow from the mammary arteries) vs patients who underwent proximal anastomosis with a proximal facilitate device, vs patients where a proximal clamp on the aorta was adopted. No-touch aortic technique and proximal facilitate device were associated with a statistically significant reduction of perioperative stroke when compared with proximal side clamp.[33]

In OPCAB, we generally do not leave temporary epicardial atrial and ventricular wires, unless patients were paced during surgery or have low left ventricular ejection fraction.

Beating Heart Coronary Artery Bypass

Performing coronary artery bypass with CPB support is especially useful in certain clinical scenarios, such as acute coronary syndromes with cardiogenic shock or in patients with severe left ventricular dysfunction.[34],[35] In many instances, these patients already have tenuous hemodynamics and will not tolerate cardiac positioning and displacement during routine OPCAB maneuvers. Another approach is to cannulate the ascending aorta and right atrium and complete the anastomoses on CPB with the heart beating but decompressed. This provides for hemodynamic support and also eliminates the global ischemic insult associated with cardioplegic arrest. Grafting can then be performed in a similar sequence as OPCAB. This approach has been supported in several recent studies,[36],[37] specifically in the setting of acute coronary syndrome, as it provided improved outcomes with lower postoperative morbidity and mortality in patients undergoing emergency myocardial revascularization.

Nonsternotomy OPCAB Approach

The competitive status of percutaneous transluminal coronary angioplasty and stenting has stimulated an interest in minimally invasive direct coronary artery bypass grafting. This is allowed by absence of cardiopulmonary bypass, permitting surgeons to pursue a less invasive approach that spares the sternum, expedites the postoperative course, and decreases the length of stay. A left thoracotomy was adopted in the 1990s as an alternative route to offer OPCAB LITA-to-LAD grafting. Initially, specialized retractors and instruments simplified the LITA harvesting and allowed surgery through a minimally invasive direct thoracotomy (MIDCAB). Internal thoracic artery harvesting was further improved by the creation of the da Vinci Surgical System (Intuitive Surgical), which facilitates harvesting of single or double ITAs, and also allows completion of anastomosis to the anterior and lateral wall of the heart. We note that the left thoracotomy approach is also used in patients undergoing a reoperation CABG as an alternate route of access to the lateral wall, thereby avoiding a repeat sternotomy and potential injury to patent grafts.

Clinical outcomes after OPCAB and ONCAB have been compared for more than a decade, with enrollment across many centers and including hundreds of thousands of patients. Despite the abundance of literature, there is still no consensus regarding the optimal bypass strategy, especially in low-risk patients. In higher risk patients, it appears in recent studies that OPCAB may reduce both morbidity and mortality.

Studies may be divided into prospective randomized trials and observational retrospective analyses. Prospective trials provide the most accurate comparison between groups and avoid selection bias and confounding inherent to retrospective and observational analyses. However, due to resource constraints, these studies are smaller and thus statistically underpowered to detect incremental differences in morbidity or mortality rates following CABG. This remains true despite the recent completion and publication of 3 large, multicentered randomized trials—ROOBY,[39] GOPCAB,[40] and CORONARY[41],[42]—as the patient sample sizes required to demonstrate a significant difference in mortality would be more than 50000 patients, and similar sample sizes would be required to detect differences in stroke and myocardial infarction.[43] Retrospective and observational analyses provide the large cohort size and long duration of follow-up to sufficiently power these studies to detect small but important differences in outcomes. However, retrospective studies are inherently limited by biases, the most important of which is selection bias, which persists despite the use of propensity matching and other advanced statistical methodologies designed to control for confounding. Taken together, both types of studies can provide valuable information to guide clinical practice.

Early Outcomes

Three recent large, multicentered randomized trials have consistently reported no difference between OPCAB and ONCAB with regard to 30-day mortality, nonfatal stroke, nonfatal myocardial infarction, and new renal failure within 30 days after randomization (Table 2). These studies varied in terms of patient populations and surgeon expertise, and thus, the similar near-term mortality and morbidities across these studies support the belief that both techniques are similarly effective in the near-term, at least for low-risk and mixed-risk populations of patients.

More specifically, the CORONARY trial (the CABG Off or On Pump Revascularization Study)[41],[42] and the GOPCABE[40] trial (German Off-Pump CABG Trial in Elderly Patients) reported results following randomization of 4752 and 2539 patients, respectively. In comparison with the previous individual largest randomized trial (ROOBY Trial with 2203 patients),[39] the CORONARY and GOPCABE trials were designed to enroll sicker, older, and an appropriate proportion of female patients: these changes resulted in a cohort more similar to the composition of patients actually operated on, thereby increasing external validity of the results. In all 3 trials, approximately 60% to 70% of patients had 3-vessel disease. The predicted 30-day mortality risk was 1.9% in the ROOBY, 3.8% in the GOPCABE, and 80% of patients in the CORONARY had a EuroSCORE of 0.5 (ie, a predicted mortality of approximately 2%). Both the 30-day and the 1-year outcomes for the 3 trials are individually summarized in Tables 2 and 3, respectively. For the 3 trials, there was no significant difference in the 30-day composite primary endpoint between OPCAB and ONCAB, or in the individual incidence of death, MI, stroke, or need for new renal dialysis. Both the CORONARY and GOPCABE trials reported an increased risk of repeat revascularization within 30 days for OPCAB, with respective figures of 0.7% vs 0.2%, and 1.3% vs 0.4%.

Though controversial, compelling evidence supports the theory that OPCAB is associated with decreased operative mortality and morbidity (Figure 6), particularly for high-risk patients, such as those with high risk scores based on Society of Thoracic Surgeon (STS) risk scoring calculations (Figure 7), older patients, patients with renal failure, patients with reduced left ventricular ejection fraction, and patients who have undergone redo CABGs.[9],[11],[12],[39],[40],[41],[42],[43] Less controversial is that OPCAB offers a variety of anaortic techniques that have been shown to be complimentary in reducing the risk of stroke.[31]

Figure 6

Risk-adjusted odds ratios for major adverse events in all patients undergoing on-pump coronary artery bypass (ONCAB, patterned bar) vs off-pump procedures (OPCAB, filled bar). MACE indicates major adverse cardiac events; MI, myocardial infarction.

Figure 7

Comparison of observed mortality rates following off-pump coronary artery bypass grafting (OPCAB) and coronary artery bypass grafting (CABG) on cardiopulmonary bypass (CPB), based on patient preoperative Society of Thoracic Surgery (STS) predicted mortality risk score.

Long-Term Outcomes

Two potential benefits of ONCAB relative to OPCAB are increased completeness of revascularization and higher quality anastomoses, both of which would likely impact long-term survival following CABG. Randomized trials have consistently reported increased number of grafts with ONCAB vs OPCAB, even when studies require very experienced surgeons,[39],[40],[41] and decreased completeness of revascularization has been shown to be associated with decreased long-term survival.[45] However, 3 randomized trials that followed patients to 5 years and beyond reported no survival difference and almost identical survival curves.[39],[42],[46],[47] Furthermore, long-term follow-up of patients randomized to either treatment in modern studies has consistently shown no difference in individual outcomes, including stroke, repeat revascularization, myocardial infarction, need for reintervention, renal failure requiring dialysis, or composite endpoints of the above.[39],[42],[46],[47] Whereas the ROOBY trial—the first of the modern multicentered randomized trials—found superior graft patency at 1 year with ONCAB,[39] this study allowed inexperienced surgeons, and even residents, as the primary surgeon, a design element that has been widely questioned and corrected in subsequent trials.

A meta-analysis of 10709 patients across 3 randomized trials and 10 retrospective studies who were followed for more than 5 years found that OPCAB was associated with increased long-term mortality (hazard ratio, 1.05 [95% CI, 1.00-1.13]).[48] Of note, randomized controlled trials contributed less than 3% of the weight to this meta-analysis outcome, and this analysis did not include the results of CORONARY. Therefore, it is unclear whether the discrepancy between randomized trials and this meta-analysis is due to increased power due to larger sample size with the meta-analysis, or selection bias among the predominantly observational cohort. (It has been repeatedly noted that nonrandomized patients undergoing OPCAB generally have higher preoperative comorbidities than those undergoing ONCAB, as surgeons and centers tend to avoid CPB in higher-risk patients.)

Converted Patients

Conversion rates of off-pump to on-pump CABG among randomized trials have ranged from 0% to over 20%,[43] and reported mortality for this converted group ranged from 6% to 15%. In the recent CORONARY study, 8% of the off-pump cohort were converted, one-half of whom were converted due to hypotension or ischemia, and the other half were converted for small or intramuscular coronaries.[48] Conversions are a crucial outcome due to association with increased mortality and morbidity both in the near term and long term, particularly when patients are converted emergently.[48]